Exploring Nonnatural Evolutionary Pathways by Saturation Mutagenesis: Rapid Improvement of Protein Function

Random point mutagenesis does not access a large fraction of protein sequence space corresponding to primarily nonconservative amino acid substitutions. The cost of this limitation during directed evolution is unknown. Random point mutagenesis over the entire gene encoding the psychrophilic protease subtilisin S41 identified a pair of residues (Lys211 and Arg212) where mutations provided significant increases in thermostability. These were subjected to saturation mutagenesis to test whether the amino acids not easily accessible by point mutagenesis provide even better ``solutions' to the thermostabilization challenge. A significant fraction of these variants surpassed the stability of the variants with point mutations. DNA sequencing revealed highly hydrophobic residues in the four most stable variants (Pro/Ala, Pro/Val, Leu/Val, and Trp/Ser). These nonconservative replacements, accessible only by multiple (two to three) base substitutions in a single codon, would be extremely rare in a point mutation library. Such replacements are also extremely rare in natural evolution. Saturation mutagenesis may be used advantageously during directed evolution to explore nonnatural evolution pathways and enable rapid improvement in protein traits. Received: 15 March 1999 / Accepted: 28 June 1999     

Hydrophobic interactions between the secondary structures on the molecular surface reinforce the alkaline stability of serine protease

We employed random mutagenesis to determine the region of the initial unfolding of hyper-alkaline-sensitive subtilisin, ALP I, that precedes the denaturation of the entire protein under highly alkaline conditions. This region comprises two α-helices and a calcium-binding loop. Stabilization of the region caused the stabilization of the entire protein at a high alkaline pH 12. The alkaline stability of this region was most effectively improved by hydrophobic interactions, followed by ionic interactions with Arg residues. The effect of mutations on the improvement was different with regard to the alkaline stability and thermostability. This indicated that different strategies were necessary to improve the alkaline stability and thermostability of the protein.     

Improvement in the alkaline stability of subtilisin using an efficient random mutagenesis and screening procedure

An efficient random mutagenesis procedure coupled to a replica plate screen facilitated the isolation of mutant subtilisins from Bacillus amyloliquefaciens that had altered autolytic stability under alkaline conditions. Out of about 4000 clones screened, approximately 70 produced subtilisins with reduced stability (negatives). Two clones produced a more stable subtilisin (positives) and were identified as having a single mutation, either Ile107Val or Lys213Arg (the wild-type amino acid is followed by the codon position and the mutant amino acid). One of the negative mutants, Met50Val, was at a site where other homologous subtilisins contained a Phe. When the Met50Phe mutation was introduced into the B. amyloliquefaciens gene, the mutant subtilisin was more alkaline stable. The double mutant (Ile107Val/Lys213Arg) was more stable than the isolated single mutant parents. The triple mutant (Met50Phe/Ile107Val/Lys213Arg) was even more stable than Ile107Val/Lys213Arg (up to two times the autolytic half-time of wild-type at pH 12). These studies demonstrate the feasibility for improving the alkaline stability of proteins by random mutagenesis and identifying potential sites where substitutions from homologous proteins can improve alkaline stability.     

Assignment of histidine resonances in the 1H NMR (500 MHz) spectrum of subtilisin BPN' using site-directed mutagenesis

A spin-echo pulse sequence has been used to resolve the six histidine C-2H protons in the 500-MHz NMR spectrum of subtilisin BPN'. Five of these residues have been substituted by site-directed mutagenesis, and this has enabled a complete assignment of these protons to be obtained. Analysis of the pH titration curves of these signals has provided microscopic pKas for the six histidines in this enzyme. The pKas of the histidine residues in subtilisin BPN' have been compared with the values obtained for the histidines in the homologous enzyme from Bacillus licheniformis (subtilisin Carlsberg). Four of the five conserved histidines titrate with essentially identical pKa's in the two enzymes. It therefore appears that the assignments made for these residues in subtilisin BPN' can be transferred to subtilisin Carlsberg. On the basis of these assignments, the one histidine that titrates with a substantially different pKa in the two enzymes can be assigned to histidine-238. This difference in pKa has been attributed to a Trp to Lys substitution at position 241 in subtilisin Carlsberg.     

The maintenance of high affinity plasminogen binding by group A streptococcal plasminogen-binding M-like protein is mediated by arginine and histidine residues within the a1 and a2 repeat domains

Subversion of the plasminogen activation system is implicated in the virulence of group A streptococci (GAS). GAS displays receptors for the human zymogen plasminogen on the cell surface, one of which is the plasminogen-binding group A streptococcal M-like protein (PAM). The plasminogen binding domain of PAM is highly variable, and this variation has been linked to host selective immune pressure. Site-directed mutagenesis of full-length PAM protein from an invasive GAS isolate was undertaken to assess the contribution of residues in the a1 and a2 repeat domains to plasminogen binding function. Mutagenesis to alanine of key plasminogen binding lysine residues in the a1 and a2 repeats (Lys98 and Lys111) did not abrogate plasminogen binding by PAM nor did additional mutagenesis of Arg101 and His102 and Glu104, which have previously been implicated in plasminogen binding. Plasminogen binding was only abolished with the additional mutagenesis of Arg114 and His115 to alanine. Furthermore, mutagenesis of both arginine (Arg101 and Arg114) and histidine (His102 and His115) residues abolished interaction with plasminogen despite the presence of Lys98 and Lys111 in the binding repeats. This study shows for the first time that residues Arg101, Arg114, His102, and His115 in both the a1 and a2 repeat domains of PAM can mediate high affinity plasminogen binding. These data suggest that highly conserved arginine and histidine residues may compensate for variation elsewhere in the a1 and a2 plasminogen binding repeats, and may explain the maintenance of high affinity plasminogen binding by naturally occurring variants of PAM.     

Contribution of a Salt Bridge Triad to the Thermostability of a Highly Alkaline Protease from an Alkaliphilic Bacillus Strain*

Summary Crystallographic analysis of the highly alkaline M-protease from an alkaliphilic Bacillus strain shows the occurrence of a unique salt bridge triad Arg19–Glu271–Arg275 (in subtilisin BPN′ numbering), which is not found in less alkaline true subtilisins BPN′ and Carlsberg from Bacillus amyloliquefaciens and Bacillus licheniformis, respectively. Because the corresponding residues are all Gln residue in the subtilisin BPN′, Gln residue was engineered into the position(s) 19, 271 and/or 275 in M-protease by site-directed mutagenesis. Disruptions of the salt bridge caused the reduction of the thermostability of the mutant proteins at alkaline pH with the following decreasing order of thermal inactivation rate; the wild-type > Arg275 → Gln > Glu271 → Gln > Arg19 → Gln/Glu271 → Gln/Arg275 → Gln > Arg19 → Gln. This result provides the evidence that the salt bridge triad contributes to the thermostability and structural rigidity of the highly alkaline M-protease.     

Recombinant chymotrypsin inhibitor 2: expression, kinetic analysis of inhibition with alpha-chymotrypsin and wild-type and mutant subtilisin BPN', and protein engineering to investigate inhibitory specificity and mechanism

The serine protease inhibitor chymotrypsin inhibitor 2 (CI2 or BSPI2) has been expressed in Escherichia coli with the pINIIIompA3 expression vector to produce 20-40 mg/L of culture. Recombinant CI2 purified from this system has been characterized and found to be identical with CI2 from barley. Slow-binding kinetics were observed for the interaction between CI2 and subtilisin BPN', with Ki = 2.9 x 10(-12) M. Analysis of slow-binding data indicates that binding of the inhibitor follows the simplest model of E + I = EI with no kinetically detectable intermediate steps or proteolytic cleavage of the reactive site bond in CI2 (Met-59-Glu-60). This, in agreement with crystallographic data, indicates that the enzyme-inhibitor adduct is the Michaelis complex, which is not chemically processed by the enzyme. Three mutant CI2 molecules with new P1 residues have also been examined with a range of serine proteases, including a mutant subtilisin. In agreement with earlier studies, we find the P1 amino acid an important determinant of specificity. CI2 Met----Lys-59 was found to be a temporary inhibitor of subtilisin BPN' but an effective inhibitor of subtilisin Carlsberg and subtilisin BPN'(Glu----Ser-156). The structural reasons for this are discussed in relation to mechanisms of inhibition of serine proteases.     

Modified Proteases for Peptide Synthesis in Organic Media

We showed that modified proteases could catalyze synthesis of a wide variety of peptides of various lengths and structures both in solution and on solid phase in organic solvents. The following modified proteases were studied as catalysts for enzymatic peptide synthesis in polar organic solvents (acetonitrile, dimethylformamide, and ethanol): pepsin sorbed on celite, a noncovalent complex of subtilisin with sodium dodecylsulfate, and subtilisin or thermolysin covalently immobilized on a cryogel of polyvinyl alcohol. The use of the noncovalent complex of subtilisin with sodium dodecylsulfate and immobilized subtilisin is especially promising for the segment condensation of peptide fragments containing residues of trifunctional amino acids with unprotected ionogenic groups in side chains, such as Lys, Arg, His, Glu, and Asp.     

Modified proteinases in peptide synthesis in organic media

We showed that modified proteases could catalyze synthesis of a wide variety of peptides of various lengths and structures both in solution and on solid phase in organic solvents. The following modified proteases were studied as catalysts for enzymatic peptide synthesis in polar organic solvents (acetonitrile, dimethylformamide, and ethanol): pepsin sorbed on celite, a noncovalent complex of subtilisin with sodium dodecylsulfate, and subtilisin or thermolysin covalently immobilized on a cryogel of polyvinyl alcohol. The use of the noncovalent complex of subtilisin with sodium dodecylsulfate and immobilized subtilisin is especially promising for the segment condensation of peptide fragments containing residues of trifunctional amino acids with unprotected ionogenic groups in side chains, such as Lys, Arg, His, Glu, and Asp.     

Binding, proteolytic, and crystallographic analyses of mutations at the protease-inhibitor interface of the subtilisin BPN'/chymotrypsin inhibitor 2 complex

A series of mutants of chymotrypsin inhibitor 2 (CI2), at residues that interact with the inhibited enzyme subtilisin BPN', were studied to determine the relative importance of intermolecular contacts on either side of the scissile bond. Mutants were tested for inhibition of subtilisin, rates of hydrolysis by subtilisin, and ability to acylate subtilisin. Additionally, crystal structures of the mutant CI2 complexes with subtilisin were obtained. Ordered water molecules were found to play an important role in inhibitor recognition, and features of the crystal structures, in combination with biochemical data, support a transition-state stabilization role for the P(1) residue in subtilisin catalysis. Consistent with the proposed mechanism of inhibition, in which rapid acylation is followed by religation, leaving-group contacts with the enzyme were found to be more critical determinants of inhibition than acylating-group contacts in the mutants studied here.     

Effect of Glu-143 and His-231 substitutions on the catalytic activity and secretion of Bacillus subtilis neutral protease

On the basis of the homology with the Bacillus thermoproteolyticus zinc endopeptidase thermolysin, we hypothesized that Glu-143 and His-231 are the key residues for the catalytic activity of the Bacillus subtilis neutral protease. To test this possibility by site-directed mutagenesis, we substituted these two residues with Ala, Ser, Trp and Arg, and Leu, Val and Cys respectively. All these substitutions dramatically affected the amount of secreted mutant proteins, as determined by immunological methods, and their catalytic activities. No appreciable secretion was observed with the three Glu mutants Trp, Ser and Arg, whereas the Glu----Ala mutant enzyme was secreted at a level of a few hundred micrograms per litre of culture. The His mutants were all secreted at higher levels (in the order of a few milligrams per litre) and their residual catalytic activity could be determined using Z-Ala-Leu-Ala as substrate. Our results confirm the key role played by Glu-143 and His-231 in catalysis and moreover suggest the existence of a relationship between the catalytic activity of the enzyme and the extent of its secretion. In this context, we present data suggesting an autoproteolytic mechanism of cleavage of the precursor form of the enzyme, analogous to the one previously reported for the B. subtilis subtilisin.     

Improved autoprocessing efficiency of mutant subtilisins E with altered specificity by engineering of the pro-region.

Modification of substrate specificity of an autoprocessing enzyme is accompanied by a risk of significant failure of self-cleavage of the pro-region essential for activation. Therefore, to enhance processing, we engineered the pro-region of mutant subtilisins E of Bacillus subtilis with altered substrate specificity. A high-activity mutant subtilisin E with Ile31Leu replacement (I31L) as well as the wild-type enzyme show poor recognition of acid residues as the P1 substrate. To increase the P1 substrate preference for acid residues, Glu156Gln and Gly166Lys/Arg substitutions were introduced into the I31L gene based upon a report on subtilisin BPN' [Wells et al. (1987) Proc. Natl. Acad. Sci. USA 84, 1219-1223]. The apparent P1 specificity of four mutants (E156Q/G166K, E156Q/G166R, G166K, and G166R) was extended to acid residues, but the halo-forming activity of Escherichia coli expressing the mutant genes on skim milk-containing plates was significantly decreased due to the lower autoprocessing efficiency. A marked increase in active enzyme production occurred when Tyr(-1) in the pro-region of these mutants was then replaced by Asp or Glu. Five mutants with Glu(-2)Ala/Val/Gly or Tyr(-1)Cys/Ser substitution showing enhanced halo-forming activity were further isolated by PCR random mutagenesis in the pro-region of the E156Q/G166K mutant. These results indicated that introduction of an optimum arrangement at the cleavage site in the pro-region is an effective method for obtaining a higher yield of active enzymes.     

Evidence that both arginine 102 and arginine 747 are involved in substrate binding to neutral endopeptidase (EC 3.4.24.11)

Neutral endopeptidase (EC 3.4.24.11, NEP) is a Zn-metallopeptidase involved in the degradation of biologically active peptides, notably the enkephalins and atrial natriuretic peptide. Recently, the structure of the active site of this enzyme has been probed by site-directed mutagenesis, and 4 amino acid residues have been identified, namely 2 histidines (His583 and His587), which act as zinc-binding ligands, a glutamate (Glu584) involved in catalysis, and an arginine residue (Arg102), suggested to participate in substrate binding. Site-directed mutagenesis has now been used to investigate the role of 4 other arginine residues (Arg408, Arg409, Arg659, and Arg747) that have been proposed as possible active site residues and to further analyze the role of Arg102. In each case, the arginine was replaced with a methionine, and both enzymatic activity and the IC50 values of several NEP inhibitors were measured for the mutated enzymes and compared to wild-type enzyme. The results suggest that 2 arginines, Arg102 and Arg747, could both be important for substrate and inhibitor binding. Arg747 seems to be positioned to interact with the carbonyl amide group of the P'1 residue and can be modified when the enzyme is treated with the arginine-specific reagents phenylglyoxal and butanedione. Arg102 could be positioned to interact with the free carboxyl group of a P'2 residue in some substrates and inhibitors and can be modified by phenylglyoxal but not by butanedione. The results could explain the dual dipeptidylcarboxypeptidase and endopeptidase nature of NEP.     

Identification of basic residues in the heparin-binding exosite of factor Xa critical for heparin and factor Va binding

We recently demonstrated that a template mechanism makes a significant contribution to the heparin-accelerated inactivation of factor Xa (FXa) by antithrombin at physiologic Ca(2+), suggesting that FXa has a potential heparin-binding site. Structural data indicate that 7 of the 11 basic residues of the heparin-binding exosite of thrombin are conserved at similar three-dimensional locations in FXa. These residues, Arg(93), Lys(96), Arg(125), Arg(165), Lys(169), Lys(236), and Arg(240) were substituted with Ala in separate constructs in Gla domainless forms. It was found that all derivatives cleave Spectrozyme FXa with similar catalytic efficiencies. Antithrombin inactivated FXa derivatives with a similar second-order association rate constant (k(2)) in both the absence and presence of pentasaccharide. In the presence of heparin, however, k(2) with certain mutants were impaired up to 25-fold. Moreover, these mutants bound to heparin-Sepharose with lower affinities. Heparin concentration dependence of the inactivation revealed that only the template portion of the cofactor effect of heparin was affected by the mutagenesis. The order of importance of these residues for binding heparin was as follows: Arg(240) > Lys(236) > Lys(169) > Arg(165) > Lys(96) > Arg(93) >/= Arg(125). Interestingly, further study suggested that certain basic residues of this site, particularly Arg(165) and Lys(169), play key roles in factor Va and/or prothrombin recognition by FXa in prothrombinase.     

Role of the intramolecular hydrogen bond network in the inhibitory power of chymotrypsin inhibitor 2

A series of mutants of chymotrypsin inhibitor 2 (CI2), at residues involved in intramolecular interactions that shape and constrain the binding loop, were studied to determine their relative importance for inhibition of the serine protease subtilisin BPN', and for resistance of the inhibitor to proteolysis. These functional properties were investigated in tandem with the crystal structures of the mutant inhibitor-enzyme complexes. A dense hydrogen bonding network that supports the binding loop in the vicinity of the scissile bond was found to be important both for enzyme affinity and for stability to proteolysis. Structural analysis, in combination with biochemical measurements, allows differentiation of the structural components most important for resistance to proteolysis and/or binding. The most critical participating residues in the network were found to be Thr-58, Glu-60, Arg-65, and Gly-83. Glu-60 is more important for resistance to proteolysis than for binding, while Arg-65 and two other Arg residues play a greater role in binding than in resistance to proteolysis. Structural comparisons reveal a wide variety of subtle conformational changes in response to mutation, with built-in robustness in the hydrogen bond network, such that loss of one contact is compensated by other new contacts.     

Interaction of subtilisin BPN' and recombinant fungal protease inhibitor F from silkworm with substituted P1 site residues

Fungal protease inhibitor F (FPI-F) from silkworm inhibits subtilisin and fungal proteases. FPI-F mutants P(1) residues of which, Thr(29), were replaced with Glu, Phe, Gly, Leu, Met, and Arg, were prepared. The inhibitory activities of mutated FPI-F against subtilisin and other mammalian proteases indicated that FPI-F might be a specific inhibitor toward subtilisin-type protease.     

Enhancement of the thermostability of subtilisin E by introduction of a disulfide bond engineered on the basis of structural comparison with a thermophilic serine protease

Sites for Cys substitutions to form a disulfide bond were chosen in subtilisin E from Bacillus subtilis, a cysteine-free bacterial serine protease, based on the structure of aqualysin I of Thermus aquaticus YT-1 (a thermophilic subtilisin-type protease containing two disulfide bonds). Cys residues were introduced at positions 61 (wild-type, Gly) and 98 (Ser) in subtilisin E by site-directed mutagenesis. The Cys-61/Cys-98 mutant subtilisin appeared to form a disulfide bond spontaneously in the expression system used and showed a catalytic efficiency equivalent to that of the wild-type enzyme for hydrolysis of a synthetic peptide substrate. The thermodynamic characteristics of these enzymes were examined in terms of enzyme autolysis (t1/2) and thermal stability (Tm). The half-life of the Cys-61/Cys-98 mutant was found to be 2-3 times longer than that of the wild-type enzyme. Similar results were obtained by differential scanning calorimetry. The disulfide mutant showed a Tm of 63.0 degrees C, which was 4.5 degrees C higher than that observed for the wild-type enzyme. Under reducing conditions, however, the characteristics of the mutant enzyme were found to revert to those of the wild-type enzyme. These results strongly suggest that the introduction of a disulfide bond by site-directed mutagenesis enhanced the thermostability of subtilisin E without changing the catalytic efficiency of the enzyme.     

Identification of the allosteric regulatory site in bacterial phosphoribulokinase

Bacterial phosphoribulokinases (PRKs) are octameric members of the adenylate kinase family of enzymes. The enzyme is allosterically activated by NADH and allosterically inhibited by AMP. We have determined the crystal structure of PRK from Rhodobacter sphaeroides bound to the ATP analogue AMP-PCP to a resolution of 2.6 A. The structure reveals that the ATP analogue does not bind to the canonical ATP site found in adenylate kinase family members. Rather, the AMP-PCP binds in two different orientations at the interface of three of the monomers in the octamer. This interface was previously characterized as having an unusually large number of arginine residues. Of the five arginine residues that are near the bound nucleotide, one (Arg 221) is highly conserved in both prokaryotic and eukaryotic (nonallosterically regulated) PRKs, two (Arg 234 and Arg 257) are on a second subunit and conserved in only prokaryotic PRKs, and two (Arg 30 and Arg 31) are on a third subunit with only one of them (Arg 31) conserved in prokaryotic PRKs. Each of these arginine residues was converted by site-directed mutagenesis to alanine. Fluorescence binding data suggest that none of these arginines are involved in active site ATP binding and that Arg 234 and Arg 257 on the second subunit are directly involved in NADH binding, while the other arginines have a minimal effect on NADH binding. While the wild-type enzyme exhibits low maximal activity and hyperbolic kinetics with respect to ATP in the absence of NADH and high maximal activity and sigmoidal kinetics in the presence of NADH, the R31A mutant exhibits identical hyperbolic kinetics with respect to ATP in the presence or absence of NADH. Thus, the transmission of allosteric information from one subunit to another is conducted through a single path that includes NADH and Arg 31.     

Alanine scanning mutagenesis of a high-affinity nitrate transporter highlights the requirement for glycine and asparagine residues in the two nitrate signature motifs

Common to all of the nitrate nitrite porter family are two conserved motifs in transmembrane helices 5 and 11 termed NS (nitrate signature) 1 and NS2. Although perfectly conserved substrate-interacting arginine residues have been described in transmembrane helices 2 and 8, the role of NSs has not been investigated. In the present study, a combination of structural modelling of NrtA (nitrate transporter from Aspergillus nidulans) with alanine scanning mutagenesis of residues within and around the NSs has been used to shed light on the probable role of conserved residues in the NSs. Models show that Asn168 in NS1 and Asn459 in NS2 are positioned approximately midway within the protein at the central pivot point in close proximity to the substrate-binding residues Arg368 and Arg87 respectively, which lie offset from the pivot point towards the cytoplasmic face. The Asn168/Arg368 and Asn459/Arg87 residue pairs are relatively widely separated on opposite sides of the probable substrate translocation pore. The results of the present study demonstrate the critical structural contribution of several glycine residues in each NS at sites of close helix packing. Given the relative locations of Asn168/Arg368 and Asn459/Arg87 pairs, the validity of the models and possible role of the NSs together with the substrate-binding arginine residues are discussed.     

Identification of electrostatic interactions that determine the phosphorylation site specificity of the cAMP-dependent protein kinase

"Charged-to alanine" scanning mutagenesis of the catalytic subunit of the Saccharomyces cerevisiae cAMP-dependent protein kinase (C1) identified three glutamate residues, E171, E214, and E274, that are involved in the recognition of a peptide substrate, kemptide (Leu1Arg2Arg3Ala4Ser5Leu6Gly7). These glutamate residues are conserved or conservatively substituted with asparate in the serine/threonine protein kinases that have a requirement for basic residues on the N-terminal side of their phosphorylation sites. Alanine replacement mutants in C1 were subjected to kinetic analysis using alanine-substituted peptides as substrates. The additivity or nonadditivity of the effects of the alanine substitutions on the catalytic efficiency (kcat/Km) was analyzed. This allowed the identification of electrostatic interactions between the three glutamate residues in the enzyme and the two arginine residues present in the peptide substrate. The data suggest that E171 interacts with Arg2 in the substrate and that E214 and E274 both interact with Arg3. This may be a general method for identifying simple intermolecular interactions involving proteins when there is no three-dimensional structure available of the complex of interacting species. The identification of these interactions provides the potential for rational protein engineering of enzymes with alternative specificities.