Molecular characterisation of the thermostability and catalytic properties of enzymes from hyperthermophiles

J.H.G. Lebbink

Research output: Thesisinternal PhD, WU


<p>Hyperthermophilic organisms are able to survive and reproduce optimally between 80°C and 113°C. Most of them belong to the domain of the Archaea, although several hyperthermophilic Bacteria have been described. One of the major questions regarding hyperthermophiles concerns the molecular mechanisms that determine the extreme stability of their macromolecules. In particular, enzymes on the one hand should be flexible in order to efficiently perform catalysis, while on the other hand they should be sufficiently rigid in order to prevent thermal unfolding and inactivation. In other words, they have to be thermostable as well as thermoactive. Therefore, the research described in this thesis has been focused on analysing the molecular determinants of enzyme thermostability and thermoactivity.</p><p>In the first chapter of this thesis, hyperthermophilic microorganisms are introduced, and the origin of the hyperthermophilic phenotype is discussed in the light of established and emerging phylogenetic, biochemical and structural biological considerations. Furthermore, thermodynamic and kinetic protein stability is explained, and an overview is given of the strategies that have been suggested to confer thermostability to enzymes from hyperthermophiles, by analysing amino acid sequences, homology-based structural models, and three-dimensional structures. One of the proposed general strategies for achieving hyperthermostability relates to the number of ion-pairs and their organisation in extensive networks in hyperthermostable enzymes. While biochemical studies have shown that isolated ion-pairs at room temperature do not contribute significantly to protein stability, several theoretical considerations indicate that arrangements of ion-pairs into networks at high temperatures may play an important role in protein stabilisation. In contrast to the need for maximum stability at high temperatures, it is the requirement for flexibility that enables an enzyme to efficiently perform its catalytic function. The mechanisms that determine hyperthermostability on the one hand, hyperthermoactivity on the other hand, and the interplay between these two characteristics, have been investigated in the research described in this thesis, using three different approaches of protein engineering, namely (i) domain swapping, (ii) site-directed mutagenesis, and (iii) random mutagenesis and directed evolution.</p><p>In order to study the molecular determinants of enzyme thermostability and thermoactivity, we decided to include representatives of both hyperthermophilic Archaea and Bacteria in the analysis. <em>Pyrococcus furiosus</em> was selected as hyperthermophilic Archaeon, since it optimally grows at 100°C . Moreover, <em>P.furiosus</em> is very well studied with regard to its physiology, biochemistry and genomics. As representative of the hyperthermophilic Bacteria we choose <em>Thermotoga maritima</em> that grows optimally at 80°C . Both organisms are able to ferment mixtures of polypeptides and a variety of oligosaccharides. One of the enzymes that was chosen as model, isβ-glucosidase, a glycosyl hydrolase involved in the hydrolysis of disaccharides and oligosaccharides into the corresponding monomeric sugars that are further metabolised. The second model enzyme is glutamate dehydrogenase (GDH), a key enzyme in a pathway that is involved in the disposal of reducing equivalents that are generated during sugar fermentation. In both <em>P.furiosus</em> and <em>T.maritima</em> these two key enzymes are expected to perform similar functions.</p><p>GDH is extensively introduced in Chapter 2 and is a useful model enzyme since it is well studied, catalyses an important reaction, and has been isolated and characterised from many organisms in all three kingdoms of life, ranging from psychrophiles to hyperthermophiles. Multiple amino acid sequences are available, as well as high resolution three-dimensional structures from GDH of various sources including the hyperthermophiles <em>P.furiosus</em> and <em>T.maritima</em> . Furthermore, the <em>gdh</em> genes from these hyperthermophiles can be easily overexpressed in <em>Escherichia coli</em> , allowing efficient enzyme purification and mutagenesis (Chapter 3). GDH is in general a hexameric enzyme, composed of six identical subunits that are arranged in two trimers that are stacked upside down on top of each other . Each subunit is composed of a substrate and a nicotinamide cofactor binding domain, linked by a flexible hinge region and separated by a deep cleft in which the active site is located. During catalysis, the hinge region mediates the opening and closing of the cleft by a rotation of the cofactor-domain with respect to the substrate-binding domain. In this way an adequate hydrophobic environment for hydride transfer during catalysis is created.</p><p>Chapter 3 describes the functional expression of the <em>gdh</em> gene from <em>P.furiosus</em> in <em>E.coli</em> . The <em>P.furiosus</em> GDH amounted to 20 % of total <em>E.coli</em> cell protein, with the majority of the expressed protein in the hexameric conformation. Following activation by a heat-treatment, the GDH that could be purified from <em>E.coli</em> was indistinguishable from that purified from <em>P.furiosus</em> . The role of the GDH substrate and cofactor binding domains in conferring thermoactivity and thermostability were studied by exchanging them between the GDHs of the hyperthermophilic archaeon <em>P.furiosus</em> and the mesophilic bacterium <em>C.difficile</em> . Hybrid genes were constructed and successfully expressed in <em>E.coli</em> . One of the resulting hybrid proteins, containing the glutamate-binding domain of the <em>C.difficile</em> and the cofactor-binding domain of the <em>P.furiosus</em> enzyme, did not show detectable activity. In contrast, the complementary hybrid, containing the <em>P.furiosus</em> glutamate and the <em>C.difficile</em> cofactor binding domain was found to be a catalytically active hexamer that showed a reduced substrate affinity but maintained efficient cofactor binding with the specificity found in the <em>C.difficile</em> enzyme. Compared to the <em>C.difficile</em> GDH, the latter archaeal-bacterial hybrid is slightly more thermoactive, less thermostable but much more stable towards guanidinium chloride-induced inactivation and denaturation.These results indicate (i) that the cofactor-binding domain is structurally independent from the substrate-binding domain with respect to affinity for the coenzyme, and (ii) that thermal and chemical stability may be uncoupled.</p><p>The elucidation of the three-dimensional structure of GDH from <em>P.furiosus</em> and its comparison with the homologous GDH from the mesophilic bacterium <em>C.symbiosum</em> , has highlighted the formation of extensive ion-pair networks at domain and subunit interfaces as a possible explanation for the superior thermostability of the archaeal enzyme . In order to assess whether this is a more general stabilising strategy which is not only employed in GDH from <em>P.furiosus</em> , but also in enzymes from other hyperthermophilic Archaea and Bacteria, an homology-based modelling study was carried out using GDH amino acid sequences derived from species spanning a wide spectrum of optimal growth temperatures. Chapter 4 describes the observed correlation between the amount and extent of ion-pair networks and the thermal stability of the studied enzymes, which is consistent with a role for the involvement of such networks in the adaptation of enzymes to extreme temperatures.</p><p>In order to analyse the role of ion-pair networks in more detail, we decided to rationally engineer the networks that are found in the GDH from <em>P.furiosus</em> , into the GDH from <em>T.maritima</em> , in which most of the networks are largely reduced or absent and which is much less thermostable than the pyrococcal GDH . In the study that is described in Chapter 5, we focused on the flexible hinge region that is connecting the two domains. While in GDH from <em>P.furiosus</em> a five-residue ion-pair network is present, this is reduced to two single ion-pairs in <em>T.maritima</em> GDH. Using a site-directed mutagenesis approach, the missing charged residues were introduced into <em>T.maritima</em> GDH.</p><p>The resulting mutant GDHs as well as the wild-type enzyme were overproduced in <em>Escherichia coli</em> and subsequently purified. Elucidation of the three-dimensional structure of the double mutant N97D/G376K at 3.0 Å, showed that the designed ion-pair interactions were indeed formed. Moreover, because of interactions with an additional charged residue, a six-residue network is present in this double mutant. Melting temperatures of the mutant enzymes N97D, G376K and N97D/G376K, as determined by differential scanning calorimetry, did not differ significantly from that of the wild-type enzyme. Identical transition midpoints in guanidinium chloride-induced denaturation experiments were found for the wild-type and all mutant enzymes. Thermal inactivation at 85°C occured more than two-fold faster for all mutant enzymes than for the wild-type GDH. At temperatures of 65°C and higher, the wild-type and the three mutant enzymes showed identical specific activities. However, at 58°C the specific activity of N97D/G376K and G376K was found to be significantly higher than that of the wild-type and N97D enzymes. These results suggest that the engineered ion-pair interactions in the hinge region do not affect the stability towards temperature- or guanidinium chloride-induced denaturation but rather affect the specific activity of the enzyme and the temperature at which it functions optimally.</p><p>Chapter 6 focuses on the role of an 18-residue ion-pair network, which is present at the subunit interface of the GDH from <em>P.furiosus</em> . This network has been studied by introducing four new charged amino acid residues into the subunit interface of GDH from <em>T.maritima.</em> Amino acid substitutions were introduced as single mutations as well as in several combinations. Elucidation of the crystal structure of the quadruple mutant S128R/T158E/N117R/S160E <em>T.maritima</em> GDH showed that all anticipated ion-pairs were formed and that a 16-residue ion-pair network was present. Enlargement of existing networks by single amino acid substitutions unexpectedly resulted in a decrease in resistance towards thermal inactivation and thermal denaturation. However, combination of destabilising single mutations in most cases restored stability, indicating the need for balanced charges at subunit interfaces and high cooperativity between the different members of the network. Combination of the three destabilising mutations in triple mutant S128R/T158E/N117R resulted in an enzyme with a 30 minutes longer half-life of inactivation at 85°C, a 3°C higher temperature optimum for catalysis, and a 0.5°C higher apparent melting temperature than that of wild type GDH. These findings confirm the hypothesis that large ion-pair networks do indeed stabilise enzymes from hyperthermophilic organisms.</p><p>Regarding subunit interfaces in GDH, a more or less complete picture of the role of ion-pair networks has emerged by now. The modelling study that involved GDHs derived from different sources spanning the complete temperature spectrum, showed a progressive increase in formation of networks at the subunit interfaces with increasing temperature. The removal of central ion-pairs from the large 18-residue ion-pair network in <em>P. kodakaraensis</em> GDH has been reported to reduce the half-life for thermal inactivation.</p><p>Furthermore, there are now multiple examples of reconstituted large networks in less thermostable GDHs that significantly increase half-life values for thermal inactivation and apparent melting temperatures (Chapter 6). Ion-pair networks therefore play an important role in the thermostabilisation and maintenance of a correct structure of this enzyme. However, except for the studies described in this thesis (Chapter 5 and 6), three-dimensional structures of mutated GDHs have not been determined. This implies that only for the charged residues introduced into the subunit interface and hinge region of <em>T.maritima</em> GDH it has actually been proven that the predicted networks are indeed formed.</p><p>The role of ion-pairs in the flexible hinge region of GDH is less well resolved. These networks are apparently involved in determining thermoactivity and kinetic parameters, which is consistent with the hinge region being involved in positioning of the domains with regard to each other during catalysis. However, the introduction of a 6-residue ion-pair network in the hinge of <em>T.maritima</em> GDH did not increase the thermostability of this enzyme. This result may reflect the considerable differences between the archaeal and the bacterial enzyme, which share 55% amino acid identity. This consideration is supported by the fact that (i) in <em>Thermococcus litoralis</em> GDH, which has a much higher homology (87% identity) with <em>P.furiosus</em> GDH, a restored ion-pair network was found to be stabilising only in combination with a second site mutation , and (ii) stabilisation of the <em>T.maritima</em> GDH <em></em> subunit interface is only observed after the combination of multiple, destabilising single mutations (Chapter 6).</p><p>Apparently, different strategies for stabilisation have evolved within the hyperthermostable members of this protein family. While the subunit interfaces of the most thermostable, archaeal members are highly charged, the subunit interface of the <em>T.maritima</em> GDH is optimised by hydrophobic interactions . In spite of this, we have shown that it is possible to successfully introduce an archaeal feature into a bacterial enzyme that is stabilised by a different mechanism. The electrostatic optimisation of the subunit interfaces in the archaeal GDHs, on the one hand, and the hydrophobic optimisation of the bacterial GDH subunit interface on the other hand, is not a universal division. Remarkably, exactly the opposite situation has been observed in the superoxide dismutase family, which shows electrostatic optimisation of the subunit interfaces in the enzyme from the bacterium <em>Aquifex pyrophilus</em> , and hydrophobic optimisation in the enzyme of the archaeon <em>Sulfolobus acidocaldarius</em> .</p><p>The thesis continues with an overview of the current knowledge on the well-studied second model enzyme, theβ-glucosidase CelB from <em>P.furiosus</em> (Chapter 7). CelB is involved in the hydrolysis ofβ-(1,4)-linked disaccharides during growth on sugars, and is highly thermostable and thermoactive, with a half-life of thermal inactivation of 85 hours at 100°C and an optimum temperature for catalysis of 102-105°C . Heterologous expression and enzyme purification procedures have been developed that allow for routine site-directed mutagenesis and <em>in vitro</em> applications (Chapter 7).</p><p>Directed evolution is a potent approach to evolve a desired property into a biocatalyst, as well as a promising method to study structure-function and structure-stability relationships. So far, this approach was restricted to model enzymes from mesophilic and thermophilic sources. Chapter 7 describes the development of a directed evolutionary approach for the thermostable CelB by random mutagenesis, <em>in vitro</em> recombination and a rapid screening procedure. The constructed library contains over 6000 random CelB clones with an average of 1 or 2 amino acid substitutions per enzyme. The ability of CelB to hydrolyse chromogenic glucose- and galactose derivatives was used to isolate mutants with a changed ratio in the hydrolysis of these enantiomeric sugars. One of the isolated mutants was characterised in detail and found to contain a single amino acid substitution of asparagine to serine at position 206 in the active site. In the wild-type enzyme this asparagine forms a hydrogen bond with the hydroxyl group of the substrate at the C2 position. Biochemical analysis of the mutant using different substrates provided insight in the active-site architecture of CelB, and indicated that the interactions between CelB and its substrates are similar to those observed in three-dimensional structures of mesophilic homologues complexed with substrate analogues.</p><p>Chapter 8 describes a three-dimensional model of CelB and the validation of this model by engineering substrate specificity. The model is based on 3.3 Å X-ray diffraction data and on the tetramericβ-glycosidase LacS from <em>Sulfolobus solfataricus</em> and the monomeric 6-phospho-β-galactosidase from <em>Lactococcus lactis</em> (LacG) as search models. CelB shows high structural homology with LacS and contains subunits that adopt the common (αβ) <sub>8</sub> barrel motif and are arranged in a slightly twisted tetramer. The difference in substrate specificity betweenβ-glycosidases and 6-phospho-β-glycosidases seems to be determined by three residues in the active site that are conserved within, but not between, both types of glycosidases. This hypothesis was tested by replacing these residues in CelB with the corresponding substitutes, as present in LacG. Mutant CelB containing all three substitutions showed a reduced catalytic activity on non-phosphorylated sugars and increased efficiency on 6-phospho-β. These results indicate that the phosphate-binding site is functional at elevated temperatures.</p><p>Low-temperature activity and substrate specificity of CelB was studied in Chapter 9 using the random CelB library described in Chapter 7. Screening of this library at room temperature on the artificial substrate pNp-glucose resulted in the isolation and characterisation of mutants with up to three-fold increased rates of hydrolysis of this aryl-glucoside. Amino acid substitutions were identified in the active site region, at subunit interfaces, at the enzyme surface, and buried in the interior of the monomers. Characterisation of the mutants revealed that the increase in low-temperature activity was achieved in different ways, including altered substrate specificity and increased flexibility by an apparent overall destabilisation of the enzyme. Kinetic characterisation of active site mutants showed that in all cases the catalytic efficiency at 20°C on <em>p</em> -nitrophenol-β<KLEIN>D</KLEIN>-glucose as well as on the disaccharide cellobiose, was up to two-fold increased. In most cases, this was achieved at the expense ofβ-galactosidase activity at 20°C, and of total catalytic efficiency at 90°C. Substrate specificity is affected by many of the observed amino acid substitutions, only some of which are located in the vicinity of the active site. The largest change in the ratio of <em>p</em> -nitrophenol-β-<SMALL>D</SMALL>-glucopyranoside/ <em>p</em> -nitrophenol-β-<SMALL>D</SMALL>-galactopyranoside hydrolysis is a 7.5-fold increase, which was observed in N415S CelB. This asparagine at position 415 interacts with active site residues that stabilise the hydroxyl group at the C4 position of the substrate, the conformation of which is equatorial in glucose-containing and axial in galactose-containing substrates.</p><p>In conclusion, this random screening approach using a hyperthermostable enzyme resulted in the identification of residues that are critical in determining thermostability, low-temperature activity, and substrate recognition. It was shown that low-temperature activity may be engineered into a hyperthermostable enzyme without affecting its extreme stability. Several mutants display evolved properties that could enable them to perform more efficiently in industrial or biotechnological applications than the wild-type CelB. It is feasible that N415S CelB with its increased catalytic efficiency on cellobiose but unchanged stability, or T371A CelB with its much better affinity for cellobiose, could be applied instead of wild-type CelB in biosensors for the detection of cellobiose at low temperatures. Not only is directed evolution a potent approach to evolve a desired property into a biocatalyst of potential industrial interest, but, as is shown here, it may enhance our understanding of the mechanisms determining enzyme catalysis, stability, and substrate recognition at physiological or extreme conditions.</p><p>In Chapter 10 the molecular basis for the extreme thermostability of CelB has been addressed by comparative analysis of the three-dimensional model of CelB and the crystal structure of the homologousβ-glycosidase LacS from <em>S.solfataricus</em> . Furthermore, this chapter describes the biochemical characterisation of wild-type CelB and mutants thereof, containing substitutions or deletions of charged residues at the C-terminus, which may be involved in the formation of ion-pair networks. All mutant CelB enzymes show decreased optimal temperatures for activity and increased rates of thermal inactivation, indicating that ion-pairs at the C-terminus of CelB play a role in determining the extreme thermostability of this enzyme. Moreover, increased thermal inactivation of CelB in the presence of sodium chloride and at extreme pHs, supports the role of electrostatic interactions in this enzyme. The CelB structure indicates that part of CelB stability is derived from structural adaptations that allow oligomerisation and from the presence of solvent-filled, hydrophilic cavities, which have earlier been identified in LacS. However, structural comparison of several large ion-pair networks that are present in LacS, reveals that a five-residue and a six-residue membered network is not present in CelB and a large 16-membered network at the subunit interface, extensively cross-linking the subunits, has been fragmented.</p><p>The picture emerging from the structural and biochemical characterisation of wild-type, site-directed, and random mutants of CelB, is that, compared to homologous enzymes from mesophilic sources, active site architecture, substrate recognition, acid-base catalysis and transition state stabilisation have been preserved, while thermoactivity and thermostability have been adapted to the high-temperature conditions in which the enzyme has to function optimally.</p><p>An evaluation of the obtained results and conclusions against the original aims as set in the introduction of this thesis reveals the following. With the GDH model we have established and validated a strategy for enzyme stabilisation that seems to be employed by many different hyperthermophiles. Furthermore, we have shown that these features can be successfully engineered into less thermostable enzymes, in order to raise their thermostability. In addition, we have increased our knowledge of enzyme catalysis and substrate recognition at high temperatures and in relation with low-temperature catalysis, using site-directed mutagenesis and directed evolution of the hyperthermostableβ-glucosidase. The directed evolutionary approach, in particular, has been shown to be a potent technique which will in the near future rapidly increase our knowledge of enzyme characteristics and may be employed to increase the biotechnological potential of the enzymes under study.</p>
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • de Vos, W.M., Promotor
  • van der Oost, John, Promotor
Award date19 Nov 1999
Place of PublicationS.l.
Print ISBNs9789058080936
Publication statusPublished - 1999


  • thermophilic microorganisms
  • enzymes

Fingerprint Dive into the research topics of 'Molecular characterisation of the thermostability and catalytic properties of enzymes from hyperthermophiles'. Together they form a unique fingerprint.

  • Cite this