Formation of Heterodimers Between Wild Type and Mutant Trp Aporepressor Polypeptides of Eschem'chia Coli Thomas J

PROTEINS: Structure, Function, and Genetics 4:173-181 (1988)

Formation of Heterodimers Between Wild Type and Mutant trp Aporepressor Polypeptides of Eschem'chia coli Thomas J. Graddis,' Lisa S. Klig,' Charles Yanofsky,' and Dale L. Oxender' 'Department of Biological Chemistry, University of Michigan, Ann Arbor, Michigan 48109 and 'Department of Biological Sciences, Stanford University, Stanford, California 94305

ABSTRACT Availability of the three-dimen- tophan.1-6 The trp aporepressor is activated by the sional structure of the trp repressor of Escherichia binding of two molecules of its corepressor. Once ac- coli and a large group of repressor mutants has per- tivated, trp repressor binds to the operators of the trp, mitted the identification and analysis of mutants aroH, and trpR operons, regulating transcription ini- with substitutions of the amino acid residues that tiati~n.~,'The aroH and trp operons encode biosyn- form the tryptophan binding pocket. Mutant apore- thetic enzymes, whereas the trpR operon encodes the pressors selected for study were overproduced using trp aporepressor. The trpR gene has been cloned and a multicopy expression plasmid. Equilibrium di- its nucleotide sequence determined.3 The aporepres- alysis with ''C-tryptophan and purified mutant and sor and repressor have been purified and the crystal wild type aporepressors was employed to determine structures of both have been solved at high resolu- tryptophan binding constants. The results obtained ti~n.~-"The trp aporepressor is a dimer of identical indicate that replacement of threonine 44 by methi- 107 residue polypeptides." The existence of the three- onine (TM44) or arginine 84 by histidine (RH84) low- dimensional structures, and the availability of many ers the affinity for tryptophan approximately two- repressor mutants,2i12has facilitated this analysis of and four-fold, respectively. Replacement of arginine the in vivo and in vitro tryptophan-binding activity 54 by histidine (RH84) or glycine 85 by arginine of mutant repressors with changes of those amino (GR85) results in complete loss of tryptophan bind- acid residues predicted to participate in tryptophan ing activity. Purified mutant and wild type apore- binding. pressors were used in in vitro heterodimer studies. A class of trpR mutants was identified that have a The trp repressor of E. coli functions as a stable transdominant effect when present in a strain bear- dimer. A large number of trp repressor mutants pro- ing a chromosomal copy of the wild type repressor duces defective repressors that are transdominant to gene." The transdominant phenotype (trpR-d) is the wild type repressor in vivo. The transdominance thought to result from the formation of heterodimers presumably results from the formation of inactive or defective in repressor activity.13 Analogous results slightly active heterodimers between the mutant and have been observed with mutant forms of other re- wild type polypeptide subunits. An in vitro assay was pressors, such as the lac repres~or.'~,~~The chain ter- developed to detect and measure heterodimer for- minating trpR nonsense mutant Qam68 produces a mation. Heterodimer formation was thermally in- truncated polypeptide lacking the C-terminal 40- duced, and heterodimers were separated on amino acid residues that contain the polypeptide seg- nondenaturing polyacrylamide gels. Aporepressors ment involved in DNA recognition. Since the Qam68 readily formed heterodimers upon treatment at 65°C mutant displays strong transdominance, it is likely for 3 minutes. Heterodimer formation was signifi- that the first 67 residues of the trpR polypeptide are cantly retarded by the presence of the corepressor, sufficient for negative complementation. L-tryptophan. Indole-3-propionic acid, 5-methyl The majority of the trpR-d missense mutations tryptophan, and other analogs of tryptophan, as well (71%) map to a gene segment corresponding to the D- as indole, also inhibited heterodimer formation. helix-turn-E-helix region of the repressor, which is These results indicate that the presence of the indole believed to interact directly with operator DNA. A moiety in the corepressor binding pocket increases second group of transdominant missense mutations the stability of the dimer. results in the substitution of residues that form the surface of the corepressor binding Key words: DNA binding protein, ligand binding, In the present study we have purified four mutant equilibrium dialysis, dimer, stability trp aporepressors that have substitutions of residues believed to comprise the corepressor binding site. Us- INTRODUCTION The trp repressor of Escherichia coli regulates Received December 14, 1987; revision accepted July 28, 1988. expression of genes required for the de novo biosyn- Address reprint requests to Dr. Dale L. Oxender, Department of Biological Chemistry, University of Michigan Medical thesis of tryptophan in response to changes in the School, M-531910606, 1301 Catherine Road, Ann Arbor, MI intracellular concentration of its corepressor, L-tryp- 48109-0606. 0 1988 ALAN R. LISS, INC. 174 T.J.GRADDIS ET AL. ing equilibrium dialysis analyses, we showed that yields were obtained with each repressor, indicating these mutant aporepressors are partially or totally that none was particularly labile. Dialyzed aporepres- defective in L-tryptophan binding. We also developed sor preparations were concentrated using an Amicon an in vitro procedure for generating heterodimers ultrafiltration apparatus and P10 membranes. Pro- from purified wild type and mutant trp aporepressors. tein concentration was determined spectrophotomet- The net charge differences between wild type apore- ri~a1ly.l~The repressor was stored frozen at -20°C pressor and some mutant species, and their hetero- in buffer containing 0.2M NaCl at a protein concen- dimers, allowed us to separate heterodimers elec- tration of 4-7 mg/ml. trophoretically on nondenaturing polyacrylamide gels, and to quantitate their formation. We have dem- Equilibrium Dialysis onstrated that L-tryptophan and certain tryptophan The affinity of L-tryptophan for pure wild type and analogs inhibit heterodimer formation. mutant trp aporepressors was determined using equilibrium dialysis measurements with L-[methylene- MATERIALS AND METHODS 14C] tryptophan (sp. act. 53.5 mCi/mmol, Amersham). Chemicals Strains Plasmids, and Media An eight-place Hoeffer equilibrium microdialyzer L-tryptophan, D-tryptophan, L-phenylalanine, D,L- module with EMD104 membranes (6,000-8,000 dal- tryptophan, D,L-5-methyltryptophan,tryptamine, in- ton cutoff),driven by a rotary motor, was used in this dole, and indole-3-propionic acid were purchased from study. Each chamber had a capacity of 100 pl on each Sigma. Overexpressing plasmids were constructed by side of the dialysis membrane. The dialysis solution incorporating BamHI fragments containing the var- contained 10 mM Tris-HC1, pH 7.8, and 0.1 M NaCl. Aporepressor at a final concentration of 40 pM was ious mutant trpR alleles into expression plasmid ptacterm, as describe9.16 In this plasmid, the tac pro- placed on one side of the chamber, and various con- moter drives trpR mRNA production. The recipient centrations of L-[methylene-14C] tryptophan were placed on the other side. Dialysis was carried out strain used for overproduction of the trp aporepressor overnight (16-18 hours) at 4°C. All chambers were was CY15071 (W31 10 tnaAZAtrpR thr lac@). This emptied, and 10 p1 aliquots of each were counted in a strain has trpR deleted and carries lacZQ for control scintillation counter. of the tac promoter of the expression plasmid. Each expression plasmid was introduced into CY15071, and trp aporepressor synthesis was induced by addition of Samples for Heterodimer Formation IPTG, as described.I6 Standard conditions for formation of heterodimers Repression was measured in vivo using a lysogenic were as follows: two purified trp aporepressors, at the recipient, strain CY15058 (W3110 tnaA2 trpR thr- same concentration, were combined, so that 0.15 AlacU169/XTLFl),in which the single integrated copy nmoles of total dimeric protein (3.75 pg protein) were of the lambda genome contains a trp promoter/opera- present in a final volume of 3-6 pl(25-50 pM) in 100 tor trpL'-'lacZ gene fusion. Production of hybrid (3- mM Tris buffer (pH 6.81, 50-100 mM NaC1. Where galactosidase by this strain is controlled by repres- indicated, L-tryptophan was added. Samples were sion of transcription initiation at the trp promoter of then heated in water baths at various temperatures XTLF1. To measure in vivo repression by any of the for various times. After heating, samples were cooled mutant repressors, derivatives of pRLK13" contain- rapidly in an ice bath. Three microliters of 80% glyc- ing the various mutant alleles were introduced into erol with 0.0025% marking dye were added to each CY15058, and bacterial cultures were grown in min- sample before loading the entire sample on a gel. The imal mediurnl7 containing 0.2% glucose, 0.2% casein rate of heterodimer formation was not affected by hydrolysate, with or without 40 pg/ml tryptophan, as changing the pH of the Tris buffer from 6.8 to 8.8. indicated. Culture media generally contained either Samples to be examined for heterodimer formation 30 pg/ml chloramphenicol or 100 pg/ml ampicillin to induced by ethanol were prepared essentially as stabilize the appropriate plasmid. Plasmid pRLK13 above, except that tryptophan was added prior to the is a derivative of pACYC184 in which trpR expression ethanol, and samples were incubated at either 22°C is driven by the tet promoter." or 37°C for 15 minutes. (3-GalactosidaseAssays Nondenaturing Polyacrylamide Gel Strain CY15058 containing plasmids pRLK13, Electrophoresis pTM44, pRH54, pRH84, or pGR85 was grown at 31°C (the XTLFl prophage is a XcI857 derivative) in the Both alkaline continuous and neutral discontin- media described above and (3-galactosidase assays uous native polyacrylamide gel buffer systems were were performed as described by Miller.I8 used.20*21Neither of these buffer systems induced heterodimer formation. On the other hand, alkaline dis- Purification of Mutant trp Aporepressors continuous and neutral continuous electrophoretic Mutant aporepressors were overproduced in E. coli conditions gave results that were not always repro- and purified as previously described. l6 Comparable ducible (data not shown). TRP REPRESSOR HETERODIMER FORMATION 175

Measurement of Heterodimer Formation by Tryptopan Binding by Mutant and Wild-Type trp Densitometry Repressors Coomassie stained gels were scanned with a Biomed The four mutant aporepressors were overproduced Instruments Soft Laser Scanning Densitometer. The to the same extent in E. coli, suggesting that they are data were plotted and integrated using a Biomed equally stable. They also behaved similarly to the Instruments Videophoresis Analysis Program in con- wild type repressor during purification. The electo- junction with an Apple IIe computer system. The phoretic mobility of the mutant polypeptides during inset in Figure 1C shows that the fraction of total SDS-PAGE was indistinguishable from that of the protein determined spectrophotometrically19 agreed wild type polypeptide, indicating that they are nor- with the fraction of the total signal obtained from a mal length. The electrophoretic mobilities of the mu- densitometry trace of stained bands produced from tant aporepressors in nondenaturing gels were serially diluted protein. The linearity of signal re- consistent with their being stable dimers, with the sponse permitted the fraction of total protein per sam- expected charge differences (Figs. 1and 2). Formation ple to be determined. The fractional abundance of of heterodimers with intermediate inobilities (see be- homodimers and heterodimers per sample was calcu- low) supports this interpretation. Negative comple- lated by scanning the total signal from each sample. mentation analyses in vivo12 also indicate that mutant aporepressor polypeptides can readily form heterodimers with the wild type polypeptide. Taken RESULTS together, these findings suggest that the dimeric mu- In Vivo Activity of Mutant trp Repressors tant aporepressors are structurally similar to the wild The in vivo activities of four mutant trp repressors, type protein. constitutively expressed from the tet promoter of The binding of L-tryptophan by pure wild type and pACYC184, were determined using cultures growing mutant repressors was measured using equilibrium in medium with and without tryptophan supplemen- dialysis (see Materials and Methods). Wild type, tation. It can be seen in Table I that the four mutant TM44, and RH84 trp aporepressors bind two mole- repressors have decreased activity in vivo, in the cules of tryptophan in a noncooperative manner. The presence or absence of exogenous tryptophan. Two Ks value of 23 + 5 pM for wild type repressor is repressors, TM44 and RH84, showed a slight trypto- essentially indistinguishable from the previously re- phan response in vivo, while GR85, which was slightly ported binding constants measured at 4°C."2,,23The active, was unaffected by the presence of the trypto- Ks values obtained with the four mutant aporepres- phan supplement. Preliminary in vitro filter binding sors are listed in Table I. All four are defective in experiments indicate that RH54 repressor does not tryptophan binding; aporepressors RH54 and GR85 bind operator DNA within the limits of detection (L.S. are devoid of tryptophan binding activity in the pres- Klig, unpublished results). We would not expect any ence of 1 mM L-tryptophan, while the TM44 and of these mutant repressors to bind operator DNA in RH84 aporepressors have only slightly reduced bind- vitro. ing activity. The TM44 and RH84 repressors respond

TABLE I. The Amino Acid Replacements in trpR Mutants Presumed Altered in the L-Tryptophan Binding Site, Tryptophan Binding, and In Vivo Repressor Activity*

Amino acid In vivo 0-gal activity Wild type trpR Allele KS (mutant) Position on plasmid -TrP +TrP L-tryptophan (pM) None 12,000 12,000 - Wild type 500 10 23 + 5 Threonine 44 TM44 9,000 5,900 39 i 7 (methionine) Arginine 54 RH54 11,000 11,200 N.B. (histidine) Arginine 84 RH84 10,100 5,300 88 i 4 (histidine) Glycine 85 GR85 2,100 2,000 N.B. (arginine) *Wild type and mutant plasmids were introduced into a trpR- strain containing a repression indicator (A trp promoterloperator-trpL'-' lac2 fusion), and 0-galactosidase (&gal) activities were determined using cultures grown with and without 20 pglml L-tryptophan. The Ks values for tryptophan binding, obtained from duplicate equilibrium dialysis measurements with pure aporepressors, are presented in the last column. N.B. stands for no detectable binding up to a concentration of 1 mM L-tryptophan. 176 T J. GRADDIS ET AL A abcde f gh i j k Im WT-1

RH84- - Time (min) 5500.5 I2 5 10‘- -65-2237506580 Tern p (OC

0.5 - B

MINUTES AT 65°C 1 2 ’ I IO-’g PROTEIN 0.0 ’ ’ 0 1 2 3 4 5

MINUTES AT 65°C

Fig. 1. A: Native alkaline continuous polyacrylamide gel RH84 protein (concentration determined spectrophotometrically) stained with Coomassie brilliant blue R. Samples of wild-type against the fraction of total signal (n).A least squares calculation (WT) and mutant (RH84) aporepressors were treated at times and of these data yielded the line marked “observed.” The fraction of temperatures indicated. 6: Densitometer traces used to obtain total protein is denoted by the line marked “predicted.” It is the this plot. C: Plot of percent total protein from the integrated values amount of protein per sample in the serial dilution (determined obtained from a densitometer trace: 0,WTIRH84 heterodimer; spectrophotometrically) divided by the total amount of protein in A,WT hornodimer; U,RH84 homodimer. Inset shows a plot of the combined samples. The arrow denotes one-half the total the data obtained from a densitometer trace of a serial dilution of protein concentration per sample.

slightly to tryptophan in vivo, in agreement with the dimer bands are observed, as shown in Figure lA, results of these in vitro binding studies. lane c. If the mixture was heated at 65°C for several minutes, a third heterodimer band appeared, which Heterodimer Formation had an intermediate mobility reflecting its interme- Homodimers of purified mutant and wild type (WT) diate charge (see lanes d-h in Fig. 1A). Lanes a and b aporepressors may be separated according to their show that homodimers of WT or mutant protein do charge differences on either an alkaline continuous not produce this heterodimer band when treated in- polyacrylamide gel (Figs. 1A and 2A) or a neutral dividually at 65°C. Homodimer depletion was con- discontinuous nondenaturing polyacrylamide gel (Fig. comitant with heterodimer formation, as shown by 3). Mutant RH54 and RH84 repressors have two ad- the densitometry trace in Figure 1B and by the plot ditional negative charges per homodimer relative to of the integrated values presented in Figure 1C. All wild type dimer and thus migrate to the anode faster, combinations of repressor mixtures assayed yield a whereas mutant repressor GR85 has gained two pos- final equilibrium distribution of homodimer to heter- itive charges and migrates more slowly. Mutant apo- odimer to homodimer of 1:2:1, which reflects the sta- repressor TM44 is isoelectric with wildtype and tistical distribution expected if association of therefore exhibits wild type mobility. We chose mu- individual polypeptide chains was random. The rate tant repressor RH84 for our initial studies of hetero- of heterodimer formation was similar for all combi- dimer formation with the WT aporepressor. Many of nations tested, implying that this property is inde- the conclusions reached with this mixture are appli- pendent of ability to negatively complement WT cable to mixtures with other mutant repressors. When repressor when tested in vivo. Complete inhibition of purified samples of mutant and WT aporepressors heterodimer formation was achieved at high salt con- were mixed and treated at 22T, only the two homo- centration, i.e., 0.6M NaC1. The high salt concentra- TRP REPRESSOR HETERODIMER FORMATION 177 erodimer formation. The Ks for tryptophan for wild type and RH84 repressors are 23 pM and 88 pM, abcdefg respectively. The data in Figure 2A show that as the WT- tryptophan concentration was increased, heterodimer formation was inhibited. A concentration of 2 mM L- 4 ro tryptophan, determined from the plot presented in RH84' Figure 2B, was required to give one-half the maximal inhibition of heterodimer formation (Ksr). This con- om5a5 I 3 5 20 centration of tryptophan was required for all mixtures that included the WT aporepressor. Tryptophan A x 10-% TRYPTOPHAN was effective in inhibiting heterodimer formation whenever one homodimer species could bind corepressor. For example, RH54 by itself does not bind core- 0.5 q pressor, but in a WT-RH54 mixture, the presence of i 20 mM L-tryptophan completely inhibits the formation of the heterodimer. Mutant RH54 and GR85 aporepressors do not bind tryptophan. When a mixture of these aporepressors was incubated at 65°C for 3 minutes in the presence or absence of 20 mM L-tryptophan, the rate of heterodimer formation was not affected (data not shown). Mutant aporepressors RH84 and TM44 bind L-tryptophan with dissociation constants of 88 pM and 39 pM at 25"C, respectively. When mutant TM44 and RH84 aporepressors were mixed, a somewhat higher r~o-'~TRYPTOPHAN concentration of tryptophan, 5 mM, than that required with a mixture of RH84 and WT aporepressors was required to give 50% inhibition of heterodimer formation (data not shown). These results correlate 0.1 I ' ' . ' ' ' ' ' 10 15 20 well with the corepressor binding properties of these mutant proteins and support the contention that in- B x IO-~PJ TRYPTOPHAN hibition of heterodimer formation results specifically from L-tryptophan binding. Fig. 2. A: Native alkaline continuous polyacrylamide gel stained with Coornassie brilliant blue R. Mixtures of WT and RH84 pro- Ethanol-Induced Heterodimer Formation teins were treated for 3 minutes at 65OC.The indicated concentration of L-tryptophan was added prior to heating. The arrow indicates the heterodirner band. 6: Plot of protein content of each The addition of ethanol to a homodimer mixture band as a function of the tryptophan concentration of the reaction also results in heterodimer formation. The extent of mixture obtained from integrated values of densitorneter traces: heterodimer formation was dependent on the concen- U, WTIRH84 heterodirner; A,WT homodirner; W, RH84 horno- dirner. The value of 50% maximal inhibition of heterodimer for- tration of ethanol and on temperature. Complete het- mation (KsJ was determined from this plot to be 2 mM tryptophan. erodimer formation between mixtures of WT-RH84 or The inset is the densitorneter trace used to obtain this plot. WT-RH54 was achieved with the addition of 26% tion may prevent dimer dissociation by strengthening ethanol at 22T, while 20% ethanol was sufficient at hydrophobic interactions between polypeptides of the 37°C (data not shown). As with thermal denatura- dimer. tion, the presence of 0.6 M NaCl prevented dimer Figure 1A also shows the effect of varying temper- dissociation induced by ethanol. These results rein- ature on heterodimer formation at a constant time of force the conclusion that heterodimer formation in- 3 minutes. Heterodimer was not detected when sam- volves disruption of hydrophobic interactions between ples were incubated at 50°C for 3 minutes, although the polypeptides of trp aporepressor. substantial formation occurs at 60°C and maximum L-tryptophan prevented heterodimer formation in- levels were reached at 65°C. An incubation tempera- duced by ethanol. When a WT-RH84 mixture was ture of 80°C did not change the final distribution of assayed at 37°C in 20% ethanol, 4 mM tryptophan dimer species but increased the rate, so that maxi- completely prevented formation of heterodimer (data mum formation occurs within 1 minute. Samples not shown). When 20 mM tryptophan was present in treated at 37°C for up to 8 hours failed to show a 65°C sample assay, heterodimer formation was also significant heterodimer formation (data not shown). prevented however, the addition of 5% ethanol induced heterodimer formation. Effect of Tryptophan on Heterodimer Formation Tryptophan was added to the mixture of RH84 and Dissociation of Heterodimer wild type aporepressors to examine its effect on het- We also examined dissociation of heterodimer to 178 T. J. GRADDIS ET AL a b C d

Wl= RH84- Temperat urerc] Tryptophan

Fig. 3. Native neutral discontinuous polyacrylamide gel at 65OC for 3 minutes (lanes a, c, and d). The sample in lane d stained with Coomassie brilliant blue R. A heterodimer band was was soaked in running buffer to which 20 mM tryptophan was excised from an unstained native alkaline continuous gel, soaked added. The arrow indicates the heterodimer band. in running buffer of the neutral gel for 15 minutes, and incubated

form the respective homodimers. Bands representing Compound Structure K,(PM) kXmM) heterodimer formed between WT and RH84 subunits ,COT were excised from samples run on an alkaline contin- L-tryptophan 2 uous gel and then incubated with and without L- tryptophan and electrophoresed on a neutral discon- 5-methyl-L- 3 0.2 tinuous gel. The results of this experiment are shown tryptophan in Figure 3. In lane b, the heterodimer did not disso- indole-3- 0.8 ciate when samples were not heated. The emergence propionic acid of homodimers from heterodimers occurred when

samples were heated at 65°C (lane c). If the gel slices indole 25 4 were infused with 20 mM L-tryptophan, the thermal disruption of WTRH84 heterodimer was retarded (lane d). We observed similar results with the WT/ RH54 heterodimer. This latter species has only one wild type corepressor binding site, which apparently D-tryptophan 340 NA &I::N is all that is required for tryptophan inhibition of subunit dissociation. Fig. 4. Half-maximal inhibition of WT/RH84 heterodimer for- Effect of Tryptophan Analogs on Dimer Stability mation (Ksl) by L-tryptophan and its analogs at 65OC. The Ks, values obtained are listed with their structures alongside the The trp aporepressor shows significant affinity for dissociation constants (Ks) determined at 4OC by Marmorstein, et al.23 The Ksl values were derived from integration of densito- a number of tryptophan analog^.'^ The KSI values meter traces much the same as shown in Figure 28. The Ksl reported in Figure 4 are the concentrations required value given for 5-methyl-L-tryptophan is based on the assumption for half-maximal inhibition of heterodimer formation that the D isomer does not bind. The range of L-tryptophan Ks, values are within 15% of the value listed. NA stands for no affect with a mixture of WT and RH84 aporepressors, as- at a concentration of 20 mM. sayed following 3 minutes of incubation at 65°C. The WT aporepressor binding constants for these analogs, obtained at 4"C, were previously reported by Mar- morstein, et aLZ3Aporepressor affinity for these li- tryptophan, Ks of 3 pM at 4"CZ3As shown in Figure gands corresponds in general to the concentrations 4, the Ksr for 5-methyltryptophan is one-tenth that of required to inhibit heterodimer formation. These re- tryptophan. Indole-3-propionic acid (IPA) binds to trp sults show a high specificity for L-tryptophan and aporepressor to yield an inactive complex.25IPA binds certain analogs, since neither 20 mM D-tryptophan to the aporepressor 1.5 times more tightly than core- nor 20 mM L-phenylalanine prevent WT/RH84 het- pressor.25 The KSI of IPA (0.8 mM) reflects this tigh- erodimer formation. The analog 5-methyltryptophan ter binding relative to the KsI of 2 mM observed for functions as a corepressor in v~vo.'~This analog binds L-tryptophan. At 6 mM, IPA completely inhibited to the WT aporepressor with greater affinity than L- heterodimer formation. The analogs tryptamine and TRP REPRESSOR HETERODIMER FORMATION 179 r4 f4

ARC

Fig. 5. Stereo view of one of the L-tryptophan binding sites in residues surrounding the corepressor, tryptophan. The figure was the trp repressor showing the position of the four amino acid generated by the drawing program of A. Lesk and K. Hardman." indole have Ksr's of 4 mM in the heterodimer assay; tryptophan binding site is contributed by the DNA these KsI's correspond to the aporepressor's weaker recognition helix, helix E, it is conceivable that the affinity for these species, 25 pM and 41 pM, mutant Met residue displaces this helical region. Re- re~pectively.~~ garding the RH84 change, Arg 84 is within helix E, the helix that is presumed to interact specifically DISCUSSION with the operator. Thus the RH84 change may cause From the three-dimensional structure of the L-tryp- localized positional changes in helix E that are detri- tophan-trp aporepressor complex (the trp repressor), mental to operator binding, in addition to slightly the following amino acid residues are predicted to reducing tryptophan binding. participate in specific binding of the corepressor, L- It is interesting to note that the GR85 mutation tryptophan: Thr 44, Arg 54, Arg 84, and Gly 85 (Fig. allows some repressor activity in vivo (Table I) despite 5). Each of the two L-tryptophan binding sites in the the fact that we cannot detect tryptophan binding in trp aporepressor is formed from different segments of vitro. Examination of a space-filling atomic model of the two identical polypeptides. Thus, for a given co- the tryptophan binding site reveals that the base of repressor binding pocket, Thr 44 is from one polypep- the indole moiety of the corepressor is situated close tide chain, while Arg 54, Arg 84, and Gly 85 are from to Gly 85. Thus one would predict that virtually any the second. Among the set of repressor mutants iso- substitution at this position would inactivate the re- lated previously, mutants were obtained that had pressor. Perhaps the bulky arginine residue intro- amino acid changes at each of these positions.'2 The duced at position 85 in GR85 causes conformational amino acid changes in these mutants are listed in shifts in the protein segments normally responsible Table I. for corepressor binding. Other amino acid changes at Repressor activities measured in vivo indicate that position 85, GK85 (Gly-tLys) and GE85 (Gly+Glu) none of the mutant aporepressors is appreciably acti- totally inactivate the repressor." Further analyses of vated by tryptophan. Equilibrium dialysis analyses the effects of selected amino acid substitutions at show that only two of the mutant repressors, TM44 these positions in the trp repressor may clarify the and RH84, have near-normal tryptophan binding ac- requirements for the critical conformational changes tivity. Of the four mutant repressors studied, only in the protein that normally accompany tryptophan these respond to tryptophan in vivo, and the response binding. is slight. What is perhaps most striking about these The trp repressor forms a highly stable symmetric results is that two repressors with only a modest dimer composed of identical 107 residue polypeptides. reduction in tryptophan affinity nevertheless have Each polypeptide subunit is composed of six or-helices very little repressor activity in vivo in the presence (A-F), five of which make contacts with the apposing of tryptophan. Clearly these proteins are being satu- subunit." The hydrophobic interactions between sub- rated with corepressor, but are ineffective as repres- units of the trp repressor are extensive. If we assume sors. Thus either Thr 44 and Arg 84 have other an individual subunit has the same conformation essential roles in repressor function, or the residues found in the trpR dimer, then only 4% of its residues replacing them, Met 44 and His 84, respectively, dis- are excluded from solvent, atypical of the 20% found turb essential structural features of the protein while in monomers of most oligomeric proteins.23 In the still allowing high-affinity tryptophan binding. Con- dimeric form, however, 18% of the residues of the sidering these changes separately, little can be said repressor are buried.23 This architecture presumably about the TM44 replacement. Since one surface of the accounts for the marked dimer stability, since the 180 T.J.GRADDIS ET AL. hydrophobic forces usually found in the interior of a erodimer formation presumably depends on its hydro- globular protein are shared between subunits in the phobic nature and/or its ability to form a hydrogen trp repressor. Nuclear magnetic resonance (NMR) bond at its heterocyclic nitrogen. It has been proposed analysis suggests that as temperature is increased that the indole nitrogen of tryptophan forms a hydro- from 25°C to 70°C the trp aporepressor retains its gen bond with the aporepressor in a nonpolar environ compact, folded structure and that its melting tem- of the corepressor binding pocket, resulting in large perature is greater than 70°C.26,27In our study, how- negative AS and AH values associated with its bind- ever, wild type and mutant homodimers of trp ing." Structural analysis of repressor shows that the repressor can exchange subunits to form heterodi- indole nitrogen of tryptophan forms a hydrogen bond mers in vitro by heating at 65°C or by addition of with an ordered water molecule.28 20%ethanol at 37"C, conditions that reduce intersub- Perhaps the hydrophobic nature of the indole moiety unit hydrophobic forces. The dissolution of hydropho- of tryptophan, IPA, tryptamine, or indole itself ac- bic bonds between polypeptide subunits of the trp counts for increased homodimer stability. When the repressor apparently permits the exchange of sub- corepressor binds, it displaces several water mole- units between homodimers. This exchange leads to cules and causes the microenvironment of the binding an equilibrium distribution of homodimers and site to be less polar in the liganded specie^.^ Since the heterodimer. corepressor binding site is formed between both sub- Even though the binding of tryptophan is thermally units, the ability of indole to stabilize the dimer may sensitive, significant binding activity is detectable at be due to the addition of a hydrophobic patch to the 65°C. An extrapolation of the van't Hoff plot pre- subunit interface. It is also possible that corepressor sented by Arvidson et a1.22 suggests that aporepres- binding may inhibit heterodimer formation indirectly sor would bind tryptophan with a Ks of 0.7 mM at by strengthening bonds that already exist between 65°C. In this study, we find that upon binding L- subunits. By shielding preexisting intersubunit elec- tryptophan, or one of its analogs, the thermal stabil- trostatic or ionic bonds from the solvent, tryptophan ity of the repressor dimer increases significantly. would significantly strengthen subunit interaction. Since the affinity for tryptophan is reduced at ele- The fact that the analog concentration required for vated temperatures, concentrations as high as 20 mM inhibition of heterodimer formation correlates with tryptophan are required to completely inhibit subunit affinity for the aporepressor suggests that every fea- dissociation. Significant retardation of subunit disso- ture that contributes to analog binding also influ- ciation is observed when the heterodimer RH84/WT, ences stability. Thus the additional aporepressor- which has one wild type site and one low-affinity site, ligand interactions responsible for higher affinity is isolated and heated a second time in the presence binding of analogs such as 5-methyltryptophan and of L-tryptophan. This result suggests that only one of IPA would also be expected to increase stability of the the two corepressor binding sites per trpR dimer need analog-containingrepressor. be occupied by L-tryptophan in order to stabilize intersubunit binding. The concentrations of tryptophan analogs required to inhibit heterodimer formation (Fig. 4) correlate ACKNOWLEDGMENTS well with their affinity for the aporepressor. However, the molecular details by which bound tryptophan or The assistance of Jan Paluh with the ptacterm plas- its analogs stabilize subunit interaction are un- mid and in the purification of aporepressors is greatly known. First of all, the increased repressor stability appreciated, We would like to thank Cathy Lawson in the presence of tryptophan would be expected to for supplying the stereo drawing of the tryptophan result primarily from interactions in the vicinity of binding pocket. This work was supported by the Na- the corepressor binding site, since crystallographic tional Science Foundation (DMB8703685) and The studiesghave shown the absence of quaternary shifts American Heart Association (69-015). D.L.O. was an along essentially the entire subunit interface when American Cancer Society Scholar. L.S.K. was a fellow tryptophan binds. It is also reasonable to presume of the Jane Coffin Childs Memorial Fund for Medical that the same mechanism of stabilization of the di- Research. C.Y. is a Career Investigator of the Ameri- mers applies to all of the analogs tested. The crystal can Heart Association. structure of pseudorepressor supports this assumption, since the indole ring of IPA occupies approximately the same position as that of L-tryptophan in REFERENCES repressor, only it is "flipped over" 180" with respect to the long axis of the indole moiety." Comparison of 1. Bennett, G.N., Schweingruber, M.E., Brown, K.D., Squires, C., Yanofsky, C. Nucleotide sequence of region preceding the structures of the analogs indicates that the indole trp mRNA initiation site and its role in promoter and oper- moiety is sufficient to stabilize subunit interaction, ator function. Proc. Natl. Acad. Sci. USA 73:2351-2355, i.e., the a-carbon functional groups of bound trypto- 1976. 2. Bennett, G.N., Yanofsky, C. Sequence analysis of operator phan are not required for inhibition of subunit dis- constitutive mutants of the tryptophan operon of Esche- sociation. The action of indole in preventing het- richia coli. J. Mol. Biol. 121:179-192, 1978. TRP REPRESSOR HETERODIMER FORMATION 18 1

3. Gunsalus, R.P., Yanofsky, C. Nucleotide sequence and 16. Paluh, J.L., Yanofsky, C. High level production and rapid expression of Escherichia coli trpR, the structural gene for purification of the E. coli trp repressor. Nucleic Acids Res. the trp repressor. Proc. Natl. Acad. Sci. USA 77:7117-7121, 14:7851-7860, 1986. 1980. 17. Vogel, H.J., Banner, D.M. Acetylornithinase of Escherichia 4. Zurawski, G., Gunsalus, R.P., Brown, K.D., Yanofsky, C. colt Partial purification and some properties. J. Biol. Chem. Structure and regulation of aroH, the structural gene for 218:97-106, 1956. tryptophan-repressible 3-deoxy-D-arabino-heptulosonicacid- 18. Miller, J.H. “Experiments in Molecular Genetics.” Cold 7-phosphate synthetase of Escherichia coli J. Mol. Biol. Spring Harbor: Cold Spring Harbor Laboratory, 1972. 145:47-73, 1981. 19. Joachimiak, A., Kelley, R.L., Gunsalus, R.P., Yanofsky, C., 5. Grove, C.L., Gunsalus, R.P. Regulation of the aroH operon Sigler, P.B. Purification and characterization of trp apore- of Escherichia coli by tryptophan repressor. J. Bacterial. pressor. Proc. Natl. Acad. Sci. USA 80:668-672, 1983. 169:2158-2164,1987. 20. Williams, D.E., Reisfeld, R.A. Disc electrophoresis in poly- 6. Klig, L.S., Crawford, I.P., Yanofsky, C. Analysis of trp re- acrylamide gels: Extension to new conditions of pH and pressor-operator interaction by filter binding. Nucleic Acids buffer. Ann. N.Y. Acad. Sci. 121:373-381, 1962. Res. 155339-5351,1987. 21. Blackshear, P.J. Systems for polyacrylamide gel electropho- 7. Yanofsky, C. Comparison of regulatory and structural re- resis. Methods Enzymol. 104:237-255, 1984. gions of genes of tryptophan metabolism. Mol. Biol. Evol. 22. Arvidson, D.N., Bruce, C., Gunsalus, R.P. Interaction of the 1:143-161, 1984. Escherichia coli trp aporepressor with its ligand, L-trypto- 8. Brown, K.D. Regulation of aromatic amino acid biosyn- phan. J. Biol. Chem. 261:238-243, 1986. thesis in Eschericia coli K12. Genetics 60:31-48. 1968. 23. Marmorstein, R.Q., Joachimiak, A,, Sprinzl, M., Sigler, P.B. 9. Zhang, R.G., Joachimiak, A,, Lawson, C.L., Otwinowski, Z., The structural basis for the interaction between L-trypto- Sigler, P.B. The crystal structure of trp aporepressor at 1.8A phan and Escherichia coli trp aporepressor. J. Biol. Chem. shows how binding tryptophan enhances DNA affinity. Na- 262:4922-4927, 1987. ture 327:591-597,1987. 24. Doolittle, W.F., Yanofsky, C. Mutants of Escherichia coli 10. Lawson, C.L., Zhang, R.G., Schevitz, R.W., Otwinowski, Z., with altered tryptophanyl-transfer ribonucleic acid synthe- Joachimiak, A,, Sigler, P.B. Flexibility of the DNA-binding sis in viva. J. Bacteriol. 951283-1294, 1971. domains of trp repressor. Proteins 3:18-31, 1988. 25. Morse, D.E., Mostellar, R.D., Yanofsky, C. Dynamics of 11. Schevitz, R.W., Otwinowski, Z., Joachimiak, A., Lawson, synthesis, translation, and degradation of trp operon mes- C.L., Sigler, P.B. The three-dimensional structure of trp senger RNA in E. coli. Cold Spring Harbor Symp. Quant. repressor. Nature 317:782-786, 1985. Biol. 34:725-740, 1969. 12. Kelly, R.L., Yanofsky, C. Mutational studies with the trp 26. Lane, A.N., Jardetzky, 0. NMR studies of the trp repressor repressor of Escherichia coli support the helix-turn-helix from Escherichia coli. Eur. J. Biochem 152:395-404, 1985. model of repressor recognition of operator DNA. Proc. Natl. 27. Lane, A.N., Jardetzky, 0. Unfolding of the trp repressor Acad. Sci. USA 82:483-487, 1985. from Escherichia coli monitored by fluorescence, circular 13. Herskowitz, I. Functional inactivation of genes by domi- dichroism and nuclear magnetic resonance. Eur. J. Biochem nant negative mutations. Nature 329:219-222, 1987. 164:389-396,1987. 14. Sadler, J.R., Novick A. The properties of repressor and the 28. Lawson, C.L., Sigler, P.B. The structure of trp pseudore- kinetics of its action. J. Mol. Biol. 12:305-326, 1965. pressor at 1.65A shows why indole propionate acts as a trp 15. Adler, K., Beyreuther, K., Fanning, E., Geisler, N., Gronen- “inducer.” Nature 333:869-871, 1988. horn, B., Klemm, A,, Miiller-Hill, B., Pfahl, M., Schmitz, A. 29. Lesk, A.M., Hardman, K.D. Computer-generated schematic How lac repressor hind to DNA. Nature 237:322-327,1972. diagrams of protein structures. Science 216:539-540, 1982.