Keywords

1 Introduction

The first significant report related to cold-active enzymes dates back to more than a quarter of a century with the publication of the properties of a heat-labile alkaline phosphatase isolated from an Antarctic bacterium (Kobori et al. 1984). This paper can be considered as a seed paper since the two main properties of cold-active catalysts were already correctly described, i.e., a high specific activity at low temperature and a rather high-temperature sensitivity. The biotechnological potential of this enzyme was also underlined since this cold-active enzyme could be, contrary to calf intestinal ATPase, rapidly inactivated by mild heat treatment offering a very convenient method for the 5′-end radio-labelling of DNA fragments. Cold-active enzymes are produced by psychrophiles, a term introduced by Schmidt-Nielsen (1902), but their existence in cold environments was considered as hypothetical due to the fact that their optimum growth temperature was generally above 20 °C. They were considered as cold tolerant or psychrotolerant rather than cold loving or psychrophiles (Ingraham and Stokes 1959). This confusion still persists nowadays since some investigators erroneously consider that the so-called optimum temperature based on the rate of growth is the best possible temperature for microorganisms. It was the merit of Hess (1934) to point out that numerous bacteria, although their maximum growth rate occurred around 20 °C, were true psychrophiles simply because in broth cultures, the largest cell populations were obtained at 5 °C rather than at 20 °C. This was confirmed by more recent works such as that of Margesin (2009) who stated that «cultivation temperatures close to the maximum growth temperature are not appropriate for studying psychrophiles». It was also shown that the production of extracellular enzymes was the highest at temperatures close to that of the environment of the microorganism and far below the so-called optimum temperature (Feller et al. 1994). An overview on the concept of psychrophily and psychrotolerance is reported in Chap. 1.

The adaptative modifications of enzymatic properties as a function of environmental temperature have been discussed long time ago by pioneer scientists such as Peter Hochachka and George Somero (Hochachka and Somero 1973; Somero 1977). They defined three possible strategies for the adaptation of enzymes to low temperatures: first, the organism can increase the concentration of the enzyme in the cell or secreted; second, the activity can also be modulated by increasing the substrate-binding ability; and third, the organism can modify in a permanent way the catalytic efficiency of the enzyme. The first strategy is, at first sight, improbable due to its energy cost in a cold environment; the second strategy will not be efficient in the case of enzymes working at saturated substrate concentration that probably concerns most of the extracellular enzymes. In the case of intracellular enzymes, it has been shown that the Km values of phosphoenolpyruvate for pyruvate kinases of fishes adapted to different temperatures were the lowest in the range of temperatures usually experimented by the organisms and that at the respective cell temperatures, the substrate-binding affinities were close to each other (Somero 1977). So, only the third strategy, i.e., the production of enzymes specifically adapted to low temperatures, seemed to be the only one appropriate. These first information, as well as an easier accessibility to Arctic and Antarctic sites, have stimulated the interest for these enzymes and in the 90s several groups were formed, thanks to the support of the European Union, and focused on psychrophiles with the aims to shed some light on the fundamentals aspects and on the biotechnological potential of these extreme organisms. Therefore, significant progresses in the understanding of the molecular adaptation to cold of cold-active enzymes were made. In this context, the characteristics of a cold-active xylanase from the Antarctic yeast, Cryptococcus adeliensis (formerly Cryptococcus adeliae), was investigated in 1996 (Petrescu et al. 2000). When compared to its mesophilic counterpart, C. adeliensis, a maximum cell density was obtained by cultivation at 4 °C and exceeded by sixfold the cell density reached by C. albidus at the same temperature. By contrast, at 20 °C, the cell density obtained with the mesophilic strain was 2.5 times as large as that of the psychrophilic organism illustrating the cold-adapted character of C. adeliensis. Concomitant differences were observed at the level of the respective catalytic properties of the xylanases with the cold-active enzyme displaying a kcat at 5 °C three times as high as that of the mesophilic counterpart. Interestingly also, the cold-active enzymes displayed an activation energy, ΔG*, about 5 % lower than that of the mesophilic enzyme. This mainly resulted from a significantly lower activation enthalpy, ΔH*, partially compensated by a more negative value of the activation entropy ΔS*. The cold-active enzyme was also rapidly inactivated after exposure at 30 °C contrarily to the mesophilic enzyme.

2 Catalytic Properties and Substrate Binding

Few years ago, (Garcia-Viloca et al. 2004), a generalized expression of reaction rate for enzyme-catalysed reactions, based on the transition state theory developed by Eyring (1935), was introduced. This expression, kcat = γ(T)kBT/h exp (−ΔG*/RT), through its coefficient γ(T), takes into account the possibility that the activated complex ES* could recross the energy barrier under the influence of various parameters among which the viscosity of the medium should play an important role in inducing a value for γ(T) that could differ from 1 in the case of cold-active enzymes. This coefficient, an extended version of the old transmission coefficient κ, has been usually neglected, but the intracellular space represents a very crowded medium with viscosity increasing from 2.5 cP at 20 °C to 5.0 cP at 0 °C. This high viscosity affects reaction rates as demonstrated by studies on lactate dehydrogenases (Demchenko et al. 1989) and alkaline phosphatase (Homchaudhuri et al. 2006). In this latter study, it was also demonstrated that, next to the influence of viscosity, the high fractional cell volume occupancy by the numerous macromolecules could also severely influence reaction rates of enzyme-catalysed reaction in reducing the frequency of enzyme–substrate encounter. Therefore, the catalytic parameters of cold-active enzymes measured in test tubes are probably far from those prevailing in cells. Taking these considerations apart, numerous investigations on cold-active enzymes have clearly demonstrated that these enzymes display a specific activity usually much higher than their mesophilic and a fortiori thermophilic counterparts at low and moderate temperatures as shown in Fig. 15.1. This figure concerns, from left to right, the temperature dependence of the activity of ornithine transcarbamylase (OTCase) from psychrophilic (Moritella abyssi) and mesophilic bacteria (Escherichia coli) and hyperthermophilic archaea (Pyrococcus furiosus) (Xu et al. 2003). One can see that the apparent optimum temperature of the OTCases of these microorganisms considerably differs from each other with a difference of nearly 40 °C between the psychrophilic and mesophilic enzymes. The specific activity of the cold-active enzyme is much higher than that of the mesophilic counterpart in the range of 0–30 °C, and the enzyme is also rapidly inactivated by heat since the apparent optimum does not exceed 25 °C. A large majority of cold-active enzymes displays similar properties as shown in Table 15.1. In a limited number of cases, the specific activity of the cold-active enzyme does not exceed that of the mesophilic one as demonstrated by arginine kinase that shows an activity similar to that of the mesophilic enzyme (Suzuki et al. 2012). This type of anomaly has also been mentioned in the case of chitinase (Lonhienne et al. 2001a), and isocitrate reductase (Fedoy et al. 2007), for example. In this latter case, the thermal stability also seems higher than that of the mesophilic counterpart. These data tend to indicate that the enzyme from the psychrophilic bacterium Desulfotalea psychrophila is not well adapted to low temperature. It would be interesting to compare the catalytic properties of other enzymes from D. psychrophila to those of homologous enzymes from the mesophilic bacterium Desulfitobacterium hafniense to see whether this apparently incomplete adaptation to cold can be extended to other enzymes and possibly to the whole organism. In the case of chitinase, the type of substrate is probably important since chitins are structurally very different from each other. As far the arginine kinase from the deep-sea clam Calyptogena kaikoi is concerned, the significance of the data is uncertain since, at 25 °C, the activation enthalpy is, as expected, lower than that of the mesophilic enzyme from clams of the genus Corbicula. This favourable term is, however, over-compensated by a much more negative activation entropy, the difference leading to a less favourable activation energy. As this enzyme is intracellular, one has in fact to consider the physiological efficiency kcat/Km rather than the kcat alone because these enzymes are not probably working at saturated substrate concentration. If this is taken into consideration, the cold-active arginine kinase is still, at 10 °C, three times less efficient than the mesophilic enzyme.

Fig. 15.1
figure 1

Relative activities of psychrophilic (Moritella abyssi; black circles) mesophilic (E. coli; black squares) and thermophilic (Pyrococcus furiosus; black triangles) OTCases as a function of temperature. Reproduced with permission from Xu et al. (2003)

Full size image
Table 15.1 Km and kcat values of some psychrophilic (P) and mesophilic (M) enzymes
Full size table

Also interesting in Table 15.1 are the values of the Michaelis constant, Km. With the possible exception of lactate dehydrogenase from the icefish Champsocephalus gunnarii, the Km of cold-active enzymes are systematically higher than the values observed in the case of mesophilic counterparts and the tendency is the same for extracellular and intracellular enzymes. The larger values can be translated into a lower affinity of the cold-active one, provided that one neglects the rate constants not directly involved in the dissociation of the enzyme–substrate complex. In the case of intracellular enzymes, the catalytic efficiency or physiological efficiency is better evaluated by the ratio Km/kcat. That leads to largely spread situations, favourable for cold-active endonuclease and lactate dehydrogenase but unfavourable for DNA ligase, Isocitrate dehydrogenase and OTCase. So, the usual increase in the rate constant, kcat, of cold-active enzyme is not systematically large enough to compensate for the loss of substrate-binding affinity.

That could indicate that the adaptation of some cold-active enzymes is not complete and depends on the enzyme concerned and probably also on the evolution history of the organism. For extracellular enzymes, the situation appears more favourable since they are probably generally working at saturated substrate concentrations. As already discussed by Somero (1977), the Km values are rather constant in the range of temperature experimented by the organism but can rapidly increase for temperatures exceeding this zone. This loss in substrate-binding affinity contributes to fix the limit of the thermal tolerance of living organisms. The nearly systematic increase in the Km values of cold-active enzymes is not fortuitous and constitutes a consequence of the structural modifications needed to improve the catalytic efficiency of these enzymes at low temperatures.

3 The Activity Challenge

A quick look to the modern form of the Eyring equation mentioned above indicates that the best way to improve the catalytic efficiency of enzymes at low temperature would be to further decrease the activation energy ΔG* of the catalysed reaction when compared to the reaction catalysed by the mesophilic counterpart. Simply because the rate constant is exponentially related to the activation energy and that even a small decrease in this value would considerably affect the velocity of the reaction. The activation energy is related to the energy changes necessary to transform the ground state of the enzyme–substrate complex into an activated state. This eventually leads to the transformation of substrate into product. The energy change resulting from the interaction of the enzyme with the substrate has also to be taken into consideration since it is associated with a negative energy change that places the enzyme substrate complex in an unfavourable deep pit of energy as shown in Fig. 15.2. As the Km values of cold-active enzymes are higher than their mesophilic counterparts, one can consider that the ground state of the ES complex of the cold-active enzyme has a higher free energy level that is favourable in limiting the energy change necessary to reach the activated state. The fact that the ground state of the complex ES occupies a higher level of energy could also be favourable for the activation of ES into ES*. Indeed, if less weak bonds or looser bonds are implicated in the formation of the complex ES, less costly structural modifications will be necessary to reach the activated state. That could lead to the situation represented in Fig. 15.2 showing a significant decrease in the activation energy in the case of the cold-active enzyme. Experimentally, the activation energy of some cold-active enzymes has been measured and compared to that of their mesophilic counterpart and the data are presented in Table 15.2. The activation energy ΔG* of cold-active enzymes is systematically lower that that of mesophilic homologues to an extent ranging roughly from 1 (amylases) to 10 % (cellulases and bacterial xylanases). These rather moderate figures result from an important decrease in the activation enthalpies ΔH*, that, for example, amounts to 25 % in the case of the amylases, systematically compensated by a much more negative value of the activation entropy ΔS*. This indicates that the activated state of the psychrophilic enzymes requires a much more important reordering of the ground state. The activation entropy values of three mesophilic enzymes, i.e., the chitobiase, endonuclease and lysozyme, are positive. That could mean that contrarily to the other enzymes, the activation state ES* is characterized by more disorder than the ground state or, alternatively, that the rearrangement of water molecules, under the form of a release, induces a positive value of the entropy change in the course of the activation. The negative values of the activation entropy of the other cold-active enzymes confirm that they display a rather loose structure as suggested by the nearly systematic increase in the Km values. So the high catalytic efficiency of cold-active enzymes seems to be associated with a rather open structure that has also been associated with a certain loss of specificity of these enzymes. Indeed multi-substrate cold-active enzymes seem to display a broader specificity when compared to their mesophilic counterpart. This has been demonstrated in the case of elastases (Smalas et al. 2000), alcohol dehydrogenase (Tsigos et al. 1998) and α-amylases (D’Amico et al. 2006). These enzymes have the propensity to more easily accept bulky substrates but are less efficient in the case of small-size substrates.

Fig. 15.2
figure 2

Energetics of the activation of an enzyme-catalysed reaction for mesophilic enzyme (continuous line) and psychrophilic counterpart (dotted line). The activation in psychrophilic enzymes is rendered easier by a decrease in the affinity of the enzyme for the substrate (higher level of ES) and by a possibly lower energetic level of ES*

Full size image
Table 15.2 Activation parameters of a few psychrophilic enzymes (P) compared to those of mesophilic counterparts (M)
Full size table

The notion of flexibility in relation with the high specific activity and the generally low thermal stability of cold-active enzymes has been introduced long time ago and notably by Somero (1977). This correlation has been demonstrated in a large majority of cases, but a more refined view of the linkage between the catalytic properties and overall thermal stability of enzymes is probably now emerging. In fact, the catalytic efficiency of a protein is mainly related to the mobility–flexibility of crucial parts of the molecular edifice. An improvement of this flexibility usually leads to an overall decrease in the thermal stability but in a limited number of cases this can occur without significant modifications or visible changes of the thermal stability. Intuitively, one can indeed think that substitutions of amino acids in existing loops, for example, can modify the conformational mobility–flexibility of regions crucial for catalysis without altering the thermal stability. A clear relationship between the activity and local flexibility in psychrophilic and mesophilic enzymes has been recently demonstrated in carbonic anhydrases. It was shown that the cold-active enzyme with a higher specific activity displays a higher flexibility in the region that controls the folding of the protein contrarily to the mesophilic counterpart (Chiuri et al. 2009). EPR spectroscopy has also shown that the catalytic function of a cold-active alkaline phosphatase was closely related to the relative mobility of the helix carrying cysteine 67 and nucleophilic serine 65 (Heidarsson et al. 2009). In subtilisin-like proteases, the relative flexibility of a beta-hairpin close to the active site correlates well with the differential activity of psychrophilic, mesophilic and thermophilic enzymes, but in this example, the homologous enzymes also display a different thermostability (Tiberti and Papaleo 2011).

Protein dynamics obviously play a crucial role in catalytic processes; a systematic investigation using molecular dynamics of 5 pairs of psychrophilic and mesophilic or thermophilic enzymes has shown that amino acid sequences common to both enzymes are generally more flexible in cold-active enzymes, this in relation with differences in the rate or amplitude of opening of the respective active sites. However, in the case of α-amylases, the overall flexibility of the mesophilic enzyme was higher than that of the psychrophilic one due to the insertion of additional loops. It was also shown that the flexibility of orthologous xylanases was approximately the same at the respective apparent optimum temperatures (Spiwok et al. 2007). Experiments using NMR relaxation have also demonstrated that in adenylate kinases, the rate-limiting step was strictly correlated with flexibility in relation with the opening and closing of the active-site lids that shows slower rates in the thermophilic enzyme when compared to the mesophilic counterpart (Wolf-Watz et al. 2004). An apparent increase in the flexibility of a region important for the catalytic efficiency was noticed in the isocitrate dehydrogenase from the cold-adapted D. psychrophila and that seems to be induced by the formation of a methionine cluster, but the catalytic efficiency of this enzyme is lower than that of the mesophilic counterpart from D. hafniense. Its activation energy ΔG* is also higher at 5 °C as well as its thermostability. The Km of the cold-active enzyme for isocitrate is strongly affected by temperatures over 25 °C and is, in any case, higher than that of the mesophilic counterpart (Fedoy et al. 2007). The interpretation of these data seems rather complex since, apart from the Km value and possibly the local flexibility, this enzyme does not show the usual characteristics of a cold-adapted enzyme, i.e., a higher catalytic efficiency at low temperature and a lower thermal stability. As already mentioned above, it would certainly be interesting to study the biochemical characteristics of other enzymes from this organism. The notion of flexibility is rather complex since the term can refer to structural modifications as a function of time which can be assimilated to a «respiration» of the molecular edifice or to the amplitude of the conformational changes. Both are probably important for the accommodation of the substrates at low temperature and the release of products. The flexibility of proteins has been evaluated by various techniques. Molecular dynamics simulations (MDS) have been applied with a certain success in the case for instance of subtilisin-like proteases from organisms living in different temperature environments (Tindbaek et al. 2004; Tiberti and Papaleo 2011). It was shown that the higher catalytic efficiency of cold-active enzymes was correlated to a higher flexibility of regions important for catalysis, itself correlated to an evolution towards a lower thermal stability from high to low-temperature environments. Recent investigations using MDS have corroborated the idea that the adaptation to cold, in terms of function, is correlated with an improvement of the flexibility of crucial regions in various proteins such as chitinases (Ramli et al. 2012), frataxins (Roman et al. 2013), haemoglobins (Stadler et al. 2012) and beta-tubulins (Chiappori et al. 2012). It has also been shown from studies carried out on homologous 3-isopropylmalate dehydrogenases and thermolysin-like proteases that, at their respective environment temperature, enzymes from mesophilic and thermophilic organisms maintain a balance between their overall rigidity and local flexibility important for their catalytic properties (Radestock and Gohlke 2011) in agreement with the strong correlation between resistance to unfolding and evolutionary adaptation temperature (Hochachka and Somero 2002). Amide hydrogen–deuterium exchanges have provided controversial data. For example, a psychrophilic 3-isopropylmalate dehydrogenase was found to be more rigid than its mesophilic counterpart (Svingor et al. 2001). The «anomalies» observed with the H/D exchange techniques may be due to the use of non-relevant temperatures or time scales. The temperature factor or B-factor has also been used to evaluate the relative flexibility of proteins. It represents the spread of electron densities and can be calculated from known three-dimensional structures. Average B-factors are not systematically correlated with the activity-flexibility relationship (Sun-Yong et al. 1999) but relative B-factors representing the ratio between the B-factors of local parts of the proteins and the average B-factor apparently correlate well with the relative activity and thermostability of psychrophilic and mesophilic enzymes (Russell et al. 1998; Sun-Yong et al. 1999). A recent systematic comparison of the B-factors of twenty pairs of psychrophilic and mesophilic enzymes has demonstrated that psychrophilic enzymes are more flexible in 5-turn and strand secondary structures; they have also larger average cavity sizes and these cavities are lined with an increased frequency of acidic groups (Paredes et al. 2011). Fluorescence quenching of trytophane fluorescence using acrylamide has been used successfully to evaluate the relative flexibility of psychrophilic and mesophilic enzymes. The data have shown that psychrophilic enzymes are much more permeable than their mesophilic counterparts in the case of Ca2+-Zn2+ protease (Chessa et al. 2000), α-amylase (D’Amico et al. 2003; Cipolla et al. 2012), DNA ligase (Georlette et al. 2000, 2003), xylanase (Collins et al. 2003), cellulase (Sonan et al. 2007), aminopeptidase (Huston et al. 2008) and thermolysin (Xie et al. 2009). An example of the comparison of tryptophane fluorescence quenching of psychrophilic and mesophilic enzyme is shown in Fig. 15.3. One can see that the cold-active cellulase presents a much more opened structure when compared to the mesophilic enzyme. Moreover, the mutants that are characterized by a progressive shortening of the linker region (Cel5GΔ1P, Cel5GΔ2P), which induces a progressive increase in the melting point, display a more compact structure, further enhanced in the case of the variant Cel5GΔ2P mutCC carrying an additional disulphide bridge and of the catalytic module Cel5GCM. This once again illustrates the good correlation that generally exists between the low stability and flexibility of cold-active enzyme responsible for their higher catalytic efficiency at low temperature (Sonan et al. 2007).

Fig. 15.3
figure 3

Quenching of tryptophane fluorescence at 10 °C of cold-active cellulase from Antarctic Pseudoalteromonas haloplanktis (Cel5G, black dots) and variants differing by a shorter length of the linker region (Cel5GΔ1P, open circles; Cel5GΔ2P, black squares) as well as mutants characterized by higher melting points (Cel5GΔ2PmutCC, open squares; Cel5GCM, diamonds) and mesophilic cellulase from Erwinia chrysanthemi (Cel5A, black triangles). F0/F represents the ratio between the fluorescence in the absence (F0) and presence (F) of acrylamide at various concentrations. Adapted from Sonan et al. (2007)

Full size image

4 Conformational and Thermodynamic Stability

The analysis of the thermodependence of the activity of cold-active enzyme, as illustrated in Fig. 15.1, indicates that these enzymes are inactivated at temperatures lower than that of their mesophilic counterparts, and in consequence, they seem to display a lower thermal stability. This thermal stability has been investigated by numerous techniques from the evaluation of the residual activity after exposure for a certain time at a given temperature to biophysical techniques such as circular dichroïsm, fluorescence spectroscopy or microcalorimetry. The method using residual activity is only valid if one demonstrates that the unfolding is irreversible at all temperature tested; an example of the use of this technique is shown in Fig. 15.4. In this figure, the thermodependence of the activity of a cold-active polygalacturonase from Icelandic yeast (Birgisson et al. 2003) is compared to the thermal stability of the enzyme exposed for one hour at the respective temperatures. One can see that the enzyme starts to unfold at a temperature around 25 °C, far below the so-called optimum temperature which is around 40 °C. This figure clearly demonstrates that the so-called optimum temperature is only an apparent optimum since at this temperature, the half-life of the enzyme is rather short. Another illustration of the thermodependence of the activity and stability of a cold-active enzyme is shown in Fig. 15.5. That concerns a cold-active α-amylase for the Antarctic bacterial strain Pseudoalteromonas haloplanktis (AHA) as compared with the mesophilic enzyme from pig pancreas (PPA) and the thermophilic counterpart from Bacillus amyloliquefaciens (BAA). The upper panel represents the relative activities and the lower panel the thermal unfolding as followed by fluorescence spectroscopy (D’Amico et al. 2003). A striking feature of these experiments is the fact that the inactivation of the cold-active enzymes precedes the unfolding of the structure. In the mesophilic and thermophilic enzymes, the apparent optimum is strictly correlated to the unfolding transition. That can mean that the active site is very sensitive to temperature or that the ES complex is destabilized before any significant or visible modification of the three-dimensional structure of the enzyme. Such phenomenon has been observed in other cold-active enzymes such as for instance xylanases (Collins et al. 2003, 2007) and DNA ligases (Georlette et al. 2003).The thermal sensitivity of the active site and/or of the ES complex is in agreement with the increase in Km values reported for cold-active enzymes. In general, the residues lining the catalytic cavities and involved in the interaction of the substrate are strictly conserved. This has been notably demonstrated by comparison of the crystallographic structures of the psychrophilic α-amylase from an Antarctic bacterium with that of the homologous enzyme from pig pancreas. The superimposition of the active site of the two enzymes has shown that the 24 residues forming the catalytic cleft have a position strictly conserved in both enzymes (Aghajari et al. 1998, 2002). Therefore, the flexibility of the catalytic region should originate from other regions of the protein. In other cases, the catalytic cavities have been altered so as to confer a larger opening of the catalytic region of psychrophilic enzymes that favours the accommodation of larger substrates and facilitates the release of products. This has been achieved through the modification, for example, of electric potentials that favours the attraction and orientation of substrates in citrate synthase (Russell et al. 1998), malate dehydrogenase (Kim et al. 1999), uracyl-DNA glycosylase (Leiros et al. 2003) and trypsin (Gorfe et al. 2000). In a cold-active Ca2+-Zn2+ protease, an additional Ca2+ pulls the backbone lining the entrance of the site and offers in this way a much better accessibility for the substrate (Aghajari et al. 2003). In cold-active iron superoxide dismutases, it has been shown that the residues forming the active site and the active-site channel are more mobile than in mesophilic counterparts (Merlino et al. 2010).

Fig. 15.4
figure 4

Activity as a function of temperature of polygalacturonase from the Antarctic yeast Cystofilobasidium capitatum (formerly Cystofilobasidium lari-marini) S3B (grey circles) and residual activities of the enzyme after exposure at various temperatures (black squares). Note that at the so-called optimum temperature, around 40 °C, the enzyme is already partially unfolded. Adapted from Birgisson et al. (2003)

Full size image
Fig. 15.5
figure 5

Upper panel Relative activity as a function of temperature of psychrophilic (open circles), mesophilic (open triangles), and thermophilic (open squares) α-amylases. Lower panel Concomitant thermal unfolding as recorded by fluorescence spectroscopy. Worth noting is the fact that the thermal inactivation of the psychrophilic enzyme occurs without any significant modification of the three-dimensional structure contrary to mesophilic and thermophilic counterparts. Adapted from Georlette et al. (2004)

Full size image

The thermodynamic stability of some cold-adapted proteins has been investigated assuming that the unfolding is a reversible two-state process defined by the equilibrium

$$ \Updelta {\text{G}}_{{{\text{N}} - {\text{U}}}} = - {\text{RT}}\;{ \ln }\;{\text{K}}_{{{\text{N}} - {\text{U}}}} \quad{\text{with}}\quad{\text{K}}_{{{\text{N}} - {\text{U}}}} = \left[ {\text{U}} \right]/\left[ {\text{N}} \right] $$

For a protein mainly in a native state, ΔGN−U is a positive value and can be defined as the energy necessary to unfold the protein. The thermodynamic stability of enzymes can be evaluated by non-calorimetric techniques making use of chemical denaturants (Pace and Laurents 1989; Talla-Singh and Stites 2008) or more easily in a microcalorimeter in which the heat absorbed during the unfolding of the protein is measured and defined as ΔHcal. It represents the energy necessary to disrupt the weak bonds that stabilize the three-dimensional structure and appears as a peak often symmetrical in the case of cold-active enzymes but often unsymmetric and flattened in the case of mesophilic and thermophilic counterparts. ΔHcal is equivalent to the area limited by the peak and this area significantly increases from psychrophiles to thermophiles. The symmetry is indicative of a pronounced cooperativity of unfolding and the transition appears in this case sharp contrarily to that of mesophilic and thermophilic proteins that show a flattened profile distributed over a rather broad range of temperatures (Feller 2010). As a function of temperature, the stabilization energy or free energy of unfolding is described by an equivalent of the Gibbs–Helmholtz equation:

$$ \Updelta {\text{G}}_{{({\text{T}})}} = \Updelta {\text{H}}_{\text{cal}} \left( { 1- {\text{T}}/{\text{T}}_{\text{m}} } \right) + \Updelta {\text{C}}_{\text{p}} \left( {{\text{T}} - {\text{T}}_{\text{m}} } \right) - {\text{T}}\Updelta {\text{C}}_{\text{p}} { \ln }\left( {{\text{T}}/{\text{T}}_{\text{m}} } \right) $$

In this equation, Tm is the so-called melting temperature, located at the top of the peak, and defined as the temperature of half denaturation where \( \left[ {\text{N}} \right] = \left[ {\text{U}} \right],\;\Updelta {\text{H}}_{\text{cal}} \) has been defined above and is equivalent to the van’t Hoff enthalpy in a two-state model, T is a given temperature, and ΔCp, the heat capacity change occurring during the transition from the native to the unfolded state. The calculation of the stabilization energy, over a temperature range where the native state is in excess over the unfolded state, gives access to stability curves as shown in Fig. 15.6. These curves, which are parabola-shaped, have been established for cold-active α-amylase AHA, mesophilic α-amylase PPA and thermophilic α-amylase BAA. One can see that over the whole range of temperature, the cold-active enzyme presents a much lower thermodynamic stability than its mesophilic and thermophilic counterparts, even in the low-temperature range. This has been attributed to the lower number or/and strength of the weak bonds that stabilizes the structure. A second striking feature is that the maximum stability for the three orthologous enzymes is found in the rather narrow range of 20–30 °C. This is due to the hydrophobic effect found to be maximum around room temperature (Kumar et al. 2002). Cold denaturation is also more easily achieved in the case of the psychrophilic enzyme with a Tm prediction around −10 °C. On the left side of the curve, the stabilization enthalpy and stabilization entropy are negative. That means that the residual stability of the cold-active enzyme near the environment temperature, around 0 °C (Antarctic sea water), is secured through the negative value of the stabilization entropy resulting from the hydration of the hydrophobic groups of non-polar amino acids that contributes to decrease the entropy of the unfolded state (Dias et al. 2010). Cold denaturation occurs with a release of heat contrarily to heat denaturation. It is also interesting to note that at temperatures corresponding to the most stable conditions, around 20 °C, the cold-active enzyme is marginally stable since the energy required for unfolding is around 30 kJ mol−1. The value drops to around 8 kJ mol−1 at 0 °C, the equivalent of about 2 weak bonds.

Fig. 15.6
figure 6

Stabilization energy as a function of temperature of psychrophilic α-amylase (AHA) from the Antarctic strain Pseudoalteromonas haloplanktis, of mesophilic enzyme from pig pancreas (PPA) and of the thermophilic counterpart (BAA) from Bacillus amyloliquefaciens. Adapted from D’Amico et al. (2003)

Full size image

5 Determinants of Cold Adaptation and Mutagenesis

It has been shown that the flexibility of crucial parts of the structure of cold-active enzymes is generally, but not probably always, accompanied by a decrease in the thermal stability of the protein structure. To understand cold adaptation, it is necessary to define the structural changes that are directly or indirectly involved in the adaptation to cold. The problem is rendered more complicated by the high number of neutral mutations resulting from genetic drift that leads to large differences in the amino acid sequences of homologous enzymes produced by related species. The elucidation of the three-dimensional structure of several cold-active enzymes by X-ray crystallography, more than thirty actually, as well as the production of reliable models based on the comparison of the amino acid sequences of mesophilic and psychrophilic proteins, has allowed to detect, in a first approach, the amino acid substitutions that could lead to a decrease in the thermal stability of the molecular structure of cold-active enzymes. Again, only a few of them are intuitively involved in the adaptation to cold since, in low-temperature environments, the absence a high selective pressure on thermal stability can enhance the importance of the genetic drift.

The structural parameters involved in the thermal stability of proteins have been extensively discussed elsewhere (Vieille and Zeikus 2001), and their implications in the adaptation of cold-active enzymes have already been discussed in many reviews (Somero and Low 1976; Feller et al. 1997a; Marshall 1997; Feller and Gerday 1997, 2003; Russell 2000; Smalas et al. 2000; Gerday et al. 2000; Feller 2003, 2010; Hoyoux et al. 2004; Georlette et al. 2004; Siddiqui and Cavicchioli 2006; Marx et al. 2007; Collins et al. 2007, 2008), and there is a consensus to agree that the enhanced flexibility of crucial parts of these enzymes is mainly secured through a weakening of intramolecular weak bonds that stabilize the structure of mesophilic counterparts associated or not to entropic factors that tend to increase the entropy of the folded state. In general, however, the structural changes leading to cold adaptation are rather discrete since the three-dimensional structure of psychrophilic and mesophilic homologous enzymes can be nearly superimposed. Hydrophobic interactions play important roles in driving the folding of proteins in aqueous media and in stabilizing the nascent polypeptide chain. In cold-active enzymes, the number of hydrophobic interactions can be lower and their strength can also be affected by a decrease in the size of the hydrophobic groups and a reduction in the number or/and size of hydrophobic clusters that contribute to decrease the compactness of the molecule in creating cavities that can be occupied by water molecules (Paredes et al. 2011). In other cases, there is an increase in the proportion of hydrophobic groups at the surface of the protein; this induces a destabilization of the molecular edifice through the reduction in the overall entropy of the system through the formation of clathrate-like structures around these groups. The stabilizing effect of salt bridges can be altered through a modification of the orientation of ionized groups and of their reduction in number (Papaleo et al. 2007). The presence of ionized groups carrying similar charge, in general negative, can also modify the stability through repulsion and by increasing the interaction with the solvent (Narinx et al. 1997; Feller et al. 1999; Adekoya et al. 2006). In cold-active uracil-DNA N-glycosylase, however, a more positively charged surface near the active site seems to be important for the adaptation to cold (Moe et al. 2004). In citrate synthases, psychrophilic, mesophilic and thermophilic enzymes have increasingly stronger electrostatic stabilization of the transition state (Bjelic et al. 2008). Some cold-active enzymes are also poorer in arginine residues and display a lower Arg/Arg + Lys ratio (Adekoya et al. 2006). This in relation with the fact that arginine can develop more electrostatic interactions with surrounding amino acids than lysine. α-helices can be assimilated to macroscopic electrostatic dipoles that carry a net positive charge at their N-terminal end and a net negative charge at their C-terminal end (Serrano and Fresht 1989). These secondary structures can be stabilized by various electrostatic interactions in their vicinity that can be weakened in psychrophilic enzymes such as in cold-active trypsins (Leiros et al. 2000). Also the hydrogen bonding network is often weakened in psychrophilic enzymes and found responsible for the increase in flexibility of these enzymes (Xie et al. 2009). The structure of loops connecting secondary structures seems also important in conferring the appropriate flexibility to some cold-active enzymes; their length can be increased, their level of interactions with the internal moiety can be lowered, and their flexibility can be enhanced thanks to the deletion of proline residues (Feller et al. 1997b; Gudmundsdottir 2002; Matsuura et al. 2002). Metal binding can also be involved through the reduction in binding affinity or deletion of site (Narinx et al. 1997; Almog et al. 2003). The deletion of disulphide bridges can also improve the flexibility and activity of cold-active enzymes; this has been demonstrated in cold-active α-amylase and alkaline phosphatase (D’Amico et al. 2003; Asgeirsson et al. 2007; Papaleo et al. 2007).

Following the comparison of the three-dimensional structures of psychrophilic and mesophilic proteins, mutation experiments were carried out in order either to transform psychrophilic enzymes into mesophilic counterparts or to try to force mesophilic enzymes to adopt psychrophilic characteristics.

In a first attempt, random mutagenesis was applied to mesophilic subtilisin BPN’ (Taguchi et al. 1998) and a mutant m63 carrying three mutations, V72I, A92T, G131D, was obtained. It displayed, at 10 °C, an activity, towards the synthetic small-size substrate s-AAPF para nitroanilide, identical to that of the wild-type mesophilic enzyme; only the Km was twice lower than that of subtilisin BPN’. The activity was not tested on large-size substrates and its thermal stability was slightly lower than that of the original enzyme. More success was obtained later on with the multiple mutant m 51 also carrying three mutations, A31T, A88 V and A98T (Taguchi et al. 1999). Its specific activity at 10 °C was 1.5 higher than that of subtilisin BPN’; also towards s-AAPF-pNA, the Km was slightly lower than that of the original enzyme and the thermostability was apparently unchanged. Another cold-active subtilisin (S41) from Antarctic Bacillus sp. (Davail et al. 1994) was submitted to random mutagenesis followed by in vitro recombination. Mutant libraries were screened to identify enzymes that displayed greater thermal stability without sacrificing low-temperature activity. After several generations and recombinations, a variant, 3-2G7, carrying 7 mutations was obtained. The specific activity towards a small-size synthetic substrate, s-AAPF-pNa, displayed a threefold increase at all temperatures tested with no modification of the Km values. A higher thermostability was also observed (Miyazaki et al. 2000). This variant was further engineered to give the subtilisin mutant 8-4A9 carrying 13 amino acid substitutions distributed throughout the structure leading to an increase in the affinity for Ca2+. The midpoint of unfolding was about 3 °C higher than the variant 3-2G7 with a specific activity at 10 °C slightly lower than that of 3-2G7 but nearly twice as high as that of the wild-type S41. The Km fors-AAPF-pNa steadily decreased from S41 to 3-2G7 and 8-4A9 (Wintrode et al. 2001). These experiments tended to demonstrate that it was quite possible to increase the thermal stability of an enzyme while keeping its high catalytic activity at low temperature; in other words that the two properties were not necessarily inversely related. It is, however, worth noting that these data have been obtained with non-natural, synthetic and small-size substrates. If it seems possible to modify, by directed evolution, enzymes in such a way to display a higher thermostability associated with a higher specific activity at low temperature towards artificial substrate, the situation is probably much more complicated when natural large-size substrates are concerned. We have indeed learned that the protease mutant 3-2G7 in fact displays a much lower activity than the wild-type S41 cold-active enzymes when natural and large-size substrates such as proteins were used. That does not mean that in vitro it is not possible to increase the thermostability of a cold-active enzyme while preserving or even increasing its catalytic efficiency at low temperature. A cold-active subtilisin, S39, extremely similar to the S41 enzyme, was also submitted to site-directed mutagenesis with the aim to mainly try to increase its stability (Narinx et al. 1997). Several mutations were introduced first to restore a salt bridge found in subtilisin BPN’, a hydrophobic interaction found in subtilisin Carlsberg, a disulphide bridge present in aqualysine, a mutation N136S supposed to increase the specific activity towards large-size substrates and a mutation T85D to exchange a poor Ca2+ ligand for a strong one. Most of these mutations had a positive effect on the thermostability of the enzyme, especially the mutation T85D, involved in the Ca2+ coordination of calcium-binding site 1 that increased the half-life at 50 °C by more than 50 min. As expected, the affinity of this mutant for calcium was drastically improved. At 5 °C, the specific activity of this mutant towards the synthetic substrate s-FAAF-pNa was 4 times as high as that of the wild-type enzyme; the Km value was not affected. The specific activity was, however, only slightly increased when azocasein was used as substrate. In another cold-active subtilisin-like proteinase, Ser-Ala and Xaa-Pro mutations were introduced to produce single, double and triple mutants. Most of these mutations, especially the Xaa-Pro exchanges, led to an increase in the thermostability of the variants with a concomitant decrease in the specific activities (Arnorsdottir et al. 2007). A thermophilic subtilase was also engineered in order to produce cold-active variants (Zhong et al. 2009). All successful single mutations were located within or near the active site and a variant carrying four amino acid substitutions showed a sixfold increase in specific activity towards casein in the temperature range of 15–25 °C, a decrease in thermal stability, and a shift of its apparent optimum towards low temperature by approximately 15 °C. Interestingly, this multiple mutant was less active towards the synthetic substrate s-AAPF-pNA due to a large increase in Km and decrease in kcat. This again underlines the importance, before drawing robust conclusions, of selecting appropriate substrates when studying multi-substrate enzymes by mutagenesis.

Directed evolution was also applied to a thermostable esterase from B. subtilis using p-nitrobenzyl acetate as substrate (Giver et al. 1998). After several generations, mutants of higher thermostability were obtained; they show a specific activity slightly higher, about 20 %, than that of the wild-type enzyme. In this case, the natural substrates are unfortunately unknown.

The molecular basis of the adaptation to cold of a psychrophilic α-amylase was also investigated by site-directed mutagenesis. Fourteen amino acid substitutions, including double mutations were introduced with the aim to mimic specific structural characteristics found in mesophilic and thermophilic homologous enzymes (D’Amico et al. 2001, 2002) under the form of hydrogen bonds, salt bridges, helix dipole stabilization, hydrophobic interactions and reinsertion of a disulphide bridge that connects domain A and B in mesophilic counterpart. The highest contribution to stability, in terms of both Tm and ΔHcal, was obtained by the introduction of an electrostatic interaction; a double aromatic interaction increased the stabilization energy by a factor of two at 10 °C whereas the reinforcement of hydrophobic clusters within the hydrophobic core of the enzyme induced the production of multiple calorimetric domains similar to those existing in the mesophilic counterpart from pig pancreas. Most of the introduced mutations aiming to increase the stability were efficient but led to a decrease in kcat and Km. Any attempts to further increase the catalytic efficiency at low temperature through a decrease in stability gave rise to unfolded forms suggesting that the cold-active α-amylase has reached a limit in stability precluding any improvement of the specific activity via a decrease in thermal stability. The incorporation of a disulphide bridge found in the mesophilic enzyme decreased the specific activity by a factor of two at 5 °C as well as the Km and showed a microcalorimetric profile similar to that of the mesophilic counterpart with, however, a first Tm transition lower than that of the mesophilic α-amylase. This indicates that the introduction, in the cold-active enzyme, of this disulphide bridge leads to a stabilization of some parts of the cold-adapted protein but also induces an unfavourable structural constraint in another part of this protein. Multiple mutants of this cold-active enzyme were also recently produced to force the cold-active enzyme to display mesophilic characteristics (Cipolla et al. 2012); the specific activity of the variants was lowered whereas the thermal stability was drastically improved with a concomitant decrease in the flexibility of the mutants. It was concluded that these mutants can be considered as structural intermediates between their parents psychrophilic and mesophilic enzymes. The effects of these mutations, single and multiple, on the molecular dynamics of the protein, were recently investigated (Papaleo et al. 2011), and it was shown that the above-mentioned mutants displayed a reduced flexibility in various regions, not only near the active-site and substrate-binding groove, but also at long distance such as in domain C that did not carry any mutation.

Several other cold-active enzymes have been the target of site-directed mutagenesis in order to shed some light on the residues involved in the adaptation. One can mention the work on a cold-active lipase in which the substitutions of three polar residues in the lid region by amino acid residues conserved in homologous mesophilic lipases gave rise to an increase in stability and a modification of the substrate specificity with a preference for C8 rather than C4 fatty acid chains. A substitution, Ser-Gly, increased the specificity for C12 fatty acid chain associated with a decrease in stability this probably in relation with a better accessibility of the active site and improvement of the flexibility in this region (Santarossa et al. 2005).

An interesting study on the adaptation to cold of an alkaline phosphatase from an Antarctic bacterium was carried out by directed evolution to identify amino acid residues involved in the adaptation and stability of this protein (Koutsioulis et al. 2008). Three thermostable and six thermolabile variants were obtained. All the variants showing an increase in stability were also characterized by an increase in the Tm of the first calorimetric transition and persistence, as in the case of wild type, of secondary transitions at higher temperatures; their specific activity at 15 °C was also lower than that of the parent enzyme and this was associated, as expected, with a higher activation energy. By contrast, the two variants, H135E and double mutant H135E/G149D, displaying an activity about 1.5 higher and a much higher Km than that of the wild-type enzyme had a nearly symmetrical microcalorimetric profile with only one detectable transition shifted towards higher temperature by about 6 °C when compared to the first transition of the parent enzyme. This latter, however, displayed a calorimetric domain with a Tm about 7 °C higher than those of both variants. It has been suggested that the mutation H135E distorts the Mg2+ binding site, decreases the affinity for the metal and increases the flexibility of the site presumably through the destabilization of the domain showing the highest thermostability in the wild-type enzyme. In the double mutant, the mutation G149D apparently improves the mobility of residue R148 located in a loop that facilitates the phosphate/substrate coordination during catalysis. Also, after incubation at 60 °C, these two more active variants were also more rapidly inactivated than the wild-type enzyme suggesting an increase in flexibility of the active site.

In a cold-active isocitrate lyase, the replacement of Ala 214 by a Ser residue found in some mesophilic enzyme and close to the Tim barrel involved in substrate binding induced a decrease in the catalytic activity associated with an increase in the thermostability of the mutant presumably due to the two hydrogen bonds formed between Ser and Gln119. It was concluded that Ala 214 was implicated in cold adaptation by providing more flexibility to the active site (Sato et al. 2008).

These experiments of mutagenesis indicate that, in most cases, the high specific activity of cold-active enzymes is related to an improvement of the flexibility of crucial parts of the molecular structure that induces a high thermal sensitivity of the active site and a generally better accessibility for the substrate at low temperatures. The attempts to try to simultaneously increase the thermostability and activity at low temperature have been hardly successful. Directed evolution has produced more stable and more active variants but essentially with multi-substrate enzymes and small-size and artificial substrates, and it remains to be demonstrated that the mutants are also active on large-size substrates at low temperature. One mutated cold-active subtilisin, carrying only one mutation was, however, found more stable and more active on a large-size substrate, azocasein (Narinx et al. 1997), and interestingly, two mutants of a cold-active alkaline phosphatase were found more active than the wild-type enzyme and showed a symmetrical calorimetric profile synonymous of a high cooperativity of unfolding. Although the first transition of the wild-type enzyme had a Tm lower than that of the mutants, these variants had lost the most thermostable calorimetric domain and appear to be more rapidly inactivated than the wild-type alkaline phosphatase. This is correlated to the fact that many cold-active enzymes also show a high cooperativity on unfolding whereas heat inactivation often precedes any significant changes in the three-dimensional structure.

6 Conclusions

Psychrophilic organisms living in permanently cold habitats have developed numerous adaptations that allowed them to successfully thrive in low-temperature environments. Their enzymes are a key feature of this adaptation. They are much more active than their mesophilic counterparts at low and moderate temperatures and their high specific activity is undoubtedly due to an improvement of the flexibility of the active site or/and of other regions of the structure indirectly involved. This is induced by rather discrete structural modifications which can be located at long distance from the active site and which generally lead to a higher thermal instability of these enzymes, often associated with an even faster heat inactivation. If the general strategy adopted by nature consists in a weakening of the intramolecular forces that stabilize the structure of their mesophilic and a fortiori thermophilic counterparts, this strategy appears to be specific to each enzyme; depends on the position of the enzyme within a metabolic pathway; on its localization, intra-or extra-cellular; on its structural modification capability; on the environment of the organism and on its evolution history. As a consequence, there is a continuum in the adaptation to low temperatures; some cold-active enzymes indeed display high specific activity associated with low Tm while others show rather high Tm and stronger thermal dependence like their mesophilic counterparts.