Download Structural adaptation of enzymes to low

Protein Engineering vol.14 no.3 pp.141–148, 2001
Structural adaptation of enzymes to low temperatures
Giulio Gianese1, Patrick Argos1 and
Stefano Pascarella1,2,3
1Dipartimento
di Scienze Biochimiche ‘A. Rossi Fanelli’ and Centro di
Biologia Molecolare del CNR and 2Centro Interdipartimentale di Ricerca
per l’Analisi dei Modelli e dell’Informazione nei Sistemi Biomedici (CISB),
Università ‘La Sapienza’, P.le A. Moro 5, 00185 Rome, Italy
3To
whom correspondence should be addressed, at the Dipartimento di
Scienze Biochimiche. E-mail: [email protected]
A systematic comparative analysis of 21 psychrophilic
enzymes belonging to different structural families from
prokaryotic and eukaryotic organisms is reported. The
sequences of these enzymes were multiply aligned to 427
homologous proteins from mesophiles and thermophiles.
The net flux of amino acid exchanges from meso/thermophilic to psychrophilic enzymes was measured. To assign
the observed preferred exchanges to different structural
environments, such as secondary structure, solvent accessibility and subunit interfaces, homology modeling was
utilized to predict the secondary structure and accessibility
of amino acid residues for the psychrophilic enzymes
for which no experimental three-dimensional structure is
available. Our results show a clear tendency for the charged
residues Arg and Glu to be replaced at exposed sites on
α-helices by Lys and Ala, respectively, in the direction from
‘hot’ to ‘cold’ enzymes. Val is replaced by Ala at buried
regions in α-helices. Compositional analysis of psychrophilic enzymes shows a significant increase in Ala and Asn
and a decrease in Arg at exposed sites. Buried sites in
β-strands tend to be depleted of Val. Possible implications
of the observed structural variations for protein stability
and engineering are discussed.
Keywords: cold adaptation/protein engineering/protein
stability/psychrophilic enzymes/residue substitution
Introduction
It is estimated that around 90% of the biosphere exists
at temperatures below 10°C. Indeed, the earth’s surface is
dominated by low-temperature environments such as the Arctic
and Antarctic continents, mountains regions and the marine
waters which cover 70% of its area and display, below 1000
m under the sea level, temperatures not exceeding 5°C.
Psychrophilic organisms live at such low temperatures, where
most other species cannot grow and to survive they need to
produce enzymes able to perform efficiently their catalysis
under these extreme environmental conditions. At the same
temperatures, enzymes from mesophilic or thermophilic organisms are generally unable to sustain a viable metabolism (Feller
and Gerday, 1997; Gerday et al., 2000). For these reasons,
enzymes synthesized by psychrophilic organisms have considerable biotechnological potential: their ability to work efficiently as catalysts at low temperatures offers advantageous
environmental applications and energy savings when used
in industrial processes (Marshall, 1997). Whereas important
© Oxford University Press
progress has been made in elucidating the molecular adaptation
mechanism of enzymes produced by extremophiles such as
hyperthermophiles (e.g. Jaenicke and Böhm, 1998), the
molecular basis of the cold adaptation is still relatively poorly
understood. Recent accumulation of structural data on psychrophilic enzymes is beginning to shed light on their functional
and structural characteristics (e.g. Feller and Gerday, 1997).
The commonly observed features of these cold active enzymes
are their increased catalytic efficiency at low temperatures
measured as kcat/KM and a significantly increased thermolability
which is believed to be a consequence of enhanced peptide
chain flexibility. The amount of available structural data,
primary and tertiary, on psychrophilic enzymes is now sufficient
to undertake a comprehensive and significant comparative
analysis. Studies on the structural adaptation of thermophilic
enzymes (e.g. Menéndez-Arias and Argos, 1989; Vogt et al.,
1997; Szilágyi and Závodszky, 2000) utilized comparative
analysis. This approach gave valuable indications on possible
adaptive strategies utilized by evolution to stabilize enzymes
at high temperatures, thus suggesting rules to be followed by
protein engineers to produce modified enzymes with characteristics suitable for biotechnological applications (Fontana,
1991). For example, the neutral protease from Bacillus subtilis
was successfully stabilized by two mutations Gly → Ala in
loop and helical regions (Margarit et al., 1992). These
exchanges were indicated by Menéndez-Arias and Argos
(1989) to be involved in the enzyme adaptation to hot environments. This paper describes an analysis of preferred amino
acid residue substitutions in a dataset of 21 psychrophilic
enzymes extracted from various prokaryotic and eukaryotic
species and multiply aligned to homologous mesophilic and
thermophilic sequences. To assign the observed preferred
exchanges to different structural environments (secondary
structure, solvent accessibility and subunit interfaces), homology modeling was utilized to predict the secondary structure
and accessibility of amino acid residues in those psychrophilic
enzymes for which no experimental three-dimensional structure
was available. The results suggest possible general rules for
protein engineering experiments aimed at producing enzymes
catalytically effective at low temperatures. It should be emphasized that the statistical analysis described in this work can
only detect general features of enzyme cold adaptation and
overlooks the subtle structural modifications that can be reliably
identified by detailed intra-family structural comparisons (e.g.
Maes et al., 1999).
Materials and methods
Data collection
Protein sequences were retrieved from the SWISS-PROT, PIR,
EMBL and NRL3D databanks using the Sequence Retrieval
System (SRS) (Etzold and Argos, 1993). An initial search was
carried out with the keywords: ‘psychro’, ‘cold’, ‘arctic’,
‘antarctic’ and the like. The names of cold-adapted species,
from which the selected protein sequences were extracted,
141
G.Gianese, P.Argos and S.Pascarella
were used in turn as keywords to check further for the presence
of other proteins from the same source in the databanks. A
literature scrutiny was undertaken to ensure that only proteins
with proved ‘cold-adapted’ features and clear enzymatic activity were considered among those previously retrieved. Subsequently, each of the selected cold-adapted proteins was used
as query sequences in the program FASTA (Pearson and
Lipman, 1988) to collect homologous mesophilic and thermophilic counterparts from the databanks. Within each protein
family, the sequences were then multiply aligned using the
programs CLUSTAL W (Thompson et al., 1994) or PILEUP
in the Genetics Computer Group suite (Deveraux et al., 1984).
Sequences sharing less than 35% residue identity to the
psychrophilic protein were removed and the remainder were
realigned. Such an identity threshold guarantees a sufficiently
accurate alignment and structural homology (Vogt et al.,
1995). For the same reason, we also excluded incomplete and
ambivalently homologous sequences. Proteins from plants
were not taken into consideration owing to the ambiguous
definition of ‘optimum temperature’ for such organisms. To
limit the comparisons to functional enzymatic units, only
sequences of mature protein were considered (e.g. signal
sequences were removed). Only psychrophilic proteins for
which at least one homologous sequence had a known threedimensional structure were used in the analysis. Structural
data were taken from the Brookhaven Protein Data Bank
(PDB) (Sussman et al., 1998). Multiple sequence alignments
were manually refined to optimize the localization of insertion/
deletions. For each protein family, only one cold-adapted
representative was chosen to avoid oversampling of the same
amino acid exchanges. Among similar psychrophilic enzymes
belonging to the same family, the protein with known crystallographic structure or with the lowest optimum growth temperature was selected.
The optimum growth temperature assigned to each protein
corresponds to the normal living environmental temperature
(or to the average of a range of temperatures of the normal
habitat) for monocellular and ectothermic organisms and to
body temperature for homeothermic organisms. Bacterial
optimum temperatures were taken from Bergey’s Manual of
Systematic Bacteriology (Krieg and Holt, 1984; Sneath et al.,
1986; Staley et al., 1989; Williams et al., 1989), whereas
yeasts and filamentous fungi optimum temperatures were
taken from the Web site of the Deutsche Sammlung von
Mikroorganismen und Zelikulturen GmbH (DSMZ) (URL:
http://www.gbf-braunschweig.de/dsmz/dsmzhome.htm). The
use of host environmental temperatures has several precedents
in the literature (e.g. Querol et al., 1996; Menéndez-Arias and
Argos, 1989). Only homologous proteins from organisms with
growth temperatures 艌22°C were considered.
Preferred amino acid substitutions
Favored amino acid substitutions were calculated from the
multiple alignments using the method of Argos et al. (1979),
adopting the modifications introduced by Menéndez-Arias and
Argos (1989) to remedy the excess of data due to very
similar sequences from related species with the same growth
temperature. Therefore, to cope with statistical overestimation
of residue exchanges, all sequences having an identity of 85%
or more to each other and the same growth temperature were
merged into one. In the amino acid substitution evaluation,
this approach counts exchanges in every alignment position
only once for each different residue type found, irrespective
142
of its occurrences. For example, if only a Val was observed
at an alignment position in four closely related sequences and
the psychrophilic one had Ala, only one exchange Val → Ala
was counted. In the presence of different amino acids, each
exchange type counted once.
The substitution matrix was calculated by comparing each
protein sequence in a multiple alignment with the psychrophilic
counterpart. If the possible pairwise sequence comparisons
were n, the cij elements of a temperature-weighted average
exchange matrix can be calculated according to
Σ (∆T )a
n
cij ⫽
ij
n
Σ
(1)
(∆Tn)2
n
where aij is the number of times an amino acid of type i in a
mesophilic or thermophilic protein has changed to one of type
j in the homologous psychrophilic protein in a pairwise
comparison. ∆T is the absolute difference between the optimum
growth temperatures of the two species in degrees Celsius.
Equation 1 shows that the weight (∆Tn)2 of identified substitutions is variable according to temperature differences.
The favored residue substitutions in a single protein family
can be calculated by Equation 1. The overall exchange matrix
for k protein families was calculated according to
Cij ⫽
Σ (c )
(2)
ij k
k
which defines elements of a matrix C giving equal weight to
all the families, irrespective the number of sequences or the
following equation (Menéndez-Arias and Argos, 1989):
[
Σ [Σ
Σ Σ (∆T )a
k
C⬘ij ⫽
n
n
k
]
]
ij
k
(3)
(∆Tn)2
n
k
which defines elements of a matrix C⬘ assigning larger weight
to families with more members. Then, the cross-elements in
the C or C⬘ matrices were subtracted from each other to obtain
a new matrix Dij (equal to: Cij – Cji) or D⬘ij (equal to: C⬘ij –
C⬘ji), representing the net residue substitution flow in the
direction from non-cold to cold-adapted species. The standard
deviations for the non-zero elements of matrices D and D⬘
were determined. The significance R of each exchange was
calculated by dividing each of the D or D⬘ elements by the
standard deviation of the respective matrix.
Equation 3 was also used in the analysis of the amino acid
composition of the sequences (Böhm and Jaenicke, 1994; Vogt
et al., 1997). In this case, the aij coefficient was substituted
by the difference in amino acid composition between the coldadapted enzyme and thermo/mesophilic counterparts in each
comparison. For each protein family of k members, where the
kth is the psychrophilic protein, the average composition
difference ci for each of the 20 amino acids ai was calculated as
k–1
n
ci ⫽
[
Σ ∆T N Σ (a ) –N Σ (a )
n⫽1
–1
k
–1
n
i l
l
k–1
Σ (∆T
n⫽1
n
)2
]
i m
m
(4)
Cold adaptation of enzymes
where N is the size of the sequence considered and l and m
are the number of times the amino acid ai is present in the
psychrophilic and in the related protein, respectively.
The significance of the observed differences in amino
acid composition (temperature weighted and extended to all
families) was measured by transforming its value into a Zscore (difference between each value and the overall mean
divided by the standard deviation).
Model building
Secondary structures, solvent accessibility and subunit interface
residues were assigned to the psychrophilic proteins with
unknown three-dimensional structure by comparative
modeling. Homology models were based on the available
crystal structures in the selected families. Template and target
sequences were multiply aligned and the resulting alignments
were checked by visual inspection of superposition of template
three-dimensional structures. The program MODELLER
version 4.0 (Šali and Blundell, 1993) was used to build the
models. Four models at the highest optimization level were
built for each target protein. The model displaying the lowest
‘objective function’ value was selected among the four. Model
quality was assessed with the program ProsaII (Sippl, 1993).
Whenever applicable, multimeric biological units were recreated from the monomers using the symmetry information in
the template structures. Secondary structures were determined
with the program DSSP (Kabsch and Sander, 1983) and were
assigned to α-helix (DSSP symbols H, G, I), β-strand (B and
E) or coil (the rest). Solvent accessibility was calculated from
atom coordinates with the program NACCESS (Hubbard and
Thornton, 1993). Structure sites displaying not more than 0.05
and not less 0.25 fractional accessibility were considered
buried and exposed, respectively (Pascarella et al., 1998).
During accessibility calculations, only physiological ligands
(cofactors, ions, etc.) were mantained. In allowance for residues
located at subunit interfaces, they were considered to be located
at the interface if they lost at least 15% of the accessible
surface area of all their atoms upon subunit association
(Menéndez-Arias and Argos, 1989). Water molecules were
always excluded from the structures in all calculations.
Amino acid exchanges and composition in different
structural environments
Structural environments taken into consideration were (i)
secondary structure (α-helix, β-strand or random coil); (ii)
accessibility state (buried or exposed) and (iii) subunit interface.
Propensities Pij for a residue exchange from type i to type j
were calculated according to
Pij ⫽
(C⬘ij)env /(C⬘ij)tot
(Na)env /(Na)tot
(5)
The terms (C⬘ij)env and (C⬘ij)tot represent elements of the
exchange matrices calculated for residues observed in the
structural environments (env) (i), (ii) or (iii) and for the whole
sequences (tot). (Na)env and (Na)tot are the number of amino acids
counted in the structural environments and in all sequences,
respectively. Residue compositional differences at the structural
environments (i) and (ii) were calculated with Equation 4.
Results
The data selection procedure yielded 21 psychrophilic enzymes
from different families and 427 homologous sequences (Table I)
accounting for 112 117 amino acid comparisons. Fifty homo-
logs, distributed among 15 protein families (Table I), are from
thermophilic bacteria able to grow above 60°C and 27 of these
belong to hyperthermophilic organisms adapted at temperatures
above 80°C (Jaenicke and Böhm, 1998). Six psychrophilic
enzymes have known three-dimensional structure while the
remaining 15 families include at least one homologous member
with solved spatial structure (Table II). Entire D⬘ matrix and
counts of amino acid exchanges observed in our dataset are
shown in Figure 1. While matrix D is computed without any
correction for family size, D⬘ relies on a weighting scheme
that gives more emphasis to families bearing a higher number
of sequences. Comparison of the two matrices should indicate
the residue exchanges biased by statistical noise. Therefore,
only the residue exchanges scored by both matrices with a
significance higher than 2.0 were considered. Seven exchanges
fulfill such a condition and, among these, three have a
significance above 3.0 (Table III). To emphasize the residue
exchanges affecting protein stability at extreme temperatures,
we compared only the thermophilic and psychrophilic enzymes
and recalculated the two matrices D and D⬘. This reduced data
set contains 65 sequences, 15 from psychrophilic and 50
from thermophilic organisms, accounting for 18 338 residue
exchanges. The results confirmed the top four exchanges
reported in Table III at a significance level higher than 2.0.
Preferred structural environments for the exchanges reported
in Table III were analyzed. Secondary structure and solvent
accessibility assignments to 15 psychrophilic proteins with no
available spatial structure required the application of homology
modeling. The percentage of residue identity between the
structural templates and the target sequences ranged from 34
to 73% and eight cold enzymes bore more than one structural
homolog (Table II). This assured a sufficient degree of modeling
accuracy (Šali et al., 1995). Template–target alignments were
taken from the literature. In all cases, however, they were
checked and, when appropriate, alignments were modified at
insertion/deletion regions to optimize the target model or to
include more templates (an example is shown in Figure 2).
Table III reports the propensities of the most significant
exchanges for different structural environments. Fractions of
α-helix, β-strand and coil observed in our sample are 33, 23
and 44%, respectively. Sites displaying not more than 0.05
and not less than 0.25 fractional accessibility were considered
buried and exposed, respectively (Pascarella et al., 1998).
Propensities at subunit interfaces were calculated in the databank subset represented by the 12 enzymes displaying a
quaternary structure (Table II) with 263 homologous sequences.
Discussion
We have described a systematic comparative analysis of
21 psychrophilic enzymes from different families and 427
homologous sequences and structures. This is the largest
dataset of psychrophilic enzymes analyzed to date. Our analysis
is aimed at detecting general features of structural adaptation
of enzymes to low temperatures. Enzyme-specific strategies
can only be identified by a detailed intra-family structural
comparison.
The results indicate three residue exchanges (namely Glu
→ Ala, Val → Ala, Arg → Lys) scored at a significance level
higher than 3.0 by both matrices D and D⬘. Statistical tendencies
are also observed for four other residue substitutions involving
charged residues (Lys → Ser, Lys → Asn, Arg → Ser) and
hydrophobic residues (Val → Ile). These observations strongly
suggest that charged residues Glu, Arg and Lys (Table III)
143
G.Gianese, P.Argos and S.Pascarella
Table I. List of psychrophilic enzymes used in this work
No. Family name
Species
Growth
temperature (°C)
Databank code or
referencea
Number of
homologsb
1
2
3
Bacillus psychrosaccharolyticus
Alteromonas haloplanctis A23
Vibrio sp. 2693
15
4
6
embl:ab021683
sw:amy_altha
sw:pyrb_vibs2
8 (1)
39
13 (5)
383
453
310
6
sw:pyri_vibs2
4 (1)
153
5
6
7
8
9
10
11
12
13
14
15
16
17
Alanine racemase
α-Amylase
Aspartate carbamoyltransferase
(catalytic chain)
Aspartate carbamoyltransferase
(regulatory chain)
β-Galactosidase
β-Lactamase
Chymotrypsin A
Citrate synthase
DNA ligase
Elastase
Isocitrate dehydrogenase 1
3-Isopropylmalate dehydrogenase
L-Lactate dehydrogenase P
Malate dehydrogenase
Ornithine carbamoyltransferase
Pyruvate kinase
Serralysin (alkaline protease)
18
19
20
21
Subtilisin
Triosephosphate isomerase
Trypsin I
Xylanase
4
aNotation
bNumbers
Vibrio sp. 2693
Arthrobacter sp. B7
Psychrobacter immobilis A5
Gadus morhua
Antarctic bacterium DS2-3R
Pseudoalteromonas haloplanktis
Salmo salar
Vibrio sp. ABE-1
Vibrio sp. I5
Bacillus psychrosaccharolyticus
Aquaspirillium arcticum
Vibrio sp. 2693
Bacillus psychrophilus
Pseudomonas aeruginosa
TACII18
Bacillus sp. TA39
Vibrio marinus
Salmo salar
Cryptococcus adeliae TAE85
15
4
4
5
4
4
4
15
15
4
6
15
4
sw:bgal_artsp
sw:ampc_psyim
sw:ctra_gadmo
sw:cisy_abds2
embl:af126866
nrl3d:1elt
sw:idh1_viba1
Wallon et al. (1997)
sw:ldhp_bacps
sw:mdh_aquar
sw:otca_vibs2
sw:kpyk_bacpy
embl:psy17314
2 (1)
11
6
13 (2)
22 (8)
15
10 (4)
45 (6)
42 (6)
9 (1)
25 (7)
51 (3)
9
1015
362
245
379
672
236
414
360
318
329
301
586
463
4
15
4
4
sw:subt_bacs9
sw:tpis_vibma
sw:try1_salsa
embl:cay15434
19 (1)
41 (3)
35
8 (1)
309
256
222
338
is ‘databank:protein code’.
in parentheses denote the number of homologous sequences from thermo/hyperthermophilic organisms.
tend to be replaced in psychrophilic enzymes. These substitutions occur mainly at exposed sites within α-helices or coil
regions. Only the substitution Lys → Asn is favoured at the
subunit interface with propensity 1.38. Compositional analysis
indicates that Asn is more frequent in psychrophilic than in
meso/thermophilic enzymes (Z-score 2.0). Asn is a thermolabile residue and its increased frequency in psychrophiles was
already noted in aspartate carbamoyltransferase (Xu et al.,
1998). The substitution Arg → Lys was already observed to
occur in α-helices in the thermophilic → mesophilic direction
(Menéndez-Arias and Argos, 1989). It has been suggested by
several authors (e.g. Vogt et al., 1997; Xiao and Honig, 1999)
that ion pairs and H-bonds and more generally electrostatic
interactions in which charged residues are involved play an
important role in protein stabilization, particularly at high
temperature, also because of the influence on polypeptide
chain flexibility. Likewise, other authors (Feller et al., 1996;
Feller and Gerday, 1997; Marshall, 1997) state, on the basis
of comparative single-family analysis, that decreased Arg
content or decreased Arg/(Lys ⫹ Arg) molar ratio is a feature
of cold adaptation. Our results support the view that substitution
of charged residues Glu, Arg and Lys be one of the mechanisms
of low-temperature adaptation shared by most of the families
included in our databank and possibly by most of the coldadapted proteins.
The two substitutions involving exclusively hydrophobic
residues, namely Val → Ala and Val → Ile, prefer buried sites
in α-helices and β-strands, respectively. The replacement Val
→ Ile was detected in the direction thermophile → mesophile
by Menéndez-Arias and Argos (1989) in α-helices. Vogt et al.
(1997) assigned it to the opposite direction mesophile →
thermophile in β-strands using a different dataset smaller than
that used in the present work. Both residues are highly preferred
in β-strands (Levitt, 1978). Ile can establish larger Van der
Waals contacts with the surrounding residues. It should be
144
Sequence length
(monomer)
mentioned that a mutation Ile → Val in a short β-strand of
chymotrypsin inhibitor 2 (Jackson et al., 1993) stabilized the
structure. It is reasonable to expect that the inverse mutation
may destabilize some proteins. Val → Ala was also detected
by Vogt et al. (1997) in α-helices, although in the opposite
direction mesophile → thermophile. Ala has a clear preference
for α-helical conformations while Val displays a propensity
for β-strands. Psychrophiles show avoidance for Val in buried
regions of β-sheets. Compositional analysis indicates that Val
is less frequent in buried regions (Z-score –3.0) and in βstrand conformations (Z-score –2.0) of psychrophilic enzymes
than in the corresponding regions of thermo/mesophiles. The
overall effect of the two mutations considered, Val → Ala,
Val → Ile, is the decrease in the number of carbon atoms in
the hydrophobic core. It can be calculated from Table III that
six side-chain carbon atoms are replaced by five carbons.
Structural comparison between salmon and bovine trypsins
(Smålas et al., 1994) revealed that the combined effect of
three of the six residue exchanges located in the interior of
salmon trypsin is the reduction of the side-chain volume.
Compositional analysis indicates increased content of Ala
particularly at exposed sites in coil regions (Z-score 2.0). An
increased presence of Ala at exposed sites in place of hydrophilic residues (see the most significant Glu → Ala substitution
in Table III) may enhance surface hydrophobicity. Favorable
solvation of the non-polar surface at low temperatures should
destabilize the entire structure (Creighton, 1991). This destabilization may increase flexibility. A significant increase in the
apolar surface exposed to solvent was observed in citrate
synthase from Antarctic bacterium (Russell et al., 1998), in
trypsin from Atlantic salmon (Smalås et al., 1994) and in αamylase from Alteromonas haloplanctis (Aghajari et al., 1998).
These enzymes display a reduced number of charged residues
on the surface compared with their mesophilic and thermophilic
counterparts. It is a general assumption that thermophilicity is
Cold adaptation of enzymes
Table II. (A) List of structurally solved homologs used as templates to build homology models of psychrophilic proteins and (B) list of psychrophilic
enzymes with known structure
(A)
Protein name
Species
PDB code
Structure
resolution (Å)
Identitya
(%)
Number of
subunits
Alanine racemase
Aspartate carbamoyltransferase (catalytic chain)
Aspartate carbamoyltransferase (regulatory chain)
β-Galactosidase
β-Lactamase
Bacillus stearothermophilus
Escherichia coli
Escherichia coli
Escherichia coli
Enterobacter cloacae
Escherichia coli
Bos taurus
Bacillus stearothermophilus
Escherichia coli
Thiobacillus ferrooxidans
Bacillus coagulans
Salmonella typhymurium
Escherichia coli
Thermus thermophilus
Bacillus stearothermophilus
Lactobacillus casei
Thermotoga maritima
Pyrococcus furiosus
Escherichia coli
Pseudomonas aeruginosa
Escherichia coli
Oryctolagus cunuculus
Saccaromyces cerevisiae
Leishmania mexicana
Pseudomonas aeruginosa
Serratia marcescens
Serratia sp. E-15
Bacillus sp. KSM-K16
Bacillus lentus (subtilisin Savinase)
Bacillus lentus (subtilisin BL)
Bacillus subtilis (subtilisin DY)
Penicillium simplicissimum
Thermoascus aurantiacus
1bd0
8atc
8atc
1bgl
1bls
2bls
1ab9
1b04
1ai3
1a05
2ayq
1cnz
1cm7
1xaa
1ldn
1llc
1a5z
1a1s
2otc
1ort
1pky
1a5u
1a3x
1pkl
1kap
1sat
1srp
1mpt
1gci
1st3
1bh6
1b31
1tax
1.60
2.50
2.50
2.50
2.30
2.00
1.60
2.80
1.90
2.00
3.00
1.76
2.06
2.10
2.50
3.00
2.10
2.70
2.80
3.00
2.50
2.35
3.00
2.35
1.64
1.75
2.00
2.40
0.70
1.40
1.75
1.75
1.14
57
73
54
34
39
37
67
48
73
62
55
53
53
51
70
56
42
43
36
36
48
43
41
41
67
56
55
41
41
41
38
40
38
2
6
6
4
1
1
1
1
2
2
2
2
2
2
4
4
4
12
3
12
4
4
4
4
1
1
1
1
1
1
1
1
1
Protein name
Species
PDB code
Structure
resolution (Å)
Number of
subunits
α-Amylase
Citrate synthase
Elastase
Malate dehydrogenase
Triosephosphate isomerase
Trypsin I
Alteromonas haloplanktis A23
Antarctic bacterium DS2-3R
Salmo salar
Aquaspirillium arcticum
Vibrio marinus
Salmo salar
1aqm
1a59
1elt
1b8p
1aw2
2tbs
1.85
2.09
1.61
1.90
2.65
1.80
1
2
1
2
2
1
Chymotrypsin A
DNA ligase (N-terminal domain)
Isocitrate dehydrogenase 1
3-Isopropylmalate dehydrogenase
L-Lactate
dehydrogenase P
Ornithine carbamoyltransferase
Pyruvate kinase
Serralysin (alkaline protease)
Subtilisin
Xylanase
(B)
aPercentage
of residue identity shared with the sequence of the psychrophilic homolog.
correlated with rigidity of the protein and that psychrophilicity
should be reflected by a more flexible protein structure, the
consequence of which is considered by many authors as
thermolability. However, the question remains as to whether
the thermal instability is a real consequence of the structure
flexibility or is correlated with the lack of selective pressure
related to the stability. A more flexible structure, in fact,
reduces the energetic cost of the conformational changes
required to interact with the substrate. The higher specific
activity can be explained by a lower activation energy resulting
from an easier accommodation of the substrate at low and
moderate temperatures. However, theoretical analyses were
unable to detect an overall increase in flexibility in salmon
trypsin compared with bovine trypsin (Heimstad et al., 1995)
but suggested that there may be significant differences on a
more detailed level.
It has been suggested that enhancement of flexibility to
achieve optimum enzymatic activity at low temperatures
implies a decreased Pro content, especially at loop regions
(Aghajari et al., 1998). No significant exchange involves Pro
residues in our data set, although the exchange Pro → Ala has
significance equal to 2.0 only in matrix D⬘. Compositional
analysis suggests a decreased Pro content only with a marginal
statistical significance (Z-score –0.7). It cannot be excluded,
however, that the lack of significant residue exchanges involving Pro is a result of scant statistics. Indeed, net residue
substitution flux calculated over single families indicates that
Pro → Ala exchange is significant only in the α-amylase
family. Family specific adaptations are also evident in the
psychrophilic triose-phosphate isomerase (Alvarez et al., 1998)
that displays a Ser → Ala substitution. Site-directed mutagenesis experiments proved that this substitution is relevant for
145
G.Gianese, P.Argos and S.Pascarella
Fig. 1. (a) Significance of the D⬘ matrix exchanges and (b) counts of residue substitutions observed from thermo/mesophilic to psychrophilic enzymes. Amino
acid residues are indicated with the one-letter code.
146
Cold adaptation of enzymes
Table III. Most significant amino acid substitutions
From meso/thermophile to
psychrophile
Glu → Ala
Val → Ala
Arg → Lys
Lys → Ser
Lys → Asn
Arg → Ser
Val → Ile
Significancea
Propensityb
D⬘
D
Helix
Sheet
Coil
5.7
3.9
3.4
2.6
2.4
2.2
2.1
5.0
4.1
3.0
3.4
2.5
2.6
2.2
2.20
1.39
1.48
1.09
1.45
1.23
0.92
0.10
0.97
0.88
0.71
0.51
0.49
2.38
0.55
0.72
0.70
1.08
0.91
1.09
0.35
Favored
regions
Helix
Helix
Helix
Helix/coil
Helix
Helix/coil
Sheet
Propensityc
Buried
Exposed
0.20
1.79
0.16
0.11
0.43
0.16
2.05
1.87
0.46
1.75
1.84
1.58
1.78
0.22
Favored accessibility
state
Exposed
Buried
Exposed
Exposed
Exposed
Exposed
Buried
aThe
significance of residue substitutions is reported as the ratio between the value and the standard deviation of the non-zero elements of the matrices D and
D⬘. Only substitutions with a significance ⬎2.0 in both matrices D and D⬘ are reported.
bPropensity is defined as the ratio between the fraction of an amino acid exchange at each type of secondary structure and the fraction of secondary structural
content in all the three-dimensional structures.
cPropensity is defined as the ratio between the fraction of an amino acid exchange at a buried or exposed position and the fraction of residues found at the
respective position in all the three-dimensional structures.
Fig. 2. Comparison between (a) part of the multiple amino acid sequence
alignment utilized in the present work for model building and (b) the
corresponding portion reported by Wallon et al. (1997). Target sequence is
3-isopropylmalate dehydrogenase from the psychrophilic Vibrio sp. I5 (code
VIBI5) and it is aligned to the structural templates from Thermus
thermophilus (THETH), Bacillus coagulans (BACCO), Thiobacillus
ferrooxidans (THIFE), Salmonella typhimurium (SALTY) and Escherichia
coli (ECOLI). The sequence segment encompassed by positions 35–40 is
folded as a β-strand. Residues are indicated in one-letter code and
numbering system refers to the Vibrio sp. I5 sequence.
low-temperature adaptation. Indeed, it is clear that adaptation
to extreme environments can be achieved with different strategies in different enzyme families (Argos et al., 1979; Jaenicke
and Böhm, 1998). For example, the exchange Ala → Gly was
detected in an α-helix of malate dehydrogenase (MDH) from
Aquaspirillium arcticum (Kim et al., 1999). Compared with
the Thermus flavus MDH, this Gly represents the only mutation
among the residues interacting with the oxaloacetate substrate.
The authors interpreted this observation in terms of increased
local flexibility that should contribute to the catalytic efficiency.
The evolutionary strategy of adaptation to low temperatures
does not seem to be merely the inverse of adaptation to
high temperature. Psychrophilic enzymes evolved at the low
boundaries of the biological temperature range and had to face
peculiar thermodynamic challenges (Gerday et al., 1997).
While thermophilic proteins need to optimize thermostability
to prevent hot denaturation, psychrophiles need to compensate
for the reduction in chemical reaction rate inherent to low
temperatures and to resist cold denaturation. Two of our seven
significant residue exchanges (Table III), namely Arg → Lys
and Val → Ile, were detected by Menéndez-Arias and Argos
(1989) and Vogt et al. (1997) in the equivalent direction
thermophile → mesophile. In particular, Arg → Lys was
observed by Menéndez-Arias and Argos (1989) in α-helices,
as in our case, while Vogt et al. (1997) assigned it to both
α-helices and β-strands. Val → Ile was observed by MenéndezArias and Argos (1989) in α-helices while in our sample it
occurs in β-strands. It can be speculated that the psychrophilicspecific substitutions meet the peculiar requirements of enzyme
function at low temperatures such as resistance to cold denaturation, local flexibility, balancing of excess flexibility and
the like. Our research suggests that a common adaptive
mechanism of enzymes to low temperature consists of reduction
of charged residues (mainly Arg, Glu and Lys) at exposed
sites in α-helix or coil regions. Vogt et al. (1997) stress that
the strategy to gain stability in thermophilic enzymes exploits
the increase of the number of H-bonds. The preferred amino
acid exchanges observed from thermo/mesophiles to psychrophiles and compositional analysis indicate also a decreased
number of side-chain potential H-bonds and salt bridges in
cold-adapted enzymes. Indeed, it can be calculated from Table
III that 10 side-chain N ⫹ O atoms in meso/thermophiles are
replaced by five N ⫹ O side-chain atoms in psychrophiles. In
this respect, the strategy of cold adaptation seems to use the
same principle as the hot adaptation, namely an increase in
the number of H-bonds/ion pairs on going from low to
high temperatures. Indeed, a smaller number of electrostatic
interdomain interactions was found in α-amylase from
Alteromonas haloplanctis (Aghajari et al., 1998), whereas a
significant decreased number of intersubunit ion pairs and ion
pair networks was observed in malate dehydrogenase from
Aquaspirillium arcticum (Kim et al., 1999) and in citrate
syntase from Antarctic bacterium (Russell et al., 1998).
It is interesting to note that the recalculation of amino acid
exchanges using only the comparison among thermophilic and
psychrophilic sequences confirmed the top four exchanges
reported in Table III. The data set now contains only 15
families with a total of 65 sequences, yet the statistical trend
is mantained. This strengthens the view that the transition
from thermo- to psychrophilic enzymes is achieved mainly by
a decrease in electrostatic interactions and possibly by alteration
of the tight packing of side-chains in the hydrophobic core.
It is suggested that protein engineering aimed at producing
a cold-adapted enzyme should plan at first the replacement of
one or more of the charged residues Arg, Glu, Lys at exposed
sites on α-helices with one of the amino acids indicated in
147
G.Gianese, P.Argos and S.Pascarella
Table III. Simultaneously or alternatively, replacement of Val
at buried sites on α-helices with Ala can also be tested.
It should be mentioned that it has been demonstrated by
‘evolutionary engineering’ (Taguchi et al., 1999) that several
alternative strategies are practicable to achieve psychrophiliclike enzymes. Indeed, a psychrophilic-like subtilisin ‘evolved’
from a mesophilic one upon incorporation of three mutations
(two Ala → Thr and Ala → Val), none of which is observed
in our sample (Table III). However, this cold-adapted enzyme
displayed only a 70% increase in kcat/KM at 10°C over the
starting mesophilic enzyme, while normally psychrophilic
enzymes display a several-fold increase compared with the
mesophilic counterpart at the same temperature. Perhaps Nature
adopted the most expedient mutations amongst different alternatives.
Acknowledgements
This work was partially supported by grants from the Consiglio Nazionale
delle Ricerche (grant No. 97.02338.12) and from the Ministero dell’Università
e della Ricerca Scientifica e Tecnologica. We are indebted to Professor
G.Marino and Dr L.Birolo for introducing us to the problem of psychrophilic
enzyme adaptation during a scientific collaboration. We are grateful to
Professor F.Bossa for support and critically reviewing the manuscript. This
work will be submitted by G.G. in partial fulfilment of the requirements of
the degree of Dottorato di Ricerca at the Università di Roma La Sapienza.
References
Aghajari,N., Feller,G., Gerday,C. and Haser,R. (1998) Structure, 6, 1503–1516.
Alvarez,M., Zeelen,J.P., Mainfroid,V., Rentier-Delrue,F., Martial,J.A.,
Wyns,L., Wierenga,R.K. and Maes,D. (1998) J. Biol. Chem., 273, 2199–
2206.
Argos,P., Rossmann,M.G., Grau,U.M., Zuber,H., Frank,G. and Tratschin,J.D.
(1979) Biochemistry, 18, 5698–5703.
Böhm,G. and Jaenicke,R. (1994) Int. J. Pept. Protein Res., 43, 97–106.
Creighton,T.E. (1991) Curr. Opin. Struct. Biol., 1, 5–16.
Deveraux,J., Haeberli,P. and Smithies,O. (1984) Nucleic Acids Res., 12,
387–395.
Etzold,T. and Argos, P. (1993) Comput. Appl. Biosci., 9, 45–57.
Feller,G. and Gerday,C. (1997) Cell. Mol. Life Sci., 53, 830–841.
Feller,G., Narinx,E., Arpigny,J.L., Aittaleb,M., Baise,E., Genicot,S. and
Gerday,C. (1996) FEMS Microbiol. Rev., 18, 189–202.
Fontana A. (1991) Curr. Opin. Biotechnol., 2, 551–560.
Gerday,C., Aittaleb,M., Arpigny,J.L., Baise,E., Chessa,J.-P., Garsoux,G.,
Petrescu,I. and Feller,G. (1997). Biochim. Biophys. Acta, 1342, 119–131.
Gerday,C. et al. (2000) Trends Biotechnol., 18, 103–107.
Heimstad,K.S., Hansen,L.K. and Smalås,A.O. (1995). Protein Eng., 8, 379–
388.
Hubbard,S.J. and Thornton,J.M. (1993) NACCESS, Computer Program.
Department of Biochemistry and Molecular Biology, University College
London.
Jackson,S.E., Moracci,M., elMasry, N., Johnson, C.M. and Fehrst, A.R. (1993)
Biochemistry, 32, 11259–11269.
Jaenicke,R. and Böhm,G. (1998) Curr. Opin. Struct. Biol., 8, 738–748.
Kabsch,W. and Sander,C.(1983) Biopolymers, 22, 2577–2637.
Kim,S.-Y., Hwang,K.Y., Kim,S.-H., Sung,H.-C., Han,Y.S. and Cho,Y. (1999)
J. Biol. Chem., 271, 11761–11767.
Krieg,N.R. and Holt,J.G. (eds) (1984) Bergey’s Manual of Systematic
Bacteriology, vol. 1. Williams and Wilkins, Baltimore, MD.
Levitt,M. (1978) Biochemistry, 17, 4277–4285.
Maes,D. et al. (1999) Proteins, 37, 441–453.
Margarit,I., Campagnoli,S., Frigerio,F., Grandi,G., De Filippis,V. and
Fontana,A. (1992) Protein Eng., 5, 543–550.
Marshall,C.J. (1997) Trends Biotechnol., 15, 359–364.
Menéndez-Arias,L. and Argos, P. (1989) J. Mol. Biol., 206, 397–406.
Pascarella,S., De Persio,R., Bossa,F., Argos,P. (1998) Proteins, 32, 190–199.
Pearson,W.R. and Lipman,D.J. (1988) Proc. Natl Acad. Sci. USA, 85, 2444–
2448.
Querol,E., Perez-Pons,J.A. and Mozo-Villarias,A. (1996) Protein Eng., 9,
265–271.
Russell,R.J.M., Gerike,U., Danson,M.J., Hough,D.W. and Taylor,G.L. (1998)
Structure, 6, 351–361.
Šali,A. and Blundell,T.L. (1993) J. Mol. Biol., 234, 779–815.
148
Šali,A., Potterton,L., Yuan,F., van Vlijmen,H. and Karplus, M. (1995) Proteins,
23, 318–326.
Sippl,M. (1993) Proteins, 17, 355–362.
Smalås,A.O., Heimstad,E.S., Hordvik,A., Willassen,N.P. and Male,R. (1994)
Proteins, 20, 149–166.
Sneath,P.H.A, Mair,N.S., Sharpe,M.E. and Holt,J.G. (eds) (1986) Bergey’s
Manual of Systematic Bacteriology, vol. 2. Williams and Wilkins,
Baltimore, MD.
Staley,J.T., Bryant,M.P., Pfennig,N. and Holt,J.G. (eds) (1989) Bergey’s Manual
of Systematic Bacteriology, vol. 3. Williams and Wilkins, Baltimore, MD.
Sussman,J.L., Lin,D., Jiang,J., Manning,N.O., Prilusky,J., Ritter,O. and
Abola,E.E. (1998) Acta Crystallogr., D54, 1078–1084.
Szilágyi,A. and Závodszky,P. (2000) Structure, 8, 493–504.
Taguchi,S., Ozaki,A., Nonaka,T., Mitsui,Y. and Momose,H. (1999) J. Biochem.,
126, 689–693.
Thompson,J.D., Higgins,D.G. and Gibson,T.J. (1994) Nucleic Acids Res., 22,
4673–4680.
Vogt,G., Etzold,T. and Argos,P. (1995) J. Mol. Biol., 249, 816–831.
Vogt,G., Woell,S. and Argos,P. (1997) J. Mol. Biol., 269, 631–643.
Wallon,G., Lovett,S.T., Magyar,C., Svingor,A., Szilagyi,A., Zàvodszky,P.,
Ringe,D. and Petsko,G.A. (1997) Protein Eng., 10, 665–672.
Williams,S.T., Sharpe,M.E. and Holt,J.G. (eds) (1989) Bergey’s Manual of
systematic Bacteriology, vol. 4. Williams and Wilkins, Baltimore, MD.
Xiao,L. and Honig,B. (1999) J. Mol. Biol., 289, 1435–1444.
Xu,Y., Zhang,Y., Liang,Z., Van de Casteele,M., Legrain,C. and Glansdorff,N.
(1998) Microbiology, 144, 1435–1441.
Received July 28, 2000; revised November 17, 2000; accepted December
20, 2000