Review Molecular adaptations to cold in psychrophilic enzymes

CMLS, Cell. Mol. Life Sci. 60 (2003) 648–6621420-682X/03/040648-15DOI 10.1007/s00018-003-2155-3© Birkhäuser Verlag, Basel, 2003CMLS Cellular and Molecular Life SciencesReviewMolecular adaptations to cold in psychrophilic enzymesG. FellerLaboratory of Biochemistry, Institute of Chemistry B6, University of Liège, 4000 Liège (Belgium),Fax +32 4 366 33 64, e-mail: gfeller@ulg.ac.beReceived 21 June 2002; received after revision 9 September 2002; accepted 10 September 2002Abstract. Psychrophiles or cold-loving organisms successfullycolonize cold environments of the Earth’s biosphere.To cope with the reduction of chemical reactionrates induced by low temperatures, these organisms synthesizeenzymes characterized by a high catalytic activityat low temperatures associated, however, with low thermalstability. Thanks to recent advances provided by X-ray crystallography, protein engineering and biophysicalstudies, we are beginning to understand the molecularadaptations responsible for these properties which appearto be relatively diverse. The emerging picture suggeststhat psychrophilic enzymes utilize an improved flexibilityof the structures involved in the catalytic cycle,whereas other protein regions if not implicated in catalysismay or may not be subjected to genetic drift.Key words. Psychrophile; extremophile; enzyme kinetics; crystal structure; folding; mutagenesis; biophysics.The current contextSince our last review in this journal [1], significant advanceshave been made in the study of proteins and enzymesfrom cold-adapted organisms, the so-called psychrophiles,demonstrating an intense research activity inthe field. Such interest is certainly motivated by the factthat life at extremely low temperatures is no longer regardedas a biological curiosity. Indeed, the cold-activeand heat-labile psychrophilic enzymes possess interestingbiotechnological potential [2–6] and have beenshown to be useful models for protein folding studies [7]and for the analysis of the adaptability of the structureand function of enzymes [8]. In addition, a low-temperatureorigin of life is still a plausible hypothesis [9]. Moreexotic reasons for this interest should also be mentioned.For example, living remnants of ancestral microorganisms,such as those expected to survive below 3500 m ofice in the Antarctic subglacial Lake Vostok [10], shoulddisplay psychrophilic characters, as should the possibleliving forms on the planet Mars [11]. Psychrophiles werefirst referred to as cold-adapted bacteria [12]. However, ifa psychrophile is defined as an organism living permanentlyat temperatures close to the freezing point of waterin thermal equilibrium with the medium, such a definitionincludes de facto a large range of species: yeast[13–16], algae [17–19], marine invertebrates and insectsor polar fish, which are the largest psychrophiles [20, 21].These examples underline that psychrophiles are numerous,taxonomically diverse and have a widespread distribution.One should not forget that the major part of Earthis exposed to low temperatures if one considers the vastextent of permanently cold environments such as theAntarctic, Arctic and mountain regions as well as thedeep-sea waters covering three-quarters of the planet surface.To define the properties of a cold-active enzyme, the effectof temperature on the activity of psychrophilic andmesophilic enzymes is illustrated in figure 1. This graphreveals three basic features which will form the body ofthe next sections.1) To compensate for the slow reaction rates at low temperatures,psychrophiles synthesize enzymes with an

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 649Figure 1. Effect of temperature on the activity of psychrophilic ()and mesophilic () enzymes. The thermodependence of the activity(of a-amylases in this case) illustrates the three main adaptive traitsof psychrophilic enzymes.up to tenfold higher specific activity in this temperaturerange. This is in fact the main physiological adaptationto cold at the enzyme level.2) The temperature for apparent maximal activity forcold-active enzymes is shifted towards low temperatures,reflecting the weak stability of these proteinswhich are prone to unfolding and inactivation at moderatetemperatures.3) Finally, the adaptation to cold is frequently not perfect.Figure 1 illustrates that the specific activity of the psychrophilicenzyme at low temperatures, although veryhigh, remains generally lower than that of themesophilic enzymes at 37°C.Some reviews have been published recently by our group[22–26] but other excellent reviews should also be consultedfor a complete coverage of this topic [4, 11,27–33].Recombinant psychrophilic enzymesA major achievement in the study of psychrophilic enzymeswas the availability of recombinant enzymes thathave been produced in sufficient amount to determine X-ray structures and for their analysis by sophisticated biophysicalmethods. This has also opened the way for mutagenesisstudies. The first cold-active enzymes werecloned in Escherichia coli and selected by the activity ofthe gene product expressed by the recombinant colonies[34–36]. Later, the high level of sequence identity withmesophilic homologues demonstrated by psychrophilicenzymes prompted the isolation by the polymerase chainreaction of other genes that cannot be detected in expressionlibraries [37, 38]. Accordingly, the study of rare enzymeswas also made possible, such as the elongationfactors EF-2 [39–41] and Tu [42] or a DNA ligase [43]which only occurs as 200 molecules per cell, thereforeprecluding its direct purification in reasonable amounts.The availability of these genes also allowed an explorationof the overexpression of these psychrophilic enzymesin a mesophilic host. Various E. coli strains wereused for this purpose, whereas expression in Bacillusspecies [44, 45] or yeast was rarely addressed. The firstconcern of workers in the field was to establish if an enzymeexpressed near 0°C by the wild-type strain maintainsits properties when expressed and folded at highertemperatures that are lethal for the parent microorganisms.Indeed, when expressed at a sufficiently low temperature(15–18°C), a recombinant psychrophilic enzymeis undistinguishable from its wild-type parent moleculeas regards kinetic parameters, folding properties[46] and three-dimensional (3D) structure [47]. By contrast,at 37°C, recombinant psychrophilic enzymes arefrequently not expressed or are quickly inactivated. In addition,when expressed at this temperature, the propertiesof the recombinant psychrophilic enzyme seem to be alteredas a result of misfolding or the occurrence of an inactivatedpopulation [46]. Affinity tags added in fusion toa recombinant psychrophilic enzyme for purification purposesare also a concern. A His-tagged phosphoglyceratekinase (PGK) has intermediate kinetic properties betweenthe native psychrophilic PGK and yeast PGK, adifference possibly arising from the stabilization centerintroduced by the affinity tag [48]. As a practical conclusion,it is therefore advisable to express recombinant psychrophilicenzymes at the lowest possible temperature, tocleave and remove affinity tags, to perform an affinitychromatography step (removing the possible inactivefraction) and to perform active-site titration to calculatethe kinetic parameters.Recently, a recombinant psychrophilic enzyme has beenexpressed in psychrophilic bacteria [49]. For this purpose,a broad host range expression vector was constructedusing a cold-adapted replication element derivedfrom a multicopy cryptic plasmid of the Antarctic Gramnegativebacterium Pseudoalteromonas haloplanktis.This study demonstrated that production of recombinantproteins can be performed at low temperatures, revealinga new biotechnological potential for psychrophilicstrains.Structural factors related to activityat low temperatureTo date, four crystal structures of bacterial psychrophilicenzymes have been solved by X-ray crystallography[50–53] as has the structure of three serine proteasesfrom cold-adapted fish [54–56]. Other structures shouldbecome available soon as crystals of three other cold-ac-

650 G. Feller Cold-active enzymestive enzymes have been obtained [57–59]. Unfortunately,only four of these X-ray structures have been comparedwith mesophilic or thermophilic homologues and analyzedin detail with regards to the structural parametersrelated to cold activity. These include the bacterial a-amylase [47], citrate synthase [53], malate dehydrogenase[52] and the fish trypsin [54]. Nevertheless, this limitedset of data has already provided valuable insights intothe molecular basis of cold adaptation [30, 60–62]. As inferredfrom their amino acid sequences and homologymodels, psychrophilic enzymes do not display unusual orexotic conformations. On the contrary, the overall fold isvery close to that of their mesophilic homologues. For example,when the 3D structures of mammalian a-amylasesare superimposed on the psychrophilic a-amylase, theroot mean square deviation is only about 0.5 Å, reflectinga high similarity of the tertiary structures [47].As far as the active site of psychrophilic enzymes is concerned,all reactive side chains as well as most side chainspointing towards the catalytic cavity are strictly conserved.This means that the overall catalytic mechanismand reaction pathway are not modified in cold-active enzymes.This aspect is not really surprising as the specificreaction mechanism of enzymes is not prone to drasticvariation. However, worth mentioning is that all residuesforming the active site can be conserved in a psychrophilicenzyme. This was demonstrated by the X-raystructure solved at high resolution of both the cold-activea-amylase and of its closest structural homologue frompig in complex with acarbose, a pseudosaccharide inhibitormimicking the transition state intermediate [63,64]. Comparison of residues with direct or water-mediatedH bonds to the inhibitor revealed that all 24 residuesinvolved are strictly conserved in the cold-active a-amylase.This outstanding example of active-site identity (interms of side chain identity) demonstrates that the specificproperties of psychrophilic enzymes can be reachedwithout any amino acid substitution in the reaction center.As a consequence, changes occurring elsewhere in themolecule are responsible for the optimization of the catalyticparameters of this enzyme, as demonstrated by sitedirectedmutagenesis [65], probably by modifying the dynamicand conformational properties of the conservedresidues.Another important feature of the active site refers to itsaccessibility, because in many cases, the catalytic cavityappears to be larger and more accessible to ligands. Thislarger opening of the catalytic cleft is achieved in variousways, including small deletions in loops bordering the activesite or by distinct conformation of these loops, and byreplacement of bulky side chains for smaller groups at theentrance. This aspect has been well illustrated for thecold-active citrate synthase [53]. The reasons for such abetter accessibility are still not clearly understood butcold-active enzymes are thought to accommodate substratesat lower energy cost, as far as the conformationalchanges are concerned, and therefore reduce the activationenergy required for the formation of the enzyme-substratecomplex. The larger active site may also facilitateeasier release and exit of products and thus may alleviatethe effect of a rate-limiting step on the reaction rate. Thislarger size of the active site and the better accessibilityhave at least two consequences. First, non-specific enzymesaccepting various substrates should have a broaderspecificity, because substrates with slightly distinct conformationsor sizes can fit and bind to the site. For example,the observed differences in substrate specificity betweenAtlantic salmon and mammalian elastases havebeen interpreted in terms of a somewhat wider and deeperbinding pocket for the cold-adapted elastase [28]. Thebroad specificity of a psychrophilic alcohol dehydrogenase,oxidizing large bulky alcohols, was also assigned toa highly flexible active site [66]. As a second consequence,substrates should bind less firmly in the bindingsite, if no point mutations have occurred, giving rise tohigher K m values. This is certainly one of the structuralexplanations for the observation that numerous coldactiveenzymes, which are strongly homologous tomesophilic counterparts, have lower binding constants.Finally, differences in electrostatic potentials in andaround the active site of psychrophilic enzymes appear tobe a crucial parameter for activity at low temperatures.Interestingly, the cold-active citrate synthase [53], malatedehydrogenase [52] and trypsin [28, 67, 68] are characterizedby marked differences in electrostatic potentialsnear the active-site region when compared with mesophilicor thermophilic counterparts. In the case of malatedehydrogenase for example, the increased positive potentialat and around the oxaloacetate-binding site and thesignificantly decreased negative surface potential at theNADH-binding region may facilitate the interaction ofthe oppositely charged ligands with the surface of the enzyme.In all cases, the differences were caused by discretesubstitutions in non-conserved charged residues resultingin local electrostatic potential differing in both sign andmagnitude. Future work in the field will certainly refinethe structural and functional relationships between electrostaticpotentials and catalytic efficiency in psychrophilicenzymes.Structural factors linked to stabilityThe numerous homology-based models and especiallythe recently solved X-ray structures of psychrophilic enzymeshave deciphered the molecular origin of the weakstability characterizing these proteins. All structural factorscurrently known to stabilize the protein molecule canbe attenuated in strength and number in the structure ofcold-active enzymes [28, 30, 62]. This involves the clus-

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 651tering of glycine residues (providing local mobility), thedisappearance of proline residues in loops (providing enhancedchain flexibility between secondary structures), areduction in arginine residues which are capable of formingmultiple salt bridges and H bonds, as well as a lowernumber of ion pairs, aromatic interactions or H bondscompared to mesophilic enzymes. The size and relativehydrophobicity of non-polar residue clusters forming theprotein core are frequently smaller, lowering the compactnessof the protein interior. The N and C caps of a-helices are also altered (weakening the charge-dipole interaction)and loose or relaxed protein extremities appearto be preferential sites for unzipping. The binding of stabilizingions, such as calcium, can be extremely weak,with binding constants differing from mesophiles by severalorders of magnitude. Insertions and deletions aresometimes responsible for specific properties such as theacquisition of extra surface charges (insertion) or theweakening of subunit interactions (deletion).Calculation of the solvent-accessible area showed thatsome psychrophilic enzymes expose a higher proportionof non-polar residues to the surrounding medium [47,53]. This is an entropy-driven destabilizing factor causedby the reorganization of water molecules around exposedhydrophobic side chains. Calculations of the electrostaticpotential revealed in some cases an excess of negativecharges at the surface of the molecule and, indeed, the pIof cold-active enzymes is frequently more acidic than thatof their mesophilic or thermophilic homologues. This hasbeen related to improved interactions with the solvent,which could be of prime importance in the acquisition offlexibility near zero degrees [69]. Besides the balance ofcharges, the number of salt bridges covering the proteinsurface is also reduced. There is now a clear correlationbetween surface ion pairs and temperature adaptation,since these weak interactions significantly increase innumber from psychrophiles to mesophiles, thermophilesand hyperthermophiles, the latter showing arginine-mediatedmultiple ion pairs and interconnected salt bridgenetworks [70, 71]. Such an altered pattern of electrostaticinteractions at the molecular surface is thought to improvethe resilience [72] or the ‘breathing’ of the externalshell of cold-active enzymes.Of course, all these factors are not found in every cold-activeenzyme: each enzyme adopts its own strategy by usingone or a combination of these altered structural factorsto improve the local or global mobility of the proteinedifice. An example is given in table 1 which provides asummary of the structural factors possibly involved in thelow stability of a psychrophilic a-amylase. A recent comparativestructural analysis of psychrophilic, mesophilicand thermophilic enzymes reports that each protein familydisplays a different structural strategy to adapt to temperature.However, some common trends are observed:the number of ion pairs, the side chain contribution to theTable 1. Structural parameters potentially involved in the stabilityof a-amylase conformation as deduced from the X-ray structures[47, 63, 74]a-amylase Psychrophile MesophileProline content 13 21Salt bridges 18 26Arginine content 13 28Arg-mediated salt bridges 8 17Arg-mediated amino-aromatic 2 13interactionsArg-mediated H bonds 20 42Disulfide bonds 4 5Core cluster hydrophobicity (S H i ) 47 103Non-constrained residues at 6 0C terminusAromatic-aromatic interactions 19 20Amino-aromatic interactions 2 13Oxygen-aromatic interactions 24 19Sulfur-aromatic interactions 2 5a helix dipole – charges in Na 2 4+ charges in Ca 3 6Calcium-binding constant (M –1 ) 5.0 ¥ 10 7 2.0 ¥ 10 11Chloride-binding constant (M –1 ) 1.6 ¥ 10 2 3.5 ¥ 10 3Data for the psychrophilic a-amylase differ slightly from Aghajariet al. [47] as only one crystal form is considered and bond lengthlimits are different.exposed surface and the apolar fraction of the buried surfaceshow a consistent decrease with decreasing optimaltemperatures [62].Kinetic optimization of cold-active enzymesAs mentioned, improving the turnover number k cat is themain physiological adaptation, because it offsets the inhibitoryeffect of low temperatures on reaction rates andtherefore provides adequate raw metabolic activity to thegrowing organism. However, both k cat and K m are fundamentalkinetic parameters characterizing an enzymaticreaction and the specificity constant k cat /K m is generally abetter indication of catalytic efficiency than k cat alone. Inprinciple, cold-adapted enzymes can optimize the k cat /K mratio by increasing k cat , decreasing K m or by changes inboth k cat and K m . Inspection of the available data, althoughlimited [for compilations see refs 4 and 28], reveals twomain trends: cold-active enzymes that improve k cat at theexpense of K m (both k cat and K m increase) and those improvingthe k cat /K m ratio (increase of k cat and decrease ofK m ). As will be discussed in the next sections, both kineticbehaviors seem to be related to cold-active enzymes devoidof adaptive mutations within the active site and toenzymes displaying adaptive mutations within the catalyticcenter, respectively.

652 G. Feller Cold-active enzymesImproving the turnover number k cat :a thermodynamic challengeThe catalytic constant k cat corresponds to the maximumnumber of substrate molecules converted to product peractive site per unit of time. In the simple Michaelis-Menten mechanism, the k cat parameter is the first-orderrate constant for the chemical conversion of the enzymesubstratecomplex to enzyme and product. The transitionstate theory assumes the existence of a stable activatedcomplex ES # in equilibrium with the ground state ES:k catE + S ´ ES ´ ES# æÆ E + P (1)and the temperature dependence of the catalytic rate constantis given by the following relation which is formallyequivalent to the Arrhenius law:k B Tk cat = k 7 e – DG# /RT(2)hwhere k is the transmission coefficient generally close to1, k B is the Bolzmann constant (1.3805 ¥ 10 –23 J K –1 ), hthe Planck constant (6.6256 ¥ 10 –34 J s) and DG # the freeenergy of activation or the variation of the Gibbs energybetween the activated enzyme-substrate complex ES # andES. The transition state theory contains several wellknownsimplifications and interpretation of the absolutevalue of the thermodynamic parameters of activationshould be taken with caution. In the context of coldadaptation has come the proposal to compare the variationof the thermodynamic parameters [25] between psychrophilic( p ) and mesophilic ( m ) enzymes, namelyD(DG # ) p–m , D(DH # ) p–m and D(DS # ) p–m . Such parametersare compiled in table 2. It should be stressed that commonlyfound negative values provide useful insights intothe way psychrophilic enzymes have evolved.As inferred from equation 2, the D(DG # ) p–m parametersimply reflects the fact that psychrophilic enzymes aremore active than their mesophilic homologues. However,the small value of this parameter originates from largedifferences in the enthalpic and entropic contributions,D(DH # ) p–m and TD(DS # ) p–m , respectively, between coldadaptedand mesophilic enzymes. To appreciate the informativecontent of D(DH # ) p–m , the exponential term ofequation 2 can be rewritten as:k B Tk cat = 7 e – (DH# /RT – DS # /R) (3)hIn this form, one can note that either a decrease of DH # oran increase of DS # results in an increase of k cat . However,the contribution of the DH # /RT term to the decrease of k catis larger when the temperature T is lowered. Interestingly,all psychrophilic enzymes studied so far possess lowerTable 2. Activation parameters for the activity of psychrophilic and mesophilic enzymes [25].Enzyme Source T k cat DG # DH # TDS # D(DG # ) p-m D(DH # ) p-m TD(DS # ) p-m(°C) (s –1 ) (kJ/mol) (kJ/mol) (kJ/mol) (kJ/mol) (kJ/mol) (kJ/mol)Trypsin psychrophile Greenland cod 5 82.0 57.7 33.0 –24.5 –2.5 –20.9 –18.3mesophile bovine 27.7 60.2 53.9 –6.2Trypsin psychrophile Antarctic cod 15 0.97 70.5 40.6 –29.9 –0.9 –42.0 –41.2mesophile bovine 0.68 71.4 82.6 11.3Amylase psychrophile Alteromonas 15 495.2 55.6 73.6 18.0 –4.1 –26.0 –21.9mesophile Bacillus 90.5 59.7 99.6 39.9Subtilisin sychrophile Bacillus TA41 15 25.4 62.7 36 –26.7 –3.7 –10.0 –6.3mesophile B. subtillis 5.4 66.4 46 –20.4Glucose- psychrophile Antarctic fish 0 106.5 56.1 36.8 –19.3 –2.1 –15.1 –13.06-phosphate mesophile human 42.4 58.2 51.9 –6.3dehydrogenaseXylanase psychrophile Antarctic yeast 5 14.8 61.7 45.4 –16.3 –2.6 –4.5 –1.9mesophile yeast 4.9 64.3 49.9 –14.4Chitobiase psychrophile Arthrobacter 15 98.0 59.5 44.7 –14.8 –4.0 –26.8 –22.8mesophile S. marcescens 18.0 63.5 71.5 8.0Chitinase A psychrophile Arthrobacter 15 1.7 69.2 60.2 –9.0 2.0 –14.1 –16.1mesophile S. marcescens 3.9 67.2 74.3 7.1Glutamate psychrophile Antarctic fish 5 3.8 64.8 33.4 –31.4 –3.3 –25.6 –22.3dehydrogenase mesophile bovine 0.9 68.1 59.0 –9.1

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 653DH # values (table 2) and therefore reduce the temperaturedependence of k cat and the inhibitory effect of low temperatureson enzyme reaction rates. Thus, the decrease ofthe activation enthalpy in the enzymatic reaction of psychrophilicenzymes can be considered as the main adaptivecharacter to low temperatures. This decrease in activationenergy is achieved structurally by a decrease in thenumber of enthalpy-driven interactions that have to bebroken during the activation steps. These interactions alsocontribute to the stability of the folded protein conformationand, as a corollary, the structural domain of the enzymebearing the active site should be less stable. This isthe first insight suggesting that activity and stability arelinked in the process of thermal adaptation. As far as theactivation entropy is concerned, the negative value ofD(DS # ) p–m can also be interpreted assuming a more resilientactive site from psychrophilic enzymes [25]. As aconsequence of active-site flexibility, the ground state enzyme-substratecomplex ES occupies a broader distributionof conformational states translated into increased entropyof this state compared to that of the mesophilic homologues,leading to a negative value of D(DS # ) p–m .Adaptive drift and adaptive optimization of K mAs inferred from the previous section, a broader distributionof the ground state ES should be accompanied by aweaker substrate binding strength if increases in k cat arereached through active-site resilience. In other words,there is an evolutionary pressure on K m to increase in orderto maximize the overall reaction rate [73]. If the substrateis weakly bound, the ground state ES complex willfall in a less deep energy pit and, as a consequence, the activationenergy of the reaction (DG # ) will be smaller, thatis k cat will be increased (eq. 2). This is precisely what isfound in psychrophilic enzymes which display identicalsubstrate-binding site and active-site architecture (interms of side chain identity) when compared with theirmesophilic homologues. This fundamental notion wasfirst introduced in the study of fish lactate dehydrogenases(A 4 -LDHs). The active-site residues are fully conservedin A 4 -LDHs from South American and Antarcticnotothenoid fish, whereas the cold-adapted enzymes arecharacterized by higher k cat and K m constants. The conclusionwas that temperature-adaptive increases in k cat occurconcomitantly with increases in K m in cold-activeLDHs [73]. A second and even more conclusive exampleis given by the above-mentioned structures of cold-activeand porcine a-amylases in complex with a large inhibitor[63, 64]. Again, all amino acids forming the active site arestrictly conserved, although the cold-active enzyme possesseshigher k cat and K m . Worth mentioning is that this aspectseems to be valid for binding constants for cofactors,allosteric effectors and ions [37, 42, 44, 74].Several enzymes counteract this adaptive drift of K m tomaintain or to improve the substrate-binding affinity byamino acid substitutions within the active site. The firstreason for these enzymes to ‘react’ against the drift is obviouswhen considering the regulatory function associatedwith K m , especially for intracellular enzymes. Thesecond reason is related to the temperature dependence ofweak interactions. Substrate binding is an especially temperature-sensitivestep because both the binding geometryand interactions between binding site and ligand aregoverned by weak interactions having sometimes oppositetemperature dependencies. Hydrophobic interactionsform endothermically and are weakened by a decrease intemperature. In contrast, interactions of an electrostaticnature (ion pairs, hydrogen bonds, Van der Waals interactions)form exothermically and are stabilized at low temperatures.Therefore, low temperatures not only inhibitthe enzyme activity (k cat ) but can also severely alter thesubstrate-binding mode and strength according to thetype of interaction involved.The chitobiase from Antarctic bacteria nicely illustratesboth aspects as well as the extent of the kinetic optimizationthat can be reached during cold adaptation of enzymes[75]. As shown in figure 2, the k cat of the cold-activechitobiase is 8 times higher than that of a mesophilicchitobiase at 5°C. However, the K m for the substrate is 25times lower at this temperature and as a result, the k cat /K mfor the cold-active enzyme is nearly 200 times greater atlow temperature. In addition, the cross-shaped plot of K mshows that the K m of each enzyme tends to minimal andoptimal values in the range of the corresponding environmentaltemperatures, reflecting the fine tuning of this parameterreached in the course of thermal adaptation. Inthe case of the mesophilic chitobiase, the 3D structure indicatesthat two tryptophan residues are the main substrate-bindingligands and perform hydrophobic interactionswith the substrate. This can be related to thedecrease of K m with temperature, according to the abovementionedthermal dependence of hydrophobic interactions.Interestingly, the two tryptophan residues are notfound in the cold-active chitobiase but are replaced by polarresidues that are able to perform stronger interactionsas the temperature decreases. Such replacements may explainthe cross-shaped plot, revealing a subtle adaptivestrategy that takes advantage of the thermodynamic propertiesof the bonds involved in substrate binding at lowtemperatures.A working hypothesisThe previous sections give the impression that two maintypes of adaptation occur in psychrophilic enzymes. Onthe one hand, enzymes that possess strictly identicalresidues in the active site as their mesophilic homo-

654 G. Feller Cold-active enzymesFigure 3. Correlation of the kinetic parameters in a psychrophilicenzyme (AHA), its stabilized mutants (closed circles) and amesophilic homologue (PPA). The general trend displayed by thestabilized mutants of AHA is to decrease both k cat and K m values[65].can counteract the adaptive drift of K m to meet other physiologicalrequirements.Mutational analysis of cold-activity determinantsFigure 2. Temperature dependence of the kinetic parameters forpsychrophilic () and mesophilic () chitobiases (from Arthrobactersp. and Serratia marcescens, respectively). Data for the catalyticrate constant k cat (a), the Michaelis parameter K m (b), note the differentscales used and the relative catalytic efficiency k cat /K m (psychrophile/mesophile)(c). The cold-adapted chitobiase is characterizedby a higher activity, an optimal K m value at low temperaturesand a 200-fold higher catalytic efficiency at 7°C [75].logues, possibly because the constraints imposed by thestructure and the enzyme function do not allow any adjustmentsat this level. On the other hand, some enzymespossess amino acid replacements in the active site thatcan contribute to cold activity. According to the abovementionedenergetic aspects of enzyme-catalyzed reactionsat low temperature, a working hypothesis can be formulated.The high catalytic activity of psychrophilic enzymescorresponds to a decrease of DG # of the catalyzedreaction (eq. 2). For all cold-active enzymes, this decreaseof DG # is achieved by a decrease of DH # (table 2)that renders the catalyzed reaction less temperature sensitive,but requires a more flexible and therefore heat-labileactive site. As a result of this active site resilience, the entropiccontribution becomes larger and substrate bindingis weakened for enzymes devoid of adaptive mutations inthe active site. This weak binding also contributes to thedecrease in DG # as the ES complex falls in a less deep energypit. However, specific mutations in the active siteAn experimental demonstration of the relationships betweenk cat and K m was recently provided using psychrophilicand pig pancreatic a-amylases as models [65].Stabilizing weak interactions found in the porcine a-amylase, but absent in the cold-active a-amylase, wereengineered by site-directed mutagenesis in the psychrophilicenzyme at positions far from the catalytic center.As shown in figure 3, stabilizing and rigidifying thecold-active a-amylase tends to decrease the k cat valuesand to decrease concomitantly the K m values of the mutantenzymes, revealing the high correlation betweenthese two kinetic parameters. This is also another strongindication that structural flexibility is an essential featurerelated to catalysis at low temperatures in psychrophilicenzymes. One should note that other mutational studiesof cold-active enzymes have been less conclusive. Citratesynthase [76], subtilisin [44], alkaline phosphatase [77,78], ribonuclease [79] and chitinase (unpublished results)have been engineered to check the involvement of specificadaptive mutations within or close to the active site.These studies have revealed a complex pattern of effectson kinetic parameters, activation energy and stability thatcannot be interpreted in simple terms. Double mutantshave also demonstrated an unsuspected synergy betweenindividual mutations sometimes giving rise to non-additiveor opposite effects. These interesting studies mainlyhighlight our current inability to predict the effect of singleside chain replacement based on rational design of the

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 655active site, not only in psychrophilic catalysts but in enzymesin general. However, a single mutation in the calcium-bindingsite of subtilisin [44] and in the phosphatebindinghelix of triose phosphate isomerase [51] wereshown to increase drastically the thermostability of thepsychrophilic mutants. These results underline the stronginvolvement of bound cofactors in the modulation of stabilityand activity.Table 3. Thermodynamic parameters for the irreversible inactivationat 52°C of Antarctic fish and bovine L-glutamate dehydrogenases[80].Enzyme k i E a DG # DH # TDS #(min –1) (kJ/mol) (kJ/mol) (kJ/mol) (kJ/mol)C. aceratus 0.230 659 94.9 657 562Ox 0.043 493 99.4 490 391Heat-labile activity and weak structural stabilityThe fast inactivation of cold-active enzymes by temperaturehas been recognized for decades. As a rule, the halflifeof activity for most psychrophilic enzymes at50–60°C is about 10 min, and a few hours at 37°C. However,some cold-active enzymes already denature at20–25°C. As it counteracts the increase of activity withtemperature, this marked heat lability is responsible forthe shift of the apparent optimal temperature of activityshown in figure 1. The irreversible denaturation betweenthe native state N and the denatured inactive state I can berepresented by a simple model characterized by a first-orderrate constant k i .k iN æÆ I (4)Surprisingly, systematic analyses of this well-recognizedproperty are scarce and generally fragmental. Such ananalysis has been performed [80] for L-glutamate dehydrogenase(GDH) from the Antarctic fish Chaenocephalusaceratus (the white-blooded icefish) and for thebovine enzyme (table 3). As anticipated, the activation energyDG # is lower for the fish enzyme, because it undergoesa faster denaturation. Because the cold-active GDHis inactivated in a narrow temperature range, the slope ofthe Arrhenius plot is steep, leading to very high values ofboth E a and DH # and, by difference, to a high value ofTDS # . Therefore, the activity loss is entropically driven(the low DG # value arises from the unfavorable entropycontribution). At a given temperature, the activated stateof the psychrophilic enzyme is reached through a largerentropy variation, probably reflecting an increased localdisorder of the active site, which is the main driving forceof heat inactivation. On the other hand, the larger enthalpyvariation accompanying this process possibly reflectsthe higher cooperativity of heat inactivation in thepsychrophilic enzyme, originating from the lower numberof interactions required to disrupt the active conformation.The origin of the heat lability is obviously found in thelow stability of the structural elements forming the activesite (localized flexibility) or in the weak conformationalstability of cold-active enzymes (global flexibility).Global and localized stability of enzymes may haveevolved independently [81]. Extrinsic stabilization bysolutes of lactate dehydrogenase from warm- and coldadaptedfish indicates that the stability of the enzymemolecules is affected to the same extent whereas the kineticparameters of cold-active lactate dehydrogenases ismore affected by cosolvents. This effect has been attributedto a flexible active site in cold-adapted enzymes thatcan be more compacted by preferential exclusion ofsolutes. Direct evidence of differences in stability betweenthe active site and the whole structure has beenprovided by microcalorimetric studies of multidomainproteins, as shown in figure 4. The cold-adapted chitobiasefrom the psychrophile Arthrobacter TAD 20 has amodular architecture consisting of an active-site domain,a galactose-binding domain and an immunoglobulin-likedomain, the latter two probably being involved in thebinding onto the chitinous exoskeleton of crustaceans.Differential scanning calorimetry (DSC) thermograms ofthis cold-active chitobiase display both a heat-labile anda heat-stable domain corresponding to the catalytic domainand to the binding domains, respectively, as demonstratedby saccharide-binding experiments [75]. The precursorof the psychrophilic a-amylase consists of the nativeenzyme and an extra domain involved in the bacterialsecretion [82]. The unfolding profiles of the precursor(fig. 4b) also display a heat-labile and a heat-stable domaincorresponding to the native enzyme and to the secretiondomain, respectively. These unfolding patternsclearly demonstrate that heat lability only affects thecalorimetric domain containing the active site while leavingstability of the non-catalytic domains less affected orunchanged. Thus heat lability is not simply the result ofthe lack of selective pressure for stable protein in a coldenvironment (as demonstrated by the stable non-catalyticdomains) but seems to be the consequence of the improvedplasticity required around the active site for efficientactivity at low temperatures. Another stability patterndetected in a cold-active enzyme is also worth mentioning.The psychrophilic PGK from PseudomonasTACII 18 is a globular enzyme composed of two domainsconnected by a hinge region that bends during catalysisforming the active site at the interface of the two domains.Surprisingly, one domain is heat labile whereas the seconddomain has the same stability as mesophilic PGKs(fig. 4c). The heat-labile domain may provide the re-

656 G. Feller Cold-active enzymesFigure 5. Thermal unfolding of a-amylases recorded by DSC.When compared with the psychrophilic AHA (from P. haloplanktis),the heat-stable enzymes PPA (from pig pancreas) and BAA(from Bacillus amyloliquefaciens) are characterized by higher T m(top of the transition) and DH cal (area under the transition) values,by a flattening of the transition and by the occurrence of calorimetricdomains indicated by deconvolutions in thin lines [65].Figure 4. Three examples of psychrophilic enzymes exhibiting localflexibility. In chitobiase (a) and the a-amylase precursor (b), theactive site is embedded in the heat-labile domain, whereas the stabilityof the non-catalytic domains is unaffected. In phosphoglyceratekinase (c), the active site is formed at the interface of both heatlabileand heat-stable domains by a hinge-bending mechanism[48, 75].quired flexibility around the active site and favor the reactionrate, whereas the heat-stable domain could improvesubstrate binding (as indicated by low K m values) asa result of its rigidity [48].Conformational stability of psychrophilic enzymesAs a result of their weak stability, psychrophilic enzymesare valuable model compounds for folding studies and areconsequently increasingly investigated by more sophisticatedtechniques. Among these, DSC offers a definitiveadvantage because several thermodynamic parametersrelated to protein stability are recorded directly and notextrapolated from indirect data. Figure 5 displays thecalorimetric records, or thermograms, of heat-inducedunfolding for psychrophilic, mesophilic and heat-stablea-amylases. Several interesting observations can alreadybe made from these records.1) The unfolding of the cold-adapted enzyme occurs atlower temperatures as indicated by the T m values whichcorrespond to the temperature of half-denaturation(for a two-state process) and are given by the top of thetransition.2) Protein unfolding is an endothermic process. The areaunder the curves (fig. 5) provides the total amount ofheat absorbed during unfolding and is given by thecalorimetric enthalpy, DH cal . This parameter correspondsto all enthalpic contributions involved in maintainingthe compact structure and is markedly lowerfor the psychrophilic enzyme. In addition, there is aclear trend for increasing DH cal values in the order psychrophile< mesophile < thermophile.3) The transition for the psychrophilic enzyme is sharpand symmetric whereas other enzymes are characterizedby a flattening of the thermograms. This is indicativeof a pronounced cooperativity during unfoldingof the psychrophilic enzyme: its structure is stabilizedby fewer weak interactions and disruption ofsome of these interactions strongly influences the entiremolecular edifice and promotes its unfolding.4) The psychrophilic enzyme unfolds according to an allor-noneprocess, revealing a uniformly low stability ofits architecture. By contrast, all other homologous enzymesdisplay two to three transitions (either observableor indicated by deconvolution of the heat capacityfunction in fig. 5). Therefore, the conformation ofthese mesophilic and thermophilic enzymes containsstructural blocks or units of distinct stability that unfoldindependently.

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 6575) Although not illustrated by figure 5, the unfolding ofthe psychrophilic enzyme is freely reversible whereasall other homologous enzymes are irreversibly unfoldedafter a DSC experiment. The weak hydrophobicityof the core clusters in the cold-adapted enzymeand the low melting temperature, at which hydrophobicinteractions are restrained, certainly account forthis reversible character because, unlike mesophilic a-amylases, aggregation does not occur.From this microcalorimetric study, one can conclude thatthe psychrophilic enzyme investigated possesses a fragilemolecular edifice that is uniformly unstable and stabilizedby fewer weak interactions than homologous mesophilic orthermophilic proteins [65, 69]. The contribution of individualweak interaction to the above-mentioned behaviorwas analyzed by site-directed mutagenesis of the psychrophilica-amylase [65]. Fourteen mutants of this enzymewere constructed, each bearing an engineeredresidue forming a weak interaction found in mesophilic a-amylases but absent in the cold-active a-amylase. Singleamino acid side chain substitutions were found to significantlymodify the melting point T m , the calorimetric enthalpyDH cal , the cooperativity and reversibility of unfoldingbut also the thermal inactivation rate constant k i and thekinetic parameters k cat and K m . Thus, a fascinating aspect ofthis study was the ability to dissect the contribution of individualamino acid side chains to the structural propertiesof heat-labile and heat-stable proteins.Thermodynamic stability of psychrophilic enzymesThe reversible unfolding pathway of some proteins betweenthe native state N and the unfolded state U is characterizedby an equilibrium constant K.KN ¨Æ U (4)The free energy of conformation or free energy of unfoldingis therefore:DG N–U = – RT lnK= DH N–U – TDS N–U(6)The latter relation can be rewritten for any temperature(T) using the parameters determined experimentally byDSC:DG N–U (T) = DH cal (1–T/T m ) + DCp(T–T m )– TDCp ln (T/T m ) (7)where DCp is the difference in heat capacity between thenative and the unfolded state. This parameter reflects thehydration of non-polar groups that are exposed to waterupon unfolding and is determined by another set ofcalorimetric experiments [83].Figure 6. Gibbs free energy of unfolding or conformational stability.Stability curves for a psychrophilic a-amylase (heavy line), severalmesophilic proteins (continuous lines) and a thermophilic protein(dashed line). To compare proteins of various sizes, the free energyis expressed in specific units per mole of residue [69].The stability curves, i.e. the Gibbs free energy of unfoldingas a function of temperature, for psychrophilic,mesophilic and thermophilic proteins are drawn in figure6. These curves correspond to the energy required to disruptthe protein structure at any given temperature. This isa parabolic function limited by the temperature of cold denaturationT m’ (below 0°C, not shown) and by the high temperaturemelting point T m , at which by definition DG N–U =0 [84]. Again, these curves provide valuable insights intothe molecular adaptations of psychrophilic enzymes.1) In principle, the low stability of a protein can bereached via three ways (or their combination): by shiftingthe stability curve toward low temperatures (with asimilar maximal value of DG N–U as mesophiles but atmuch lower temperatures), by increasing the steepnessof the stability curve (with a similar maximal value ofDG N–U as mesophiles, at the same temperature but witha lower T m ) or by a global collapse of the stabilitycurve. Each of these alternatives corresponds to acombination of alterations in the DH cal , T m or DCp valuesin equation 7. Obviously, the stability curve of thecold-active a-amylase follows the third alternative: thestrikingly weak conformational energy (DG N–U max =22 cal/mol of residue or 92 J/mol of residue for 453amino acids) over the entire temperature range wherethe native state prevails, i.e. between both meltingpoints, arises from low values of both DH cal and T m ,whereas DCp is not significantly affected.2) The temperature for maximal protein stability does notusually correspond to the environmental temperatureof psychrophiles, mesophiles or thermophiles but is,rather, close to room temperature [7]. The temperatureexperienced by mesophiles and thermophiles lies onthe right side of the bell-shaped stability curve, pro-

658 G. Feller Cold-active enzymesviding evidence that their enzymes reach the necessarybalance between optimal stability (leading to a compactmolecule) and flexibility (required for the catalyticfunction) by using the thermal dissipative force,responsible for unfolding at higher temperatures. Incontrast, the environmental temperatures, around 0°C,that are encountered by psychrophiles are found far onthe left side of the stability curve. Mention needs to bemade at this point that the decrease in stability withinthis region of the curve shown in figure 6, and ultimatelyleading to cold denaturation, arises from thehydration of protein groups which destabilizes the proteinat decreasing temperature [85]. It follows that theorigin of flexibility of psychrophilic enzymes at lowtemperatures involves mainly group hydration, and istherefore drastically different from mesophilic andthermophilic proteins, the latter taking advantage ofthe conformational entropy rise with temperature togain mobility. The improved interactions with the solventsuggested by the crystal structure of some coldactiveenzymes reinforce the idea that group hydrationplays an essential role in the acquisition of flexibilityin psychrophilic proteins.3) A surprising consequence of the free energy functionfor the psychrophilic a-amylase shown in figure 6 isits weak stability at low temperatures when comparedwith mesophilic and thermophilic proteins, whereasintuitively, cold-active proteins were also expected tobe cold stable. This protein is in fact both heat and coldlabile. Assuming constant properties of the solvent below0°C (i.e. no freezing) and the absence of protectiveeffects from cellular components, this a-amylaseshould unfold at –10°C. Incidentally, this temperatureclosely corresponds to the probable lower limit of bacterialgrowth and, therefore, cold denaturation of somekey enzymes in psychrophiles is an additional, thoughunsuspected factor fixing the lower limit of life at lowtemperatures.4) The psychrophilic a-amylase has reached a state closeto the lowest possible stability of the native state [65,69]. This fact indicates that the cold-active a-amylasecould not be any less stable than it is. If psychrophilicenzymes have indeed gained in flexibility at the expenseof stability in the course of evolution, this impliesthat the actual native state precludes further adaptationtoward a more mobile structure. This aspect accountsfor the imperfect adaptation of the catalyticfunction in psychrophilic enzymes, mentioned at thebeginning of this review and illustrated in figure 1.Natural versus directed evolutionDirected evolution has contributed significantly to explorethe role of multiple and dispersed substitutions thatact synergistically to modify enzyme properties and functions[86, 87]. Basically, random mutations are introducedin a gene and the resulting library is screened forthe acquisition or improvement of a specific property.This approach reveals mutations responsible for immediateand drastic adaptation in response to a strong selectionimposed in the laboratory and selected by a physicochemicalproperty. In contrast, natural evolution is a veryslow process driven by complex environmental and metabolicconstraints, and selected by the viability or performanceof a living organism. What can we learn from theresults of laboratory evolution and from the general traitsof psychrophilic enzymes?Laboratory evolution has demonstrated that enzyme activityand stability are not physically linked in proteins aspreviously thought, giving rise to the proposal that thelow stability of cold-active enzymes results from geneticdrift originating from the lack of selective pressure forstable proteins [8]. However, mutations that simultaneouslyincrease activity and stability are very rare [88] andproved to be extremely complex to obtain [89]. In fact, innature, these environmental constraints do not exist exceptfor thermophilic enzymes that catalyze the conversionof labile metabolic intermediates. As a matter of fact,phosphoribosyl anthranilate isomerase from Thermotogamaritima was found to be both extremely stable and active[90]. When random mutants are only screened forhigh activity at low temperature during directed evolution,the selected enzymes invariably display the canonicalproperties of psychrophilic enzymes [91–96]. Thiscan be understood from the above-mentioned results, asimprovement of activity at low temperatures associatedwith lost of stability appears to be the most frequent andaccessible strategy at the level of the protein molecule.An elegant demonstration of this aspect was given usingornithine carbamoyltransferase from Pyrococcus furiosusbecause random mutagenesis was associated with a functionalscreening procedure involving the survival of thehost strain [94]. Consequently, the low stability of coldactiveenzymes may be the simplest way to acquire activityin the absence of selection for stable proteins. Thus,the occurrence of heat-stable psychrophilic enzymesshould not be ruled out.ConclusionsThe most widely accepted hypothesis accounting for thedominant adaptive traits of cold-active enzymes, i.e. thehigh activity and the weak stability, suggests that there is acorrelation between the activity, the flexibility and the stabilityof the enzyme molecule. The flexible structure ofpsychrophilic enzymes can provide enhanced ability to undergodiscrete and fast conformational changes at low temperaturesimposed by the catalytic events. But the price to

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 659pay for such plasticity is, of course, the low stability of thenative enzyme structure [1]. This link between activity,flexibility and stability, although intuitively satisfying, remainsto be proven experimentally in cold-active enzymes.Nonetheless, these relationships have found strong experimentalsupport from the comparison of thermophilic andmesophilic enzymes [97–99]. In contrasts, very few attemptshave been made to quantify and compare the expectedflexibility of cold-adapted proteins. The rate of hydrogen/deuteriumexchange was recorded by Fourier transforminfrared spectroscopy for two cold-active enzymes[32, 100] but no significant exchange rate was foundbetween psychrophilic enzymes compared with themesophilic homologues. The nature of the protein motioninvolved in temperature adaptation is complex and remainslargely hypothetical [101]. For example, amide hydrogenexchange experiments using rubredoxin from P. furiosus,the most thermostable protein presently known, indicatethat conformational fluctuations for solvent access do notdiffer from those of mesophilic homologues at the millisecondtime scale [102], therefore questioning the involvementof conformational rigidity in the extreme heatstability of this protein. Incidentally, this unexpected resultalso raises some doubts about the connection between exchangerates and conformational flexibility. Directed evolutionsuggests that low stability results from genetic driftbut, in contrast, the biophysical studies mentioned in thisreview indicate that flexibility is an integral component ofthe adaptive strategy. At this stage, the emerging picturesuggests that psychrophilic enzymes utilize an improvedflexibility of the structures involved in the catalytic cycle(as indicated by the common heat-labile activity), whereasother protein regions, if not implicated in catalysis, may ormay not be subjected to genetic drift, as nicely illustratedby the cold-active chitobiase and the a-amylase precursor(fig. 4). A better understanding of enzyme evolution andprotein dynamics is obviously the next challenge for biochemicalstudies of extremophiles.Acknowledgements. The author acknowledges with thanks thefinancial support of the European Union, the Région wallonne,Belgium and the Fonds National de la Recherche Scientifique,Belgium. The support of the Institut Français de Recherche et deTechnologie Polaire has also been highly appreciated.1 Feller G. and Gerday C. (1997) Psychrophilic enzymes: molecularbasis of cold adaptation. Cell. Mol. Life Sci. 53: 830–8412 Gerday C., Aittaleb M., Bentahier M., Chessa J. P., Claverie P.,Collins T. et al. (2000) Cold-adapted enzymes: from fundamentalsto biotechnology. Trends Biotechnol. 18: 103–1073 Margesin R. and Schinner F. (1999) Biotechnological Applicationsof Cold-Adapted Organisms, Springer, Heidelberg4 Cavicchioli R., Siddiqui K. S., Andrews D. and Sowers K. R.(2002) Low-temperature extremophiles and their applications.Curr. Opin. Biotechnol. 13: 253–2615 Russell N. J. (1998) Molecular adaptations in psychrophilicbacteria: potential for biotechnological applications. Adv.Biochem. Eng. Biotechnol. 61: 1–216 Allen D., Huston A. L., Weels L. E. and Deming J. W. (2001)Biotechnological use of psychrophiles. In: Encyclopedia ofEnvironmental Microbiology, pp. 1–17, Bitton G. (ed.),Wiley, New York7 Kumar S., Tsai C. J. and Nussinov R. (2002) Maximal stabilitiesof reversible two-state proteins. Biochemistry 41:5359–53748 Wintrode P. L. and Arnold F. H. (2000) Temperature adaptationof enzymes: lessons from laboratory evolution. Adv. ProteinChem. 55: 161–2259 Levy M. and Miller S. L. (1998) The stability of the RNAbases: implications for the origin of life. Proc. Natl. Acad. Sci.USA 95: 7933–793810 Price P. B. (2000) A habitat for psychrophiles in deep Antarcticice. Proc. Natl. Acad. Sci. USA 97: 1247–125111 Deming J. W. (2002) Psychrophiles and polar regions. Curr.Opin. Microbiol. 5: 301–30912 Morita R. Y. (1975) Psychrophilic bacteria. Bacteriol. Rev. 39:144–16713 Sabri A., Bare G., Jacques P., Jabrane A., Ongena M., VanHeugen J. C. et al. (2001) Influence of moderate temperatureson myristoyl-CoA metabolism and acyl-CoA thioesterase activityin the psychrophilic Antarctic yeast Rhodotorula aurantiaca.J. Biol. Chem. 276: 12691–1269614 Petrescu I., Lamotte-Brasseur J., Chessa J. P., Ntarima P.,Claeyssens M., Devreese B. et al. (2000) Xylanase from thepsychrophilic yeast Cryptococcus adeliae. Extremophiles 4:137–14415 Gomes J., Gomes I. and Steiner W. (2000) Thermolabile xylanaseof the Antarctic yeast Cryptococcus adeliae: productionand properties. Extremophiles 4: 227–23516 Deegenaars M. L. and Watson K. (1998) Heat shock responsein psychrophilic and psychrotrophic yeast from Antarctica.Extremophiles 2: 41–4917 Devos N., Ingouff M., Loppes R. and Matagne R. F. (1998)Rubisco adaptation to low temperatures: a comparative studyin psychrophilic and mesophilic unicellular algae. J. Phycol.34: 655–66018 Loppes R., Devos N., Willem S., Bartelemy P. and MatagneR. F. (1996) Effect of temperature on two enzymes from apsychrophilic Chloromonas (chlorophyta). J. Phycol. 32:276–27819 Willem S., Srahna M., Devos N., Gerday C., Loppes R. andMatagne R. F. (1999) Protein adaptation to low temperatures:a comparative study of alpha-tubulin sequences in mesophilicand psychrophilic algae. Extremophiles 3: 221–22620 Prisco G. di, Pisano E. and Clarke A. (1998) Fishes of Antarctica:A Biological Overview, Springer, Milan21 Eastman J. T. (1993) Antarctic Fish Biology: Evolution in aUnique Environment, Academic Press, San Diego22 D’Amico S., Claverie P., Collins T., Georlette D., Gratia E.,Hoyoux A. et al. (2002) Molecular basis of cold adaptation.Phil. Trans. R. Soc. Lond. B 357: 917–92523 D’Amico S., Claverie P., Collins T., Feller G., Georlette D.,Gratia E. et al. (2001) Cold-adapted enzymes: an unachievedsymphony. In: Protein Adaptations and Signal Transduction,pp. 31–42, Storey K. B. and Storey J. M. (eds), Elsevier, Amsterdam24 Georlette D., Bentahir M., Claverie P., Collins T., D’Amico S.,Delille D. et al. (2001) Cold-adapted enzymes. In: Physics andChemistry Basis of Biotechnology, pp. 177–196, De CuyperM. and Bulte J. W. M. (eds), Kluwer, Dordrecht25 Lonhienne T., Gerday C. and Feller G. (2000) Psychrophilicenzymes: revisiting the thermodynamic parameters of activationmay explain local flexibility. Biochim. Biophys. Acta.1543: 1–1026 Zecchinon L., Claverie P., Collins T., D’Amico S., Delille D.,Feller G. et al. (2001) Did psychrophilic enzymes really winthe challenge? Extremophiles 5: 313–321

660 G. Feller Cold-active enzymes27 Somero G. N. (1995) Proteins and temperature. Annu. Rev.Physiol. 57: 43–6828 Smalas A. O., Leiros H. K., Os V. and Willassen N. P. (2000)Cold adapted enzymes. Biotechnol. Annu. Rev. 6: 1–5729 Sheridan P. P., Panasik N., Coombs J. M. and Brenchley J. E.(2000) Approaches for deciphering the structural basis of lowtemperature enzyme activity. Biochim. Biophys. Acta. 1543:417–43330 Russell N. J. (2000) Toward a molecular understanding of coldactivity of enzymes from psychrophiles. Extremophiles 4:83–9031 Margesin R., Feller G., Gerday C. and Russell N. J. (2002)Cold-adapted microorganisms: adaptation strategies andbiotechnological potential. In: Encyclopedia of EnvironmentalMicrobiology, vol. 2, pp. 871–885, Bitton G. (ed.), Wiley,New York32 Fields P. A. (2001) Protein function at thermal extremes: balancingstability and flexibility. Comp. Biochem. Physiol. AMol. Integr. Physiol. 129: 417–43133 Margesin R. and Schinner F. (1999) Cold-Adapted Organisms:Ecology, Physiology, Enzymology and Molecular Biology,Springer, Heidelberg34 Feller G., Thiry M., Arpigny J. L. and Gerday C. (1991)Cloning and expression in Escherichia coli of three lipaseencodinggenes from the psychrotrophic Antarctic strainMoraxella TA144. Gene 102: 111–11535 Feller G., Thiry M. and Gerday C. (1991) Nucleotide sequenceof the lipase gene lip2 from the Antarctic psychrotrophMoraxella TA144 and site-specific mutagenesis of the conservedserine and histidine residues. DNA Cell. Biol. 10:381–38836 Narinx E., Davail S., Feller G. and Gerday C. (1992) Nucleotideand derived amino acid sequence of the subtilisinfrom the Antarctic psychrotroph Bacillus TA39. Biochim.Biophys. Acta 1131: 111–11337 Davail S., Feller G., Narinx E. and Gerday C. (1994)Cold adaptation of proteins: purification, characterization,and sequence of the heat-labile subtilisin from the Antarcticpsychrophile Bacillus TA41. J. Biol. Chem. 269: 17448–1745338 Rentier-Delrue F., Mande S. C., Moyens S., Terpstra P., MainfroidV., Goraj K. et al. (1993) Cloning and overexpression ofthe triosephosphate isomerase genes from psychrophilic andthermophilic bacteria: structural comparison of the predictedprotein sequences. J. Mol. Biol. 229: 85–9339 Siddiqui K. S., Cavicchioli R. and Thomas T. (2002) Thermodynamicactivation properties of elongation factor 2 (EF-2)proteins from psychrotolerant and thermophilic Archaea. Extremophiles6: 143–15040 Thomas T. and Cavicchioli R. (2000) Effect of temperature onstability and activity of elongation factor 2 proteins fromAntarctic and thermophilic methanogens. J. Bacteriol. 182:1328–133241 Thomas T., Kumar N. and Cavicchioli R. (2001) Effects of ribosomesand intracellular solutes on activities and stabilitiesof elongation factor 2 proteins from psychrotolerant and thermophilicmethanogens. J. Bacteriol. 183: 1974–198242 Masullo M., Arcari P., Paola B. de, Parmeggiani A. and BocchiniV. (2000) Psychrophilic elongation factor Tu from theAntarctic Moraxella sp. Tac II 25: biochemical characterizationand cloning of the encoding gene. Biochemistry 39:15531–1553943 Georlette D., Jonsson Z. O., Van Petegem F., Chessa J., VanBeeumen J., Hubscher U. et al. (2000) A DNA ligase from thepsychrophile Pseudoalteromonas haloplanktis gives insightsinto the adaptation of proteins to low temperatures. Eur. J.Biochem. 267: 3502–351244 Narinx E., Baise E. and Gerday C. (1997) Subtilisin from psychrophilicAntarctic bacteria: characterization and site-directedmutagenesis of residues possibly involved in the adaptationto cold. Protein. Eng. 10: 1271–127945 Wintrode P. L., Miyazaki K. and Arnold F. H. (2000) Coldadaptation of a mesophilic subtilisin-like protease by laboratoryevolution. J. Biol. Chem. 275: 31635–3164046 Feller G., Le Bussy O. and Gerday C. (1998) Expression ofpsychrophilic genes in mesophilic hosts: assessment of thefolding state of a recombinant alpha-amylase. Appl. Environ.Microbiol. 64: 1163–116547 Aghajari N., Feller G., Gerday C. and Haser R. (1998) Structuresof the psychrophilic Alteromonas haloplanctis a-amylasegive insights into cold adaptation at a molecular level.Structure 6: 1503–151648 Bentahir M., Feller G., Aittaleb M., Lamotte-Brasseur J.,Himri T., Chessa J. P. et al. (2000) Structural, kinetic, andcalorimetric characterization of the cold-active phosphoglyceratekinase from the Antarctic Pseudomonas sp. TACII18. J.Biol. Chem. 275: 11147–1115349 Tutino M. L., Duilio A., Parrilli R., Remaut E., Sannia G. andMarino G. (2001) A novel replication element from an Antarcticplasmid as a tool for the expression of proteins at low temperature.Extremophiles 5: 257–26450 Aghajari N., Feller G., Gerday C. and Haser R. (1998) Crystalstructures of the psychrophilic a-amylase from Alteromonashaloplanctis in its native form and complexed with an inhibitor.Protein Sci. 7: 564–57251 Alvarez M., Zeelen J. P., Mainfroid V., Rentier-Delrue F., MartialJ. A., Wyns L. et al. (1998) Triose-phosphate isomerase(TIM) of the psychrophilic bacterium Vibrio marinus: kineticand structural properties. J. Biol. Chem. 273: 2199–220652 Kim S. Y., Hwang K. Y., Kim S. H., Sung H. C., Han Y. S. andCho Y. J. (1999) Structural basis for cold adaptation: sequence,biochemical properties, and crystal structure ofmalate dehydrogenase from a psychrophile Aquaspirilliumarcticum. J. Biol. Chem. 274: 11761–1176753 Russell R. J., Gerike U., Danson M. J., Hough D. W. and TaylorG. L. (1998) Structural adaptations of the cold-activecitrate synthase from an Antarctic bacterium. Structure 6:351–36154 Smalas A. O., Heimstad E. S., Hordvik A., Willassen N. P. andMale R. (1994) Cold adaptation of enzymes: structural comparisonbetween salmon and bovine trypsins. Proteins 20:149–16655 Berglund G. I., Willassen N. P., Hordvik A. and Smalas A. O.(1995) Structure of native pancreatic elastase from north Atlanticsalmon at 1.61 Å resolution. Acta Cryst. D 51: 925–93756 Karlsen S., Hough E. and Olsen R. (1998) Structure and proposedamino-acid sequence of a pepsin from Atlantic cod(Gadus morhua). Acta Cryst. D 54: 32–4657 Villeret V., Chessa J. P., Gerday C. and Van Beeumen J. (1997)Preliminary crystal structure determination of the alkalineprotease from the Antarctic psychrophile Pseudomonasaeruginosa. Protein Sci. 6: 2462–246458 Mandelman D., Bentahir M., Feller G., Gerday C. and HaserR. (2001) Crystallization and preliminary X-ray analysis of abacterial psychrophilic enzyme, phosphoglycerate kinase.Acta Cryst. D 57: 1666–166859 Leiros I., Lanes O., Sundheim O., Helland R., Smalas A. O.and Willassen N. P. (2001) Crystallization and preliminary X-ray diffraction analysis of a cold-adapted uracil-DNA glycosylasefrom Atlantic cod (Gadus morhua). Acta Cryst. D 57:1706–170860 Schroder Leiros H. K., Willassen N. P. and Smalas A. O.(2000) Structural comparison of psychrophilic and mesophilictrypsins: elucidating the molecular basis of coldadaptation.Eur. J. Biochem. 267: 1039–104961 Leiros H. K., Willassen N. P. and Smalas A. O. (1999) Residuedeterminants and sequence analysis of cold-adapted trypsins.Extremophiles 3: 205–219

CMLS, Cell. Mol. Life Sci. Vol. 60, 2003 Review Article 66162 Gianese G., Bossa F. and Pascarella S. (2002) Comparativestructural analysis of psychrophilic and meso- and thermophilicenzymes. Proteins 47: 236–24963 Qian M., Haser R., Buisson G., Duee E. and Payan F. (1994)The active center of a mammalian alpha-amylase: structure ofthe complex of a pancreatic alpha-amylase with a carbohydrateinhibitor refined to 2.2 Å resolution. Biochemistry 33:6284–629464 Aghajari N., Roth M. and Haser R. (2002) Crystallographicevidence of a transglycosylation reaction: ternary complexesof a psychrophilic alpha-amylase. Biochemistry 41: 4273–428065 D’Amico S., Gerday C. and Feller G. (2001) Structural determinantsof cold adaptation and stability in a large protein. J.Biol. Chem. 276: 25791–2579666 Tsigos I., Velonia K., Smonou I. and Bouriotis V. (1998) Purificationand characterization of an alcohol dehydrogenasefrom the Antarctic psychrophile Moraxella sp. TAE123. Eur.J. Biochem. 254: 356–36267 Gorfe A. A., Brandsdal B. O., Leiros H. K., Helland R. andSmalas A. O. (2000) Electrostatics of mesophilic and psychrophilictrypsin isoenzymes: qualitative evaluation of electrostaticdifferences at the substrate binding site. Proteins 40:207–21768 Brandsdal B. O., Smalas A. O. and Aqvist J. (2001) Electrostaticeffects play a central role in cold adaptation of trypsin.FEBS Lett. 499: 171–17569 Feller G., d’Amico D. and Gerday C. (1999) Thermodynamicstability of a cold-active a-amylase from the Antarctic bacteriumAlteromonas haloplanctis. Biochemistry 38: 4613– 461970 Yip K. S., Stillman T. J., Britton K. L., Artymiuk P. J., Baker P.J., Sedelnikova S. E. et al. (1995) The structure of Pyrococcusfuriosus glutamate dehydrogenase reveals a key role for ionpairnetworks in maintaining enzyme stability at extreme temperatures.Structure 3: 1147–115871 Vetriani C., Maeder D. L., Tolliday N., Yip K. S., Stillman T.J., Britton K. L. et al. (1998) Protein thermostability above100 degreesC: a key role for ionic interactions. Proc. Natl.Acad. Sci. USA 95: 12300–1230572 Aguilar C. F., Sanderson I., Moracci M., Ciaramella M., NucciR., Rossi M. et al. (1997) Crystal structure of the beta-glycosidasefrom the hyperthermophilic archeon Sulfolobus solfataricus:resilience as a key factor in thermostability. J. Mol.Biol. 271: 789–80273 Fields P. A. and Somero G. N. (1998) Hot spots in cold adaptation:localized increases in conformational flexibility in lactatedehydrogenase A4 orthologs of Antarctic notothenioidfishes. Proc. Natl. Acad. Sci. USA 95: 11476–1148174 Feller G., Payan F., Theys F., Qian M., Haser R. and Gerday C.(1994) Stability and structural analysis of a-amylase from theAntarctic psychrophile Alteromonas haloplanctis A23. Eur. J.Biochem. 222: 441–44775 Lonhienne T., Zoidakis J., Vorgias C. E., Feller G., Gerday C.and Bouriotis V. (2001) Modular structure, local flexibilityand cold-activity of a novel chitobiase from a psychrophilicAntarctic bacterium. J. Mol. Biol. 310: 291–29776 Gerike U., Danson M. J. and Hough D. W. (2001) Cold-activecitrate synthase: mutagenesis of active-site residues. ProteinEng. 14: 655–66177 Tsigos I., Mavromatis K., Tzanodaskalaki M., Pozidis C.,Kokkinidis M. and Bouriotis V. (2001) Engineering the propertiesof a cold active enzyme through rational redesign of theactive site. Eur. J. Biochem. 268: 5074–508078 Mavromatis K., Tsigos I., Tzanodaskalaki M., Kokkinidis M.and Bouriotis V. (2002) Exploring the role of a glycine clusterin cold adaptation of an alkaline phosphatase. Eur. J. Biochem.269: 2330–233579 Ohtani N., Haruki M., Morikawa M. and Kanaya S. (2001)Heat labile ribonuclease HI from a psychrotrophic bacterium:gene cloning, characterization and site-directed mutagenesis.Protein Eng. 14: 975–98280 Ciardiello M. A., Camardella L., Carratore V. and Prisco G. di(2000) L-Glutamate dehydrogenase from the Antarctic fishChaenocephalus aceratus: primary structure, function andthermodynamic characterisation: relationship with cold adaptation.Biochim. Biophys. Acta 1543: 11–2381 Fields P. A., Wahlstrand B. D. and Somero G. N. (2001) Intrinsicversus extrinsic stabilization of enzymes: the interactionof solutes and temperature on A4-lactate dehydrogenaseorthologs from warm-adapted and cold-adapted marinefishes. Eur. J. Biochem. 268: 4497–450582 Feller G., D’Amico S., Benotmane A. M., Joly F., Van BeeumenJ. and Gerday C. (1998) Characterization of the C-terminalpropeptide involved in bacterial wall spanning of alphaamylasefrom the psychrophile Alteromonas haloplanctis. J.Biol. Chem. 273: 12109–1211583 Privalov P. L. (1979) Stability of proteins: small globular proteins.Adv. Protein Chem. 33: 167–24184 Becktel W. J. and Schellman J. A. (1987) Protein stabilitycurves. Biopolymers 26: 1859–187785 Makhatadze G. I. and Privalov P .L. (1995) Energetics of proteinstructure. Adv. Protein Chem. 47: 307–42586 Petrounia I. P. and Arnold F. H. (2000) Designed evolution ofenzymatic properties. Curr. Opin. Biotechnol. 11: 325–33087 Tobin M. B., Gustafsson C. and Huisman G. W. (2000) Directedevolution: the ‘rational’ basis for ‘irrational’ design.Curr. Opin. Struct. Biol. 10: 421–42788 Giver L., Gershenson A., Freskgard P. O. and Arnold F. H.(1998) Directed evolution of a thermostable esterase. Proc.Natl. Acad. Sci. USA 95: 12809–1281389 Cherry J. R., Lamsa M. H., Schneider P., Vind J., Svendsen A.,Jones A. et al. (1999) Directed evolution of a fungal peroxidase.Nat. Biotechnol. 17: 379–38490 Sterner R., Kleemann G. R., Szadkowski H., Lustig A., HennigM. and Kirschner K. (1996) Phosphoribosyl anthranilateisomerase from Thermotoga maritima is an extremely stableand active homodimer. Protein Sci. 5: 2000–200891 Taguchi S., Ozaki A. and Momose H. (1998) Engineeringof a cold-adapted protease by sequential random mutagenesisand a screening system. Appl. Environ. Microbiol. 64:492–49592 Lebbink J. H., Kaper T., Bron P., Oost J. van der and Vos W. M.de (2000) Improving low-temperature catalysis in the hyperthermostablePyrococcus furiosus beta-glucosidase CelB bydirected evolution. Biochemistry 39: 3656–366593 Merz A., Yee M. C., Szadkowski H., Pappenberger G.,Crameri A., Stemmer W. P. et al. (2000) Improving the catalyticactivity of a thermophilic enzyme at low temperatures.Biochemistry 39: 880–88994 Roovers M., Sanchez R., Legrain C. and Glansdorff N. (2001)Experimental evolution of enzyme temperature activity profile:selection in vivo and characterization of low-temperature-adaptedmutants of Pyrococcus furiosus ornithine carbamoyltransferase.J. Bacteriol. 183: 1101–110595 Yasugi M., Amino M., Suzuki T., Oshima T. and Yamagishi A.(2001) Cold adaptation of the thermophilic enzyme 3-isopropylmalatedehydrogenase. J. Biochem. (Tokyo) 129:477–48496 Lonn A., Gardonyi M., Zyl W. van, Hahn-Hagerdal B. andOtero R. C. (2002) Cold adaptation of xylose isomerase fromThermus thermophilus through random PCR mutagenesis:gene cloning and protein characterization. Eur. J. Biochem.269: 157–16397 Kohen A., Cannio R., Bartolucci S. and Klinman J. P. (1999)Enzyme dynamics and hydrogen tunnelling in a thermophilicalcohol dehydrogenase. Nature 399: 496–49998 Zavodszky P., Kardos J., Svingor and Petsko G. A. (1998) Adjustmentof conformational flexibility is a key event in the

662 G. Feller Cold-active enzymesthermal adaptation of proteins. Proc. Natl. Acad. Sci. USA 95:7406–741199 Wrba A., Schweiger A., Schultes V., Jaenicke R. and ZavodskyP. (1990) Extremely thermostable D-glyceraldehyde-3-phosphatedehydrogenase from the eubacterium Thermotoga maritima.Biochemistry 29: 7584–7592100 Svingor A., Kardos J., Hajdu I., Nemeth A. and Zavodszky P.(2001) A better enzyme to cope with cold: comparative flexibilitystudies on psychrotrophic, mesophilic, and thermophilicIPMDHs. J. Biol. Chem. 276: 28121–28125101 Jaenicke R. (2000) Do ultrastable proteins from hyperthermophileshave high or low conformational rigidity? Proc.Natl. Acad. Sci. USA 97: 2962–2964102 Hernandez G., Jenney F. E. Jr, Adams M. W. and LeMaster D.M. (2000) Millisecond time scale conformational flexibility ina hyperthermophile protein at ambient temperature. Proc.Natl. Acad. Sci. USA 97: 3166–3170