Portal de Periódicos da CAPES

The Effects of Modifying the Surface Charge on the Catalytic Activity of a Thermolysin-like Protease

2002; Elsevier BV; Volume: 277; Issue: 18 Linguagem: Inglês

10.1074/jbc.m200807200

1083-351X

Arno de Kreij, Bertus van den Burg, Gerard Venema, Gert Vriend, Vincent G. H. Eijsink, Jens Erik Nielsen,

Enzyme Structure and Function

The impact of long range electrostatic interactions on catalysis in the thermolysin-like protease fromBacillus stearothermophilus was studied by analyzing the effects of inserting or removing charges on the protein surface. Various mutations were introduced at six different positions, and double-mutant cycle analysis was used to study the extent to which mutational effects were interdependent. The effects of single point mutations on the kcat/Kmwere non-additive, even in cases where the point mutations were located 10 Å or more from the active site Zn2+ and separated from each other by up to 25 Å. This shows that catalysis is affected by large electrostatic networks that involve major parts of the enzyme. The interdependence of mutations at positions as much as 25 Å apart in space also indicates that other effects, such as active site dynamics, play an important role in determining active site electrostatics. Several mutations yielded a significant increase in the activity, the most active (quadruple) mutant being almost four times as active as the wild type. In some cases the shape of the pH-activity profile was changed significantly. Remarkably, large changes in the pH-optimum were not observed. The impact of long range electrostatic interactions on catalysis in the thermolysin-like protease fromBacillus stearothermophilus was studied by analyzing the effects of inserting or removing charges on the protein surface. Various mutations were introduced at six different positions, and double-mutant cycle analysis was used to study the extent to which mutational effects were interdependent. The effects of single point mutations on the kcat/Kmwere non-additive, even in cases where the point mutations were located 10 Å or more from the active site Zn2+ and separated from each other by up to 25 Å. This shows that catalysis is affected by large electrostatic networks that involve major parts of the enzyme. The interdependence of mutations at positions as much as 25 Å apart in space also indicates that other effects, such as active site dynamics, play an important role in determining active site electrostatics. Several mutations yielded a significant increase in the activity, the most active (quadruple) mutant being almost four times as active as the wild type. In some cases the shape of the pH-activity profile was changed significantly. Remarkably, large changes in the pH-optimum were not observed. thermolysin-like protease from B. stearothermophilus thermolysin-like protease 4-morpholineethanesulfonic acid The acceleration of reaction rates by enzymes is one of the essential prerequisites for life as we know it, and the multitude and diversity of enzymes shows that it should be possible to design an enzyme that will catalyze almost any reaction under almost any set of conditions. To achieve a high rate of acceleration, enzymes rely on charged groups in their active site that stabilize the transition state or function as acid or base catalysts in the reaction. The kinetic parameters of enzymes therefore display a significant pH dependence, which is determined by the pKa values of the active site groups.Catalysis depends on intricate electrostatic interactions, which may be noticeable over distances that are large compared with short range of interactions such as hydrogen bonds and hydrophobic contacts. Thus, larger parts of an enzyme may be involved in optimizing its catalytic center than previously thought. The long range character of electrostatic effects is illustrated by a, very limited, number of examples in the literature, showing that changes in surface charge at locations as far as 15 Å from a catalytic center may affect enzyme activity (1Russell A.J. Fersht A.R. Nature. 1987; 328: 496-500Crossref PubMed Scopus (299) Google Scholar, 2Jackson S.E. Fersht A.R. Biochemistry. 1993; 32: 13909-13916Crossref PubMed Scopus (79) Google Scholar). Unfortunately, electrostatic interactions are hard to handle theoretically not only because of their long range character but also because of intrinsic theoretical difficulties. For example, most electrostatic models still use a single rigid protein structure and, at most, two dielectric constants to account for all the dynamics of the protein. This clearly is an oversimplification of reality (3Nielsen J.E. Andersen K.V. Honig B. Hooft R.W.W. Klebe G. Vriend G. Wade R.C. Protein Eng. 1999; 12: 657-662Crossref PubMed Scopus (111) Google Scholar).We have studied the contribution of long range electrostatic interactions to catalysis by analyzing the effects of a series of charge mutations scattered over a larger part of the surface of a thermolysin-like protease from Bacillus stearothermophilus(TLP-ste).1 Thermolysin-like proteases (TLPs) are members of the peptidase family M4 (4Beaumont A. Beynon R.J. Handbook of Proteolytic Enzymes. Academic Press, Inc., New York1998: 350-369Google Scholar) of which thermolysin (EC 3.4.24.27) is the prototype. One of their characteristics is a zinc ion bound in the catalytic center. The amino acid sequences of several TLPs have been determined (see Ref. 4Beaumont A. Beynon R.J. Handbook of Proteolytic Enzymes. Academic Press, Inc., New York1998: 350-369Google Scholar, or the Merops data base at www.merops.co.uk/merops/famcards/m4.htm), and the three-dimensional structures of TLPs isolated from several bacteria (Bacillus thermoproteolyticus (5Weaver L.H. Kester W.R. Matthews B.W. J. Mol. Biol. 1977; 114: 119-132Crossref PubMed Scopus (183) Google Scholar), Bacillus cereus (6Stark W. Pauptit R.A. Wilson K.S. Jansonius J.N. FEBS Eur. J. Biochem. 1992; 984: 1-11Google Scholar), Pseudomonas aeruginosa (7Thayer M.M. Flaherty K.M. McKay D.B. J. Biol. Chem. 1991; 266: 2864-2871Abstract Full Text PDF PubMed Google Scholar), andStaphylococcus aureus (8Banbula A. Potempa J. Travis J. Fernandez C.C. Mann K. Huber R. Bode W. Medrano F.J. Structure. 1998; 6: 1185-1193Abstract Full Text Full Text PDF PubMed Scopus (71) Google Scholar)) have been solved. TLPs consist of an α-helical C-terminal domain and a β-rich N-terminal domain. These two domains are connected by a central α-helix, which is located at the bottom of the active site cleft and which contains several of the catalytically important residues. From x-ray structures of thermolysin·inhibitor complexes (5Weaver L.H. Kester W.R. Matthews B.W. J. Mol. Biol. 1977; 114: 119-132Crossref PubMed Scopus (183) Google Scholar, 9Gaucher J.F. Selkti M. Tiraboschi G. Roques B.P. Tomas A. Fournié-Zaluski M.-C. Biochemistry. 1999; 38: 12569-12576Crossref PubMed Scopus (48) Google Scholar, 10Hausrath A.C. Matthews B.W. J. Biol. Chem. 1994; 269: 18839-18842Abstract Full Text PDF PubMed Google Scholar, 11Holmes M.A. Tronrud D.E. Matthews B.W. Biochemistry. 1983; 22: 236-240Crossref PubMed Scopus (43) Google Scholar, 12Jin Y. Kim D.H. Bioorg. Med. Chem. Lett. 1998; 8: 3515-3518Crossref PubMed Scopus (15) Google Scholar, 13Kester W.R. Matthews B.W. Biochemistry. 1977; 16: 2506-2516Crossref PubMed Scopus (204) Google Scholar), the active site residues have been identified and a mechanism has been proposed (13Kester W.R. Matthews B.W. Biochemistry. 1977; 16: 2506-2516Crossref PubMed Scopus (204) Google Scholar, 14Hangauer D.G. Monzingo A.F. Matthews B.W. Biochemistry. 1984; 23: 5730-5741Crossref PubMed Scopus (250) Google Scholar, 15Matthews B.W. Acc. Chem. Res. 1988; 21: 333-340Crossref Scopus (668) Google Scholar). Recently, an alternative mechanism has been proposed that has gained some support (16Mock W.L. Aksamawati M. Biochem. J. 1994; 302: 57-68Crossref PubMed Scopus (69) Google Scholar, 17Mock W.L. Stanford D.J. Biochemistry. 1996; 35: 7369-7377Crossref PubMed Scopus (77) Google Scholar). In both proposed mechanisms residues Glu-143, His-231, Tyr-157, and a Zn2+ bound water play important roles during catalysis.Mutations were introduced at six surface positions in TLP-ste, located at 10–15 Å from the catalytic center. The single and multiple mutants that were obtained displayed varying effects on catalytic efficiency, including considerable increases in activity. Double-mutant cycle analysis (18Carter P.J. Winter G. Wilkinson A.J. Fersht A.R. Cell. 1984; 38: 835-840Abstract Full Text PDF PubMed Scopus (515) Google Scholar) was used to study the additivity of mutational effects, revealing remarkable interdependence of the mutated residues. The results provide insight in the complexity of predicting and interpreting electrostatic effects in catalysis.DISCUSSIONModification of the surface charge should, according to electrostatic theory, lead to a change in the active site electrostatics (1Russell A.J. Fersht A.R. Nature. 1987; 328: 496-500Crossref PubMed Scopus (299) Google Scholar). Therefore, the catalytic performance of an enzyme may be modified without structurally disturbing the active site. Here, we have engineered a considerable increase in activity by modifying surface charges in TLP-ste. The most active mutants were approximately four times more active than the wild type. It is noteworthy that this increase was achieved by mutations that are all far from the active site. The distances between the catalytically important Zn2+ and the mutated residues varies from 10 to 15 Å.Mutational effects on the activity toward tripeptide substrate were more pronounced than mutational effects on specific activity toward casein (Tables I and II). The latter effects were marginal in most cases, but two of the mutants that were most active toward the tripeptide also displayed increased specific activity toward casein (Table II). The pH optimum of these mutants was the same regardless of the substrate used, 2A. de Kreij, unpublished observations.indicating that there are no fundamental differences in the ways in which the different substrates are hydrolyzed. In most published studies on engineering protease activity, only peptide substrates were used for mutant characterization. In the few cases in which proteinaceous substrates were used (e.g. Refs. 37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar, 38de Kreij A. van Den B.B. Veltman O.R. Vriend G. Venema G. Eijsink V.G. Eur. J. Biochem. 2001; 268: 4985-4991Crossref PubMed Scopus (20) Google Scholar; and by others in Refs. 39Mei H.-C. Li Y.-C. Liaw Y.-C. Wang D.-C. Takagi H. Tsai Y.C. Protein Eng. 1998; 11: 109-117Crossref PubMed Google Scholar, 40Graycar T.P. Bott R. Caldwell R.W. Dauberman J.L. Pushkaraj J.L. Power S.D. Sagar I.H. Silva R.A. Weiss G.L. Woodhouse L.R. Estell D.A. Anna. N. Y. Acad. Sci. 1992; 673: 70-79Crossref PubMed Scopus (29) Google Scholar) one generally observed that the results differed considerably from those obtained with peptide substrates. Usually, mutational effects were much less pronounced for proteinaceous substrates, as is also the case in the present study. One possible explanation may be that the contribution of binding to catalysis is much larger for complex substrates, thus veiling effects ofe.g. subtle changes in active site electrostatics. Analysis of mutational effects on activity toward complex proteinaceous substrates is intrinsically complex, due to the large number of titratable (ionized) groups in the substrate and the fact that the substrate has many different productive binding modes. Clearly, analysis of mutational effects on active site electrostatics requires the use of short substrates that have only one productive binding mode and whose binding to the enzyme has a marginal (and constant) effect on the dielectric constant and charge in the active site.Several elements in the results indicate that the effects of surface charge mutations on catalysis result from factors that are independent of the changes in charge per se. In general, with few exceptions, the observed changes in the pH-activity profile and activity do not correlate with expected changes in the pKa values. This suggests that values other than those of Δcharge-induced ΔpKa effects, that is, effects that are not accounted for in the software used for calculating pKa values, are important.One indication for the occurrence of other than charge effects comes from the studies on additivity of mutational effects. A priori, one may expect the effects of surface charge mutations to be additive as long as the residues do not have direct interactions (2Jackson S.E. Fersht A.R. Biochemistry. 1993; 32: 13909-13916Crossref PubMed Scopus (79) Google Scholar,41Russell A.J. Thomas P.G. Fersht A.R. J. Mol. Biol. 1987; 193: 803-813Crossref PubMed Scopus (170) Google Scholar). Accordingly, we observed increased effects upon combination of several mutations. However, the mutational effects were far from additive, as shown by the double-mutant cycle analysis. This observation is not surprising for residues 116 and 119, which share a hydrogen bond in the wild type enzyme. However, it is highly surprising to find that all residues mutated in this study affect each other, that is, they seem to interact over distances as long as 25 Å. The fact that the interdependent residues are so far apart excludes the possibility of direct contacts and also makes it unlikely that charge effects alone account for the mutational effects.Another indication of the occurrence of non-charge effects is the discrepancy between the observed effects on activity, and the observed effects on the pH optimum. If the change in activity is a result of a change in active site electrostatics, then a change in the pH optimum would also be expected, at least in some cases. Although changes in the shape of the pH profile were observed, the pH optimum itself was not changed in the most active enzyme variants.The notion that complex long range interactions may affect activity without having direct effects on the pKa values of reactive groups is supported by earlier work in which replacing uncharged residues with other uncharged residues resulted in changes in activity and in the pH-activity profile that were similar to, or larger than, the changes observed after introducing or removing charges (42Nielsen J.E. Beier L. Otzen D. Borchert T.V. Frantzen H.B. Andersen K.V. Svendsen A. Eur. J. Biochem. 1999; 264: 816-824Crossref PubMed Scopus (57) Google Scholar). It has been suggested that changes in active site dynamics are in part responsible for such counter-intuitive results. Alternatively, the electrostatic network on the surface (and throughout) an enzyme may be of an unappreciated complexity that cannot yet be properly described, making the effects of a surface charge mutation unpredictable.One might argue that changes in active site dynamics would cause changes in thermal stability and the temperature optimum for activity. However, none of the mutants described in this study displayed large changes in thermal stability or temperature optimum (results not shown). The lack of effects on thermal stability is not surprising, because it is known that the thermal stability of thermolysin-like proteases is mainly determined by local unfolding in a surface-located region in the N-terminal domain (Refs. 36Vriend G. Berendsen H.J.C. van den Burg B. Venema G. Eijsink V.G.H. J. Biol. Chem. 1998; 273: 35074-35077Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar, 43Eijsink V.G.H. Veltman O.R. Aukema W. Vriend G. Venema G. Nat. Struct. Biol. 1995; 2: 374-379Crossref PubMed Scopus (100) Google Scholar, 44van den Burg B. Vriend G. Veltman O.R. Venema G. Eijsink V.G.H. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2056-2060Crossref PubMed Scopus (260) Google Scholar; see also 45Kidokoro S. Miki Y. Endo K. Wada A. Nagao H. Miyake T. Aoyama A. Yoneya T. Kai K. Ooe S. FEBS Lett. 1995; 367: 73-76Crossref PubMed Scopus (51) Google Scholar, 46Zhao H. Arnold F.H. Protein Eng. 1999; 12: 47-53Crossref PubMed Scopus (265) Google Scholar). None of the mutations described here are located in this surface region. The lack of stability effects does not necessarily mean that the mutations did not affect flexibility. We have previously made a series of Gly → Ala mutations in the enzyme that were aimed at reducing flexibility (37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar). Some of these mutations (e.g.G135A and G136A) had profound effects on activity, but the mutations (eight in total) did not significantly affect stability, apart from one that was located in the stability-determining region in the N-terminal domain (37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar). These mutations did not significantly affect the temperature optimum of the enzyme either. 3B. van den Burg, O. R. Veltman, and V. G. H. Eijsink, unpublished observations. Interestingly, there is recent evidence in the literature suggesting that an increase in catalytically relevant flexibility does not necessarily result in decreased stability (47Hernandez G. Jenney Jr., F.E. Adams M.W. LeMaster D.M. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 3166-3170Crossref PubMed Scopus (168) Google Scholar).The fact that residues 25 Å apart are coupled has serious consequences for modeling and calculation of the effects of charge mutations, because it implies that the effect of any mutation on the surface of an enzyme is dependent on the rest of the surface residues. The coupling also implies that larger parts of an enzyme are involved in optimizing its catalytic center. If this is true, then nature probably optimized considerably more than just the active site of enzymes during evolution. The large size of enzymes could be explained by the need to balance all the interactions on the surface of an enzyme and their influence on the catalytic center.Here we show that considerable increases in catalytic activity can be obtained by modification of surface charge. The most active mutant obtained was almost four times as active as wild type TLP-ste. However, the results show that this achievement cannot easily be rationalized,e.g. by correlating it to changes in the pKa values of active site residues. Factors that are not included in current electrostatic models, e.g.conformational dynamics and unknown complexities in electrostatic networks, probably play major roles. Reliable models and predictions as to how to modify the activity and pH-activity profile of an enzyme requires better understanding of these factors. The acceleration of reaction rates by enzymes is one of the essential prerequisites for life as we know it, and the multitude and diversity of enzymes shows that it should be possible to design an enzyme that will catalyze almost any reaction under almost any set of conditions. To achieve a high rate of acceleration, enzymes rely on charged groups in their active site that stabilize the transition state or function as acid or base catalysts in the reaction. The kinetic parameters of enzymes therefore display a significant pH dependence, which is determined by the pKa values of the active site groups. Catalysis depends on intricate electrostatic interactions, which may be noticeable over distances that are large compared with short range of interactions such as hydrogen bonds and hydrophobic contacts. Thus, larger parts of an enzyme may be involved in optimizing its catalytic center than previously thought. The long range character of electrostatic effects is illustrated by a, very limited, number of examples in the literature, showing that changes in surface charge at locations as far as 15 Å from a catalytic center may affect enzyme activity (1Russell A.J. Fersht A.R. Nature. 1987; 328: 496-500Crossref PubMed Scopus (299) Google Scholar, 2Jackson S.E. Fersht A.R. Biochemistry. 1993; 32: 13909-13916Crossref PubMed Scopus (79) Google Scholar). Unfortunately, electrostatic interactions are hard to handle theoretically not only because of their long range character but also because of intrinsic theoretical difficulties. For example, most electrostatic models still use a single rigid protein structure and, at most, two dielectric constants to account for all the dynamics of the protein. This clearly is an oversimplification of reality (3Nielsen J.E. Andersen K.V. Honig B. Hooft R.W.W. Klebe G. Vriend G. Wade R.C. Protein Eng. 1999; 12: 657-662Crossref PubMed Scopus (111) Google Scholar). We have studied the contribution of long range electrostatic interactions to catalysis by analyzing the effects of a series of charge mutations scattered over a larger part of the surface of a thermolysin-like protease from Bacillus stearothermophilus(TLP-ste).1 Thermolysin-like proteases (TLPs) are members of the peptidase family M4 (4Beaumont A. Beynon R.J. Handbook of Proteolytic Enzymes. Academic Press, Inc., New York1998: 350-369Google Scholar) of which thermolysin (EC 3.4.24.27) is the prototype. One of their characteristics is a zinc ion bound in the catalytic center. The amino acid sequences of several TLPs have been determined (see Ref. 4Beaumont A. Beynon R.J. Handbook of Proteolytic Enzymes. Academic Press, Inc., New York1998: 350-369Google Scholar, or the Merops data base at www.merops.co.uk/merops/famcards/m4.htm), and the three-dimensional structures of TLPs isolated from several bacteria (Bacillus thermoproteolyticus (5Weaver L.H. Kester W.R. Matthews B.W. J. Mol. Biol. 1977; 114: 119-132Crossref PubMed Scopus (183) Google Scholar), Bacillus cereus (6Stark W. Pauptit R.A. Wilson K.S. Jansonius J.N. FEBS Eur. J. Biochem. 1992; 984: 1-11Google Scholar), Pseudomonas aeruginosa (7Thayer M.M. Flaherty K.M. McKay D.B. J. Biol. Chem. 1991; 266: 2864-2871Abstract Full Text PDF PubMed Google Scholar), andStaphylococcus aureus (8Banbula A. Potempa J. Travis J. Fernandez C.C. Mann K. Huber R. Bode W. Medrano F.J. Structure. 1998; 6: 1185-1193Abstract Full Text Full Text PDF PubMed Scopus (71) Google Scholar)) have been solved. TLPs consist of an α-helical C-terminal domain and a β-rich N-terminal domain. These two domains are connected by a central α-helix, which is located at the bottom of the active site cleft and which contains several of the catalytically important residues. From x-ray structures of thermolysin·inhibitor complexes (5Weaver L.H. Kester W.R. Matthews B.W. J. Mol. Biol. 1977; 114: 119-132Crossref PubMed Scopus (183) Google Scholar, 9Gaucher J.F. Selkti M. Tiraboschi G. Roques B.P. Tomas A. Fournié-Zaluski M.-C. Biochemistry. 1999; 38: 12569-12576Crossref PubMed Scopus (48) Google Scholar, 10Hausrath A.C. Matthews B.W. J. Biol. Chem. 1994; 269: 18839-18842Abstract Full Text PDF PubMed Google Scholar, 11Holmes M.A. Tronrud D.E. Matthews B.W. Biochemistry. 1983; 22: 236-240Crossref PubMed Scopus (43) Google Scholar, 12Jin Y. Kim D.H. Bioorg. Med. Chem. Lett. 1998; 8: 3515-3518Crossref PubMed Scopus (15) Google Scholar, 13Kester W.R. Matthews B.W. Biochemistry. 1977; 16: 2506-2516Crossref PubMed Scopus (204) Google Scholar), the active site residues have been identified and a mechanism has been proposed (13Kester W.R. Matthews B.W. Biochemistry. 1977; 16: 2506-2516Crossref PubMed Scopus (204) Google Scholar, 14Hangauer D.G. Monzingo A.F. Matthews B.W. Biochemistry. 1984; 23: 5730-5741Crossref PubMed Scopus (250) Google Scholar, 15Matthews B.W. Acc. Chem. Res. 1988; 21: 333-340Crossref Scopus (668) Google Scholar). Recently, an alternative mechanism has been proposed that has gained some support (16Mock W.L. Aksamawati M. Biochem. J. 1994; 302: 57-68Crossref PubMed Scopus (69) Google Scholar, 17Mock W.L. Stanford D.J. Biochemistry. 1996; 35: 7369-7377Crossref PubMed Scopus (77) Google Scholar). In both proposed mechanisms residues Glu-143, His-231, Tyr-157, and a Zn2+ bound water play important roles during catalysis. Mutations were introduced at six surface positions in TLP-ste, located at 10–15 Å from the catalytic center. The single and multiple mutants that were obtained displayed varying effects on catalytic efficiency, including considerable increases in activity. Double-mutant cycle analysis (18Carter P.J. Winter G. Wilkinson A.J. Fersht A.R. Cell. 1984; 38: 835-840Abstract Full Text PDF PubMed Scopus (515) Google Scholar) was used to study the additivity of mutational effects, revealing remarkable interdependence of the mutated residues. The results provide insight in the complexity of predicting and interpreting electrostatic effects in catalysis. DISCUSSIONModification of the surface charge should, according to electrostatic theory, lead to a change in the active site electrostatics (1Russell A.J. Fersht A.R. Nature. 1987; 328: 496-500Crossref PubMed Scopus (299) Google Scholar). Therefore, the catalytic performance of an enzyme may be modified without structurally disturbing the active site. Here, we have engineered a considerable increase in activity by modifying surface charges in TLP-ste. The most active mutants were approximately four times more active than the wild type. It is noteworthy that this increase was achieved by mutations that are all far from the active site. The distances between the catalytically important Zn2+ and the mutated residues varies from 10 to 15 Å.Mutational effects on the activity toward tripeptide substrate were more pronounced than mutational effects on specific activity toward casein (Tables I and II). The latter effects were marginal in most cases, but two of the mutants that were most active toward the tripeptide also displayed increased specific activity toward casein (Table II). The pH optimum of these mutants was the same regardless of the substrate used, 2A. de Kreij, unpublished observations.indicating that there are no fundamental differences in the ways in which the different substrates are hydrolyzed. In most published studies on engineering protease activity, only peptide substrates were used for mutant characterization. In the few cases in which proteinaceous substrates were used (e.g. Refs. 37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar, 38de Kreij A. van Den B.B. Veltman O.R. Vriend G. Venema G. Eijsink V.G. Eur. J. Biochem. 2001; 268: 4985-4991Crossref PubMed Scopus (20) Google Scholar; and by others in Refs. 39Mei H.-C. Li Y.-C. Liaw Y.-C. Wang D.-C. Takagi H. Tsai Y.C. Protein Eng. 1998; 11: 109-117Crossref PubMed Google Scholar, 40Graycar T.P. Bott R. Caldwell R.W. Dauberman J.L. Pushkaraj J.L. Power S.D. Sagar I.H. Silva R.A. Weiss G.L. Woodhouse L.R. Estell D.A. Anna. N. Y. Acad. Sci. 1992; 673: 70-79Crossref PubMed Scopus (29) Google Scholar) one generally observed that the results differed considerably from those obtained with peptide substrates. Usually, mutational effects were much less pronounced for proteinaceous substrates, as is also the case in the present study. One possible explanation may be that the contribution of binding to catalysis is much larger for complex substrates, thus veiling effects ofe.g. subtle changes in active site electrostatics. Analysis of mutational effects on activity toward complex proteinaceous substrates is intrinsically complex, due to the large number of titratable (ionized) groups in the substrate and the fact that the substrate has many different productive binding modes. Clearly, analysis of mutational effects on active site electrostatics requires the use of short substrates that have only one productive binding mode and whose binding to the enzyme has a marginal (and constant) effect on the dielectric constant and charge in the active site.Several elements in the results indicate that the effects of surface charge mutations on catalysis result from factors that are independent of the changes in charge per se. In general, with few exceptions, the observed changes in the pH-activity profile and activity do not correlate with expected changes in the pKa values. This suggests that values other than those of Δcharge-induced ΔpKa effects, that is, effects that are not accounted for in the software used for calculating pKa values, are important.One indication for the occurrence of other than charge effects comes from the studies on additivity of mutational effects. A priori, one may expect the effects of surface charge mutations to be additive as long as the residues do not have direct interactions (2Jackson S.E. Fersht A.R. Biochemistry. 1993; 32: 13909-13916Crossref PubMed Scopus (79) Google Scholar,41Russell A.J. Thomas P.G. Fersht A.R. J. Mol. Biol. 1987; 193: 803-813Crossref PubMed Scopus (170) Google Scholar). Accordingly, we observed increased effects upon combination of several mutations. However, the mutational effects were far from additive, as shown by the double-mutant cycle analysis. This observation is not surprising for residues 116 and 119, which share a hydrogen bond in the wild type enzyme. However, it is highly surprising to find that all residues mutated in this study affect each other, that is, they seem to interact over distances as long as 25 Å. The fact that the interdependent residues are so far apart excludes the possibility of direct contacts and also makes it unlikely that charge effects alone account for the mutational effects.Another indication of the occurrence of non-charge effects is the discrepancy between the observed effects on activity, and the observed effects on the pH optimum. If the change in activity is a result of a change in active site electrostatics, then a change in the pH optimum would also be expected, at least in some cases. Although changes in the shape of the pH profile were observed, the pH optimum itself was not changed in the most active enzyme variants.The notion that complex long range interactions may affect activity without having direct effects on the pKa values of reactive groups is supported by earlier work in which replacing uncharged residues with other uncharged residues resulted in changes in activity and in the pH-activity profile that were similar to, or larger than, the changes observed after introducing or removing charges (42Nielsen J.E. Beier L. Otzen D. Borchert T.V. Frantzen H.B. Andersen K.V. Svendsen A. Eur. J. Biochem. 1999; 264: 816-824Crossref PubMed Scopus (57) Google Scholar). It has been suggested that changes in active site dynamics are in part responsible for such counter-intuitive results. Alternatively, the electrostatic network on the surface (and throughout) an enzyme may be of an unappreciated complexity that cannot yet be properly described, making the effects of a surface charge mutation unpredictable.One might argue that changes in active site dynamics would cause changes in thermal stability and the temperature optimum for activity. However, none of the mutants described in this study displayed large changes in thermal stability or temperature optimum (results not shown). The lack of effects on thermal stability is not surprising, because it is known that the thermal stability of thermolysin-like proteases is mainly determined by local unfolding in a surface-located region in the N-terminal domain (Refs. 36Vriend G. Berendsen H.J.C. van den Burg B. Venema G. Eijsink V.G.H. J. Biol. Chem. 1998; 273: 35074-35077Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar, 43Eijsink V.G.H. Veltman O.R. Aukema W. Vriend G. Venema G. Nat. Struct. Biol. 1995; 2: 374-379Crossref PubMed Scopus (100) Google Scholar, 44van den Burg B. Vriend G. Veltman O.R. Venema G. Eijsink V.G.H. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2056-2060Crossref PubMed Scopus (260) Google Scholar; see also 45Kidokoro S. Miki Y. Endo K. Wada A. Nagao H. Miyake T. Aoyama A. Yoneya T. Kai K. Ooe S. FEBS Lett. 1995; 367: 73-76Crossref PubMed Scopus (51) Google Scholar, 46Zhao H. Arnold F.H. Protein Eng. 1999; 12: 47-53Crossref PubMed Scopus (265) Google Scholar). None of the mutations described here are located in this surface region. The lack of stability effects does not necessarily mean that the mutations did not affect flexibility. We have previously made a series of Gly → Ala mutations in the enzyme that were aimed at reducing flexibility (37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar). Some of these mutations (e.g.G135A and G136A) had profound effects on activity, but the mutations (eight in total) did not significantly affect stability, apart from one that was located in the stability-determining region in the N-terminal domain (37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar). These mutations did not significantly affect the temperature optimum of the enzyme either. 3B. van den Burg, O. R. Veltman, and V. G. H. Eijsink, unpublished observations. Interestingly, there is recent evidence in the literature suggesting that an increase in catalytically relevant flexibility does not necessarily result in decreased stability (47Hernandez G. Jenney Jr., F.E. Adams M.W. LeMaster D.M. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 3166-3170Crossref PubMed Scopus (168) Google Scholar).The fact that residues 25 Å apart are coupled has serious consequences for modeling and calculation of the effects of charge mutations, because it implies that the effect of any mutation on the surface of an enzyme is dependent on the rest of the surface residues. The coupling also implies that larger parts of an enzyme are involved in optimizing its catalytic center. If this is true, then nature probably optimized considerably more than just the active site of enzymes during evolution. The large size of enzymes could be explained by the need to balance all the interactions on the surface of an enzyme and their influence on the catalytic center.Here we show that considerable increases in catalytic activity can be obtained by modification of surface charge. The most active mutant obtained was almost four times as active as wild type TLP-ste. However, the results show that this achievement cannot easily be rationalized,e.g. by correlating it to changes in the pKa values of active site residues. Factors that are not included in current electrostatic models, e.g.conformational dynamics and unknown complexities in electrostatic networks, probably play major roles. Reliable models and predictions as to how to modify the activity and pH-activity profile of an enzyme requires better understanding of these factors. Modification of the surface charge should, according to electrostatic theory, lead to a change in the active site electrostatics (1Russell A.J. Fersht A.R. Nature. 1987; 328: 496-500Crossref PubMed Scopus (299) Google Scholar). Therefore, the catalytic performance of an enzyme may be modified without structurally disturbing the active site. Here, we have engineered a considerable increase in activity by modifying surface charges in TLP-ste. The most active mutants were approximately four times more active than the wild type. It is noteworthy that this increase was achieved by mutations that are all far from the active site. The distances between the catalytically important Zn2+ and the mutated residues varies from 10 to 15 Å. Mutational effects on the activity toward tripeptide substrate were more pronounced than mutational effects on specific activity toward casein (Tables I and II). The latter effects were marginal in most cases, but two of the mutants that were most active toward the tripeptide also displayed increased specific activity toward casein (Table II). The pH optimum of these mutants was the same regardless of the substrate used, 2A. de Kreij, unpublished observations.indicating that there are no fundamental differences in the ways in which the different substrates are hydrolyzed. In most published studies on engineering protease activity, only peptide substrates were used for mutant characterization. In the few cases in which proteinaceous substrates were used (e.g. Refs. 37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar, 38de Kreij A. van Den B.B. Veltman O.R. Vriend G. Venema G. Eijsink V.G. Eur. J. Biochem. 2001; 268: 4985-4991Crossref PubMed Scopus (20) Google Scholar; and by others in Refs. 39Mei H.-C. Li Y.-C. Liaw Y.-C. Wang D.-C. Takagi H. Tsai Y.C. Protein Eng. 1998; 11: 109-117Crossref PubMed Google Scholar, 40Graycar T.P. Bott R. Caldwell R.W. Dauberman J.L. Pushkaraj J.L. Power S.D. Sagar I.H. Silva R.A. Weiss G.L. Woodhouse L.R. Estell D.A. Anna. N. Y. Acad. Sci. 1992; 673: 70-79Crossref PubMed Scopus (29) Google Scholar) one generally observed that the results differed considerably from those obtained with peptide substrates. Usually, mutational effects were much less pronounced for proteinaceous substrates, as is also the case in the present study. One possible explanation may be that the contribution of binding to catalysis is much larger for complex substrates, thus veiling effects ofe.g. subtle changes in active site electrostatics. Analysis of mutational effects on activity toward complex proteinaceous substrates is intrinsically complex, due to the large number of titratable (ionized) groups in the substrate and the fact that the substrate has many different productive binding modes. Clearly, analysis of mutational effects on active site electrostatics requires the use of short substrates that have only one productive binding mode and whose binding to the enzyme has a marginal (and constant) effect on the dielectric constant and charge in the active site. Several elements in the results indicate that the effects of surface charge mutations on catalysis result from factors that are independent of the changes in charge per se. In general, with few exceptions, the observed changes in the pH-activity profile and activity do not correlate with expected changes in the pKa values. This suggests that values other than those of Δcharge-induced ΔpKa effects, that is, effects that are not accounted for in the software used for calculating pKa values, are important. One indication for the occurrence of other than charge effects comes from the studies on additivity of mutational effects. A priori, one may expect the effects of surface charge mutations to be additive as long as the residues do not have direct interactions (2Jackson S.E. Fersht A.R. Biochemistry. 1993; 32: 13909-13916Crossref PubMed Scopus (79) Google Scholar,41Russell A.J. Thomas P.G. Fersht A.R. J. Mol. Biol. 1987; 193: 803-813Crossref PubMed Scopus (170) Google Scholar). Accordingly, we observed increased effects upon combination of several mutations. However, the mutational effects were far from additive, as shown by the double-mutant cycle analysis. This observation is not surprising for residues 116 and 119, which share a hydrogen bond in the wild type enzyme. However, it is highly surprising to find that all residues mutated in this study affect each other, that is, they seem to interact over distances as long as 25 Å. The fact that the interdependent residues are so far apart excludes the possibility of direct contacts and also makes it unlikely that charge effects alone account for the mutational effects. Another indication of the occurrence of non-charge effects is the discrepancy between the observed effects on activity, and the observed effects on the pH optimum. If the change in activity is a result of a change in active site electrostatics, then a change in the pH optimum would also be expected, at least in some cases. Although changes in the shape of the pH profile were observed, the pH optimum itself was not changed in the most active enzyme variants. The notion that complex long range interactions may affect activity without having direct effects on the pKa values of reactive groups is supported by earlier work in which replacing uncharged residues with other uncharged residues resulted in changes in activity and in the pH-activity profile that were similar to, or larger than, the changes observed after introducing or removing charges (42Nielsen J.E. Beier L. Otzen D. Borchert T.V. Frantzen H.B. Andersen K.V. Svendsen A. Eur. J. Biochem. 1999; 264: 816-824Crossref PubMed Scopus (57) Google Scholar). It has been suggested that changes in active site dynamics are in part responsible for such counter-intuitive results. Alternatively, the electrostatic network on the surface (and throughout) an enzyme may be of an unappreciated complexity that cannot yet be properly described, making the effects of a surface charge mutation unpredictable. One might argue that changes in active site dynamics would cause changes in thermal stability and the temperature optimum for activity. However, none of the mutants described in this study displayed large changes in thermal stability or temperature optimum (results not shown). The lack of effects on thermal stability is not surprising, because it is known that the thermal stability of thermolysin-like proteases is mainly determined by local unfolding in a surface-located region in the N-terminal domain (Refs. 36Vriend G. Berendsen H.J.C. van den Burg B. Venema G. Eijsink V.G.H. J. Biol. Chem. 1998; 273: 35074-35077Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar, 43Eijsink V.G.H. Veltman O.R. Aukema W. Vriend G. Venema G. Nat. Struct. Biol. 1995; 2: 374-379Crossref PubMed Scopus (100) Google Scholar, 44van den Burg B. Vriend G. Veltman O.R. Venema G. Eijsink V.G.H. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2056-2060Crossref PubMed Scopus (260) Google Scholar; see also 45Kidokoro S. Miki Y. Endo K. Wada A. Nagao H. Miyake T. Aoyama A. Yoneya T. Kai K. Ooe S. FEBS Lett. 1995; 367: 73-76Crossref PubMed Scopus (51) Google Scholar, 46Zhao H. Arnold F.H. Protein Eng. 1999; 12: 47-53Crossref PubMed Scopus (265) Google Scholar). None of the mutations described here are located in this surface region. The lack of stability effects does not necessarily mean that the mutations did not affect flexibility. We have previously made a series of Gly → Ala mutations in the enzyme that were aimed at reducing flexibility (37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar). Some of these mutations (e.g.G135A and G136A) had profound effects on activity, but the mutations (eight in total) did not significantly affect stability, apart from one that was located in the stability-determining region in the N-terminal domain (37Veltman O.R. Eijsink V.G.H. Vriend G. de Kreij A. Venema G. van den Burg B. Biochemistry. 1998; 37: 5305-5311Crossref PubMed Scopus (27) Google Scholar). These mutations did not significantly affect the temperature optimum of the enzyme either. 3B. van den Burg, O. R. Veltman, and V. G. H. Eijsink, unpublished observations. Interestingly, there is recent evidence in the literature suggesting that an increase in catalytically relevant flexibility does not necessarily result in decreased stability (47Hernandez G. Jenney Jr., F.E. Adams M.W. LeMaster D.M. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 3166-3170Crossref PubMed Scopus (168) Google Scholar). The fact that residues 25 Å apart are coupled has serious consequences for modeling and calculation of the effects of charge mutations, because it implies that the effect of any mutation on the surface of an enzyme is dependent on the rest of the surface residues. The coupling also implies that larger parts of an enzyme are involved in optimizing its catalytic center. If this is true, then nature probably optimized considerably more than just the active site of enzymes during evolution. The large size of enzymes could be explained by the need to balance all the interactions on the surface of an enzyme and their influence on the catalytic center. Here we show that considerable increases in catalytic activity can be obtained by modification of surface charge. The most active mutant obtained was almost four times as active as wild type TLP-ste. However, the results show that this achievement cannot easily be rationalized,e.g. by correlating it to changes in the pKa values of active site residues. Factors that are not included in current electrostatic models, e.g.conformational dynamics and unknown complexities in electrostatic networks, probably play major roles. Reliable models and predictions as to how to modify the activity and pH-activity profile of an enzyme requires better understanding of these factors.