Tài liệu Báo cáo khoa học: Hyperactive antifreeze protein in flounder species The sole freeze protectant in American plaice docx

Hyperactive antifreeze protein in flounder species
The sole freeze protectant in American plaice
Sherry Y. Gauthier
1
, Christopher B. Marshall
1
, Garth L. Fletcher
3
and Peter L. Davies
1,2
1 Department of Biochemistry, Queen’s University, Kingston, ON, Canada
2 Protein Function Discovery Group, Queen’s University, Kingston, ON, Canada
3 Ocean Sciences Centre, Memorial University of Newfoundland, St. John’s, NF, Canada
Antifreeze proteins (AFPs) are functionally defined by
their ability to bind to the surface of ice and inhibit its
growth, which causes a lowering of the freezing tem-
perature below the equilibrium freezing ⁄ melting point
[1]. This thermal hysteresis (TH) effect of AFPs enables
teleost fishes to live in ice-laden polar and subpolar
oceans where temperatures can reach the freezing point
of the seawater ()1.9 °C), which is over 1 C° colder
than the freezing temperature of their hypotonic body
fluids ()0.7 to )0.9 °C). Without the protection of
AFPs, fish are effectively supercooled in these waters
and will freeze on contact with ice [2,3]. Thus, the
recent acquisition of AFPs during the late stages of the
teleost radiation has enabled some species to survive in
and ⁄ or expand into the relatively new niche created by
sea-level glaciation 1–20 million years ago [4].
Type I AFPs are small, monomeric, alanine-rich
single a-helices that have an 11-amino-acid periodicity.

They are one of five distinct nonhomologous types
of AFP found in fishes [5] and they are present in
some righteye flounders, including the winter flounder
(Pseudopleuronectes americanus), yellowtail flounder
(Limanda ferruginea) and Alaskan plaice (Pleuronectes
quadritaberulatus) [6–8]. The AFP isoforms in the win-
ter flounder have been particularly well characterized.
One of them, the 37-amino-acid HPLC-6, was the first
AFP to have its structure solved [9,10] and to have the
ice plane to which it binds defined by ice-etching [8].
The differences between isoforms mainly lie in their
length (the number of 11-amino-acid repeats being
either three or four) and in the amino acid replace-
ments on the less well conserved hydrophilic side of
Keywords
alpha-helix; antifreeze protein; freezing point
depression; ice
Correspondence
P. L. Davies, Department of Biochemistry,
and Protein Function Discovery Group,
Queen’s University, Kingston, ON,
K7L 3N6, Canada
Fax: +1 613 5332497
Tel: +1 613 5332983
E-mail:
(Received 24 March 2005, revised 8 July
2005, accepted 12 July 2005)
doi:10.1111/j.1742-4658.2005.04859.x
The recent discovery of a large hyperactive antifreeze protein in the blood
plasma of winter flounder has helped explain why this fish does not freeze

in icy seawater. The previously known, smaller and much less active type I
antifreeze proteins cannot by themselves protect the flounder down to the
freezing point of seawater. The relationship between the large and small
antifreezes has yet to be established, but they do share alanine-richness
(> 60%) and extensive a-helicity. Here we have examined two other right-
eye flounder species for the presence of the hyperactive antifreeze, which
may have escaped prior detection because of its lability. Such a protein is
indeed present in the yellowtail flounder judging by its size, amino acid
composition and N-terminal sequence, along with the previously character-
ized type I antifreeze proteins. An ortholog is also present in American
plaice based on the above criteria and its high specific antifreeze activity.
This protein was purified and shown to be almost fully a-helical, highly
asymmetrical, and susceptible to denaturation at room temperature. It is
the only detectable antifreeze protein in the blood plasma of the American
plaice. Because this species appears to lack the smaller type I antifreeze
proteins, the latter may have evolved by descent from the larger antifreeze.
Abbreviations
AFP, antifreeze protein; ApAFP, American plaice antifreeze protein; IAP, ice affinity purification; TH, thermal hysteresis.
FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS 4439
these amphipathic helices. Indeed, the conservation of
the opposite alanine-rich side was helpful in identifying
this as the ice-binding face of the AFP [11].
Two puzzling issues have persisted in this field for
many years. One is that AFPs have not been found in
some righteye flounders that inhabit the same geogra-
phical range as AFP-producing species [12]. Here, an
answer could be that fish living at greater depths are
less likely to encounter ice and might be able to sur-
vive in a supercooled state by avoiding nucleation. The
other conundrum is that levels of type I AFP in the

winter flounder ( 10 mgÆ mL
)1
) appear to be too low
to guarantee protection down to the freezing point of
seawater )1.9 °C [7]. The TH activity produced by this
concentration of type I AFP is only 0.7 C°, which
when added to the colligative freezing point depression
of solutes in the blood (0.7–0.9 C°) fails to reach the
critical )1.9 °C. We have solved the latter issue with
the recent discovery in winter flounder blood plasma
of a much larger dimeric (2 · 17 kDa ¼ 34 kDa) ala-
nine-rich AFP that is extremely effective in TH and
contributes more than enough activity to protect the
fish to the freezing point of seawater [13,14]. The main
reason this isoform has escaped detection for 30 years
is that it is very thermolabile and denatures at room
temperature. This discovery has prompted us to exam-
ine the distribution of this new hyperactive AFP,
which might have been missed in other species.
The yellowtail flounder is known to produce type I
AFP [7] but the presence of the larger, hyperactive
AFP was not previously observed or anticipated. The
American plaice (Hippoglossoides platessoides) has pre-
viously been reported to exhibit antifreeze activity [12],
but the AFP type that is present there was not identi-
fied. In light of the recent discovery of the hyperactive
but labile ‘5a-like’ 17 kDa AFP in winter flounder we
have re-examined the American plaice. We note the
absence of low M
r

AFPs, but have found a high M
r
form (ApAFP) with properties very similar to those of
the newly discovered large flounder AFP. This ortho-
log is very active and sufficiently abundant to protect
the fish down to the freezing point of seawater. The
absence of the small type I AFPs in the American
plaice argues that these single helix antifreezes of the
other flounder species may have their evolutionary ori-
gins in the larger AFP.
Results
Purification of AFP from American plaice plasma
Purification of American plaice antifreeze protein
(ApAFP) from thawed blood plasma was initially done
entirely by sequential size-exclusion chromatography.
The first Sephadex G-75 column chromatography run
at 4 °C produced a fairly typical A
230
profile for fish
plasma, with most of the proteins eluting in the void
peak (fractions 30–40 in Fig. 1A), but without any
prominent smaller peaks in the region where the type I
AFPs elute in winter flounder plasma [13]. A single
peak of thermal hysteresis activity, which was associ-
ated with spindle-shaped ice crystal morphology
(Fig. 2A, upper inset), was found on the trailing edge
of the void peak (fractions 40–45). There was no trace
of activity thereafter, even in the region (fractions 70–
80) where type I AFPs would typically elute. The
unfractionated plaice plasma had thermal hysteresis

values of 700–1000 mOsm, which corresponds to 1.3–
1.8 C° (Fig. 2A). Allowing for dilution during the
chromatography, the peak seen in Fig. 1A can account
for the activity loaded onto the column.
Rechromatography of the activity peak from
Fig. 1A on the same Sephadex G-75 column at 4 °C
gave an A
230
profile with two peaks, a void peak and a
well-defined peak on the trailing edge (fractions 40–50)
in the region where the thermal hysteresis activity was
detected in the first column (Fig. 1B). Again, the ther-
mal hysteresis activity was concentrated in just a few
fractions (41–43). In a subsequent calibration chroma-
tography of this column, horse myoglobin (M
r
17 000)
eluted around fraction 60 as shown by the vertical
arrow.
A third size-exclusion chromatography was done
at room temperature on an FPLC apparatus using a
Superose-12 column. Fractions 41–43 from Fig. 1B
gave rise to one major peak around 12–12.5 mL with a
prominent shoulder at around 11.5 mL (Fig. 1C). The
thermal hysteresis activity was entirely concentrated
under the shoulder on the leading edge of the peak
(shown by the horizontal bar). Despite the short time-
span of the chromatography, much of the ApAFP
appears to have denatured to give rise to the large
peak with a lower apparent M

r
. Calibration of the
Superose-12 column with protein standards (Fig. 1C)
showed that the TH-active shoulder chromatographed
as if it were a 65–70 kDa globular protein (coeluted
with BSA) and the inactive main peak behaved as a
37 kDa protein. However, MALDI mass spectrometry
showed that only one species was present in both the
peak and the shoulder. The mass of this protein was
17 843 Da (Fig. 3).
Refinement of the purification procedure
With the development of ice affinity purification [15]
we were able to incorporate a different sequence of
Hyperactive antifreeze protein in fish S. Y. Gauthier et al.
4440 FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS
chromatography steps to avoid procedures that resul-
ted in denaturation of the AFP. The initial size-exclu-
sion chromatography on Sephadex G-75, followed by
ice affinity purification and passage through DEAE-
Sephacel, produced ApAFP that was extremely active
and almost pure as judged by SDS ⁄ PAGE (Fig. 3, lane
3). ApAFP did not bind to the anion exchanger, pre-
sumably because its pI is higher than the pH of the
bicarbonate buffer (7.0).
A
B
C
Fig. 1. Purification of American plaice AFP by size-exclusion chro-
matography. (A) Fractionation of American plaice plasma on a
Sephadex G-75 column. (B) Refractionation of pooled active AFP

fractions from (A) on Sephadex G-75. Absorbance at 230 nm (black
dot) and thermal hysteresis activity (grey triangle) are shown across
each of the elution profiles. Fraction volume (5 mL). (C) Chromato-
graphy of pooled active AFP from (B) on an FPLC Superose-12 col-
umn (absorbance at 280 nm vs. elution volume in mL). Active
ApAFP fractions are indicated by the bar. Inset: calibration of the
Superose-12 column (Log M
r
vs. elution volume) with protein
standards (I: bovine gamma globulin, 158 kDa; II: bovine serum
albumin, 67 kDa; III: horse skeletal muscle myoglobin, 17 kDa; IV:
bovine heart cytochrome c, 12.3 kDa and V: type III AFP, 7 kDa
[23]. The elution position of the main ApAFP peak is shown on
the calibration curve by j and the higher molecular mass shoulder
by h.
B
A
Fig. 2. Thermal hysteresis activity of American plaice plasma and
AFP as a function of dilution. (A) Serial twofold dilutions of plasma
were made into 100 m
M ammonium bicarbonate (pH 7.9). Data
points are the average of two readings. Inset above the activity
curve: image of the ice crystal shape obtained with hyperactive
American plaice antifreeze protein. Inset below the activity curve:
images of the crystal shapes obtained with hyperactive antifreeze
proteins of winter flounder and yellowtail flounder compared to the
bipyramidal shape produced by type I AFP. (B) Serial dilutions of
pure ApAFP were made into 100 m
M ammonium bicarbonate
(pH 7.9) containing 1 mgÆmL

)1
BSA. Data points are the average of
at least two readings. Inset is an expansion of the data points for
the low concentration readings.
S. Y. Gauthier et al. Hyperactive antifreeze protein in fish
FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS 4441
Characterization of ApAFP
ApAFP migrated as a 17 kDa band on SDS ⁄ PAGE.
This band was blotted onto a poly(vinylidene difluo-
ride) membrane, excised, and sequenced by N-terminal
Edman degradation (Table 1). The 12 residues
determined were clearly homologous to, but distinct
from, the start of the winter flounder 5a-like 17 kDa
AFP and the theoretical gene product (5a) after which
it was named [16]. For example, the seventh residue in
ApAFP is Lys. The equivalent residue in 5a is also
Lys but is Arg in the winter flounder 5a-like protein.
The fifth, sixth and tenth residues in ApAFP are Gly,
Thr and Ser, respectively, but are Ala in the alignment
of the other two sequences.
The mass of ApAFP determined by MALDI-TOF
mass spectrometry was 17 843.33 ± 0.06 Da, which is
1160 Da larger than the 16 683 Da winter flounder
5a-like AFP (Fig. 3). Its amino acid composition is
quite similar to that of the latter winter flounder AFP
and to that of the predicted 16 267 Da product of the
5a gene (Table 2). In particular, all three proteins have
Ala as the most abundant amino acid by a large mar-
gin. It makes up 60–65% of the compositions of 5a
and 5a-like protein and 55% of ApAFP. Of the other

amino acids, Thr (13%) is the next most abundant,
and there are very low amounts of aromatic and long-
chain aliphatic amino acids.
American plaice AFP is unusually active
compared to other fish AFPs
Amino acid analysis of ApAFP also provided an
opportunity to correlate its TH activity with an accu-
rately determined protein concentration. At a concen-
tration of 0.4 mgÆmL
)1
, ApAFP had a TH activity
of 2.2 C°. This compares favourably with the winter
Fig. 3. Comparison of hyperactive AFP orthologs in three righteye
flounders. Mass spectra of yellowtail flounder (1), winter flounder
(2) and American plaice (3) hyperactive AFPs shown above the
SDS ⁄ PAGE analysis of the three samples. Approximately 1 lgof
each protein was loaded on a 10% polyacrylamide SDS gel in the
Tris ⁄ Tricine buffer system. M refers to the molecular mass markers
of 26.6 kDa and 20 kDa.
Table 1. N-terminal sequences of hyperactive AFPs from righteye
flounders. Residue X at the N terminus of the 16 683 Da, winter
flounder AFP could not be conclusively identified by Edman degra-
dation but is likely to be Asn, Thr or Ala. The N-terminal sequence
of the 5a hypothetical gene product was predicted from the previ-
ously sequenced flounder gene, 5a [16], using a signal peptide pre-
diction program.
Sequence Source
XIDPAARAAAAA Winter flounder AFP
IDPAAKAAAAA 5a hypothetical gene product
SIDPGTKAASAA American plaice AFP

NIDPAVKAAAA Yellowtail flounder AFP
Table 2. Amino acid compositions of the large, hyperactive AFPs
from righteye flounders and the putative 5a gene product.
Amino
acid
5a gene
product
(mol%)
Winter
flounder
(mol%)
American
plaice (mol%)
(1st analysis)
American
plaice (mol%)
(2nd analysis)
Yellowtail
flounder
mol%
Ala 64.6 60.6 55.8 59.6 61.5
Thr 13.3 9.3 13.9 13.9 10.7
Val 4.6 4.2 1.9 1.9 3.6
Asx 4.1 6.6 3.6 3.2 4.6
Ile 3.6 4.1 3.8 4.3 4.4
Lys 3.6 3.0 4.0 3.6 2.8
Glx 1.5 1.7 2.4 1.2 2.0
Ser 1.5 4.4 3.3 2.7 2.9
Pro 1.5 1.8 3.4 3.2 1.4
Leu 0.5 1.3 2.8 2.6 2.2

Phe 0 0.7 0.7 0.6 0.4
Tyr 0 0.6 0.4 0.2 0.4
Gly 1.0 0.5 2.6 1.8 2.5
Met 0 0.5 0.5 0.6 0.1
Arg 0 0.8 0.8 0.8 0.4
Total 99.8 100.1 99.9 100.2 99.9
Hyperactive antifreeze protein in fish S. Y. Gauthier et al.
4442 FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS
flounder 5a-like AFP, which has 2.2 C° of TH activity
at a protein concentration of 0.4 mgÆmL
)1
[13]. The
much weaker type I AFP (the 37 amino acid HPLC-6
isoform) [17] shows only  0.1 C° of TH at equivalent
concentrations. The thermal hysteresis activity of
American plaice plasma as a function of concentration
shows the usual rectangular hyperbolic shape at high
concentrations but the TH activity is unusually weak
at low concentrations (Fig. 2A). This property has
been observed for the 5a-like 17 kDa flounder AFP
[14], although the explanation is not apparent. It could
be that the AFP undergoes denaturation at low con-
centrations, or it might act cooperatively and require a
threshold concentration to begin to be effective at ice
binding. To eliminate the possibility that denaturation
occurs at low total protein concentrations, thermal
hysteresis readings were done on dilutions of pure
ApAFP in 1 mgÆmL
)1
BSA. The plot of TH as a func-

tion of ApAFP concentration with albumin present
(Fig. 2B) has the same shape as the plasma dilution
series (Fig. 2A), and the inset (Fig. 2B) shows the slow
build up of activity at low AFP concentration that is
peculiar to this antifreeze.
Another diagnostic tool for classifying and compar-
ing AFPs is the ice crystal morphology they produce
during thermal hysteresis measurements. The ice crys-
tals formed by ApAFP are spindle- or lemon-shaped
(Fig. 2A, upper inset). They are thick in the centre but
show a convex tapering off towards the crystal tips
that lie on the c-axis. These crystals lack the well-
defined facets seen with the much smaller type I AFPs,
which produce the classic hexagonal bipyramid with
flat surfaces and a 3.3 : 1 ratio of c-toa-axis lengths
(Fig. 2A, lower inset).
CD analysis
The CD spectrum of a 3 lm sample collected at 4–5 °C
between 260 and 180 nm resembled that of other
alanine-rich a-helical proteins [14,18–20], with a maxi-
mum near 190 nm followed by a negative minimum near
208 nm and a second more intense minimum at longer
wavelength (Fig. 4A). The maximum ellipticity at
189.2 nm had an amplitude of 94 000 degÆcm
)2
Ædmol
)1
while the two minima occurred at 208.5 and 220.1 nm
with amplitudes of 35 100 and 44 600 degÆcm
)2

Ædmol
)1
,
respectively. The ratio of the two minima
([h]
220.1
⁄ [h]
208.5
) was 1.27, typical of alanine-rich helices
(including other type I AFPs) at low temperature.
Deconvolution by circular dichroism neural networks
(CDNN) of the ApAFP CD spectra at 4–5 °C shows
that the protein is almost fully a-helical (94.7% a-helix,
7% b-turn, 1.4% random coil, with no substantial con-
tribution from b-sheet, 0.5%: total ¼ 103.6%). As the
sample was warmed from 4 to 8 °C, there was a small
decrease in the intensity of the spectrum; however, from
8to16°C there was a very large decrease in the ampli-
tude of the signal (Fig. 4A). The magnitude of the maxi-
mum at 189 nm decreased by 47% at 12 °C and 63% at
16 °C while the 221.5 nm minimum decreased by 42%
B
A
Fig. 4. Circular dichroism spectra of American plaice AFP at low
temperature and the effect of heating. (A) The averaged spectrum
at 4–5 °Cofa3l
M sample of ApAFP is shown in black. The sam-
ple was then very slowly heated and averaged spectra are presen-
ted for the following temperature ranges: 4–8 °C, red; 8–12 °C,
green; 12–16 °C, yellow; 16–20 °C, blue; 20–24 °C, pink; 24–30 °C,

cyan; 30–40 °C, grey; 40–50 °C, dark red; 50–60 °C, dark yellow;
and 60–75 °C, dark blue. (B) In a separate experiment, the mean
residue ellipticity (222 nm) was plotted against temperature as a
sample was warmed (d) then subsequently cooled (n)asdes-
cribed in Experimental procedures. The mean residue ellipticity
following 40 h at 4 °C is indicated by a triangle. Also shown on
this plot are the mean residue ellipticity (222 nm) values of winter
flounder AFP as a function of temperature as a sample was
warmed (ÆÆÆjÆÆÆ) then subsequently cooled (h) as described in
Experimental procedures. The mean residue ellipticity following
overnight incubation at 4 °C is indicated by a square.
S. Y. Gauthier et al. Hyperactive antifreeze protein in fish
FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS 4443
and 56% at these two temperatures, respectively. The
([h]
222
⁄ [h]
208
) ratio decreased as the ApAFP sample was
warmed. At 16 °C, the protein had lost roughly half of
its helicity (48.6% a-helix, 16.1% b-turn, 12.8% random
coil, 9% b-sheet: total ¼ 86.4%). Changes in the magni-
tude of the spectrum were small between 16 and 40 °C.
[At room temperature (20–24 °C) helicity was 40.5%
where the total for all structures was 88%.] However,
the measured ellipticities underwent a second, rapid
decrease in magnitude as the temperature increased
above 40 °C. Between 4 and 40 °C, the spectra intersect
near an isodichroic point at  201 nm with a magnitude
of  )16 000 degÆcm

)2
Ædmol
)1
. Above 40 °C, the ellip-
ticity at this wavelength decreased, possibly due to pro-
tein precipitation.
To better determine the temperature at which the
protein melts, to investigate the nature of the trans-
ition and to determine the extent to which denatura-
tion is reversible, melting and renaturation curves of
ApAFP were derived using CD spectroscopy (Fig. 4B).
The ellipticity at 222 nm ([h]
222
) was measured as the
sample was warmed from 4 to 30 °C. The magnitude
decreased slightly when the sample was warmed above
4 °C, although the first substantial changes occurred at
7 °C as the magnitude of the [h]
222
decreased by  5%.
As the sample was warmed to 12 °C, there was a very
large transition during which the [h]
222
decreased by
50%. The apparent T
m
of this transition was 9 °C.
There was no appreciable further change in the signal
as the sample was warmed from 13 to 30 °C.
To investigate whether the denaturation of this

extremely thermolabile protein is reversible, CD was
measured while the sample was cooled back down to
4 °C over 2 h. At each temperature,  80% of the
ellipticity that was lost upon warming was recovered,
thus producing a renaturation curve that was remark-
ably parallel to the melting curve. Because the [h]
222
was not completely restored at 4 °C on this short time-
scale, the sample was allowed to renature at 4 °C for
40 h and the [h]
222
was measured again. At this point
the [h]
222
was 96% of that initially produced by the
sample at 4 °C.
We have contrasted this denaturation ⁄ renaturation
profile with that obtained using winter flounder 5a-like
protein. The winter flounder AFP melts sharply at a
higher temperature ( 20 °C) but does not renature as
rapidly or completely as the ApAFP (dotted lines in
Fig. 4B). Consistent with these differences in renatura-
tion, ApAFP recovers much more activity (60–100%
after 70 min at room temperature) than the winter
flounder AFP does, but its TH values show consider-
able deviations from reading to reading and sample to
sample.
Isolation of the hyperactive AFP from yellowtail
flounder plasma
A plasma sample collected from yellowtail flounder in

March, when the seasonal production of AFP is still
high, had 500 mOsm (0.93 C°) of TH activity. This
TH value is lower than those of the winter flounder
and American plaice plasmas, but higher than would
be produced by type I AFP alone, particularly as yel-
lowtail flounder type I AFPs are less active than those
produced by the winter flounder [21]. The morphology
of the ice crystals formed in the plasma was spindle-
like and resembled that produced by the American
plaice and winter flounder 17 kDa AFPs, albeit slightly
more elongated (Fig. 2A). This elongation probably
reflects the presence of type I AFP in the plasma sam-
ple. The major protein present after purification migra-
ted on SDS ⁄ PAGE marginally faster than the winter
flounder AFP with an apparent M
r
of  16 kDa
(Fig. 3, lane 1). The monomeric mass of this protein
was 16 283.8 Da as determined by MALDI-TOF MS
(Fig. 3). Like the 17 kDa winter flounder and Ameri-
can plaice AFPs, the amino acid composition of this
protein was rich in alanine (61.5%) and threonine
(10.7%), and had few hydrophobic residues (Table 2).
The N-terminal sequence (NIDPAVKAAAA) was also
similar to those of the other two AFPs, but with
a valine substitution preceding the basic residue
(Table 1).
Discussion
Biochemical characterization of the American plaice
AFP clearly shows that it is an ortholog of the newly

discovered hyperactive (5a-like) 17 kDa AFP in winter
flounder [13]. It has a similar monomeric mass
(17 843 Da compared to 16 683 Da for the flounder
protein) and, like the flounder protein, it has a similar,
much larger apparent molecular mass (67 kDa) by
size-exclusion chromatography, suggesting an extended
conformation. Also, its amino acid composition, N-ter-
minal sequence and the CD spectrum it produces are
all characteristic of the new 5a-like flounder AFP.
Another diagnostic feature is its thermolability, which
was clearly demonstrated by the CD spectra and
FPLC chromatography on Superose-12. Indeed,
ApAFP denatures at 9 °C, making it even more labile
than the homolog in winter flounder, which melts at
20 °C. Our interpretation of the FPLC profile is that
the small residual amount of undenatured ApAFP
gives rise to the high molecular mass shoulder with
thermal hysteresis activity. Based on observations of
the winter flounder protein this species is likely to be a
Hyperactive antifreeze protein in fish S. Y. Gauthier et al.
4444 FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS
rod-like dimer [14]. The main peak is probably dena-
tured monomer, which explains its lower apparent
molecular mass, its loss of activity, and the fact that
both the main peak and the shoulder only contain one
protein type, the 17 843 Da species.
The room temperature lability of these hyperactive
proteins can perhaps explain why they have been
missed before now. Because the winter and yellowtail
flounders also have the more resilient smaller type I

AFPs, it is possible that any loss of activity of the
larger AFP went unnoticed or was attributed to poor
recovery of activity during purification. The thermal
lability of the larger AFP contrasts with the ‘elasticity’
of the smaller AFP. Although these short, monomeric
helical AFP are 50% denatured at room temperature
they completely regain their helicity and activity on
cooling [22]. Obviously, antifreeze activity can only be
measured near 0 °C because ice has to be present for
the assay. Purified ApAFP is also capable of regaining
helicity upon cooling but this does not correlate with
complete recovery of activity. This is probably due to
inefficient reassembly of dimeric quaternary structure,
a feature that appears to be required for TH activity
in the winter flounder homolog. Nevertheless, the ther-
mal denaturation of ApAFP, even though it occurs at
a lower T
m
, is a more reversible process than that of
the 17 kDa flounder AFP, as measured by both CD
and TH activity.
Until now, the American plaice was thought to have
no, or low levels of, antifreeze activity in their blood
plasma [12]. Indeed, because we are dealing with a wild
species residing in different geographical locations,
inhabiting a range of depths, it is entirely possible that
some populations of American plaice produce little, if
any, AFP. Extreme variation in AFP production has,
for example, been seen before in ocean pout from two
different locations [23]. However, because of the labil-

ity of ApAFP, it is also likely that some loss of activity
occurred on previous occasions during collection or
assay of the plasma.
Although the new AFP is 10–100-times more active
in TH than the small type I AFPs, they are quite likely
related as paralogs. The original 5a ‘pseudogene’
sequence, which is clearly homologous (based on its
predicted N-terminal sequence) to the large hyperactive
AFPs of winter flounder, American plaice and yellow-
tail flounder was isolated from a winter flounder
genomic library by hybridization to a type I AFP
cDNA probe [16,24]. Moreover, the 5a and type I
AFP genes share substantial sequence identity in the 5¢
and 3¢ untranslated regions of their genes. Thus, the
argument for the hyperactive 17 kDa AFPs being
homologous to the type I AFPs is strong even though
it is indirect and relies on the 5a gene as a common
link. That there are no small plasma AFPs in the
American plaice argues that they were probably never
developed in this fish rather than that they were selec-
tively lost. The gene family coding for type I AFPs in
winter flounder plasma has 30–40 members, many of
which are organized in tandem head-to-tail arrays,
which are a hallmark of gene amplification resulting
from strong selective pressure [25]. It seems unlikely
that such a robust gene family, which encodes func-
tionally important proteins, would be completely lost
or silenced when it is very highly expressed in closely
related species. If this is the case, then the larger AFP
could be the progenitor of the small type I AFPs.

The presence of larger, highly active AFPs in the
plasma of both winter and yellowtail flounder that are
fully capable of protecting the fish from freezing below
the freezing point of seawater calls into question the
functional significance of the small type I AFP in these
species. Considering the levels of these AFPs in
the plasma it seems unlikely that they are excess to the
needs of the fish. In addition, the abundance of
the skin-type AFP in epithelial tissues also attests to
the argument that these small AFPs are important to
the survival of winter flounder and other species in
subzero ice-laden waters [26,27].
Although the information that we have on the num-
ber of species that possess the larger AFP is limited, it
is evident that the two species (winter flounder, yellow-
tail flounder) that have both the large and the small
AFP normally reside in water that is considerably shal-
lower than that inhabited by the American plaice [12].
Thus they are much more likely to be exposed to ice.
The presence of the small skin-type AFP in the epithe-
lial tissues of winter flounder suggests that they are
required to protect the cells and tissues that would
come into direct contact with ice [27,28]. The presence
of skin-type AFP in a variety of species inhabiting the
relatively shallow waters along the coast of Newfound-
land during the winter makes a case for their physiolo-
gical significance [29–31].
The larger 5a-like AFP produced by winter flounder
is a dimer of > 30 kDa and behaves as a much larger
protein during size-exclusion chromatography. There-

fore it may not be capable of penetrating or diffusing
readily into poorly vascularized tissues such as the skin
epithelia. One other point worthy of mention in this
regard is the freeze protection of urine. AFPs have
been found in the urine of all AFP producing species
that possess functional glomeruli (winter flounder, sea
raven (Hemitripterus americanus), Atlantic cod (Gadus
morhua) and ocean pout (Macrozoarces americanus)
[32]. In the case of the winter flounder the AFP
S. Y. Gauthier et al. Hyperactive antifreeze protein in fish
FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS 4445
isolated from urine were of the small helical variety.
The problems associated with freeze protection of
extracellular fluids such as urine, and external epithelia
could have been the impetus for the evolution of the
smaller AFP.
The significance of this report is three-fold: (a) large
hyperactive AFPs are found in at least three genera of
righteye flounder; (b) because only two of these three
species have the smaller type I AFPs, it raises the pos-
sibility that the type I AFPs have been derived from
the large AFPs; and (c) if there are other flounder spe-
cies which have been scored as AFP-negative because
they do not have the type I AFPs they should be
re-examined for the presence of the hyperactive,
thermolabile large AFP.
Experimental procedures
Winter flounder (Pseudopleuronectes americanus) and yellow-
tail flounder (Limanda ferrugenia) were collected from New-
foundland coastal waters by commercial fishers and SCUBA

divers and transported live to aquaria where they were main-
tained in sea water until blood was sampled. American plaice
(Hippoglossoides platessoides) were collected from the St.
Lawrence estuary by otter trawl, transported live to St. Jolie,
PQ where they were blood sampled. Blood samples were
collected from a caudal blood vessel using 21 or 23-gauge
syringe needles, transferred to heparinized test tubes and cen-
trifuged to remove the blood cells. The resulting plasma was
stored frozen at )20 °C prior to analysis.
All measures were taken to minimize pain and discomfort
during animal experiments. Guidelines followed were those
of the Canadian Council on Animal Care (CCAC).
Purification procedures
The purification of a 17 kDa AFP from winter flounder
has been described previously [13,14] Briefly, plasma was
fractionated by size-exclusion chromatography, fractions
containing the 17 kDa AFP were pooled and subjected to
three successive rounds of ice affinity purification (IAP)
[15]. The few remaining contaminants in the ice fraction
were removed by batch adsorption to DEAE-Sephacel in
50 mm ammonium bicarbonate buffer (adjusted to pH 7.0
with HCl immediately before the chromatography). The
17 kDa AFP did not bind to the anion exchanger and was
recovered by filtering the resin into a column. All steps
were performed at or below 4 °C. The final product was
examined for purity by SDS ⁄ PAGE and mass spectro-
metry.
Yellowtail flounder antifreeze proteins were partially
purified by two rounds of ice affinity purification followed
by the DEAE-Sephacel step. Frozen plasma was thawed

and cleared by centrifugation. An aliquot (1.5 mL) from an
individual fish was diluted with 30 mL of cold 20 mm
ammonium bicarbonate (pH 7.9). Growth of an ice hemi-
sphere in the plasma solution was initiated from a cold-
finger at a temperature of )0.5 °C and the ice was grown
to two-thirds of the sample volume (20 mL) by gradually
lowering the temperature to )2.0 °C over 18 h. The ice
fraction was then removed and melted, adjusted to 20 mm
ammonium bicarbonate and a second round of IAP was
performed prior to the removal of impurities using DEAE-
Sephacel.
Blood plasma from American plaice (pooled from several
fish) was initially purified by three consecutive size-exclusion
chromatography steps. An aliquot (3 mL) of plasma was dilu-
ted with an equal volume of running buffer and loaded onto
a 0.5 L (2.6 · 100 cm) Sephadex G-75 size-exclusion column
at 4 °C and fractionated at a flow rate of 0.7 mLÆmin
)1
with
100 mm ammonium bicarbonate (pH 7.9) as the running
buffer. The fractions (5 mL) exhibiting TH activity were
pooled, lyophilized, resuspended in running buffer (6 mL)
and reapplied to the same column. The proteins were eluted
as above. Active fractions were pooled and lyophilized. This
pooled material was resuspended in 100 mm ammonium
bicarbonate (pH 7.9) running buffer (0.2 mL) and loaded
on a FPLC Superose-12 column (Pfizer, New York, NY,
USA; 1.0 · 30 cm) at room temperature and eluted at
0.5 mLÆmin
)1

. The Superose-12 column was calibrated with
standard proteins of known molecular mass (Fig. 1 legend).
To avoid the chromatography step carried out at room
temperature, an alternative purification procedure for ApA-
FP was developed. The plasma was fractionated by one
cycle of Sephadex G-75 size-exclusion chromatography as
described above. AFP was further enriched from the
TH-active fractions [diluted to 42.5 mL with 100 mm
ammonium bicarbonate (pH 7.9)] by one round of IAP
using a cooling regime of )0.6 °Cto)3.6 °C in 24 h. Con-
taminants were removed using DEAE-Sephacel as described
above. Dilute samples of AFP were concentrated by retent-
ion in an Amicon (Danvers, MA, USA) Ultra-15 centrifu-
gal filter device (5 kDa molecular mass cut-off) at 4 °C.
Amino acid analysis and N-terminal sequencing
by Edman degradation
The protein corresponding to the  17 kDa AFP in all
three righteye flounders was partially or completely purified
as described above and then electrophoresed on
SDS ⁄ PAGE. After transferring the protein band(s) onto
poly(vinylidene difluoride) membrane using 10 mm CAPS,
10% (v ⁄ v) methanol as transfer buffer (pH 11.0), the region
corresponding to the 17 kDa band was visualized using
Coomassie Blue R-250, cut out of the membrane and
sent for N-terminal sequencing by Edman degradation
(Advanced Protein Technology Centre, Hospital for Sick
Children, Toronto, ON).
Hyperactive antifreeze protein in fish S. Y. Gauthier et al.
4446 FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS
Amino acid analyses were performed by the Advanced

Protein Technology Centre, Toronto on the AFPs from
winter flounder and American plaice after they had been
purified to homogeneity following the DEAE-Sephacel step.
These analyses were used to determine both the composi-
tion and concentration of these proteins. Amino acid ana-
lysis of the yellowtail flounder AFP was done directly on
the  17 kDa band blotted from the SDS ⁄ PAGE gel.
Mass spectrometry
The mass of the hyperactive winter flounder AFP was
determined previously [13]. The masses of the American
plaice and yellowtail flounder AFPs were determined by
MALDI-TOF mass spectrometry using protein samples
taken following the DEAE-Sephacel purification step. The
ApAFP sample was analyzed with a Micromass (Waters
Ltd., Mississauga, Canada) Q-TOF Ultima instrument using
a matrix of 5 mgÆmL
)1
a-cyano-4-hydroxycinnamic acid,
0.1% trifluoroacetic acid in 70% acetonitrile. The mass of
the yellowtail flounder protein was determined with a Voy-
ager DE Pro spectrometer (Applied Biosystems, Foster City,
CA, USA) in linear mode with a matrix of 10 mgÆmL
)1
sin-
apinic acid.
Circular dichroism spectroscopy
CD spectra of ApAFP were collected using an Olis Rapid
Scanning Monochromator with a digital subtractive method
CD module (Olis, Bogart, Georgia) and a 1 mm quartz
cuvette (Hellma 121-QS). CD spectra were collected at

0.69 nm intervals between 260 and 180 nm at a scanning
rate (nmÆmin
)1
) determined by the olis software as a func-
tion of the signal intensity. The temperature of the cuvette
holder was monitored and controlled by a circulating
waterbath.
An aliquot of the purified ApAFP was dialyzed exten-
sively against 10 m m sodium phosphate (pH 7.0) at 4 ° C,
and the baseline circular dichroism of this dialysis buffer
was subtracted from the subsequent protein spectra. The
concentration of AFP in this sample was estimated by
amino acid analysis to be  3 lm . Seven spectra were col-
lected from this protein sample near 4 °C and averaged.
The temperature was then slowly increased to 75 °C (over
14.5 h) and spectra were collected continuously. The spec-
tra collected within temperature ranges of 4–10 C° were
averaged (4–8, 8–12, 12–16, 16–20, 20–24, 24–30, 30–40,
40–50, 50–60, and 60–75 °C). The measured ellipticity (mil-
lidegrees) at each wavelength was converted to mean
residue ellipticity (degÆcm
)2
Ædmol
)1
) using the total concen-
tration of all residues determined by amino acid analysis
(0.70 mm). This value corresponds closely to that deter-
mined using the estimated protein concentration and num-
ber of amino acids in the protein, which were both
predicted from the amino acid composition and molecular

mass. The components of the secondary structure of the
protein were deconvoluted from the CD spectrum (180–
260 nm) using the neural network-based software cdnn
version 2.1 (Gerald Bo
¨
hm, Institut fu
¨
r Biotechnologie,
Martin-Luther-Universita
¨
t Halle-Wittenberg, Germany) [33],
trained on the ‘simple’ set of proteins.
In a separate experiment melting and renaturation curves
were examined. A sample of protein ( 4.2 lm) was dia-
lyzed against 10 mm sodium phosphate (pH 7.0) and, as
above, the dialysis buffer was used to establish the baseline.
Circular dichroism of the sample was measured using an
AVIV 62ADS CD spectrometer (Lakewood, NJ) and a
quartz cuvette of 1 mm pathlength (Hellma 110-QS). The
temperature of the cuvette was controlled by a Peltier
device. The temperature of the sample was increased from
4to30°Cat1C° intervals every 2 min. Following equili-
bration at each temperature, the circular dichroism at
222 nm of the sample was measured. As described above,
the measured ellipticity was converted to mean residue
ellipticity using a concentration determined by amino acid
analysis.
To investigate the reversibility of the helix to coil trans-
ition that occurs between 7 and 12 °C, the sample was
cooled from 30 to 4 °C. Because there was little change in

the CD signal upon warming from 15 to 30 °C, the sample
was rapidly cooled to 15 °C before beginning collection of
CD data. The temperature was then lowered at 1 C° inter-
vals and at each temperature the ellipticity at 222 nm was
monitored until it appeared to stabilize. This required 10–
15 min for temperatures below 12 °C. When the CD signal
appeared to stabilize, it was measured for 2 min and aver-
aged before lowering the temperature an additional degree.
To determine whether the protein would continue to rena-
ture at a slow rate not detectable on a time scale of min-
utes, the sample was stored at 4 °C for 40 h and the CD
was measured again.
Antifreeze assays
Thermal hysteresis was measured as previously described
[34]. Ice crystal images were digitally collected by a Nikon
CoolPix 4500 camera mounted on a Leitz dialux 22 micro-
scope with a Leitz Wetzlar 160 ⁄ –EF10⁄ 0.25 objective
(Oberkochen, Germany).
Acknowledgements
This work was funded by grants from the Canadian
Institutes of Health Research (CIHR) and the Natural
Sciences and Engineering Research Council (NSERC)
to P.L.D. and G.L.F., respectively. P.L.D. holds a
Canada Research Chair in Protein Engineering.
C.B.M. was supported by an Ontario Graduate Schol-
arship. The authors would like to thank Madonna
S. Y. Gauthier et al. Hyperactive antifreeze protein in fish
FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS 4447
King for the shipment of fish plasma, Kim Munro and
the Protein Function Discovery Facility at Queen’s

University for circular dichroism spectroscopy, and the
Alberta Peptide Institute, Edmonton, AB and the
Advanced Protein Technology Centre at the Hospital
for Sick Children, Toronto, ON for amino acid
analysis and N-terminal sequencing. We also thank
Avi Chakrabartty and the Ontario Cancer Institute for
access to their CD machine.
References
1 Raymond JA & DeVries AL (1977) Adsorption inhibi-
tion as a mechanism of freezing resistance in polar
fishes. Proc Natl Acad Sci USA 74, 2589–2593.
2 Fletcher GL, Kao MH & Fourney RM (1986) Anti-
freeze peptides confer freezing resistance to fish. Can J
Zool 64, 1897–1901.
3 Scholander PF, Vandam L, Kanwisher JW, Hammel
HT & Gordon MS (1957) Supercooling and osmoregu-
lation in Arctic fish. J Cell Comp Physiol 49, 5–24.
4 Scott GK, Fletcher GL & Davies PL (1986) Fish anti-
freeze proteins: recent gene evolution. Can J Fish Aquat
Sci 43, 1028–1034.
5 Fletcher GL, Hew CL & Davies PL (2001) Antifreeze
proteins of teleost fishes. Annu Rev Physiol 63, 359–390.
6 Duman JG & DeVries AL (1974) Freezing resistance in
winter flounder. Nature 274, 237–238.
7 Scott GK, Davies PL, Kao MH & Fletcher GL (1988)
Differential amplification of antifreeze protein genes in
the pleuronectinae. J Mol Evol 27, 29–35.
8 Knight CA, Cheng CC & DeVries AL (1991) Adsorp-
tion of alpha-helical antifreeze peptides on specific ice
crystal surface planes. Biophys J 59, 409–418.

9 Yang DS, Sax M, Chakrabartty A & Hew CL (1988)
Crystal structure of an antifreeze polypeptide and its
mechanistic implications. Nature 333, 232–237.
10 Sicheri F & Yang DS (1995) Ice-binding structure and
mechanism of an antifreeze protein from winter floun-
der. Nature 375, 427–431.
11 Baardsnes J, Kondejewski LH, Hodges RS, Chao H,
Kay C & Davies PL (1999) New ice-binding face for
type I antifreeze protein. FEBS Lett 463, 87–91.
12 Goddard SV & Fletcher GL (2002) Physiological Ecol-
ogy of Antifreeze Proteins – a Northern Perspective. In
Molecular Aspects of Fish and Marine Biology, Vol. 1:
Fish Antifreeze Proteins. (Ewart KV & Hew CL, eds),
pp. 17–60. World Scientific, Singapore.
13 Marshall CB, Fletcher GL & Davies PL (2004) Hyper-
active antifreeze protein in a fish. Nature 429, 153.
14 Marshall CB, Chakrabartty A & Davies PL (2005)
Hyperactive antifreeze protein from winter flounder is a
very long, rod-like dimer of alpha-helices. J Biol Chem
280, 17920–17929.
15 Kuiper MJ, Lankin C, Gauthier SY, Walker VK &
Davies PL (2003) Purification of antifreeze proteins by
adsorption to ice. Biochem Biophys Res Commun 300,
645–648.
16 Davies PL & Gauthier SY (1992) Antifreeze protein
pseudogenes. Gene 112, 171–178.
17 Hew CL, Wang NC, Yan S, Cai H, Sclater A &
Fletcher GL (1986) Biosynthesis of antifreeze polypep-
tides in the winter flounder. Characterization and seaso-
nal occurrence of precursor polypeptides. Eur J Biochem

160, 267–272.
18 Wallimann P, Kennedy RJ, Miller JS, Shalongo W &
Kemp DS (2003) Dual wavelength parametric test of
two-state models for circular dichroism spectra of helical
polypeptides: anomalous dichroic properties of alanine-
rich peptides. J Am Chem Soc 125, 1203–1220.
19 Gronwald W, Chao H, Reddy DV, Davies PL, Sykes
BD & Sonnichsen FD (1996) NMR characterization
of side chain flexibility and backbone structure in the
type antifreeze protein at near freezing temperatures.
Biochemistry 35, 16698–16704.
20 Houston ME Jr, Chao H, Hodges RS, Sykes BD, Kay
CM, Sonnichsen FD, Loewen MC & Davies PL (1998)
Binding of an oligopeptide to a specific plane of ice.
J Biol Chem 273, 11714–11718.
21 Scott GK, Davies PL, Shears MA & Fletcher GL
(1987) Structural variations in the alanine-rich antifreeze
proteins of the pleuronectinae. Eur J Biochem 168, 629–
633.
22 Chao H, Houston ME Jr, Hodges RS, Kay CM, Sykes
BD, Loewen MC, Davies PL & Sonnichsen FD (1997)
A diminished role for hydrogen bonds in antifreeze pro-
tein binding to ice. Biochemistry 36, 14652–14660.
23 Hew CL, Wang NC, Joshi S, Fletcher GL, Scott GK,
Hayes PH, Buettner B & Davies PL (1988) Multiple
genes provide the basis for antifreeze protein diversity
and dosage in the ocean pout, Macrozoarces americanus.
J Biol Chem 263, 12049–12055.
24 Davies PL, Hough C, Scott GK, Ng N, White BN &
Hew CL (1984) Antifreeze protein genes of the winter

flounder. J Biol Chem 259, 9241–9247.
25 Scott GK, Hew CL & Davies PL (1985) Antifreeze pro-
tein genes are tandemly linked and clustered in the gen-
ome of the winter flounder. Proc Natl Acad Sci USA
82, 2613–2617.
26 Gong Z, Ewart KV, Hu Z, Fletcher GL & Hew CL
(1996) Skin antifreeze protein genes of the winter floun-
der, Pleuronectes americanus, encode distinct and active
polypeptides without the secretory signal and prose-
quences. J Biol Chem 271, 4106–4112.
27 Murray HM, Hew CL, Kao KR & Fletcher GL (2002)
Localization of cells from the winter flounder gill
expressing a skin-type antifreeze protein gene. Can J
Zool 80, 110–119.
Hyperactive antifreeze protein in fish S. Y. Gauthier et al.
4448 FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS
28 Murray HM, Hew CL & Fletcher GL (2003) Spatial
expression patterns of skin-type antifreeze protein in
winter flounder (Pseudopleuronectes americanus)
epidermis following metamorphosis. J Morphol 257,
78–86.
29 Evans RP & Fletcher GL (2004) Isolation and purifica-
tion of antifreeze proteins from skin tissues of snailfish,
cunner and sea raven. Biochim Biophys Acta 1700,
209–217.
30 Low WK, Miao M, Ewart KV, Yang DS, Fletcher GL
& Hew CL (1998) Skin-type antifreeze protein from the
shorthorn sculpin, Myoxocephalus scorpius. Expression
and characterization of a M
r

9,700 recombinant protein.
J Biol Chem 273, 23098–23103.
31 Low WK, Lin Q, Stathakis C, Miao M, Fletcher GL &
Hew CL (2001) Isolation and characterization of skin-
type, type I antifreeze polypeptides from the longhorn
sculpin, Myoxocephalus octodecemspinosus. J Biol Chem
276, 11582–11589.
32 Fletcher GL, King MJ, Kao MH & Shears MA (1989)
Antifreeze proteins in the urine of marine fish. Fish
Physiol Biochem 6, 121–127.
33 Bohm G, Muhr R & Jaenicke R (1992) Quantitative
analysis of protein far UV circular dichroism spectra by
neural networks. Protein Eng 5, 191–195.
34 Chakrabartty A & Hew CL (1991) The effect of enhanced
alpha-helicity on the activity of a winter flounder
antifreeze polypeptide. Eur J Biochem 202, 1057–1063.
S. Y. Gauthier et al. Hyperactive antifreeze protein in fish
FEBS Journal 272 (2005) 4439–4449 ª 2005 FEBS 4449