The Importance of a Distal Hydrogen Bonding Group in Stabilizing the Transition State in Subtilisin BPN'*

Stabilization of an oxyanion transition state is im- portant to catalysis of peptide bond hydrolysis in all proteases. For subtilisin BPN', a bacterial serine pro- tease, structural data suggest that two hydrogen bonds stabilize the tetrahedral-like oxyanion intermediate: one from the main chain NH of S e P 1 and another from the side chain NH2 of Asn'". Molecular x-ray probed ing valine, or alanine site-directed mutagenesis. tended to the and hydrogen bonding ability

Enzymes catalyze reactions by binding the transition state more tightly than the reactants or products (reviewed by Kraut, 1988). Enzymes often accomplish this using weak binding forces such as hydrogen bonds, electrostatic and hydrophobic interactions. For example, x-ray crystallographic studies of the bacterial serine protease subtilisin BPN' have indicated that two hydrogen bonds are involved in stabilizing the tetrahedral-like oxyanion transition state (Fig. 1;Robertus et al., 1972), one from the backbone amide of Ser2" and * This work was supported by National Science Foundation Postdoctoral Grant NSF-DMB-87-430 (to s. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence($ reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 2 ST
another from the amide side chain of Asn'". These structural observations were corroborated by site-directed mutagenesis (Bryan et al., 1986;Wells et al., 1986) which demonstrated that replacement of Asn'" with a variety of other side chains resulted in 3-5 kcal/mol reductions in transition state stabilization energy with only minor effects upon substrate binding.
In the x-ray structure of subtilisin BPN', the side chain 06 Thrz'" (Fig. 1) is oriented so that the y-hydroxyl is about 4 A away from the developing oxyanion in the transition state complex (E. SS). Molecular dynamic simulations and free energy calculations (Rao et al., 1987) suggested that the yhydroxyl of ThrZ2" may move to form a direct hydrogen bond with the oxyanion in the transition state. It has been suggested that enzymes can also provide oriented dipoles that stabilize charged transition states without being in direct Hbonding distance (Warshel, 1987;Warshel et al., 1988;Hwang and Warshel, 1988). ThrZZ0 is such an oriented dipole and one of the most highly conserved residues in the subtilisin family (Fig. 2 ) . Its side chain makes van der Waals contact with AS^'"^ and thus may affect the functional role of Asn'".
Therefore, to evaluate the functional importance of ThrZZo to catalysis versus substrate binding and its functional independence from AsnI5', we have systematically replaced ThrZz0 separately and together with Asn'".

EXPERIMENTAL PROCEDURES
Construction and Purification of Subtilisin Variants-The Thr"" mutations were introduced into the S24C' variant (Carter and Wells, 1987) of the Bacillus amyloliquefaciens subtilisin gene (Wells et al., 1983) cloned in to the phagemid vector pSS5 (Carter and Wells, 1988) by site-directed mutagenesis (Carter et al., 1985). Mutagenesis was performed on a single-stranded pSS5 template containing a KpnI site tides: T220S, 5' GGCGTACAACGGGAGCTCTATGGCATCTC, near codon 220 and S221C mutation with the following oligonucleo-Sac1 site; T220V, 5'GCGTACAACGGTGTGTCTATGGCTAGC CCGCACGGTG, NheI site; T220C, 5 ' G C G T A C A A C G G T m TATGGCTAGCCCGCACGTTG, NheI site; T220A, 5'GGCGTA-CAACGGCGCCTCTATGGCATCTC, NarI site, which regenerated the active site Ser21' (restriction sites introduced are underlined). Restriction-selection (Wells et al., 1986) against the KpnI site over the parent (KpnI'IS221C) template was used to enrich for mutant plasmids. The variant phagemids were verified by dideoxy sequencing (Sanger et al., 1977) and double-stranded DTJA was transformed into a protease-deficient strain (BG2036) of Bacillus subtilis (Yang et al., 1984). Subtilisin mutants were cultured (in the absence of wild type) and purified on an SP-Sephadex ion exchange column followed by activated thiol affinity chromatography as described (Carter and Wells, 1988).
Kinetic Characterization-Enzymes were assayed with the sub-' Mutants are named according to the wild type residue (using single-letter amino acid code) followed by its sequence position and then the mutant residue. For example, T220S designates that Thr"" is replaced by serine (Wetzel, 1988). Multiple mutants are indicated by the series of single mutants separated by slashes.

TABLE I Kinetic constants for wild-type and codon 220 subtilisin mutants against N-succinyl-L-Ala-L-Ala-L-Pro-L-Phe-p-nitroanilide substrate
All mutants contain S24C to aid in purification.

hydrolysis of N-succinyl-L-Ala-L-Ala-L-Pro-L-Phe-p-nitroanilide
All mutants contain S24C to aid in purification except for N155Q whose data we report from Wells et al. (1986).  strate succinyl-L-Ala-L-Ala-L-Pro-L-Phe-p-nitroanilide in 1 ml of 100 mM Tris.HC1 (pH 8.60), 4% (v/v) dimethyl sulfoxide at (25 k 0.2) "C as described previously (Carter and Wells, 1988). Enzyme concentrations were determined spectrophotometrically (e%% = 1.17, Matsubara et al., 1965). Subtilisin variants were assayed at concentrations of about 5 nM for the most active variants and 1 pM for the least active mutants. Initial rates of hydrolysis were measured at substrate concentrations in the range of 0.2-5 X KM.

RESULTS AND DISCUSSION
A series of side chain replacements was produced by sitedirected mutagenesis to probe the role of ThrZz0 in catalysis (Table I). Removal of the y-methyl group by the T220S mutation causes only a small reduction (<%fold) in kcat and essentially no change in KM. In contrast, replacing the hydrogen bond donating y-hydroxyl with a y-thiol, which is nonpolar and a poor hydrogen bond donor (Crampton, 1974;Paul, 1974), results in a 20-fold drop in kat for the T220C mutant relative to wild-type, with essentially no change in KM. The reduction in activity for the T220C mutation is not likely due to deprotonation of Cys at pH 8.6, because even at pH 7.0 the reduction in Kcat relative to wild-type is the same (data not shown). Similarly, replacing the y-hydroxyl with a y-methyl group as in the T220V mutation causes a 20-fold reduction in kc,, relative to the wild-type enzyme. The T220A mutation is slightly less deleterious than substituting the y-hydroxyl with nonpolar side chains as in the T220C and T220V mutants. Perhaps a water molecule can fit in the space left when T220 is replaced by alanine, and this polar water can stabilize the oxyanion better than a nonpolar substitutent. In any case, these data suggest that the y-hydroxyl group of ThrZz0 functions predominantly in the catalytic step to stabilize the oxyanion transition state complex.
Assuming that the enzyme mechanism and structure are not substantially altered by mutations at ThrZz0, the reduction in transition state stabilization energy ( A A G , Table I)

can
The abbreviation used is: AAC%, transition state stabilization energy. be a useful estimate of the functional importance of the ThrZz0 y-hydroxyl group (Wilkinson et al., 1983). The data suggest the hydroxyl imparts 1.8-2.0 kcal/mol to stabilize the oxyanion transition. This is substantially less than the stabilization energy imparted by estimated to be 3.5-5 kcal/ mol (Table 11; Wells et al., 1986;Bryan et al., 1986). The structural basis for these differences could reside in the fact that the N62 of is within hydrogen bonding distance of the developing oxyanion (-2.7 A), whereas the Oy of ThrzZ0 is too far away to hydrogen bond (-4.0 A). Alternatively, because of the Oy of ThrZz0 makes van der Waals contact with the role of ThrZz0 could only be to support the side chain orientation of or to polarize its amide side chain.
Double mutants were constructed to distinguish these possibilities. If the role of ThrZz0 is to support the function of then once the N62 of is altered, mutations at ThrzZ0 should no longer have an effect upon AAG% (Carter et al., 1984; for review see Wells, 1990). On the other hand, if both residues function independently then the AAG& value for the double mutant at positions 155 and 220 should be equal to the sum of the A A G values for the two single mutants. When two alanine substitutions are combined at positions 155 and 220 the effect upon the AAG% for the double mutant (N155A, T220A) is the sum of the two single mutants (Table 11). Assuming the absence of compensating effects, these mutations are functionally independent, which suggests that the y-hydroxyl of ThrzZ0 stabilizes the oxyanion transition state separately from the amide of It is important to point out that the above additivity analysis was performed using small alanine substitutions to minimize structurally disruptive or alternate H-bonding effects that could complicate the interpretation of the results. For example, when N155S is substituted for N155A (Table 11), the reduction in A A G for the double mutant (N155S/T220A) is about 1 kcal/mol less than expected (5.2 kcal/mol actual; 6.1 kcal/mol expected). By contrast, when N155Q is substi-

Oxyanion stabilization by
Th?20 tuted for N155A, the effect upon the double mutant (N155Q/ T220A) is 1.6 kcal/mol more than expected (6.5 cal/mol actual, 4.9 kcal/mol expected). Replacement of these alternative H-bond donors (serine or glutamic acid) at position 155 leads to nonadditivity when combined with the T220A mutation. Structural and kinetic analysis should help to clarify the basis for these more complicated effects. The catalytic importance of the y-hydroxyl group of ThrZz0 may be rationalized in one of two ways. Dynamic simulations over a short time period ((20 ps;Rao et al., 1987) indicate that ThrT2' can move to donate an H-bond to the oxyanion intermediate. In this case, the smaller contribution of ThrZ2' as compared with (1.8 kcal/mol uersus 4.2 kcal/mol) may represent strain or reorganization energy involved in placing the y-hydroxyl of ThrZz0 near the oxyanion.
A second possibility is that Th?20 may stabilize the oxyanion by longer range dipolar effects (Warshel, 1987;Warshel et al., 1988;Hwang and Warshel, 1988). Charge-dipole potentials fall off as l/r2 (for review see Adamson, 1979) and may provide significant electrostatic stabilization Warshel, 1978). For this effect, the contribution of ThrZz0 to transition state stabilization can be estimated in kcal/mol using equation 1 AG = 332 Qp/r2t (1) where Q is the charge on the oxyanion (-1.0; Warshel and Russell, 1986), p is the effective charge of a hydroxyl dipole (0.427;Warshe! and Russell, 1986), r is the separation distance in A (4.0 A; Matthews et al., 1975), and c is the effective dielectric constant (-4).3 For Thr220, AG is about -2.0 kcal/ ~o l . The same calculation for with a separation of 2.7 A between the Nb2 and the oxyanion, gives AG as about -4.5 kcal/mol. Thus, the smaller effect upon mutating Thr220 compared with may only reflect its being further away from the developing oxyanion. Moreover, mutations of Thr2" or AsrP5 affect kcat and not KM. Electrostatic effects can account for this preferential stabilization of the oxyanion (in the E .S$) as compared with the carbonyl (in E -S) because a charge-dipolar interaction is stronger and more long range compared with orientated dipole-dipole interaction (which decays as l/r3).
A concern in analyzing mutant enzymes with such low catalytic activity is that artifacts could occur from a consistent low level contamination by other proteases or mistranslation leading to a wild-type enzyme contamination. However, the KM values for most of the mutants analyzed usually differ significantly (albeit slightly) from wild-type enzyme and the other mutants. In addition, the multiple mutants differ substantially from each other, the single mutants, and the wild type. Each of the mutants contains an additional S24C mutation that allows affinity purification of the mutant enzyme by thiol-Sepharose chromatography (Carter and Wells, 1987). This purification procedure has been shown to reduce contamination by cysteine-free wild type enzyme below detectable limits (Carter and Wells, 1988). All subtilisin mutants were purified to homogeneity as judged by Coomassie-stained A. Warshel, personal communication.
sodium dodecyl sulfate electrophoretic gels. Thus, the activity we measure reflects catalysis from the mutated active sites.

CONCLUSIONS
The y-hydroxyl of ThrZ2' imparts an important catalytic advantage to stabilization of the oxyanion transition state. Its role is similar to but functionally independent from The stabilizing effects of ThrZ2' are mediated either via a short range charge-dipolar interaction or by dynamic fluctuations in the protein structure which permit direct hydrogen bonding with the oxyanion. Thus, polar residues that are oriented near (but not in static contact with) the developing oxyanion can have significant stabilizing effects. The introduction of such oriented dipoles by protein engineering methods may provide additional stabilization of charged transition states in enzymes.