Serine to cysteine mutations in trp repressor protein alter tryptophan and operator binding.

The tryptophan repressor regulates expression of the aroH, trpEDCBA, and trpR operons in Escherichia coli. The protein contains no cysteine residues, and the presence of this reactive side chain would allow introduction of spectral probes to monitor binding reactions. Three mutant trp aporepressors, each with a point mutation from serine to cysteine, were produced at positions 67, 86, and 88 by oligonucleotide-directed site-specific mutagenesis. This single conservative substitution affected both tryptophan and operator DNA affinities in all three purified proteins. Cysteine substitution for serine at position 67 decreased tryptophan binding by approximately 6-fold and the operator DNA affinity by approximately 50-fold. The proximity of this amino acid to Gln-68 which is involved in binding to operator DNA (Otwinowski, Z., Schevitz, R. W., Zhang, R.-G., Lawson, C. L., Joachimiak, A., Marmorstein, R. Q., Luisi, B. F., and Sigler, P. B. (1988) Nature 335, 321-329) may account for this effect. Substitution at position 86 diminished tryptophan binding by approximately 4-fold and operator DNA binding by approximately 130-fold. The participation of Ser-86 in the hydrogen bond network required for operator binding (Otwinowski, Z., Schevitz, R. W., Zhang, R.-G., Lawson, C. L., Joachimiak, A., Marmorstein, R. Q., Luisi, B. F., and Sigler, P. B. (1988) Nature 335, 321-329) presumably accounts for the DNA binding effects. The diminished corepressor activity in these two mutants may derive from distortions of the binding region, as the tryptophan and DNA binding sites are intimately related. The mutation at position 88 altered tryptophan binding the most of the three mutants (approximately 18-fold) and operator binding least (approximately 12-fold). Ser-88 forms a hydrogen bond with the amino group of bound tryptophan (Schevitz, R. W., Otwinowski, Z., Joachimiak, A., Lawson, C. L., and Sigler, P. B. (1985) Nature 317, 782-786), and alteration of the geometry of the side chain would be anticipated to perturb the topology of the binding site. The diminished operator affinity may derive from improper alignment of the tryptophan ligand, crucial for high affinity operator binding (Otwinowski, Z., Schevitz, R. W., Zhang, R.-G., Lawson, C. L., Joachimiak, A., Marmorstein, R. Q., Luisi, B. F., and Sigler, P. B. (1988) Nature 335, 321-329).(ABSTRACT TRUNCATED AT 400 WORDS)


Serine to Cysteine Mutations in Trp Repressor Protein Alter
Tryptophan and Operator Binding* (Received for publication, March 13, 1989) Wei-Yuan Chou and Kathleen Shive MatthewsS The tryptophan repressor regulates expression ofthe aroH, trpEDCBA, and trpR operons in Escherichia coli. The protein contains no cysteine residues, and the presence of this reactive side chain would allow introduction of spectral probes to monitor binding reactions. Three mutant trp aporepressors, each with a point mutation from serine to cysteine, were produced at positions 67, 86, and 88 by oligonucleotide-directed site-specific mutagenesis. This single conservative substitution affected both tryptophan and operator DNA affinities in all three purified proteins. Cysteine substitution for serine at position 67 decreased tryptophan binding by -6-fold and the operator DNA affinity by -50-fold. The proximity of this amino acid to Gln-68 which is involved in binding to operator DNA (Otwinowski, Z., Schevitz, R. W., Zhang, R.-G., Lawson, C. L., Joachimiak, A., Marmorstein The diminished corepressor activity in these two mutants may derive from distortions of the binding region, as the tryptophan and DNA binding sites are intimately related. The mutation at position 88 altered tryptophan binding the most of the three mutants (-18fold) and operator binding least (-12-fold). Ser-88 forms a hydrogen bond with the amino group of bound tryptophan (Schevitz, R. W., Otwinowski, Z., Joachimiak, A., Lawson, C. L., and Sigler, P. B. (1985) Nature 317,782-786), and alteration of the geometry of the side chain would be anticipated to perturb the topology of the binding site. The diminished operator affinity may derive from improper alignment of the tryptophan ligand, crucial for high affinity operator binding (Otwinowski, Z., Schevitz, R. W., Zhang, R.-G., Lawson, C. L., Joachimiak, A., Marmorstein, R. Q., Luisi, B. F., and Sigler, P. B. (1988) Nature 335,321-329). Despite the conservative nature of these substitutions, it is apparent that relatively minor changes in * This work was supported by Grants GM 22441 and GM 35133 from National Institutes of Health and The Robert A. Welch Foundation Grant C-576 (to K. S. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
the chemical nature of the substituent and in the geometry of the side chain have profound effects on the functional properties of the altered protein.
The trp repressor protein in Escherichia coli regulates expression of three operons: trpEDCBA, aroH, and trpR (Bennett et al., 1976;Brown, 1968;Grove and Gunsalus, 1987;Gunsalus et al., 1986). The trpEDCBA and aroH operons code for aromatic amino acid biosynthetic enzymes (Brown, 1968;Yanofsky et al., 1981), whereas the trpR operon contains the coding region of the trp repressor so that the protein is autoregulatory (Gunsalus and Yanofsky, 1980;Kelley and Yanofsky, 1982;Bogosian et al., 1984). The binding of this protein to its cognate operator DNA sequences is modulated by the interaction of the protein with the corepressor tryptophan. Trp aporepressor is a small dimeric protein of 25 kDa which binds two molecules of tryptophan to form the operator binding conformation (Gunsalus and Yanofsky, 1980;Schevitz et al., 1985;Arvidson et al., 1986;Lane, 1986). The protein has been examined in detail by crystallographic methods, and the structures of the aporepressor, holorepressor, and holorepressor-operator DNA complex have been determined at high resolution Schevitz et al., 1985;Zhang et al., 1987;Otwinowski et al., 1988;). In addition, genetic studies have identified residues which are essential for the binding activity of the protein and which participate in corepressor binding (Kelley and Yanofsky, 1985;Bass et al., 1987Bass et al., , 1988Klig and Yanofsky, 1988). The structural changes which result in an activated protein in response to corepressor binding have been elucidated, and the participation of the corepressor ligand in DNA binding has been demonstrated (Zhang et al., 1987;Otwinowski et al., 1988;Marmorstein and Sigler, 1989).
The trp repressor contains 2 intrinsic tryptophan residues/ subunit (Gunsalus and Yanofsky, 1980;Singleton et al., 1980); however, examination of their fluorescence properties in response to ligand binding is complicated by the extrinsic fluorescence of the tryptophan indole, and all analogues which bind to the repressor contain the indole moiety (Lane 1986;Marmorstein et al., 1987). Furthermore, operator binding requires the presence of tryptophan so that monitoring DNA binding is also precluded using spectroscopic methods. The trp aporepressor contains no cysteine (Gunsalus and Yanofsky, 1980), and introduction of the reactive sulfhydryl moiety was undertaken as a means of providing a site for selective modification with fluorescent probes. The sites selected for the conservative change of serine to cysteine were positions near the binding site for corepressor and DNA (Fig. 1). These substitutions resulted in marked effects on both tryptophan and operator DNA binding and illustrate the complexity of designing subtle changes in protein structure with minimal effects on function.

RESULTS
Purification of Ser-Cys Mutant Aporepressors-The Ser-Cys mutant aporepressors were purified according to the procedures for wild-type protein with the omission of the phosphocellulose column. Since the DNA binding affinity was changed in all three mutant proteins, the phosphocellulose column, which separates based on DNA binding, was excluded from the purification procedures and was replaced by two other columns, DE52 and Cibacron FBG-A blue columns. The cysteine content of the purified proteins was determined by 5,5'-dithiobis-(2-nitrobenzoic acid) reaction. Based on one cysteine per subunit, approximately 8595% of the sulfhydryl in each mutant was accessible to 5,5'-dithiobis-(2-nitrobenzoic acid) reaction (S67C, 88%; S86C, 85%; S88C, 93%); these data indicated that the sulfhydryl groups in the mutant aporepressors neither formed disulfide bonds nor were oxidized during purification.
Tryptophan Binding of Mutant Proteins-To examine the tryptophan binding for all three mutant trp aporepressors, The "Materials and Methods" are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.
The results for several measurements are summarized in Table I. Equilibrium dialysis confirmed that the tryptophan binding constants for all three mutants were >1 X

M.
The insensitivity of ANS affinity to the mutations is in contrast to effects on tryptophan binding. All three mutants have decreased affinity for tryptophan at 25 "C. The S67C mutation has the least effect (-6-fold), whereas S88C is greatest (-%fold). The Horovitz-Levitzki plot for S86C is curved, indicating more than one class of sites which may derive from slight perturbations in the conformation of the binding region. The Kd reported is that for the highest affinity binding indicated by brackets in Fig. 2B; the slope of the curve corresponding to lowest affinity binding yields a Kd similar to S88C. The sulfhydryl group of the single cysteine allows facile introduction of fluorescent labels. All three mutants were reactedwith N-(iodoacetyl)-N'-(5-sulfo-l-naphthyl)ethylenediamine (IAEDANS). IAEDANS-labeled Ser4Cys mutant aporepressors were examined for effects of tryptophan binding on fluorescence. Both IAEDANS-labeled S86C and S88C exhibited a decrease of fluorescence at 480 nm in the presence of tryptophan (Fig. 3). However, S67C modified with IAE-DANS did not show any tryptophan-dependent fluorescence change. Titration of IAEDANS-labeled S86C and S88C with tryptophan was performed by exciting at 395 nm and measuring the fluorescence intensity at 480 nm (Fig. 4). The dissociation constants are summarized in Table I. Introduction of the bulky fluorophore at Cys-86 did not significantly affect tryptophan binding, whereas the presence of this group at position 88 decreased binding almost &fold.
Operator Binding-The operator binding affinity of the three SerjCys mutants was examined by the gel retardation assay using the 90-bp operator-containing DNA used previously by Carey (1988). In addition, a 40-bp synthetic operator, which contains trpEDCBA sequences from -29 to +10 on the top strand and from -28 to +11 on the bottom strand, was utilized. These sequences encompass those indicated by DNase protection and methylation perturbation to be involved in binding to the repressor (Kumamoto et al., 1987). NO difference in affinity between the 90-and 40-bp operators was noted. Smearing behind the free operator band and the decreased intensity of the band for the operator-repressor complex indicated that the mutant repressor-operator complexes dissociated more rapidly than the wild type (Figs. 5 and 6). The equilibrium dissociation constants estimated for the mutants by this method are summarized in Table I. Measurement of tm was not possible, as this reaction occurred more rapidly than detection allowed (tm < 10 s). For the S88C mutant, an additional band appeared on the gel when mutant repressor concentration exceeded 10 nM (Fig. 6A). However, this extra band did not appear on the gel in the presence of 1 mM of DTT in the running buffer (Fig. 6B). Unlike wild-type repressor, hydroxyl radical footprinting experiments (Carey, 1989;Tullius and Dombroski, 1986) failed to show significant protection for any of the mutants. Changing the concentrations of H202 or Fe(I1) did not alter the results. Disulfide Bond Formation-Sodium dodecyl sulfate-polyacrylamide gel electrophoresis in the presence of mercaptoethanol showed only a single band a t 12.5 kDa for the mutant proteins, whereas in the absence of mercaptoethanol a weak  (Carey, 1988). Reactions were performed under the standard conditions described by Carey (1988). additional band was observed (Fig. 7); this band was more prominent for S88C. Sephadex G-75 chromatography of wildtype and S88C mutant repressor with 1 mM DTT gave a single peak eluting at -38 kDa. In contrast, in the absence of DTT, a small peak corresponding to -44 kDa and a larger peak at  in lanes 1, 3, 5, and 8. B-Mercaptoethanol (5 p l ) was added to samples in lanes 2, 4, 6, and 9; lanes 1 and 2, wildtype trp repressor; lanes 3 and 4, S67C mutant; lanes 5 and 6, S86C mutant; lane 7, molecular weight standards; and lanes 8 and 9, S88C mutant.
-29 kDa were observed for mutant protein (data not shown). The formation of a disulfide bond may trap the higher molecular weight (possibly tetrameric) form of the repressor and allow it to be detected as a separate species.

DISCUSSION
Oligonucleotide-directed site-specific mutagenesis was used to construct three serine to cysteine trp aporepressor mutants. The rationale for changing serine to cysteine residues in the aporepressor derives from the similarity between sulfhydryl and hydroxyl moieties and the reactivity of the sulfhydryl group. Following chemical modification to introduce a fluorophore at this single site in the structure, the reacted protein could be used to study the interactions of the aporepressor with its ligands. The positions chosen for mutation were based on x-ray crystallographic data: the serine residues located in or near the ligand binding sites were changed to cysteine. There are 6 serine residues in trp aporepressor (Gunsalus and Yanofsky, 1980). Three of these serines are located in the termini of the subunit (at 5, 8, and 107) which are distal to both tryptophan and operator recognition sites ( Fig. 1; Schev itz et al., 1985;Zhang et al., 1987;Otwinowski et al., 1988). A fluorescent probe attached at those positions might not monitor any conformational effects upon ligand binding because of the rigidity of the quaternary structure of the aporepressor (Zhang et al., 1987). Ser-88 is hydrogen-bonded with the tryptophan so that this mutant would be expected to affect corepressor binding. Ser-67 and -86 were selected because their side chains are not directly involved in tryptophan binding but are close to the ligand recognition sites. The 0, of Ser-86 has recently been shown to be involved in operator binding .
All three serine to cysteine mutants showed decreased affinities for both tryptophan and operator DNA. The mutation of Ser to Cys at 67 gives a small change (-6-fold) in tryptophan binding. These data are in agreement with the x-ray crystallographic information which indicates that residue 67 is removed from the tryptophan binding site (Schevitz et al., 1985). In addition, no fluorescence change of IAEDANSlabeled S67C mutant upon tryptophan binding was observed, consistent with these results. The thiol group at position 67 may result in a subtle change in conformation of trp aporepressor. Such a change may derive from the difterences in the geometry of serine (C-0 bond distance = 1. 43  The mutant S86C showed a -4-fold increase in K d for tryptophan binding and a 130-fold increase in K d for operator DNA binding. The change in dissociation constant for operator DNA binding represents a loss of about 3 kcal/mol in free energy of binding at 25 "C. Since Ser-86 in the repressor makes a hydrogen bond directly with operator DNA (Otwinowski et al., 19881, the changed geometry of cysteine appears to interfere with this contact, and the increase in free energy change may correlate to the change for hydroxyl to sulfhydryl. The curvature of a Horovitz-Levitzki plot of tryptophan binding to this mutant implies a conformational distortion in the tryptophan binding region for this protein. The absence of an effect of IAEDANS modification on corepressor binding for this mutant would be anticipated from its position based on the crystal structure analysis (Schevitz et al., 1985;Zhang et al., 1987).
The mutant S88C showed the least change in operator binding and the most change in tryptophan binding. As shown by x-ray crystallographic data (Schevitz et aL, 1985;Zhang et al., 1987), Ser-88 forms a hydrogen bond with the a-amino group of bound tryptophan. Thus, the change in the mutant from OH -N to SH . . . N appears to yield an increase in the free energy change on complex formation by 1.7 kcal/mol at 25 "C. This major effect on tryptophan affinity does not appear to alter the ability of ligand binding to generate the DNA binding conformation of the repressor. However, the somewhat decreased operator affinity may be ascribed to the effects on tryptophan binding and improper alignment of the corepressor which in turn affects the configuration required for operator binding. In addition, the sulfhydryl may influence the backbone conformation and distort Ser-86 and/or Asn-87 contacts with the operator DNA .
The magnitude of the decrease in tryptophan binding observed upon IAEDANS modification presumably derives from a significant perturbation of the binding site as well as interruption of the hydrogen bond with the a-amino group of the ligand.
The effects of these three mutations on ANS binding affinity are minimal (Table I), in contrast to the marked alterations in tryptophan binding. Since ANS and tryptophan compete for the same site on the protein (Chou et al., 1989), it is noteworthy that the effects on their respective affinities differ considerably. This difference may arise from the discriminating contacts required for tryptophan binding, specifically the indole ring and a-carboxylate (Marmorstein et al., 1987). In contrast, ANS binds to hydrophobic pockets on many proteins and in a relatively nonselective fashion (Slavik, 1982). The sulfonate moiety may contribute to the binding in a mode analogous to the carboxylate, but the geometric requirements appear to be relaxed for ANS relative to tryptophan. Similar results have been observed with mutant trp repressors having altered amino acid residues in the tryptophan binding site: the ANS binding is diminished less significantly than tryptophan binding.3 The dissociation rates for all three mutants from operator DNA are faster than 0.06 s-' which is the upper limit of resolution for the gel retardation method. These fast dissociation rates may account for the observed increases in the values of equilibrium dissociation constants for operator binding. Hydroxyl radical footprinting using these proteins did not result in significant protection of the DNA backbone. The basis for the inability to obtain a hydroxyl radical footprint may derive from loss of sequence-specific contacts, especially for the S67C and S86C mutants. From x-ray crystallographic studies, Zhang et al. (1987) indicate that the trp repressor can be divided into three domains: central core, DNA-reading head, and hydrophobic brace. The central core consists of residues in the aminoterminal half of the molecule and residues 90-104 in both subunits. This central structure is not perturbed by the binding of L-tryptophan. The DNA-reading head is comprised of the helix-turn-helix motif which involves sequence positions 66-86 in the carboxyl-terminal region. This domain exhibits the principal conformational differences between trp aporepressor and trp repressor. The hydrophobic brace contains several hydrophobic amino acids within the region from 57 to 89. Compared with other DNA binding proteins, fewer bulky side chains characteristic of hydrophobic amino acid residues are found in the helix-turn-helix of trp aporepressor (Zhang et al., 1987). This local flexibility allows movement of the DNA binding region in response to tryptophan binding. The hydroxyl to sulfhydryl changes at positions 67 and 88 may weaken hydrogen bonds of the operator with other amino acids, such as Gln-68, Arg-69, Lys-72, Ser-86, Asn-87, and Lys-90 . These changes in hydrogen bonding may yield structural shifts which result in increased equilibrium dissociation constants and rates of dissociation of mutant proteins from operator DNA.
All three mutants showed a single protein band, corresponding to 12.5 kDa, on sodium dodecyl sulfate-polyacrylamide gels in the presence of 2-mercaptoethanol. However, in the absence of 2-mercaptoethanol, all three mutants showed two protein bands, corresponding to 12.5 and 25 kDa, respectively. These data imply that the serine to cysteine mutation enables J.-J. He and K. S. Matthews, manuscript in preparation. disulfide bond formation to occur to some extent between two monomers. This reaction is most prominent in the S88C mutant protein. Whether this disulfide linkage occurs within the dimer or derives from intermolecular interactions between dimers is not clear. However, the gel filtration results suggest the possibility of tetramer in rapid equilibrium with dimer. Data from differential scanning calorimetry and fluorescence polarization measurements are consistent with this possibility (Bae et al., 1988).4 Further experiments are in progress to examine this oligomer formation.
In summary, the substitution of cysteine for serine at three different sites within the trp repressor results in alterations in both corepressor and operator affinity. The changes observed can be rationalized in terms of the known threedimensional crystal structure of the protein, but the effects illustrate clearly the complexities involved in even a relatively conservative substitution. The formation of disulfide linkage between monomers and behavior of these mutant and wildtype proteins on gel filtration suggests the possibility of tetramer formation.