DTD 5
ARTICLE IN PRESS




International Journal of Food Microbiology xx (2005) xxx–xxx

www.elsevier.com/locate/ijfoodmicro




Alcohols, esters and heavy sulphur compounds production by
pure and mixed cultures of apiculate wine yeasts

Nathalie Moreira, Filipa Mendes, Tim Hogg, Isabel VasconcelosT

Escola Superior de Biotecnologia, Universidade Catolica Portuguesa, Rua Dr. Antonio Bernardino de Almeida, 4200-072, Porto, Portugal

Accepted 29 December 2004




Abstract


Strains of Hanseniaspora uvarum, Hanseniaspora guilliermondii and Saccharomyces cerevisiae were used as pure or mixed
starter cultures in commercial medium, in order to compare their kinetic parameters and fermentation patterns. In pure and
mixed cultures, yeasts presented similar ethanol yield and productivity. Pure cultures of H. uvarum and S. cerevisiae showed a
specific growth rate of 0.38 h ; however, this value decreased when these yeasts were grown in mixed cultures with H.
1

guilliermondii. The specific growth rate of pure cultures of H. guilliermondii was 0.41 h 1 and was not affected by growth of
other yeasts. H. guilliermondii was found to be the best producer of 2-phenylethyl acetate and 2-phenylethanol in both pure and
mixed cultures. In pure cultures, H. uvarum led to the highest contents of heavy sulphur compounds, but H. guilliermondii and
S. cerevisiae produced similar levels of methionol and 2-methyltetrahydrothiophen-3-one. Growth of apiculate yeasts in mixed
cultures with S. cerevisiae led to amounts of 3-methylthiopropionic acid, acetic acid-3-(methylthio)propyl ester and 2-
methyltetrahydrothiophen-3-one similar to those obtained in a pure culture of S. cerevisiae; however, growth of apiculate yeasts
increased methionol contents of fermented media.
D 2005 Elsevier B.V. All rights reserved.

Keywords: Hanseniaspora uvarum; Hanseniaspora guilliermondii; Saccharomyces cerevisiae; Secondary fermentation products; Heavy
sulphur compounds




1. Introduction must fermentation (Kunkee, 1984; Gao and Fleet,
1988; Zironi et al., 1993; Gil et al., 1996; Fleet, 2003).
Apiculate wine yeasts (Hanseniaspora uvarum and Their intolerance to high concentrations of ethanol,
Hanseniaspora guilliermondii) have become an the high sugar concentration and the low available
object of interest as they are frequently found in oxygen conditions during fermentation are the main
grapes and are also dominators of the early stages of reasons why Saccharomyces cerevisiae becomes
dominant and keeps its activity until the end of
fermentation (Goto, 1980; Fleet et al., 1984; Heard
T Corresponding author. Tel.: +351 22 5580049; fax: +351 22
5090351. and Fleet, 1985; Martinez et al., 1989; Fleet and
E-mail address: ivasc@esb.ucp.pt (I. Vasconcelos). Heard, 1993; Schutz and Gafner, 1993; Lema et al.,
¨

0168-1605/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijfoodmicro.2004.12.029




FOOD-03317; No of Pages 10

ARTICLE IN PRESS


2 N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx



1996; Constant et al., 1997; Egli et al., 1998; Hansen production of sulphur compounds by non-Saccharo-
et al., 2001). myces yeasts. In a study performed by Romano et al.
Growth of apiculate yeasts with S. cerevisiae must (1997b), several strains of Kloeckera apiculata and H.
be considered because it may influence the sensory guilliermondii were compared according to the pro-
quality of wine. Some studies evaluated the production duction of sulphur dioxide and hydrogen sulphide in a
of fermentation compounds by pure, mixed or sequen- basal synthetic medium. These authors observed that
tial cultures of apiculate yeasts with S. cerevisiae all strains produced less than 10 mg l 1 of sulphur
strains, using either grape must or basal synthetic dioxide and that K. apiculata produced higher amounts
medium (Herraiz et al., 1990; Mateo et al., 1991; of hydrogen sulphide than H. guilliermondii.
Velazquez et al., 1991; Zironi et al., 1993; Ciani and In order to understand the effect of growth of
Picciotti, 1995; Gil et al., 1996; Romano et al., 1997a,b; apiculate yeasts and how they contribute to the final
Ciani and Maccarelli, 1998; Rojas et al., 2001, 2003; composition of fermented media, experiments were
Zohre and Erten, 2002; Romano et al., 2003). These conducted using pure and mixed cultures of H.
experiments showed that there are significant differ- guilliermondii, H. uvarum and S. cerevisiae. The H.
ences in chemical composition of the resulting wines guilliermondii strain studied was isolated from grape
or fermented media. However, there is considerable musts of the Douro region, in Portugal. Fermentation
controversy concerning the effect of growth of apicu- kinetic parameters and the production of secondary
late yeasts on the organoleptic quality of wines. metabolites were evaluated in pure and mixed
Ciani and Picciotti (1995) exclude the possibility of cultures. Special attention was given to the heavy
using apiculate yeasts in winemaking, due to the sulphur compounds production profiles of Hansenias-
production of large amounts of ethyl acetate and acetic pora strains, as, to our knowledge, analysis of heavy
acid. Gil et al. (1996) observed that wines produced sulphur compounds produced by apiculate yeasts was
with mixed cultures presented a higher concentration of never reported. A basal commercial medium was used
alcohols and acids, in contrast with those fermented to characterize the fermentation pattern of Hansenias-
with pure cultures of S. cerevisiae. However, Herraiz pora strains in order to avoid interferences of grape
et al. (1990) found a higher content in higher alcohols must composition and provide easily reproducible
in wines fermented with Saccharomyces spp. than in growth conditions. Growth of Hanseniaspora strains
those fermented with pure cultures of apiculate yeasts. on simple media may help to explain results obtained
Experiments performed by Rojas et al. (2001, 2003) on grape musts fermentations.
reported that H. guilliermondii 11104 (CECT, Spain)
was a strong producer of 2-phenylethyl acetate.
According to Romano et al. (1997a,b, 2003), the 2. Materials and methods
synthesis of secondary products is an individual and
reproducible strain characteristic. 2.1. Yeast strains
Sulphur compounds comprise a structurally diverse
class of molecules that provides a whole range of The strains used in this study were H. guillier-
characteristic aromatic notes. Generally, the aromatic mondii NCYC 2380 (National Collection of Yeast
contributions of these compounds are considered Cultures, Norwich, UK), H. uvarum PYCC 4193T
detrimental to wine quality (Anocibar Beloqui and and S. cerevisiae PYCC 3507 (Portuguese Yeast
Bertrand, 1995; Mestres et al., 2000); however, new Culture Collection, Instituto Gulbenkian da Ciencia,
developments in wine research allowed the differ- Oeiras, Portugal). Yeasts were maintained on Yeast
entiation of a family of sulphur compounds responsible Malt agar slants (YM agar, Difco Laboratories,
for a varietal aroma of wines. The formation of sulphur Detroit, IN, USA).
compounds is affected by the organic and inorganic S-
containing substances and pesticides in grape musts, by 2.2. Fermentations
the nutrient level of grape musts and by the yeast
metabolism during fermentation (Rauhut, 1993). Very Pure cultures of H. guilliermondii (Hg), H. uvarum
few reports are available in literature concerning the (Hu) and S. cerevisiae (Sc) and mixed cultures (Hu–

ARTICLE IN PRESS


N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx 3



Hg, Hu–Sc, Hg–Sc, Hu–Hg–Sc) were carried out in guilliermondii cells. For mixed cultures of all yeast
200 ml of YM medium (Difco Laboratories, Detroit, strains, the number of H. uvarum cells was obtained
IN, USA), with approximately 10 g l 1 of glucose, at by the difference between the total cell number
25 8C, under gentle agitation (80 rpm). Each experi- obtained in YM agar with cycloheximide at 25 8C
ment was reproduced four times. Inocula of each yeast and the number of H. guilliermondii cells (YM agar
strain were previously grown at 25 8C for 24 h in YM with cycloheximide at 37 8C); the number of S.
medium. The inoculation of media was carried out in cerevisiae cells was obtained by the difference
order to obtain an initial cell concentration of 105–106 between the total cell number obtained in YM agar
cfu ml 1 of each strain. at 37 8C and the number of H. guilliermondii cells.

2.3. Enumeration of yeast populations 2.4. Analytical determinations

According to the characteristics of each yeast After fermentation, yeast cells were removed by
species, as defined by Barnett et al. (1990), it was centrifugation at 8000 rpm and 4 8C for 15 min. The
possible to define selective media and incubation supernatant was analysed using chromatographic
conditions that allow the differentiation of each procedures, according to the following methods.
yeast species. The number of yeast cells, expressed as The concentration of ethanol was determined by
cfu ml , was determined using the pour plate method,
1 High Performance Liquid Chromatography using a
after incubation of plates at specific temperatures for Beckman, System Gold. Separation was performed on
48 h. The medium used was YM agar with or without an AminexR HPX-87H column (300 7.8 mm, Bio-
addition of a selective component. YM agar allows the Rad) and detection was assessed by refractive index.
enumeration of viable yeast cells of all tested strains, The mobile phase was a 0.5 mM sulphuric acid
after incubation at 25 8C; if incubation is performed at solution, with a flow rate of 0.5 ml min , at 30 8C.
1

37 8C only H. guilliermondii and S. cerevisiae will Higher alcohols, ethyl acetate and acetaldehyde
grow, due to the growth inhibition of H. uvarum at this were analysed using a Hewlett-Packard 5890 gas
temperature. Plates of YM agar with 0.01% of cyclo- chromatograph equipped with a flame ionisation
heximide (Sigma Chemical, St Louis, MO, USA) were detector and connected to an H.P. 3396 Integrator.
used as a selective medium for Hanseniaspora enu- 50 Al of 4-methyl-2-pentanol at 10 g l 1 were added
meration, after incubation at 25 8C. Incubation of plates to 5 ml of fermented medium as an internal standard.
of YM agar with 0.01% of cycloheximide at 37 8C only The sample (1 Al) was injected (split, 1:60) into a CP-
allows growth of H. guilliermondii. WAX 57 CB column (Chrompack) of 50 m 0.25 mm
When mixed cultures were tested, plates of YM and 0.2 Am phase thickness. The temperature program
agar, incubated at 25 8C, were used for total cell was 40 8C (5 min) to 180 8C (0 min) at 3 8C min . 1

enumeration. For mixed cultures of H. uvarum and S. Injector and detector temperatures were set at 250 8C.
cerevisiae, plates of YM agar at 37 8C were used for Carrier gas was H2 at 1 ml min . 1

counting S. cerevisiae cells, whereas plates of YM The determination of 2-phenylethyl acetate and 2-
agar with 0.01% of cycloheximide, incubated at 25 phenylethanol was performed in a Perkin-Elmer
8C, were used for the enumeration of H. uvarum. In Autosystem, equipped with a flame ionisation detector.
mixed cultures of H. guilliermondii and H. uvarum, it 50 ml of sample, with 4-decanol at 1.5 mg l 1 as
is only possible to count H. guilliermondii cells by internal standard, was extracted successively with 4, 2
incubating plates of YM agar with cycloheximide and 2 ml of ether–hexane (1:1 v/v) for 5 min. The
(0.01%) at 37 8C; the number of H. uvarum cells was organic phase (1 Al) was injected (splitless, 0.3 min)
estimated by the difference between total cell number into a CP-WAX 58 (FFAP)-CB column (Chrompack)
and the number of H. guilliermondii cells. A similar of 50 m 0.32 mm and 0.3 Am phase thickness.
procedure was applied for mixed cultures of H. Temperature program was 40 8C (5 min) to 220 8C
guilliermondii and S. cerevisiae; the number of S. (20 min) at 2 8C min . Injector and detector temper-
1

cerevisiae cells was estimated by the difference atures were set 250 8C. The carrier gas used was H2 at
between total cell number and the number of H. 1–2 ml min . 1

ARTICLE IN PRESS


4 N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx



2-Methyltetrahydrothiophen-3-one, acetic acid-3- compounds by H. uvarum, H. guilliermondii and S.
(methylthio)propyl ester, methionol (3-(methylthio)-1- cerevisiae in pure and mixed cultures. Experiments
propanol) and 3-(methylthio)propionic acid were were conducted using commercial medium so that
determined according to the method described by results could be easier reproduced and compared.
Moreira et al. (2004). The concentrations of commer-
cially available sulphur compounds were expressed as 3.1. Enumeration of yeast population
Ag l . For those compounds whose reference stand-
1

ard was not available, the amounts were expressed as The viable number of cells was determined for
the ratio of peak area/peak area of internal standard. each sample of fermentation broth. Results obtained
For each analysed compound, the response of the on the different media used for cell enumeration (YM
detector was obtained using several standard solutions agar and YM+cycloheximide), at two incubation
with different concentrations. The reproducibility of temperatures (25 and 37 8C), were analysed using
each method was assessed from several analyses of the ANOVA analysis. For pure and mixed cultures, in
the same sample. conditions where the strains were able to grow, no
significant differences in cell enumeration were found
2.5. Statistical analysis related to the growth medium or temperature of
incubation used.
An analysis of variance (ANOVA) was applied to When mixed cultures of H. uvarum and S. cerevi-
the experimental data; results were considered sig- siae were tested, the total cell number and the sum of S.
nificant if the associated P value was below 0.05. The cerevisiae and H. uvarum cell numbers were not
significant differences were determined by Tukey significantly different. In mixed cultures of all strains,
tests. All statistical analyses were performed using the again no significant differences were found between
software SPSS for Windows, version 10.0. the total cell number and the sum of S. cerevisiae, H.
guilliermondii and H. uvarum cell numbers.


3. Results and discussion 3.2. Growth behaviour

Experiments were performed in order to evaluate Growth kinetics and ethanol production by each
the fermentation kinetic parameters and the produc- yeast strain was followed during fermentation. Figs. 1
tion of higher alcohols, esters and heavy sulphur and 2 represent the evolution of yeast population


1e+8 0,8

0,7
(%)
0,6
1e+7
0,5
-1
ml
0,4
cfu
concentration 0,3
1e+6

0,2

(A) Ethanol 0,1 (B)

1e+5 0,0
0 10 20 30 40 0 10 20 30 40
Time (hours) Time (hours)

Fig. 1. Growth kinetics (A) and ethanol production (B) by pure cultures of S. cerevisiae (n, 5), H. guilliermondii ( .,o) and H. uvarum (E, 4) on
a commercial medium. values for colony forming units are the average values of results obtained from four fermentations. Vertical bars represent
standard deviation.

ARTICLE IN PRESS


N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx 5


1e+8 0,8 1e+8 0,8
Ethanol Ethanol
0,7 0,7

0,6 0,6
1e+7 concentration 1e+7 concentration
-1 0,5 0,5
-1
ml
0,4 ml
0,4
cfu
0,3 cfu
0,3
1e+6 1e+6
0,2 0,2
(%) (%)
(A) 0,1
(B) 0,1

1e+5 0,0 1e+5 0,0
1e+8 0,8
Ethanol 1e+8 0,8 Ethanol
0,7 0,7

0,6 0,6
1e+7 concentration
1e+7 concentration
-1 0,5
-1 0,5
ml
0,4 ml
0,4
cfu
0,3 cfu
1e+6 0,3
1e+6
0,2 (%) 0,2 (%)

(C) 0,1
(D) 0,1

1e+5 0,0 1e+5 0,0
0 10 20 30 40 0 10 20 30 40
Time (hours) Time (hours)

Fig. 2. Growth kinetics (solid symbols) and ethanol production (o) by mixed cultures of H. uvarum (E), H. guilliermondii ( ) and S. .
cerevisiae (n) on a commercial medium. (A) H. uvarum and H. guilliermondii, (B) S. cerevisiae and H. guilliermondii, (C) S. cerevisiae and H.
uvarum and (D) S. cerevisiae, H. guilliermondii and H. uvarum. values for colony forming units are the average values of results obtained from
four fermentations. Vertical bars represent standard deviation.



when respectively pure and mixed cultures were used. wine yeasts. Experiments performed by Charoenchai
After a short lag phase, yeasts started the exponential et al. (1998), in a chemically defined grape juice
growth, increasing the viable population to 107–108 medium, showed specific growth rates of 0.16–0.17
cfu ml . In pure or mixed cultures, apiculate yeasts
1 h 1 for S. cerevisiae and 0.15–0.17 h 1 for K.
achieved its highest cell mass concentration after apiculata (values estimated from plots of the log of
approximately 8–12 h of fermentation, and started the optical density against time, using the straight line of
decline phase after approximately 30 h of fermenta- the exponential growth phase). Ciani and Picciotti
tion. In general, H. uvarum attained this phase earlier (1995), using a modified grape juice, reported specific
than H. guilliermondii, whereas S. cerevisiae kept its growth rates of 0.14 h 1 for H. uvarum, 0.23 h 1 for
activity for a longer period. 1
K. apiculata and 0.26 h for S. cerevisiae.
In pure cultures, the specific growth rate of S. Under the conditions tested, ethanol productivity
cerevisiae and H. uvarum was 0.38 h 1 (Table 1). H. and ethanol yield obtained for apiculate yeasts were
guilliermondii presented a slightly higher specific similar to those found in pure cultures of S. cerevisiae
growth rate value of 0.41 h . In mixed cultures, the
1 (Table 1). The main difference was found in the
specific growth rate of H. guilliermondii was not fermentation carried out with a mixed culture of
affected; however, in mixed culture of all yeasts, the apiculate yeasts. This culture exhibited the lowest
specific growth rates of S. cerevisiae and H. uvarum values for ethanol yield (41%) and ethanol produc-
decreased to 0.33 h 1 and 0.26 h , respectively.
1 tivity (0.14 g l 1 h ). According to Ciani and
1

Only a few studies reported the kinetic parameters of Picciotti (1995), H. uvarum, K. apiculata and S.

ARTICLE IN PRESS


6 N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx


Table 1
Fermentation parameters of pure and mixed cultures of H. uvarum (Hu), H. guilliermondii (Hg) and S. cerevisiae (Sc) on a commercial medium

Fermentation Pure cultures Mixed cultures
parameters
Hu Hg Sc Hu–Hg Hu–Sc Hg–Sc Hu–Hg–Sc

Ethanol production 0.60 0.64 0.67 0.54 0.62 0.65 0.61
(%, v/v) (0.03) (0.02) (0.02) (0.01) (0.02) (0.02) (0.02)
Yeth (%, w/w) 45 48 50 41 46 48 46
Qeth (g l 1 h ) 1 0.15 0.16 0.17 0.14 0.16 0.16 0.16


Hu Hg Hu Sc Hg Sc Hu Hg Sc

lx (h )
1 0.38 0.41 0.38 0.33 0.40 0.39 0.40 0.42 0.37 0.26 0.41 0.33
(0.02) (0.02) (0.04) (0.06) (0.05) (0.04) (0.02) (0.03) (0.03) (0.01) (0.03) (0.05)

values in parenthesis are standard deviations from four determinations; Yeth=ethanol yield (ratio between the maximum ethanol level produced
and the initial sugar concentration of the media); Qeth=ethanol productivity (ratio between maximum ethanol produced and fermentation time,
considered as the moment where ethanol concentration became constant); lx=maximum specific growth rate (slope of the least square
regression line of the natural logarithm of cell number vs. time data during the exponential growth phase).



cerevisiae presented ethanol yields of 0.47, 0.53 and concentration of fermented media. Similar results
0.61 (ml g ) respectively, and ethanol productivities
1 were obtained by Romano et al. (1997a) using a basal
of 0.48, 0.51 and 1.36 ml l 1 h 1 respectively in a synthetic medium, where apiculate yeasts produced
modified grape juice. 1.4–6.1 mg l 1 of 1-propanol, 5.2–11.2 mg l 1 of 2-
methyl-1-propanol, 7.7–12.6 mg l 1 of 2-methyl-1-
3.3. Composition of the fermented media butanol and 18.5–23.2 mg l 1 of 3-methyl-1-butanol.
In wines, higher alcohols are quantitatively dominant
3.3.1. Higher alcohols and important in the sensory properties and quality.
No significant differences in 1-propanol, 2-methyl- Below 300 mg l 1 higher alcohols contribute pos-
1-butanol and 3-methyl-1-butanol concentrations were itively to wine quality, while excessive amounts
observed in the media fermented by pure or mixed (higher than 400 mg l 1 ) may detract quality
cultures (Table 2). The fermented media by a pure (Soufleros and Bertrand, 1979; Rapp and Versini,
culture of S. cerevisiae presented a higher content in 1991; Lambrechts and Pretorius, 2000).
2-methyl-1-propanol than those obtained from pure
cultures of H. uvarum and H. guilliermondii (Fig. 3.3.2. Acetaldehyde and ethyl acetate
3A); however, the content in 2-methyl-1-propanol was No significant differences in ethyl acetate were
similar in mixed cultures. No significant differences obtained in media fermented by pure or mixed
were also obtained for the higher alcohols total cultures (Table 2). Variations in acetaldehyde con-


Table 2
Concentration of major volatile compounds produced by pure and mixed cultures of H. uvarum (Hu), H. guilliermondii (Hg) and S. cerevisiae
(Sc) on a commercial medium

Concentration (mg l ) 1 Pure cultures Mixed cultures Sig.

Hu Hg Sc Hu–Hg Hu–Sc Hg–Sc Hu–Hg–Sc

1-Propanol 5.06 (0.43) 6.42 (1.90) 6.94 (1.21) 5.01 (0.79) 6.11 (1.46) 5.06 (0.43) 6.15 (0.86) ns
2-Methyl-1-butanol 10.8 (3.1) 11.6 (0.9) 11.5 (0.5) 9.78 (0.68) 9.86 (2.15) 10.9 (2.3) 10.1 (0.5) ns
3-Methyl-1-butanol 14.8 (4.3) 19.5 (4.6) 11.8 (2.4) 12.8 (2.6) 17.2 (3.3) 14.8 (3.2) 17.2 (3.0) ns
Total higher alcohols 39.4 (8.8) 45.0 (9.2) 42.8 (5.4) 33.2 (4.5) 40.6 (8.5) 39.5 (6.9) 41.7 (6.0) ns
Ethyl acetate 9.07 (0.12) 9.74 (0.21) 9.85 (1.46) 6.77 (0.11) 9.33 (2.11) 9.07 (0.12) 9.30 (1.91) ns
Acetaldehyde 68.7 (7.2)a,b 78.7 (12.6)b 93.4 (3.2)b 81.1 (5.2)b 39.5 (1.3)a 68.7 (3.3)a,b 57.7 (5.4)a,b y

values in parenthesis are standard deviations from four determinations; Sig.: significance, y displays the significance at 1%, ns—not significant;
values not sharing the same superscript letter (a, b) within the horizontal line are different according to the Tukey test.

ARTICLE IN PRESS


N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx 7



) 16
-1 l b (A) 2-methyl-1-propanol
12
(mg ab a ab ab
a

8 a


4

Concentration
0

) 10
-1 l c
bc (B) 2-phenylethanol
8
(mg
b
6
b
4

2 a a a

Concentration
0
) 16
-1 l c (C) 2-phenylethyl acetate
12
(mg

8

b
ab
4
ab

Concentration a a a
0
Hu Hg Sc Hu-Hg Hu-Sc Hg-Sc Hu-Hg-Sc

Fig. 3. Concentration of (A) 2-methyl-1-propanol, (B) 2-phenylethanol and (C) 2-phenylethyl acetate in pure and mixed cultures of H. uvarum
(Hu), H. guilliermondii (Hg) and S. cerevisiae (Sc) on a commercial medium. values not sharing the same superscript letter on top bar are
different according to the Tukey test. Vertical bars represent standard deviation.



centrations were observed, but it was not possible to 3.3.3. 2-Phenylethanol and 2-phenylethyl acetate
correlate them to the experiments performed. The analysis of variance of data shows a
Romano et al. (1997a,b) also found different con- significant effect of the yeast strain on the amount
tents in acetaldehyde and ethyl acetate in synthetic of 2-phenylethanol and 2-phenylethyl acetate in the
media fermented by different apiculate yeast strains. fermented media (Fig. 3B and C). The highest
Studies performed using grape musts inoculated with concentrations of these compounds were observed
apiculate yeasts showed that the resulting wines in media fermented by pure and mixed cultures of H.
presented large amounts of these compounds, includ- guilliermondii. In pure cultures, this species was able
ing acetic acid (Benda, 1982; Herraiz et al., 1990; to produce 6.30 mg l 1 of 2-phenylethanol, while H.
Ciani and Maccarelli, 1998; Schutz and Gafner, ¨ uvarum and S. cerevisiae produced less than
1993). Excessively high contents of ethyl acetate do 1.12 mg l . In mixed cultures with H. guilliermon-
1

not improve the aroma of young wines, but at low dii, a high content in 2-phenylethanol was obtained,
contents (50–80 mg l ) it contributes to wine 1 reaching 7.50 mg l 1 in media fermented by
quality (Ribereau-Gayon et al., 2000); it has also apiculate yeasts. 2-Phenylethyl acetate was only
been reported that the negative effect of high levels detected in pure and mixed cultures of H. guillier-
of this compound may be reduced during bottle mondii. In pure culture, H. guilliermondii produced
aging (Lilly et al., 2000). 11.1 mg l 1 of this compound. Rojas et al. (2001),

ARTICLE IN PRESS


8 N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx



using a synthetic medium, under anaerobic condi- bringing fruity and flowery flavours (Rapp and
tions, also reported that H. guilliermondii 11104 Mandery, 1986). Non-Saccharomyces wine yeasts
produced 28.9 mg l 1 of 2-phenylethyl acetate, are good producers of esters and their use has been
whereas fermented media by pure cultures of H. suggested as mixed starters together with S. cerevi-
uvarum 1444 and S. cerevisiae T73 only presented siae to improve the sensory properties of wine. It
0.22 mg l 1 and 0.25 mg l , respectively. From an 1 was reported that yeast strains of H. guilliermondii
oenological point of view, these compounds pro- are able to promote the esterification of various
duced during fermentation contribute significantly to alcohols such as ethanol, geraniol, isoamyl alcohols
the desirable aspects of the bouquet of wine, and 2-phenylethanol (Rojas et al., 2001).


2,5
) (A) methionol
-1 l 2,0 b b b
b
(mg b
1,5 ab

1,0
a
0,5
Concentration
0,0
750
) (B) 3-methylthiopropionic acid
-1 l c
600
g(
450 b b
b b

300

150
Concentration
0 a a

0,4
area (C) acetic acid-3-(methylthio)propyl ester

0,3 b
peak b b
area/
0,2
a
peak standard a
a
0,1
a

internal 0,0
100
) b
-1 (D) 2-methyltetrahydrothiophen-3-one
l
g( 80

60
a
a a
40
a a

20 a
Concentration
0
Hu Hg Sc Hu-Hg Hu-Sc Hg-Sc Hu-Hg-Sc

Fig. 4. Concentration of heavy sulphur compounds in pure and mixed cultures of H. uvarum (Hu), H. guilliermondii (Hg) and S. cerevisiae (Sc)
on a commercial medium. (A) Methionol, (B) 3-methylthiopropionic acid, (C) acetic acid-3-(methylthio)propyl ester, (D) 2-methyltetrahy-
drothiophen-3-one. values not sharing the same superscript letter on top bar are different according to the Tukey test. Vertical bars represent
standard deviation.

ARTICLE IN PRESS


N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx 9



3.3.4. Heavy sulphur compounds with a pure culture of S. cerevisiae. Pure cultures of
Under the conditions tested, production of heavy H. guilliermondii and S. cerevisiae showed similar
sulphur compounds was influenced by the yeast strain contents of methionol and 2-methyltetrahydrothio-
used. In general, pure cultures of H. uvarum led to the phen-3-one, but H. guilliermondii produced higher
highest production of heavy sulphur compounds. In levels of acetic acid-3-(methylthio)propyl ester and 3-
pure cultures, the fermented medium by S. cerevisiae methylthiopropionic acid. Concentrations of heavy
presented 470 Ag l 1 of methionol, whereas higher sulphur compounds were also higher in a pure culture
amounts were found in fermented media by apiculate of H. uvarum than in a pure culture of S. cerevisiae.
yeasts (Fig. 4A). Growth of apiculate yeasts increased Nevertheless, except for methionol, levels of heavy
methionol content in mixed cultures. 3-Methylthio- sulphur compounds in mixed cultures of apiculate
propionic acid (Fig. 4B) was not detected in a pure yeasts with S. cerevisiae were similar to those
culture of S. cerevisiae, which also presented a low obtained in a pure culture of S. cerevisiae.
content in acetic acid-3-(methylthio)propyl ester (Fig. Although further research is needed, results
4C). A lower production of acetic acid-3-(methyl- obtained in this work on growth of Hanseniaspora
thio)propyl ester was also obtained in mixed cultures strains on simple media, and reports in literature on
with S. cerevisiae. The highest content in 2-methyl- growth of apiculate yeasts on grape musts, suggest
tetrahydrothiophen-3-one was obtained for a pure that the use of mixed cultures in wine fermentation
culture of H. uvarum; however, growth of this strain processes, combined with vinification technology,
had no effect on 2-methyltetrahydrothiophen-3-one may lead to the production of wines with different
concentration in mixed cultures (Fig. 4D). Methionol characteristics.
is present in wines at concentrations up to 5 mg l , 1

and above its threshold value (1.2 mg l 1 in hydro-
alcoholic solution) it attributes a cauliflower aroma. 2- Acknowledgements
Methyltetrahydrothiophen-3-one (metallic, natural gas
odour) and acetic acid-3-(methylthio)propyl ester The authors gratefully acknowledge the financial
(cooked potatoes aroma) are usually found in wines support from FCT and FSE (III Quadro Comunitario
at levels below their threshold value. The concen- de Apoio) and PAMAF (INIA, Project 2025).
tration of 2-methyltetrahydrothiophen-3-one in wines
is usually lower than 60 Ag l ; however, in reduced
1

wines, with disagreeable odours, it is present at
References
concentrations higher than 100 Ag l . 3-Methylthio-
1

propionic acid has a limit of perception of 50 Ag l 1
Anocibar Beloqui, A., Bertrand, A., 1995. Study of sulfur
(in hydroalcoholic solution), but it can be found in compounds in wine: preliminary results. Italian Journal of Food
wines at higher contents, up to 310 Ag l 1 (Anocibar Science 3, 279–288.

Beloqui and Bertrand, 1995; Ribereau-Gayon et al., Barnett, J.A., Payne, R.W., Yarrow, D., 1990. Yeasts, Characteristics
and Identification. Cambridge University Press, Cambridge.
2000; Mestres et al., 2000).
Benda, I., 1982. Wine and brandy. In: Reed, G. (Ed.), Prescott and
Dunn’s industrial microbiology. Avi Technical Books, Westport,
pp. 292–402.
4. Conclusions Charoenchai, C., Fleet, G.H., Henschke, P.A., 1998. Effects of
temperature, pH, and sugar concentration on the growth rates
and cell biomass of wine yeasts. American Journal of Enology
Results presented here show that apiculate yeasts
and Viticulture 49, 283–288.
are able to grow in the presence of S. cerevisiae and Ciani, M., Maccarelli, F., 1998. Oenological properties of non-
influence the composition of the fermented media. H. Saccharomyces yeasts associated with wine-making. World
guilliermondii showed desirable growth kinetic Journal of Microbiology and Biotechnology 14, 199–203.

parameters and the ability to produce high levels of Ciani, M., Picciotti, G., 1995. The growth kinetics and fermentation
behaviour of some non-Saccharomyces yeasts associated with
2-phenylethyl acetate and 2-phenylethanol. Growth of
wine-making. Biotechnology Letters 17, 1247–1250.
apiculate yeasts together with S. cerevisiae reduced Constant, M., Poblet, M., Arola, L., Mas, A., Guillamon, J., 1997.
the content of 2-methyl-1-propanol, when compared Analysis of yeast populations during alcoholic fermentation in a

ARTICLE IN PRESS


10 N. Moreira et al. / International Journal of Food Microbiology xx (2005) xxx–xxx


newly established winery. American Journal of Enology and aroma of wine. International Journal of Food Microbiology 14,
Viticulture 48, 339–344. 153–160.
Egli, C.M., Edinger, W.D., Mitrakul, C.M., Henick-Kling, T., 1998. Mestres, M., Busto, O., Guasch, J., 2000. Analysis of organic sulfur
Dynamics of indigenous and inoculated yeast populations and compounds in wine aroma. Journal of Chromatography, A 881,
their effect on the sensory character of Riesling and Chardonnay 569–581.
wines. Journal of Applied Microbiology 85, 779–789. Moreira, N., Guedes de Pinho, P., Vasconcelos, I., 2004. Method for
Fleet, G.H., 2003. Yeast interactions and wine flavour. International analysis of heavy sulphur compounds using gas chromatog-
Journal of Food Microbiology 86, 11–22. raphy with flame photometric detection. Analytica Chimica
Fleet, G.H., Heard, G.M., 1993. Yeasts-growth during fermentation. Acta 513, 183–189.
In: Fleet, G.H. (Ed.), Wine Microbiology and Biotechnology. Rapp, A., Mandery, H., 1986. Wine aroma. Experientia 42,
Harwood Academic Publishers, Chur, Switzerland, pp. 27–55. 873–884.
Fleet, G.H., Lafon-Lafourcade, S., Ribereau-Gayon, P., 1984. Rapp, A., Versini, G., 1991. Influence of nitrogen compounds in
Evolution of yeasts and lactic acid bacteria during fermentation grapes on aroma compounds of wines. In: Rantz (Ed.),
and storage of Bordeaux wines. Applied and Environmental Proceedings of the International Symposium on Nitrogen in
Microbiology 48, 1034–1038. Grapes and Wine. American Society for Enology and Viti-
Gao, C., Fleet, G.H., 1988. The effects of temperature and pH on culture, Davis, CA, pp. 156–164.
the ethanol tolerance of the wine yeasts. Saccharomyces Rauhut, D., 1993. Production of sulfur compounds. In: Fleet, G.H.
cerevisiae, Candida stellata, and Kloeckera apiculata. Journal (Ed.), Wine Microbiology and Biotechnology. Harwood Aca-
of Applied Bacteriology 65, 405–409. demic Publishers, Chur, Switzerland, pp. 183–242.
Gil, J., Mateo, J., Jimenez, M., Pastor, A., Huerta, T., 1996. Aroma Ribereau-Gayon, P., Glories, Y., Manjean, A., Dubourdieu, D., 2000.
compounds in wine as influenced by apiculate yeasts. Journal of Handbook of Enology. The chemistry of wine stabilization and
Food Science 61, 1247–1266. treatments, vol. 2. John Wiley and Sons Ltd, Chichester.
Goto, S., 1980. Changes in the wild yeast flora of sulphited grape Rojas, V., Gil, J., Pinaga, F., Manzanares, P., 2001. Studies on
musts. Journal of Institute of Enology and Viticulture Yamana- acetate ester production by non-Saccharomyces wine yeasts.
shi University 15, 29–32. International Journal of Food Microbiology 70, 283–289.
Hansen, H., Nissen, P., Sommer, P., Nielsen, J.C., Arneborg, N., Rojas, V., Gil, J., Pinaga, F., Manzanares, P., 2003. Acetate
2001. The effect of oxygen on the survival of non-Saccha- ester formation in wine by mixed cultures in laboratory
romyces yeasts during mixed cultures fermentations of grape fermentations. International Journal of Food Microbiology 86,
juice with Saccharomyces cerevisiae. Journal of Applied 181–188.
Microbiology 91, 541–547. Romano, P., Suzzi, G., Comi, G., Zironi, R., Maifreni, M., 1997a.
Heard, G.M., Fleet, G.H., 1985. Growth of natural yeast flora Glycerol and other fermentation products of apiculate wine
during the fermentation of inoculated wines. Applied and yeasts. Journal of Applied Microbiology 82, 615–618.
Environmental Microbiology 50, 727–728. Romano, P., Suzzi, G., Domizio, P., Fatichenti, F., 1997b. Secondary
Herraiz, T., Reglero, G., Herraiz, M., Martn-Alvarez, P., Cabezudo, products formation as a tool for discriminating non-Saccharo-
M.D., 1990. The influence of the yeast and type of culture on the myces wine strains. Antonie van Leeuwenhoek 71, 239–242.
volatile composition of wines fermented without sulfur dioxide. Romano, P., Fiore, C., Paraggio, M., Caruso, M., Capece, A., 2003.
American Journal of Enology and Viticulture 41, 313–318. Function of yeasts species and strains in wine flavour. Interna-
Kunkee, R.E., 1984. Selection and modification of yeasts and tional Journal of Food Microbiology 86, 169–180.
lactic acid bacteria for wine fermentation. Food Microbiology Schqtz, M., Gafner, J., 1993. Analysis of yeast diversity during
1, 315–332. spontaneous and induced alcoholic fermentations. Journal of
Lambrechts, M.G., Pretorius, I.S., 2000. Yeast and its importance to Applied Bacteriology 75, 551–558.
wine aroma. A review. South African Journal of Enology and Soufleros, E., Bertrand, A., 1979. Role de la souche de levure dans
Viticulture 21, 97–129. la production des substances volatiles en cours de la fermenta-
Lema, C., Garcia-Jares, C., Orriols, I., Angulo, L., 1996. Contribution
tion du jus du raisin. Connaissance de la Vigne et du Vin 13,
of Saccharomyces and non-Saccharomyces populations to the 181–198.
production of some components of Albarino wine aroma. Velazquez, J., Longo, E., Sieiro, C., Cansado, J., Calo, P., Villa, T.,
American Journal of Enology and Viticulture 47, 206–216. 1991. Improvement of the alcoholic fermentation of grape juice
Lilly, M., Lambrechts, M.G., Pretorius, I.S., 2000. Effect of with mixed cultures of Saccharomyces cerevisiae wild strains.
increased yeast alcohol acetyltransferase activity on flavour Negative effect of Kloeckera apiculate. World Journal of
profiles of wine and distillates. Applied and Environmental Microbiology and Biotechnology 7, 485–489.
Microbiology 66, 744–753. Zironi, R., Romano, P., Suzzi, G., Battistutta, F., Comi, G., 1993.
Martinez, J., Millan, C., Ortega, J.M., 1989. Growth of natural flora Volatile metabolites produced in wine and sequential cultures of
during the fermentation of inoculated musts from dPedro Hanseniaspora guilliermondii or Kloeckera apiculata and
XimenezT grapes. South African Journal of Enology and Saccharomyces cerevisiae. Biotechnology Letters 15, 235–238.
Viticulture 10, 31–35. Zohre, D.E., Erten, H., 2002. The influence of Kloeckera apiculata
Mateo, J., Jimenez, M., Huerta, T., Pastor, A., 1991. Contribution of and Candida pulcherrima yeasts on wine fermentation. Process
different yeasts isolated from musts of Monastrell grapes to the Biochemistry 38, 319–324.


Process Biochemistry 40 (2005) 2781–2791
www.elsevier.com/locate/procbio




Modelling of the alcohol dehydrogenase production in baker’s yeast

A. Vrsalovic Presecki, . Vasic-Racki *
ˇ ˇ

Faculty of Chemical Engineering and Technology, University of Zagreb, Savska c. 16, HR-10000 Zagreb, Croatia

Received 24 May 2004; received in revised form 8 November 2004; accepted 11 December 2004




Abstract


A mathematical model was formulated to simulate cell growth and enzyme production during the aerobic and micro-aerobic culture of the
yeast Saccharomyces cerevisiae. Model development was based on three simplied metabolic events in the yeast: glucose fermentation,
glucose oxidation and ethanol oxidation. Cell growth was expressed as a composite of these metabolic events. Their contributions to the total
specic growth rate depended on the activities of the pacemaker enzyme pools of the individual pathways. The effect of substrate
concentrations on the specic growth rate was described by a Michaelis–Menten equation. It was assumed that enzyme formation is cell
growth associated. The model successfully predicted the dynamics of cell growth, glucose consumption, ethanol metabolism and alcohol
dehydrogenase (ADH) production. A good agreement between model simulations and experimental data was achieved. It was observed that
ADH production depends on the available oxygen concentration in the medium. In general, the proposed model appears to be useful for the
design, scale-up, control and optimization of alcohol dehydrogenase production.
# 2005 Elsevier Ltd. All rights reserved.


Keywords: Baker’s yeast; Enzyme production; Alcohol dehydrogenase; Modelling; Fermentation; Oxygen




1. Introduction One of the products of the baker’s yeast metabolism is the
enzyme, complex structured protein alcohol dehydrogenase
Saccharomyces cerevisiae is the most popular industrial (ADH). ADH belongs to oxidoreductases. The isolated
microorganism because it utilises cheap materials for growth enzyme has a molecular weight of about 150,000 [11],
and production. Wild type yeasts are considered as excellent quaternary structure stabilised by Zn-ions. Therefore, it
objects for research and development, because they are robust belongs to a methaloenzyme group. It is dependent on the
and sufciently stable in cultivations over an extended period coenzyme nicotinamide adenine dinucleotide, which is
of time [1]. It is the most thoroughly investigated yeast as an involved, in two-electron oxidations or reductions. YADH
experimental model due to its distinct feature of glucose catalyses reversible reaction of alcohol oxidation and is
metabolism [2–5] and importance in industry. This organism active mainly on unbranched primary (C2–C12) and
has been already accepted as non-pathogenic, i.e. safe secondary (C3–C14) aliphatic alcohols [12], making it
producer, which can be easily manipulated genetically and useful in biotransformation processes. The advantage of
grown on simple and cheap media compared to that of animal YADH is its steroselectivity, which enables enantioselective
cell cultivation. Also, it is capable of post-translation reductions of 2-ketones to the S isomer of the corresponding
modications of produced proteins (e.g. glycosylation, secondary alcohols [13–16]. Enzyme YADH is also used for
phosphorylation and acetylation). Thus, apart from the preparation of aldehydes by primary alcohols oxidation [17].
classical applications in production of wine, beer, baker’s Due to its possibility to oxidise ethanol, the most important
yeast, and ethanol, S. cerevisiae is also a potential expression application of the enzyme YADH is for the determination of
system of recombinant proteins [6–10]. ethanol in foodstuffs like alcoholic beverages, fruit juices,
vinegar, chocolates, jam, honey etc. [18]. As analytical
enzyme, it is also used for the ethanol determination in

* Corresponding author. Tel.: +385 1 4597104; fax: +385 1 4597133. blood. Recently, a semi industrial scale for the hexanol
E-mail address: dvracki@marie.fkit.hr (. Vasic-Racki).
ˇ continuous production from hexanal in a gas phase by whole

0032-9592/$ – see front matter # 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.procbio.2004.12.020

2782 A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ


cell yeast ADH has been established [19]. The same enzyme available information of baker’s yeast behaviour [27–30] in
was incorporated in that process, as NAD(H) regenerating mathematical model for the optimising baker’s yeast growth
enzyme with ethanol added as cosubstrate. In a novel and the production of enzyme alcohol dehydrogenase.
processes for the production of heterologus proteins in high
yields (hirudins, miniproinsulins, leptons), yeast ADH 2.1. Model assumptions
promoter system is also applied [20].
The production of alcohol dehydrogenase (ADH) is Our proposed model is based on the following assump-
directly connected to dissolved oxygen concentration, tions [28]:
carbon source and ethanol, which is the product of a
reductive metabolic pathway. ADH is the enzyme that is - reactor contains two phases (gas–liquid system), micro-
produced in the cell to synthesise and to oxidize ethanol. The organisms are not considered as a separate phase because
production of the ethanol takes place only if the oxidation of of their small size and water-like density;
pyruvate through the tricarboxylic acid cycle is stopped due - the reactor contents are considered homogenous in axial
to a lack of oxygen, which is the main driving force for the and radial directions;
operation of this cycle or alternatively too much pyruvate is - energy balance were not considered since effective
being produced from glucose for it to be handled in the temperature control was accomplished;
mitochondrion [21]. Whereas the production of pyruvate is - the mass transfer between gas and liquid phase is
connected with the formation of reducing equivalents explained by lm model which is also incorporated into
(NADH), which must be reoxidized, a cell needs to nd the process model [29].
some other mechanism than oxidative phosporylation to
reoxidize it. So, the conversion of pyruvate to ethanol is a In order to describe the change of the enzyme activity of
good solution because it is a reductive process linked in nal the enzyme alcohol dehydrogenase the following two ass-
steps to the reoxidation of NADH (Fig. 1). Taking the umptions were introduced:
previous fact into consideration it can be assumed that the
highest ADH production would be achieved during the - enzyme production is following the biomass growth, i.e.
oxidative–reductive baker’s yeast cultivations. The oxida- enzyme production is growth associated [23];
tive pathway is necessary for the formation of biomass - the enzyme deactivation caused by inappropriate condi-
because the ADH is intracellular product and the reductive tions in a bioreactor (temperature, pH) can be described by
for the ethanol synthesis [22]. the rst order kinetic [26].


2.2. Mass balances in batch bioreactor
2. Modelling
Taking model assumptions into consideration, balance
Kinetic studies form a vital part of the overall equations for biomass, glucose, ethanol oxygen and volume
investigation of the product formation in fermentations. A activity of the enzyme ADH for the batch reactor have been
mathematical model allows easy data analysis and provides derived [27,28]:
a strategy for optimising the production. Models for dcX
describing enzyme production are usually based on the mOX mRED mEt cX (1)
dt
Luedeking & Piret study [23] for the product formation with
dcG
some modications that are characterised for the specic rG;OX rG;RED cX (2)
dt
enzyme [24–26]. The intention of this work was to collect all
dcEt
rEt;pr rEt;OX cX (3)
dt
dcO
kLacO;S cO rO cX (4)
dt
dAV dcX
a b AV (5)
dt dt
In expressions (1)–(5) left sides of equations present the
accumulation of substance in time.


2.3. Kinetic equations


The biomass growth is an autocatalytic reaction
expressed as:

Fig. 1. Metabolism of glucose under anaerobic conditions. rX m cX (6)

A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ 2783


and the total specic biomass growth rate is the sum of the reactor conguration. The empirical correlation [29]
particular growth rate in different metabolisms: for the basic value (kLa)0 is given by:
m mOX mRED mEt (7) " 0:4 #
P
:5
i.e. kLa0 3600 0:026 u0gas (17)
V
m YX=G rG;OX YX=G
OX RED rG;RED YX=Et rEt;OX (8)
P/V in the equations represents the characteristic of the
The glucose uptake rate follows the Michaelis–Menten bioreactor that is determined by its volume and power input.
kinetics: A power input directly depends on the stirrer rpm, which was
cG changed during the experiment; hence an average value was
rG rG;max : (9)
KG cG evaluated based on literature available data [31]. The linear

However, glucose could be utilised through two gas velocity ugas is given by:

metabolic pathways, the oxidative and the reductive. The 4 qVair
ugas : (18)
oxidative metabolic pathway depends on the availability of 3600D2r
the dissolved oxygen in the reaction medium, and therefore
A correlation to take into the effect of temperature and
the quantity of glucose that can be utilised, oxidatively
biomass is given by:
corresponds to the oxidative capacity, the rate of which can
20
be expressed as follows: kLa kLa01 0:07 cX 1:022T : (19)

cO
rO;lim rO;max (10)
KO cO
From these two equations the rate of glucose utilisation in 3. Materials and methods

the oxidative metabolic pathway can be written as:
A pure culture of S. cerevisiae cells (obtained from
rG
rG;OX min (11) microorganisms collection of Faculty of Chemical Engi-
rO;lim=YO=G neering and Technology, Zagreb in the doc. PhD Briski ˇ
and the rest of glucose is utilised by the reductive metabolic laboratory) was stored in the plates containing malt agar at
pathway: the +4 8C. The culture was reactivated by inoculating a S.

rG;RED rG rG;OX (12) cerevisiae cells from the solid substrate into an Erlenmeyer
ask with 100 mL medium, which was incubated at 30 8C on
As the ethanol is formed by reductive pathway of glucose
a rotary shakers for 24 h.
specic rate of ethanol production is given by:
The fermentation and shake asks medium [32] for
rEt;pr YEt=G rG;RED (13) experiment contained: FeSO4 7H2O 0.012 g dm , (NH4)2- 3

3 3
The ethanol consumption follows the Michaelis–Menten SO4 7.5 g dm , CaCl2 0.03 g dm , CuSO4 5H2O 0.001 g
3 3 3 and
kinetics, and it depends on the availability of dissolved dm , KH2PO4 1.5 g dm , ZnSO4 7H2O 0.006 g dm
3 in distilled water. As a carbon
oxygen as well. However, the consumption of ethanol could MgSO4 7H2O 0.691 g dm
source, different initial concentrations of glucose were used
proceed only in the oxidative metabolic pathway. Yeast has
3
an emphasised priority toward glucose, so that ethanol will (cGO = 5, 10, 30 and 50 g dm ). Shake-ask growth
medium contained 10 g dm 3 glucose.
not metabolise as long as measurable quantities of glucose
The batch growth of the baker’s yeast was carried out in a
can be found in the reaction medium. The mathematical
5 dm3 reactor (Drasler, Slovenia) containing 3.5 dm3 med-
description of the ethanol uptake rate is given by the
ium. The bioreactor was equipped with standard control units
following equations:
for pH, temperature, aeration and stirrer speed. Computer
cEt
rEt;up rEt;max (14) aquisition to collect and monitor DO (dissolved oxygen) data
KEt cEt every 10 s was established. All fermentations were carried out
rEt;up at the temperature of 30 8C. Dissolved oxygen concentration
rEt;OX min : (15)
rO;lim rG;OX YO=G=YO=Et was kept about 10% (micro-aerobic conditions) and about

Oxygen consumption is given by the following equa- 40% (aerobic conditions) concentration of saturation by

tion: variation of stirring (200–700 RPM) and air-ow rate (3–
10 dm3 min ). The fermentation medium and the bioreactor
1
rO YO=G rG;OX YO=Et rEt;OX: (16) were sterilised during the 20 min at 120 8C, except glucose,
which was sterilised during the 30 min at 110 8C.
2.4. Oxygen transfer rate A biomass wet weight change was monitored on the
spectrophotometer at the wavelength 660 nm [33] with a
The value of kLa that determines the air ux between the calibration curve. Changes in ethanol concentration and
liquid and gas phase, is mostly dependent on the size of the ADH activity were measured by the BOEHRINGER test
air bubbles blown through the reactor, the stirrer speed an [34]. The test sample for ADH activity was prepared by

2784 A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ


permeabilization the yeast cells with cetyltrimethylammo- Table 1

nium bromide [35]. Glucose concentration was measured by Parameters of the mathematical model

standard colorenzymatic method (PAP). Oxygen concentra- Parameter Literature values [27,28]

tion was calculated according to the correlation: (a) From the literature
1
cO rO,max (h ) 0.0384
DO (20) rEt,max (h )
1 0.07176
cO;S 3
KG (g dm ) 0.612
3
where cO,S represents saturation oxygen concentration KO (g dm ) 9.6 10 5

3
3 KEt (g dm ) 0.1012
(cO,S = 0.00712 g dm , calculated according to the Henry’s
law) that was assumed to be constant during the experiment
Parameter Estimate values Literature values [27,28]
[28].
(b) Estimated in this work
The model parameters were estimated by non-linear
rG,max (h )
1 0.212 0.0124 0.8856
regression analysis using the Nelder–Mead method [36]. YX/Et (gWW g ) 1 0.293 0.0131 0.7173
The numerical values of the parameters were evaluated by YO/Et (g g )
1 2.838 0.5955 0.8904
tting the model to the experimental data with the 1
YEt/G (g g ) 0.049 0.0051 0.4856
1
‘‘Scientist’’ [37] software. The model equations were YO/G (g g ) 0.515 0.0676 0.3858
OX 1 1.521 0.3662 1.6898
solved numerically by the fourth order Runge-Kutta YX=G (gWW g )
RED 1
algorithm, which is also offered in the same software. YX=G (gWW g ) 1.051 0.0691 0.1667

1
The set of optimum parameters was used for the simulation. a (U gWW ) 0.400 0.0532 –
b (h ) 1 0.015 0.0014 –
The calculated data were compared with the experimental
data, recalculated in the optimization routine and fed again
to the integration step until minimal error between
b were taken from the literature for the similar processes
experimental and integrated values was achieved (built-in OX
Scientist). The residual sum of squares was dened as the [27,28]. The initial values for the parameters YX/Et, YX=G,
RED
sum of the squares of the differences between experimental YX=G , YEt/G were approximately determined from the
experimental results. The Michaelis–Menten parameters
( yi) and calculated data ( yi ,calc ).
for the ethanol uptake (rEt,max, KEt) and for the limited
n
oxygen respiration(rO,max, KO) and glucose constant of the
sum Xyi yi
;calc 2 (21)
saturation (KG) were taken from the literature [27,28]
i1
considering the assumption that the dry cell weight is 30% of
For each data set the Pearson product–moment correla-
wet biomasss (Table 1(a)). Parameter a was assessed from
tion coefcient was determined (built-in Scientist). This
the maximal specic activity (U of ADH/g of wet biomass).
correlation between two variables X and Y is dened by the
Parameters like maximum specic glucose consumption rate
expression:
Pn (rG,max), yields of oxygen on glucose and ethanol (YO/Et, YO/
xi xyi y ) and parameter b for the enzyme deactivation were
r qq2 i1 G
Pn Pn (22)
estimated by using the least square method to minimize
i1 xi x2 i1 yi y
difference between experimental and calculated values of
where x and y are the means of X and Y, n is the number of state variables (condence was set at 95%). The list of
point. parameters from the literature along with the evaluated one
The ‘‘Episode’’ algorithm for stiff system of differential (together with the condence intervals) is given in
equations, implemented in the ‘‘Scientist’’ software pack- Table 1(b). Condence intervals of almost all parameters
age, was used for the simulations. It uses variable coefcient are less than 10% except the yield of oxygen on ethanol (YO/
Adams-Moulton and Backward Differentiation Formula Et ) that is responsible for the ethanol uptake rate. A yield of
methods in the Nordsieck form, treating the Jacobian matrix ethanol on glucose is about 10-fold lower than the value in
as full or banded. literature. The possible reason is that wild type yeast, which
was used, is not a good ethanol producer.
The results of comparison of model and experiment are
4. Results and discussion presented in Fig. 2. Biomass growth can be divided into two
phases (Fig. 2b): exponential phase and linear phase. During
4.1. Parameter estimation the exponential phase baker’s yeast growth glucose is
consumed aerobically and anaerobically as well. Anaerobic
Two baker’s yeast cultivations with the initial glucose conditionsare presentonly whenthe respiratory is insufcient
concentration of 30 g dm 3 and under micro-aerobic to metabolise all sugar consumption aerobically. Ethanol is
conditions (dissolved oxygen concentration was kept about accumulated during the yeast growth under anaerobic
10% concentration of saturation) were carried out for the conditions (Fig. 2c) by reducing the acetaldehyde with the
parameter estimation. Initial values of all parameters that are enzyme alcohol dehydrogenase. Therefore, the volume
used to describe baker’s yeast growth except parameter a and activity of ADH increases as well (Fig. 2d). The cause of

A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ 2785




Fig. 2. Glucose (a), biomass wet weight (b), ethanol (c), volume activity (d) and dissolved oxygen (e) (cG0 30 g dm , micro-aerobic conditions, DO = 10%)
3

changes with time. *, experiment 1; &, experiment 2; , mean of the experiment 1 and 2; —, model (interval of one standard deviations of the
experimental results).


slight decrease of volume activity is enzyme deactivation. 10, 30 and 56 g dm , DO = 10%) and aerobic (cG0
3 10
Whereas the model results are placed within a standard and 50 g dm , DO = 40%) conditions were carried out. The
3

deviation of experimental results and correlation coefcient comparison of experiment results and simulation results
of each variable is acceptable, it has been concluded that the done by the package program SCIENTIST using previous
proposed model shows good tting with this cultivation. parameters, together with the corresponding correlation
As it could be seen from the results of simulations volume coefcient, is shown in Figs. 3 and 4. By increasing the
activity follows biomass growth, and with higher biomasss initial glucose concentration under micro-aerobic condi-
concentration higher enzyme activity was expected. A higher tions, a proportionally higher biomass and ethanol
nal biomass concentration could be achieved by higher concentration and enzyme activity were obtained. A good
initial glucose concentration or by increasing oxygen supply. agreement of the result of the experiment and simulation was
Therefore, the model has beenvalidated using different initial achieved. Hence, proposed model is valid for the cultivation,
concentration of glucose and under aerobic conditions. which are done under micro-aerobic conditions with the
initial glucose concentration between 5–50 g dm . 3

4.2. Model validation Under aerobic conditions, a lower ethanol concentration,
but a higher biomass concentration was obtained in
To validate the model, several cultivations with different comparison to the same initial glucose concentration using
initial glucose concentration under micro-aerobic (cG0 5, lower oxygen supply rate. According to the results and

2786 A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ




Fig. 3. Glucose (a), biomass wet weight (b), ethanol (c), volume activity (d) and dissolved oxygen (e) changes with time under micro-aerobic conditions
(DO = 10%). *, cG0 5 g dm ; &, cG0
3 10 g dm ; ~, cG0
3 30 g dm ; !, cG0
3 56 g dm ; —, model.
3




correlation coefcients, the model also describes that very oxidative–reductive metabolism was accomplished, a lower
well. The average values of variables P and qVair that are level of ethanol productivity under aerobic conditions
higher under aerobic conditions (Table 2) directly inuence caused a less specic activity of enzyme under same
the kLa value (Eqs. (17)–(19)) and thereby the oxygen conditions (Fig. 5a). For this reason the lower ratio of
concentration and the oxidative capacity respectively enzyme activity and biomass concentration was assigned
(Eq. (11)). Higher oxidative capacity enables more glucose (Fig. 5b) and value of parameter a for the aerobic baker’s
to be consumed oxidatively. Since the yield of biomass on yeast cultivations is about four fold lower (a = 0.099
glucose is higher on the oxidative path than on the reductive 0.0121 U g ) than the one under micro-aerobic conditions
1

one, higher biomass concentration has been achieved. Lower (Table 1b).
ethanol production is observed because the less glucose is
consumed reductively at the higher oxygen concentration. 4.3. Sensitivity analysis
The parameter a that describes the enzyme activity was
not valid for the cultivations under aerobic conditions. In In order to qualify the sensitivity of model predictions
Fig. 5a it could be seen that ethanol productivity follows the with respect to estimated parameter errors, a sensitivity
increase of enzyme activity while yeast is growing on analysis has been performed. Model parameter values
glucose (Fig. 5a, circles). After all glucose was spent and (Table 1) were taken as nominal values. The solutions of the
yeast has used ethanol, ADH activity has appeared to remain model equations were then computed with relative para-
constant (Fig. 5a, triangles). Although in both cases the meter errors ranging from 50 to +50% of the nominal

A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ 2787




Fig. 4. Glucose (a), biomass wet weight (b), ethanol (c), volume activity (d) and dissolved oxygen (e) changes with time under aerobic conditions (DO = 40%).
&, cG0 10gdm ; !, cG0 50gdm ; —, model.
3 3




values. The nal concentrations of biomass, cX, the maximal 45%) had a more inuence on the nal biomass concentra-
ethanol concentration, cEt and the maximal alcohol tion than the yield of biomass on glucose consumed
dehydrogenase volume activity, AV, were than compared oxidatively, YX=G, (max 9%), because most glucose under
OX

to the nominal concentrations [38]. All simulations were this conditions is spent by reductive path. A variation of
performed for the period of 35 h. The results are shown in yield of biomass on ethanol, YX/Et did not bring large
Fig. 6. changes of a biomass concentration because the concentra-
Since this analysis has been done by taking the variables tion of the produced ethanol is very low with respect to nal
P and G for the micro-aerobic conditions, a change of yield a biomass concentration. A lower yield of oxygen on glucose,
biomass on glucose consumed reductively, YX=G , (max RED YO/G caused a decreasing amount of glucose consumed
reductively and therefore the ethanol concentration as well.
Table 2 The change of the parameter YO/Et has no impact on the nal
Average values of air ow rate, stirrer rate and stirrer power and maximal concentration of the products, because it only
Variables Micro-aerobic Aerobic determines the ethanol consumption rate, i.e. the time that is
conditions conditions necessary to spend the entire ethanol amount. The yield of
G m3 h
1 0.21 0.60 ethanol on glucose, YEt/G, inuences proportionally the
n min 1 400 550 ethanol concentration, but it does not induce the changes in
P kW 0.0135 0.0321 other products from the reasons mentioned before.

2788 A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ




Fig. 5. Dependence the ethanol volume productivity of the specic ADH activity (a) and the specic ADH activity time change (b) under the micro-aerobic
(DO = 10%) and aerobic (DO = 40%) conditions (cG0 50 g dm ). Black symbols, micro-aerobic conditions; white symbols, aerobic conditions.
3




Fig. 6. Sensitivity analysis for the parameters of the model. The changes in the nal biomass wet weight and maximal ethanol concentration and maximal ADH
activity with respect to the deviation of the nominal value of the considered parameter. —, biomass; - - -, ethanol; , ADH volume activity.

A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ 2789


Michaelis–Menten parameters for the glucose consumption
(rG,max, KG) and the oxidative capacity (rO,max, KO) caused
the expected changes of the product concentrations because
they both dene the ratio of glucose spent reductively and
oxidatively. As it can be seen from the model, parameters a
and b determine only the volume activity of the enzyme
alcohol dehydrogenase. Sensitivity analysis for the Michae-
lis–Menten parameters of the ethanol consumption (rEt,max,
KEt) shows that the change of these parameters had not
induced any evident deviations from the nominal values.
Eq. (14) takes place only when the ethanol concentration is
very low, and in all other cases ethanol consumption is
Fig. 7. Parameter a as empirical function of dissolved oxygen in the
limited by oxygen capacity (Eq. (15)). medium.


4.4. Predictions of the enzyme alcohol dehydrogenase associated it also causes inhibition of its production and for
production at different oxygen concentrations that reason simulation at lower oxygen concentration was
not made.
Previous comparisons of experimental results and model Fig. 8b shows dependence of the specic activity (ratio of
simulation have shown that the proposed model describes volume activity and biomass concentration) to the amount of
alcohol dehydogenase production in the growing yeast cells oxygen for two initial glucose concentrations. It can be
in the range of glucose concentration 5–50 g dm 3 quite observed that the specic enzyme activity decreased by
well. It can also be seen that by using estimated parameters increasing the oxygen concentration, while the initial
biomass, glucose and ethanol time change at different glucose concentration has no signicant inuence to the
oxygen concentration in the medium can be predicted. change of the specic enzyme activity.
Parameter a that directly describes enzyme production in the
yeast cell depends on the oxygen supply rate. Therefore, to
4.5. Prediction of the enzyme alcohol dehydrogenase
estimate activity of the produced enzyme, non-linear
production in a different reaction medium volume
empirical dependence of that parameter to the oxygen
concentration was assumed (Fig. 7). This way to describe
To predict enzyme ADH production in the different
this dependencies was chosen to avoid negative value of
reactor volumes, when there is no change in the hydro-
parameter a at higher oxygen concentration.
dynamic properties (density, viscosity) of the reaction
Using the assumed parameter a and previously estimated
medium, and with assumptions that stirring and temperature
other parameters (Table 1), simulations of volume activity
control are not different than in the case above, Eq. (5) can
change with time at different oxygen concentration were
be rewritten as follows:
made (Fig. 8a).
It can be observed that the highest volume activity of dA dcX
a V b A (23)
ADH is achieved at the lowest oxygen concentration and it dt dt
decreased by increasing the oxygen concentration. Under
V represents the volume of reaction medium. Necessary
the anaerobic conditions, i.e. during the baker’s yeast growth
volume of the bioreactor can be estimated by equation:
by minimal oxygen supply, the ethanol is produced to a
higher amount causing the inhibition of the biomass 4
VR V: (24)
production [39]. Since the enzyme production is cell growth 3




Fig. 8. Simulations of the volume (a), and specic (b) ADH activity at the different oxygen concentrations. —, cG0 = 10 g dm ; - - -, cG0 = 50 g dm .
3 3

2790 A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ


DO relativedissolvedoxygenconcentration(%,-)
Dr reactor diameter (m)
KEt ethanol saturation constant (g dm ) 3

3
KG glucose saturation constant (g dm )
kLa total volumetric mass transfer coefcient
(h ) 1

KO oxygen saturation constant (g dm ) 3

n stirrer rate (min ) 1

P stirrer power (kW)
QP volume productivity (g dm 3 h ) 1

qVair airow rate (m3 h ) 1

r Pearson product–moment correlation
Fig. 9. Simulations of the ADH activity in the different volumes of the coefcient (-)
reaction medium.
rEt,max maximal ethanol oxidation rate
Experimental and simulation results indicate that max- (g gWW 1 h ) 1

imal enzyme production has been achieved using initial rEt,OX specic ethanol oxidation rate
glucose concentration cG0 = 50 g dm , and the oxygen
3 (g gWW 1 h ) 1

concentration 10% of the concentration of saturation. rEt,pr specic ethanol production rate rate (h ) 1

Accordingly, simulations of enzyme productions in a rEt,up specic ethanol consumption rate rate
different medium volume have been done using these (g gWW 1 h ) 1

conditions (Fig. 9). rG specic glucose consumption rate
Based on these results it can be predicted that about (g gWW 1 h ) 1

7,000,000 U (enzyme activity units) ADH can be produced rG,max maximal glucose consumption rate
in the reaction volume of 300 dm3 that corresponds to (g gWW 1 h ) 1

bioreactor volume of VR 400 dm3. If the assumed yield rG,OX specic glucose consumption rate by
during the isolation of enzyme from the yeast cell is 5%, oxidative path
about 350,000 U of the enzyme alcohol dehydrogenase (g gWW 1 h ) 1

would be produced in a 400 dm3 bioreactor. rG,RED specic glucose consumption rate by
reductive path (g gWW 1 h ) 1

rO specic oxygen consumption rate
Acknowledgements (g gWW 1 h ) 1

rO,lim oxygen capacity (g gWW 1 h ) 1

This research was supported by the Ministry of Science rO,max maximal oxygen capacity (g gWW 1 h ) 1

and Technology by grant 0125-021/2002. We are grateful to t time (h)
the PhD F. Briski, Faculty of Chemical Engineering and
ˇ ugas linear gas velocity (m h ) 1

Technology, University of Zagreb, Croatia, for the gift of V reactor volume (m3)
Saccharomyces cerevisiae isolates. YEt/G yield of ethanol on glucose (g g ) 1

YO/Et yield of oxygen on ethanol (g g ) 1

YO/G 1
yield of oxygen on glucose (g g )
Appendix A YX/Et yield of biomass on ethanol (gWW g ) 1

YX=G
OX yield of biomass on glucose consumed
List of symbols oxidatively (gWW g ) 1

a model parameter for the enzyme production YX=G
RED yield of biomass on glucose consumed
(U gWW ) 1 reductively (gWW g ) 1

A activity, U
Greek symbols
AS specic activity of the ADH enzyme 1
1 m total specic growth rate (h )
(U mgWW )
mEt specic growth rate on ethanol (h ) 1
AV volume activity of the ADH enzyme 1
3 mOX specic growth rate by oxidative path (h )
(U cm )
mRED specic growth rate by reductive path (h ) 1
b model parameter for the enzyme
deactivation (h ) 1

cEt ethanol concentration (g dm ) 3

cG glucose concentration (g dm ) 3
References
cO 3
oxygen concentration (g dm )
cO,S oxygen concentration of saturation (g dm ) 3
[1] Sonnleitner B, Hahnemann U. Dynamics of the respiratory bottleneck
cX biomass wet weight (gWW dm ) 3
of Saccharomyces cerevisiae. J Biotechnol 1994;38:63–79.

A.V. Presecki, . Vasic-Racki / Process Biochemistry 40 (2005) 2781–2791
ˇ ˇ 2791


[2] Fiechter A, Seghezzi W. Regulation of glucose metabolism in growing [20] Badziong W, Habermann P, Moeller J, Werner A. Process for using the
yeast cells. J Biotechnol 1992;27:27–45. yeast ADH II promoter system for the production of heterologus
[3] Kappeli O. Regulation of carbon metabolism in Saccharomyces cere-
¨ proteins in high yields, US Patent 5,866,371, 1999.
visiae and related yeasts. Adv Microb Physiol 1986;28:181–209. [21] Ratledge C. Yeast physiology: a micro-synopsis. Bioprocess Eng
[4] Lagunas R. Energy metabolism of Saccharomyces cerevisiae. Dis- 1991;6:195–203.
crepancy between ATP balance and known metabolic functions. [22] Fowler MW. Enzyme Technology: Biotechnology for engineers.
Biochim Biophys Acta 1976;440:661–74. Wiley & Sons; 1988. p. 180–5.
[5] Lagunas R, Dominguez C, Busturia A, Saez MJ. Mechanism of [23] Luedeking R, Piret EL. A kinetic study of the lactic acid fermenta-
appearance of the Pasteur effect in Saccharomyces cerevisiae: inacti- tion. Batch process at controlled pH. Biotechnol Bioeng 2000;67:
vation of sugar transport systems. J Bacteriol 1982;152:19–25. 633–44.
[6] Broker M, Bauml O, Gottig A, Ochs J, Bodenbenner M, Amann E.
¨ ¨ ¨ [24] Kekos D, Marcis BJ. Kinetic studies of amylase and biomass produc-
Expression of the human blood coagulation protein Factor XIIIa in tion by Calvatia gigantean. Biotechnol Bioeng 1987;30:681–4.
Saccharomyces cerevisiae: dependence of the expression levels from [25] Santos M, Teixeira J, Rodrigues A. Production of dextransucrose,
host-vector systems and medium conditions. Appl Microbiol Biotech- dextran and fructose from sucrose using Leuconostoc mesenteroides.
nol 1991;34:756–64. Biochem Eng J 2000;4:177–88.
[7] Carlsen M, Jochumsen VK, Emborg C, Nielsen J. Modeling the growth [26] Miron J, Gonzalez MP, Pastrana L, Murado MA. Diauxic production
and proteinase A production in continuous cultures of recombinant of glucose oxidase by Aspergillus niger in submerged culture. Enzyme
Saccharomyces cerevisiae. Biotechnol Bioeng 1997;55:447–54. Microb Tech 2002;31:615–20.
[8] Coppella S, Dhurjati P. A mathematical description of recombinant [27] Sonnleitner B, Kappeli O. Growth of Saccharomyces cerevisiae is
¨
yeast. Biotechnol Bioeng 1990;35:356–74. controlled by its limited respiratory capacity: formulation and ver-
[9] Tttrup HV, Carlsen S. A process for the production of human ication of a hypothesis. Biotechnol Bioeng 1986;28:927–37.
proinsulin in Saccharomyces cerevisiae. Biotechnol Bioeng 1990; [28] Pertev C, Turker M, Berber R. Dynamic modelling sensitivity analysis
¨
35:339–48. and parameter estimation of industrial yeast fermenters. Comput
[10] Vasavada A. Improving productivity of heterologous protein in recom- Chem Eng 1997;21:5739–44.
binant Saccharomyces cerevisiae. Adv Applied Microbiol 1995;41: [29] Kristiansen B. Integrated design of fermentation plant: the production
25–54. of baker’s yeast. NY: VCH, 1989. p. 27–52.
[11] Leskovac V, Trivic S, Pericin D. The three zinc-containing alcohol
ˇ [30] Pham HTB, Larsson G, Enfors SO. Growth and energy metabolism in
dehydrogenase from baker’s yeast Saccharomyces cerevisiae. FEMS aerobic fed-batch cultures of Saccharomyces cerevisiae: simulation
Yeast Res 2002;2:481–94. and model verication. Biotechnol Bioeng 1998;60:474–82.
[12] DickinsonFM,DackS.TheactivityofyeastADHIandADHIIwithlong [31] Plavljanic C. Effect of mixing on oxygen transfer rate in fermentation
ˇ

chain alcohols and diols. Chem Biol Interact 2001;130–132:417–23. processes, Master Thesis: University of Zagreb, 1994.
[13] Orlich B, Berger H, Lade M, Schomacker R. Stability and activity of
¨ [32] Thatipamala R, Rohani S, Hill GA. Effects of high product and
alcohol dehydrogenases in w/o-microemulsions: enantioselective substrate inhibitions on the kinetics and biomass and product yields
reduction including cofactor regeneration. Biotechnol Bioeng 2000; during ethanol batch fermentation. Biotechnol Bioeng 1992;40:289–
70:638–46. 97.
[14] Molinari F, Bertolini C, Aragozzini F. Microbial biotransformations in [33] Thatipamala R, Rohani S, Hill GA. Spectrophotometric method for
biphasic systems: formation of R-2-alkanols by methylketone reduc- high biomass concentration measurements. Biotechnol Bioeng
tion. Biocatal Biotransform 1998;16:87–94. 1991;38:1007–11.
[15] Gervais TR, Carta G, Gainer JL. Effect of aeration during cell growth [34] Biochemica information, Boehringer Manheim, GmbH, W Germany,
on ketone reactions by immobilized yeast. Biotechnol Prog 2000; 1973.
16:208–12. [35] Bhat N, Bhat SG, Gowda LR, Nana N. Detergent permeabilized yeast
[16] Grifn DR, Gainer JL, Carta G. Asymmetric ketone reduction with cells as the source of intracellular enzymes for estimation of biomo-
immobilized yeast in hexane: biocatalyst deactivation and regenera- lecules. Enzyme Microb Tech 1993;15:796–800.
tion. Biotechnol Prog 2001;17:304–10. [36] Nelder JA, Mead R. A simplex method for function minimisation.
[17] Dickinson FM, Monger GP. A study of the kinetics and mechanism of Compu J 1965;7:308–31.
1
yeast alcohol dehydrogenase with a variety of substrates. Biochem J [37] SCIENTIST handbook, MicroMath , Salt Lake City, 1986–1995.
1973;F131(2):261–70. [38] Bernard O, Bastin G, Stentelaire C, Lesage-Meessen L, Asther M.
[18] Methods of Biochemical Analysis and Food Analysis, Boehringer Mass balance modelling of vanillin production from vanillic acid by
Manheim, GmbH, W Germany, 1989. cultures of the fungus Pycnoporus cinnabarinus in bioreactors. Bio-
[19] Maugard T, Lamare S, Legoy MD. Gas phase biotransformation technol Bioeng 1999;65:558–71.
reaction catalyzed by baker’s yeast. Biotechnol Bioeng 2001;73: [39] Sonnleitner B. Dynamics of yeast metabolism and regulation. Biopro-
164–8. cess Eng 1991;6:187–93.


Journal of Food Engineering 69 (2005) 115–123

www.elsevier.com/locate/jfoodeng




Eect of fermentation conditions and immobilization
supports on the wine making

Y. Kourkoutas a,* , M. Kanellaki , A.A. Koutinas , C. Tzia
a a b


a Department of Chemistry, Food Biotechnology Group, Section of Analytical Environmental and Applied Chemistry, University of Patras,
GR-26500 Patras, Greece
b Laboratory of Food Chemistry and Technology, School of Chemical Engineering, National Technical University of Athens,
5 Iroon Polytechniou Street, Zografou, 15780 Athens, Greece

Received 19 March 2004; accepted 5 August 2004




Abstract


Batch and continuous fermentations at two dierent temperatures using immobilized cells on apple, quince and pear pieces were
carried out separately and the eect of fermentation temperature, type of fermentation and immobilization support on kinetic
parameters and on quality characteristics of the produced wines was investigated. Duncans multiple range test showed that higher
temperature (30 C) resulted in higher wine productivity. Statistically higher wine productivity was also observed when quince-
immobilized biocatalyst was used in continuous fermentation compared to batch system at the same temperature. In contrast, wine
productivity was statistically higher in batch fermentations compared to continuous process when apple and pear-supported biocat-
alysts were employed in wine making at 15 C. Although volatile acidity in wines produced by immobilized cells on pear pieces was a
little increased and statistically higher compared to wines produced by immobilized cells on apple and quince pieces, it still remained
in levels usually found in dry wines. No vinegar odor was observed. Ethyl acetate concentration was higher in wines produced at
lower temperature (15 C) contributing to the fruity character of the product. The content of methanol, amyl and higher (propanol-1
and isobutanol) alcohols in all wines was found in very low levels, leading to improved quality products. Low-temperature (15 C)
wine-fermentation resulted in improved quality and the produced wines had a distinctive aromatic prole. Apple appears to have an
advantage over quince and pear, as it is widely cultivated, it is considered more compatible with wine taste, while continuous process
using immobilized cells on apple pieces resulted in high wine productivity.
2004 Elsevier Ltd. All rights reserved.


Keywords: Immobilized cells; Apple; Quince; Pear; Wine fermentation




1. Introduction suitable for the wine industry have to meet additional
prerequisites such as food-grade purity, low cost,
Cell immobilization in alcoholic fermentation is a abundance, non-degradable nature and suitability for
rapidly expanding research area because of its attractive low-temperature fermentation. Although many immobi-
technical and economic advantages compared to the lization supports have been proposed for use in wine
conventional free cell system (Margaritis & Merchant, making (Bakoyianis, Kanellaki, Kaliafas, & Koutinas,
1984; Stewart & Russell, 1986). Immobilization supports 1992; Bardi & Koutinas, 1994; Fumi, Trioli, &
Colagrande, 1987; Shimobayashi & Tominaga, 1986)
industrial application of the technology is still uncertain.

* Many support systems such as inorganic materials or
Corresponding author. Tel.: +30 (2)610 997104; fax: +30 (2)610
997105. alginates were considered inappropriate for wine ma-
E-mail address: jkou@chemistry.upatras.gr (Y. Kourkoutas). king and were therefore eventually abandoned.

0260-8774/$ - see front matter 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2004.08.003

116 Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123


Continuous processes are favoured in many elds of 2.2. Cell immobilization and grape must fermentation
the fermentation technology for a series of obvious rea-
sons, including the economic advantages of having unin- Cell immobilization on apple (Kourkoutas et al.,
terrupted operation for longer periods (Jackson, 1994). 2001), quince (Kourkoutas et al., 2003) and pear
Continuous processes can generally achieve substantial (Mallios et al., 2004) pieces was carried out as described
improvement in eciency of the process and subse- earlier. In brief, pieces of apple, quince and pear were
quently higher productivity and lower operating cost. introduced separately in synthetic medium consisting
However, references for continuous wine making are of glucose and the appropriate amount of biomass and
scarce in literature (Iconomou, Kanellaki, Voliotis, was allowed to ferment. When the fermentation was
Agelopoulos, & Koutinas, 1996; Loukatos et al., 2000; completed, the liquid was decanted; the immobilized
Maicas, Pardo, & Ferrer, 2001). biocatalysts were washed twice with grape must and
The development of such a continuous wine produc- employed in fermentations.
tion technology using immobilized cells has not been Grape must preparation and repeated batch and con-
achieved due to several reasons including the lack of a tinuous fermentations using immobilized cells on apple
suitable low-cost support material, undetermined taste (Kourkoutas et al., 2001; Kourkoutas et al., 2002)
and aroma quality for the produced wine and lower via- quince (Kourkoutas et al., 2002; Kourkoutas et al.,
bility of the immobilized system. 2003) and pear (Mallios et al., 2004) pieces were carried
The use of fruits in developing such a biocatalyst was out as described in previous studies.
an obvious alternative that has been used in several food-
related industries. Apple (Kourkoutas, Komaitis, 2.3. Analytical assays
Koutinas, & Kanellaki, 2001; Kourkoutas, Koutinas,
Kanellaki, Banat, & Marchant, 2002) quince (Kourkou- Alcohol was distilled and measured using a Gay-Lus-
tas et al., 2002; Kourkoutas et al., 2003) and pear sac alcoholmeter and wine and ethanol productivity
(Mallios et al., 2004) pieces have been already used suc- were calculated as the grams of wine/ethanol per liter to-
cessfully for immobilization of Saccharomyces cerevisiae tal volume produced per day (g/Ld1).
yeast in both batch and continuous wine making and the Total acidity was estimated by titration of samples
nal product had improved quality. In addition, they are with 0.1M NaOH solution and volatile acidity by titra-
cheap, of food grade purity, abundant in nature and tion with 0.1M NaOH of distillates obtained by steam
would possibly lead to a product with improved taste distillation of wine samples (Zoeklein, Fugelsang,
and aroma due to potential transfer of some of their con- Gump, & Nury, 1990).
stituents into the wine. It is obvious, therefore, that a
comparative study is necessary in order to investigate 2.4. Volatile by-products determination
the optimum immobilization support and fermentation
conditions. Acetaldehyde, ethyl acetate, propanol-1, isobutanol
In this paper we studied the eect of the type of fer- and amyl alcohols (2-methyl-butanol and 3-methyl-but-
mentation (batch or continuous system), fermentation anol) were determined by gas chromatography using a
temperature and immobilization support (apple, quince stainless steel column, packed with Escarto-5905 con-
and pear pieces) on daily wine, daily ethanol productiv- sisting of Squalene 5%, Carbowax-300 90% and di-
ity and on quality characteristics of the products, in ethyl-hexyl sebacate 5% (v/v) (Cabezudo et al., 1978).
order to conclude for the optimum immobilization sup- Nitrogen was used as carrier gas at 20mL/min. Injec-
port and fermentation conditions facilitating scale-up of tion port and FID detector temperatures were 210
the process. and 220 C, respectively. The column temperature was
70 C. In all cases, the internal standard was butanol-
1 at a concentration of 0.5% (v/v). Samples of 4lL of
2. Materials and methods wine were directly injected into the column and the
concentrations of the above volatile compounds were
2.1. Yeast strain and must preparation determined using standard curves. Methanol was also
determined by gas chromatography using Porapac S
Saccharomyces cerevisiae AXAZ-1 is an alcohol tole- column. Nitrogen was used as carrier gas at 40mL/
rant and psychrophilic strain isolated from the Greek min. The column temperature was programmed at
agricultural area by Argiriou et al. (1992). It was grown 120–170 C at a rate 10 C/min. The temperatures of
on a medium consisting of glucose 4%, yeast extract the injector and FID detector were 210 and 220 C,
0.4%, (NH4)2S04 0.1%, KH2P04 0.1% and MgSO4 respectively. For the methanol determination, 2lL
0.5%. Pressed wet weight cells (15–20g) were prepared samples were injected directly into the column and
accordingly to Argiriou et al. (1992) and employed the concentration of methanol was determined using
directly in the fermentation process. standard curves. Butanol-1 was used as internal stand-

Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123 117


ard at a concentration of 0.5% (v/v). All analyses were
conducted in triplicate and the mean values are
anol )
presented (max deviation for all values was about
±5%). 6
Meth (mg/L 40 36 Tr 43 44


2.5. Experimental design and statistical analysis ols

alcoh
In the experiments conducted the eect of type of fer- )

mentation, immobilization support and fermentation 90
Amyl (mg/L 246 195 167 141 166
temperature on daily wine and daily ethanol producti-
vity, total and volatile acidity and concentrations of
acetaldehyde, ethyl acetate, propanol-1, isobutanol, tanol
L)
amyl alcohols and methanol were examined. The exper-
iments were designed and analyzed statistically by Isobu (mg/ 24 21 45 41 27 46

ANOVA. At least ve repeated batch fermentations
were carried out and at least four samples were collected
in continuous fermentation for each experiment and the L)
opanol-1
presented results are the mean values. Duncans multiple Pr (mg/ 15 26 53 29 24 49
range test was used to determine signicant dierences
among results (coecients and the ANOVA tables were e

computed using Statistica). The factors and the interac-
acetat
tions between the factors are symbolized by the proper yl L)
combination of letters. 87 65 79 75 85
Eth (mg/ 109



3. Results and discussion ehyde
)

In this study the eect of the type of fermentation,
Acetald (mg/L 35 17 78 60 75 76
fermentation temperature and immobilization support
on kinetic parameters and quality characteristics of the y L)
produced wines were studied. The strategy adopted
was to use immobilized cells on apple, quince and pear acidit acid/
C tic
pieces separately for batch and continuous wine making 51 tile
ace
at room and mild temperature and to examine potential ta Vola (g 0.29 0.12 0.96 0.41 0.17 0.98
dierences in productivity and quality characteristics of g

the products. As it is well known that productivity and
makin /L)
product quality are the main factors determining indus-
ine acid
trial application of immobilized cells, parameters such as ity
w
daily wine and daily ethanol productivity, total and on acid aric port.

volatile acidity and volatile by-products (acetaldehyde, ort tart sup

ethyl acetate, propanol-1, isobutanol, amyl alcohols Total (g 5.0 5.1 4.8 5.0 4.9 4.8
supp
and methanol) of the produced wines were studied be-
cause they seem to play the most important role in oper- tion vity
ating cost and wine quality. wine
ucti
ly ) immobilization
obiliza
91 60
Dai prod (g/L 200 188 163 260
3.1. Eect of type of fermentation and immobilization imm IS:,

support on wine making at 15 C and
tion
ion S) entation
Batch and continuous fermentations using immobi- (I
ferm
lized cells on apple, quince and pear pieces separately entat obiliza
port le nce le nce of
were examined in order to study the eect of the type ferm Imm sup App Qui Pear App Qui Pear pe
ty
of fermentation and immobilization support on daily of

wine productivity, total and volatile acidity and volatile TF:
type
by-products of the produced wines. The experimental 1 of tation
of in. in. in. aces,
conditions and the results are summarized in Table 1. tr

Table 2 shows the results of the statistical analysis Table Eect Type fermen (TF) Batch Batch Batch Cont Cont Cont Tr:

118 Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123


related to the signicance of the studied factors and of
**
wine- their interactions.
F 0.4033 9.0313 0.0338 The type of fermentation had very signicant eect
during propanol-1,
on volatile acidity, acetaldehyde, isobutanol and amyl
PR:
alcohols content (p < 0.01), while total acidity and con-
E 16.5308
197.1920
M MS 4416.0078 centrations of ethyl acetate, propanol-1 and methanol
by-products acetate,

** ** ** were not signicantly aected. On the other hand,
ethyl
immobilization support aected all parameters studied
volatile
EA:
of 20.5975 30.2531
F 121.1542 except total acidity. Total acidity was not inuenced nei-
ther by the type of fermentation, nor by the immobiliza-
tion support. On the contrary, acetaldehyde, isobutanol
M cetaldehyde,a
6006.5498 8822.2578 and amyl alcohols concentrations were aected by both
concentrations A MS 35330.4180
AC: factors (p < 0.01) and in addition there was a signicant
and
** ** ** interaction between them (p < 0.01). Strong interaction
acidity,
acidity 7.1740 between the type of fermentation and the immobili-
F 22.6449 57.5075
zation support was also observed in daily wine produc-
volatile
volatile tivity (p < 0.01), ethyl acetate (p < 0.01) and propanol-1
VA: (p < 0.05) content.
BI 472.3920 149.6558
MS 1199.6558
acidity, Type of fermentation and immobilization support
** * acidity,
total seemed to aect signicantly daily wine productivity.

1.1390 4.3368 total The statistically highest wine productivity was recorded
F 51.7491
uctivity, TA: in continuous fermentation using immobilized cells on
quince pieces (Table 1). In batch fermentations, immobi-
prod
lized cells on apple pieces resulted in the highest produc-
wine RP 55.7780
212.3808
MS 2534.2422
productivity, tivity followed by immobilized cells on quince pieces
daily * **
with almost equal value and immobilized cells on pear
on wine
F 2.2576 4.8004 8.8286 pieces. Batch fermentation using immobilized cells on
daily apple and on pear pieces resulted in statistically higher
support
wine productivities compared to those reported in con-
DWP:
A 387.2000 823.3307
E MS 1514.2230 tinuous system using the same immobilized biocatalysts.
Fermentations using immobilized cells on pear pieces re-
** ** ** support,
immobilization sulted in lower wine productivities both in batch and
d
an F 26.2423 14.7617 11.4279 continuous fermentation compared to the other two
supports, but still higher than those of traditional fer-

immobilization mentation (Jackson, 1994). The lower wine productivi-
C IS: ties might be attributed to possible lower amount of
fermentation A MS 5222.9121 2937.9558 2274.4480
of yeast cells immobilized on pear pieces, since apple and
** **

type quince are fruits with macroscopically higher diameter
2.4748
12.2810 pores.
F 806.7855 fermentation,
of When apple-supported biocatalyst was used, statisti-
rameters:
pa type
A cally higher volatile acidities were observed in wines pro-
the V MS 0.0275 1.8084 0.0055
TF: duced by continuous fermentation compared to those
of
produced by batch fermentations. Wines produced using
ratio,
F 0.0830 0.5227 0.0944 immobilized cells on quince pieces had very low volatile
acidity, indicating an improved quality product, while
signicance

the TA MS 0.0146 0.0918 0.0166 MS/error:F methanol. wines produced by immobilized cells on pear pieces gave
to statistically higher values, but still in levels usually
** ** ME:
square, present in dry wines (up to 1g/L) (Ribereau-Cayon,

related
0.0308
24.3319 Glories, Maujean, & Dubourdieu, 2000). However,
F 154.9417
mean alcohols, there was no indication of vinegar odor in the nal
analysis
MS:
amyl 0.01.
0.05. product.
<
11.5520 < p
p Acetaldehyde is the major aldehyde found in wines. It
9138.6328 AM:
statistical C DWP MS 58193.4023 for
freedom, for is more likely that acetaldehyde contributes positively to
the 15 of
the characteristic odor of the wines (Etievant, 1991).

of at Df 1 2 2
2
degree isobutanol, However, above threshold values, it usually is consid-
Signicant
Signicant
Table Results making Eect TF IS TF-IS Df: IB: * ** ered as o odor (Jackson, 1994). Duncans multiple

Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123 119


range test showed that wines produced by batch fermen-
tation using apple and quince-supported biocatalysts
L)
ethanol contained lower amounts of acetaldehyde compared to
6
M (mg/ 39 32 47 43 44 the ones produced by continuous process using the same
immobilized biocatalysts and by batch fermentation
)
using pear-supported biocatalyst, as well (Table 1). In
74 90
Amyl alcohols (mg/L 123 113 141 166 most cases, acetaldehyde content was in levels usually
found in dry wines (up to 75ppm) (Koutinas & Pefanis,
1994).
tanol
L) In batch fermentations, the statistically highest ethyl

Isobu (mg/ 19 18 29 41 27 46 acetate concentration was reported in wines produced
by apple-supported biocatalyst, followed by wines pro-
duced by quince and pear-supported biocatalysts. In
anol-1 L) continuous fermentations, no statistical dierences in
ethyl acetate content were found. However, in all wines
Prop (mg/ 22 25 18 29 24 49
ethyl acetate concentration was in levels usually found
in dry wines (up to 150ppm) (Jackson, 1994) and a fru-

acetate ity aroma was predominant.
)
Continuous fermentation resulted in statistically
Ethyl (mg/L 59 69 44 79 75 85 lower concentrations of amyl alcohols when immobi-
lized cells on apple and quince pieces were used, contrib-
yde uting to an improved quality of the produced wines. The
type of fermentation did not aect amyl alcohols con-
aldeh L)
tent in wines produced by pear-immobilized biocatalyst.
Acet (mg/ 27 42 64 60 75 76 Average amyl alcohols content between batch and con-
tinuous fermentation was higher when immobilized cells
/L) on apple pieces were used, followed by immobilized cells
acidity acid on pear pieces, while quince-immobilized biocatalyst re-
ic
ported the lowest value. In all cases, higher (propanol-1
acet
g and isobutanol) and amyl alcohols ranged in low levels
Volatile (g 0.13 0.17 0.78 0.41 0.17 0.98
(Etievant, 1991), indicating an improved quality

makin
L) product.

wine acid/
s

acidity 3.2. Eect of fermentation temperature and
tal
ntinuou tartaric immobilization support on continuous wine making
co To (g 5.2 4.3 4.6 5.0 4.9 4.8

on
Productivity is considered as the most important fac-
ort nol
tor aecting the use of immobilized cells because it
etha
supp uctivity ort. determines the cost eectiveness of the process. Since
5 the highest daily wine productivity was reported when
tion Daily prod (g/L) 21 66 35 14 23 supp
continuous system was used (Tables 1 and 2), it was
y thought to study continuous wine making using immo-
obiliza
wine bilized cells at higher temperature, as it is well known
imm uctivit
ly obilization that high temperatures usually result in higher fermenta-
60
and Dai prod (g/L) 258 750 403 163 260 imm tion rates.
IS: In order to study the eect of fermentation tempera-
rature ion re, ture and immobilization support on kinetic parameters
(IS) and quality characteristics of the produced wines, con-
tempe bilizat
ort peratu
nce nce tinuous wine making using immobilized cells on apple,
tem
Immo supp Apple Qui Pear Apple Qui Pear quince and pear pieces separately was carried out at
entation ion 30 C and compared to results obtained at 15 C. Table
ion 3 presents the experimental conditions and the results,
ferm re
entat
3 fo C) while Table 4 shows the results of the statistical analysis
t entat (
ble peratu ferm related to the signicance of the studied factors and of
Ta Eec Ferm tem (FT) 30 30 30 15 15 15 FT: their interactions.

120 Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123


*
IB: Fermentation temperature and immobilization sup-

F 1.0567 1.1611 4.1823 port had a signicant eect on daily wine and ethanol
productivity (p < 0.01), volatile acidity and concentra-
propanol-1,
making tions of acetaldehyde, higher (propanol-1 and isobuta-
E
488.0720 536.3327 PR:
M MS 1931.8250
wine nol) and amyl alcohols (p < 0.01) as indicated in Table

** ** ** acetate, 4. A signicant interaction between these two factors
aecting all parameters studied except total acidity and
ethyl
continuous F 59.9592 92.9853 10.2819
EA: acetaldehyde content was also observed (p < 0.01). Total
during acidity was aected only by immobilization support
M (p < 0.05), while ethyl acetate content was not aected
A MS 6174.0981 9574.8574 1058.7500

acetaldehyde, by this factor. In contrast, ethyl acetate concentration
** ** **
by-products
AC: was signicantly inuenced by fermentation temperature
5.8247
38.7752
volatile F 120.0033 (p < 0.01). Methanol concentration was independent by
acidity,
of both fermentation temperature and immobilization sup-
volatile port, but there was a strong interaction between them
89.0923
BI 593.0923
MS 1835.5280
VA: (p < 0.05).
concentrations Higher fermentation temperature resulted in higher
** ** **


and acidity,
9.2912 daily wine and daily ethanol productivities. Duncans
F 40.4649 24.1766 tal
to
multiple range test clearly showed that in both fermenta-
acidity
TA:
tion temperatures studied, immobilized cells on quince
volatile RP 259.0019 673.9481
MS 1128.0020 pieces gave statistically higher productivities compared
to the other two immobilized supports (Table 3). Immo-
** ** productivity,
acidity,
bilized cells on pear pieces resulted in higher daily wine
total 1.2201 7.5400
F 34.9920 ethanol and ethanol productivity compared to immobilized cells
daily on apple pieces at 30 C, while at 15 C they were lower.

A Average daily wine productivity at the two fermentation
productivity, 124.4692 769.2077 DEP:
E MS 3569.7920
e temperatures studied, was higher when immobilized cells
win ** **
on quince pieces were used, followed by immobilized
daily
2.2485
F 30.6771 10.0786 productivity,
on cells on pear and then apple pieces with almost similar
wine value. The same trend was observed for daily ethanol
support daily productivity.
C 364.1000
A MS 4967.5518 1632.0308 Duncans multiple range test showed that wines pro-
DWP:
** ** ** duced by apple and pear immobilized biocatalysts at

immobilization 15 C contained statistically higher amounts of ethyl ace-
86.1586 24.1116 support,
F 708.3039
and tate, compared to those produced at 30 C by the same

A biocatalysts (Table 4), contributing to the fruity charac-
V MS 0.1763 1.4497 0.0494
ter and ascertaining the distinctive aroma of the prod-
immobilization
temperature *
IS: ucts observed mainly in wines produced in lower
F 2.6283 4.9730 2.9777 temperatures.

fermentation
TA MS 0.3976 0.7523 0.4505 temperature,
3.3. Eect of fermentation temperature, type of
fermentation and immobilization support on wine making
** ** **


parameters:
fermentation
the 768.0461 258.3960
F 1808.1758
of FT: Finally, in a smaller scale experiment, batch and con-
tinuous fermentations using immobilized cells on quince
ratio,
and pear pieces separately were carried out at 30 C and
748.2250
signicance DEP MS 5235.8481 2223.9941
15 C, in order to study the eect of fermentation tem-
the
** ** **
to 17 16 16 MS/error:F perature, type of fermentation and immobilization
10 10 10 methanol.
support on daily wine productivity and quality charac-
related square,
F 1.31 4.99 1.76 ME:
teristics of the produced wines. The strategy adopted
mean
analysis was to use the immobilized biocatalysts that previously
MS: alcohols,
0.05. 0.01. reported in average the higher (quince-immobilized bio-
< <
93015.0469 amyl p p
DWP MS 688947.1875 263575.0625
statistical catalyst) and lower (pear-immobilized biocatalyst) wine
freedom, for for
AM:
the Df 2 2 of productivity during wine-making at 15 C, as productiv-
of
4 ity is probably the most important factor for industrial
degree
Signicant Signicant
Table Results Eect FT1 IS FT–IS Df: isobutanol, application of immobilized cells and it is also believed
* **

Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123 121


that low-temperature fermentation leads to higher qual-
ity products.
The experimental conditions and the corresponding L)
ethanol
results are summarized in Table 5, while the statisti- 6
M (mg/ 24 15 32 47 36 Tr 44
cal analysis related to the signicance of the studied
factors and of their interactions are presented in ols

Table 6.
alcoh
All three factors aected signicantly daily wine pro- yl L)

ductivity and acetaldehyde concentration (p < 0.01). 74 90
Am (mg/ 219 173 113 195 167 166
Strong interaction between fermentation temperature
and immobilization support (p < 0.01), fermentation
tanol )
temperature and type of fermentation (p < 0.01) and
between fermentation temperature, type of fermenta- Isobu (mg/L 18 39 18 29 21 45 27 46
tion and immobilization support (p < 0.01) aecting
daily wine productivity was observed. In addition,
)
strong interaction between all three factors aecting sig- anol-1

nicantly ethyl acetate concentration was also observed
Prop (mg/L 18 19 25 18 26 53 24 49
(p < 0.01). Total acidity was signicantly aected by
)
the type of fermentation (p < 0.01), whereas volatile
acidity, ethyl acetate and higher alcohols (propanol-1
Ethyl acetate (mg/L 56 43 69 44 87 65 75 85
and isobutanol) concentrations were aected by both
e
fermentation temperature and immobilization support
(p < 0.01). Amyl alcohols content was only temperature
)
dependent (p < 0.05). However, a strong interaction aldehyd

between fermentation temperature and type of fermen- Acet (mg/L 12 78 42 64 17 78 75 76
tation and between type of fermentation and immobili-
zation support aecting both amyl alcohols and /L)
ing
acetaldehyde concentrations was observed. Both type acidity acid
of fermentation (p < 0.05) and immobilization support mak ic
ort.
(p < 0.01) aected methanol concentration and there acet
wine
was a strong interaction between fermentation temper- Volatile (g 0.10 0.79 0.17 0.78 0.12 0.96 0.17 0.98 supp
on
ature and immobilization support (p < 0.01). ion
L)
Continuous fermentation resulted in much higher
y
daily wine productivity than repeated batch fermenta- support acid/

tions when quince-supported biocatalyst was used in acidit immobilizat
both temperatures studied. The same results were ob- tartaric
IS:
served at fermentations carried out at 30 C using Total (g 5.8 5.8 4.3 4.6 5.1 4.8 4.9 4.8

pear-supported biocatalyst. In contrast, at 15 C daily immobilization y ation,
wine productivity was slightly lower when immobilized
and wine
cells on pear pieces were employed in continuous pro- uctivit
ly ferment
cess compared to batch system (Table 5). Continuous 91 60
ation Dai prod (g/L) 667 175 750 403 188 260 of
fermentation at 30 C using immobilized cells on quince
type
pieces resulted in the statistically highest daily wine pro-
ferment
ductivity. Duncans multiple range test showed that of (IS) TF:

quince-supported biocatalyst used in batch system at re,
obilization ort
nce nce nce nce
30 C resulted in even higher daily wine productivity type
Imm supp Qui Pear Qui Pear Qui Pear Qui Pear
compared to pear-supported biocatalyst employed in peratu
ure,
continuous process at 30 C. tem

Methanol concentration was statistically higher in tation tion
temperat of n. n. n. n.
wines produced by immobilized cells on quince pieces
enta
at 15 C in comparison with wines produced by immobi-
tation Type fermen (TF) Batch Batch Conti Conti Batch Batch Conti Conti
ferm
lized cells of pear pieces at the same temperature, prob-
ably due to possible higher pectin content in quince. FT:
fermen
However, in all cases, methanol content was in extre- 5 of C)
entation rature (
mely low levels (Jackson, 1994), indicating a high qual- traces,

ity product. Table Eect Ferm tempe (FT) 30 30 30 30 15 15 15 15 Tr:

122 Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123


3.4. Technological consideration
* ** **
making

F 1.5617
wine 4.7227 7.5526 1.1931 9.9075 0.8056 1.1970 propanol-1, Supports used in wine making should be cheap, of
PR: food grade purity, abundant in nature, stable and easy
during
to use at industrial scale. Fruit pieces appear to meet
E
660.1932 504.3571 340.5712 506.0389
M MS 1996.4767 3192.7981 4188.3257 acetate,
all the above prerequisites and could be used in low-tem-
ethyl perature improved quality wine making. The results
* ** **
by-products
EA:
1.4445 1.4018 8.6296 2.7004 0.3177 showed that these supports produced wines with accep-
F 85.5310 31.9403
volatile table characteristics. In particular, the overall aroma of
of
the wines produced at lower temperature (15 C) was im-
acetaldehyde,
M 253.1409
1150.9299 1116.9142 6875.6104 2151.5032 proved and could be attributed to enhanced volatiles
A MS 68146.3047 25448.2090 AC:
prole due to increased solubility (Jackson, 1994). The
concentrations ** ** *
produced wines were characterized as novel, special type
and acidity,
0.1241 4.9351 1.7886 3.4055 0.3392
F 19.7779 92.6524 wines, with a pleasant, soft, distinctive aroma and an
acidity volatile improved taste, which may lead to new and dierent
VA:
5.5119 wine products. Although possible transfer of fruit con-
volatile 79.4584 15.0678
BI 878.6143 219.2379 151.2883
MS 4116.0000 ity,
stituents to the wines may have occurred, a pretreatment
acid
acidity, ** ** ** of the fruit pieces before their use as supports was not
total
total necessary because (i) most of the fruit constituents (glu-
0.0379 3.6042 2.6066 0.6306
55.0620 84.2148
F 132.0385 TA:
cose, fructose, organic acids, phenolic compounds, etc)
are normal constituents of the wine and (ii) fruit consti-
productivity, 1.0780
RP 74.2118 17.9535 tuents may contribute to the avor and to a distinctive
102.6150 productivity,
MS 3759.2322 1567.6543 2397.6575
wine aroma of the product and therefore their extraction in
wine
** ** **
daily the wine might be useful. The aforementioned fruits
daily
on 2.3224 0.2179 3.1922 2.1107 9.3272
F 49.4538 11.5617 have the necessary hardness required for production of
DWP: an immobilized biocatalyst with high operational stabi-
support

on lity, suitable for industrial application. Even though
A 31.9174
340.2261 467.6591 309.2213
E MS 7244.9692 1693.7883 1366.4308 support, quince is the most appropriate fruit, its cultivation is

** ** ** * ** limited. Therefore, the choice is restricted between apple
immobilizati
and pear pieces. Apple seems to have an advantage due
8.1355 5.2687 2.3110 0.7621
and F 16.3934 71.4274 35.0796
immobilization to higher wine productivity, which is also strengthened
IS: by the fact that it is considered more compatible with
wine taste. Finally, continuous process is favored due
fermentation C 514.0662 169.5244
1809.6504 3646.5386 1171.9685 7803.0742
A MS 15888.2393
of to ecient monitoring of the process. The experience
fermentation,
type ** ** **
of that has been acquired enables proposition of a rela-
1.8440 0.0019 1.1186 0.2089 tively small (5000–10,000L) ‘‘Multi Stage Fixed Bed
14.4416 12.0171 type
F 859.8963

TF: Tower’’ (MFBT) bioreactor (Bakoyianis & Koutinas,
temperature, 5

10 1996; Koutinas, Bakoyianis, Argiriou, Kanellaki, &

A S Voliotis, 1997; Loukatos et al., 2000) that could be used
V M 0.1022 0.0130 6.0824 1.33 0.0850 0.0079 0.0015
temperature, for industrialization of immobilized cells for wine mak-
fermentation
6

** 10 ** ing, as handling of the above supports at this scale could

be performed without any problems.
F 1.6534 24.7527 3.74 20.1416 1.0943 0.6945 0.1198
parameters: fermentation

the 7 FT:

of 10

ratio, 4. Conclusions
TA MS 0.3869 5.7921 8.75 4.7131 0.2561 0.1625 0.0280

signicance ** ** ** ** ** ** MS/error Factors such as fermentation temperature, type of
the :F
to 0.3279 methanol.
22.8972 13.5327 54.5460 11.4211 fermentation and selection of a suitable immobilization
F 361.8371 239.6084
ME:
square, support should be seriously considered as they seem to
related
play an important role in wine making by determining
mean
alcohols,
analysis 1251.4020 the cost eectiveness of the process and the quality of
87396.2578 51652.9141 43593.3242 MS:
914560.5000 208196.2813
DWP MS 1381094.6250 amyl 0.05. 0.01. the produced wines. Low-temperature (15 C) wine-fer-
< <
p p
AM:
statistical Df 1 1 1 1 1 1 mentation resulted in improved quality and the pro-
freedom, for for

the of duced wines had a distinctive aromatic prole. Apple
of
6 appears to have an advantage over quince and pear, as
degree isobutanol,
Signicant Signicant
Table Results Eect FT1 TF IS FT–TF FT–IS TF–IS FT–TF–IS Df: IB: it is widely cultivated, it is considered more compatible
* **

Y. Kourkoutas et al. / Journal of Food Engineering 69 (2005) 115–123 123


with wine taste, while continuous process using immobi- Kourkoutas, Y., Komaitis, M., Koutinas, A. A., Kaliafas, A.,

lized cells on apple pieces resulted in high wine Kanellaki, M., Marchant, R., & Banat, I. M. (2003). Wine
production using yeast immobilized on quince biocatalyst at
productivity.
temperatures between 30 and 0 C. Food Chemistry, 82, 353–360.
Kourkoutas, Y., Komaitis, M., Koutinas, A. A., & Kanellaki, M.
(2001). Wine production using yeast immobilized on apple pieces at
low and room temperatures. Journal of Agricultural and Food
References Chemistry, 49, 1417–1425.
Kourkoutas, Y., Koutinas, A. A., Kanellaki, M., Banat, I. M., &
Argiriou, T., Kaliafas, A., Psarianos, K., Kana, K., Kanellaki, M., & Marchant, R. (2002). Continuous wine fermentation using a
Koutinas, A. A. (1992). New alcohol resistant strains of S. psychrophilic yeast immobilized on apple cuts at dierent temper-
cerevisiae species for potable alcohol production using molasses. atures. Food Microbiology, 19, 127–134.
Applied Biochemistry and Biotechnology, 36, 153–161. Koutinas, A. A., Bakoyianis, V., Argiriou, T., Kanellaki, M., &
Bakoyianis, V., Kanellaki, M., Kaliafas, A., & Koutinas, A. A. (1992). Voliotis, S. (1997). A qualitative outline to industrialize alcohol
Low temperature wine making by immobilized cells on mineral production by catalytic multistage xed bed tower (MFBT)
Kissiris. Journal of Agricultural and Food Chemistry, 40(7), bioreactor. Applied Biochemistry and Biotechnology, 66, 121–131.
1293–1296. Koutinas, A. A., & Pefanis, S. (1994). Biotechnology of Foods and
Bakoyianis, V., & Koutinas, A. A. (1996). A catalytic multistage xed- Drinks. Patras, Greece: University of Patras, p. 74.
bed tower bioreactor in an industrial-scale pilot plant for alcohol Loukatos, P., Kiaris, M., Ligas, I., Bourgos, G., Kanellaki, M.,
production. Biotechnology and Bioengineering, 49, 197–203. Komaitis, M., & Koutinas, A. A. (2000). Continuous wine making
Bardi, E., & Koutinas, A. A. (1994). Immobilization of yeast on by c-alumina supported biocatalyst. Quality of the wine and
delignied cellulosic material for room temperature and low- distillates. Applied Biochemistry and Biotechnology, 89, 1–13.
temperature wine-making. Journal of Agricultural and Food Chem- Maicas, S., Pardo, I., & Ferrer, S. (2001). The potential of positively-
istry, 42, 221–226. charged cellulose sponge for malolactic fermentation of wine, using
Cabezudo, M. D., Gorostiza, E. F., Herraiz, M., Fernadez-Biarange, Oenococcus oeni. Enzyme and Microbial Technology, 28, 415–419.
J., Garcia-Dominguez, J. A., & Molera, M. J. (1978). Mixed Mallios, P., Kourkoutas, Y., Iconomopoulou, M., Koutinas, A. A.,
columns made to order in gas chromatography. IV. Isothermal Psarianos, C., Marchant, R., & Banat, I. M. B. (2004). Low
selective separation of alcoholic and acetic fermentation products. temperature wine-making using yeast immobilized on pear pieces.
Journal of Chromatographic Science, 16, 61–67. Journal of the Science of Food and Agriculture, 84, 1615–1623.
Etievant, P. X. (1991). Wine. In Henk Maarse, (Ed.), Volatile
Margaritis, A., & Merchant, F. J. A. (1984). Advances in ethanol
compounds in foods and beverages (pp. 483–546). production using immobilized cell systems. CRC Critical Reviews in
Fumi, M., Trioli, G., & Colagrande, O. (1987). Preliminary assessment Biotechnology, 1(4), 339–393.
on the use of immobilized yeast cells in sodium alginate for Ribereau-Cayon, P., Glories, Y., Maujean, A., & Dubourdieu, D.

sparkling wine processes. Biotechnology Letters, 9(5), 339–342. (2000). Formation of volatile acidity by bacteria. In Handbook of
Iconomou, L., Kanellaki, M., Voliotis, S., Agelopoulos, K., & enology. The chemistry of wine. Stabilization and treatments
Koutinas, A. A. (1996). Continuous wine making by delignied (pp. 214). West Sussex, England: John Wiley and Sons Ltd.
cellulosic materials supported biocatalyst. Applied Biochemistry and Shimobayashi, Y., & Tominaga, K. (1986). Application of biotech-
Biotechnology, 60, 303–313. nology in the food industry. I. Brewing of white wine by a
Jackson, R. S. (1994). Wine science. Principles and applications. San bioreactor. Hokaidoritsu Kogyo Shikenjo Hokoku, 285, 199–204.
Diego, California: Academic Press (pp. 184, 199, 201, 233, 250, Stewart, G. G., & Russell, I. (1986). One hundred years of yeast
255). research and development in the brewing industry. Journal of the
Kourkoutas, Y., Douma, M., Koutinas, A. A., Kanellaki, M., Banat, Institute of Brewing, 92, 537–558.
I. M., & Marchant, R. (2002). Room and low temperature Zoeklein, B., Fugelsang, K., Gump, B., & Nury, F. (1990). Volatile
continuous wine making using yeast immobilized on quince pieces. Acidity. In Van Norstrand Reinhold (Ed.), Production wine
Process Biochemistry, 39, 143–148. analysis (pp. 105–110). New York.


Developmental and Comparative Immunology 27 (2003) 835–844
www.elsevier.com/locate/devcompimm




Purication and binding properties of porcine plasma colin
that binds Actinobacillus pleuropneumoniae

A.S. Brooks, J.P. DeLay, M.A. Hayes*

Department of Pathobiology, University of Guelph, Guelph, Ontario, Canada N1G 2W1

Received 9 July 2002; revised 22 January 2003; accepted 25 February 2003




Abstract

Previous studies demonstrated that porcine plasma colin binds the important pig pathogen Actinobacillus
pleuropneumoniae (APP) in an N-acetylglucosamine-dependent manner. In the present study, attempts to characterize the
bacterial-binding properties of colin indicated colin is the major porcine plasma protein that binds directly to epoxy-activated
chromatography matrices. We developed an efcient method for purifying colin using epoxy-activated Toyopearl and
compared these with forms retrieved from other chromatography matrices and from intact APP. Puried colins retained their
GlcNAc- and bacterial-binding properties, and migrated as two high molecular weight multimers composed of 38, 40 and
42 kDa reduced forms (pI 5.2–6.0). An N-acetylated amine-activated Toyopearl matrix bound colin, and colin was
dissociated from this matrix with acetamide. Acetate, acetamide, and GlcNAc, but not glucose or glucosamine, dissociated
plasma colin from the surface of intact APP serotype 5b, which contains N-acetylated saccharides in the capsule. These studies
indicate that porcine colin binds APP 5b and an N-acetylated matrix in a similar manner, supporting the view that N-acetyl
groups may be important for binding of porcine plasma colin to some microbial surfaces.
q 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Ficolin; Pattern recognition proteins; Actinobacillus pleuropneumoniae; Lectins; N-acetylglucosamine; Bacterial polysaccharides;
Plasma proteins; Pigs; Innate immunity




1. Introduction proteins that bind molecular patterns on microbial
surfaces [1,4,5], either directly with carbohydrate-
Mammals have several evolutionarily conserved binding domains [6], or indirectly by immunoglobulin
collagenous defense proteins, such as the collectins binding, in the case of C1q [7]. The microbial surface
[1], structurally-related colins [2], and complement recognition properties underlie their agglutination,
component C1q [3]. These are large multivalent opsonization, and complement xation functions that
contribute to innate and acquired resistance to
Abbreviations: GlcNAc, N-acetylglucosamine; GalNAc, N- bacterial infection [1,8,9].
acetylgalactosamine; Glc, glucose; NaOAc, sodium acetate; APP, Ficolins are lectins that bind N-acetylglucosamine
Actinobacillus pleuropneumoniae; MBL, mannan-binding lectin.
[10,11] and were rst identied in the pig uterus
* Corresponding author. Tel.: 1-519-824-4120x54637; fax:
1-519-824-5930. [12], but have since been characterized in humans
E-mail address: ahayes@uoguelph.ca (M.A. Hayes). [10,13–15], rodents [16,17], hedgehogs [18], and

0145-305X/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0145-305X(03)00057-0

836 A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844


invertebrates [19]. Ficolins contain a C-terminal from Amersham-Pharmacia (Baie d’Urfe, Que,
collagenous domain and an N-terminal brinogen- Canada). The Bio-Rad Protein Assay kit was used
like domain and multiple subunit-trimers form a for determining protein concentrations (Bio-Rad
‘bundle of tulips’ structure similar to C1q and certain Laboratories, Ont, Canada). Rabbit anti-porcine
collectins such as MBL [17,20–22]. Ficolins bind colin antiserum was previously prepared against
GlcNAc at the C-terminal brinogen-like domain colin 38, 40 and 42 kDa subunits [28]. Goat
[23,24], and the clustering of numerous brinogen- anti-rabbit polyclonal HRP was obtained from the
like domains in the colin polymer implies a role in Dako Corporation (Santa Barbara, CA, USA).
binding to multivalent polysaccharide targets [6,9].
For example, colins may bind some 2.2. Ficolin afnity-purication
microbial surface polysaccharides since human
plasma L-colin binds Salmonella typhimurium and Plasma was obtained from healthy adult York-
E.coli in a carbohydrate-dependent manner [10,25]. shire-cross sows. Whole blood was collected into
Furthermore, human L- and H-colin (Hakata 3.8% buffered sodium citrate (pH 7.4, 9:1 blood:
antigen) are complexed with MBL-associated serine citrate ratio) and centrifuged at 1000g for 30 min at
proteases and may activate complement upon room temperature. Platelet-poor plasma fractions
binding bacterial surfaces [26,27]. In contrast, direct were stored at 270 8C. For afnity-purication and
evidence implicating porcine colins in similar ligand binding studies, Toyopearl AF-Epoxy 650M
functional roles is lacking. and epoxy-activated Sepharose 6B were conjugated
In our recent search for porcine plasma lectins that to GlcNAc and APP LPS, respectively, according
bind bacteria, colin was found as the major to the manufacturer’s protocol. Five grams of
porcine plasma protein that bound some isolates of epoxy-activated Sepharose 6B were conjugated
Actinobacillus pleuropneumoniae (APP) in a with 75 mg of LPS that was extracted from APP
GlcNAc-dependent manner [28]. Identication of as described [31], in 0.1 M NaOH (pH 12) for 16 h
innate defense mechanisms against Actinobacillus with shaking at 37 8C. The coupled matrix was
sp. is of interest because these pathogens are washed with 500 ml of 0.1 M NaOH (pH 12), and
responsible for economically-signicant pneumonic non-reacted epoxide groups were blocked using
and septicemic disease in young pigs [29,30]. In this 1 M ethanolamine (pH 8) at 37 8C for 16 h.
report we describe the purication of bacterial-bind- GlcNAc (150 mg) was conjugated to rehydrated
ing forms of porcine plasma colin. Afnity Toyopearl AF-Epoxy 650M (5 gm) in 0.1 M NaOH
chromatography and binding studies with APP (pH 11) for 16 h with at 37 8C with shaking, and
serotype 5b indicate that N-acetyl groups may be residual non-reacted epoxide groups were similarly
important for binding of porcine plasma colin to blocked with ethanolamine. Ligand-adducted
some microbial surfaces. matrices were washed extensively with MilliQ
H2O and equilibrated in PBS pH 7.4.
An N-acetyl Toyopearl afnity matrix
2. Materials and methods was prepared as described [32]. Briey, Toyopearl
AF-Amino 650M (10 g) was washed twice with
2.1. Chemicals and reagents 500 ml of distilled water and incubated with 8 ml of
0.2 M sodium acetate and 4 ml of acetic anhydride for
All chemicals were obtained from Fisher Scientic 30 min on ice. An additional 4 ml of acetic anhydride
(Ottawa, Ont, Canada), except the monosaccharides was added to the mixture and the incubation continued
which were obtained from Sigma (Oakville, Ont, for another 30 min. The resin was washed sequentially
Canada). Toyopearl AF-Epoxy 650M and Toyopearl with several volumes MilliQ water, 1 M NaOH,
AF-Amino 650M were obtained from TosoHaas, and then equilibrated in PBS pH 7.4.
(Montgomery, PA, USA). Epoxy-activated Sepharose Unmodied and ethanolamine-blocked epoxy-
6B, Sepharose 6B (agarose), HiTrap Protein G, and activated Toyopearl and epoxy-activated Sepharose
chemiluminescence detection reagents were obtained 6B matrices were used as controls. Epoxide groups

A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844 837


were blocked with ethanolamine as above. Amicon, Beverly, MA, USA), and then characterized
Unmodied Toyopearl AF-Epoxy 650M and Epoxy- by non-denaturing (native)- and reducing
activated Sepharose 6B matrices were used as SDS–PAGE, and Western blots.
supplied, with neither ligand nor ethanolamine A simplied purication protocol was developed
exposure. Sepharose 6B and Toyopearl AF-Amino for routine colin purication with the unmodied
650M matrices that are not epoxy-activated were also Toyopearl AF-Epoxy 650M matrix. Citrated plasma
used as controls. Table 1 contains a list of the (20 ml) was applied to the matrix, which was washed
chromatography matrices used in the present study. in PBS to baseline absorbance, and bound proteins
All afnity chromatography was performed at 4 8C. were sequentially eluted with 1 M NaCl pH 7.4, and
Absorbance (280 nm), conductivity, and pH of the 50 mM NaOAc pH 3.9. NaOAc elutions containing
elutions were continuously recorded. Citrated plasma colin were collected into 1 M Tris–HCl pH 8.0,
(50 ml) was loaded onto each afnity matrix and buffer exchanged into PBS pH 7.4, lyophilized, and
loaded columns were washed with equilibration buffer stored at 220 8C. NaOAc elutions contained
(50 mM Tris–HCl pH 7.4) to baseline absorbance, additional contaminating proteins, and these were
then sequentially eluted with 50 mM Tris–HCl pH removed in a second purication step with a HiTrap
7.4 containing 1 M NaCl (salt), 50 mM EDTA Protein G column (1 ml bed volume) (Amersham
(EDTA), 300 mM a-D-glucose (glucose), 300 mM Pharmacia). Briey, NaOAc eluents were applied to
a-D-GlcNAc (GlcNAc), 300 mM sodium acetate the Protein G column, which was then washed with
(acetate), 300 mM acetamide (acetamide), and a PBS pH 7.4 to baseline absorbance. Bound proteins
nal stripping buffer of 50 mM acetic acid (NaOAc) were sequentially eluted with 1 M NaCl, 300 mM
pH 3.5 (acid). In some chromatography runs, EDTA GlcNAc, and 20 mM glycine (pH 2.8). Ficolin was
and glucose elutions were omitted, and GlcNAc, eluted from this matrix with 300 mM GlcNAc,
acetamide, acetate, and acid elutions were whereas the protein contaminants and excess colin
ordered selectively to determine respective ability to were removed in the unbound fractions. To determine
dissociate colin bound to the chromatography if colin puried from plasma either by afnity to
matrix. Elutions containing absorbance peaks were unmodied Toyopearl AF-Epoxy 650M or to intact
collected, concentrated (Centricon 10 concentrators, APP serotype 5b retained their binding properties,


Table 1
Summary of colin afnity-purication

Afnity matrix Adducted ligand Ficolin bound Ficolin elution



Sepharose 6B None No –

Epoxy-activated Sepharose 6B None (unmodied) Yes NaOAc pH 3.5
Ethanolamine Yes NaOAc pH 3.5
Lipopolysaccharide Yes NaOAc pH 3.5

Toyopearl AF Epoxy 650 M None (unmodied) Yes Acetate/acetamide/GlcNAc/
NaOAc pH 3.5
Ethanolamine Yes GlcNAc/NaOAc pH 3.5
GlcNAc Yes GlcNAc/NaOAc pH 3.5

Toyopearl AF Amino 650 M None (unmodied) No –
Acetyl-groups Yes Acetamide

HiTrap Protein G Protein G Yes GlcNAc

Ficolin bound the epoxy-activated Sepharose 6B and epoxy-activated Toyopearl matrices regardless of whether they were conjugated with a
ligand, blocked with ethanolamine, or left unmodied. Toyopearl AF Amino 650M did not bind colin and could be converted to a colin
binding target by N-acetylation. Ficolin binding to HiTrap Protein G was used in a second purication step.

838 A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844


GlcNAc elutions containing puried colin Membranes were washed in ve changes of PBS-T
were dialyzed into PBS pH 7.4, re-applied to and exposed to goat anti-rabbit immunoglobulins
the unmodied Toyopearl AF-Epoxy 650M matrix, conjugated to HRP (Dako) (1:2000 dilution in
and re-eluted with 300 mM GlcNAc. PBS-T). Western blots were then developed by
chemiluminescence.
2.3. Binding to A. pleuropneumoniae


Previous studies demonstrated that porcine plasma 3. Results
colins bind to some serotypes of APP [28].
To determine if colins puried by afnity The results of various colin purication strategies
chromatography retained the ability to bind bacteria, in the present study are summarized in Table 1.
a binding assay was performed with an isolate of Initially we attempted to afnity-purify bacterial-
APP 5b (VSB 1104), (serotyping performed by binding colins from porcine plasma using
Dr M. Gottschalk, University of Montreal). Briey, LPS coupled to epoxy-activated Sepharose 6B.
bacteria were grown on blood agar plates sup- Ficolin was present in the acid (NaOAc pH 3.5), but
plemented with 100 ml of 10% NAD and single not the GlcNAc elutions, from the LPS-conjugated
colonies were selected for culture in BHI broth and the control (unmodied) epoxy-activated Sepha-
(Difco Laboratories, Detroit, MI, USA) containing rose 6B matrices. The 38, 40 and 42 kDa (reduced)
0.02% NAD. Bacteria (1 108 CFU) were harvested forms of the colin puried with these epoxy-
by centrifugation (10,000g 10 min at 4 8C), washed activated Sepharose 6B matrices are illustrated in
twice in PBS pH 7.4, and incubated with whole Fig. 2, in comparison with other afnity-puried
plasma (1.5 ml) or NaOAc elutions containing colin colins. These results indicated that porcine plasma
(20 mg total protein) (after buffer exchange into PBS colin binds directly to epoxy-activated Sepharose 6B
pH 7.4) for 1 h with gentle agitation at 4 8C. Bacteria irrespective of a conjugated ligand.
were pelleted and resuspended in PBS four times, and Because colin was potentially interacting directly
bound proteins were sequentially eluted from the with uncharacterized structures on epoxy-activated
bacterial surface in 400 ml of 1 M NaCl, 300 mM Sepharose 6B, we used an epoxy-activated acrylic
GlcNAc, and 50 mM NaOAc (pH 3.5). In separate (non-glycan) matrix (Toyopearl AF-Epoxy 650M).
bacterial-binding studies, bound proteins were eluted Ficolins eluted from the GlcNAc-conjugated
with either 300 mM sodium acetate, 300 mM acet- Toyopearl AF-Epoxy 650M and the control (unmo-
amide, 300 mM a-D-glucose, 300 mM a-D-glucosa- died) Toyopearl AF-Epoxy 650M matrices in either
mine, or 300 mm a-D-GlcNAc in PBS pH 7.4. 300 mM GlcNAc or acid (NaOAc pH 3.5) elutions
All eluents were concentrated (Centricon 10 concen- (Table 1). Ficolin did not bind the non-activated
trators, Amicon, Beverly, MA, USA), and then Sepharose 6B or the Toyopearl AF-Amino 650M
characterized by reducing SDS–PAGE. matrices, neither of which contain epoxide-spacer arm
adducts (data not shown). Ficolin afnity-puried on
2.3.1. Electrophoresis and Western blots the epoxy-activated Toyopearl matrices exhibited
Eluents from afnity chromatography and bacterial similar major and minor, high molecular weight
binding assays were characterized by reducing SDS-, forms that migrated similarly to human a2-macro-
non-denaturing (native)-, and two-dimensional globulin (720 kDa) on non-denaturing (native) PAGE
(2D)-PAGE. First dimension isoelectric focusing was (Fig. 1). Under reducing conditions, afnity-puried
performed using non-linear pH 3–10 immobilized pH colins migrated as multiple subunits with apparent
gradient strips (Amersham Pharmacia). For western sizes of 38, 40 and 42 kDa (Fig. 2) which were
blots, proteins separated by 1D- and 2D-PAGE were immuno-reactive with anti-porcine colin antiserum
electroblotted onto nitrocellulose membranes. Mem- (Fig. 2). The 38, 40 and 42 kDa bands were excised
branes were blocked in PBS-T (PBS pH 7.4 containing separately and N-terminal sequence over 16 amino
1% BSA, and 0.1% Tween-20) and exposed to rabbit acid residues were consistent with porcine colin a
anti-colin antiserum diluted 1:1000 in PBS-T. for each band [28].

A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844 839


The direct binding of porcine colin, independent
of a conjugated ligand, to the epoxy-activated
Toyopearl AF-Epoxy 650 M and epoxy-activated
Sepharose 6B matrices indicated these were
unsuitable for studying colin–ligand interactions,
but a simple method to purify porcine plasma colin
was developed from these studies. The unmodied
Toyopearl AF-Epoxy 650 M matrix was the least
complex, most efcient matrix for purication of
porcine plasma colins and was chosen for routine
purication (Fig. 3). Approximately 0.5–1 mg of
colin could be obtained in the 50 mM NaOAc pH
3.5 elution from 20 ml of plasma which is consistent
with reported colin concentrations of 10–80 mg/ml
[33]. The epoxy-activated Toyopearl AF-Epoxy
650M efciently depleted immuno-reactive colin
Fig. 1. Non-denatured forms of colin obtained by afnity from plasma (Fig. 3). The NaOAc elution (Fig. 3)
chromatography. Coomassie blue-stained non-denaturing (native) also contained variably abundant additional proteins
gradient PAGE (5–10%). Afnity-puried colins migrated as two
with reduced sizes of ,200, 81, 63, 55, and 25 kDa.
high molecular weight bands relative to human a2-macroglobulin
(MW , 720 kDa). Lane 1: human a2-macroglobulin. Lane 2: N-terminal sequence of the 81 and 63 kDa bands
GlcNAc elution of colins that bound GlcNAc-Toyopearl. Lane 3: were consistent with porcine IgM heavy chain and
GlcNAc elution of colins that bound ethanolamine–Toyopearl. albumin, respectively [28]. However, all of these
Lane 4: GlcNAc elution of colins that bound unmodied epoxy-
could be removed by binding colin to Protein G
activated Toyopearl.
Sepharose and eluting it with 300 mM GlcNAc
(Fig. 3C). The forms of colin puried by this
method closely resembled the native and reduced




Fig. 2. Reduced forms of colin puried by chromatography. (A) Coomassie-stained reducing SDS–PAGE (12%). Similar 38, 40, and 42 kDa
colin bands (labelled*) were present in the 50 mM NaOAc pH 3.5 elutions (lanes 2–4) from various epoxy-activated Sepharose 6B matrices,
and in the GlcNAc elutions (lanes 5–7) from various epoxy-activated Toyopearl matrices. Lane 1: molecular weight marker. Lane 2:
LPS–Sepharose 6B. Lane 3: ethanolamine–Sepharose 6B. Lane 4: unmodied epoxy-activated Sepharose 6B. Lane 5: GlcNAc–Toyopearl.
Lane 6: ethanolamine–Toyopearl. Lane 7: unmodied epoxy-activated Toyopearl. (B) Western blot of (A) (same lane loads as in (A)) with
rabbit anti-porcine colin antiserum. Similar immuno-reactive colin forms are present in the GlcNAc and 50 mM NaOAc pH 3.5 elutions
depicted in (A).

840 A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844




Fig. 3. Method for purifying plasma colin. Application of plasma onto the unmodied epoxy-activated Toyopearl matrix depleted of immuno-
reactive colin from plasma. (A) Silver-stained SDS–PAGE (12%) Lane 1: molecular weight marker. Lane 2: plasma (30 mg total protein)
before exposure to the epoxy-activated Toyopearl. Lane 3: plasma (30 mg total protein) after exposure to the epoxy-activated Toyopearl. Lane
4: NaOAc pH 3.5 elution containing 38, 40 and 42 kDa colin bands (total protein load 5 mg), as well as additional protein bands at ,200, 81,
63, 55, and 25 kDa. (B) Western blot with rabbit anti-porcine colin antiserum. Lane 5: plasma (30 mg total protein) before exposure to the
epoxy-activated Toyopearl. Lane 6: plasma (30 mg total protein) after exposure to the epoxy-activated Toyopearl showing absence of colin.
Lane 7: NaOAc pH 3.5 elution containing colin (total protein load 0.1 mg). (C) Silver stain SDS–PAGE (12%) depicting the 300 mM GlcNAc
elution containing puried colins (38, 40 and 42 kDa colin bands) that bound the HiTrap Protein G Sepharose.




Fig. 4. Comparison of puried and plasma colin. Ficolin puried with the unmodied epoxy-activated Toyopearl resembled colin forms in
the original plasma. (A) Western blot of non-denaturing (native) gradient PAGE (5–10%) with rabbit anti-porcine colin antiserum. Lane 1:
plasma. Lane 2: NaOAc elution from Toyopearl AF-Epoxy 650M chromatography. (B) Silver stained 2D-PAGE of afnity-puried colin
revealed an array of colin subunit forms (apparent MW 38–42 kDa, pI range 5.2–6.0) similar to those in whole plasma (pI range 5.4–6.0)
detected by Western blot (C) with rabbit anti-porcine colin antiserum.

A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844 841


matrix, in its unmodied state, did not bind
porcine plasma colin (data not shown). Accordingly,
putative colin targets were coupled to this matrix.
Acetic anhydride was used to N-acetylate the amine-
activated Toyopearl matrix [32], and colin in porcine
plasma subsequently bound to this N-acetylated
matrix and was eluted with 300 mM acetamide
(data not shown). Also, acetate or acetamide dis-
sociated plasma colin from the unmodied epoxy-
activated Toyopearl AF-Epoxy 650 M (data not
shown).
The effect of various glucose derivatives and small
acetyl-containing compounds on colin binding was
further assessed in bacterial-binding assays. APP
serotype 5b was incubated with citrated plasma,
washed, and then exposed to equi-molar concen-
trations of either sodium acetate, acetamide, glucose,
Fig. 5. Puried colin retains bacterial-binding properties. Silver
stained reducing SDS–PAGE (12%). APP serotype 5b was glucosamine, or N-acetylglucosamine (GlcNAc).
incubated with porcine colin puried with the unmodied Characteristic 38, 40 and 42 kDa colin bands
epoxy-activated Toyopearl matrix. Lane 1: molecular weight (reduced) were eluted from the bacterial surface by
marker. Lane 2: Ficolin (NaOAc elution) which was incubated
sodium acetate, acetamide, and N-acetylglucosamine,
with bacteria. Lane 3: unbound proteins. Lane 4: Final PBS wash of
the bacteria. Lane 5: 1 M NaCl elution. Lane 6: 300 mM GlcNAc
elution containing 38, 40 and 42 kDa colin bands. Lane 7: 50 mM
NaOAc pH 3.5 elution.



colin forms (pI range 5.4–6.0) present in the
original plasma (Fig. 4).
Ficolin puried from plasma with the Toyopearl
AF-Epoxy 650 M subsequently bound to intact APP
serotype 5b (Fig. 5). Plasma colin that eluted from
the unmodied-Toyopearl AF-Epoxy 650M matrix
was incubated with intact APP serotype 5b.
The bacteria were washed extensively and sequen-
tially eluted with 1 M NaCl, 300 mM GlcNAc, and
50 mM NaOAc pH 3.5. Ficolin was consistently
detected in GlcNAc elutions from the bacterial surface
indicating that GlcNAc-dependent bacterial-binding
properties were retained during purication (Fig. 5).
Furthermore, plasma colin that eluted with GlcNAc
from the unmodied-Toyopearl AF-Epoxy 650M
matrix, or intact bacteria, retained the ability to bind
Fig. 6. Acetyl-containing compounds dissociate colin from the
the unmodied Toyopearl matrix (data not shown). bacterial surface. Silver stained reducing SDS–PAGE (12%). After
Ficolin which was further puried by Protein G incubating APP serotype 5b in porcine plasma, colin was

Sepharose chromatography also bound to APP dissociated from the bacterial surface with acetate, acetamide, and
GlcNAc, but not glucose or glucosamine. Lane 1: 300 mM acetate
serotype 5b (data not shown).
elution; Lane 2: 300 mM acetamide elution; Lane 3:. 300 mM
Unlike the epoxy-activated Toyopearl, the ami- glucose elution; Lane 4: 300 mM glucosamine elution; Lane 5:
ne-activated Toyopearl AF-Amino 650 M afnity 300 mM N-acetylglucosamine elution.

842 A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844


but not glucose or glucosamine (Fig. 6). Similar This limitation is similar to human L-colin which
results were obtained by incubating the bacteria with binds CNBr-activated Sepharose 6B non-specically
afnity-puried colin instead of plasma (data not [34]. Toyopearl AF-Epoxy 650M is composed of
shown). epoxide groups separated from acrylic beads by a
spacer-arm. Epoxy-activated Sepharose 6B is
a galactan polymer also coated with epoxide-carrying
4. Discussion spacer-arms. The addition of epoxide-carrying spacer
arms onto the base matrix polymer may have
The present studies reveal that porcine introduced colin binding sites since colin did not
plasma colin binds to several commonly used bind the unmodied Toyopearl AF-Amino 650M or
epoxy-activated chromatography matrices in the Sepharose 6B which are composed of the same acrylic
absence of specic ligand adducts. While this limits and galactan base polymers, respectively.
the use of these matrices as a means of studying colin However, specic colin–ligand interactions of
ligand interactions, these studies provided a simple interest could be studied with Toyopearl AF-Amino
and rapid method to obtain pure quantities of colin 650M, which lacked direct colin-binding properties.
which retained the ability to bind the important pig Porcine plasma colin also bound to hydroxysuccini-
pathogen APP. However, colin bound specically mide-activated Protein G Sepharose suggesting that
to an N-acetylated Toyopearl matrix and similar there are various derivatives of Sepharose which bind
acetyl-containing compounds dissociated colin from porcine colin. This might be of practical importance
this matrix and from APP serotype 5b. This suggests because Protein G-Sepharose is commonly used
that N-acetyl structures may be important in the to purify immunoglobulins, so colins could con-
binding interactions of porcine plasma colins. taminate these preparations. However, such colin
Our objective was to purify and characterize the contamination could be avoided by including a
binding properties of porcine plasma colin that bind GlcNAc elution prior to eluting the immunoglobulin.
to bacteria. In a previous report, porcine plasma colin Multiple carbohydrate-binding brinogen-like
lost GlcNAc-binding properties during chromatog- domains are tightly localized at one pole of the colin
raphy on a GlcNAc–agarose matrix and smaller molecule [20,21], which suggests that colin may
molecular weight forms of colin were generated preferentially bind multivalent targets, similar to
[21]. Whether these colin forms bind bacteria was other collagenous defense lectins such as the collec-
not described [21]. In our experiments, coupling tins [1,6,35]. Porcine plasma colin may bind targets
potential bacterial saccharide binding targets to containing N-acetyl groups since introduction of
an afnity matrix was unnecessary because the N-acetyl groups onto the amine-activated Toyopearl
unmodied epoxy-activated Toyopearl AF-Epoxy converted it into a colin binding target. Also, simple
650M provided efcient purication of porcine acetyl-containing molecules such as acetate,
colins that retained binding functions towards acetamide, and GlcNAc dissociated porcine plasma
GlcNAc, bacteria, and the preparative matrix. All of colin from the surface of APP serotype 5b, but the
the immuno-reactive colin could be removed from precursor sugars glucose and glucosamine, that lack
plasma by this method, and the puried colin the N-acetyl group, did not. The capsular polysac-
exhibited similar native and subunit forms when charides and LPS O-chains of APP are characterized.
compared with colin in plasma. The ability to purify In the capsule of APP serotype 5b, N-acetyl groups are
bacterial-binding forms of porcine colin enables present in the form of GlcNAc residues [36], and the
further study of the potential roles of these proteins in capsule also contains glucose and 3-deoxy-D-manno-
phagocytosis, complement activation, and the patho- octulosonic acid [36]. Although the LPS O-chain of
genesis of porcine pleuropneumonia. this serotype is a linear polymer of galactose, some
The basis for colin binding to the epoxy-activated isolates may contain O-acetyl modications in the
Sepharose and Toyopearl matrices is not O-chain [37]. While the precise structure(s) on
understood, but use of these matrices could lead to the surface of APP serotype 5b bound by colin are
mis-identication of porcine colin ligands. not revealed in these studies, it is possible that acetyl

A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844 843


groups in the surface polysaccharides may be Acknowledgements
important for colin binding to this serotype.
The acetyl-binding property may be evolutionarily We would like to thank Bette Anne Quinn for
conserved in porcine colins. The amino acid technical assistance and advice, and NSERC, the
sequence of invertebrate colin p40 of the solitary Canadian Research Network on Bacterial Pathogens
ascidian Halocynthia roretzi is up to 52% homologous in Swine, OMAFRA, the Ontario Ministry of
with porcine colin a, and p40 binds the N-acetyl Education and Training, and the Ontario
group of GlcNAc and GalNAc [19]. The horseshoe Veterinary College for stipend support. We are
crab tachylectins 5A and B are brinogen-like indebted to Dr M Gottschalk (University of Montreal)
proteins that lack a collagen-like domain but have for performing bacterial serotyping, and

44% amino acid sequence homology with the Dr J. MacInnes (University of Guelph) for supplying

brinogen-like domain of porcine colin a [20,32]. the isolate of APP serotype 5b, and for serotyping

TL-5A/B bind the acetyl structure present in many additional bacterial strains.

compounds such as acetate, acetamide, acetylsalicylic
acid, GlcNAc, and microbial surfaces [32]. The crystal
structure of TL-5A complexed with the N-acetyl References

group of GlcNAc [38,39] has recently been solved,
[1] Lu J, Teh C, Kishore U, Reid KB. Collectins and colins:
and the amino acid sequence homology suggests
sugar pattern recognition molecules of the mammalian innate
porcine colins could interact with acetyl groups in a
immune system. Biochim Biophys Acta 2002;1572:387–400.
similar fashion [38,39]. [2] Matsushita M, Fujita T. The role of colins in innate
N-acetyl groups, such as those in GlcNAc, are immunity. Immunobiology 2002;205:490–7.

present in many varieties of microbial cell [3] Kishore U, Reid KB. C1q: structure, function, and receptors.
Immunopharmacology 2000;49:159–70.
envelopes that contain peptidoglycan [40], chitin
[4] Janeway Jr. CA, Medzhitov R. Innate immune recognition.
[41], LPS core regions [42], and some bacterial Annu Rev Immunol 2002;20:197–216.
capsules [36] and may present a suitable molecular [5] Medzhitov R, Janeway Jr. CA. Decoding the patterns of self

pattern recognizable by porcine plasma colins. and nonself by the innate immune system. Science 2002;296:
298–300.
However, the core and antennary structures of host
[6] Weis WI, Taylor ME, Drickamer K. The C-type lectin
N-linked glycans, some glycosaminoglycans such
superfamily in the immune system. Immunol Rev 1998;163:
as hyaluronic acid [43], and sialic acids also 19–34.
contain acetyl modications [43], but it is unknown [7] Kishore U, Kojouharova MS, Reid KB. Recent progress in the

whether colins bind host glycans. The spatial understanding of the structure–function relationships of the
globular head regions of C1q. Immunobiology 2002;205:
distribution and multiplicity of carbohydrate bind-
355–64.
ing targets on microbial surfaces facilitates binding [8] Reid KB, Colomb MG, Loos M. Complement component C1
by other collagenous defense lectins such as MBL and the collectins: parallels between routes of acquired and

[6,9,35]. Thus other factors such as ligand density, innate immunity. Immunol Today 1998;19:56–9.
[9] Hoffmann JA, Kafatos FC, Janeway CA, Ezekowitz RA.
accessibility, charge and steric hindrance may be
Phylogenetic perspectives in innate immunity. Science 1999;
important factors in colin binding. To understand
284:1313–8.
colin function, it is important to understand how [10] Matsushita M, Endo Y, Taira S, Sato Y, Fujita T, Ichikawa N,
colins might distinguish between polymeric targets Nakata M, Mizuochi T. A novel human serum lectin with

given that acetyl structures are commonly found in collagen- and brinogen-like domains that functions as an
opsonin. J Biol Chem 1996;271:2448–54.
biomolecules. While there is some evidence that
[11] Matsushita M, Fujita T. Ficolins and the lectin complement
human colins may function as opsonins and pathway. Immunol Rev 2001;180:78–85.
complement-activators, the functions of porcine [12] Ichijo H, Ronnstrand L, Miyagawa K, Ohashi H, Heldin CH,

colins are unknown [2]. The ability to purify Miyazono K. Purication of transforming growth factor-beta 1
binding proteins from porcine uterus membranes. J Biol Chem
porcine colins with dened bacterial-binding
1991;266:22459–64.
properties permits further investigation into the [13] Harumiya S, Omori A, Sugiura T, Fukumoto Y, Tachikawa H,
functions of these proteins in the pig. Fujimoto D. EBP-37, a new elastin-binding protein in human

844 A.S. Brooks et al. / Developmental and Comparative Immunology 27 (2003) 835–844


plasma: structural similarity to colins, transforming growth [27] Matsushita M, Kuraya M, Hamasaki N, Tsujimura M, Shiraki
factor-beta 1-binding proteins. J Biochem (Tokyo) 1995;117: H, Fujita T. Activation of the lectin complement pathway by
1029–35. H-colin (Hakata antigen). J Immunol 2002;168:3502–6.
[14] Edgar PF. Hucolin, a new corticosteroid-binding protein from [28] DeLay J. Biochemical characterization and bacterial binding
human plasma with structural similarities to colins, trans- functions of porcine colins. DVSc Thesis. University of
forming growth factor-beta 1-binding proteins. FEBS Lett Guelph; 1999
1995;375:159–61. [29] Fenwick B, Henry S. Porcine pleuropneumonia. J Am Vet
[15] Lu J, Tay PN, Kon OL, Reid KB. Human colin: cDNA Med Assoc 1994;204:1334–40.
cloning, demonstration of peripheral blood leucocytes as the [30] MacInnes JI, Desrosiers R. Agents of the suis-ide diseases of
major site of synthesis and assignment of the gene to swine: Actinobacillus suis, Haemophilus parasuis, and
chromosome 9. Biochem J 1996;313:473–8. Streptococcus suis. Can J Vet Res 1999;63:83–9.
[16] Fujimori Y, Harumiya S, Fukumoto Y, Miura Y, Yagasaki K, [31] Darveau RP, Hancock RE. Procedure for isolation of bacterial
Tachikawa H, Fujimoto D. Molecular cloning and character- lipopolysaccharides from both smooth and rough Pseudomo-
ization of mouse colin-A. Biochem Biophys Res Commun nas aeruginosa and Salmonella typhimurium strains.
1998;244:796–800. J Bacteriol 1983;155:831–8.
[17] Ohashi T, Erickson HP. Oligomeric structure and tissue [32] Gokudan S, Muta T, Tsuda R, Koori K, Kawahara T, Seki N,
distribution of colins from mouse, pig and human. Arch Mizunoe Y, Wai SN, Iwanaga S, Kawabata S. Horseshoe crab
Biochem Biophys 1998;360:223–32. acetyl group-recognizing lectins involved in innate immunity
[18] Omori-Satoh T, Yamakawa Y, Mebs D. The antihemorrhagic are structurally related to brinogen. Proc Natl Acad Sci USA
factor, erinacin, from the European hedgehog (Erinaceus 1999;96:10086–91.
europaeus), a metalloprotease inhibitor of large molecular size [33] Seebaransingh R. Plasma colins and acute phase proteins of
possessing colin/opsonin P35 lectin domains. Toxicon 2000; young pigs. MSc Thesis. University of Guelph; 1999
38:1561–80. [34] Le Y, Tan SM, Lee SH, Kon OL, Lu J. Purication and
[19] Kenjo A, Takahashi M, Matsushita M, Endo Y, Nakata M, binding properties of a human colin-like protein. J Immunol
Mizuochi T, Fujita T. Cloning and characterization of novel Methods 1997;204:43–9.
colins from the solitary ascidian, Halocynthia roretzi. J Biol [35] Weis WI, Drickamer K. Structural basis of lectin–carbo-
Chem 2001;276:19959–65. hydrate recognition. Annu Rev Biochem 1996;65:441–73.
[20] Ichijo H, Hellman U, Wernstedt C, Gonez LJ, Claesson-Welsh [36] Altman E, Brisson JR, Gagne SM, Perry MB. Structure of the
L, Heldin CH, Miyazono K. Molecular cloning and capsular polysaccharide of Actinobacillus pleuropneumoniae
characterization of colin, a multimeric protein with brino- serotype 5b. Eur J Biochem 1992;204:225–30.
gen- and collagen-like domains. J Biol Chem 1993;268: [37] Altman E, Grifth DW, Perry MB. Structural studies of the O-
14505–13. chains of the lipopolysaccharides produced by strains of
[21] Ohashi T, Erickson HP. Two oligomeric forms of plasma Actinobacillus (Haemophilus) pleuropneumoniae serotype 5.
colin have differential lectin activity. J Biol Chem 1997;272: Biochem Cell Biol 1990;68:1268–71.
14220–6. [38] Kairies N, Beisel HG, Fuentes-Prior P, Tsuda R, Muta T,
[22] Hansen S, Holmskov U. Structural aspects of collectins and Iwanaga S, Bode W, Huber R, Kawabata S. The 2.0-A crystal
receptors for collectins. Immunobiology 1998;199:165–89. structure of tachylectin 5A provides evidence for the common
[23] Le Y, Lee SH, Kon OL, Lu J. Human L-colin: plasma origin of the innate immunity and the blood coagulation
levels, sugar specicity, and assignment of its lectin systems. Proc Natl Acad Sci USA 2001;98:13519–24.
activity to the brinogen-like (FBG) domain. FEBS Lett [39] Kawabata S, Tsuda R. Molecular basis of non-self recognition
1998;425:367–70. by the horseshoe crab tachylectins. Biochim Biophys Acta
[24] Teh C, Le Y, Lee SH, Lu J. M-colin is expressed on 2002;1572:414–21.
monocytes and is a lectin binding to N-acetyl-D-glucosamine [40] van Heijenoort J. Formation of the glycan chains in the
and mediates monocyte adhesion and phagocytosis of synthesis of bacterial peptidoglycan. Glycobiology 2001;11:
Escherichia coli. Immunology 2000;101:225–32. 25R–36R.
[25] Lu J, Le Y, Kon OL, Chan J, Lee SH. Biosynthesis of human [41] Munro CA, Gow NA. Chitin synthesis in human pathogenic
colin, an Escherichia coli-binding protein, by monocytes: fungi. Med Mycol 2001;39:41–53.
comparison with the synthesis of two macrophage-specic [42] Rietschel ET, Kirikae T, Schade FU, Mamat U, Schmidt G,
proteins, C1q and the mannose receptor. Immunology 1996; Loppnow H, Ulmer AJ, Zahringer U, Seydel U, Di Padova F,
89:289–94. et al. Bacterial endotoxin: molecular relationships of structure
[26] Matsushita M, Endo Y, Fujita T. Cutting edge: comp- to activity and function. Faseb J 1994;8:217–25.
lement-activating complex of colin and mannose-binding [43] Reuter G, Gabius HJ. Eukaryotic glycosylation: whim of
lectin-associated serine protease. J Immunol 2000;164: nature or multipurpose tool? Cell Mol Life Sci 1999;55:
2281–4. 368–422.


Biochimica et Biophysica Acta 1635 (2003) 10–19
www.bba-direct.com




Characterization and partial purification of protein fatty
acyltransferase activity from rat liver

Abel Hiola,1, Joan M. Caronb, Charles D. Smithc, Teresa L.Z. Jonesa,*

a Metabolic Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, USA
b Department of Physiology, University of Connecticut Health Center, Farmington, CT, USA
c Department of Pharmacology, Pennsylvania State University, Hershey, PA, USA

Received 10 March 2003; received in revised form 22 September 2003; accepted 3 October 2003




Abstract


The acylation of proteins through the addition of palmitate to cysteine residues is a common posttranslational modification for a variety
of proteins, but the enzymology of this reversible modification has resisted elucidation. We developed a strategy to purify protein fatty
acyltransferase (PAT) activity from rat livers that took advantage of recent knowledge on the cellular location and inhibition of PAT activity.
We determined that three different thiolases have PAT activity in the presence of imidazole and therefore started the purification with a
plasma membrane fraction to minimize the contamination with these enzymes. After detergent extraction of the plasma membrane fraction,
the PAT activity was enriched about 90-fold by sequential chromatography including affinity chromatography to a cerulenin-based inhibitor
of palmitoylation. The partially purified PAT activity (1) was lost with treatments to degrade or denature proteins, (2) could acylate tubulin,
Gai and RGS16 and (3) showed a preference for palmitate and to a lesser degree other long-chain fatty acids. This purification procedure is
a significant advance over previous efforts at PAT purification and a starting point for a proteomic approach for identification of
mammalian PAT.
Published by Elsevier B.V.


Keywords: Palmitoylation; Protein fatty acyltransferase; Heterotrimeric G protein; RGS protein; Tubulin; Liver




1. Introduction glycine residues through a stable amide bond and palmitoy-
lation that occurs on cysteine residues through a reversible
A wide variety of cellular and viral proteins undergo thioester bond. The dynamic nature of palmitoylation makes
posttranslational modification by the covalent attachment of it a critical regulator of G protein signaling, synaptic
fatty acids (reviewed in Refs. [1–4]). Acylation can alter the transmission and other cellular processes.
properties of a protein by changing its protein interactions, In contrast to myristoylation, in which the N-myristoyl
affinity and orientation toward the membrane and targeting transferase is well-characterized, the enzymology of palmi-
within membranes. The two most common forms of acyl- toylation is less clear. A strong candidate for the enzyme
ation are myristoylation that occurs on amino-terminal that removes palmitate is acyl protein thioesterase 1 [5,6], a
member of the a/h hydrolase family [7]. Purification of a
protein fatty acyltransferase (PAT) has proven much more
Abbreviations: CoA, coenzyme A; cpm, counts per minute; PAT, difficult because: (1) the enzymatic activity is labile and lost
protein fatty acyltransferase; RGS16, regulator of G protein signaling 16; early in the purification process [8,9]; (2) in vitro PAT
S.E., standard error of the mean; SDS-PAGE, sodium dodecyl sulfate-
activity has been shown for proteins that are or are likely
polyacrylamide gel electrophoresis
* Corresponding author. Division of Diabetes, Endocrinology and to be involved in fatty acid metabolic pathways. A protein
Metabolic Diseases, NIDDK/NIH, 6707 Democracy Blvd., Room 651, from the silkworm Bombyx mori, p260/270, and its mouse
Bethesda, MD 20892, USA. Tel.: +1-301-435-2996; fax: +1-301-480-3503. homolog, fatty acid synthase, can transfer palmitate to
E-mail address: tlzj@helix.nih.gov (T.L.Z. Jones). synthetic peptides [10] and GAP-43 [11], respectively.
1 Present address: Institut Mediterraneen de Recherche en Nutrition,

Purification of PAT activity toward H-Ras yielded a protein
Laboratoire de Chimie, Biologique Appliquee (UMR A 1111), Faculte des
Sciences et Techniques de St. Jerome Case 431, 13397 Marseille Cedex 20,
that was subsequently determined to be the peroxisomal 3-
France. oxoacyl-CoA thiolase A [12,13]. Thiolases are involved in

1388-1981/$ - see front matter. Published by Elsevier B.V.
doi:10.1016/j.bbalip.2003.10.001

A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19 11


both lipid degradative and biosynthetic pathways; (3) non- volume of 100 Al for 45 min at 30 jC. In some experiments,
enzymatic palmitoylation can occur in vitro and makes palmitoyl CoA was synthesized (see below) and used as the
determination of the enzymatic component of palmitoyla- fatty acid substrate and the immunoprecipitate of Gai,
tion more difficult and has led to the suggestion that purified RGS16 (15 Ag) or microtubule protein purified
autoacylation occurs in vivo [4,14,15]. from bovine brain [22] were used as the protein substrates.
Recently, two reports show the PAT activity of proteins The reaction was stopped by the addition of sample buffer
from Saccharomyces cerevisiae—Akr1p toward casein ki- (Invitrogen) and the proteins were separated by sodium
nase, Yck2p [16] and Erf2p/Erf4p complex toward Ras2p dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-
[17]. In addition, the skinny hedgehog protein in Drosophila PAGE) and stained by MicrowaveBlue (Protiga). The results
melanogaster was found to catalyze amino-terminal palmi- were analyzed by excising the gel bands containing tubulin
toylation of the hedgehog proteins [18]. These studies and dissolving them in 0.5 ml of 30% (v/v) H2O2 at 50 jC
significantly advance our understanding of the enzymology for 2 h in scintillation vials. Ten milliliters of scintillation
of palmitoylation. Both Akr1p and Erf2p contain a con- fluid (CytoScint) was added, and the radioactivity deter-
served Asp–His–His–Cys cysteine-rich domain [19]. This mined by scintillation counting. Background values were
domain may be a useful handle for discovering mammalian obtained by taking the mean counts per minute (cpm) of at
PAT, yet genomic analysis may be inadequate because at least three bands of similar size to the tubulin band that were
least for the Erf2/Erf4 complex, Erf4 is required for PAT cut from areas not showing Coomassie blue staining.
activity but it lacks any mammalian homologs. We Alternatively, the gels were prepared for fluorography by
addressed the problem of identifying a mammalian PAT transferring proteins to nitrocellulose membranes and coated
because a full understanding of palmitoylation requires with EA-wax (EA Biotech Ltd., Scotland) per the manu-
elucidation of the enzyme(s) responsible for the modifica- facturer’s instructions. The membranes were exposed to
tion. We developed a new purification strategy that used a Kodak MS film for 2–5 days at 70 jC.
recently discovered inhibitor of palmitoylation [20] to
partially purify PAT activity to a significantly higher level 2.3. Rat liver plasma membrane isolation
of purity than previous reports [8,9]. With this partially
purified preparation, we evaluated its protein and fatty acid Livers (6–9 g) from male rats CD strain, (Charles River
specificity. Laboratories) were rapidly excised, washed in homogeniza-
tion (H) buffer consisting of 0.25 M sucrose, 5 mM Tris–
HCl pH 7.4 and 1 mM MgCl2, and cut into 2–3 mm pieces.
2. Materials and methods The rat liver pieces were homogenized in five volumes of H
buffer by 10 strokes in a 40 ml Dounce homogenizer. The
2.1. Materials volume of the homogenate was adjusted to 5 ml of H buffer
per gram of liver wet weight and filtered through four layers
3-Ketoacyl-CoA thiolase purified from pig heart [21] of moistened gauze. The filtered homogenate underwent
was a gift from Dr. Horst Schulz (City University of New centrifugation at 300 g for 5 min, the supernatant was
York). Compound 7e was synthesized as described [20]. saved and the homogenization and centrifugation of the
Palmitoyl CoA-agarose, tubulin, fatty acids, acyl CoA pellet was repeated in half the original volume of H buffer.
synthetase and cerulenin were from Sigma-Aldrich Chem- The supernatants were combined and centrifuged at 1500
ical Co. Tubulin was also purified from bovine brain as g for 10 min. The pellet (whole membrane fraction) was
described [22]. Octyl Sepharose and Q-Sepharose were resuspended in a volume equal to two ml of H buffer per g
purchased from Amersham Pharmacia Biotech. The bacu- of liver wet weight and fractionation (F) buffer, 2 M
lovirus containing Gai was a gift from Dr. Tohru Kozasa sucrose, 5 mM Tris–HCl pH 7.4, 1 mM MgCl2, was added
(University of Illinois, Chicago, IL). Purified human wild- to achieve a final sucrose concentration of 1.42 M. The
type RGS16 and the cysteine mutants [23] were kindly suspension was transferred to a nitrocellulose tube and
provided by Dr. Kirk Druey (NIAID, Rockville, MD). overlaid with H buffer. After centrifugation for 1 h at
120,000 g in a Beckman SW28 rotor, the material at the
2.2. PAT activity assay interface of 0.25 and 1.42 M sucrose was suspended in H
buffer in a volume equal to four times the initial pellet
Tubulin was dissolved in the assay buffer, 20 mM Tris– volume and washed by centrifugation at 3000 g for 10
HCl, pH 7.4, 150 mM KCl and 1 mM EDTA, and centri- min. The washed material was resuspended in buffer con-
fuged at 100,000 g for 1 h. The supernatant was diluted in sisting of 1.35 M sucrose, 5 mM Tris–HCl, pH 7.4 and 1
the same buffer for a final protein concentration of 7 mg/ml mM MgCl2, overlaid by H buffer and underwent centrifu-
and 20 Ag were incubated with the membrane preparations gation at 100,000 g for 150 min. The material from the
(10 Ag), 200 AM Coenzyme A (CoA), 2 mM ATP and 5 ACi interface was washed twice in H buffer with centrifugation
of [9,10] [3H]palmitate (American Radiolabeled Chemicals, at 3000 g for 10 min and resuspended in H buffer at 1.5
specific activity 30–60 Ci/mmol) and assay buffer in a final mg protein/ml (plasma membrane fraction). This fraction

12 A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19


showed a 7.3 F 1.7-fold (mean F S.E., three experiments) gation as described above. Triton X-100 was adjusted to
enrichment in Na+/K+ATPase, a plasma membrane protein, 0.15% and this final preparation was partially purified PAT
as determined by immunoblotting and densitometry using a (ppPAT). The enzyme was aliquoted and stored at 80
Umax model UTA-II scanner (Umax Data Systems) and jC for at least 3 months without loss of activity.
Scion Image software (Scion Corporation). Proteins from
this fraction were solubilized by adding Triton X-100 in 20 2.5. Protein expression and PAT activity determination in a
mM Tris–HCl pH 7.4, 150 mM KCl, 1 mM EGTA and 10 reticulocyte lysate system
Ag/ml leupeptin to give a final Triton X-100 concentration
of 0.15% (v/v), followed by gentle agitation for 1 h at 4 jC. The human cytosolic acetoacetyl-CoA thiolase [24] and
The sample was then centrifuged at 100,000 g in a Beck- human mitochondrial acetoacetyl-CoA thiolase [25] were
man TLA-45 rotor for 1 h. The supernatant was diluted with expressed from the full-length cDNAs (gift of Dr. Toshiyuki
the same buffer at 0.8 mg protein/ml and 0.15% Triton X- Fukoa, Gifu University School of Medicine, Japan) using
100 (PATI fraction). the combined TnT transcription and translation reticulocyte
lysate expression system according to the manufacturer
2.4. Preparation of partial purified PAT instructions (Promega). The in vitro expression was per-
formed in the presence or absence of 20 ACi/ml of [35S]me-
The PATI fraction was adjusted to 1 M KCl in equil- thionine (Amersham Pharmacia Biotech, specific activity 10
ibration (E) buffer, 20 mM Tris–HCl pH 7.4, 1 mM Ci/mmol). The proteins were separated by SDS-PAGE and
EGTA, 10 Ag/ml leupeptin and 0.15% (v/v) Triton X- the gels prepared for fluorography as described [26]. The
100, and applied at 0.8 ml/min to an Octyl Sepharose gels were exposed to Kodak MS film for 1–3 days at room
column (5 10 cm) previously equilibrated with E buffer temperature.
containing 1 M KCl. The column was washed with E
buffer without Triton X-100 until no protein was detected 2.6. Preparation of [3H]palmitoyl CoA
by spectrophotometer readings at 280 nm. A linear gradi-
ent in which the KCl concentration varied from 1 to 0 M [3H]palmitoyl CoA was prepared as described [27] with
and simultaneously Triton X-100 from 0.15% to 0.5% in E some modifications. Three millicuries of [9,10-3H] palmi-
buffer was applied. The fractions showing the greatest PAT tate were dried under N2 and resuspended in 99 Al of
activity using tubulin as the substrate were concentrated by ethanol. The solution was incubated for 5 min under
ultrafiltration (Amicon 10 PM filters) to 3 ml and the vigorous agitation at 30 jC with 900 Al of 10 mM sodium
resulting solution was dialyzed against E buffer containing phosphate buffer pH 7.4, 5 mM ATP, 2 mM CoA, 5 mM
150 mM KCl. The dialyzed concentrate was loaded onto MgCl2 and 0.05% Triton X-100. One microliter (2 units) of
the cerulenin analog HiTrap affinity column (1 5 cm) acyl CoA synthetase was added and after a 2-h incubation,
(Amersham Pharmacia Biotech) that had been equilibrated 10 ml of chloroform/methanol (1:1) was added. The
in the same buffer. The column was prepared by first reaction mixture was centrifuged at 2000 g for 10 min.
cross-linking the cerulenin analog inhibitor of palmitoyla- The supernatant was taken and separated in two phases by
tion, compound 7e, [20] to N-hydroxysuccinimide at high the addition of 5 ml of chloroform and 2.5 ml of water. The
pH and then on the HiTrap column following the manu- pH of the aqueous phase containing [3H]palmitoyl CoA
facturer’s instructions. The column was washed with E was decreased to 6.2 with 10 mM sodium dihydrogen
buffer and the active fractions were eluted by a linear phosphate and the solution was centrifuged at 2000 g at
gradient of palmitoyl CoA (0–50 mM) in E buffer. The 4 jC for 30 min. Palmitoyl CoA was added to the aqueous
fractions showing PAT activity were collected and dialyzed phase for a specific activity of about 3000 cpm/pmol.
against E buffer alone overnight. The sample was loaded Aliquots of 200 Al were stored at 80 jC. The radio-
onto a Q-Sepharose column (5 10 cm) that had been chemical purity was >95% as determined by thin layer
equilibrated with E buffer. The column was washed and chromatography [28].
PAT activity was eluted by a linear gradient of KCl from 0
to1 M in E buffer. The fractions were kept overnight at 4 2.7. Miscellaneous
jC and then the PAT assay was performed. Fractions
containing PAT activity were pooled and concentrated to SDS/PAGE was performed on 10% or 4–20% acrylam-
0.5 ml by ultrafiltration as described above. The sample ide gels (Invitrogen). Immunoblotting was performed with
was applied by gravity flow to a column (0.8 5 cm) of rabbit antibodies raised against thiolase A, a gift from Dr.
palmitoyl CoA-agarose pre-equilibrated with E buffer Paul P. Van Veldhoven (University of Leuven, Belgium)
containing 150 mM KCl. The column was washed one [29] or Na+/K+ATPase (Biomol Research Laboratories),
time with 10 mM palmitoyl CoA in E buffer. One milliliter after transferring the proteins from the gels to nitrocellulose
fractions were eluted by a step gradient of 20 mM membranes. Blots were developed by electrochemilumines-
palmitoyl CoA in E buffer. The fractions containing PAT cence (ECL) detection reagents according to the protocol
activity were combined and concentrated by ultracentrifu- supplied by the manufacturer (Amersham Pharmacia Bio-

A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19 13


3. Results and discussion

3.1. Determination of PAT activity

In order to purify a PAT, we developed an assay that
measured the cell-free incorporation of [3H]palmitate into
proteins. Purified tubulin was incubated with either the
untreated or boiled plasma membrane fraction from rat liver
and either [3H]palmitate or [3H]palmitoyl CoA. We obtained
similar results using [3H]palmitoyl CoA prepared from
purified fatty acyl-CoA synthetase (Fig. 1D) or with
[3H]palmitate and including CoA and ATP in the buffer to
allow synthesis of [3H]palmitoyl CoA from the endogenous
acyl-CoA synthetase (Fig. 1A,C). Under these conditions, a




Fig. 1. Protein palmitoylation of tubulin by a rat liver plasma membrane
extract. (A) The plasma membrane (PM) fraction from rat liver (15 Ag),
either untreated or after boiling (PMb) by heating to 95 jC for 5 min, was
incubated with tubulin and [3H]palmitate in a reaction buffer, as described
in Materials and methods, for 45 min at 30 jC. The proteins were separated
by SDS-PAGE and prepared for fluorography at 70 jC for 2 days. The
molecular weight markers in kilodaltons are to the left. (B, C and D) The
PAT assay was performed as described above except that the reaction was
stopped after 0, 10, 20, 30, 40, 60 min and [3H]palmitoyl CoA was used in
D. As a control for the level of nonenzymatic palmitoylation, a boiled
plasma membrane sample (60b) was used. The gels were stained with
Coomassie brilliant blue (B), prepared for fluorography as described above
(C) or stained and the tubulin bands excised and prepared for scintillation
counting as described in Materials and methods (D). The background
Fig. 2. PAT activity and cellular location of thiolase proteins. (A) Two
counts were determined by performing scintillation counting on gel bands
micrograms of purified 3-keto-CoA thiolase from pig heart or ppPAT (4 Ag)
of similar size from areas of the gel not containing Coomassie blue staining. were incubated with tubulin and [3H]palmitate in the PAT assay buffer for
Shown are the net values after subtraction of the background. The filled 45 min in the absence ( ) or the presence (+) of 5 mM imidazole. The
circles show the mean results ( F S.E.) using untreated membranes and the
proteins were separated by SDS-PAGE and prepared for fluorography. (B)
open circles using boiled membranes from three separate experiments. (E)
Thiolases were expressed in a reticulocyte lysate system by adding the
The PAT assay was performed in duplicate as described above except that
cDNAs for mitochondrial thiolase (Tm) or cytosolic thiolase (Tc) or no
after SDS-PAGE, the gels were incubated in 1 M Tris–HCl, pH 7.5 ( ) or cDNA as the control. Proteins were metabolically labelled with [35S]me-
1 M hydroxylamine, pH 7.5 (+) for 1 h, washed and then prepared for
thionine and 2 Al of the lysate mixture was separated by SDS-PAGE and
fluorography.
prepared for fluorography. (C) The PAT activity was determined with 2 Al of
the reticulocyte lysate reactions expressing the thiolases, RGS16 (15 Ag)

tech). The protein concentrations were determined by the and [3H]palmitate in the PAT assay buffer with and without 5 mM
imidazole. (D) The PAT activity was determined with RGS16 (15 Ag) and
Bio-Rad DCk (detergent compatible) protein assay with [3H]palmitoyl CoA with and without 5 mM imidazole and with buffer alone
immunoglobulin G as the standard. Maintenance and infec- or with the reticulocyte lysate mixture expressing the cytosolic thiolase or
tion of Sf9 cells with a baculovirus containing the cDNA for without DNA (2 Al). (E) Ten micrograms of protein from each membrane
Gai [30] and the immunoprecipitation of Gai from the fraction from rat liver was separated by SDS-PAGE and underwent

cytosol of the infected Sf9 cells was performed as described immunoblotting with an antibody raised against thiolase A and enhanced
chemiluminescence for detection. Lane 1, whole membrane; lane 2, crude
previously [26,31]. The crude microsomal and mitochon-
microsomal membrane; lane 3, mitochondrial/peroxisomal membrane; lane
drial/peroxisomal membrane fractions were isolated as de- 4, plasma membrane. The molecular weight markers in kilodaltons are
scribed previously [22]. shown to the left.

14 A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19


3.2. Thiolases have PAT activity in the presence of imidazole

A previous attempt at purification of PAT isolated
peroxisomal 3-oxoacyl-CoA thiolase A [12], an enzyme
that was dependent on the presence of imidazole for PAT
activity (Gelb, M.H., personal communication). We tested
whether other thiolases had PAT activity and its depen-
dence on imidazole because the presence of these enzymes
could complicate purification of PAT. Incubation of puri-
fied 3-ketoacyl-CoA thiolase from pig heart with tubulin in
the reaction buffer led to a significant increase in [3H]pal-
mitate incorporation into tubulin only in the presence of
imidazole (Fig. 2A). We also expressed a cytosolic ace-
toacetyl-CoA thiolase, Tc [24], and a mitochondrial ace-
toacetyl-CoA thiolase, Tm [25], in a coupled transcription
and translation rabbit reticulocyte lysate system. Both of
these enzymes were expressed as seen by the incorporation
of [35S]methionine into protein bands that migrate at about
41 kDa in lysates containing the cDNAs (Fig. 2B).
Incubation of the lysates expressing the thiolases in the
PAT assay buffer led to incorporation of [3H]palmitate into
the regulator of G protein signaling (RGS16) that was
increased in the presence of imidazole (Fig. 2C). The




Fig. 3. Purification of ppPAT. (A) Chromatography elution profile of the
Triton X-100 extract of plasma membranes applied to an Octyl Sepharose
column and eluted with decreasing concentrations of KCl. (B) Chromatog-
raphy elution profile from a column with compound 7e, a cerulenin analog
inhibitor of protein palmitoylation, that had been cross-linked to a HiTrap
column. Open diamond, the spectrophotometric reading at an optical
density of 280 nm (OD280); filled diamond, KCl concentration; filled
square, [3H]palmitate incorporation into tubulin. The PAT assay was
performed by using 10–30 Al of the column fractions.



small amount of tritium was incorporated into tubulin in the
presence of the boiled plasma membrane fractions and
represents nonenzymatic palmitoylation (Fig. 1A,C). In
contrast, the untreated plasma membrane fractions signifi-
cantly increased the incorporation of tritium into tubulin in a
time-dependent fashion (Fig. 1B,C). In order to quantify the
[3H]palmitate incorporation, the bands containing tubulin
were cut out of the gels after SDS-PAGE and the tritium
incorporation determined by scintillation counting. This Fig. 4. Inhibition of protein palmitoylation. The PAT assay was performed
in the absence or presence of the indicated concentrations of cerulenin or
method shows a good correlation with the intensity of the
compound 7e, a derivative of cerulenin and in (A) a rat liver membrane
bands after fluorography (Fig. 1D). The tritium signal can preparation [22] and [3H]palmitate and in (B) boiled membranes,
be removed from the protein band with hydroxylamine [3H]palmitoyl CoA and as indicated 100 AM compound 7e. The proteins
showing the thioester linkage of [3H]palmitate to the protein were separated by SDS-PAGE and the gels prepared for fluorography and

(Fig. 1E). Previous work using a similar assay has shown exposure to film at 70 jC for 1 day. The molecular weight marker in
kilodaltons is shown to the left. In A, the two bands are the a and h forms
that the tritium signal is due to tritiated palmitate [22]. This
of tubulin that were separated by SDS-PAGE in this experiment by the use
assay gives us a practical and accurate method to determine of higher concentrations of running buffer solute, 50 mM Tris and 0.4 M
PAT activity during the purification. glycine.

A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19 15


Table 1 nal sections [24]. A similar reaction may take place for
Purification of PAT activity from rat liver protein acylation in which an acyl-enzyme intermediate
Step Total Total Recovery Specific Fold forms to facilitate the transfer to the sulfhydryl group of
protein activity (%) activity the cysteine on the protein substrate. The function of
(mg) (U/mg)
imidazole in promoting this reaction for thiolases was
Plasma membrane 1750 4200 100 2.4 1 not solved in this work. Nonetheless, the minimal PAT
extract (PATI) 1650a 3630 100 2.2 1
activity without imidazole and the cellular location of
Octyl Sepharose 69 1918 46 27.8 12
(PATII) thiolases to peroxisomes, mitochondria and cytosol make
Inhibitor affinity 22 959 23 43.6 18 them unlikely candidates for the physiologic PAT. These
(PATIII) results also show the feasibility of expressing PAT candi-
Q-Sepharose 9 522 12.4 58 24 dates in a reticulocyte lysate system and determining their
(PATIV)
PAT activity.
Palmitoyl CoA- 0.7 160 3.8 228 95
Agarose (ppPAT) 1.2a 226 6.2 188 85
3.3. Purification of PAT from rat livers
The PAT activity in each step was assayed under standard conditions as
described in Materials and methods using tubulin as the substrate and
subtracting the background that was determined by the [3H]palmitate Our purification protocol was considerably different than
incorporation into tubulin in the absence of membranes. previous methods and took advantage of new knowledge
a The values in italics below the dashed lines are the mean results from
and reagents for protein palmitoylation. We started with a
two purifications using [3H]palmitoyl CoA instead of [3H]palmitate.
plasma membrane fraction from rat livers because PAT
activity toward G proteins is enriched in the plasma mem-
lysate mixture without DNA or the buffer alone showed a brane fraction [8]. The plasma membrane fraction is also
trivial amount of PAT activity with and without imidazole relatively free of thiolases, which are located in the cytosol
(Fig. 2D). The [3H]palmitate incorporation into RGS16 in and mitochondria. We did not detect thiolase A in the
the absence of added imidazole may be due to the presence plasma membrane by immunoblotting (Fig. 2E).
of residual imidazole from the purification of RGS16. We used a low level of Triton X-100 (0.15%) for
These results indicate that the PAT activity of thiolases detergent extraction of the plasma membrane because we
in the presence of imidazole may be a general property of obtained a relatively high level of extraction of PAT activity
these enzymes. Thiolases catalyze a number of different without the problems of protein detection during chroma-
reactions, but they all form an acyl-S-enzyme intermediate tography or PAT activity inhibition that we found with 1%
through a conserved cysteine residue in their amino-termi- Triton X-100 and as has been reported by others [32]. The




Fig. 5. Protein staining and PAT activity of ppPAT. (A) The PAT activity was determined on the initial detergent extract from plasma membrane (PATI) and the
final preparation (ppPAT) using 10 Ag of the membrane preparations, tubulin, and [3H]palmitate. (B,C) 10 Ag of the PATI and ppPAT fraction for B and 15 Ag
of the ppPAT fraction for C were separated by SDS-PAGE on 4–20% gradient acrylamide gels and stained with Coomassie brilliant blue. The molecular
weight marker in kilodaltons are shown to the left.

16 A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19


detergent extract of the plasma membrane fraction was
applied to the Octyl Sepharose column and eluted with a
KCl linear gradient (Fig. 3A). The peak PAT activity was
reproducible eluted at about 300 mM KCl. The fractions
enriched in PAT activity were concentrated and applied to a
column in which a cerulenin-based inhibitor of PAT, com-
pound 7e, [20] had been cross-linked on the HiTrap affinity
column and the proteins eluted with palmitoyl CoA (Fig.
3B). The site of attachment to the column matrix is likely to
be the amide group that is opposite to the acyl chain on
compound 7e [20]. This compound can be considered the
palmitoyl analog of cerulenin because the octadienyl side
chain of cerulenin was substituted with a saturated palmitoyl
side chain. Compared to a series of cerulenin analogues, it
was the most active in inhibiting palmitoylation [20] and
inhibited [3H]palmitate incorporation into tubulin with a
half-maximal dose of about 10 AM (Fig. 4A). Compound 7e
did not inhibit the nonenzymatic incorporation of [3H]pal-
mitate into tubulin (Fig. 4B). Despite its relatively low
potency at inhibiting palmitoylation of tubulin in vitro, this
chromatography step enriched the PAT activity.
The PAT activity was further purified by sequential
chromatography steps using Q-Sepharose and palmitoyl-
CoA agarose columns. The latter is a new approach to
attempt to isolate the enzyme based on its substrate speci-
ficity. Nonspecific interactions based on the hydrophobicity
of palmitoyl CoA and compound 7e could also be a factor in
the purification. We found a loss of PAT activity if we
measured PAT activity directly after concentrating samples
from Q-Sepharose chromatography. The PAT activity was
restored by storing the fractions at 4 jC for about 12 h. This Fig. 6. PAT activity of ppPAT was enzymatic. (A,B) The PAT assay was
performed with ppPAT (10 Ag), tubulin and [3H]palmitate or [3H]palmitoyl
observation suggests the possibility that time was needed for
CoA under the indicated conditions and the proteins separated by SDS-
PAT to undergo conformational changes or re-association
PAGE. The gels were (B) prepared for fluorography with exposure to film
with another protein or cofactor. With this purification at 70 jC for 2 days or (A) stained with Coomassie blue and the
method, we achieve about a 90-fold purification of PAT radioactivity in the tubulin band determined. The results shown are the

(Table 1) (Fig. 5A). A limited number of proteins were mean F S.E. from three separate experiments. Con, all components; no
ppPAT, without the membrane preparation; B, ppPAT was heated at 95 jC
detected by Coomassie blue staining of ppPAT after sepa-
for 5 min; S, ppPAT was pretreated with 0.05% SDS for 30 min at 4 jC; T,
ration on SDS-PAGE (Fig. 5B,C). Further attempts at
ppPAT was pretreated with 3.2 mg/ml of trypsin for 1 h at 30 jC and then
purification on gel filtration columns were unsuccessful soybean trypsin inhibitor was added at 10 mg/ml; T + STI, ppPAT was
because of an irreversible loss of activity. pretreated with the above concentrations of trypsin and soybean trypsin
inhibitor added together for 1 h at 30 jC; no tub., without tubulin. The filled
bars are with [3H]palmitate and the open bars are with [3H]palmitoyl CoA
3.4. PAT activity is enzymatic
(C). The PAT assay was performed with tubulin and [3H]palmitoyl CoA and
the indicated amounts of ppPAT as described in the legend to Fig. 1. Shown
Protein palmitoylation can occur under the appropriate are the mean results ( F S.E.) using untreated ppPAT (filled circles) and
conditions in the presence of palmitoyl CoA alone, without boiled ppPAT (open circles) from three separate experiments.

a source of enzymes, because the sulfhydryl group on
cysteine is a good nucleophile [4,14,15]. We tested whether of ppPAT (Fig. 6C). We were also concerned that the PAT
the PAT activity of this partially purified preparation from activity seen with this preparation may be due to increased
rat liver was enzymatic by determining the PAT activity after enzymatic synthesis of palmitoyl CoA. However, we saw
treatments that would destroy protein activity. Treatment similar results with these experiments when we used
with heat, a strong ionic detergent and trypsin all signifi- [3H]palmitate or [3H]palmitoyl CoA (Fig. 6A). These results
cantly reduced PAT activity (Fig. 6A,B). Soybean trypsin indicate that the PAT activity seen with this partially purified
inhibitor could partially block the effect of trypsin when preparation was likely to be enzymatic and are consistent
they were added together (Fig. 6A) and had no effect on with kinetic studies showing that nonenzymatic palmitoyla-
PAT activity when added alone (data not shown). In addi- tion is too slow for the rapid thioacylation cycle in the cell
tion, PAT activity increased with increasing concentrations [33]. The recent discovery of yeast proteins with PAT

A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19 17


activity also points to palmitoylation as an enzymatic
process in the cell [16,17].

3.5. Protein and fatty acid specificity of PAT activity

We then tested the specificity of the PAT activity for
different protein and fatty acid substrates. In addition to
tubulin, ppPAT showed PAT activity towards two other
protein substrates, the Ga protein, Gai and RGS16. For
Gai, the PAT activity of ppPAT was determined with Gai
after expression in the cytosol of Sf9 cells and immuno-
Fig. 8. The fatty acid specificity of ppPAT. The PAT assay was performed
precipitation. The incorporation of tritiated palmitate into with ppPAT (4 Ag), tubulin, [3H]palmitate with and without 30 AM of fatty
Gai was only seen for Gai in its native state. Denaturing acids C8, caprylate; C12, laurate; C14, myristate; C16, palmitate; C18,
Gai or ppPAT by boiling led to a loss of PAT activity (Fig. stearate; C18:1, oleate; C20, arachidate. The palmitate incorporation was
7A). Human RGS16 has seven cysteine residues, but Cys- determined as described in Materials and methods. Shown is one

2, Cys-12 and Cys-98 are the likely sites for palmitoyla- representative experiment from three experiments performed in duplicate.

tion. Cys-2 and Cys-12 are critical for palmitate incorpo-
ration within the cell [34] and a proteolytic cleavage study
showed [3H]palmitate incorporation into Cys-98 [23]. an accurate terminology for this modification because other
Incubation of ppPAT in the reaction buffer with purified, long-chain fatty acids including 14:0, 18:0, 18:1 and 18:2
recombinant, wild-type RGS16 expressed in bacteria have been detected on heart, liver and brain proteins after
showed incorporation of [3H]palmitate (Fig. 7B). Mutation base hydrolysis [35,36] and on specific proteins using
of Cys-2 and Cys-12 in RGS16 led to a marked reduction metabolic labeling of radiolabeled fatty acids [1]. However,
in [3H]palmitate incorporation and a smaller reduction for the fatty acids added to proteins do not just reflect the acyl
mutation of Cys-98 alone. This result indicates that ppPAT CoA pool in the cell, but show cell-type specificity [36,37].
showed specificity for cysteine residues that are critical for The most likely fatty acid substrate for these and the
in vivo palmitoylation. other experiments in this report is the acyl-CoA that is
We also investigated the specificity of the fatty acid catalyzed by acyl-CoA synthetase from the fatty acid, CoA
substrate of ppPAT by testing the ability of fatty acids with and ATP. We have obtained only a partial purification of
different chain lengths to compete with [3H]palmitate for PAT, so the most likely reason that we see tritium incorpo-
incorporation into tubulin. Addition of C16 palmitate ration into proteins using [3H]palmitate is that our partially
showed the greatest inhibition of [3H]palmitate incorpora- purified material is contaminated with a small amount of
tion with some inhibition for the other fatty acids except for acyl-CoA synthetase activity. Acyl-CoA synthetase has
C8, caprylate (Fig. 8). Results from studies on other similar properties to ppPAT in that it is found in the
membrane preparations enriched in PAT activity [8,9] and particulate fraction and solubilized with Triton X-100
the yeast Erf2p/Erf4p complex also show the same pattern [38]. The enzyme is stable and active at a pH used in our
of fatty acid incorporation [17]. Protein palmitoylation is not assay conditions [38]. An isoform of acyl-CoA synthetase,




Fig. 7. PAT activity of ppPAT toward Gai and RGS16. (A) The PAT assay was performed with [3H]palmitate, ppPAT (4 Ag) and Gai immunoprecipitated from
400 Ag of the cytosol of Sf9 cells that had been infected with a baculovirus containing the cDNA for Gai. As indicated, ppPAT or the Gai immunoprecipitate
were heated to 95 jC for 5 min ( + b). The arrowhead points to Gai. (B) The PAT assay was performed with ppPAT (4 Ag), [3H]palmitate and 15 Ag of either the
wild-type (WT) or cysteine mutants of RGS16, cysteines at residues 2 and 12 mutated to alanine (C2/12A) or cysteine 98 mutated to alanine (C98A) that had
been expressed and purified from bacteria. The proteins were separated by SDS-PAGE and the gels prepared for fluorography.

18 A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19


ACS1, is an intrinsic membrane protein found in rat liver in [7] Y. Devedjiev, Z. Dauter, S.R. Kuznetsov, T.L.Z. Jones, Z.S. Dere-

microsomal and plasma membrane fractions [39]. ACS1 is wenda, Crystal structure of the human acyl protein thioesterase I
from a single X-ray data set to 1.5, Struct. Fold. Des. 8 (2000)
likely to be abundant in liver because of evidence that it
1137–1146.
provides acyl-CoA for the synthesis of triacylglycerol. The [8] J.T. Dunphy, W.K. Greentree, C.L. Manahan, M.E. Linder, G-pro-
partition of the different isoforms of acyl-CoA synthetase to tein palmitoyltransferase activity is enriched in plasma membranes,
different sites in the cell with overlapping and separate J. Biol. Chem. 271 (1996) 7154–7159.

functions has led to the proposal that acyl-CoA synthetase [9] L. Berthiaume, M.D. Resh, Biochemical characterization of a palmi-
toyl acyltransferase activity that palmitoylates myristoylated proteins,
may directly interact with specific acyltransferases [39].
J. Biol. Chem. 270 (1995) 22399–22405.
[10] K. Ueno, Y. Suzuki, p260/270 expressed in embryonic abdominal leg
3.6. Concluding remarks cells of Bombyx mori can transfer palmitate to peptides, J. Biol.
Chem. 272 (1997) 13519–13526.

The recent discoveries of PAT in S. cerevisiae [16,17] are [11] K. Ueno, Involvement of fatty acid synthase in axonal development in
mouse embryos, Genes Cells 5 (2000) 859–869.
a significant advance in our understanding of the enzymol-
[12] L. Liu, T. Dudler, M.H. Gelb, Purification of a protein palmitoyltrans-
ogy of palmitoylation. Both Akr1p and Erf2p contain a ferase that acts on H-Ras protein and on a C-terminal N-Ras peptide
conserved Asp–His–His–Cys cysteine-rich domain [19]. (correction), J. Biol. Chem. 274 (1999) 3252.
This domain may be a useful handle for discovering [13] L. Liu, T. Dudler, M.H. Gelb, Purification of a protein palmitoyltrans-

mammalian PAT, but genomic analysis alone may be insuf- ferase that acts on H-Ras protein and on a C-terminal N-Ras peptide,
J. Biol. Chem. 271 (1996) 23269–23276.
ficient to characterize the mammalian PAT because, for
[14] M.C. Bano, C.S. Jackson, A.I. Magee, Pseudo-enzymatic S-acylation
example, the yeast protein Erf4 does not have a mammalian of a myristoylated yes protein tyrosine kinase peptide in vitro may
homolog. We report here the partial purification of PAT reflect non-enzymatic S-acylation in vivo, Biochem. J. 330 (Pt 2)
activity from rat liver using novel purification steps and (1998) 723–731.

characterization of its activity. In the future, this purification [15] O.A. Bizzozero, J.F. McGarry, M.B. Lees, Autoacylation of myelin
proteolipid protein with acyl coenzyme A, J. Biol. Chem. 262 (1987)
could be a useful starting point for a proteomic investigation
13550–13557.
of PAT. A combination of genomic, proteomic and classic [16] A.F. Roth, Y. Feng, L. Chen, N.G. Davis, The yeast DHHC cysteine-
biochemical approaches will likely be necessary to fully rich domain protein Akr1p is a palmitoyl transferase, J. Cell Biol. 159
characterize the enzymology of the elusive PAT. (2002) 23–28.
[17] S. Lobo, W.K. Greentree, M.E. Linder, R.J. Deschenes, Identification
of a Ras palmitoyltransferase in Saccharomyces cerevisiae, J. Biol.
Chem. 277 (2002) 41268–41273.
Acknowledgements
[18] Z. Chamoun, R.K. Mann, D. Nellen, D.P. von Kessler, M. Bellotto,
P.A. Beachy, K. Basler, Skinny hedgehog an acyltransferase required
We thank Dr. Samuel W. Cushman for the rat livers, Dr. for palmitoylation and activity of the hedgehog signal, Science 293
Paul P. Van Veldhoven for the thiolase antibody, Dr. Horst (2001) 2080–2084.

Schulz for the purified thiolase, Dr. Toshiyuki Fukua for the [19] T. Putilina, P. Wong, S. Gentleman, The DHHC domain: a new
highly conserved cysteine-rich motif, Mol. Cell. Biochem. 195
thiolase cDNAs; Dr. Tohru Kosaza for the Gai baculovi-
(1999) 219–226.
ruses and Dr. Kirk M. Druey for purified RGS16. This work [20] D.S. Lawrence, J.T. Zilfou, C.D. Smith, Structure–activity studies of
was supported in part by grants from Lea’s Foundation for cerulenin analogues as protein palmitoylation inhibitors, J. Med.
Leukemia Research, Inc., and the American Heart Associ- Chem. 42 (1999) 4932–4941.

ation, Heritage Affiliate, to J.M.C. [21] H. Schulz, H. Staack, 3-Ketoacyl-CoA-thiolase with broad chain
length specificity from pig heart muscle, Methods Enzymol. 71
(Pt C) (1981) 398–403.
[22] J.M. Caron, Posttranslational modification of tubulin by palmitoy-
References lation: I. In vivo and cell-free studies, Mol. Biol. Cell 8 (1997)
621–636.
[1] M.D. Resh, Fatty acylation of proteins: new insights into membrane [23] A. Hiol, P.C. Davey, A.A. Waheed, J.C. Osterhout, E. Fischer,
targeting of myristoylated and palmitoylated proteins, Biochim. Bio- C.-K. Chen, G. Milligan, K.M. Druey, T.L.Z. Jones, Palmitoyla-
phys. Acta 1451 (1999) 1–16. tion and RGS16 signaling: I. Mutation of amino-terminal cysteine
[2] M.E. Linder, R.J. Deschenes, New insights into the mechanisms of residues on RGS16 prevents targeting to lipid rafts and palmitoy-
protein palmitoylation, Biochemistry 42 (2003) 4311–4320. lation of an internal cysteine residue, J. Biol. Chem. 278 (2003)
[3] M.F.G. Schmidt, Fatty acylation of proteins, Biochim. Biophys. Acta 19301–19308.
988 (1989) 411–426. [24] X.Q. Song, T. Fukao, S. Yamaguchi, S. Miyazawa, T. Hashimoto, T.
[4] R. Qanbar, M. Bouvier, Role of palmitoylation/depalmitoylation re- Orii, Molecular cloning and nucleotide sequence of complementary
actions in G-protein-coupled receptor function, Pharmacol. Ther. 97 DNA for human hepatic cytosolic acetoacetyl-coenzyme A thiolase,
(2003) 1–33. Biochem. Biophys. Res. Commun. 201 (1994) 478–485.
[5] J.A. Duncan, A.G. Gilman, A cytoplasmic acyl-protein thioesterase [25] M. Kano, T. Fukao, S. Yamaguchi, T. Orii, T. Osumi, T. Hashimoto,
that removes palmitate from G protein alpha subunits and p21(RAS), Structure and expression of the human mitochondrial acetoacetyl-
J. Biol. Chem. 273 (1998) 15830–15887. CoA thiolase-encoding gene, Gene 109 (1991) 285–290.
[6] J.A. Duncan, A.G. Gilman, Characterization of Saccharomyces cere- [26] T.L.Z. Jones, W.F. Simonds , J.J. Merendino Jr., M.R. Brann, A.M.
visiae acyl-protein thioesterase 1, the enzyme responsible for G pro- Spiegel, Myristoylation of an inhibitory GTP-binding protein alpha
tein alpha subunit deacylation in vivo, J. Biol. Chem. 277 (2002) subunit is essential for its membrane attachment, Proc. Natl. Acad.
31740–31752. Sci. U. S. A. 87 1990, pp. 568–572.

A. Hiol et al. / Biochimica et Biophysica Acta 1635 (2003) 10–19 19


[27] J.A. Duncan, A.G. Gilman, Autoacylation of G protein alpha subu- peptide sequences by long-chain acyl-CoAs, Biochemistry 36 (1997)
nits, J. Biol. Chem. 271 (1996) 23594–23600. 5546–5553.
[28] A.K. Hajra, J.E. Bishop, Preparation of radioactive acyl coenzyme A, [34] K.M. Druey, O. Ugur, J.M. Caron, C.K. Chen, P.S. Backlund, T.L.Z.
Methods Enzymol. 122 (1986) 50–53. Jones, Amino-terminal cysteine residues of RGS16 are required for
[29] V.D. Antonenkov, P.P. Van Veldhoven, E. Waelkens, G.P. Mannaerts, palmitoylation and modulation of Gi and Gq-mediated signaling, J.
Substrate specificities of 3-oxoacyl-CoA thiolase A and sterol carrier Biol. Chem. 274 (1999) 18836–18842.
protein 2/3-oxoacyl-CoA thiolase purified from normal rat liver pero- [35] J.C. De Mar Jr., R.E. Anderson, Identification and quantitation of the
xisomes. Sterol carrier protein 2/3-oxoacyl-CoA thiolase is involved fatty acids composing the CoA ester pool of bovine retina, heart, and
in the metabolism of 2-methyl-branched fatty acids and bile acid liver, J. Biol. Chem. 272 (1997) 31362–31368.
intermediates, J. Biol. Chem. 272 (1997) 26023–26031. [36] O.A. Bizzozero, M.B. Lees, Fatty acid composition of myelin proteo-
[30] T. Kozasa, A.G. Gilman, Purification of recombinant G proteins lipid protein during vertebrate evolution, Neurochem. Res. 24 (1999)
from Sf9 cells by hexahistidine tagging of associated subunits. 269–274.
Characterization of alpha 12 and inhibition of adenylyl cyclase by [37] H. Reverey, M. Veit, E. Ponimaskin, M.F. Schmidt, Differential fatty
alpha z, J. Biol. Chem. 270 (1995) 1734–1741. acid selection during biosynthetic S-acylation of a transmembrane
[31] T.L.Z. Jones, C. Woodard, A.M. Spiegel, Baculovirus expression and protein (HEF) and other proteins in insect cells (Sf9) and in mamma-
purification of a soluble, mutant G-protein alpha subunit, Protein lian cells (CV1), J. Biol. Chem. 271 (1996) 23607–23610.
Expr. Purif. 4 (1993) 64–71. [38] T. Tanaka, K. Hosaka, M. Hoshimaru, S. Numa, Purification and
[32] M.F.G. Schmidt, R.A. McIlhinney, G.R. Burns, Palmitoylation of properties of long-chain acyl-coenzyme-A synthetase from rat liver,
endogenous and viral acceptor proteins by fatty acyltransferase Eur. J. Biochem. 98 (1979) 165–172.
(PAT) present in erythrocyte ghosts and in placental membranes, [39] R.A. Coleman, T.M. Lewin, C.G. Van Horn, M.R. Gonzalez-Baro, Do
Biochim. Biophys. Acta 1257 (1995) 205–213. long-chain acyl-CoA synthetases regulate fatty acid entry into syn-
[33] R. Leventis, G. Juel, J.K. Knudsen, J.R. Silvius, Acyl-CoA binding thetic versus degradative pathways? J. Nutr. 132 (2002) 2123–2126.
proteins inhibit the nonenzymic S-acylation of cysteinyl-containing


Protein Expression and Purication 34 (2004) 208–214

www.elsevier.com/locate/yprep




Purication of a plant cell wall bronectin-like adhesion
protein involved in plant response to salt stress

Delphine Pellenc, Eric Schmitt, and Olivier Gallet*

ERRMECe, Universite de Cergy-Pontoise, 2 avenue Adolphe Chauvin, B.P. 222, Pontoise, 95302 Cergy-Pontoise cedex, France

Received 26 September 2003, and in revised form 14 November 2003




Abstract


The structural role of extracellular-matrix (ECM) has been recognized in both plants and animals as a support and anchorage-
inducing cell behavior. Unlike the animal ECM proteins, the proteins that have been identied in plant ECM have not yet been
puried from whole plants and cell wall. As several immunological data indicate the presence of animal ECM-like proteins in plants
cell wall, especially under salt stress or water decit, we propose a protocol to purify a bronectin-like protein from the cell wall of
epicotyls of young germinating peas. The process consists of a combination of gelatin and heparin anity chromatography, close to
the classical one used for human blood plasma bronectin purication. Proteins with anity for gelatin and heparin, immuno-
logically related to human bronectin, are found in the cell wall of epicotyls grown under salt stress or not. Total amount of puried
proteins is 3–4 times more enriched in salt stressed epicotyls. SDS–PAGE and Western blot with antibodies directed against human
blood plasma bronectin give evidence that the cell wall proteins puried by gelatin/heparin anity chromatography are closely
related to human bronectin. The present protocol leads us to purify 17 (control) or 65 (salt stress) micrograms of protein per g of
fresh starting material. Our results suggest that plant cell wall proteins can provide better anchorage of the cell to its cell-wall during
salt stress or water decit and could be considered not only as cell adhesion but also as signaling molecules.
2003 Elsevier Inc. All rights reserved.


Keywords: Plant cell wall; Plant signaling; Fibronectin-like proteins; Fibronectin purication; ECM; Purication; Anity-chromatography




Introduction von Willebrand factor, and osteopontin [1]. Fibronectin
(FbN) is a family of high molecular weight glycopro-
Animal cell adhesion to the ECM1 is mediated by teins (450–500 kDa) with two disulde-bonded chains [2]
cell-surface receptors, interacting with cell proteins found in mammalian connective tissues under brillar
bound in the matrix support. Mammalian cells also forms or in complexes with other macro proteins such as
possess dierent kinds of receptors at their membrane brillar collagens. FbN chains are composed of domains
level which can recognize either carbohydrates or pro- which interact with various biological molecules such as
teins inside the ECM or close to other cells. The role of plasma brin, collagens, and glycosaminoglycans of the
animal ECM glycoproteins in the anchoring of the cell extracellular matrix, and specic receptors from the cell
to its matrix support has been well characterized for membrane called integrin complexes [3,4]. Due to its
several proteins such as bronectin (FbN), vitronectin, anity for several ligands, mammalian FbN is thought
collagens, laminin, brinogen, thrombospodin, to play an important role, not only in cell adhesion, but
also in other cell and extracellular matrix behavior like
cell spreading [5]. Cell adhesion to matrix proteins im-
* Corresponding author. Fax: +33-1-34-25-65-52. plies dierent domains but the well-known attachment
E-mail address: olivier.gallet@bio.u-cergy.fr (O. Gallet). site is the tripeptide RGD, a motif often conserved in
1 Abbreviations used: FbN, bronectin; SDS–PAGE, sodium matrix proteins, to get selective cell–matrix adhesion. At
dodecyl sulfate polyacrylamide gel electrophoresis; ECM, extracellular
the cell plasma membrane, some integrins mediate such
matrix; EDTA, ethylenediaminetetraacetic acid tetrasodium salt;
PMSF, phenylmethylsulfonyl uoride; PVDF, polyvinylidene adhesion by the RGD binding domains. Others bind to
diuoride. ligands via non-RGD binding domain like a4b1 that

1046-5928/$ - see front matter 2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.pep.2003.11.011

D. Pellenc et al. / Protein Expression and Purication 34 (2004) 208–214 209


selectively recognized the LDV sequence on connective Plant material
segment 1 of FbN [3]. Adhesive interactions between cells
and the surrounding ECM play a major role in the con- Bean seeds of Pisum sativum L. cv Primavil (Vilmo-
struction of multicellular organisms during each devel- rin, France) were grown at 20 °C in darkness in ver-
opment stage. Adhesion proteins not only have a function miculite irrigated with either water (control) or 150 mM
in the attachment but also contribute to cell signaling via NaCl (salt stress). Epicotyls were excised from 10-day-
internal transduction pathways [6]. Changes in cell ad- old plants and kept on ice for further protein extraction.
hesive interactions are well known in dierent human
pathologies involving cell dissemination, attachment to Solubilization of cell wall-associated proteins
other tissues, and tumor formation [7].
Several data suggest great homology between animal Cell wall protein extracts were obtained from either
and plant cell behavior, for instance, between the axon epicotyls or hypocotyls by salt extraction [18] with
guidance and the pollen tube targeting [6]. Implication modications. After weighing (5–15 g), organs were
of cell wall adhesive proteins in plant resistance to os- ground in a mortar in 20 ml of 50 mM Tris–HCl,
motic stress was rst identied in NaCl-adapted plant 330 mM mannitol, and 3 mM PMSF at pH 7.4. The
cells [8] and during water decit [9]. Results indicate the solution was centrifuged at 10,000g for 10 min at 8 °C
presence of wall proteins closely related to animal and the pellets were homogenized by vortexing with
vitronectin or FbN [8,10,11] or containing the RDG 20 ml of 50 mM Tris–HCl, 150 mM NaCl, pH 7.4. The
sequence or an RGD-like conguration [9,12]. RGD- procedure was repeated four times. At this point, the
dependent ECM-plasma membrane interactions have crude cell wall pellets were weighed, extracted in 2 ml/g
been described in plants [13–16]. Both ECM attachment cell wall residue of 50 mM Tris–HCl, pH 7.4, 1.5 M
proteins and plasma membrane receptors have yet to be NaCl, and 25 mM EDTA for 2 h on ice, and hand-mixed
puried and classied comparing to animal homologues. every 15 min. Extracts were collected by centrifugation
As described for animals, plant cell attachment on at 10,000g for 10 min at 8 °C.
ECM should play a fundamental role in all steps of
development. This was described for pollen tube tar- Dialysis and concentration
geting [6] and for developmental information during
embryogenesis [10,17]. The NaCl extracts were desalted by extensive dialysis
The purpose of the present work is not only to against 50 mM Tris–HCl, 1 mM EDTA, pH 7.4
establish a purication process of cell wall proteins (2 5 liter changes at 8 °C for 18 h). When the volume
with young germinations as models, but also to pro- was greater than 5 ml, the dialysate was concentrated by
pose a convenient protocol based upon the blood reverse dialysis against dry sucrose [19] at 8 °C until the
plasma bronectin purication previously described desired volume was reached and dialyzed further against
[2]. According to previous works, overexpression of 50 mM Tris–HCl, 1 mM EDTA, pH 7.4. Dialysates were
cell wall adhesion proteins could be induced by salt or carefully recovered and amounts of proteins were
water stress [8,9]. Such stress conditions are imposed determined.
on plants seeds. We also explore the possibility that
cell walls overexpress proteins as a response to envi- Gelatin anity chromatography
ronmental stress. It suggests that such proteins func-
tion as linkers, signaling elements, and provide Dialysate was loaded onto a 5-ml laboratory Luer-
physical anchoring to the cell during salt stress or lock connections empty column (MoBiTec, Gottingen,
water decit. Germany) lled with 1.5 ml gelatin–Cellune or gelatin–
Sepharose and equilibrated in 50 mM Tris–HCl,
150 mM NaCl, and 1 mM EDTA, at pH 7.4 (gelatin-
Materials and methods stabilization buer). The resin was carefully mixed with
the dialysate in the column by hand mixing and de-
All extraction steps were carried out at 4 °C or on ice canted for 15 min on ice. After four washing steps with
to minimize proteolytic degradation. Anity chroma- 5 ml stabilization buer (until no protein was detected in
tography purication steps were carried out at room eluate fractions), bound proteins were eluted with
temperature. All common chemicals meet American 2 5 ml of 50 mM Tris–HCl, 150 mM NaCl, and 6 M
Chemical Society specications. Gelatin and heparin– urea at pH 7.4. Eluted proteins were immediately con-
Cellune resins were from Interchim (Montlucon, centrated by sucrose reverse dialysis and extensively
France). Gelatin–Sepharose was from Pharmacia-Bio- dialyzed for 18 h against (2 5 liters) 50 mM Tris–HCl,
tech. Human blood plasma bronectin was puried in 1 mM EDTA at pH 7.4. After chromatography, the
our laboratory according to the process previously de- column was washed with gelatin-stabilization buer
scribed [2]. supplemented with 8 M urea, prior to reequilibration.

210 D. Pellenc et al. / Protein Expression and Purication 34 (2004) 208–214


Heparin anity chromatography (Bio-Rad, France) using human blood plasma bro-
nectin and bovin serum albumin as protein standards.
Eluted gelatin anity fractions were loaded onto a
10-ml laboratory Luer-lock connections empty column
(MoBiTec, Gottingen, Germany) lled with 5 ml hepa-
Results
rin–Cellun in 50 mM Tris–HCl, 150 mM NaCl, and
1 mM EDTA at pH 7.4 (heparin-stabilization buer). Cell wall purication and extraction of wall-associated
Bound proteins were eluted with stabilization buer protein
supplemented with increasing sodium chloride concen-
trations (0.15, 0.3, and 1 M). After chromatography, the To determine any dierence in the total amount of
column was washed with stabilization buer supple- cell wall crude fraction and the associated proteins, we
mented with 7 M urea, followed by stabilization buer. compare the data obtained for epicotyls grown in water
(control) to those for epicotyls grown in the presence of
Electrophoresis 150 mM NaCl. As shown in Table 1 no signicant dif-
ference was observed. The yield of total cell wall ranges
Gel electrophoresis was performed by SDS–PAGE from 0.32 0.08 (control) to 0.35 0.14 (salt stress) g
[20] using 0.75 mm thick polyacrylamide gels (3% per g of fresh epicotyl material. The dierence is not any
stacking gel, pH 6.8, 10% separating gel 10%, pH 8.8) in more signicant for the total amount of salt extracted
a Mighty Small II gel apparatus (Hoefer Pharmacia proteins: 0.79 0.21 (control) and 0.94 0.36 (salt
Biotech) cooled by water ow. Samples were adjusted to stress) mg of extracted protein per g of crude cell wall
3.5% SDS and 10% glycerol. Under reducing conditions fraction. The same results are observed when starting
the samples were treated with 2.5% (v/v) of 2-mercap- from hypocotyls (data not shown). The present extrac-
toethanol. Samples were never boiled as described in tion protocol allows us to recover enough cell wall
previous bronectin studies [2,21]. The molecular mas- protein for further purication. Starting from 10 g of
ses were estimated from the relative migration of stan- fresh epicotyls (whatever the farming conditions) we
dard mixtures of known proteins (High Molecular obtain 2.5–3 mg cell wall extracted proteins.
Weight standard markers, Sigma). Proteins were visu-
alized after electrophoretic separation by silver staining Eect of salt stress on expression of cell wall proteins
(Bio-Rad). puried by gelatin-anity chromatography

Protein blotting and immunodetection Gelatin-anity chromatography is widely used to
purify preliminary FbN from dierent sources. Fig. 1
Protein blotting [22] from polyacrylamide gels to presents the data obtained for epicotyl cell wall proteins
PVDF membranes was carried out with a semi-dry puried onto gelatin–Sepharose. Epicotyls from un-
procedure in Nova Blot apparatus (Pharmacia Biotech). treated germinations contain in their wall some proteins
Proteins blotted on PVDF (Bio-Rad) were detected with with high anity for gelatin. The mean amount of
anti-human bronectin rabbit polyclonal serum from protein recovered for eight dierent extractions from
Sigma, reference F-3648, diluted to 1:10,000. The sec- the control batch is 16.9 7.2 lg per g of fresh mate-
ondary antibody was an anti-rabbit IgG alkaline phos- rial. This means 50 lg per g of crude cell wall. Relative
phatase conjugate from Sigma, reference A-3687, to control, 150 mM NaCl stressed epicotyls expressed
diluted to 1:15,000 and the alkaline phosphatase activity 3- or 4-fold more proteins with anity for gelatin.
was detected using Bio-Rad, AP color reagent kit. Total recovery for 12 dierent extractions is
65.1 17.7 lg per g of fresh material This means 200 lg
Protein assay proteins per g of crude cell wall. The dierence ob-
served between the two sets of epicotyls is highly sig-
Quantity determination of protein in dierent frac- nicant as determined by Students t test with a
tions was performed with Bradford assay procedure probability of 0.05%. The gelatin anity proteins


Table 1
Cell wall and cell wall proteins purication in both control epicotyls and salt stressed epicotyls of young germinating peas

Cell wall fraction in g per g of fresh material Cell wall proteins in mg per g of cell wall

Pea epicotyls grown with water 0.32 0.08 (n 8) 0.79 0.21 (n 8)
Pea epicotyls grown with 150mM NaCl 0.35 0.14 (n 12) 0.94 0.36 (n 12)

Total quantity of cell wall fraction is estimated in g per g of fresh epicotyls (from 5 to 15g of fresh material). Cell wall proteins were extracted
from cell wall fraction by high NaCl and EDTA treatment and results are given in mg protein (determined by Bradford assay) per g of cell wall pellet.
Data are expressed as SD and n is the number of dierent extracts.

D. Pellenc et al. / Protein Expression and Purication 34 (2004) 208–214 211




Fig. 1. Eect of salt stress on expression of epicotyl cell wall proteins
with anity for gelatin–Sepharose. Cell wall proteins from either water
grown (control) or salt stressed epicotyls (NaCl stress) were puried by
gelatin–Sepharose anity chromatography. Proteins eluted with 6 M
urea were dialyzed and amounts were determined by Bradford assay.
Fig. 2. SDS–PAGE comparison of epicotyl cell wall proteins, puried
Data represent means of eight dierent purications for controls and
by gelatin anity chromatography, with mammalian bronectins.
12 purications for epicotyls grown under 150mM NaCl. The results
Samples (2lg) collected after gelatin–Cellun (lanes A and B) or gel-
are expressed in micrograms of protein with high anity for gelatin per
atin–Sepharose (lanes F and G) anity chromatography were run on a
g of fresh epicotyls SD. The asterisk indicates that the results are
10% SDS–PAGE under reducing conditions. The gel was silver
signicantly, dierent from the untreated control at the 0.005 level
stained. MW indicate molecular weights ranging from the top to the
(Students t test).
bottom: 180, 116, 97, 66, 48, and 33kDa. Lanes C, D, and E related,
respectively, to human (C : 2 lg; D: 5lg) and rabbit (E: 5 lg) blood
plasma FbN pre-puried on gelatin–Cellun without any further
puried from salt stressed epicotyls were preferentially heparin anity chromatography step.
tested and the results are presented above. Comparable
results were nevertheless obtained with untreated epi-
cotyl cell wall gelatin anity proteins. Furthermore, 5 lg of each kind of protein was run on
SDS–PAGE but silver staining should be stopped earlier
Identication of the cell wall protein puried by gelatin- for pea proteins than for animal FbN to prevent over-
and heparin-anity chromatography exposure. This suggests that the Bradford assay under-
estimates the real amount of pea proteins compared to
For gelatin-anity chromatography the stabilization animal bronectin.
buer contained 150 mM NaCl to prevent non-specic Two 4 and 8 lg of salt stressed epicotyl cell wall pro-
binding of proteins on both porous matrices: cellulose teins with anity for gelatin and 8 lg of human blood
for gelatin–Cellun or agarose for gelatin–Sepharose. plasma FbN were run on the same SDS–PAGE under
We used the two supports to check the implication of reducing or nonreducing conditions (Fig. 3). The
the nature of the gel in any anity background inter-
actions, especially when starting from plant cell wall
proteins, which could recognize a cellulose matrix. Since
no dierence is observed between the two gelatin ma-
trices on the electrophoretic patterns of the puried
proteins, as shown on SDS–PAGE (Fig. 2), we conclude
that the support matrix does not interfere with the pu-
rication process. The results are compared to animal
FbN, which had been pre-puried with gelatin–Cellun
matrix. At this rst step of purication, human and
rabbit blood plasma FbN share numerous contamina-
tion bands corresponding to dierent plasma proteins
having anity for gelatin (e.g., matrix metallopro-
teases). This justies the inclusion of other steps for
animal plasma bronectin purication. Such contami-
Fig. 3. SDS–PAGE salt stressed pea epicotyl cell wall proteins puried
nations seem not to occur when pea cell wall proteins by gelatin anity chromatography. Samples of gelatin anity puried
are puried by single gelatin anity chromatography. salt stressed epicotyl cell wall proteins (lanes 1 and 5, 2 lg; lanes 2 and
The electrophoretic pattern of cell wall protein fractions 6, 4lg; and lanes 3 and 7, 8 lg) and gelatin/heparin anity puried

eluted with 6 M urea on gelatin chromatography shows human plasma FbN (lanes 4 and 8, 8lg) were run on a 10% SDS–
PAGE and silver stained under reducing conditions (lanes 1–4) or not
the presence of a dark single band under reducing con-
(lanes 5–8). Numbers indicate apparent molecular weights (kDa).
ditions at 220 kDa close to mammalian bronectin Double arrow indicates position of the human blood plasma single-
subunits. chain under reducing conditions in both parts of the same gel.

212 D. Pellenc et al. / Protein Expression and Purication 34 (2004) 208–214


electrophoretic patterns obtained for reducing condi- Heparin-anity chromatography was used to check
tions (lanes 1–4) in the presence of 2-mercaptoethanol if the proteins eluted from gelatin have the same anity
indicate that pea cell wall proteins migrate at the same for heparin as human bronectin. The total elution of
220 kDa size as the human FbN subunit (lane 4). As proteins bound to heparin–Sepharose occurred between
mentioned before for protein determination, when 300 and 500 mM NaCl as previously described for hu-
comparing 8 lg human bronectin to the same amount man plasma bronectin.
of Bradford determined cell wall protein, the pattern Pea proteins puried by gelatin and heparin anity
indicates a stronger signal for pea proteins with over- were blotted from polyacrylamide gels to PVDF mem-
stained bands. Even with 2 lg cell wall proteins (lane 1), branes and probed with polyclonal antibodies directed
the 220 kDa band is stronger than the 8 lg human FbN against human plasma bronectin. Immunoblots
band (lane 4). (Fig. 4) indicate that the pea cell wall proteins puried
Under non-reducing conditions (lanes 5–8) both by gelatin-anity alone (lane 4) or by gelatin then
proteins (pea cell wall proteins or human plasma bro- heparin-anity (lane 3) are well recognized by anti-
nectin) migrate slower. The present SDS–PAGE is not bodies directed against human FbN as compared with
precise enough to identify the exact apparent molecular controls: human plasma (lane 1) and human cell (lane 2)
weight but we can assume that the 450 kDa molecular bronectin. A single band is detected on the Western
weight for human bronectin under non-reducing con- blot with the antibodies. Similar data were obtained
dition is visible (lane 8) and no dierence between both during the purication steps when using dot–blot assay
animal and plant proteins can be detected (lanes 5–7). with polyclonal antibodies directed against human FbN
This presumes that the pea cell wall protein puried with but no signal occurred when using polyclonal antibodies
the present protocol shares size similarity with human directed against other blood plasma proteins such as
plasma bronectin: not only the molecular weight but brinogen or von Willebrand factor.
also the possibility having two subunit chains as de-
scribed for mammalian bronectin.
Because we successfully get human plasma FbN with Discussion
a purication process consisting of gelatin and heparin
anity chromatography [2], we used the same combi- Adhesion molecules similar to those found in animal
nation for the present work even though a single gelatin connective tissues have been immunologically detected
anity chromatography step was sucient. in dierent plant species. Evidence for the presence of an
RGD-cell binding domain is supported by various
studies [8–17]. Nevertheless, to our knowledge, nobody
has tried to purify such proteins directly from whole
plant tissues by using protocol described for animal
MEC proteins. Supporting the idea that ECM proteins
found in plant or in animal tissues share great functional
and structural similarities, this paper adapts our human
plasma bronectin purication protocol to plant tissues
proteins.
The main problem found with plant tissues is the
presence of dierent pigments (especially chlorophyll
and carotenoids), which interfere when optical density
measuring is needed. Preliminary experiments done
with whole green adult leaves of dierent plants such
as spinach, sugar beat or bean did not give good re-
sults because of the large amount of chlorophylls and
carotenoid pigments which interfere with both gel af-
Fig. 4. Immunoblot detection of salt stressed pea cell wall gelatin af- nity and protein assay. This led us to choose etio-
nity puried proteins with polyclonal antibodies directed against lated organs of plants grown in darkness. Such plants
human plasma bronectin. Cell wall proteins from NaCl stressed pea
lacking high pigment levels are more suitable. Fur-
epicotyls were puried by gelatin anity chromatography (lane 4) or
by combination of gelatin and heparin anity chromatography (lane thermore, young organs such as epicotyls or hypoco-
3), separated under reduced conditions on SDS–PAGE, transferred on tyls are more convenient for physiological experiments.
PVDF membrane, and probed with polyclonal antibodies directed Epicotyls are widely used by plant physiologists as
against human blood plasma FbN. Human blood plasma FbN puried models in studies dealing with environmental factor
with the gelatin/heparin anity chromatography (lane 1) and com-
impacts. In such organs, during the rst steps of de-
mercial cellular FbN (lane 2) were used as positive controls. Arrows
indicate the 220kDa position of the human blood plasma bronectin velopment, cell elongation occurs (auxesis) before cell
subunits. division (meresis).

D. Pellenc et al. / Protein Expression and Purication 34 (2004) 208–214 213


In 1993, Zhu et al. described the enrichment of vi- [5] S.A. Corbett, C.L. Wilson, J.E. Schwarzbauer, Changes in cell

tronectin- and bronectin-like proteins in tobacco cell spreading and cytoskeletal organization are induced by adhesion
to a bronectin–brin matrix, Blood 88 (1996) 158–166.
culture with excess NaCl. Immunological binding sup-
[6] R. Palanivelu, D. Preuss, Pollen tube targeting and axon guidance:
ported this hypothesis. Our data not only conrm such a parallels in tip growth mechanisms, Trends Cell Biol. 10 (2000)
hypothesis but also propose a convenient method to 517–524.
purify signicant amount of such proteins. Evidence for [7] F. Carreiras, B. Thiebot, J. Leroy-Dudal, S. Maubant, M.F.

the adaptation of young pea germinations to salt stress Breton, H. Darbeida, Involvement of avb 3 integrin and disrup-
tion of endothelial bronectin network during the adhesion of the
is given when comparing the 3- to 4-fold increase of
human ovarian adenocarcinoma cell line IGROV1 on the human
recovery of gelatin-anity proteins from 150 mM NaCl umbilical vein cell extracellular matrix, Int. J. Cancer 99 (2002)
stressed peas compared to water irrigated peas. All these 800–808.
data led us to conclude that we puried bronectin-like [8] J.K. Zhu, J. Shi, U. Singh, S. Wyatt, R. Bressan, P. Hasegawa, N.

protein localized at the cell wall level. Further investi- Carpita, Enrichment of vitronectin- and bronectin-like proteins
in NaCl-adapted plant cells and evidence for their involvement in
gations should be carried out to get more structural
plasma membrane-cell wall adhesion, Plant J. 3 (1993) 637–
information on this protein. By physical approaches we 646.
could determine the structures of such proteins and [9] B. Garcia-Gomez, F. Campos, M. Hernandez, A. Covarrubias,
correlate structure to function as done in our laboratory Two bean cell wall proteins more abundant during water decit

for human blood plasma FbN [23]. Furthermore, cell are high in proline and interact with a plasma membrane protein,
Plant J. 22 (2000) 277–288.
physiological adhesion tests [7] will be required to con-
[10] V.T. Wagner, L. Brian, R.S. Quatrano, Role of a vitronectin-like
rm the role of this protein in anchoring the cell to its molecule in embryo adhesion of the brown alga Fucus, Proc. Natl.
physiological or articial support. We expect a struc- Acad. Sci. USA 89 (1992) 3644–3648.
tural and functional homology between animal FbN [11] L.C. Sanders, C. Wang, L.L. Walling, E. Lord, A homolog of the

and plant FbN-like protein that could explain several substrate adhesion molecule vitronectin occurs in four species of
owering plants, Plant Cell 3 (1991) 629–635.
cell physiological signaling pathways. Animal FbN and
[12] F. Bernier, A. Berna, Germins and germin-like proteins: plant do-
other ECM proteins are also considered as anchoring all proteins. But what do they do exactly?, Plant Physiol. Biochem.
elements for pathogens, especially in septic and noso- 39 (2001) 545–554.
comial diseases [24]. Previous data from our laboratory [13] A.M. Laboure, A. Faik, P. Mandaron, D. Falconet, RGD-

clearly showed evidence for the presence of FbN-bind- dependent growth of maize calluses and immunodetection of an
integrin-like protein, FEBS Lett. 42 (1999) 123–128.
ing proteins at the surface of pathological Pseudomonas
[14] H. Barthou, M. Petitprez, C. Briere, A. Souvre, RGD-mediated
[25,26]. Similar pathogenic processes were also described membrane-matrix adhesion triggers agarose-induced embryoid
in plant infection: Fungal appressorium formation is formation in sunower protoplasts, Protoplasma 206 (1999) 143–
induced in plant tissues by specic elements provided by 151.

the host plant [27]. Fungal appressorium dierentiation [15] V. Laval, M. Chabannes, M. Carriere, H. Canut, A. Barre, P.
Rouge, R. Pont-Lezica, J.P. Galaud, A family of Arabidopsis
on inductive surfaces could be blocked with peptides
plasma membrane receptors presenting animal b-integrin do-
containing the RGD sequence as competitors. These mains, Biochim. Biophys. Acta 1435 (1999) 61–70.
data suggest that integrin and bronectin-like proteins [16] A. Faik, A.M. Laboure, D. Gulino, P. Mandaron, D. Falconet, A
are involved not only for pathogenic linkage but also as plant surface protein sharing structural properties with animal

sensing elements to induce dierentiation and patho- integrins, Eur. J. Biochem. 253 (1998) 552–559.
[17] C. Brownlee, F. Berger, Extracellular matrix and pattern in plant
genic development. FbN-like proteins identied in the
embryos on the lookout for developmental information, Trends
plant cell wall of pea epicotyls and hypocotyls (data not Genet. 11 (1995) 344–348.
shown) could be not only targets for plant pathogen [18] G.J. McDougall, Procedure for selection of cell wall-associated
invasion but also adaptive signaling elements for plant glycoproteins, Phytochemistry 45 (1997) 633–636.

cell responses to environmental stresses such as water [19] P.A. Underwood, P.A. Bean, S.M. Mitchell, J.M. Witelock,
Specic anity of cell adhesion molecules and growth factors
decit and salt stress.
from serum, J. Immunol. Methods 247 (2001) 217–224.
[20] J.P. Doucet, J.M. Trifaro, A discontinuous and highly porous
sodium dodecyl sulfate–polyacrylamide slab gel system of high
resolution, Anal. Biochem. 168 (1988) 265–271.
References [21] S.I. Miekka, Heat-induced fragmentation of human plasma
bronectin, Biochim. Biophys. Acta 748 (1983) 374–380.
[1] E. Ruoslahti, B. Obrink, Common principles in cell adhesion, [22] W.N. Burnette, Western blotting: Electrophoretic transfer pro-
Exp. Cell. Res. 227 (1996) 1–11. tein from dodecyl sulfate–polyacrylamide gels to unmodied
[2] L. Poulouin, O. Gallet, M. Rouahi, J.M. Imho, Plasma nitrocellulose and radiographic detection with antibody and
bronectin: three steps to purication and stability, Protein. radioiodinated protein, Anal. Biochem. 112 (1981) 195–
Expression. Purif. 17 (1999) 146–152. 203.
[3] S.A. Mousa, D.A. Cheresh, Recent advances in cell adhesion [23] E. Pauthe, J. Pelta, S. Patel, D. Lairez, F. Goubard, Temperature-
molecules and extracellular matrix proteins: potential clinical induced b-aggregation of bronectin in aqueous solution, Bio-
implications, Drug Discovery Today 2 (1997) 187–199. chim. Biophys. Acta 1597 (2002) 12–21.
[4] H. Mohri, Fibronectin and integrins interactions, J. Invest. Med. [24] E. Brouillette, G. Grondin, L. Shkreta, P. Lacasse, B.G. Talbot,
44 (1996) 429–441. In vivo and in vitro demonstration that Staphylococcus aureus is

214 D. Pellenc et al. / Protein Expression and Purication 34 (2004) 208–214


an intracellular pathogen in the presence or absence of bronectin- [26] J. Rebiere-Huet, J. Guerillon, A.L. Pimenta, P. Di Martino, N.

binding proteins, Microbial. Pathogen. 35 (2003) 159–168. Orange, C. Hulen, Porins of Pseudomonas uorescens MFO as
[25] J. Rebiere-Huet, P. Di Martino, O. Patrick Gallet, C. Hulen,
bronectin-binding proteins, FEMS Microbiol. Lett. 215 (2002)
Interaction of Pseudomonas aeruginosa outer membrane protein 121–126.
with plasmatic bronectin. A root for new bacterial adhesions, CR [27] H.B. Deising, S. Werner, M. Wernitz, The role of fungal
Acad. Sci. 322 (1999) 1071–1080. appressoria in plant infection, Microb. Infect. 2 (2000) 1631–1641.


International Journal of Biological Macromolecules 34 (2004) 233–240




Purication and characterization of human hemoglobin:
effect of the hemolysis conditions

C.T. Andrade, L.A.M. Barros, M.C.P. Lima, E.G. Azero

Instituto de Macromoléculas Professora Eloisa Mano, Universidade Federal do Rio de Janeiro,
P.O. Box 68525, 21945-970 Rio de Janeiro, RJ, Brazil

Received 6 May 2004




Abstract

Human hemoglobin was isolated and puried by anion exchange chromatography. To isolate hemoglobin, outdated red blood cells (RBC)
were transformed into carbonylhemoglobin, by reaction with carbon monoxide, and submitted to washing/centrifugation procedures, to
eliminate other plasma proteins. Albumin was quantied in each supernatant, by the bromcresol green method. Hemolysis was performed
in three different hypotonic media (water, 0.01 M NaCl and 5 mM Tris/HCl buffer at pH 7.4), at 8 C for 24 h. Sonication for 5 min was also
used to lyse RBC. After isolation of hemoglobin, additional purication was carried out by anion exchange chromatography on AG MP-1,
Q-SFF and both exchangers. Hemoglobin concentration of hemolysates and of puried solutions were determined by the hemiglobincyanide
method. Residual phospholipids were extracted from the four different hemolysates, as well as from the puried hemoglobin solutions,
and were analyzed by high performance liquid chromatography. Native and SDS-polyacrylamide gel electrophoresis experiments were
performed on puried hemoglobin samples to verify the presence of proteins other than hemoglobin. According to the results, the hemolysis
conditions have inuence on the purication of hemoglobin.
2004 Elsevier B.V. All rights reserved.

Keywords: Human hemoglobin; Purication; Ion exchange chromatography; High performance liquid chromatography; Electrophoresis




1. Introduction diation therapy, and for preservation of organs and tissues
for transplantation [5,8,9].
Intensive efforts, mainly motivated by blood shortage as- Some human and bovine hemoglobin based oxygen carri-
sociated with increasing demand [1], have been made to de- ers are currently under clinical trials [4,5,7,10]. Hemoglobin
velop safe injectable uids that could be used in place of is a tetrameric protein composed of two identical α and
blood, and that could be capable of transporting and deliv- two identical β globin chains, each bound to a heme group
ering the respiratory gases, oxygen and carbon dioxide, for capable of binding one molecule of oxygen inside the red
a limited period of time. Such products, often called arti- blood cell (RBC). This structure is stabilized by hydrogen
cial blood, blood substitutes or, more accurately, oxygen bondings, van der Waals forces, intra- and intermolecular
carriers, would offer substantial clinical advantages over the salt bonds. Under normal conditions in man, oxygen re-
transfusion of red blood cells, including a long shelf life, lease and delivery to tissues is modulated by the presence of
minimal risk of transmitting infectious agents, and no need 2,3-diphosphoglycerate (2,3-DPG), which acts as allosteric
for cross matching [2–7]. Oxygen carriers would nd major effector. Outside the RBC, human hemoglobin dissociates
applications in general surgery, as a treatment for ischemia into α, β dimers and looses 2,3-DPG. As a result, cell-free
associated with myocardial infarction or stroke in which hemoglobin has a short half-life in circulation, which causes
blood is not indicated, in the initial resuscitation stage of kidney damage, and high oxygen afnity. Also, some un-
acute trauma or hemorrhaging patients, as enhancers of ra- desirable and toxic effects of hemoglobin solutions have
been suggested as due to membrane phospholipids and non-
hemoglobin protein contaminants [11]. To circumvent these

Corresponding author. Tel.: +55 21 2562 7208; effects, hemoglobin from outdated RBC should be isolated,
fax: +55 21 2270 1317. puried and chemically modied before its use as acellular
E-mail address: ctandrade@ima.ufrj.br (C.T. Andrade). oxygen carriers.


0141-8130/$ – see front matter 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijbiomac.2004.05.003

234 C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240


RBC washing, lysing, thermal treatment, centrifugation [17]. The resulting suspension was washed with an equal
and ltration procedures are usually carried out to isolate weight of isotonic saline solution (0.9% NaCl, w/v), and
hemoglobin. Purication demands attention, since phos- centrifuged at 1000 × g for 30 min in a Hermle Centrifuge
pholipid components of the red cell membrane are potential model Z 383 K from National Labnet Company Inc. (New
toxins if present even in small amount in a product to be ad- Jersey, USA). The supernatant was discarded and the wash-
ministered by intravenous infusion [12]. Anion and cation ing/centrifugation procedure was repeated twice.
exchange chromatography have been used to eliminate To verify the efciency of the washing/centrifugation pro-
residual phospholipids [13–16]. cess, each supernatant was analyzed at 626 nm in a Ther-
In the present work, four methods were used to lyse molyne Turner spectrophotometer, model SP-870 (Dubuque,
RBC. In order to prevent oxidation, oxyhemoglobin inside USA) according to the bromcresol green method at pH 4
the RBC was transformed into thermally stable carbonyl- [18], using a calibration curve prepared with dilute solutions
hemoglobin prior to the isolation process [17]. The result- of bovine serum albumin.
ing hemolysates were further puried by anion exchange Hemolysis was performed in water, in 0.01 M NaCl, and
chromatography, characterized and compared, in relation to in 5 mM Tris/HCl buffer at pH 7.4, using a 1:2 (w/w) ratio
hemoglobin concentration, residual phospholipids and other of RBC/hypotonic solution at 8 C for 24 h. Hemolysis was
protein contaminants. The aim was to verify the effect of the also carried out by sonication at 8 C for 5 min with a 750 W
hemolysis conditions on the concentration of the more abun- Cole Parmer Processor (Vernon Hills, USA), equipped with
dant phospholipids in hemolysates and puried samples. a standard pin of 13 mm diameter at 40% amplitude.
After hemolysis, the suspension of lysed cells was sub-
mitted to heating at 60 C for 1 h in a water bath in the dark,
2. Materials and methods and centrifuged at 2000 × g for 1 h. The bottom layer was
discarded. The hemoglobin solution (approximately 50 ml)
2.1. Materials was removed from the top layer of the tube, diluted with an
equal volume of water, saline or buffer solution (depending
Outdated red blood cells were kindly donated by the on the hemolysis medium), and subjected to another cen-
Hemotherapy Sector of Hospital Universitario Clementino trifugation at 2000×g for 1 h. The resulting hemoglobin so-
Fraga Filho, Universidade Federal do Rio de Janeiro lution, removed from the center layer, was ltered through
(HUCFF-UFRJ). The dye-binding reagent used for albumin carbon powder and 0.22 m Millipore membranes.
quantication (2.5 mM bromcresol in 0.82 M lactic acid at
pH 4 containing 30 ml/1 of Tween 80), the stock solution 2.3. Purication of hemoglobin by ion exchange
of bovine serum albumin at 0.4 g/1 stabilized with sodium chromatography
azide, Drabkin’s reagent and hemiglobincyanide (HiCN)
were purchased from Doles Reagentes (Goiania, Brazil). A Flash 12i chromatography system, purchased from Bio-
Phospholipid standards phosphatidylethanolamine (PE), tage, Division of Dyax Corporation (Charlottesville, USA)
phosphatidylcholine (PC) and sphingomyelin (SM), bovine was used. Polyethylene columns of 75 mm × 12 mm were
serum albumin and MW-SDS-70 Kit for Electrophoresis prepared by allowing 10 g of the anion exchangers AG MP-1
were purchased from Sigma (Saint Louis, USA) and used or Q-SFF to settle under gravity. Mixed columns were also
as received. Acetonitrile and methanol (HPLC grade) were prepared by packing equal amounts of both resins succes-
provided by Vetec Quimica Fina Ltda. (Rio de Janeiro, sively. Prior to use, each column was equilibrated at room
Brazil). temperature and 103 kPa with 0.01 M Tris/HCl pH 7.4 for
Anion exchangers, AG MP-1 from BioRad Labs (Brom- approximately 12 h. Hemoglobin lysates (1 ml) were eluted
ley, England) and Q Sepharose Fast Flow (Q-SFF) from with the same buffer at rates that varied according to the
Pharmacia Biotech (Wikstroms, Sweden) were used after pu- type of lysate and the type of resin. Fractions of 1 ml were
rication and conditioning. The AG MP-1 resin was washed collected and analyzed by spectrophotometry to determine
with ethyl alcohol and with deionized water, treated with hemoglobin concentration.
0.1 N NaOH for 15 min and neutralized by successive wash-
ings, and nally treated with 0.1 N HC1 for 15 min and neu- 2.4. Determination of hemoglobin concentration
tralized by extensive washings.
All other reagents and solvents (PA grade) were supplied Hemoglobin concentrations were determined by the
by Vetec (Rio de Janeiro, Brazil) and used as received. hemiglobincyanide (HiCN) method [19]. Hemoglobin
lysates and puried hemoglobin solutions (20 l) was mixed
2.2. Isolation of hemoglobin with 5 ml of 1:100 diluted Drabkin’s reagent, homogeneized
for 3 min and analyzed by absorption spectrophotometry at
RBC (100 g) was submitted to carbonylation reaction 540 nm. In each case, the hemoglobin concentration was
with CO gas under mild shaking for 180 s, to convert determined in relation to a calibration curve, prepared with
oxyhemoglobin (HbO2) to carbonylhemoglobin (HbCO) diluted solutions of standard HiCN.

C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240 235


2.5. Specication of isolated and puried hemoglobins alcohol aqueous solution. For native PAGE [22], 10 l
samples of puried hemoglobin solution were diluted
2.5.1. Extraction of phospholipids at 1 g/l in a buffer containing 8.8 ml 1 M Tris/HC1 pH
Phospholipids were extracted from the hemolysates and 6.8, 1.2 ml glycerol at 87% (v/v) and 0.5 ml bromphenol
from hemoglobin solutions after purication by anion ex- blue at 0.02% (w/v) and electrophoresed on discontinu-
change chromatography, according to the literature [20]. To ous poly(acrylamide/bis-acrylamide) gels for 80–90 min.
20 ml of a solution prepared by diluting 4 g of hemoglobin SDS-PAGE was performed according to the procedure of
lysate with 50 ml of deionized water, 50 ml methyl alco- Laemmli [23]. Samples at 1 g/l concentration were prepared
hol and 25 ml methylene chloride were added. After stir- by heat denaturation at 100 C for 5 min in a buffer con-
ring for 10 min, 25 ml of methylene chloride and 25 ml of taining 1.0 ml 1 M Tris/HCl pH 6.8, 4 ml deionized water,
2 M KCl were added. The resulting mixture was stirred for 1.6 ml SDS at 10% (w/v), 0.8 ml glycerol at 87% (v/v) and
10 min and transferred to a decantation funnel. The lower 0.2 ml bromphenol blue at 0.05% (w/v). Hemoglobin and
layer was collected and dried at 40 C. In the case of puried MW-SDS-70 solutions (10 l) were applied to each lane of
hemoglobin solutions, 1 g was taken from the two most con- the gel and processed at the same electrophoretic conditions.
centrated fractions, to which 10 ml of deionized water were MW-SDS-70 markers are composed of lysozyme (MW
added, and the procedure described above was followed. The 14,300), ( -lactoglobulin (subunits of MW 18,400),
residue obtained by drying the lower layer was redissolved trypsinogen (MW 24,000), pepsin (MW 34,700), egg
in 1 ml methylene chloride, ltered in poly(triuorethylene) albumin (MW 45,000), and bovine albumin (MW
0.22 m membranes and used to analyze the presence of 66,000).
phospholipids by high performance liquid chromatography.

2.5.2. High performance liquid chromatography (HPLC) 3. Results and discussion
Normal phase HPLC was carried out with a Pharma-
cia LKB-HPLC pump model 2248 from Pharmacia-Biotech The carbonylation of hemoglobin was carried out for
(Uppsala, Sweden), equipped with a HP 3396 Series II in- 180 s, a longer period of time than suggested by other authors
tegrator from Hewlett Packard (Palo Alto, USA). Separa- [17]. To verify the efciency of the washing/centrifugation
tions were performed on a stainless-steel column (250 mm procedures to eliminate plasma proteins contaminants, the
× 4.6mm i.d.), packed with 60 A TSK Gel Silica-60, from supernatant was assayed for albumin as a marker. Initially,
TOSOH Biosep LLC (Montgomeryville, USA) at 30 C and the methodology was repeated ve times; after each cycle,
a constant ow rate of 1.1 ml/min. Solutions of the stan- the resulting suspension was analyzed by spectrophotome-
dard phospholipids phosphatidylethanolamine (PE), phos- try at 626 nm. Fig. 1 shows the effect of each cycle of wash-
phatidylcholine (PC) and sphingomyelin (SM) were pre- ing/centrifugation on the albumin concentration, which con-
pared in methylene chloride at 1 g/1. A stock solution of sistently decreases from 0.5 g/dl to undetectable levels. As
mixed standard phospholipids was prepared with 2 ml PE no albumin was detected after the third cycle, three cycles
(250 g/ml), 1 ml PC (125 g/ml) and 5 ml SM (625 g/ml) were considered sufcient to eliminate plasma proteins from
solutions, and used to quantify residual phospholipids from RBC.
hemolysates and puried hemoglobin solutions. The sam-
ples were applied with a Hamilton syringe via a 20 l Rheo-
dyne 7125 injector (Cotati, USA) and eluted with acetoni-
trile/methyl alcohol/phosphoric acid in 900:95:5 volume ra-
tio [17,21]. The same solvent mixture was used as mo-
bile phase for the analysis of phospholipids extracted from
hemoglobin experimental samples. The elution was mon-
itored with a UV–vis Shimadzu model SPD-10AV from
Shimadzu Scientic Instruments (Columbia, USA), set at
210 nm.

2.6. Electrophoresis


Polyacrylamide gel electrophoresis (PAGE) and sodium
dodecylsulfate-polyacrylamide gel electrophoresis (SDS-
PAGE) were carried out in a single-sided vertical Owl Sci-
entic Inc. system, model P81 (Woburn, USA), equipped
with a Electrophoresis Power Supply E 835 from Con-
sort nv (Turnhout, Belgium), at 2 W. Gels were stained in Fig. 1. Effect of the number of washing/centrifugation procedures on the
Comassie blue G-250 and destained in acetic acid/ethyl albumin concentration of RBC samples.

236 C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240


The methodology used for hemolysis was chosen as the
best conditions, after previous experiments (not shown),
in which the RBC/hypotonic medium ratio (w/w), temper-
ature and period of time were varied, and the resulting
hemolysates were analyzed by optical microscopy [24]. As
specied in the experimental section, the hemolysis pro-
cess was performed in water, in 0.01 M NaCl, and in 5 mM
Tris/HCl buffer pH 7.4, at 8 C and 1:2 ratio (w/w) for 24 h.
Ultrasound radiation was also used to lyse RBC at the same
temperature, for 5 min. Cellular debris and denatured pro-
tein contaminants were eliminated by centrifugation and l-
tration.
Additional purication was carried out by anion exchange
chromatography, using the Flash 12i system, with AG MP-1,
Q-SFF and two layers of each resin as stationary media.
Elution curves were obtained for the hemoglobin eluates.
Since columns were loaded manually, slightly different ow
rates were attained, although the same pressure was applied.
Typical chromatograms are shown in Fig. 2 for different
hemoglobin solutions that had been puried on AG MP-1,
Q-SFF and both resins. In all cases, only one peak was
detected. Hemolysis by sonication gave rise to hemoglobin
fractions at higher concentrations (6–14 g/1) compared to
the other methods (2–5 g/1).
Phospholipids were extracted from the most concentrated
fractions and quantied by normal phase HPLC, in com-
parison to a mixed solution of PE, PC and SM standards.
PE, PC and SM, together with phosphatidylserine (PS, not
analyzed in the present work) constitute the major phospho-
lipids present in the red cell membrane, and are frequently
used to monitor stromal contamination [13,17]. Fig. 3 shows
HPLC chromatograms of phospholipids obtained from
the solution of standards, a RBC suspension, an isolated
hemoglobin solution obtained after hemolysis in water (wa-
ter hemolysate) and from a water hemolysate additionally
puried with Q-SFF resin at 0.80 ml/min ow rate (denoted
W/Q-SFF/0.80). Elution times of 8.9, 19.3 and 27.4 min
were observed for PE, PC and SM, respectively (Fig. 3a).
SM sometimes appears as a large [17] or a double peak
[12,21]. According to the chromatogram of Fig. 3d, sample
W/Q-SFF/0.80 still contains a signicant amount of PC.
Also, two other peaks were detected at 22.6 and 24.2 min
elution times. These peaks, denoted Ul and U2, were ob-
served by other workers and their origin has not been fully
established. It was suggested that Ul and U2 might identify
degraded diacyl glyceride products of phospholipids, which
were proved to further degrade with longer storage [21].
Fig. 2. Chromatography on AG MP-1: (a) at 0.59 ml/min ( ), 0.30 ml/min
Table 1 shows an estimation of phospholipids concentra-
( ), and 0.17 ml/min ( ) ow rates; on Q-SFF (b) at 1.58 ml/min ( ),
tions detected by HPLC for hemolysates in water, saline and and 1.45 ml/min ( ) ow rates; and AG MP-1/Q-SFF resins (c) at 0.21 ml/
buffer solutions, and the hemolysate obtained by sonication, min ( ), and 0.03 ml/min ( ) ow rates, of hemoglobin eluates previously

in g/mg of hemoglobin. A high phospholipid concentration lysed in water, by sonication, and in saline solution, respectively.

in the hemolysate (before further purication) may indicate
a strong phospholipid/hemoglobin interaction. For example, sonication led to the lowest concentrations of PE, PC and
water seems to favor PC/hemoglobin interaction, whereas SM, and to the lowest total phospholipids concentration.
in saline and buffered hemolysates, hemoglobin maintains a Acoustic energy, not absorbed by molecules, via the in-
higher interaction with SM. On the contrary, hemolysis by direct complex phenomenon known as cavitation, can break

C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240 237




Fig. 3. Normal phase HPLC chromatograms of phospholipids from a solution of standards (a), a RBC suspension (b), a water hemolysate (c), and a
water hemolysate additionally puried on Q-SFF resin at 0.80 ml/min ow rate. Peaks Ul and U2 were not fully identied (see text).


the cohesion of liquids and cause fragmentation and/or ero- considered as the best hypotonic medium for hemolysis [26],
sion of solids [25]. In the present case, hemolysis by sonica- dilution of hemoglobin solutions in pure water seems not to
tion had a markedly different effect on the RBC membrane, contribute to purication by anion exchange chromatogra-
probably by collapsing the phospholipid membrane bilayer, phy, at least on AG MP-1 and Q-SFF resins, since substantial
when compared to hypotonic media, which cause the mem- amounts of PC were not retained by these resins. Both resin
brane to swell and rupture. type and ow rate inuenced the efciency of the process.
Table 2 shows an estimation of phospholipids concentra- AG MP-1 and mixed AG MP-1/Q-SFF showed improved
tions detected by HPLC for hemoglobin solutions, obtained efciency in comparison to Q-SFF alone. As expected, in
from hemolysis in water and additional purication by an- all cases, the lower the ow rate, the lower the amount of
ion exchange chromatography. Although cold water may be residual PC in the puried hemoglobin solutions.



Table 1
Estimation of phospholipids concentrationsa in RBC and hemolysates, determined by normal phase HPLC

Sampleb PE ( g/mg Hb) PC ( g/mg Hb) SM ( g/mg Hb) Total ( g/mg Hb)

RBC 62.1 592.4 122.1 776.6
Water hemolysate 0.1 54.5 ndc 54.6
Saline hemolysate ndc 0.1 64.4 64.4
Buffered hemolysate 1.0 1.3 119.0 121.3
Ultrasound hemolysate 0.8 0.8 0.4 2.0

a PE, phosphatidylethanolamine; PC, phosphatidylcholine; SM, sphingomyelin; Hb, hemoglobin.
b RBC, red blood cells; Water hemolysate, hemoglobin solution isolated after hemolysis in water; Saline hemolysate, hemoglobin solution isolated
after hemolysis in 0.01 M NaCl; Buffered hemolysate, hemoglobin solution isolated after hemolysis in 5 mM Tris/HC1 pH 7.4; Ultrasound hemolysate,
hemoglobin solution isolated after hemolysis by sonication.
c nd, nondetected.

238 C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240


Table 2
Estimation of phospholipids concentrationsa determined by normal phase HPLC for puried hemoglobin (Hb) solutions from water hemolysates

Sampleb PE ( g/mg Hb) PC ( g/mg Hb) SM ( g/mg Hb) Total ( g/mg Hb)

W/AG MP-1/0.59 ndc 6.1 nd 6.1
W/AG MP-1/0.30 nd 1.7 nd 1.7
W/Q-SFF/0.80 nd 24.9 nd 24.9
W/Q-SFF/0.65 nd 14.7 nd 14.7
W/AG MP-1/Q-SFF/0.89 nd 36.6 nd 36.6
W/AG MP-1/Q-SFF/0.55 nd 20.1 nd 20.1

a PE, phosphatidylethanolamine; PC, phosphatidylcholine; SM, sphingomyelin.
b Each sample was denoted as water hemolysate (W)/type of anion exchanger/ow rate (ml/min), where water hemolysate refers to hemoglobin
solution isolated after hemolysis in water.
c nd, nondetected.



Lysing in hypotonic saline or buffered solutions, and by Puried hemoglobin samples were analyzed by elec-
sonication, with further purication on anion exchangers, trophoresis. It was of special interest to verify the effect
led to hemoglobin solutions in which the main phospho- of the sonication procedure on the hemolysate and on the
lipids PE, PC and SM were not detected by normal phase resulting puried hemoglobin solutions. Fig. 5 shows native
HPLC. As an example, Fig. 4 shows HPLC chromatograms PAGE (a) and SDS-PAGE (b) analyses for the hemolysate
of phospholipids extracts obtained from saline hemolysates obtained by sonication (lane 2) and corresponding puri-
before and after additional purication by anion exchange ed hemoglobin solutions on AG MP-1 at 0.14, 0.15 and
chromatography on AG MP-1, Q-SFF and both resins. As 0.13 ml/min ow rates (lanes 3, 4 and 5, respectively), on
may be observed, the peak that was eluted at 27.4 min in Q-SFF at 0.63 and 0.52 ml/min ow rates (lanes 6 and
the hemolysate (Fig. 4a) has vanished in the other chro- 7, respectively) and both resins at 0.17 and 0.05 ml/min
matograms. This result indicates the complete retention of ow rates (lanes 8 and 9, respectively), in comparison with
SM by the resins, independently of ow rate. the standard solution, prepared with MW-SDS-70 markers




Fig. 4. Normal phase HPLC chromatograms of phospholipids from saline hemolysates before additional purication (a), and after additional purication
by chromatography on AG MP-1 resin at 0.07 ml/min ow rate (b), on Q-SFF resin at 1.45 ml/min ow rate (c), and on AG MP-1/Q-SFF resins at
0.03 ml/min ow rate (d). Peaks Ul and U2 were not fully identied (see text).

C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240 239


thermal treatments to denature any remnant water-soluble
protein, other than heat-stable carbonylhemoglobin, and pu-
ried by anion exchange chromatography. As expected, the
puried hemoglobin solutions obtained by hemolysis in wa-
ter, 0.01 M NaCl and 5 mM Tris/HCl pH 7.4 showed lower
hemoglobin concentrations as compared to the samples orig-
inated from hemolysis by sonication. Residual phospho-
lipids extracted from the hemolysates and from puried
hemoglobin solutions were analyzed by HPLC. While the
highest total level of phospholipids was detected for the
buffered hemolysate, hemolysis by sonication led to the low-
est total concentration of phospholipids. After further pu-
rication by anion exchange chromatography on AG MP-1,
Q-SFF and both resins, hemoglobin samples obtained from
the water hemolysate still presented signicant concentra-
tions of phosphatidylcholine. No phospholipid was detected
by HPLC for the other puried samples. Electrophoresis
analyses showed that in all cases the hemoglobin structure
was maintained after the hemolysis and purication pro-
cesses.



Acknowledgements


The authors thank Doctor Carmen Martins Nogueira,
from the blood bank of Hospital Universitario Clementino
Fig. 5. Native PAGE (a) and SDS-PAGE (b) electrophoresis analyses for
Fraga Filho, Universidade Federal do Rio de Janeiro
the hemolysate obtained by sonication (Lane 2) and corresponding puried
(HUCFF-UFRJ), for supplying outdated red blood cells,
hemoglobin samples on AG MP-1 at 0.14, 0.15 and 0.13 ml/min ow
rates (Lanes 3, 4 and 5, respectively), on Q-SFF at 0.63 and 0.52 ml/min and Professor Luiz Carlos Trugo, Instituto de Quimica
ow rates (LANES 6 and 7, respectively) and both resins at 0.17 and (IQ-UFRJ), who made the HPLC equipment available.
0.05 ml/min ow rates (Lanes 8 and 9, respectively), in comparison with The nancial support provided by Conselho Nacional de
the standard solution (Lane 1).
Desenvolvimento Cientico e Tecnologico (CNPq), Coor-
denacao de Aperfeicoamento de Pessoal de Nivel Superior
(lane 1). In native PAGE, the presence of only one pro- (CAPES), Fundacao de Apoio a Pesquisa do Estado do
tein species was observed in every lane. Although a poor Rio de Janeiro (FAPERJ), and Fundacao Universitaria Jose
resolution was observed for the standards, the position Bonifacio (FUJB) is gratefully acknowledged.
of the bands revealed that, after hemolysis, isolation and
purication, hemoglobin was not dimerized; the protein
maintained the tetrameric conguration. Sonication was
efcient to promote hemolysis without protein degradation. References

As expected, SDS-PAGE analysis showed bands in every
lane, corresponding to the globin monomer. In lanes 3, 4, [1] Vamvakas EC, Taswell HF. Transfusion 1994;34:464.
[2] Goodnough LT, Scott MG, Monk TG. Clin Orthop Relat Res
8, and 9, a small amount of material migrated slower than
1998;357:89.
the major globin subunits. This band may be attributed to
[3] Menu P, Longrois D, Faivre B, Donner M, Labrude P, Stoltz J-F,
the presence of some hemoglobin variant. Differences in Vigneron C. Transf Sci 1999;20:5.
ow rate, used for purication on anion exchangers, had no [4] Winslow RM. Adv Drug Deliv Rev 2000;40:131.

signicant inuence on the electrophoresis results. [5] Riess JG. Chem Rev 2001;101:2797.
[6] Squires JE. Science 2002;295:1002.
[7] Moore EE. J Am Coll Surg 2003;196:1.
[8] Bunn HF. Tranf Clin Biol 1995;2:433.
4. Conclusion [9] Scott MG, Kucik SDF, Goodnough LT, Monk TG. Clin Chem
1997;43:1724.

After carbonylation and three cycles of washing/centri- [10] Carmichael FJL. Transf Apheresis Sci 2001;24:17.

fugation, red blood cells were considered free from plasma [11] MacDonald VW, Winslow RM, Marini MA, Klinder MT. Biomater
proteins and were submitted to hemolysis at 8 C. Three Artif Cells Artif Org 1990;18:263.
[12] Home T, Holt-Larkin S. J Chromatogr B 1997;695:259.
different hypotonic media, as well as sonication, were used [13] Williams RC, Tsay K-Y. Anal Biochem 1973;54:137.
for hemolysis. The resulting hemolysates were submitted to [14] Winslow RM, Chapman KW. Methods Enzymol 1994;231:3.

240 C.T. Andrade et al. / International Journal of Biological Macromolecules 34 (2004) 233–240


[15] Deepthi S, Johnson A, Sathish R, Pattabhi V. Biochim Biophys Acta [21] Nakai K, Sekiguchi S. In: Tsuchida E, editor. Articial Red Cell.
2000;1480:384. Chichester: Wiley; 1997. p. 131.
[16] Hu T, Su Z. Biochem Biophys Res Commun 2002;293:958. [22] Silva JG. Eletroforese de Protenas. Rio de Janeiro: Editora Inter-

[17] Sakai H, Takeoka S, Yokohama H, Seino Y, Nishide H, Tsuchida E. ciencia; 2001.
Protein Expr Purif 1993;4:563. [23] Laemmli UK. Nature 1970;227:680.
[18] Miyada DS, Baysinger V, Notrica S, Nakamura RM. Clin Chem [24] Barros LAM. Master Dissertation. Universidade Federal do Rio de
1972;18:52. Janeiro, 2003.
[19] International Council for Standardisation in Haematology: expert [25] Kardos N, Luche J-L. Carbohydr Res 2001;332:115.
panel on haemoglobinometry. J Clin Pathol 1996;49:271. [26] Haney CR, Buehler PW, Gulati A. Adv Drug Deliv Rev 2000;40:
[20] Bligh EG, Dyer WJ. Can J Biochem Physiol 1959;37–38:911. 153.


Biochimica et Biophysica Acta 1700 (2004) 209–217
www.bba-direct.com




Isolation and purification of antifreeze proteins from skin tissues of
snailfish, cunner and sea raven

Robert P. Evans*, Garth L. Fletcher

Ocean Sciences Centre, Memorial University of Newfoundland, St. John’s, Newfoundland, Canada A1C 5S7

Received 16 December 2003; received in revised form 13 May 2004; accepted 14 May 2004
Available online 21 June 2004




Abstract


Antifreeze proteins/polypeptides (AFPs), which are found in diverse species of marine fish, are grouped into four distinct classes (types
I–IV). The discovery of skin-specific type I AFPs established that this class contains distinct isoforms, liver-type and skin-type, which are
encoded by separate gene families. In this study, type I AFPs were isolated and partially characterized from skin tissues of Atlantic snailfish
(Liparis atlanticus) and cunner (Tautogolabrus adspersus). Interestingly, evidence from this study indicates that snailfish type I AFPs
synthesized in skin tissues are identical to those circulating in their blood plasma. Furthermore, type II AFPs that are identical to those
expressed in liver for export into blood were purified from sea raven (Hemitripterus americanus) skin tissue extracts. It is clear that epithelial
tissues are an important source for antifreeze expression to enhance the complement of AFPs that protect fish from freezing in extreme cold
environments. In addition, the evidence generated in this study demonstrates that expression of AFPs in fish skin is a widespread
phenomenon that is not limited to type I proteins.
D 2004 Elsevier B.V. All rights reserved.

Keywords: Type I; Type II; Antifreeze; Protein; Skin




1. Introduction However, more recent research has shown that winter
flounder and two sculpin species synthesize type I AFPs
Numerous species of teleost fish that inhabit extreme in epithelial tissues (termed skin-type) that are distinct from
cold environments synthesize antifreeze proteins/polypep- liver expressed proteins [5–7]. These skin-type AFPs derive
tides (AFPs) or antifreeze glycoproteins (AFGPs) for pro- from a subset of genes that are separate from the liver-type
tection against freezing. AFPs are grouped into four distinct multigene family. The skin-type AFPs are synthesized as
classes based on their diverse physical and structural char- mature proteins that lack both the signal and pro-sequences
acteristics—type I, II, III and IV [1–3]. However, despite typical of liver-type (plasma) proteins, and thus could
variations in protein structure, all AFPs lower solution remain intracellular. It has been suggested that the skin-type
freezing point non-colligatively by binding to specific ice AFPs are a widespread antifreeze isoform which might
crystal surfaces and inhibiting further growth. In the pres- represent a common adaptation in many cold ocean species
ence of a seed ice crystal, the observed difference between [8]. Some authors further suggested that liver-type AFPs
the lowered freezing point and unaffected melting point is evolved from skin-type proteins [5,8]. If this hypothesis is
termed thermal hysteresis and is used as an in vitro measure accurate, then all fish that contain plasma AFPs should
of antifreeze activity [1,3,4]. contain evidence of skin expression or at least the remnants
Originally, it was generally accepted that the synthesis of of an ancestral gene.
AFPs was confined solely to liver tissue (termed liver-type) In order to better assess the distribution of skin type
for secretion into blood for extracellular freeze protection. AFPs in general, three teleost species were examined in
this study that were known to contain plasma AFPs.
Atlantic snailfish (Liparis atlanticus) belong to a family
* Corresponding author. Department of Biochemistry, University of
of marine fishes in the order Scorpaeniformes that is
Alberta, Edmonton, AB Canada T6G 2H7. Tel.: +1-780-492-3481; fax: +1-
780-492-0886. closely related to sculpins [9]. Type I AFPs that were
E-mail address: robert.evans@ualberta.ca (R.P. Evans). previously isolated and characterized from the blood

1570-9639/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbapap.2004.05.006

210 R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217


plasma of Atlantic snailfish are the largest described to ity were pooled and lyophilized. Following dialysis in 0.1 M
date (>9.3 kDa) [10]. Sea raven (Hemitripterus ameri- NH4HCO3, samples were further purified by reverse phase
canus), which contain type II AFP in blood plasma, was HPLC. The partially purified AFPs were separated on a
judged to be an ideal comparison species since they are Nucleosil C8 column (0.46 25 cm). A gradient of 35–
closely related to snailfish and sculpins which produce 65% acetonitrile (solvent A) and 0.1% trifluoroacetic acid
type I AFPs [3,9]. If sea raven also contain type I AFP in (solvent B) was used with a flow rate of 1 ml/min.
skin, this would help to clarify the phylogenetic distribu- Individual peaks were collected, lyophilized and redissolved
tion of skin type I AFPs. An unrelated species, cunner in 0.01 M NH4HCO3 for activity measurements.
(Tautogolabrus adspersus), is known to have measurable The purified proteins were separated on 16.5% tricine
antifreeze activity in epithelial tissue but no AFPs have polyacrylamide gels in a Tris–Tricine buffer system (Bio-
been isolated to date [11]. The assumption has been that Rad Laboratories, Mississauga, ON) and stained with 0.1%
cunner survives in part by depending on an epidermis that CoomassieR brilliant blue R-250 [12]. A polypeptide
is fortified by AFPs to provide a barrier to ice [3,11]. standard (Bio-Rad Laboratories) was run in each gel to
Isolation of AFPs from cunner skin would confirm the estimate the approximate molecular weight of AFPs. Ami-
earlier results and would give credence to the argument no acid analysis and mass spectrometry (MS) were per-
that skin-type proteins are ubiquitous in all species pro- formed on HPLC-purified protein samples. For
ducing AFPs. comparison, individual protein bands were also cut out
of SDS-PAGE gels and purified using Ultrafree-MCR
Centrifugal Filter Units and Zip-TipsR (Millipore Ltd.,
2. Materials and methods Nepean, ON) prior to amino acid and mass spectrometry
analyses. Amino acid analyses and mass spectrometry
2.1. Tissue sample collection (Micromass ESI-QqTOF) were performed by the Ad-
vanced Protein Technology Centre (Hospital for Sick
Twelve Atlantic snailfish (L. atlanticus) and two cunner Children, Toronto, ON).
(T. adspersus) were collected by divers near Logy Bay,
Newfoundland, Canada in the winter of 2000. Two sea 2.3. Measurement of antifreeze activity
raven (H. americanus) were collected by divers in the
winter of 1998. Live fish were brought into the laboratory Antifreeze activity was measured as thermal hysteresis
and placed into holding tanks supplied with ambient using a Clifton Nanolitre Osmometer (Clifton Technical
temperature seawater. Prior to tissue collection, fish were Physics, Hartford, NY), following the procedure of [13].
anaesthetized using MS-222 and bled using a syringe and Thermal hysteresis is defined as the difference between the
needle containing heparin. For snailfish and sea raven, skin melting and freezing temperatures (in jC) of a test solution
epithelial tissue was peeled away from the body of the containing a minute ice crystal. Samples were dissolved in
anaesthetized fish, immediately frozen in liquid nitrogen 0.01 M NH4HCO3 and centrifuged before use. For each
and stored at 70 jC. In the case of the cunner, scales sample, measurements were made in triplicate, and the
containing epithelial tissue were scraped from the body average value taken.
using a knife blade, frozen in liquid nitrogen and stored at
70 jC. 2.4. RT-PCR and sequencing of sea raven AFP RNA

2.2. Isolation and purification of skin AFPs One microgram of DNase-treated total RNA from sea
raven skin and liver tissue was combined with 70 pmol of
Frozen skin tissues were first pulverized using a mortar an anchored poly-T primer and Superscriptk II RNase
and pestle containing liquid nitrogen prior and homogenized H Reverse Transcriptase (Invitrogen Canada Inc., Bur-
in 0.1 M NH4HCO3 (pH 8.5) using a Polytronk homoge- lington, ON) was used to generate first strand cDNA in a
nizer. Samples were then centrifuged at 5000 rpm for 1-h reaction at 42 jC, as described by the manufacturer.
10 min; supernatants were transferred to new containers The forward and reverse primers were designed to span
and subsequently lyophilized. The homogenates for anti- an intron/exon boundary and were based on the published
freeze activity measurements were removed both before and sequence of sea raven type II cDNA (see Fig 4) [14,15].
after lyophilization. After redissolving in 0.1M NH4HCO3, Normally, 1/10th of the RT reaction was combined with
total protein was assayed using a Modified Lowry Protein the primers and touchdown PCR amplification was per-
Assay kit (Pierce Biotechnology Inc., Rockford, IL) as formed using ELONGaseR Enzyme Mix polymerase
described by the manufacturer. Bovine serum albumin (Invitrogen Canada) in an Eppendorf MastercyclerR ther-
(BSA) was used as a standard reference protein. Normally mocycler. The touchdown cycling conditions consisted of
1.5-ml aliquots were applied to a Sephadex G-75 gel an initial 94 jC denaturing step (1 min), followed by 10
filtration column (0.9 60 cm) and eluted with 0.1 M cycles of 94 jC (15 s), 72 jC decreased to 60jC (15 s),
NH4HCO3. Protein fractions that exhibited antifreeze activ- 72 jC (60 s) and 25 more cycles of 94 jC (15 s), 60 jC

R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217 211




Fig. 1. Outline of the purification of snailfish skin AFPs. Panel (A) shows a typical Sephadex G-75 profile of crude skin homogenates with fractions that
contained antifreeze activity indicated. Horizontal bar indicates fractions collected for further HPLC purification. (B) Sephadex G-75-purified antifreeze from
skin homogenates were separated on a Nucleosil C8 column (see Section 2 for details). Approximately 6.4 mg of total protein was loaded onto the column with
f30% recovery from the column. The sole AFP peak found is labelled as Las-AFP. Panel (C) is the SDS-PAGE separation of L. atlanticus skin AFP. Lane 1,
pooled G-75 column fractions of skin AFPs (approximately 25 Ag of total protein); lane 2, HPLC-purified protein (approximately 10 Ag of protein); lane M,
molecular weight markers; lane 3, HPLC-purified snailfish plasma AFP; lane 4, HPLC-purified snailfish skin AFP.


(15 s), 72 jC (60 s). RT-PCR reaction products were Bands containing DNA were excised from the gel and
separated on 1% agarose gels and visualized using ethidium purified using CONCERTk Gel Extraction System (Invi-
bromide. trogen Canada) prior to cloning. The pGEMR-T Easy

212 R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217


Vector System was used to clone the purified RT-PCR Table 1

products for sequencing into a pGEMR-T Easy cloning Amino acid composition (mol%) and molecular mass of snailfish (L.
atlanticus) and cunner (T. adspersus) skin AFPs
vector, as described by the manufacturer (Promega). Se-
quencing was performed on at least three independent Amino acids Snailfish (Las-AFP) Cunner (Tas-AFP)

clones using M13 Forward and M13 Reverse primers at ASX 5.5 5.0

the DNA sequencing facility in The Centre for Applied GLX 4.9 1.1
SER 4.7 2.5
Genomics (Hospital for Sick Children, Toronto, ON).
GLY 3.7 3.4
ARG 2.4 5.8
THR 10.8 7.3

3. Results ALA 45.9 54.1
PRO 2.9 2.8
VAL 4.9 1.8
Crude homogenates prepared from skin tissue from all
ILE 2.1 1.2
three fish species contained antifreeze activity that could be LEU 4.1 3.1
purified for identification of unknown antifreeze proteins. PHE – 1.1
After an initial centrifugal step to remove insoluble debris, LYS 4.1 8.2
Molecular mass (Da)a 9344, 9415 (major) 7009 (major)
snailfish skin homogenate had a thermal hysteresis of 0.18
9457, 9387, 9501 6993, 6961
jC while cunner skin was 0.17 jC and sea raven was 0.19
a Based on ESI-MS analysis of HPLC peaks.
jC. Following a concentration step to half of their original
volumes, thermal hysteresis activity of the snailfish skin
homogenate increased to 0.26 jC, cunner to 0.22 jC and sea
raven to 0.24 jC—an average increase of 35%. The new 3.2. Purification and analysis of cunner skin AFP
homogenates were then used in the further purification of
AFPs. Homogenized cunner skin was initially fractionated on
Sephadex G-75 gel filtration columns and the active frac-
3.1. Purification and analysis of snailfish skin AFP tions were collected and further purified using HPLC. Only
a single peak collected from HPLC retained antifreeze
Partially purified protein from Sephadex G-75 gel filtra- activity, which was designated as Tas-AFP (Tautogolabrus
tion chromatography could be further resolved by HPLC adspersus skin AFP; Fig. 2A,B). The active HPLC peak was
into a single major peak designated as Las-AFP (Liparis run on SDS-PAGE and shown to have a strong major band
atlanticus skin AFP) and a few minor peaks (Fig. 1A,B). with an Mr of f 6200 and some very faint larger bands
Although all peaks were initially collected and analyzed for indicating that the column had removed many impurities
the presence of antifreeze activity, only the major one from the protein fraction (Fig. 2C). Analysis by mass
contained activity. Based on the SDS-PAGE results, it spectrometry determined that the HPLC peak contained a
appeared that the collected HPLC fraction was purified to major protein with a molecular mass of 7009 Da and two
homogeneity since there was a single band on the gel with minor ones (Table 1). Based on mass spectrometry deter-
an Mr of f 6500 (Fig. 1C). However, analysis by mass mination, the molecular mass of the most prominent band
spectrometry determined that there were actually five dif- excised from the gel was identical to the major HPLC peak.
ferent proteins within the HPLC peak; two major proteins The size of the cunner skin protein is larger than all type I
(molecular mass, 9344 and 9415 Da) and three minor ones AFPs reported to date with the exception of snailfish and
(see Table 1). Analysis of the excised SDS-PAGE band gave shorthorn sculpin skin AFPs [6,10].
similar results. Alanine is the most prominent amino acid with just
The two major skin proteins have identical molecular over 54 mol% of the total, which is consistent with type I
masses compared to the type I AFPs previously isolated AFPs (Table 1). Threonine and lysine levels are also
from L. atlanticus plasma [10]. Previously, it was deter- similar to type I AFPs reported before from sculpin skin
mined that these two proteins differ by a single alanine and the percentage of proline suggests that the protein
residue at their amino terminal ends and the discrepancy contains two or three of these residues. Preliminary MS/
between electrophoretic and mass spectrometry data is MS sequence analysis of peptides prepared from the
likely a direct consequence of the structure of these proteins excised gel band showed that the antifreeze contained
[10]. The amino acid content of the snailfish skin AFPs is the sequence AAAATAEAA. The cunner skin AFPs pro-
typical of all type I AFPs in that they have very high alanine duced ice crystals that had the usual hexagonal bipyrami-
composition, around 46 mol% (Table 1). The abundance of dal shape (Fig. 2B).
alanine residues is slightly lower than the plasma AFP but
the content of other amino acids such as threonine and 3.3. Purification and analysis of sea raven skin AFP
proline is quite similar. The AFPs purified from snailfish
skin give the typical hexagonal bipyramidal shape to ice Sea raven skin homogenates that contained antifreeze
crystals when cooling in the hysteresis gap (Fig. 1B). activity were initially fractionated on Sephadex G-75 gel

R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217 213




Fig. 2. Outline of the purification of cunner skin AFPs. Panel (A) shows a typical Sephadex G-75 profile of crude skin homogenates with fractions that
contained antifreeze activity indicated. Horizontal bar indicates fractions collected for further HPLC purification. (B) Sephadex G-75-purified antifreeze from
skin homogenates were separated on a Nucleosil C8 column (see Section 2 for details). Approximately 5.2 mg of total protein was loaded onto the column with
f30% recovery from the column. The sole AFP peak found is labelled as Tas-AFP. Panel (C) is the SDS-PAGE separation of T. adspersus skin AFP. Lane 1,
pooled G-75 column fractions of skin AFPs (approximately 25 Ag of total protein); lane 2, HPLC-purified protein (approximately 10 Ag of protein).



filtration columns and the active fractions were collected gel (Fig. 3C). The results indicated that while impurities
and purified by HPLC (Fig. 3A,B). A large HPLC peak were removed from column fractions by HPLC, there
containing antifreeze activity (Has-AFP; Hemitripterus appeared to be two individual bands—a prominent one with
americanus skin AFP), which was mixed with several an Mr f 18000 and a smaller, fainter, band of f 14000.
smaller peaks, was collected and analyzed on an SDS-PAGE When the collected HPLC peak was analyzed by mass

214 R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217




Fig. 3. Outline of the purification of sea raven skin AFPs. Panel (A) shows a typical Sephadex G-75 profile of crude skin homogenates with fractions that
contained antifreeze activity indicated. Horizontal bar indicates fractions collected for further HPLC purification. (B) Sephadex G-75-purified antifreeze from
skin homogenates were separated on a Nucleosil C8 column (see Section 2 for details). Approximately 5.8 mg of total protein was loaded onto the column with
f30% recovery from the column. The sole AFP peak found is labelled as Has-AFP. Panel (C) is the SDS-PAGE separation of H. americanus skin AFP. Lane
1, pooled G-75 column fractions of skin AFPs (approximately 25 Ag of total protein); lane 2, HPLC-purified protein (approximately 10 Ag of protein).


spectrometry, the two molecular masses were determined to separately for antifreeze activity before this is absolutely
be 18345 and 14006 Da (Table 2). The HPLC-purified sea certain.
raven skin AFPs produced ice crystals that had shapes Results of amino acid analysis of the isolated proteins
similar to previous reports for sea raven plasma AFPs showed they had elevated cystine but only f 12% alanine.
(Fig. 3B). While it is likely that both gel bands correspond Clearly they were not type I AFPs but more resembled type
to true sea raven AFPs, they would need to be analyzed II AFPs that sea raven synthesize in liver for circulation in

R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217 215


Table 2 flounder [5] in addition to shorthorn and longhorn sculpins
Amino acid composition (mol%) and molecular mass of sea raven (H. [6,7].
americanus) skin and liver AFPs
The discovery that the type I AFPs expressed in snailfish
Amino acids Sea raven SR-liver SR-liver skin tissue are identical to their major plasma proteins was
(Has-AFP) (163 AA)a (129 AA)a
completely unexpected [10]. Although it is practically
ASX 12.8 8.6 8.5 impossible to completely avoid blood plasma contamination
GLX 10.5 8.0 8.5
when isolating proteins from epithelial tissues, caution was
SER 7.3 6.7 7.0
GLY 8.7 7.4 7.8 used when removing the skin in order to minimize blood
HIS 3.3 2.5 3.1 contamination. Moreover, we have data which demonstrate
ARG 3.4 1.8 2.3 that snailfish express AFPs in epithelial tissues by skin-
THR 7.3 8.6 7.8 specific genes [18]. While it is obvious that snailfish plasma
ALA 12.2 13.5 13.2
AFPs are extracellular, it is not clear whether some skin
PRO 6.6 4.9 5.4
TYR 2.0 1.2 1.6 proteins remain intracellular or are exported to blood as a
VAL 4.3 4.3 3.1 source of circulating AFPs. It is also uncertain exactly how
CYS 5.8 6.7 7.8 snailfish AFPs are secreted from the epithelial cells that
ILE 3.3 3.1 3.1 express them since their corresponding mRNA does not
LEU 6.6 8.0 6.2
contain requisite signal sequences [18]. Alternative path-
PHE 2.5 1.5 2.3
Molecular mass (Da) 18345, 14006b 17469a 13993a ways for protein export that circumvent the usual endoplas-

a mic reticulum–Golgi complex have been described
Based on published protein sequence [14,15].
b Based on ESI-MS analysis of HPLC peaks. previously [19,20]. This surprising result from snailfish
has generated numerous questions concerning the differen-
tial expression of liver- and skin-type AFPs in general.
blood plasma. When compared to the published amino acid With the isolation and partial characterization of AFPs
content of the plasma type II AFP, there was considerable from cunner skin tissues, we have confirmed an earlier
similarity between the skin and plasma AFPs (Table 2). report from Valerio et al. [11] that cunner have antifreeze
Furthermore, it is known that the circulating AFP in sea activity in skin tissues. It is apparent that cunner skin AFPs
raven is 129 amino acids long (14 kDa) and is derived from are type I since amino acid analysis and MS/MS sequencing
an initial 163-amino-acid translation product that is 17.5 kDa showed that they have an alanine-rich primary structure.
[16]. These two molecular masses also correspond well with Moreover, their high alanine content would be indicative of
the two proteins isolated from skin tissue. RT-PCR was used an a-helical secondary structure, which is also characteristic
to determine if any AFP mRNA expressed in the skin tissue of type I AFP.
was related to the known liver sequence using primers from Cunner are from the order Perciformes, which is not
the published cDNA sequence (Fig. 4A). The identity of the closely related to the other known orders of fish producing
sea raven Type II AFP mRNA was confirmed since only two skin-type AFPs. These results provide more evidence that
nucleotide differences were observed between the RT-PCR skin AFPs might be ubiquitous across all teleost orders that
result and published cDNA sequence, and their translation contain AFPs. Other data (Fletcher et al., unpublished
products were identical (Fig. 4B). results) indicate that cunner also have type I AFPs circulat-
ing in their blood, although at this time it is unclear how
these are related to the skin localized protein. While the
4. Discussion evidence presented here is informative, it would be neces-
sary to clone the corresponding cDNA, to determine the
Based on the previous results, we determined that skin nature of possible amino acid repeats which could be used to
tissue would be an appropriate source from which to isolate help clarify its relationship to other type I AFPs from skin.
and characterize novel AFPs. The results reported here Results here demonstrate that sea raven skin tissues
confirm that snailfish skin tissue contains antifreeze activity contain antifreeze activity that could be purified by gel
that could be purified by standard chromatography techni- chromatography and HPLC to two individual bands on an
ques. Although antifreeze activity of skin tissue is measur- SDS-PAGE gel. Amino acid and mass spectrometric anal-
ably low, we have data that show that even a small quantity yses indicated that these proteins are nearly identical to the
of AFPs can be effective at controlling ice growth since their previously identified circulating type II AFPs from sea
activity is enhanced by co-occurring salts [18]. The purified raven. The mature plasma sea raven AFP—129 amino acids
proteins, which gave a single peak on an SDS-PAGE gel, (14.0 kDa)—is produced from a primary 163-amino-acid
were characterized as type I AFPs based on their high levels translation product (17.5 kDa) synthesized in liver [16]. A
of alanine. This is another example of a fish from the 146-amino-acid (16 kDa) proAFP intermediate is processed
superorder Acanthopterygii which expresses type I AFP in to mature AFP through pro-peptide cleavage during or soon
epithelial tissues. Three species from this superorder were after its release into blood circulation. Given that the
previously known to produce skin type I AFPs—the winter molecular mass of the major AFP found in skin tissue

216 R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217




Fig. 4. RT-PCR results and cDNA sequence of sea raven skin AFP mRNA. (A) Lanes 1 and 2 are duplicate samples of total skin RNA; C1 and C2 are RT-PCR
controls. (B) Sequence comparison between published sea raven type II AFP cDNA and RT-PCR results.



corresponds with the hepatic expressed pre-proprotein, it is for the pre-proprotein. Since the primers spanned an intron/
unlikely to represent contaminating blood protein since no exon boundary, the product generated by RT-PCR was not
significant amount of the unprocessed AFP exists in blood due to DNA contamination.
[16]. Furthermore, RT-PCR experiments confirmed that skin Up to now it has been assumed that sea raven type II
tissue does express the requisite mRNA necessary to code AFPs are expressed specifically in liver tissue since no

R.P. Evans, G.L. Fletcher / Biochimica et Biophysica Acta 1700 (2004) 209–217 217


expression of type II AFP mRNA could be detected in skin [4] C.L. Hew, D.S. Yang, Protein interaction with ice, Eur. J. Biochem.

or gill tissue of two sample fish [17]. Evidence from snail- 203 (1992) 33–42.
[5] Z. Gong, K.V. Ewart, Z. Hu, G.L. Fletcher, C.L. Hew, Skin antifreeze
fish, however, indicates that there can be significant tissue
protein genes of the winter flounder, Pleuronectes americanus, en-
variability in AFP mRNA expression between individual code distinct and active polypeptides without the secretory signal and
fish [18] and another study has shown that individual fish prosequences, J. Biol. Chem. 271 (1996) 4106–4112.
from one population of New Brunswick ocean pout had [6] W.K. Low, M. Miao, K.V. Ewart, D.S. Yang, G.L. Fletcher, C.L. Hew,

demonstrable differences in antifreeze gene copies [21]. Skin-type antifreeze protein from the shorthorn sculpin, Myoxocepha-
lus scorpius. Expression and characterization of a Mr 9700 recombi-
Until expression analysis is performed on many individual
nant protein, J. Biol. Chem. 273 (1998) 23098–23103.
sea raven using blood cells and epithelial tissues, the extent [7] W.K. Low, Q. Lin, C. Stathakis, M. Miao, G.L. Fletcher, C.L. Hew,
and importance of the type II expression in skin will remain Isolation and characterization of skin-type, type I antifreeze polypep-
unresolved. tides from the longhorn sculpin, Myoxocephalus octodecemspinosus,

These results from sea raven represent the first clear J. Biol. Chem. 276 (2001) 11582–11589.
[8] W.K. Low, Q. Lin, K.V. Ewart, G.L. Fletcher, C.L. Hew, The skin-
report of the isolation of a class of AFP—other than type I—
type antifreeze polypeptides: a new class of Type I AFPs, in: K.V.
from skin tissues and the second example of a fish having Ewart, C.L. Hew (Eds.), Molecular Aspects of Fish and Marine Bio-
identical plasma and epithelial AFPs. These data provide logy, Fish Antifreeze Proteins, vol. 1, World Scientific Publishing,
more evidence that skin tissue is a common site of AFP Singapore, 2002, pp. 161–186.

expression and further demonstrate that AFPs can also be [9] W.B. Scott, M.G. Scott, Atlantic Fishes of Canada, University of
Toronto Press, Toronto, 1988.
expressed in skin tissues that include pre- and pro-sequen-
[10] R.P. Evans, G.L. Fletcher, Isolation and characterization of type I
ces, similar to liver-type proteins. While it is apparent that antifreeze proteins from Atlantic snailfish (Liparis atlanticus) and
unprocessed protein remains in skin tissue, it is also possible dusky snailfish (Liparis gibbus), Biochim. Biophys. Acta 1547
that the signal sequences are cleaved such that some protein (2001) 235–244.

is exported into blood. Clearly, sea raven can utilize [11] P.F. Valerio, M.H. Kao, G.L. Fletcher, Thermal hysteresis activity in
the skin of the cunner, Tautogolabrus adspersus, Can. J. Zool. 68
additional means to bolster their antifreeze complement
(1990) 1065–1067.
for protection from freezing during winter. [12] H. Schagger, G. von Jagow, Tricine–sodium dodecyl sulfate-poly-
Although the precise physiological function of epithelia acrylamide gel electrophoresis for the separation of proteins in the
expressed AFPs has not been unequivocally established, range from 1 to 100 kDa, Anal. Biochem. 166 (1987) 368–379.

they likely act as an additional extracellular barrier to ice- [13] M.H. Kao, G.L. Fletcher, N.C. Wang, C.L. Hew, The relationship
between molecular weight and antifreeze polypeptide activity in ma-
crystal propagation into peripheral tissues. More rigorous
rine fish, Can. J. Zool. 64 (1986) 578–582.
investigation is required before this puzzle can be unravelled [14] N.F. Ng, K.Y. Trinh, C.L. Hew, Structure of an antifreeze polypeptide
fully. However, these results provide more evidence that precursor from the sea raven, Hemitripterus americanus, J. Biol.
expression of AFPs in fish skin is a very widespread Chem. 261 (1986) 15690–15695.

phenomenon that is not restricted to type I proteins alone. [15] P.H. Hayes, G.K. Scott, N.F. Ng, C.L. Hew, P.L. Davies, Cystine-rich
type II antifreeze protein precursor is initiated from the third AUG
codon of its mRNA, J. Biol. Chem. 264 (1989) 18761–18767.
[16] B.P. Duncker, S.Y. Gauthier, P.L. Davies, Evidence for a proprotein
Acknowledgements intermediate during maturation of type II antifreeze protein in sea
raven, Hemitripterus americanus, Biochim. Biophys. Acta 1292

We thank M. King and Dr. M. Shears at the OSC for (1996) 312–316.
[17] Z. Gong, G.L. Fletcher, C.L. Hew, Tissue distribution of fish anti-
technical assistance and the OSC divers for sample
freeze protein mRNAs, Can. J. Zool. 70 (1992) 810–814.
collection. We also thank Dr. Ming Kao for help with [18] R.P. Evans, Characterization of Skin and Plasma Type I Antifreeze
antifreeze activity measurements. This work was supported Proteins From Atlantic (Liparis Atlanticus) and Dusky (Liparis Gib-
by a grant from NSERC. bus) Snailfish, (PhD Thesis), Memorial University of Newfoundland,
St. John’s, Newfoundland, 2003.
[19] P. Mignatti, T. Morimoto, D.B. Rifkin, Basic fibroblast growth factor,
a protein devoid of secretory signal sequence, is released by cells via a
References pathway independent of the endoplasmic reticulum–Golgi complex,
J. Cell. Physiol. 151 (1992) 81–93.
[1] G.L. Fletcher, S.V. Goddard, P.L. Davies, Z. Gong, K.V. Ewart, C.L. [20] R.P. Menon, R.C. Hughes, Determinants in the N-terminal domains of
Hew, New insights into fish antifreeze proteins: physiological signif- galectin-3 for secretion by a novel pathway circumventing the endoplas-
icance and molecular regulation, in: H.O. Po¨rtner, R.C. Playle (Eds.), mic reticulum–Golgi complex, Eur. J. Biochem. 264 (1999) 569–576.
Society for Experimental Biology Seminar Series, Cold Ocean Phys- [21] C.L. Hew, N.C. Wang, S. Joshi, G.L. Fletcher, G.K. Scott, P.H.
iology, Cambridge Univ. Press, New York, 1998, pp. 240–265. Hayes, B. Buettner, P.L. Davies, Multiple genes provide the basis
[2] K.V. Ewart, Q. Lin, C.L. Hew, Structure, function and evolution of for antifreeze protein diversity and dosage in the ocean pout, Macro-
antifreeze proteins, Cell Mol. Life Sci. 55 (1999) 271–283. zoarces americanus, J. Biol. Chem. 263 (1988) 12049–12055.
[3] G.L. Fletcher, C.L. Hew, P.L. Davies, Antifreeze proteins of teleost
fishes, Annu. Rev. Physiol. 63 (2001) 359–390.


Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25
www.elsevier.com/locate/cbpb




Purification and characterization of lysozyme from plasma of the eastern
oyster (Crassostrea virginica)

Qing-Gang Xuea, Kevin L. Scheyb, Aswani K. Voletyc, Fu-Lin E. Chud, Jerome F. La Peyrea,*

a Cooperative Aquatic Animal Health Research Program, Department of Veterinary Science, Louisiana State University Agricultural Center,
111 Dalrymple Building, Baton Rouge, LA 70803, USA
b Department of Cell and Molecular Pharmacology, Medical University of South Carolina, Charleston, SC 29425, USA
c
Division of Ecological Studies, Florida Gulf Coast University, Fort Myers, FL 33965, USA
d Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA 23062, USA

Received 23 January 2004; received in revised form 24 May 2004; accepted 25 May 2004




Abstract


Lysozyme was purified from the plasma of eastern oysters (Crassostrea virginica) using a combination of ion exchange and gel filtration
chromatographies. The molecular mass of purified lysozyme was estimated at 18.4 kDa by SDS-PAGE, and its isoelectric point was greater
than 10. Mass spectrometric analysis of the purified enzyme revealed a high-sequence homology with i-type lysozymes. No similarity was
found however between the N-terminal sequence of oyster plasma lysozyme and N-terminal sequences of other i-type lysozymes, suggesting
that the N-terminal sequences of the i-type lysozymes may vary to a greater extent between species than reported in earlier studies. The
optimal ionic strength, pH, cation concentrations, sea salt concentrations, and temperature for activity of the purified lysozyme were
determined, as well as its temperature and pH stability. Purified oyster plasma lysozyme inhibited the growth of Gram-positive bacteria (e.g.,
Lactococcus garvieae, Enterococcus sp.) and Gram-negative bacteria (e.g., Escherichia coli, Vibrio vulnificus). This is a first report of a
lysozyme purified from an oyster species and from the plasma of a bivalve mollusc.
D 2004 Elsevier Inc. All rights reserved.

Keywords: Eastern oyster; Bivalve mollusc; Plasma protein; Lysozyme; Amino acid sequence; Biochemical properties; Antibacterial properties; Invertebrate
immunity




1. Introduction Scheltinga, 1996; Fastrez, 1996; Holtje, 1996; Hultmark,
1996; Jolles, 1996; Jolles et al., 1996; Prager, 1996). There
Lysozymes are a group of proteins defined as 1,4-h-N- has been increasing interest in recent years in the
acetylmuramidases (EC 3.2.1.17). These enzymes cleave the distribution and characterization of invertebrate i-type
glycosidic bond between N-acetylmuramic acid and N- lysozymes which include lysozymes of bivalve molluscs
acetylglucosamine of peptidoglycan, a major component of (Bachali et al., 2002; Olsen et al., 2003; Takeshita et al.,
bacterial cell walls (Salton, 1957; Jolles, 1969; Chipman 2003; Zavalova et al., 2003; Bachali et al., 2004).
and Sharon, 1969). Several types of lysozyme (e.g., c-, g-, i- Lysozyme activity has been detected in the body fluids
types) which differ in their amino acid composition, and tissues of many bivalve molluscs and is believed to play
biochemical and antimicrobial properties, and gene sequen- a role in host defense and digestion (McDade and Tripp,
ces have been identified in a wide range of organisms from 1967; Rodrick and Cheng, 1974; McHenery et al., 1986;
bacteriophages to humans (Beintema and Terwisscha van Takahashi et al., 1986; Chu and La Peyre, 1989; Maginot et
al., 1989; Allam et al., 2000; Cronin et al., 2001).
Lysozymes of several bivalve molluscs have been purified

* Corresponding author. Tel.: +1 225 578 5419; fax: +1 225 578 4890. mostly from parts of the digestive system, such as the
E-mail address: jlapeyre@agctr.lsu.edu (J.F. La Peyre). crystalline style and visceral mass (McHenery and Birk-

1096-4959/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.cbpc.2004.05.011

12 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25



beck, 1979; Myrnes and Johansen, 1994; Jolles et al., 1996; and (7) determine the antibacterial properties of the purified
Ito et al., 1999; Nilsen et al., 1999; Miyauchi et al., 2000; lysozyme(s).
Montenegro-Ortega and Viana, 2000; Olsen et al., 2003).
Their molecular masses range from 11 to 22 kDa, and the
analysis of their N-terminal amino acid sequences, when 2. Materials and methods
determined, indicated that they belong to a distinct type of
lysozyme, the i-type (Jolles et al., 1996; Ito et al., 1999; 2.1. Chemicals
Nilsen et al., 1999; Miyauchi et al., 2000; Olsen et al.,
2003). Their biochemical properties vary between species, Sephadex G-25 (Superfine), CM-Sepharose Fast Flow,
and at least one lysozyme from the Manila clam, Tapes and Superdex G-75 media were purchased from Amersham
japonica, was reported to possess isopeptidase activity in Pharmacia Biotech (Piscataway, NJ). Chemicals used for
addition to 1,4-h-N-acetylmuramidase and chitinase activity sodium-dodecylsulfate-polyacrylamide gel electrophoresis
(McHenery and Birkbeck, 1982; Viana and Raa, 1992; (SDS-PAGE), electroblotting and isoelectric were purchased
Myrnes and Johansen, 1994; Ito et al., 1999; Nilsen et al., from Sigma-Aldrich (Saint Louis, MO).
1999; Miyauchi et al., 2000; Takeshita et al., 2003).
The presence of lysozyme activity in cell-free haemo- 2.2. Oysters
lymph (plasma) of the eastern oyster (Crassostrea virginica)
was first demonstrated by McDade and Tripp (1967). The Eastern oysters (C. virginica), 10–15 cm in shell length,
plasma contained an enzyme which (1) caused a reduction were collected from the coast of Louisiana between
in the turbidity of bacterial cell walls, (2) liberated reducing November 2001 and March 2002. They were transported
sugars, and (3) liberated a complex containing glucosamine to Louisiana State University, Baton Rouge, and maintained
and muramic acid, therefore fulfilling the criteria of Salton in a 1000-l recirculating seawater system at a salinity of 15
(1957) and Jolles (1964) for the designation of the enzyme ppt and a temperature of 15 8C. Haemolymph was sampled
as a lysozyme. While the effects of pH and various salts on from individual oysters within a week after their transfer to
the lytic activity on Micrococcus lysodeikteicus as well as the recirculating seawater system.
the temperature stability of the plasma lysozyme have been
determined (McDade and Tripp, 1967; Rodrick and Cheng, 2.3. Haemolymph sampling
1974), basic knowledge of biochemical and antimicrobial
properties of this enzyme in purified form is lacking. The Oyster haemolymph was withdrawn from the adductor
eastern oyster is an important commercial bivalve species muscle sinus with a 3-ml syringe equipped with a 25 gauge
along the Atlantic and Gulf of Mexico coasts of North needle through a notch on the dorsal side of the shell.
America and is threatened by diseases (Ford and Tripp, Haemolymph from about 200 oysters were pooled and
1996). The development of a procedure to purify lysozyme centrifuged at 500g for 15 min at 4 8C. Supernatant
from the plasma of eastern oysters will allow investigations (plasma, 800 ml) was collected and stored at 20 8C for
of its potential role in the oyster host defense. To our lysozyme purification.
knowledge, no lysozyme has been purified from the cell-
free haemolymph (plasma) of any bivalve mollusc. Lyso- 2.4. Lysozyme activity and protein concentration
zyme(s) purified from plasma may differ from those purified
from the digestive system in biochemical and antimicrobial Lysozyme activity was measured in 96-well plates by
properties because of their respective putative role in host mixing in each well 20 Al of sample with 180 Al of
defense and digestion. Lysozymes with different molecular Micrococcus lysodeikticus bacterial suspension prepared at
weights and biochemical properties have for instance been a concentration of 0.8 mg/ml in appropriate buffer solutions.
recently purified from blue mussels (Mytilus edulis; Olsen et Buffer solutions were selected according to the requirements
al., 2003). of the individual experiments described later in the text.
The objectives of this study were to (1) purify Absorbance of wells was immediately measured at 450 nm
lysozyme(s) from the plasma of eastern oysters, (2) estimate with a microtiter plate reader (Dynatec, Chantilly, VA).
the molecular mass(es) and isoelectric point(s) of the Absorbance was measured 5 min after the initial reading,
purified lysozyme(s), (3) determine the N-terminal amino and the decrease in absorbance at 450 nm/min was
acid sequence(s) of the purified lysozyme(s) by automatic calculated. Assays were performed at room temperature
Edman degradation, (4) analyze the purified lysozyme(s) by (RT, 20 8C) unless otherwise indicated. In this study, one
mass spectrometry to compare primary amino acid sequen- unit of lysozyme was defined as that quantity which caused
ce(s) with other proteins and to verify molecular mass(es), a decrease in absorbance of 0.001/min of M. lysodeikticus
(5) determine the optimal ionic strength, pH, cation suspended in 0.18 M (I=0.180) acetate buffer at pH 5.5. All
concentrations, sea salt concentrations, and temperature lysozyme measurements in 96-well plates were done in
for activity of the purified lysozyme(s), (6) determine triplicates. Sample protein concentration was measured
temperature and pH stability of the purified lysozyme(s), using the Micro BCA Protein Assay Reagent Kit from

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 13



Pierce Biotech (Rockford, IL). All protein measurements in was subjected to SDS-PAGE as described below. Fractions
96-well plates were done in triplicates. showing a single protein band were pooled, and the salts in
the solution were eliminated by gel filtration in a Sephadex
2.5. Lysozyme purification G-25 column (2.630 cm) equilibrated with Millipore
water. The protein obtained by washing the column with
2.5.1. Sample preparation Millipore water was designated as dpurified oyster
Oyster plasma (800 ml) was freeze-dried (FTS systems, lysozymeT. It was concentrated to 2 mg/ml by ultrafiltration
Stone Ridge, NY) and resuspended in 80 ml of Millipore using a Centriprep YM-10 filter at 2800g, 4 8C, and stored
water. The preparation was clarified by centrifugation at at 20 8C as stock solution until use.
3000g for 30 min at 4 8C. The supernatant was divided
into two 40-ml samples, and these were applied onto 2.6. Molecular mass by SDS-PAGE and isoelectric point
Sephadex G-25 columns (2.630 cm) equilibrated with determination
0.02 M sodium acetate buffer, pH 5.0. The columns were
then washed with 0.02 M sodium acetate buffer, pH 5.0, at a The purity and approximate molecular mass of the
linear superficial flow rate (LSFR) of 60 cm/h. The elution lysozyme sample were estimated by SDS-PAGE under
was monitored for its absorbance at 280 nm with an Econo reduced condition in a protean III vertical slab gel unit (Bio-
UV Monitor from Bio-Rad Laboratories. Fractions collected Rad, Richmond, CA) with a 12.5% running gel and a 4%
in the first peak were pooled. This pool contained plasma stacking gel. The low-range (14.4–97.4 kDa) protein
proteins including lysozyme in 0.02 M sodium acetate molecular mass markers from Bio-Rad were used as
buffer, pH 5.0, and was designated as dcrude plasma standards to calculate molecular mass. The isoelectric point
lysozyme sampleT. of the purified lysozyme was determined by isoelectric
focusing in a MINI IEF Cell (Model 111, Bio-Rad) using
2.5.2. Initial ion (IE) exchange chromatography isoelectric point standards (4.45–9.6) from Bio-Rad for
The crude plasma lysozyme sample was loaded onto a reference.
CM-Sepharose Fast Flow column (1.630 cm) equili-
brated with 0.02 M sodium acetate buffer, pH 5.0. The 2.7. N-terminal amino acid sequencing
column was successively washed with 0, 0.1, 0.3, and 0.6
M of NaCl in 0.02 M sodium acetate buffer, pH 5.0, at an The purified lysozyme in the solution was sent to the
LSFR of 60 cm/h. The elution was monitored for Protein Chemistry Laboratory of the University of Texas
absorbance at 280 nm. Fractions from the 0.6 M NaCl Medical Branch, Galveston, for N-terminal amino acid
eluted peak, which contained lysozyme activity, were sequencing. The N-terminal amino acid sequence was
pooled and concentrated by centrifugation at 2800g at analyzed by automatic Edman degradation using an Applied
48C using Centriprep YM-10 filters (Millipore, Bedford, Biosystems Procise 494/HT protein sequencer (Applied
MA). The pooled sample was designated as dlysozyme- Biosystems, Foster City, CA) after reduction and alkylation
enriched IE sampleT. of the sample. The sample was reduced in 50 Al of 6 M
guanidine–HCl, 0.25 M Tris buffer at pH 8.5, 1 mM EDTA
2.5.3. Gel-filtration (GF) chromatography in the presence of 2.5 Al of 10% h-mercaptoethanol (Hawke
The lysozyme-enriched IE sample was applied to a and Yuan, 1987). The reduction was allowed to proceed for
Superdex G-75 column (1.660 cm) equilibrated with 0.1 2 h at room temperature under argon in the dark. One
M sodium acetate buffer, pH 5.0. The column was eluted microliters of 4-vinylpyridine was added, and incubation
with the same buffer at an LSFR of 30 cm/h. The elution was continued for an additional 20 min in the dark at RT
was monitored for its absorbance at 280 nm, and the under argon (Hawke and Yuan, 1987; Andrews and Dixon,
fractions of the second peak with lysozyme activity were 1987; Tempst et al., 1990). The sample was cleaned prior to
pooled. The buffer of the pooled fractions was changed to sequencing on a prosorb sample preparation cartridge
0.02 M acetate buffer, pH 5.0, using Sephadex G-25 (Applied Biosystems). Another sample of the purified
columns (2.630 cm), and the pooled fractions were lysozyme on PVDF membrane was also submitted for N-
designated as dlysozyme GF-separated sampleT. terminal amino acid sequencing. Following an SDS-PAGE
as described earlier, the lysozyme was electroblotted onto
2.5.4. Final ion-exchange chromatography Sequi-blotk PVDF membrane (Bio-Rad) in CAPS buffer
Lysozyme GF-separated sample was loaded onto a CM- (0.01M CAPS, 10% methanol, pH 11) using a Bio-Rad Mini
Sepharose Fast Flow column (0.810 cm). The column was TransblotR Electrophoretic cell (Bio-Rad). The membrane
washed with a linear gradient of NaCl, 0.3 to 0.7 M in 0.02 was stained with Coomassie blue R-250 to locate lysozyme
M sodium acetate buffer, pH 5.0, at an LSFR of 60 cm/h. on the membrane. The PVDF membrane with the lysozyme
The elution was monitored for its absorbance at 280 nm. was cut, washed six times in Millipore water, and sent to the
The lysozyme activity of each fraction was tested, and an Protein Chemistry Laboratory of the University of Texas
aliquot from each fraction containing high lysozyme activity Medical Branch. Sequence similarity between the N-

14 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25



terminal amino acid sequence of oyster plasma lysozyme accomplished on a Voyager-DE STR instrument (Applied
and proteins in the GenBank databases of the National Biosystems) operating in linear mode.
Center for Biotechnology Information (NCBI) was inves- The resulting peptide sequences were subjected to
tigated using the BLAST program. BLAST searches of the National Center for Biotechnology
Information (NCBI) GenBank database to ascertain
2.8. Primary amino acid sequencing using tandem mass sequence homology. The peptide sequences were then
spectrometry and molecular mass determination by matrix- aligned with sequences of four bivalve mollusc lysozymes
assisted laser desorption ionization (MALDI) aligned according to Bachali et al. (2002). The four
sequences were for lysozymes from the cold-seep clam,
A 5-Al aliquot of purified lysozyme (555 pmol) was Calyptogena sp. (Bachali et al., 2002; GenBank accession
reduced with 50 Al of 5 mg/ml dithiothreitol in 6 M no. AF334667), from the Mediterranean mussel, Mytilus
guanidine–HCl, 1.5 M Tris, pH 8.4 (buffer A) at 37 8C for galloprovincialis (Bachali et al., 2002; GenBank accession
35 min followed by alkylation with 50 Al of 15 mg/ml no. AF334665), from the Icelandic scallop, Chlamys
iodoacetamide in buffer A at 37 8C for 45 min. Excess islandica (Nilsen and Myrnes, 2001; GenBank accession
reagents were removed by step elution over a 2.1100 mm no. CAC34834), and from the Manila clam, T. japonica (K.
C18 Brownlee Aquapore column (Perkin-Elmer, Boston, Takeshita et al., unpublished, GenBank accession no.
MA). After sample injection, the column was washed with AB091383).
5% acetonitrile, 0.1% trifluoroacetic acid for 5 min at a flow
of 200 Al/min. Lysozyme was eluted with 85% acetonitrile, 2.9. Determination of pH and ionic strength optima and pH
0.1% TFA. Absorbance was monitored at 214 nm, and stability
collected fractions dried in a speed vac.
The reduced and alkylated lysozyme-containing fraction The lytic activity of purified lysozyme on M. lysodeikti-
was solubilized in 100 mM ammonium bicarbonate buffer, cus was measured at 12 pHs (3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5,
pH 7.8 (50 Al). Trypsin (100 ng) was added, and the 7.0, 7.5, 8.5, 9.5, and 10.5) and 10 ionic strengths (0.005,
digestion allowed to proceed for 18 h at 37 8C. Tryptic 0.010, 0.020, 0.040, 0.070, 0.100, 0.140, 0.180, 0.220,
peptides were desalted using C18 ZipTips (Millipore) and 0.260, and 0.280). A total of 120 buffer solutions were
eluted with 2 Al nanospray solvent [water/acetonitrile/acetic prepared with sodium hydroxide–acetic acid (pH 3.5–5.5),
acid (49:49:2, v/v/v)]. Na2HPO4–NaH2PO4H2O (pH 6.0–8.5), and boric acid–
.

Nanospray mass spectrometry was carried out on a NaOH (9.5–10.5). The pHs of all buffers were measured
tandem quadrupole/time-of-flight mass spectrometer and adjusted to the expected pH whenever slight pH
(QSTAR, Applied Biosystems) equipped with a Protana deviation (b0.1) was found from the calculated pH. The
nanospray source. Approximately 1.75 Al of the tryptic buffers were used to dilute M. lysodeikticus stock suspension
digest solution was loaded into custom-pulled, gold- (16 mg/ml in water) to 0.8 mg/ml, and the activity of
coated quartz capillaries. Tryptic peptides were selected
for sequencing by tandem mass spectrometry according to
their observed molecular ions. Molecular ions were
selected and subjected to collision-induced dissociation
in the collision quadrupole filled with nitrogen. The
product ion spectra were recorded with the high-
resolution time-of-flight mass spectrometer. Typically
100 spectra were accumulated to produce data of high
signal-to-noise ratio. In addition, tandem time-of flight-
mass spectrometry (Applied Biosystems 4700 Proteomics
Analyzer) was carried out to obtain confirmatory and
additional sequence information. A 0.50-Al aliquot of the
desalted tryptic digest was mixed (1:3, v/v) with matrix
(a-cyano-4-hydroxycinnamic acid, 50 mM in 70% aceto-
nitrile, 0.1% TFA). Matrix-assisted laser desorption
ionization (MALDI) with an Nd-YAG laser was used to
ionize tryptic peptides, and ions of interest were selected
for collision-induced dissociation. Typically 200–5000
laser shots were used to generate the product ion spectra.
Manual interpretation of both Q-TOF and TOF-TOF data
Fig. 1. SDS-PAGE of purified oyster (C. virginica) plasma lysozyme and
was carried out. Purified lysozyme was prepared for
protein molecular markers after staining with Coomassie blue. The
MALDI analysis as described above to obtain the molecular mass of oyster lysozyme under reducing conditions was
molecular mass of the intact protein. This analysis was estimated at 18.4 kDa. Both lanes contained about 2.5 Ag of protein.

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 15




Fig. 2. Purification of plasma lysozyme from eastern oysters (C. virginica) by ion exchange and gel filtration chromatographies. (A) Ion exchange
chromatography of crude plasma lysozyme sample on CM-Sepharose FF column after a stepwise elution with NaCl. Fractions covered by peak IV were
recovered and pooled as dlysozyme-enriched IE sampleT. (B) Gel filtration chromatography of lysozyme-enriched IE sample on Superdex G-75 column.
Fractions in peak II were collected and pooled as dlysozyme GF-separated sampleT. (C) Ion exchange chromatography of lysozyme GF-separated sample on
CM-Sepharose FF column. The column was eluted with a linear gradient of NaCl from 0.3 M to 0.7 M in 0.02 M acetate buffer at pH 5.0. Fractions from peaks
IV and V were pooled and designated dpurified lysozymeT, because they showed a single protein band with the same molecular weight by SDS-PAGE and
Coomassie blue staining. Eluted proteins (solid line) were monitored at an absorbance of 280 and lysozyme activity (broken line) was measured against M.
lysodeikticus suspension.

16 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25


Table 1
Summary of the oyster (C. virginica) lysozyme purification

Sample Total protein (mg) Total activity (U) Specific activity (U/mg protein) Yield (%)

Crude plasma lysozyme sample 4240.0 7.55105 1.78102
Lysozyme-enriched IE sample 28.8 5.16105 1.79104 68.34
Lysozyme GF-separated sample 6.4 4.00105 6.25104 52.98
Purified oyster lysozyme 1.0 1.52105 1.52105 20.13


lysozyme at 5 Ag/ml was measured as described earlier. ionic strength of 0.115, 0.130, 0.145, 0.160, 0.175, 0.190,
Results were expressed as percent activity of the highest 0.205, 0.220, 0.235, 0.250, and 0.265 was used as reference
activity measured in the experiment. All measurements in for changes in ionic strength caused by the addition of
microplates were done in triplicates. CaCl2 or MgCl2. All of these solutions were used to prepare
The pH stability was tested by diluting the lysozyme M. lysodeikticus suspension. The lytic activity of lysozyme
stock solution to 50 Ag/ml in buffers of pH 2.0 to 13.0 in at a final concentration of 5 Ag/ml on M. lysodeikticus
increments of one pH unit. All buffers had ionic strength of suspensions was measured in microplate in triplicates as
I=0.180. Following incubation for 10 and 30 min, respec- described above. Results were expressed as the percent of
tively, at each pH, the lysozyme was diluted to 5 Ag/ml in the activity of lysozyme on M. lysodeikticus in 0.1 M
0.18 M ammonium acetate buffer, pH 5.5, and the activity ammonium acetate buffer at pH 5.5 and an ionic strength of
measured as described earlier. Results were expressed as 0.100 which was taken to represent 100%.
percent activity with the activity from lysozyme diluted The effects of seawater were tested by measuring the
directly to 5 Ag/ml with 0.18 M ammonium acetate buffer, lytic activity of purified oyster lysozyme (5 Ag/ml in water)
pH 5.5, as 100%. All microplate measurements were done on M. lysodeikticus in artificial seawater diluted to 2, 4, 6, 8,
in triplicates. 10, 15, 20, 25, or 30 parts per thousand (ppt). The salinity of
full-strength seawater is about 35 ppt (i.e., 35 g of sea salts
2.10. Determination of the effects of cations and seawater per kilogram of water). Eastern oysters thrive in estuarine
waters between 5 and 15 ppt and have a wide tolerance to
The effects of Na+, Ca2+, and Mg2+ on oyster lysozyme salinity ranging from 5 to 40 ppt (Galtsoff, 1964; Berrigan et
activity were determined by adding NaCl, CaCl2, or MgCl2 al., 1991). Artificial seawater (pH 8.5) was prepared with
at different concentrations to M. lysodeikticus in 0.1 M hw Marinemix Professional sea salts (Hawaiian Marine
ammonium acetate buffer at pH 5.5 and an ionic strength of Imports, Houston, TX), and the salinities adjusted with a
0.100. The selection of a buffer at suboptimal ionic strength refractometer (Aquatic Ecosystems, Apopka, FL). The
(0.100) allowed the demonstration of changes in activity measurements were performed in microplates and results
due to salts, above or below an intermediary level of expressed as percent of the activity of lysozyme on M.
activity. NaCl was added to the final concentrations of lysodeikticus in 0.18 M ammonium acetate buffer at pH 5.5.
0.025, 0.050, 0.075, 0.100, 0.125, 0.150, 0.200, 0.250, All measurements were performed in triplicates.
0.300, 0.350, and 0.400 M, and CaCl2 or MgCl2 were added
to the final concentrations of 0.005, 0.010, 0.015, 0.020, 2.11. Determination of temperature optimum, thermal
0.025, 0.030, 0.035, 0.040, 0.045, 0.050, and 0.055 M. Two stability, and comparison with chicken egg white lysozyme
sets of ammonium acetate buffer at pH 5.5 were also
prepared with increasing concentrations of ammonium The effects of temperature on oyster lysozyme were
acetate to be used as reference buffers for the increase in determined using 1.5-ml-capacity disposable microcuvettes
ionic strength. One set of buffers with ionic strengths of with 1 cm light path, and absorbance was measured with a
0.125, 0.150, 0.175, 0.200, 0.225, 0.250, 0.400, 0.450, and Shimadzu UV 600 spectrophotometer (Shimadzu, Kyoto,
0.500 was used as a reference for changes in ionic strength Japan). This larger volume was used to minimize temper-
caused by the addition of NaCl. Another set of buffers with ature variation during measurements of lysozyme activity



Fig. 3. Amino acid sequence of purified oyster (C. virginica) plasma lysozyme. (A) N-terminal sequence determined by automatic Edman degradation. Purified
lysozyme was submitted to Edman degradation analysis both in liquid form and after being electroblotted onto PVDF membrane. The resulting sequences from
the two preparations were identical. N-terminal sequences of bivalve species other than C. virginica were according to references of Jolles (1996) for
Calyptogena sp. and M. galloprovincialis, Nilsen et al. (1999) for C. islandica, and Ito et al. (1999) for T. japonica. (B) Sequences of oyster lysozyme tryptic
peptides determined by tandem mass spectrometry. (C) Alignment of amino acid sequences of purified plasma lysozyme with that of four bivalve mollusc
species. Peptide sequences were subjected to BLAST searches of the National Center for Biotechnology Information (NCBI) GenBank database. Seven oyster
lysozyme peptide sequences were aligned with the sequences of four bivalve mollusc lysozymes obtained from GenBank, from Calyptogna sp. (Bachali et al.,
2002—GenBank accession no. AF334667), from the Mediterranean mussel, M. galloprovincialis (Bachali et al., 2002—GenBank accession no. AF334665),
from the Iceland scallop, C. islandica (Nilsen et al., 1999—GenBank accession no. CAC34834), and from the Manila clam, T. japonica (K. Takeshita et al.,
unpublished—GenBank accession no. AB091383) aligned by Bachali et al. (2002). The boxed region encompasses the aligned sequences. Identical residues
shared by all sequences are in boldface. The first three N-terminal amino acids determined by Edman degradation are shaded.

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 17



at different temperatures. The enzyme temperature opti- reaction was initiated by mixing 0.1 ml of purified
mum was determined by diluting M. lysodeikticus stock lysozyme (2 Ag/ml in water) with 1.4 ml of M.
suspension to an absorbance of 0.7 at 450 nm with 0.18 M lysodeikticus suspension at each temperature and measur-
ammonium acetate buffer at pH 5.5 and equilibrated to ing the absorbance at 450 nm every 20 s for 2 min. The
temperatures of 0 to 70 8C in increment of 5 8C. The decrease in absorbance per minute was calculated and used

18 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25



to compare activities at different temperatures. Results buffer were analyzed by paired t tests. All other data were
were expressed as percent activity relating to the highest analyzed by one- and two-factor analysis of variance,
activity in the experiment. All measurements were followed by SNK’s multiple comparison of means when
performed in triplicates. significant differences (pb0.05) were found.
Thermal stability of the purified lysozyme was tested by
incubating lysozyme (2 Ag/ml in water) at 20, 40, 60, 80, or
100 8C for 10 and 30 min separately. The lytic activity on 3. Results
M. lysodeikticus was then measured at 20 8C. The results
were expressed as the percent of the activity of lysozyme 3.1. Lysozyme purification, molecular mass by SDS-PAGE
maintained on ice during the duration of the experiment and and isoelectric point
measured at 20 8C. All measurements were performed in
triplicates. A protein with high lytic activity against M. lysodeik-
The lytic activities of purified oyster lysozyme (2 Ag/ml ticus was purified from oyster plasma by a combination
in water) and chicken egg white lysozyme (HEWL, 10 Ag/ of ion exchange and gel filtration chromatographies. The
ml in water) on M. lysodeikticus were compared at 10, 15, protein appeared as a single band with a molecular mass
20, 25, 30, 35, and 40 8C using 1.5-ml-capacity disposable of 18.4 kDa determined by SDS-PAGE and Coomassie
microcuvettes as described above. The enzyme concen- blue staining (Fig. 1). Sixty eight percent of the lysozyme
trations were selected to give the same activity for oyster from the crude plasma lysozyme sample was recovered
lysozyme and HEWL at 20 8C. Results were expressed as following the first ion exchange chromatography, and the
percent activity relating to the activity of each enzyme at specific activity of the lysozyme enriched IE sample
20 8C. (peak IV, Fig. 2A) was 100-fold greater than that of the
crude plasma lysozyme sample (Table 1). This first ion
2.12. Antibacterial activities exchange chromatography was quite effective as a first
purification step, because the bulk of the nonlysozyme
The concentrations of oyster lysozyme inhibiting the plasma proteins was eluted during sample loading and
growth of three Gram-positive bacteria (Lactococcus column washing with 0.02 M sodium acetate buffer at
garvieae, Streptococcus iniae, Enterococcus sp.) and four pH 5.0 (peak I, Fig. 2A; whole peak range was not
Gram-negative bacteria (Escherichia coli, Vibrio vulnifi- shown). Some proteins bound to CM-Sepharose Fast
cus, Aeromonas hydrophila, Edwarsiella ictaluri) were Flow were washed down with 0.1 and 0.3 M NaCl, but
determined. Bacterial species were obtained from Dr. John no lysozyme activity was detected in these fractions
Hawke, Dr. Richard Cooper, or the late Dr. Ronald (peak II and III, Fig. 2A). Gel filtration chromatography
Siebeling at the Louisiana State University, Baton Rouge. of the lysozyme-enriched IE sample yielded lysozyme
All bacteria were grown in nutrient broth containing 5 g GF-separated sample with a 3.5 greater specific activity
beef extract, 2 g neopeptone, 0.1 g bactose dextrose, 1 g (Table 1, peak II, Fig. 2B). Finally, several peaks were
yeast extract, and 10 g NaCl per liter of water and were observed when proteins of the lysozyme GF-separated
harvested in log phase. The bacteria were resuspended in sample were purified by ion exchange using a linear
phosphate buffer saline (PBS) to a density of about 107 NaCl gradient elution (Fig. 2C). Lysozyme activity was
bacteria/ml, and 20 Al were added to 20 Al of twofold detected in peaks III–V (Fig. 2C). Fractions from peaks
serially diluted lysozyme (400–0.4 Ag/ml) in PBS or to 20 IV and V were pooled and designated purified oyster
Al of PBS alone (control) in 96-well plates in duplicate lysozyme, because they showed a single protein band
wells. After 2 h of incubation at RT, 160 Al of nutrient with the same molecular mass by SDS-PAGE and
broth were added to each well, and the plates were Coomassie blue staining. Fractions from peak III showed
incubated at 28 8C. Bacterial growth was measured at 640 three proteins and were not used. The purified lysozyme
nm with a microtiter plate reader (Chantilly, VA) at 12 h had a specific activity 854 times greater than the specific
for E. coli and V. vulnificus, 24 h for A. hydrophila and activity of the crude oyster plasma preparation (Table 1).
E. ictaluri, and 36 h for the slower growing Gram- About 1.0 mg of purified lysozyme was obtained from
positive L. garvieae, S. iniae, and Enterococcus sp. 800 ml of oyster plasma which represented 20% of
Results were expressed as the minimum concentration of lysozyme from the original plasma sample (Table 1). The
lysozyme which significantly inhibited bacterial growth isoelectric point of purified lysozyme was greater than
compared to control (PBS only). The experiment was 10, the highest isoelectric point that could be measured
repeated twice. the ampholytes used.

2.13. Statistical analysis 3.2. N-terminal amino acid sequence

Data on the effects of each cation on lysozyme activity The N-terminal sequence of the purified oyster plasma
compared to the activity of lysozyme in appropriate control lysozyme was analyzed to the 43rd amino acid residue by

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 19




Fig. 4. Molecular mass determination of purified oyster (C. virginica) lysozyme by MALDI-MS. Purified lysozyme was mixed with matrix and molecular mass
measurements made by MALDI-MS. Two signals, m/z 17771.5 and m/z 17861.0, were observed in a broad peak indicating possibly two forms of the protein.


automatic Edman degradation (Fig. 3A). Both the liquid approximately 60% of the lysozyme sequence based on
sample and the PVDF membrane sample showed the same the number of residues in homologous species (Fig. 3B).
amino acid sequence. No similarity was found between the Three tryptic peptides contained sequences identified by
N-terminal sequence of oyster plasma lysozyme and other Edman degradation, and when the Edman sequence is
sequences from GenBank. combined with the tandem mass spectrometry sequence,
approximately 70% of the sequence is covered. The
3.3. Primary amino acid sequence using tandem mass protein sequence data reported in this paper will appear
spectrometry in the SWISS-PROT and TrEMBL knowledgebase under
the accession no. P83673. The high-sequence homology
Mass spectrometric analysis yielded high-quality observed allowed positive identification of a lysozyme via
sequence information on nine tryptic peptides covering BLAST searching of interpreted sequences (Fig. 3C). Note




Fig. 5. Activity of purified oyster (C. virginica) plasma lysozyme as a function of pH and ionic strength. Data were obtained on the activity of purified
lysozyme in 120 buffers covering a pH ranging from 3.5 to 10.5 and ionic strength ranging from I=0.005 to 0.280. A maximum lysozyme specific activity of
1.76105 U/mg protein was observed at a combination of pH 5.9 and I=0.180. Activities were expressed as a percentage of that observed at maximum activity
and a contour plot of lysozyme activity in 10% increment was generated.

20 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25



that tandem mass spectrometry, as performed in this study, much broader pH and ionic strength ranges. Lysozyme
cannot distinguish between isoleucine/leucine and lysine/ retained 60% of its maximum activity within the pH
glutamine residues; therefore, assignments were made on range of 5.0–7.5 and ionic strength range of I=0.070–
sequence homology. Two molecular ions 17771.5 and 0.260 (Fig. 5). Preincubation of purified lysozyme in
17861.0 Da were observed by MALDI for purified buffers of pH 2–13 for 10 or 30 min had no effect on its
lysozyme, indicating two potential forms of the enzyme activity.
(Fig. 4).
3.5. Effects of cations and seawater
3.4. pH and ionic strength optima
Lysozyme activity increased as the NaCl concentration
Optimal pH and ionic conditions for the purified was increased from 0 to 0.1 M and then decreased from
lysozyme were observed at pHs between 5.5 and 6.0 and its maximum activity at 0.1 M NaCl with further increase
ionic strengths between I=0.180 and 0.200. Within this in NaCl concentrations (Fig. 6A). The lowest activity was
range of pHs and ionic strengths, the purified lysozyme detected at NaCl concentrations greater than 0.25 M (Fig.
expressed more than 90% of its maximum activity (Fig. 6A). The same trend was observed when lysozyme
5). Lysozyme activity remained relatively high within a activity was measured in the reference ammonium acetate




Fig. 6. Effects of cations and seawater on the activity of the purified oyster (C. virginica) plasma lysozyme. (A) Effects of Na+ and (B) effects of Ca2+ and
Mg2+: in both cases, effects were observed by measuring the activity of lysozyme in 0.1 M ammonium acetate buffer at pH 5.5 supplemented with NaCl,
CaCl2, or MgCl2 at different concentrations. Controls were two sets of ammonium acetate buffer at pH 5.5 with the ionic strengths corresponding to the buffers
after addition of Na+, Ca2+, or Mg2+. The 100% activity in both experiments represented a specific lysozyme activity of 7.4104 U/mg protein in 0.1 M
ammonium acetate buffer at pH 5.5. (C) Effects of diluted seawater at different salinities; full-strength seawater has a salinity of 35 ppt. The activity of
lysozyme in 0.18 M ammonium acetate buffer at pH 5.5 was used to represent 100% and had a specific activity of 1.42105 U/mg protein.

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 21



buffers. The activity of purified lysozyme measured in
reference buffers was the same as the activity of lysozyme
measured in buffers containing NaCl at each ionic
strength (Fig. 6A). In contrast, the activity of lysozyme
in buffers containing CaCl2 and MgCl2 were significantly
higher than the activity of lysozyme in the reference
buffers at each ionic strength ( pb0.01; Fig. 6B).
Lysozyme activity in CaCl2-containing buffers increased
by 45.38–99.22% relative to lysozyme activity in refer-
ence buffers. The activity of lysozyme in MgCl2-contain-
ing buffers increased by 19.01–86.05% relative to the
activity of lysozyme in reference buffers. Maximum
lysozyme activity was measured in buffers with an ionic
strength of I=0.205 regardless of the presence or absence
of CaCl2 or MgCl2 (Fig. 6B).
No lytic activity was detected when the purified
lysozyme was in distilled water (Fig. 6C). Lysozyme
activity was highest when measured in seawater diluted
to 2 ppt. At 2 ppt, lysozyme activity was about 20%
higher than lysozyme activity in the reference buffer
(0.18 M ammonium acetate, pH 5.5). Lysozyme activity
decreased rapidly with further increase in seawater
salinity. Lysozyme activity measured at 10 ppt was
reduced to 20%, and activity was not detected at 25
ppt (Fig. 6C).

3.6. Temperature optimum, thermal stability, and temper-
ature effects

Activity of the purified lysozyme increased with increas-
ing temperature from 0 to 45 8C and decreased markedly at
temperatures greater than 55 8C (Fig. 7A). No decrease in
activity was measured after incubation of lysozyme at 20 8C
for 10 and 30 min, and no activity was detected after a 30
min incubation of lysozyme at 100 8C (Fig. 7B). In the
temperature range of 10 to 40 8C, the activity of oyster
lysozyme was significantly less affected by temperature
change than the activity of hen egg white lysozyme (Fig.
7C). The activity of oyster lysozyme at 10 8C was
68.3%F2.7% of the activity at 20 8C, and at 40 8C,
138.9%F2.2% of its activity at 208C (Fig. 7C). In contrast,
the activity of hen egg white lysozyme at 10 and 40 8C was
51.7%F3.5% and 170.2%F5.5%, respectively, of the
activity at 20 8C.

3.7. Antibacterial activities

Purified lysozyme at concentrations of 0.8 and 3.1 Ag/ Fig. 7. Effects of temperature on the activity of purified oyster (C.
virginica) plasma lysozyme. Assays were carried out using 1.5-ml-capacity
ml significantly inhibited the growth of two Gram-positive
disposable microcuvettes with a 1 cm light path in a Shimadzu UV 600
bacteria, L. garvieae and Enterococcus sp., respectively. spectrophotometer. M. lysodeikticus stock suspension was diluted to an
The purified lysozyme at concentrations of 6.3 and 25 Ag/ absorbance of 0.7 at 450 nm in ammonium acetate buffer at 0.18 M and pH
ml significantly inhibited the growth of two Gram-negative 5.5 to measure lysozyme activity. (A) Optimal temperature: the maximum

bacteria, E. coli and V. vulnificus, respectively. The growth activity was observed at 40 8C, which represented 100% activity. (B)
Thermal stability: lysozyme activity at 20 8C was used to represent 100%
of Aeromona hydrophila was inhibited only at a high
activity. (C) Comparison of activities of purified oyster lysozyme and hen
concentration of 400 Ag/ml. No growth inhibition of S. egg white lysozyme (HEWL) between 10 and 40 8C; activity level at 20 8C
iniae and Edwardsiella ictaluri by purified lysozyme was for both lysozymes were used to represent 100% activity.

22 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25


Table 2 preceding the sequence—GMVSQKCLRCICNVES—that
Antibacterial activities of purified oyster (C. virginica) lysozyme shows a strong similarity to the N-terminal amino acid
Minimum inhibitory sequences of other i-type lysozymes. The 49 amino acids
concentration (Ag/ml) account for about 30% of the molecular mass of oyster
Gram-positive plasma lysozyme and explains its higher molecular mass
Lactococcus garvieae 0.8
compared to all other i-type lysozymes except for some of
Enterococcus sp. 3.1
the multiple forms of blue mussel (M. edulis) lysozymes. It
Streptococcus iniae N400
Gram-negative will be interesting to compare the N-terminal amino acid
Escherichia coli 6.3 sequence of oyster plasma lysozyme to that of the largest
Vibrio vulnificus 25 (18, 22 kDa) purified blue mussel lysozymes when their N-
Aeromona hydrophila 400
terminal sequences are eventually determined (McHenery
Edwardsiella ictaluri N400
and Birkbeck, 1979; Olsen et al., 2003).
Mass spectrometry (i.e., MALDI-TOF) analysis of oyster
noted at the maximum concentration tested (400 Ag/ml; lysozyme indicated two potential forms of the enzyme with
Table 2). molecular masses of 17771.5 and 17861 Da. Multiple forms
of lysozymes which showed different biochemical proper-
ties have been recently identified in blue mussels and the
4. Discussion medicinal leech, Hirudo medicinalis (Olsen et al., 2003;
Zavalova et al., 2003). The consistent N-terminal amino
A protein with high lytic activity against M. lysodeikticus acid sequence of our sample and the absence of secondary
was purified from the plasma of eastern oysters by a activity optimum under a wide range of conditions,
combination of ion exchange and gel filtration chromatog- however, suggest the two forms of the enzyme are similar
raphies as indicated by SDS-PAGE and N-terminal sequenc- in biochemical properties and differ in sequence by only one
ing. The purified protein lytic activity against M. or a few amino acids. This is in contrast to secondary pH
lysodeikticus, low molecular mass, high isoelectric point, and ionic strength optima observed for lysozymes isolated
and heat stability suggested the protein was a lysozyme from blue mussels and the Manila clam, Ruditapes
(Jolles, 1969). Mass spectrometric sequence analysis con- philippinarum (T. japonica synonym; McHenery and
firmed that the purified protein was a lysozyme and that it Birkbeck, 1982; Maginot et al., 1989). Future research is
belonged to the invertebrate type of lysozymes (i-type). This needed to separate the two potential forms of lysozyme to
is a first report of a lysozyme purified from an oyster species test our hypothesis.
and from the plasma of a bivalve mollusc. The isoelectric point of our purified lysozyme was
No similarity was found between the N-terminal amino greater than 10. The high isoelectric point is typical for
acid sequence of purified oyster plasma lysozyme and N- most lysozymes which are usually basic proteins (Jolles and
terminal amino acid sequences of other purified i-type Jolles, 1984). Data on the isoelectric points of bivalve
lysozymes (Jolles and Jolles, 1975; Jolles et al., 1996; mollusc lysozymes are limited. The isoelectric point of
Fradkov et al., 1996; Ito et al., 1999; Nilsen et al., 1999). lysozyme purified from blue mussels was 9.2 (McHenery
Primary amino acid sequencing of the purified protein using and Birkbeck, 1979) and that of lysozyme purified from
tandem mass spectrometry, however, confirmed the N- pismo clams (Tivela stutorum) was 7.7 (Montenegro-Ortega
terminal amino acid sequence obtained by Edman degrada- and Viana, 2000). In contrast to our results, Feng (1974)
tion. Moreover, the sequence of a 2536.6-MW tryptic reported that lysozyme-like activity from the plasma of
peptide (i.e., CCVPSSSNSGSFSTGMVSQKCLR) we eastern oysters was mostly associated with acidic proteins of
obtained, overlapped with the N-terminal amino acid different electrophoretic mobilities. It is likely that oyster
sequence of our purified protein determined by Edman plasma lysozyme, which has a very high isoelectric point
degradation (i.e., CCVPSSSN–) and showed sequence and constitutes a fraction of oyster plasma proteins as
homology with i-type lysozymes (i.e.,–GMVSQKCLR). determined in our study, may have associated with plasma
Our study suggests the N-terminal amino acid sequence of i- acidic proteins in Feng’s study and as a result carried an
type lysozymes may vary to a greater extent between overall negative charge (McHenery and Birkbeck, 1979).
species than reported in earlier studies and may not always Hen egg white lysozyme, for example, is known to strongly
be a reliable criterion to identify i-type lysozymes. It is associate with proteins with low isoelectric points (Essink et
noteworthy that the N-terminal sequences of i-type lyso- al., 1985). Alternatively, an acid form of oyster lysozyme
zymes described in earlier studies have all been for proteins may exist as reported for some i-type lysozymes. The
about a third smaller (~ 13 kDa) than oyster plasma isoelectric point predicted from the amino acid composition
lysozyme (~ 18 kDa; Jolles et al., 1996; Ito et al., 1999; of lysozyme purified from the Icelandic scallop C. islandica
Nilsen et al., 1999; Zavalova et al., 2003). Amino acid was 6.9 (Nilsen et al., 1999). Moreover, Zavalova et al.
sequencing of purified oyster lysozyme by Edman degra- (2003) recently purified a new acid form of destabilase
dation and mass spectrometry identified 49 amino acids lysozyme from medicinal leeches, while other forms of

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 23



destabilase lysozyme were basic proteins. The isoelectric tested (I=0.1–0.265). No such increase in lysozyme activity
points of i-type lysozymes are therefore quite variable and over the reference ammonium acetate buffer was measured
may be related to different functions of the protein (e.g., in the presence of Na+ (0.025–0.4 M). In either the presence
host defense, digestion). or absence of the cations tested in ammonium acetate buffer
Optimal pH and ionic conditions for the purified (pH 5.5), the activity of purified oyster lysozyme was
lysozyme were observed at pHs between 5.5 and 6.0 and always greatest when the final ionic strength was 0.2,
ionic strengths between I=0.180 and 0.200. This range falls thereby confirming our earlier results. Seawater diluted to
between the pH optima of 6.5 and 5.0 of eastern oyster 10 ppt, which has a similar ionic strength, markedly
plasma lysozyme activity reported in two earlier studies inhibited the activity of purified lysozyme. A large part of
(McDade and Tripp, 1967; Cheng and Rodrick, 1974). this inhibition was likely due to the high pH of the freshly
Differences in the type and molarity of buffers used in the made seawater used (i.e., pH 8.5). At pH 8.5, the optimal
earlier studies likely accounted for the variation in optimal ionic strength for purified lysozyme was about 0.07 and
pH. In our study, the optimal activity of the purified oyster much closer to the ionic strength of diluted seawater at 2
lysozyme at each pH tested varied markedly with the ionic ppt, a salinity at which the activity of purified lysozyme was
strength of the buffer used. The activity of purified greatest. Some inhibition of purified lysozyme by seawater
lysozyme at low pHs was highest at the highest ionic components may also have occurred. Lysozyme purified
strengths tested, while the activity of purified lysozyme at from the crystalline styles of blue mussels was also reported
high pHs was highest at the lower ionic strengths. A similar to be very sensitive to seawater (McHenery and Birkbeck,
phenomenon was observed with hen egg white lysozyme 1982). The activity of lysozyme in oysters, however, would
activity (Davies et al., 1969). The optimal pH for hen egg be greater than in seawater, because oyster hemolymph has
white lytic activity moved from approximately pH 5 to 9 as a lower pH than seawater.
the ionic strength decreased from I=0.2 to 0.02, and it was The activity of purified oyster lysozyme was optimal at
hypothesized that binding of lysozyme to the cell wall of M about 45 8C and in the range reported for other bivalve
.lysodeikticus requires a certain electrostatic condition mollusc lysozymes (McHenery and Birkbeck, 1982; Ito et
(Davies et al., 1969). This might be a general phenomenon al., 1999; Nilsen et al., 1999; Miyauchi et al., 2000;
for lysozymes (Saint-Blancard et al., 1970; Jensen and Montenegro-Ortega and Viana, 2000). This range varied
Kleppe, 1972). greatly from a low optimal temperature of 20 8C for
A unique feature of the purified oyster lysozyme activity lysozyme purified from Icelandic scallops (Nilsen et al.,
was its relatively high optimal ionic strength compared to 1999) to a high optimal temperature of 75 8C for lysozyme
hen egg white lysozyme (Davies et al., 1969). There is, purified from Manila clams (Ito et al., 1999). An interesting
however, limited information to compare our results to the feature of purified oyster lysozyme is that at 5 8C it retained
ionic strengths optima of purified lysozymes of other about 50% of its activity observed at 45 8C, indicating a low
bivalve molluscs. In most studies, in which the effects of Q10 value for this enzyme. This feature is similar to that
ionic strengths or molarities were determined, measure- observed for lysozyme of the Icelandic scallop (Nilsen et al.,
ments were not systematically made at different pHs (Ito et 1999). Change in temperature, for example, between 10 and
al., 1999; Nilsen et al., 1999; Montenegro-Ortega and 40 8C, had a much greater effect on the activity of hen egg
Viana, 2000; Viana and Raa, 1992). It is important to note white lysozyme than on the activity of lysozymes from
that the activity of purified oyster lysozyme remained either eastern oyster (this study) or from the Icelandic
relatively high within relatively broad pH and ionic strength scallop (Viana and Raa, 1992).
ranges. Purified oyster lysozyme retained about 60% of its Purified oyster lysozyme had antibacterial activities
maximum activity within an ionic strength range of against both Gram-positive and Gram-negative bacteria.
I=0.070–0.260 and pH range of 5.5–7.5. Activity of the Information on the antimicrobial properties of lysozymes
purified lysozyme in these ranges of ionic strength and pH purified from bivalve mollusks is limited to only two
would be advantageous for eastern oysters because (1) studies, but both reported antibacterial activities against
oyster plasma ionic strength can be expected to vary Gram-positive and Gram-negative bacteria (Nilsen et al.,
considerably, because oysters thrive in estuaries between 5 1999; Montenegro-Ortega and Viana, 2000). Lysozyme
and 15 ppt and are osmoconformers (Berrigan et al., 1991; purified from the Icelandic scallop at moderate concen-
Shumway, 1996); and (2) oyster haemolymph pH has been trations (2–10 AM) inhibited the growth of several Gram-
reported in the range of 6.8 to 8.0 (Jones et al., 1995; Boyd positive bacteria (Listeria monocytogenes, Bacilus cereus,
and Burnett, 1999). Staphylococcus epidermis, Enterococcus faecalis) and
The effects of a wide range of concentrations of Na+, Gram-negative bacteria (E. coli, Pseudomonas aeruginosa,
Ca2+, and Mg2+, which are the major cations in oyster Proteus mirabilis, Vibrio salmonicida), which are associated
plasma, were tested on purified oyster lysozyme activity. with infection diseases in humans and animals (Nilsen et al.,
Lysozyme activity in the presence of the divalent cations 1999). Montenegro-Ortega and Viana (2000) tested partially
Ca2+ or Mg2+ (0.005–0.055 M) was always greater than in purified lysozyme of the pismo clam (Tivela stultorum)
reference ammonium acetate buffer at each ionic strength against the Gram-positive bacteria Micrococcus luteus,

24 Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25



Staphylococcus aureus, and Streptococcus alfa, and the Bachali, S., Bailly, X., Jolles, J., Jolles, P., Deutsch, J.S., 2004. The

Gram-negative bacteria E. coli, Pseudomonas putrefaciens, lysozyme of the starfish Asteria rubens. A paradygmatic type i
lysozyme. Eur. J. Biochem. 271, 237–242.
and Vibrio parahaemolyticus, but only found antimicrobial
Beintema, J.J., Terwisscha van Scheltinga, A.C., 1996. Plant lysozymes. In:
activity against M. luteus, S. alfa, and E. coli. While Jolles, P. (Ed.), Lysozymes: Model Enzymes in Biochemistry and
lysozymes are generally considered more active against Biology. Birkhauser Verlag, Basel, pp. 75–86.
Gram-positive bacteria because their cell walls are largely Berrigan, M., Candies, T., Cirino, J., Dugas, R., Dyer, C., Gray, J.,

made of peptidoglycan (i.e., 90%), there is increasing Herrington, T., Keithly, W., Leard, R., Nelson, J.R., Van Hoose, M.,
1991. The Oyster Fishery of the Gulf of Mexico, United States: A
evidence that lysozymes are also active against Gram-
regional Management Plan. Gulf States Marine Fisheries Commission,
negative bacteria through mechanisms not related to their Ocean Springs, MS.
enzymatic activity (Pelligrini et al., 1992; During et al., Boyd, J., Burnett, L.E., 1999. Reactive oxygen intermediate production
1999; Ibrahim et al., 2001). It remains to be determined by oyster hemocytes exposed to hypoxia. J. Exp. Biol. 202,

whether oyster lysozyme antibacterial activity is related to 3135–3143.
Cheng, T.C., Rodrick, G.E., 1974. Identification and characterization of
its enzymatic activity or to other properties such as its
lysozyme from the hemolymph of the soft-shelled clam, Mya arenaria.
significant positive charge as proposed by Pelligrini et al. Biol. Bull. 147, 311–320.
(1992) for hen egg white lysozyme. Chipman, D.M., Sharon, N., 1969. Mechanism of lysozyme. Science 165,
In summary, an i-type lysozyme was purified from the 454–465.

plasma of eastern oysters, and some of its biochemical and Chu, F.-L.E., La Peyre, J.F., 1989. Effect of environmental factors and
parasitism on hemolymph lysozyme and protein of American oysters
antibacterial properties were characterized. This study is a
(Crassostrea virginica). J. Invertebr. Pathol. 54, 224–232.
first step towards elucidating the role this enzyme plays in Cronin, M.A., Culloty, S.C., Mulcahy, M.F., 2001. Lysozyme activity and
the eastern oyster’s host defense or digestion. The anti- protein concentration in the haemolymph of the flat oyster Ostrea edulis
bacterial properties of the purified oyster lysozyme observed (L.). Fish Shellfish Immunol. 11, 611–622.

in our study imply a possible role of plasma lysozyme in the Davies, R.C., Neuberger, A., Wilson, B.M., 1969. The dependence of
lysozyme activity on pH and ionic strength. Biochim. Biophys. Acta
oyster host defense, but further research is needed to
178, 294–503.
identify its antibacterial mechanism(s) of action. Future During, K., Porsch, P., Mahn, A., Brinkmann, O., Gieffers, W., 1999. The
identification of the lysozyme gene and characterization of non-enzymatic microbicidal activity of lysozymes. FEBS Lett. 449,
its expression will assist in identifying the protein function 93–100.

in eastern oysters. Essink, A.W., Arkesteijn, G.L., Notermans, S., 1985. Interference of
lysozyme in the sandwich enzyme-linked immunosorbent assay
(ELISA). J. Immunol. Methods 80, 91–96.
Fastrez, J., 1996. Phage lysozymes. In: Jolles, P. (Ed.), Lysozymes:
Acknowledgements Model Enzymes in Biochemistry and Biology. Birkhauser Verlag,
Basel, pp. 35–64.
We thank Chwan-Hong Foo, Yanli Li, and Kim Nguyen Feng, S.Y., 1974. Lysozyme-like activities in the haemolymph of

of the Department of Veterinary Science at Louisiana State Crassostrea virginica. In: Cooper, E.L. (Ed.), Contemporary Topics
in Immunobiology, Invertebrate Immunology, vol. 4. Plenum Press,
University for technical assistance. We thank Steve Smith at
New York, pp. 225–231.
the Protein Chemistry Laboratory of the University of Texas Ford, S., Tripp, M.R., 1996. Diseases and defense mechanisms. In:
Medical Branch, Galveston for N-terminal amino acid Kennedy, V.S., Newell, R.I.E. (Eds.), The Eastern Oyster Crassostrea
sequencing. We acknowledge the MUSC Mass Spectrom- virginica. Maryland Sea Grant College Program. University of Mary-

etry Facility for access to mass spectrometry instrumenta- land, College Park, MD, pp. 581–660.
Fradkov, S., Berezhnoy, S., Barsova, E., Zavalova, L., Lukyanov, S.,
tion. This research was funded by the Louisiana Sea Grant
Baskova, I., 1996. Enzyme from the medicinal leech (Hirudo
College Program and the National Sea Grant Oyster Disease medicinalis) that specifically spits endo-q(g-Glu)-Lys isopeptide
Research Program. Approved by the Director of the bonds: cDNA cloning and protein primary structures. FEBS Lett.
Louisiana Agricultural Experiment Station as manuscript 390, 145–148.

number 04-44-0298. Galtsoff, P.S., 1964. The American oyster Crassostrea virginica Gmelin.
US Fish Bull. 64, 1–480.
Hawke, D., Yuan, P., 1987. Applied Biosystems User Bulletin, vol. 28.
Holtje, J.-V., 1996. Bacterial lysozymes. In: Jolles, P. (Ed.), Lysozymes:
References Model Enzymes in Biochemistry and Biology. Birkhauser Verlag,
Basel, pp. 65–74.
Allam, B., Paillard, C., Auffret, M., 2000. Alterations in hemolymph and Hultmark, D., 1996. Insect lysozymes. In: Jolles, P. (Ed.), Lysozymes:
extrapallial fluid parameters in the Manila clam, Ruditapes philippina- Model Enzymes in Biochemistry and Biology. Birkhauser Verlag,
rum, challenged with the pathogen Vibrio tapetis. J. Invertebr. Pathol. Basel, pp. 87–102.
76, 63–69. Ibrahim, H.R., Thomas, U., Pelligrini, A., 2001. A helix-loop-helix peptide
Andrews, P.C., Dixon, J.E., 1987. A procedure for in situ alkylation of at the upper lip of the active site cleft of lysozyme confers potent
cystine residues on glass-fiber prior to protein microsequence analysis. antimicrobial activity with membrane permeabilization action. J. Biol.
Anal. Biochem. 161, 524–528. Chem. 276, 43767–43774.
Bachali, S., Jager, M., Hassanin, A., Schoentgen, F., Jolles, P., Fiala- Ito, Y., Yoshikawa, A., Hotani, T., Fukuda, S., Sugimura, K., Imoto, T.,
Medioni, A., Deutsch, J.S., 2002. Phylogenetic analysis of invertebrate 1999. Amino acid sequences of lysozymes newly purified from
lysozymes and the evolution of lysozyme function. J. Mol. Evol. 54, invertebrates imply wide distribution of a novel class in the lysozyme
652–664. family. Eur. J. Biochem. 259, 456–461.

Q.-G. Xue et al. / Comparative Biochemistry and Physiology, Part B 139 (2004) 11–25 25


Jensen, H.B., Kleppe, K., 1972. Effects of ionic strength, pH amines and Nilsen, I.W., Overbo, K., Sandsdalen, E., Sandaker, E., Sletten, K., Bjornar,
divalent cations on the lytic activity of T4 lysozyme. Eur. J. Biochem. M., 1999. Protein purification and gene isolation of chlamysin, a cold
28, 116–122. active lysozyme-like enzyme with antibacterial activity. FEBS Lett.
Jolles, P., 1964. Recent developments in the study of lysozymes. Angew. 464, 153–158.
Chem. 3, 28–36. Olsen, a.M., Nilsen, I.W., Sletten, K., Myrnes, B., 2003. Multiple
Jolles, P., 1969. A chapter in molecular biology. Angew. Chem. 8, 227–294. invertebrate lysozymes in blue mussel (Mytilus edulis). Comp.
Jolles, P., 1996. From the discovery of lysozyme to the characterization Biochem. Physiol. B136, 107–115.
of several lysozyme families. In: Jolles, P. (Ed.), Lysozymes: Model Pelligrini, A., Thomas, U., von Fellenberg, R., Wild, P., 1992. Bactericidal
Enzymes in Biochemistry and Biology. Birkhauser Verlag, Basel, activities of lysozyme and aprotinin against gram-negative and gram-
pp. 3–5. positive bacteria related to their basic character. J. Appl. Bacteriol. 72,
Jolles, J., Jolles, P., 1975. The lysozyme from Asteria rubens. Eur. J. 180–187.
Biochem. 54, 19–23. Prager, E.M., 1996. Adaptive evolution of lysozyme: changes in amino acid
Jolles, P., Jolles, J., 1984. What’s new in lysozyme research? Mol. Cell. sequence, regulation of expression and gene number. In: Jolles, P. (Ed.),
Biochem. 63, 165–189. Lysozymes: Model Enzymes in Biochemistry and Biology. Birkhauser
Jolles, J., Fiala-Medioni, A., Jolles, P., 1996. The ruminant digestion model Verlag, Basel, pp. 323–345.
using bacteria already employed early in evolution by symbiotic Rodrick, G.E., Cheng, T.C., 1974. Kinetic properties of lysozyme from the
mollusks. J. Mol. Evol. 43, 523–527. hemolymph of Crassotrea virginica. J. Invertebr. Pathol. 24, 41–48.
Jones, T.O., Whyte, J.N.C., Townsend, L.D., Ginther, N.G., Iwama, G.K., Saint-Blancard, J., Chuzel, P., Mathieu, Y., Perrot, J., Jolles, P., 1970.
1995. Effects of domoic acid on haemolymph pH, PCO2 and PO2 in the Influence of pH and ionic strength on the lysis of Micrococcus
Pacific oyster, Crassostrea gigas and the California mussel, Mytilus lysodeikticus cells by six human and four avian lysozymes. Biochim.
californianus. Aquat. Toxicol. 31, 43–55. Biophys. Acta 220, 300–306.
Maginot, N., Samain, J.F., Daniel, J.Y., Le Coz, J.R., Moal, J., 1989. Salton, M.R.J., 1957. The properties of lysozyme and its action on
Kinetic properties of lysozyme from the digestive glands of Ruditapes microorganisms. Bacteriol. Rev. 21, 82–99.
philippinarum. Oceanis 15, 451–464. Shumway, S.E., 1996. Natural environmental factors. In: Kennedy, V.S.,
McDade, J.E., Tripp, M.R., 1967. Lysozyme in the hemolymph of the Newell, R.I.E. (Eds.), The Eastern Oyster Crassostrea virginica.
oyster Crassostrea virginica. J. Invertebr. Pathol. 9, 531–535. Maryland Sea Grant College Program, University of Maryland, College
McHenery, J.G., Birkbeck, T.H., 1979. Lysozyme of the mussel, Mytilus Park, MD, pp. 467–513.
edulis (L.). Mar. Biol. Lett. 1, 111–119. Takahashi, K., Mori, K., Nomura, T., 1986. Occurrence and character-
McHenery, J.G., Birkbeck, T.H., 1982. Characterization of the lysozyme of ization of lysozyme from marine bivalves. Bull. Jpn. Soc. Sc. Fish. 52,
Mytilus edulis (L). Comp. Biochem. Physiol. B71, 583–589. 863–868.
McHenery, J.G., Allen, J.A., Birkbeck, T.H., 1986. Distribution of Takeshita, K., Hashimoto, Y., Ueda, T., Imoto, T., 2003. A small
lyzozyme-like activity in 30 bivalve species. Comp. Biochem. Physiol. chimerically bifunctional monomeric protein: Tapes japonica lysozyme.
B85, 581–584. Cell. Mol. Life Sci. 60, 1944–1951.
Miyauchi, K., Matsumiya, M., Mochizuki, A., 2000. Purification and Tempst, P., Link, A.J., Riviere, L.R., Fleming, M., Elicone, C., 1990.
characterization of lysozyme from brackishwater clam Corbicula Internal sequence-analysis of proteins separated on polyacrylamide gels
japonica. Nippon Suisan Gakkaishi 66, 275–281. at the submicrogram level-improved methods, applications and gene
Montenegro-Ortega, R.E., Viana, M.T., 2000. Biochemical and bacterio- cloning strategies. Electrophoresis 11, 537–553.
logical properties of the lysozyme of the clam Tivela stultorum. Cienc. Viana, M.T., Raa, J., 1992. Lysozyme-like enzyme from the scallop
Mar. 26, 225–251. Chlamys islandica. Cienc. Mar. 18, 93–107.
Myrnes, B., Johansen, A., 1994. Recovery of lysozyme from scallop waste. Zavalova, L.L., Artamonova, I.I., Berezhnoy, S.N., Tagaev, A.A., Baskova,
Prep. Biochem. 24, 69–80. I.P., Andersen, J., Roepstorff, P., Egorov, T.A., 2003. Multiple forms of
Nilsen, I.W., Myrnes, B., 2001. The gene of chlamysin, a marine medicinal leech destabilase-lysozyme. Biochem. Biophys. Res. Com-
invertebrate-type lysozyme, is organized similar to vertebrate chicken- mun. 306, 318–323.
type lysozyme genes. Gene 269, 27–32.


Journal of Chromatography B, 816 (2005) 175–181




A novel matrix for high performance afnity chromatography and its
application in the purication of antithrombin III

Rui Zhaoa , Jia Luoa , Dihua Shangguana, Guoquan Liua
, ,b

a Center for Molecular Science, Institute of Chemistry, Chinese Academy of Sciences, Zhongguancun, 100080 Beijing, P.R China
b Graduate School of Chinese Academy of Sciences, Beijing 100039, P.R. China

Received 1 July 2004; accepted 15 November 2004
Available online 7 December 2004




Abstract


Viscose ber, a regenerated cellulose, was evaluated for using as a novel matrix for high performance afnity chromatography. With a
one-step activation with epichlorohydrin, heparin can be readily covalently attached to the matrix. This heparin–viscose ber material was
used for purifying antithrombin III (AT III) from human plasma. The purity of the AT III from this one-step purication is 93% as measured
by SDS-PAGE and the protein recovery yield is about 90%. This column is highly specic as described by the dissociation constant of the
complex of immobilized heparin and AT III, which was 2.83 × 105 mol/L. And more important, this viscose ber material demonstrated its
excellent mechanical property that allows the ow rate to reach up to 900 cm/h or more.
2004 Elsevier B.V. All rights reserved.

Keywords: Viscose ber; High performance afnity chromatography; Heparin; Antithrombin III




1. Introduction support material is also critical important for advancements
in bioprocess technology [6].
Afnity chromatography is one of the most powerful tech- The matrix used for HPAC can be roughly divided into
niques in selective purication and isolation of a great num- two groups, namely, inorganic and organic media. Generally,
ber of compounds [1]. This technique has the purication the inorganic polymers, such as silica, have good mechan-
power to eliminate steps, increase yields and improve process ical stability and can be easily derived to introduce func-
economics [2]. High performance afnity chromatography tional groups [7], but suffer from poor chemical stability and
(HPAC) was introduced by Ohlson et al. [3], who combined high non-specic adsorption caused by its residual silanol
two chromatographic techniques of high performance liq- groups [8]. The synthetic organic polymers are suitable from
uid chromatography and afnity chromatography. Due to its a chemical stability point of view, but some of these possess
specicity, rapidity and high performance, HPAC has gained low biocompatibility due to their hydrophobic character, for
more and more attention and is being used increasingly for example, polystyrene [1].
large-scale bimolecular purications [4,5]. Success of HPAC Being an easily available natural polymer, cellulose
depends on many factors. The type of matrix is one of the has played an important role in afnity chromatography
most important factors. The physical and chemical properties [9,10]. Compared with the synthetic organic polymers,
of matrix constitute dominant effects on the chromatographic cellulose and its derivatives have hydrophilic surfaces and
performance. The availability of cost-effective and efcient are biocompatible [11,12]. Various cellulose media derived
from regenerated and non-regenerated cellulose are com-
mercially available. However, the high degree of endogenous
crystallinity in natural cellulose makes it less suitable for
Corresponding author. Tel.: +86 10 62557910; fax: +86 10 62559373.
E-mail address: zhaorui@iccas.ac.cn (R. Zhao). afnity chromatography. The crystalline regions possess


1570-0232/$ – see front matter 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jchromb.2004.11.030

176 R. Zhao et al. / J. Chromatogr. B 816 (2005) 175–181


much less accessible hydroxyl groups than amorphous of epoxy groups on the surface of epichlorohvdrin–viscose
regions and may exhibit non-specic adsorption as well as bers was measured by Keen’s method [14].
low derivatization ability. The regenerated cellulose does not Heparin sodium salt (373 mg) dissolved in 10 mL of
have such crystalline regions and possess more accessible 0.04 mol/L HCl was stirred at 100 C for 30min. After cool-
hydroxyl groups that reduce the effect of non-specic ing to room temperature, partially hydrolyzed heparin was
adsorption and enhance reactivity in chemical reactions. recovered by precipitation with four volumes of ethanol.
Covalent crosslinks in regenerated cellulose also improve its The mixture of epichlorohydrin modied viscose bers
rigidity and mechanical stability. (2.0 g) and partially hydrolyzed heparin (250 mg) in 20 mL
In the present work, viscose ber, a regenerated cellulose, of 0.5 mol/L Na2CO3 (pH 11) was incubated at 50 C for
was studied as a novel and potential matrix for HPAC. To in- 72 h with shaking. The obtained heparin–viscose bers were
vestigate its afnity chromatographic behaviors, heparin was thoroughly washed with water to remove unreacted heparin.
used as a model ligand. The afnity chromatographic behav- The residual epoxy groups on the heparin–viscose bers were
iorsofheparin–viscoseberswereevaluated.Theapplication blocked by reacting with 20 mL of 1.0 mol/L ethanolamine
of heparin–viscose bers in isolating antithrombin III from (pH 9.0) for 8 h.
human plasma was also investigated. The packing material for the control column was prepared
using the same procedures, which included a surface activa-
tionstepwithepichlorohydrin,acouplingstepwithoutthead-
2. Experimental
dition of heparin and an end-capping step with ethanolamine
as those used in the afnity column preparation.
2.1. Materials and equipment

The viscose ber was provided by Baoding Swan Chem- 2.3. High performance afnity chromatography of

ical Fiber Group Corporation (Hebei, China). Antithrombin antithrombin III on heparin immobilized viscose bers

III (AT III) was a gift from Hualan Bioengineering Company
Ltd. (Henan, China). Heparin sodium salt (140 U/mg) was The resultant heparin–viscose bers and the control pack-

purchased from Jiangsu Changzhou Biochemistry Institute ing media were suspended in ethanol and were slurry

(Jiangsu, China). packed into stainless steel columns (70 mm × 5.0 mm i.d.)
Epichlorohydrin was from Jinda Fine Chemical Fac- at 100 kg/cm2, respectively.

tory (Tianjin, China). Protein markers were purchased from Without additional mention, all chromatographic experi-

Shanghai Biochemistry Institute of CAS (Shanghai, China). ments were performed with step-wise elution mode at room

Water used to prepare aqueous buffers was triple distilled. temperature. A 30 mg/mL AT III solution was prepared with

The other chemical reagents used in the study were of ana- triple distilled water. The loading buffer was 0.01 mol/L

lytical grade. The buffers and solutions were ltered through sodium phosphate (pH 4.5) containing 0.1 mol/L NaCl. The

0.22 m lters prior to use. elution buffer was 0.01 mol/L sodium phosphate (pH 7.4)

The HPLC apparatus consists of a DIONEX AGP-1 gradi- containing 0.5 mol/L NaCl. The ow rate was 300 cm/h. A

ent system (Dionex Corporation, USA) with a Kratos Spec- 5 LaliquotofATIIIwasinjectedonthecolumn.Afterwash-

troow 757 variable-wavelength detector (ABI, USA). The ing with loading buffer for 5 min, the bound AT III was eluted

output from the detector was connected to a WDL-95 HPLC from column with elution buffer and detected at 220 nm.

Workstation (Dalian Institute of Chemical Physics, Dalian,
China). 2.4. Standard curve of antithrombin III

Scanning electron micrography (SEM) analysis was car-
ried out on a KYKY-2800 scanning electron microscope The standard curve was prepared using pure AT III as the

(KYKY Apparatus Factory, Chinese Academy of Sciences, standard. The different concentrations of pure AT III were

Beijing, China). Circular dichroism spectra were mea- preparedas0.024,0.048,0.096,0.12 mg/mL.Theabsorbance

sured with a Jasco J-810 spectropolarimeter (Jasco, Japan). was detected at 220 nm. The correlation coefcient (r) of the

UV absorbance spectra were determined on a Techcomp standard curve was equal to 0.997 (n = 4). From this curve, the

8500 UV–vis spectrometer (Tianmei Corporation, Shanghai, recovery of AT III from afnity column could be determined.

China).
2.5. Circular dichroism determination

2.2. Preparation of heparin–viscose ber and control
packing media Circular dichroism measurements were performed at
25 C on a Jasco series 810 spectropolarimeter. Quartz made
Briey, brous viscose (2.4 g) was suspended in a so- cylindrical cell with a path length of 1 cm was utilized to
lution consisting of 20 mL of 2.4 mol/L NaOH and 20 mL obtain spectral data in the 200–260 nm region. Two samples
of epichlorohydrin. The reaction mixture was incubated at were prepared for evaluating the effect of the loading and
50 C for 3 h. The unreacted epichlorohydrin was removed elution conditions on the structure of the protein. Sample “a”
by extensive washing with distilled water [13]. The content was prepared by loading pure AT III onto the heparin column

R. Zhao et al. / J. Chromatogr. B 816 (2005) 175–181 177


at pH 4.5 and eluting at pH 7.4. Sample “b” was prepared by
loading pure AT III and eluting at pH 7.4.

2.6. Purication of antithrombin III from human plasma
with heparin-viscose column


The starting material for the purication of AT III was
normal human plasma, which were screened to conrm the
absence of hepatitis B surface antigen and antibody to hu-
man immunodeciency virus. Human plasma was ltered
through a 0.22 m membrane lter to remove cryoprecipi-
tate. Vitamin K-dependent coagulation factors were removed
by adsorption on diethylaminoethyl Sephadex A-50, accord-
ing to the method described by Hoffmann [15]. Then 5 mL
Fig. 1. Scanning electron micrograph of the morphology of the viscose
of the treated plasma was loaded continuously onto the bers. Magnication: 2000×.
pre-equilibrated heparin-viscose column. The ow rate was
300 cm/h. Unbound specic compounds were washed thor-
oughly with 0.01 mol/L sodium phosphate (pH 4.5) contain- high mechanical strength. Another unique character of this

ing 0.1 mol/L NaCl and 0.5 mol/L NaCl, successively. The viscose ber is that it is soft in organic solvent but rigid in

boundATIIIwaselutedfromcolumnwith0.01 mol/Lsodium aqueous solution. After suspended in ethanol, it can be eas-

phosphate (pH 7.4) containing 0.5 mol/L NaCl at a ow rate ily slurry-packed into a stainless steel column at 100 kg/cm2.

of 900 cm/h. The elution was monitored at 280 nm. When replacing ethanol with an aqueous solution, the bers
become rigid, which leads to a lower backpressure, better

2.7. Characterization of puried antithrombin III performance and longer lifetime. The investigation of the re-
lationship between ow rate and column backpressure con-

The purity of the eluted fractions from heparin- rmedthatviscosebersexpressedgoodmechanicalstability

viscose column was analyzed by sodium dodecyl sulfate- in aqueous solution. As shown in Fig. 2, the column backpres-

polyacrylamide gel electrophoresis (SDS-PAGE) on 10% sure increased linearly with increasing ow rate during the

gels using a DYY-8C electrophoresis system (LiuYi Appara- test range from 60 to 1500 cm/h. When the ow rate was up to

tus Factory, Beijing, China) with a DYCZ-24D electrophore- 1500 cm/h. the backpressure of the column was about 14 MPa
siscell(gelsize:8.2 cm × 8.2 cm).Theseparatedproteinsand and there was no evidence that the bers were compressed to

protein markers were stained with Blue Coomassie R-250. cause fouling. Due to its excellent ow characteristics under

The following proteins were used as markers: rabbit phos- pressure, the viscose ber would be a promising support for

phorylase b, Mr 97400; bovine serum albumin, Mr, 66200; fast afnity purication.

rabbit actin, Mr 43000; bovine carbonic anhydrase, Mr 31000
and trypsin inhibitor, Mr 20100. 3.2. Immobilization of heparin on viscose bers

The gels were scanned using a GIS-2010 densitometer
(Tanon, Shanghai, China) for quantication of protein bands. Heparin is known for its anticoagulant activity and inter-
acts with AT III specically. Heparin-immobilized media has
been widely used as an afnity adsorbent for the separation

3. Results and discussion and purication of plasma components in blood-coagulation
systems [16,17]. Because viscose is lack of active group, the

3.1. Characterization of viscose bers attachment of heparin is made in two steps: activation and


The viscose ber is a kind of regenerated cellulose ber
and widely used in industries of weaving, knitting and threads
making. It is commercially available and inexpensive. In our
experiment, the viscose ber with a diameter of 30 m was
cut into short bers with an average length of about one hun-
dred micrometers. Morphology analysis by scanning electron
microscope (Fig. 1) showed that the viscose bers looked
like cylinders with characteristically concavo–convex sur-
face and there seems to be no pores on the surface. The
concavo–convex surface provides more specic surface area,
Fig. 2. Relationship between ow rate and backpressure of the control col-
which enables high ligand density and high binding capac- umn. Column: stainless steel column (70 mm × 5.0 mm i.d.) packed with
ity. The non-porous nature of viscose ber makes it possess control packing media; mobile phase: water.

178 R. Zhao et al. / J. Chromatogr. B 816 (2005) 175–181


Table 1 Usually, the suitable pH value of loading buffer for pro-
Optimization of the activation of viscose by epichlorohydrin using different tein purication was chosen as 7.4 in order to mimic the
NaOH concentrations
physiological buffer condition. But in this experiment, the
Concentration of Amount of epoxy groups optimized pH of loading buffer decreased to 4.5. Compared
NaOH (mol/L) on viscose (mmol/g bers)
with other researches, the coupling method of heparin and
1.0 0.28 matrix in this study is quite different from them. The reason
1.5 0.53
2.4 0.65 for the pH requirement at 4.5 for binding instead of 7.4 was
3.0 0.72 probably due to the pretreatment of heparin with hydrogen
chloride at 100 C. This probably caused partial hydrolysis of
heparin and loss of some negatively charged sulfate groups,
coupling. The viscose bers were rst activated with a bi- which will decrease the binding to the positive charges on
functional reagent, epichlorohydrin in an alkaline condition. AT III. Because pI value of AT III is about 5.0, a lower pH
Then heparin was attached to the viscose bers through a will make the protein more positively charged, which in turn
reaction between its amino groups and the epoxy groups on to strengthen the interactions between heparin and AT III.
the surface of epichlorohydrin–viscose. Epichlorohydrin also When the buffer pH is 7.4, AT III is more negatively charged
acted as the spacer between viscose and heparin. which in- so as to hardly interact with immobilized heparin.
creased the exibility of the ligand and allowed an effective SincetherewasalmostnoATIIIbindingtoheparinafnity
binding between heparin and its specic protein. column at pH 7.4, the elution buffer was chosen as 0.01 mol/L
In order to generate a reasonable amount of epoxy groups sodium phosphate buffer containing 0.5 mol/L NaCl at pH
on the surface of epichlorohydrin–viscose bers, the con- 7.4. The recovery of AT III from heparin–viscose ber col-
centration of NaOH in the activation step was optimized. umn was 90% at this condition, which was determined spec-
As shown in Table 1. the amount of epoxy groups increased trophotometrically by standard curve method.
along with increasing the concentration of NaOH. However, The concentration of NaCl in elution buffer was generally
considering that an extremely alkaline condition might have used about 1.0 mol/L or more in AT III purication reported
negative effect on the chemical stability of viscose bers, a by Peterson and Blackburn [18], as well as Funahashi [19].
NaOH concentration of 2.4 mol/L was used in the activation In their cases, the pH values of loading and elution buffers
reaction according to Guo’s work [12], The concentration of were the same as about 7, so high concentration of NaCl
epoxy groups on the obtained epichlorohydrin–viscose bers was needed to elute the AT III. In this study, the pH values of
was 0.65 mmol/g bers. loading and elution buffers were quite different as 4.5 and 7.4,
Before coupling, heparin was reacted with diluted HC1. respectively. According to the study of buffer pH, almost no
which made more amino groups of heparin exposed by par- adsorption of AT III on heparin afnity column was observed
tially hydrolyzing the N-sulfate from heparin. The partially at 0.01 mol/L phosphate buffer (pH 7.4), so the concentration
hydrolyzed heparin can be readily reacted with the epoxy of NaCl in the elution buffer was used as 0.5 mol/L. More-
groups of epichlorohydrin–viscose bers. The maximum over, due to the high protein recovery at this condition, the
binding capacity to AT III on the heparin media made of relatively low salt concentration used to elute the AT III from
this coupling method has been revealed in our previous work the column is reasonable.
[16]. After coupling reaction, the excess of epoxy groups The dynamic binding capacity of the media was investi-
remained on the heparin–viscose bers was deactivated by gated at different ow rate. As shown in Fig. 3, when ionic
ethanolamine. which would eliminate the non-specic ad- strength of loading buffer was used as 0.01 mol/L sodium
sorption of proteins. No epoxy groups were detected after phosphate (pH 4.5), the inuence of ow rate on binding
ethanolamine end-capping step. capacity could be neglectable. When the 0.01 mol/L sodium
phosphate (pH 4.5) containing 0.1 mol/L NaCl was used as
3.3. Optimization of high performance afnity loading buffer, the binding capacity signicantly decreased
chromatography conditions with increasing ow rates. Therefore, when loading buffer
with relatively high ionic strength was used to avoid non-
In order to maximize binding capacity, several parame- specic adsorption, the ow rate of loading step must be
ters, including pH, ionic strength and ow rate of mobile maintained at 300 cm/h or lower in order to gain the maxi-
phase, were optimized. The pH values showed signicant in- mum adsorption. Despite the necessity of maintaining rela-
uence on the effective adsorption of AT III on the heparin- tively low ow rate at loading step, fast purication could still
immobilized viscose bers. Maximum adsorption capacity be achieved at equilibrating and elution steps by using ow
was achieved at pH 4.5. The binding capacities decreased rate of 900 cm/h or more. The binding capacity was deter-
greatlyatotherpHvalues.WhenthepHvaluewashigherthan mined by overloading the column with AT III. The adsorbed
7, almost no adsorption occurred. So a 0.01 mol/L sodium AT III was eluted and measured spectrophotometrically as
phosphate solution with pH 4.5 was chosen as the loading 1.62 mg/g bers.
buffer and a 0.01 mol/L sodium phosphate solution with pH With the optimized conditions, anithrombin III exhibited
7.4 was chosen as the elution buffer. apparently specic binding activity to heparin–viscose bers,

R. Zhao et al. / J. Chromatogr. B 816 (2005) 175–181 179




Fig. 3. Effect of ow rate in the sample-loading step on the dynamic bind-
ing capacity of AT III on the heparin–viscose bers column, (a) loading Fig. 5. Circular dichroism spectra of AT III with (a) and without (b) experi-

buffer, 0.01 mol/L sodium phosphate (pH 4.5) containing 0.1 mol/L NaCl; encing pH4.5 binding step. Wavelength region was 200–260 nm. Operating

(b) loading buffer, 0.01 mol/L sodium phosphate (pH 4.5). The column was temperature was 25 C.

pre-equilibrated with loading buffer. Then 150 g (5 L) of AT III was in-
jected on the column. After washing with loading buffer for 5 min, the bound As shown in the Fig. 5, the CD spectra of AT III with and
AT III was eluted with 0.01 mol/L sodium phosphate (pH 7.4) containing
without experiencing pH4.5 binding step superimposed each
0.5 mol/L NaCl. The detection wavelength was 220 nm.
other well, suggesting that they have very similar secondary
structure characteristics. The structural similarities between
while no retention of AT III was observed on the control
AT III with and without experiencing pH 4.5 were also ver-
column at the same condition (as shown in Fig. 4).
ied by identical second derivatives of the UV absorbance
To address the concern that the AT III binding to the col-
spectra (data not shown).
umn at pH 4.5, circular dichroism (CD) measurement for
the sample with and without experiencing a pH 4.5 loading
3.4. Determination of the dissociation constant of the
condition were carried out. CD is a standard method to eval-
complex of immobilized heparin and antithrombin III
uate the secondary structural characteristics of protein [20].

The interaction between heparin–viscose bers and AT
III was evaluated by analytical HPAC [21,22]. The AT III
solutions containing different concentrations were passed
through the heparin-viscose column until plateaus of max-
imum absorbance occurred, respectively (shown in Fig. 6).
At this time, the eluate had the same concentration as the
initial applied AT III solution. The procedure was monitored
at 280 nm. The variation of elution volume V was plotted
according to the equation [23]:

1 Kd 1
V V0 = MT + MT [P]0

Where V and V0 are the elution volume at which the afn-
itymatrixishalf-saturatedandthevoidvolumeofthecolumn,
respectively (L), MT is the total amount of immobilized hep-
arin (mol), [P]0 is the initial concentration of AT III solution
(mol/L) and Kd is the dissociation constant of the complex
of immobilized heparin and AT III (mol/L). From the plot of

Fig. 4. Chromatographic behaviors of antithrombin III on the 1/ V V0 versus [P]0 shown in Fig. 6, Kd can be calcu-
heparin–viscose bers column (a) and the control column (b). The lated by the ratio of intercept/slope. Kd = 2.83 × 105 mol/L,
columns were pre-equilibrated with loading buffer. Then 150 g (5 L) which is similar to the result obtained in our previous work
of AT III was injected on the columns. After washing with loading buffer
[16]. The apparent dissociation constant obtained is within
for 5 min, the bound AT III was eluted with 0.01 mol/L sodium phosphate
(pH 7.4) containing 0.5 mol/L NaCl. The ow rate was 300 cm/h and the the range of 104–108 mol/L, which is suitable for afnity
wavelength was 220 nm. applications [24].

180 R. Zhao et al. / J. Chromatogr. B 816 (2005) 175–181




Fig. 8. SDS-PAGE analysis of fractions from human plasma separated on
heparin-viscose afnity column. Lanes: 1 and 2, eluted fractions from differ-
ent chromatographic runs; 3, standard AT III; 4, human plasma; 5, molecular
mass markers.



tate and vitamin K-dependent coagulation factors, the plasma
was directly loaded on the afnity column. After loading, the
column was washed with 0.01 mol/L sodium phosphate (pH
4.5) containing 0.1 mol/L NaCl and 0.5 mol/L NaCl until the
Fig. 6. Frontal chromatography of AT III on the column packed with
unbound compounds were completely eluted. The ow rate
heparin–viscose bers (A). The concentrations of AT III were as follows: (a)
0.52 × 105 mol/L; (b) 1.03 × 105 mol/L; (c) 1.72 × 105 mol/L and (d) was300 cm/hatsampleloadingandwashingsteps.TheATIII
2.59 × 105 mol/L. The ow rate was 300 cm/h and the loading buffer was fraction was eluted from column by changing mobile phase to
0.01 mol/L sodium phosphate (pH 4.5) containing 0.1 mol/L NaCl, wave- 0.01 mol/L sodium phosphate (pH 7.4) containing 0.5 mol/L
length: 280 nm. Plot of 1/ V V0 vs. [P]0 (B). The linear regression equa- NaCl at a ow-rate of 900 cm/h (Fig. 7).
tion and correlation coefcient were shown.
The AT III fraction was collected between the two vertical
lines indicated in Fig. 7 and was analyzed by SDS-PAGE.
3.5. Purication of antithrombin III from human plasma As the electrophoretogram in Fig. 8 shown, the puried AT
III exhibited one major band, which was consistent with the
AT III from human plasma was puried on the heparin- standard AT III band (Mr, 57000 measured by MALDI-TOF-
immobilized viscose column. After removal of cryoprecipi- MS). Densitometric scanning of the gel indicated that the
purity of the eluted AT III was 93%.



4. Conclusion


Viscose bers have been demonstrated to be suitable as
a novel matrix of HPAC. The excellent recovery and purity
of the isolated AT III from human plasma in this one step
HPAC suggest that this material is suitable for the isolation
of biological materials. The mechanical property of this ma-
terial makes it t for fast protein isolation. Furthermore, the
low cost of viscose ber makes it very attractive and great
potential for large-scale industrial applications.



Acknowledgements

Fig. 7. Chromatogram of human plasma on heparin-viscose column. The
column was equilibrated with 0.01 mol/L sodium phosphate–0.1 mol/L This work was funded by National Natural Science Foun-
NaCl, pH 4.5, and then 5 mL of human plasma was loaded. Elu- dation of China (90408018, 20035010, and 20435030) and
ents, 5–15 min, 0.01 mol/L sodium phosphate–0.1 mol/L NaCl, pH 4.5, Chinese Academy of Sciences (KJCX2-SW-H06). Authors
15–25 min,0.01 mol/Lsodiumphosphate–0.5 mol/LNaCl,pH4.5(owrate,
thank Dr. Yingxin Zhao of University of Tennessee for use-
300 cm/h), 25–30 min, 0.01 mol/L sodium phosphate–0.5 mol/L NaCl, pH
7.4 (ow rate, 900 cm/h); wavelength, 280 nm. The column efuents were ful discussion and Miss Xue Qu for SEM analysis, as well as
collected between the two vertical lines. Mr. Hanyuan Gong for helping the CD measurement.

R. Zhao et al. / J. Chromatogr. B 816 (2005) 175–181 181


References [12] W. Guo, Z. Shang, Y. Yu, L. Zhao, J. Chromatogr. A 685 (1994)
344.

[1] C. Tozzi, L. Anfossi, G. Giraudi, J. Chromatogr. B 797 (2003) 289. [13] L. Sundberg, J. Porath, J. Chromatogr. A 90 (1974) 87.

[2] N.E. Labrou, J. Chromatogr. B 790 (2003) 67. [14] R.T. Keen, Anal. Chem. 29 (1957) 1041.

[3] S. Ohlson, L. Hansson, P.O. Larssom, K. Mosbach, FEBS Lett. 93 [15] D.L. Hoffmann, Am. J. Med. 87 (Suppl. 3B) (1989) 23S.

(1978) 5. [16] Y.X. Zhao, R. Zhao, D.H. Shangguan, S.X. Xiong, G.Q. Liu,

[4] A. Serres, E. Legendre, J. Jozefonvicz, D. Muller, J. Chromatogr. B Biomed. Chromatogr. 15 (2001) 487.

681 (1996) 219. [17] I. Danishefsky, F. Tzeng, M. Ahrens, S. Klein, Thromb. Res. 8 (1976)

[5] C.K. Jana, E. Ali, J. Immunol. Methods 225 (1999) 95. 131.

[6] P. Gemeiner, M. Polakovic, D. Mislovicova, V. Stefuca, J. Chro- [18] C.B. Peterson, M.N. Blackburn, J. Biol. Chem. 260 (1985) 610.

matogr. B. 715 (1998) 245. [19] D. Josic, F. Bal, H. Schwinn, J. Chromatogr. 632 (1993) 1.

[7] B.H.J. Hofstee, Biochem. Biophys. Res. Commun. 63 (1975) [20] G. Zettlmeissl, H.S. Conradt, M. Nimtz, H.E. Karges, J. Biol. Chem.

618. 264 (1989) 21153.

[8] F.L. Zhou, D. Muller, J. Jozefonvicz, J. Chromatogr. 510 (1990) 71. [21] I.M. Chaiken, J. Chromatogr. 376 (1986) 11.

[9] D. Mislovicova, M. Chudinova, P. Gemeiner, P. Docolomansky, J. [22] D.S. Hage, J. Chromatogr. B 768 (2002) 3.

Chromatogr. B 664 (1995) 145. [23] Y. Shai, M. Flashner, I.M. Chaiken, Biochemistry 26 (1987) 669.

[10] W. Guo, E. Ruckenstein, J. Membr. Sci. 182 (2001) 227. [24] P. Mohr, K. Pommerening, Afnity Chromatography-Practical and

[11] E. Klein, Afnity Membranes, Wiley, New York, 1991, p. 18. Theoretical Aspects, Marcel Dekker, New York, 1985, p. 87.