Protein engineering on subtilisin

/nf. J. Eiochem. Vol. 25, No. 3, pp. 307-312, 1993 Printed in Great Britain

0020-7 I IX/93 624.00 + 0.00 Pergamon Press Ltd

MINIREVIEW

PROTEIN

ENGINEERING

ON SUBTILISIN

HIROSHI TAKAGI

Food

Research

and

Development Laboratories, Ajinomoto Co. Inc., 1-I Suzuki-cho, Kawasaki 210, Japan [Tel. 044-244-3921; Fax 044-244-96171

Kawasaki-ku,

(Received 14 July 1992)

The serine proteases include two distinct families: the mammalian serine proteases (e.g. chymotrypsin, trypsin and elastase) and the bacterial serine proteases (e.g. subtilisin). They differ in amino acid sequences and three-dimensional structure, despite a common active site geometry and enzymatic mechanism. Subtilisins are a family of alkaline serine endoproteases secreted by a wide variety of Bacillus species; subtilisin BPN’ from B. amyloliquefaciens, subtilisin Carlsberg from B. ficheniformis, subtilisin E from B. subtilis, and subtilisin Amylosacchariticus from B. amylosacchariticus (Markland and Smith, 1971). The enzyme having no disulfide bonds consists of a single polypeptide chain of about 275 amino acid residues (molecular was about 27,500) and possesses the “catalytic triad” of Asp, His and Ser residues which is conservative among serine proteases, as the hallmark of its active sites. The three-dimensional structure from x-ray crystallographic data (Wright et af., 1969) as well as the enzymatic properties have been studied in detail. The genes have been cloned and their DNA sequences have been determined. They exhibit high degrees of amino acid sequence similarity. In view of the industrial applications, subtilisin represents the largest commercial enzyme market, with a world market of more than 500 metric tons a year worth more than $150 millions, mainly for detergent additives and skin and leather processing. Therefore, subtilisin has been extensively investigated from both basic and applied aspects as a promising target for protein engineering. This review describes the current state of protein engineering of subtilisin, and is divided into five areas, structure-function analysis of the precursor, its catalysis, substrate specificity, stability and pH dependence, including our experimental data. STRUCTUREFUNCTlON

ANALYSIS

OF THE PRECURSOR

DNA sequence analysis of subtilisin genes has revealed the existence of a long peptide (prosequence) consisting of 77 amino acid residues between the signal peptide (pre-sequence) and the mature protease. It has been shown that subtilisin is first

synthesized as a pre-pro-form. Although it has not yet been identified in viuo, pro-subtilisin is supposed to be first secreted across the cytoplasmic membrane with the aid of a signal peptide. Subsequently, the pro-sequence is removed either autocatalytically or by existing active subtilisin to produce active mature subtilisin, which is released into the medium (Power et al., 1986) (Fig. 1). However, no biochemical or functional evidence involved in precursor forms of subtilisin has yet been presented, since only mature subtilisin is detected extracellularly. Inouye et al. have previously cloned and expressed the gene for B. subtilis subtilisin E in Escherichiu coli, and have shown using site-directed mutagenesis that active subtilisin is effectively secreted into the E. coli periplasmic space (Ikemura et al., 1987). In addition, a dramatic increase in active subtilisin production has been achieved by inducing the expression of the pre-prop-subtilisin gene with a very low inducer concentration at a low culture temperature. This indicates that the rate of protein synthesis and the culture temperature are important factors for proper folding of the proteins secreted across the cytoplasmic membrane (Takagi et a/., 1988). No protease activity has been detected despite the fact that a large amount of a gene product having the identical primary structure to subtilisin E is secreted into the periplasmic space, when the OmpA signal peptide was directly fused to the mature subtilisin sequence, although the structure of the pre-sequence did not affect the production of the active subtilisin E. These results indicate that the subtilisin pro-sequence is essential for the formation of enzymatically active subtilisin (Ikemura et al., 1987). When the aggregates of pro-subtilisin produced in E. coli were solubilized in 6 M guanidine-HCl and dialyzed against high concentrations of sodium phosphate buffer, pro-subtilisin was efficiently processed to active subtilisin by an intramolecular, self-processing mechanism with concomitant cleavage of the pro-sequence. It was also suggested that the highly charged nature of the pro-sequence plays an important role in the process of denatured pro-subtilisin refolding (Ikemura and Inouye, 1988). With the 307

308

HIR~SHITAKAGI

\

/ DNA

t v PX-

Pro-

mRNA

c

Mature

:11

Active mature subtilisin Fig. 1. Schematic representation of maturation mechanism of subtilisin. aid of exogenously added pro-sequence by either an active-centre mutant of pro-subtilisin (Asp32 to Asn). which is not processed to the active enzyme due to the prevention of the intramolecular processing (Zhu et al., 1989) or a synthetic peptide of 77 residues (Ohta et al., 1991). the denatured mature protein can be refolded to the active enzyme in an intermolecular process in oifro. Intermolecular complementation of protein folding by a pro-sequence has been observed in the acid-denatured subtilisins BPN’ and Carlsberg as well as in inactive mature subtilisin E (Zhu et al., 1989). It has been proposed that the pro-peptide functions as an intermolecular chaperone for prosubtilisin (Ohta er al., 1991). Recently, to analyze the role of the pro-sequence in subtilisin folding, local random mutagenesis using the spontaneous misincorporation rate of Taq DNA polymerase in PCR was used to obtain mutations in the pro-sequence which prevent production of active enzyme (Lerner et al., 1990). They are currently characterizing the effects of these mutations on the folding of pro-subtilisin and are trying to isolate active revertants which suppress the pro-sequence mutations by performing the same mutagenesis on the mature subtilisin. Therefore. it is of great interest to investigate the specific interactions between the pro-peptide and mature subtilisin during the renaturation process. CATALYSIS The catalytic machinery in subtilisin consists of the catalytic triad (Asp32, His64 and Ser221) and the

oxyanion binding site (Asn 155 and the main chain amide of Ser221). In the hydrolysis of peptide bonds by subtilisin, the hydroxyl oxygen of the catalytic Ser221 transfers its proton to the catalytic His64 coincident with attack on the carbonyl group of the scissile peptide bond. In serine proteases, two peptide NH groups of the polypeptide backbone form the so-called “oxyanion hole” by donating hydrogen bonds to the negatively charged oxygen atom of the tetrahedral intermediate. Crystallographic studies suggest that stabilization of this activated complex is accomplished through the donation of a hydrogen bond from the amide side group of Asn155 to the carbonyl oxygen of the peptide substrate. To investigate the function of the hole, Leu was introduced at position 155. Leu155 mutant had an unaltered K,, but a greatly reduced K,,, when assayed with a peptide substrate. This result is consistent with the Asnl55 mediating stabilization (Bryan et NI., 1986). Two independent groups (Rao et al. (1987) and Hwang and Warshel (1987) have applied a new simulation approach called free energy perturbation, to calculate the differential free energy of binding and free energy of activation for catalysis by wild-type and mutant (Asn 155 to Thr, Leu and Ala) subtilisins BPN’. Three simulations, one on the enzyme itself, one on the enzyme-substrate non-covalent complex and one on a model for the transition state for acylation catalysis, were performed which demonstrated the predictive power and utility of theoretical simulation methods in studies of the effects of site-directed mutagenesis. Carter and Wells (1990) have examined the catalytic importance and interplay between residues within the catalytic triad by individual and multiple replacement with Ala in the subtilisin BPN’ gene. Mutations in the catalytic triad greatly reduced the turnover number but little affected the K,,. However, the mutants retained activities that are at least lo”-IO+’ above the non-enzymatic rate. A possible reason for the residual activity was hydrogen bonding with the N of AsnlS5 that helps to stabilize the oxyanion. Replacing Asnl55 by Gly lowers the k,,, 150-fold with no change in the K,,,. Upon combining the Asn155Gly and Ser22lAla mutations. k,,, is actually 5-fold greater than for the Ser221Ala enzyme. Thus the catalytic role of Asn155 is dependent upon the presence of Ser221. On the other hand. the specific activity of subtilisin E was substantially increased by optimizing the amino acid residue at position 3 1 (Ile in the wild-type) in the vicinity of the catalytic triad (Takagi et al., 1988a, b). Eight non-charged amino acids were introduced at this site, which is next to catalytic Asp32. Mutant enzymes were expressed in E. coli and prepared from the periplasmic space. Only the Val and Leu substitutions gave active enzymes, and the Leu3 I mutant had a greatly increased activity compared with the wild-type enzyme. The other six mutant enzymes showed a marked decrease in activity.

Protein engineering on subtilisin Table

I.

Kinetics

of wild-type

and mutant

subtilisins

against

k,,,, Km (k,,,

309

substrates

differmg

in PI

residue

Km)* BPN’B

Subtihsin PI

reslduet

Phe

Ala Arp

Leu

Wild-type

II (21, 1.9) 4.2 (1.4, 0.33) 47 (0.93. 0.20)

E!: Leu31

60 (120. 2.0) 1.4 (2.3. 0.31)

Carlsberg

.,L&!

360 (50. 0 14) I4 (1.9. 0 15)

wild-type

5.1 (30.5.3)

2600 (250. 0.094) 40 (6 I. 0.15)

2500 (510. 0.20) 86 (14. 0.16)

22

(I IO, 5.0)

I

0.035 (0.1%

5.2)

x7 (3..1. 0.3X) 40 (30. 0 75) I40 (13. OOQ) I400 (25. 0 OIXJ

Gin Lys Me1 Tyr *k,,,(sec ‘), IiJmM). ~Succlnvl-Ala-,4la-Pro-X-r,-nltroanlllde: Phe:Val-Arg-p-nitrol~ilid~. $Takagl CI N/. (198X). $Wells <‘! rrl (19Xi).

I7

IO (2 2. 0.21 )

Glil

SPECIFICITY

Subtilisin has broad specificity and contains a large hydrophobic substrate binding cleft. A conserved Gly at position 166, located at the bottom of the substrate binding cleft, was replaced by I2 nonionic amino acids (Estell er al., 1986). In general. the catalytic efficiency toward small hydrophobic substrates was increased up to l6-fold by hydrophobic substitutions at position 166 in the binding cleft. Exceeding the optimal binding volume of the cleft (160 A), by enlarging either the substrate side chain or the side chain at position 166, evoked precipitous drops in catalytic efficiency (up to 5000-fold) as a result of steric hindrance. Wells et al. (1987b) reported general changes in substrate specificity resulting from charged amino acid substitutions at residues 156 and 166 in the PI binding site, causing electrostatic effects. This was supported by data showing that charged substitutions substantially increase the catalytic efficiency toward complementary charged PI substrates (up to 1900-fold) and decrease it toward similarly charged PI substrates.

I

2.2

1.3) 5’) (IS. 0.31) Y2 (15. 1.6)

(3.7. I 7) 160 (46, 0.29) I6 (68. 4.3) 2000 (87. 0.044) 2900 (230, 0.079)

(I 3,

I500

(76. 0 05) 3x00 (140. 0.036)

X = Phe,l ~u.C;lu.C;ln.4la.L~~,Met

The Leu31 replacement caused a prominent 226-fold increase in catalytic efficiency due to a high k,,, for peptide substrates (Table I). This result indicates that a branched-chain amino acid at position 31 is essential for the expression of subtilisin activity and that the level of the activity depends on side chain structure. Position 31 is occupied by Leu in subtilisin Carlsberg, which has a much higher activity than other subtilisins, but by Be in subtilisin BPN’ and also subtilisin E. Our results indicate that the amino acid residue at position 31 is responsible for the difference in enzymatic activity among these enzymes. SUBSTRATE

SerlS6:Alal69 Wild-type

or

Tyr.

and

henzoyl-

Subtilisin BPN’ and Carlsberg differ by 30% in primary structure and by more than 6-fold in catalytic efficiency toward various substrates. Despite large differences in sequence and specificity among the two subtilisins. only two amino acid substitutions (at position 156 and 217) occur within 4 8, of modeled substrates, and a third substitution (at position 169) is within 7 A. The three substitutions of subtilisin Carlsberg (Serl56. Ala169 and Leu217) were introduced into subtilisin BPN’ (Wells et a/.. 1987a). The substrate specificity of the triple mutant approached that of subtilisin Carlsberg when assayed with several different substrates that varied in charge, size and hydrophobicity. Thus. the substrate specificity can be altered by replacement of amino acid residues to which a substrate binds directly. We analyzed the substrate specificity of subtilisin E based on the structure of a new alkaline elastase produced by the alkalophilic Bacillus strain Ya-B, which has very high elastolytic activity (Takagi et al., 1992). Despite the high homology of the primary sequences of both enzymes (54% identical), alkaline elastase lacked four consecutive amino acids which, in subtilisin, have been shown by x-ray analysis to lie close to the Pl binding cleft. To examine the influence of such a deletion in subtilisin E on its substrate specificity, we constructed several mutants missing four amino acids by site-directed mutagenesis. When assayed with synthetic peptides, elastin and casein as substrates, a mutant lacking Ser 161 -Thr I62-Ser l63-Thrl64 showed considerably lower specific activity toward the substrates for subtilisin. and its specificity approached that of alkaline elastase. The results indicate that this region is responsible for the difference in specificity between the two enzymes.

310

HIROSHI TAKAGI

Replacement of the catalytic His64 with Ala in subtilisin BPN’ reduced the catalytic efficiency when assayed with succinyl-Phe-Ala-Ala-Phe-p-nitroanilide. Molecular modeling showed that a His side chain at the P2 position of a substrate bound at the active site could be superimposed on the catalytic His side chain. Accordingly, the His64Ala mutant enzyme can partially recover the function of the lost catalytic His from a His P2 side chain on the substrate (Carter and Wells, 1987). Such “substrateassisted catalysis” provides a new basis for engineering enzymes with very narrow and potentially useful specificities. The catalytic efficiency of this mutant was increased almost 20-fold by judicious choice of substrate and by installing three additional mutations (Glul56Ser, Gly169Ala and Tyr217Leu) which increased the activity of wild-type enzyme (Carter et ul., 1989). STABILITY

Since previous studies have implicated Met222 as a primary site for oxidative inactivation of subtilisin, all 19 amino acid substitutions at this site were introduced in the subtilisin BPN’ gene using a cassette mutagenesis (Estell et al., 1985). Mutants containing nonoxidizable amino acids (Set-, Ala and Leu) were resistant to inactivation by HzOz, whereas Met and Cys enzymes were rapidly inactivated. To enhance the thermostability, disulfide bond(s) have been introduced into the protein by site-directed mutagenes (Table 2). There have been conflicting reports concerning the thermostability of subtilisin BPN’ in which a designed disulfide bond has been introduced between positions 22 and 87 with the aid of computer modeling (Wells and Powers 1986; Pantoliano et al., 1987). Furthermore, based upon the computer program, five other disulfide variants Table

2. Thermodynamic

characteristics subtilisins

of the disulfide

Half-life* Mutant

subtilisin

cys22/cyss7t Cys22/Cys87: Cys24/Cys87t Cys26/Cys232$ Cys29:Cysl19$ Cys36!Cys210$ Cys4l,Cys8@$ Cysl48/Cys243§ Cys61 ICys98’1

ATn,( C)

Reduced -SHI-SH

Oxidized -S-S-

29 85 73

41 200 96 100 69

3 1.5 15 103

MetSOPhe Asn76Asp Gly I69Ala Gln206Cys TyrZl7Lys AsnZIE.Serj, “Values are realtive half-lives tWells et al. (1986). $Pantoliano er al. (1987). §Mitchinson ef al. (1989). 1ITakagi et al. (1990). IiPantoliano el al. (1989).

mutant

3.1

I IO

0.3 0.5

3.0 1.5 261

4.5

- 30,000

(wild-type;

(Mutant wild-type)

100).

14.3

of subtilisin BPN’ have been constructed (Cys26/ Cys232, Cys29/Cysl19, Cys36/Cys210, Cys41/Cys80 and Cys148/Cys243) (Mitchinson and Wells, 1989). The disulfides were chosen on the basis of structural homology with proteinase K, a fungal protease that contains two disulfide bonds, and the stabilizing effect of a structural calcium atom. In some cases, the disulfide bond-containing protein was stabilized relative to its reduced form. However, none of the disulfide mutants was substantially more stable than the wild-type enzyme. In our study, sites for Cys substitutions to form a disulfide bond were chosen in subtilisin E from B. subtilis based upon the structure of aqualysin I of Thermus aquaticus YT-1 (a thermophilic subtilisin-type protease containing two disulfide bonds) (Takagi et al., 1990). Cys residues were introduced at position 61 (wild-type, Gly) and 98 (Ser) in subtilisin E by site-directed mutagenesis. The Cys61/Cys98 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. The thermodynamic characteristics of these enzymes were examined in terms of enzyme autolysis (tvz) and thermal stability (T,). The half-life of the Cys61/Cys98 mutant was found to be 2-3 times longer than that of the wild-type enzyme. The mutant showed a 7’,,, of 63.OC, which was 4.5 C higher than that observed for the wild-type enzyme. Under reducing conditions, however, the characteristics of the mutant enzyme reverted to those of the wild-type. These results indicate that the introduction of a disulfide bond enhanced the thermostability of subtilisin E without changing the catalytic efficiency of the enzyme. Although eleven mutants designed using computer modeling have been produced in an attempt to engineer salt bridges into subtilisin BPN’ in view of the contributions to the thermal stability, mutants were found to be nearly identical to the wild-type in thermal stability (Erwin et al., 1990). A computer modeling prediction revealed that the low percentage of salt bridge formation is probably due to an overly simplistic electrostatic model. Six individual amino acid substitutions at separate positions in subtilisin BPN’ were found to increase the stability (Pantoliano et al.. 1989). These stabilizing changes, Asn2 18Ser, Gly 169Ala. Tyr2 17Lys, MetSOPhe, Gln206Cys and Asn76Asp, were discovered using five different approaches: (1) random mutagenesis; (2) design of buried hydrophobic side groups; (3) design of electrostatic interactions at Ca’+ binding sites; (4) sequence homology consensus; and (5) serendipity (Table 2). The combination of these six mutations was found to be 100 times more stable than the wild-type enzyme in aqueous solution at room temperature and 50 times more stable than the wild-type in anhydrous dimethylformamide (Wong et al.. 1990).

Protein engineering on subtilisin pH DEPENDENCE

Chemical modification studies have shown that the pH dependence of catalysis by serine proteases alters with changes in the overall surface charge. His64 at the active site of subtilisin acts as a general base during catalysis, accepting a proton from the nucleophilic residue Ser221 as it forms a bond with the substrate carbonyl carbon. The activity of subtilisin increases between pH 6 and 8 (pKa 7.2) as His64 is deprotonated, and decreases between pH 9.5 and 11, with a pKa of 10.5. Fersht et al. (1985) showed how the change of just one surface charge which is 14-15 A from the imidazole of His64 subtilisin BPN’ significantly affected the pH dependence of the enzyme. The mutation of surface Asp to Ser at position 99 had no significant effect on either the structure or the catalytic properties of the enzyme other than electrostatically. The mutation (Asp99Ser) lowers the pKa of the His by up to 0.4 units at low ionic strength, whereas, at high ionic strength, the mutation has essentially no effect on the pKa values due to the masking of the electrostatic effects. Another charged residue that is at a similar distance from His64 is also in an external loop that is known to tolerate Ser is Glul56, at the opposite side of the active site. A mutation of Gh.1156 to Ser also lowers the pKa of His64 by up to 0.4 units at low ionic strength (Russell et ul., 1987). Moreover, increased changes in charge were made by mutating Asp99 and Glu156 to Lys (Russel and Fersht, 1987). In fact, making a double charge change by mutating either Asp99 or Glul56 to Lys lowered the pKa of His64 by 0.6 units. The quadruple charge change in the double mutant Asp99Lys; Glul56Lys gave a shift of 1.0 unit at low ionic strength. The activity of the double mutant (Asp99Lys; Glu156Lys) for the Phe substrate as PI site is twice that of wild-type subtilisin BPN’ at high pH and IO times greater at pH 6 at low ionic strength. A series of their experiments has shown that modest changes in activity may be effected

by one or two point

mutations.

The following

changes in pH profiles by protein engineering. (1) Making the surface more negatively charged raises the pKa values of acidic groups in protein because they all lose a proton on ionization. (2) Changes will be maximized at low ionic strength. (3) Significant changes will be manifested at ionic strengths as high as 0.1 M. (4) Mutations should be designed so that they do not concentrate counterions in the active site. The spectral analysis showed that Tyr ionization correlated with the reversible loss of activity. Two Tyr residues are located in the substrate binding site, Tyr104 in the P4 binding site and Tyr217 in the PI’ binding site. Indeed, substitution of Tyr104 to Phe extended the stability in alkaline condition (pH 10.6) (Wells and Estell, 1988). rules

should

be

applied

for

designing

311

Acknowledgements--I am greatly indebted to my coresearchers Drs Masayori Inouye, Takahisa Ohta, Hiroshi Matsuzawa, Haruo Momose, Yoshimi Maeda, Makari Yamasaki, Haruo Ikemura, and to my colleagues at Ajinomoto Co., Inc.

REFERENCES

Bryan P., Pantoliano M. W., Quill S. G., Hsiao H.-Y. and Poulos T. (1986) Site-directed mutagenesis and the role of the oxyanion hole in subtilisin. Proc. n&n. Acad. Sci. U.S.A. 83, 3743-3745. Carter P. and Wells J. A. (1987) Engineering enzyme specificity by “substrate-assisted catalysis”. Science 237, 394399. Carter P. and Wells J. A. (1990) Functional interaction among catalytic residues in subtilisin BPN’. Proteins Strucf. Funct. Genef. 7, 335-342. Carter P., Nilsson B., Burnier J. P., Burdick D. and Wells J. A. (1989) Engineering subtilisin BPN’ for sitespecific proteolysis. Proteins Struct. FWICI. Genet. 6, 240-248. Erwin C. R., Barnett B. L., Oliver J. D. and Sullivan J. F. (1990) Effects of engineered salt bridges on the stability of subtilisin BPN’. Protein Engng 4, 87-97. Estell D. A., Graycar T. P. and Wells J. A. (1985) Engineering an enzyme by site-directed mutagenesis to be resistant to chemical oxidation. J. biol. Chem. 260, 65184521. Estell D. A., Graycar T. P., Miller J. V., Powers D. B., Burnier J. P., Ng P. G. and Wells J. A. (1986) Probing steric and hydrophobic effects on enzymesubstrate interactions by protein engineering. Science 233, 659463. Hwang J.-K. and Warshel A. (1987) Semiquantitative calculations of catalytic free energies in genetically modified enzymes. Biochemistry 26, 2669-2673. Ikemura H. and lnouye M. (1988) In virro processing of pro-subtilisin produced in Escherichia coli. J. biol. Chem. 263, 12959-12963. Ikemura H., Takagi H. and Inouye M. (1987) Requirement of pro-sequence for the production of active subtilisin E in Escherichia coli. J. biol. Chem. 262, 7859-7864. Lerner C. G., Kobayashi T. and Inouye M. (1990) Isolation of subtilisin pro-sequence mutations that affect formation of active protease by localized random polymerase chain reaction mutagenesis. J. hiol.Chem. 265, 20085-20086.

Markland F. S. and Smith E. L. (1971) Subtilisins: primary structure, chemical and physical properties. In The Enzymes (edited by Boyer P. D.), Vol. 3, pp. 561L608. Academic Press, New York. Mitchinson C. and Wells J. A. (1989) Protein engineering of disulfide bonds in subtilisin BPN’. Biochemistry 28, 480748 15. Ohta Y., Hojo H., Aimoto S., Kobayashi T., Zhu X., Jordan F. and Inouye M. (1991) Pro-peptide as an intermolecular chaperone: renaturation of denatured subtilisin E with a synthetic pro-peptide. Molec. Microbial. 5, 1507-1510. Pantoliano M. W., Ladner R. C., Bryan P. N., Rollence M. L., Wood J. F. and Poulos T. L. (1987) Protein engineering of subtilisin BPN’: enhanced stabilization through the introduction of two cysteines lo form a disulfide bond. Biochemistry 26, 2077-2082.

312

HIROSHI TAKAGI

Pantoliano M. W., Whitlow M., Wood J. F., Dodd S. W., Hardman K. D., Rollence M. L. and Bryan P. N. (1989) Large increases in general stability for subtilisin BPN’ through incremental changes in the free energy of unfolding. Biochemisrry 28, 7205-7213. Power S. D., Adams R. M. and Wells J. A. (1986) Secretion and autoproteolytic maturation of subtilisin. Proc. natn. Acad. Sci. U.S.A. 83, 30963100. Rao S. N., Singh U. C., Bash P. A. and Kollman P. A. (1987) Free energy perturbation calculations on binding and catalysis after mutating Asn155 in subtilisin. Nature 328, 551-553. Russell A. J. and Fersht A. R. (1987) Rational modification of enzyme catalysis by engineering surface charge. Nature 328, 496-500. Russell A. J.. Thomas I’. G. and Fersht A. R. (1987) Electrostatic effects on modification of charged groups in the active site cleft of subtilisin by protein engineering. J. molec. biol. 193, 8033813. Takagi H., Arafuka S., Inouye M. and Yamasaki M. (1992) The effect of amino acid deletion in subtilisin E. based on structural comparison with a microbial alkaline elastase. on its substrate specificity and catalysis. J. Biochem. 111, 584-588. Takagi H., Morinaga Y.. Ikemura H. and lnouye M. (1988a) Mutant subtilisin E with enhanced protease activity obtained by site-directed mutagenesis. J. biol. Chem. 263, 19592-19596. Takagi H., Morinaga Y., Tsuchiya M., Ikemura H. and lnouye M. (1988b) Control of folding of proteins secreted by a high expression secretion vector, PIN-III-ompA: 16-fold increase in production of active subtilisin E in Escherichia co/i. BioiTechnol. 6, 948-950.

Takagi H., Takahashi T., Momose H., Inouye M., Maeda Y., Matsuzawa H. and Ohta T. (1990) Enhancement of the thermostability of subtilisin E by introduction of a disulfide bond engineered on the basis of structural comparison with a thermophilic serene protease. J. biol. Chem. 265, 6874-6878. Thomas P. G., Russell A. J. and Fersht A. R. (1985) Tailoring the pH dependence on enzyme catalysis using protein engineering. Nature 318, 375-376. Wells J. A. and Estell D. A. (1988) Subtilisin-an enzyme designed to be engineered. TIBS 13, 291-297. Wells J. A. and Powers D. B. (1986) In uivo formation and stability of engineered disulfide bonds in subtilisin. J. biol. Chem. 261, 6564-6570. Wells J. A. Cunningham B. C., Graycar T. P. and Estell D. A. (1987a) Recruitment of substrate-specificity properties from one enzyme into a related one by protein engineering. Proc. nam. Acad. Sci. U.S.A. 84, 5167~-5171. Wells J. A.. Powers D. B., Bott R. R. Graycar T. P. and Estell D. A. (1987b) Designing substrate specificity by protein engineering of electrostatic interactions. Proc. natn. Acad. Sci. U.S.A. 84, 1219.-1223. Wong C-H., Chen S.-T., Hennen W. J., Bibbs J. A., Wang Y.-F.. Liu J. L.-C., Pantoliano M. W., Whitlow M. and Bryan P. N. (1990) Enzymes in organic synthesis: use of subtilisin and a highly stable mutant derived from multiple site-specific mutations. J. Am. them. Sot. 112, 945-953. Wright C. S., Alden R. A. and Kraut J. (1969) Structure of subtilisin BPN’ at 2.5A resolution. Nalure 221, 235-242. Zhu X.. Ohta Y., Jordan F. and Inouye M. (1989) Prosequence of subtilisin can guide the refolding of denatured subtilisin in an intermolecular process. Nuture 339, 483-484.