Mutant Tryptophan Aporepressors with Altered Specificities Of

Mutant Tryptophan Aporepressors With Altered Specificitiesof Corepressor Recognition

Dennis N. Arvidson,' Michael Shapiro' and Philip Youderian'

Department of Biological Sciences, University of Southern Calqornia, Los Angeles, Calqornia 90089 Manuscript received November 20, 1990 Accepted for publication January 19, 199 1

ABSTRACT The Escherichia coli trpR gene encodes tryptophan aporepressor, which binds the corepressor ligand, L-tryptophan, to form an active repressor complex. The side chain of residue valine 58 of Trp aporepressor sits at the bottom of the corepressor @-tryptophan) binding pocket. Mutant trpR genes encoding changes of Vals8 to the other 19 naturally occurring amino acids were made. Each of the mutant proteins requires a higher intracellular concentration of tryptophan for activation of DNA binding than wild-type aporepressor. Whereas wild-type aporepressor is activated better by 5- methyltryptophan (5-MT) than by tryptophan, IlesHand other mutant aporepressors prefer tryptophan to 5-MT as corepressor, and Ala5' and GIY'~prefer 5-MT much more strongly than wild-type aporepressor in vivo. These mutant aporepressors are the first examples of DNA-binding proteins with altered specificities ofcofactor recognition.

ANY regulatory proteins areallosteric, and bind intermonomer interactions stabilize its extraordinary M small metabolic intermediates that modify their hydrophobic core. The aporepressor dimer denatures activities, allowing the intracellular pattern of gene at a very high temperature (>go") (BAEet al. 1988). expression to respond to physiologicalchange. For Comparisons of the crystal structures of repressor example, the Escherichia coli tryptophan aporepressor (SCHEVITZet al. 1985; LAWSONet al. 1988;OTWI- (TrpR) protein binds DNA poorly. However, when NOWSKI et al. 1988), and aporepressor (ZHANG et al. complexed with L-tryptophan (corepressor), it binds 1987) suggest that activation of the TrpR dimer in- three sites (operators) on theE. coli genome to repress volves repositioning a-helices D and E with respect to transcriptionof the trp (BENNETTand YANOFSKY the rigid hydrophobic core determined by &-helices 1978), aroH (ZURAWSKIet al. 198 l), andtrpR operons A, B, C, and F. Crystal structures of theTrp repressor (GUNSALUSand YANOFSKY1980). Thus, Trp apore- complex show that each of the two symmetric corepressor mediates the homeostatic control of trypto- pressor binding pockets ina repressor dimeris formed phan biosynthesis. by side chains from both monomers (SCHEVITZet al. Although mutations that alleviate the requirement 1985; LAWSONet al. 1988; OTWINOWSKIet al. 1988). of DNA binding proteins for their allosteric cofactors In these structures, tryptophan participates in hydro- havebeen described (ADHYA and GARGES 1990; philic and hydrophobic interactions with theseside MENONand LEE 1990), no mutant repressors oracti- chains (Figure 1). Both the carboxylate and a-ammo- vators with altered specificities of cofactor recognition nium groups of tryptophan areinvolved in hydrogen- have been found to date. In this paper, we describe bonding networks. The carboxylate group forms hy- the first such mutants, Trp aporepressors altered in drogen bonds (H-bonds) with the guanidino group of theirinteractions with L-tryptophanand 5-methyl- Arg*4 from one monomer, and the hydroxyl group of tryptophan (5-MT), a toxic corepressor analog. Thr44' fromits partner monomer('). The ammonium The product of theE. coli trpR gene is a monomer group forms H-bondswith the Ser" hydroxyl and the of 107 amino acids that assembles as a dimer (GUN- main-chain carbonyls of Leu4" and Leu43'. The in- SALUS and YANOFSKY1980; JOACHIMIAK et al. 1983; dole ringmakes multiple hydrophobic interactions ARVIDSON, BRUCEand GUNSALUS 1986).The crystal with side chains that mold the binding pocket. The structure of Trp aporepressor (ZHANG et al. 1987) plane of the indole ring is sandwiched between the shows that each monomer is a bundle of six a-helices, A-F. Four of the six a-helices, A, B, C, and F are parallel, extended aliphatic side chains of ArgR4 and interwoven in thedimer, and multiple, symmetric Arg54, and the C5-C7 edge of the indole ring packs against the branched hydrophobic side chains ofIIe5'

I Present address: California Institute of Biological Research, 11099 N. and Val5*. The corepressor, L-tryptophan, must act Torrey Pines Road, La Jolla, California 92037. * Present address: Department of Biochemistry. University of California, like a keystone in the hydrophobic core of aporepres- Berkeley, California 94720. sor. Its hydrophobic indole ring is buried in an "argi-

Genetics 128: 29-35 (May, 1991) 30 N. D. Arvidson, M. Shapiro and P. Youderian

FIGURE1 .-The structure of the L-tryptophan (corepressor) binding pocket is shown as a stereo pair of diagrams. The y-methyl group of ValsRpoints toward the 5tarbon of bound tryptophan. Reprinted from SCHEVITLet al. (I 98.5) with permission from P. B. SIGLER. nine sandwich" and framed by hydrophobic interac- (5-MT) more tightly than tryptophan as corepressor tions, and its hydrophilic functional groups are on the (MARMORSTEINet al. 1987; D. N. ARVIDSONand R. surface of the repressor complex. P. GUNSALUS,unpublished results). However, by no means do the crystal structures of To investigate the nature of corepressor binding, tryptophan repressor give us a complete understand- we have examined the effects of all possible amino ing of how tryptophan binds as corepressor and elicits acid changes of residue valine 58. Thesemutants allow the allosteric change between aporepressor and re- us toexplore the role of this residue in amanner pressor. In part, this is because structural differences independent of the conclusions we may draw from a between aporepressor and repressor are obscured by structure in vitro. Several amino acid changes should lattice interactions in the crystals. For example, the shorten or lengthen the Va15'side chain without com- sum ofRMS (root mean square) differences in the promising its hydrophobiccharacter. Therefore, if coordinates of two different crystal forms of repressor the carbon4 of tryptophan is close to the y-methyls is greater than the sum of differences between the of Val"" in the repressor complex, then a subset of same form of aporepressor and repressor (LAWSONet Val5" changes (including Ile"Rand Ala") should alter al. 1988). the preference of aporepressor for 5-MT and trypto- In addition, the pattern of specific weak bonds that phan. Indeed, we show that manyof these mutant stabilizes tryptophan binding in the crystal structures aporepressors have altered specificities of corepressor recognition. Even more surprisingly, we find that a of repressor must be different than the patternin vivo. wide variety of amino acid changes are tolerated at The crystal structures do not explain two things. First, this residue position, suggesting that the corepressor Thr44appears to make an important bond with the binding pocket is much less rigid than the structures tryptophan carboxylate in vitro, the change Thr4' + represented in the crystals. Ala has only a subtle effect on tryptophan binding in vitro (HE andMATTHEWS 1990). Second, the distance MATERIALS AND METHODS between carbon-5 of the tryptophan indole and they- methyl carbons of Val5' in the repressor crystal struc- Bacteria, phage and plasmids: Bacterial strains are derivatives of S. typhimurium LT2 or E. coli K12. Salmonella tures (ca. 3.8A) (SCHEVITZet al. 1985; LAWSONet al. strain MSl363 (leuA-am414 SUPE)(SUSSKIND 1980) was used 1988) is tooshort to accommodate an additional for the permissive growth of challenge phages; MS1868/ methyl group. This is disturbing because the related F'lacP' (BASSet al. 1987) carrying plasmid pPY2000 was Enterobacter aerogenes and Enterobacter cloacae trypto- used for challenge phage assays. The "Trp" challenge phage used to assay DNA-binding in vivo, P22 KnY 0-ref1arc- phan aporepressors are predicted to have isoleucine, amH1605 (BASSet al. 1987) carries a minimal reference-type not valine, at this position (our unpublished results). trp operator, 5' GAACTAGTTAACTAGTTC3', control- Furthermore, aporepressorbinds 5-methyltryptophan ling ant expression. Plasmid pPY2000 is a derivative of Mutant Aporepressors 31 pPY 150 (BENSONet al. 1986) missing the sequences between Ana1 the Sal1 and sites of its pBR322 backbone. Mutant plasmld pPY 2000 derivatives of pPY2000 were made by site-directed muta- frpR genesisusing the degenerate mismatch primer,5' CGTGCCTATTNNSGAAGAGCTG 3' and the "clean-up" 1 primer, 5' CGTGCTTATTSRAGAACAGCTG 3' (TAY- Trp aporepressor LOR,OTT and ECKSTEIN1985). Mutagenized plasmids were + t transformed into E. coli strain X90/F'lacfe1 (AMANN,BRO- 4- L-tryptophan SIUS and PTASHNE 1983) andsequenced by the method of SANGER,NICKLEN and COULSON(1977), using the oligonucleotide 5'GTCGACCTGCTTAAGAATGCC 3' includ- 1 I ing codons 23 to 29 of trpR, as primer. Strain CG103 (Alac- 00 repressor Trp pro AtrpR505 ara thi recA56 srl::TnlO)/XCLGI45) (D. N. I ARVIDSONand C. G. ARVIDSON,unpublished results) carries a lambda prophage with a fusion of the trp operator/promoter region and start of trpE to lacZ. Quantitation of mutant proteins: E. coli CG103 with each mutant derivative of plasmid pPY2000 was grown in Trp "drop-out" medium (BASSet al. 1987) with 50 pg/ml proline, 100 pg/ml thiamine and 400 pg/ml carbenicillin, to a density of 2 X 1O8/rnl. Proteins were resolved by SDS polyacrylanlide gel electrophoresis (SCHAGGERand VON JA- P- galactosldaseassay challengephage assay COW 1987) and were transferred to nitrocellulose electro- reporter gene IacZ reporter gene anf phoretically (TSANG,PERALTA andSIMONS 1983). Nitrocel- lulose-bound TrpR was detected by incubation with rabbit represslonresults In lower repressionresults In higher anti-aporepressor antibody (the gift of R. P. GUNSALUS) P-galactosidaseactlvlty frequency of lysogeny (GUNSALUS,MIGUEL andGUNSALUS 1986), biotin-conju- FIGURE2.-Assays for the activities of mutant aporepressors gated goat anti-rabbit antibody, and streptavidin-conjugated in vivo. (1) Wild-type Trp aporepressor producedby plasmid pPY2000 alkaline phosphatase (HARLOWand LANE 1988).In Salmo- binds corepressor to form the active repressor complex. Repressor nella, Gly'"is unstable with or without L-tryptophan, but complex binds the trp operator, repressing transcription of a trpE/ may be produced stably in the presence of 5-MT (data not lacZ gene fusion, and low levels of P-galactosidase activity are shown). Thus,the addition of a methyl group to the 5 position of the indole ring of corepressor, or to the a-carbon produced. Mutant aporepressors blocked at some step in repressor assembly or that form an inactive complex with corepressor, cannot of Gly" (+ Ala'"), confers stability on this mutant aporepressor/corepressor complex. repress transcriptionof the trpE/lacZ fusion; in their presence, high Rapid @-galactosidaseassays: Rapid /$galactosidase as- levels of LacZ are made. (2) A challenge phage infection follows says were performed as described by GARDELLA,MOYLE and one of two developmental pathways, depending on the expression SUSSKIND(1989) with three modifications. Each CG103 of P22 antirepressor. The reporter gene in this assay is the phage strain with a pPY2000 derivative was grown in media with P22 ant (antirepressor) gene; binding of a protein to an operator or without 2 pg/ml tryptophan before assay;cells were represses ant transcription, permitting lysogenic development and host survival (what is measured). Expression of antirepressor during grown in wells of microtiter plates in 250 pl volumes without active aeration at 37"; and, a high multiplicity of T-even P22 infection triggers lytic development and death of the infected phage was used to lyse cellsfrom without (DOERMANN1952). host. Activities are given in BARRICKunits. Data varied less than 20% from experiment to experiment. ure 2). On this plasmid, the lacW5 promoter directs Challenge phage assays: Overnight cultures of MS 18681 transcription of the trpR+ gene, under the control of F'laclQ' carrying plasmid pPY2000 and each of its mutant Lacrepressor. Codon 58 of trpR was mutagenized derivatives were diluted 100-fold into LB medium with 50 with a mixed oligonucleotide, used to prime the re- pg/ml ampicillin, and grown to a density of 4 X 10"/ml. synthesis of the trpR sense strand on single-stranded Aporepressor expression was induced bthe addition of isopropyl-P-thiogalactoside(IPTG) to IO- x.M, cultures were templatepPY2000 DNA (TAYLOR,OTT and ECK- incubated at 37" for 30 min, and chilled to 4".Each culture STEIN 1985; BASS, SORRELLSand YOUDERIAN1988). was infected with challenge phage P22 Kn9 0-ref1 arc- Mutagenized DNA was transformed into strain X901 amH1605at amultiplicityof 25 phage/cell. After adswption F'lacp' which makes high levels of Lac repressor to of the phage for 15 min at 25 O, infected cells were diluted in the wells of microtiter plates, and 4 pl of eachserial prevent the expression of potentially lethal levels of dilution was spotted on green tryptophan drop-out plates mutant aporepressors (BASS, SORRELLSand YOUDER- with ampicillin, kanamycin, IPTG (10 p~),and tryptophan IAN 1988). Transformants were pooled, and plasmid (or tryptophan plus 5-MT). The efficiency of survival is the DNA was extractedfrom the pool. Becausesite-di- number of surviving cells divided by the number of input cells (assayed ongreen plates with ampicillin); averages of at rectedmutagenesis results ina highfrequency of least three independent measurements are shown. untargeted second-site mutations, a small restriction fragment with thetarget codon was generated by RESULTS cleavage of the plasmidDNA at uniqueSal1 and MluI sites, purified, and subcloned intowild-type pPY2000 Mutant aporepressors with Val5' changes:Plasmid backbone. Sequencesof this fragment in the subclones pPY2000 expresses wild-type Trp aporepressor (Fig- show that 47 mutant plasmids represent 20 different 32 D. N. Arvidson, M. andShapiro P. Youderian

TABLE 1 prophage with a trpEllacZ gene fusion (D. N. ARVID- Activities of mutant aporepressors with Val5*changes in vivo SON and C. G. ARVIDSON,unpublished results). In

~ CG103, synthesis of the LacZ fusion protein is regu- @-Galactosidase lated by the plasmid-encoded Trp aporepressor (Fig- activity CRM, % ure 2). As shown in Table 1, wild-type aporepressor Residue CodonResidue wild type fTRP-TRP produced bypPY2OOO represses LacZ expression (del) 0 1952 1874 about 50-fold without added tryptophan; with tryp- Pro (1) 2 1820 1305 tophan, lower levels of LacZ activity are made. An ASP GAC (2) 5 1841 1003 otherwise isogenic plasmidwith adeletion of trpR CGA* 7 1697 161 1 does not repress the gene fusion. Five mutant apore- Glu GAA* 14 1660 137 pressors (Cys > Ile > Ala = Ser > Thr) retain near LYS AAG (1) 21 712 187 wild-type activity, and seven mutant proteins (Met > GIY GGA* 33 1074 110 Phe > His > Asn = Tyr > Gln > Leu) have interme- GGC (3) diate activities. Even in the presence of tryptophan, UGG (2) 38 1224 229 TrP the seven most labile mutantaporepressors (Gly > Leu 46 (1) 586 68 Glu Trp Lys Asp Pro Arg) effect little Ser ACG(I) 48 146 19 > > > > > ucc (3) repression of the gene fusion. Phe E (2) 49 246 41 Activation of mutant aporepressors by corepres- Asn AAC (2) 53 391 49 sors: The interaction of Trp repressor with the wild- His CAC (2) 59 276 41 type E. coli trp operator involves the binding of mul- CYS UGC (1) 61 50 12 Ile AUC(2) 61 59 19 tiple repressor dimers to tandemsites within operators and GUNSALUS 1987;BASS et TYr UAC (2) 61 374 52 (KUMAMOTO,MILLER al. Thr ACC (3) 62 190 24 1987; ELLEDGEand DAVIS1989; STAACKEet al. 1990). Gin CAA* 79 407 56 Mutations in trpR may have pleiotropic phenotypes, CAG (1) affecting not only the ability of individual dimers to Met (2) 82 239 33 Ala GCC (3) 87 155 18 bind single sites, but also the ability of multiple dimers GCG (1) to bind tandem arraysof sites. To exclude pleiotropic Val -GUC 100 42 12 effects of trpR mutations on the tandem binding of Val58 changes reduceboth the stability and activityof Trp dimers, we studied the binding of each mutant protein aporepressor. The number of isolates of each codon is given in to a single operator site, using the challenge phage parentheses; codons with asterisks were recovered from a “clean- assay (BENSONet al. 1986). up” mutagenesis. Plasmids with underlined codons were assayed. As a negative control (del), assays were conducted on cells with a A P22 challenge phagewill lysogenize host cells that plasmid missing the 434-bp BarnHI fragment containing trpR, yet produce a repressor able to bind a phage-borne op- otherwise isogenic with pPY2000. The percentage of cross-reacting material (CRM) relative to wild-type aporepressor produced from erator. After infection of a sensitive supo host, a chal- each mutant plasmid was determined from the analysis of immu- lenge phage with a trp operator will establish lysogeny noblots. @-Galactosidaseactivity is given in BARRICKunits as de- if and only if its trp operator is bound by Trp repressor scribed by GARDELLA,MOYLE and SUSSKIND(1989). (Figure 2). Once integrated, the P22 prophage confers codons for 17119 possible amino acid changes at po- a kanamycin-resistant phenotype upon its surviving sition 58 (Table l). Plasmids producing aporepressors host. Alternatively, if the challenge phage operator is with the two remaining amino acid substitutions were free of repressor, P22 will make antirepressor, de- made using a second, less degenerate primer. velop lytically, and kill its host. Because the fractional Aporepressor made by each of the mutant plasmids occupancy of an operator increases with increasing was quantitated using rabbit anti-aporepressor antirepressor concentration, the efficiency of lysogeny of body (GUNSALUS,MICUEL and GUNSALUS1986). All a challenge phage with a trp operator increases in of the mutant proteins are made in lower amounts response to increasing levels of both aporepressorand than wild type and may be grouped into threesubsets the corepressor, L-tryptophan (BASSet al. 1987). (Table 1). Three of the 19 mutantproteins, Pro, Asp, The Salmonella host for challenge phage infection and Arg, are made in nearly undetectable amounts assays carries the Trp aporepressor-producing plas- (

07 r opment with maximum efficiencies less than wild type, 1 and are ranked by activity in the order: His > Phe > Asn > Gln. Both His58 and Asn" proteinsrequire -2-I i moretryptophan for activity than wild-type aporepressor; additionof tryptophan does notincrease their -3 t maximum ability tosupport lysogeny. Presumably, when these proteins are complexed with tryptophan, they have lower intrinsic affinities for the trp operator than the wild-type complex in vivo. The eight mutant aporepressorsthat are least active have changes of Val5' to Pro or Gly, to each of four charged amino acids (Asp, Glu, Lys andArg), or to residues with larger side chains (Lys, Tyr, Trp andArg). The tryptophan analog, 5-MT, can substitute for L- -5L tryptophan as corepressor in vivo (COHENand JACOB /"-"" 1959), and is bound by wild-type aporepressor with a -:[ ,*Y four-fold higher affinity in vitro (MARMORSTEINet al. -2 1987; D. N. ARVIDSONand R. P. GUNSALUS, unpub- I lished results). This analog is toxic to both S. typhimurium and E. coli, either because incorporation of 5- MT into essential proteins renders them inactive, or because 5-MT both causes feedback inhibition of an- [trpl Nu9 1 ml [S-MT] Ng/ml thranilate synthase and acts as corepressor, the com- bined effect of which is lethal tryptophan starvation FIGURE3.-Interactions of mutant aporepressors with corepres- (SOMERVILLE 1983). sors in vivo. The efficiency of survival (frequency of lysogeny) of cells infected with a high multiplicity of trp challenge phage is a The toxic effect of 5-MT is antagonized by small measure of how well the phage-bornereference trp operator is amounts of added tryptophan; these amounts do not occupied by plasmid-encoded TrpR protein. Trp repressor activity permit the efficient lysogenization of cellsby challenge in vivo is measured as a function of increasing amounts of added phage. Thus, we measured the frequency of lysogeny corepressor. Note that we are comparing the responses of apore- of a challenge phage as a functionof 5-MT concentra- pressors to the pure stereoisomer, L-tryptophan, with those of aporepressors to a racemic mixture of 5-MT; we might anticipate tion in thepresence of 1 pg/ml addedtryptophan that the D-isomer of 5-MT is inactive as corepressor, because this is (Figure3). As expected, wild-type aporepressor re- the case for D-tryptophan (MARMORSTEINet al. 1989; D. N. ARVID- quires less added 5-MT (0.4 pg/ml 5-MT + 1 pg/ml SON and R. P. GUNSALUS,unpublished results). tryptophan) than tryptophan (3 + 1 pg/ml) to permit efficient lysogeny. Seven mutant aporepressors permit in Figure 3, the fraction of lysogenic survivors among the Trp challenge phage to lysogenize with wild-type infected cells producing wild-type aporepressor is low efficiencies in the presence of 1 pg/ml tryptophan and in the absence of exogenous tryptophan andincreases 10 pg/ml 5-MT. These are ranked in order of their dramatically (more than 1000-fold) upon the addition increasing dependence on 5-MT: Ala > Cys > Ser > of tryptophan. Leu > Thr > Ile = Met. Four mutant aporepressors, The frequency of lysogeny of each aporepressor- in the presence of high levels of 5-MT, permit lyso- producing host depends on added tryptophan concen- geny with efficiencies less than observed withwild- tration;differences in this dependence allow us to type aporepressor, and are ranked in the order: Gly rank the requirement of each mutant aporepressor > Asn > Gln > Lys. The remainingeight mutant for tryptophan in vivo. Each of the mutant Trp apo- aporepressors (Pro, Asp, Glu, His, Phe, Tyr, Trpand repressors requires more tryptophan than Vals8 apo- Arg) do not permit efficient lysogeny at any 5-MT repressor for activity (Figure 3). Seven mutant apo- concentration. repressors permit the efficient lysogenic development of challenge phage with 100 pg/ml tryptophan. These DISCUSSION are ranked in decreasing order of activity: Ile > Cys > Thr > Ser > Ala > Leu > Met. This order agrees All of the changes at residue Val5' of E. coli Trp well with that of the ability of mutant aporepressors aporepressor reduce either the stability or activity of to control the trpE/lacZ fusion regulated by the wild- this protein. The Val5' side chain is located in the type E. coli trp operator. The five mutant aporepres- carboxyl-terminal half of a-helix C. It is no surprise sors that repress the fusion best are the most active in that the change, Va15'+Pro, predicted to disrupt this the challenge phage assay. secondary structure, results in an unstable protein. Four mutant aporepressors permit lysogenic devel- The electrostaticcharacter of the residue 58 side 34 D. N. Arvidson, M. andShapiro P. Youderian chain is also a critical determinant of corepressor The most surprising result toemerge from this recognition. Changes of the hydrophobic Val5' side study is that many different amino acid changes at chain to charged side chains (e.g., Arg, Lys, Asp, Glu) Val5' do not result in a majorloss of repressor activity. result in unstable mutant aporepressors. In a related study, WELLSet al. (1987) examined all Ile5' is the most active mutant aporepressor. Ile5' possible changes of residueMetzzz, located in the requires much more 5-MT foractivity than wild type. interior of the substrate binding pocket of subtilisin. In this case, the addition of a single methyl group to All of the changes of Metzz2 thatincrease the specific the side chain of residue 58 (Val + Ile) changes the volume of the residue 222 side chain (e.g., Phe222) preference of aporepressor from a corepressor with a result in dramatic decreases in enzyme activity. Unlike 5-methyl groupto one without. Presumably, steric the substrate-binding pocket of subtilisin, the core- exclusion of the additional &methyl of He5' by the 5- pressor-binding pocket of tryptophanaporepressor methyl of 5-MT distortsthe structureof the repressor must be less rigid to accommodate both awide variety complex. Incontrast, Ala5' aporepressorrequires of indole analogs (MARMORSTEINet al. 1987), and a more tryptophan than Val5' (wild type) for activation, wide variety of amino acid substitutions at Val5'. Al- yet is activated by 5-MT as well as the wild type. though the X-ray crystal structures of Trp repressor Whereas lengthening the Val5' side chain by the ad- protein may provide us with an initial picture of how dition of a &carbon impairs the interaction of apore- activation by corepressor might work, it is likely that pressor with 5-MT, shortening theVaI5'side chain by more detailed spectroscopic analyses of the motion of the subtraction of y-carbons impairs the interaction this flexible substructure,the tryptophan binding of aporepressor with tryptophan. Gly5' aporepressor pocket, will be necessary to fill in the details of how can be activated in the challenge phage assay only by tryptophan activates Trp aporepressor. 5-MT. The volume of the residue 58 side chain is one of the most critical determinants of corepressor rec- We thank CINDYGROVE ARVIDSON, FRED EISERLING,ANDY KUMAMOTO,TODD LANE, DENNIS SAKAI, MIMI SUSSKIND,GARY ognition specificity. STORMOand GARYTRUMP for advice and helpful discussions. We The geometry of the residue 58 side chain is also thank PAUL SIGLERfor permission to reprint Figure 1 from SCHEV- an important determinantof corepressor recognition. ITZ et al. (1 985). This work was supported by National Institutes of Aporepressors with more extended residue 58 side Health grant GM34150 to P. Y. and GM12629 to D.N.A. chains fare better with 5-MT as corepressor than with tryptophan in the challenge phage assay. Thus, LITERATURE CITED whereas Ile5' prefers tryptophan,Leu5' prefers 5-MT; ADHYA,S., and S. GARGES, 1990 Positive control. J. Biol. Chem. in addition, Lys5' is activated only by 5-MT. Of the 265: 10797-10800. four mutant repressors with large aromatic or side AMANN,E., J. BROSIUSand M. PTASHNE,1983 Vector bearing a chains, only repressors with the smallest two side hybrid trp-lac promoter useful for regulated expression of chains (Phe andHis) are activated by tryptophan, and cloned genes in Escherichia coli. Gene 25: 167-178. none are activated by 5-MT. ARVIDSON,D. N., C. BRUCE~~~R. P. GUNSALUS, 1986 Interaction of the Escherichia coli trp aporepressor with its ligand, L-tryp- Substitution of polar residues for Val5' weakens the tophan. J. Biol. Chem. 261: 238-243. interaction between aporepressor and corepressor. BAE,S. -J., W. -Y. CHOU,K. MATTHEWS and J. M. STURTEVANT, Thus, theSer5' and Thr5'aporepressors require both 1988 Tryptophan repressor of Escherichia coli shows unusual more tryptophan and more 5-MT for activity than thermal stability. Proc. Natl. Acad. Sci. USA 85: 6731-6732. BASS,S., V. SORREIS and P. YOUDERIAN,1988 Mutant Trp re- wild type. However, the volume of the side chain is pressors with new DNA-binding specificities. Science242: 240- more important than polarity, because these mutant 245. proteins bind tryptophan in the order of affinity: Val BASS,S., P. SUGIONO,D. N. ARVIDSON,R. P. GUNSALUSand P. > Thr > Ser > Ala. YOUDERIAN, 1987 DNAspecificity determinants of Esche- The hydrophobicinteraction of residue Val5*of richia coli tryptophan repressor binding. Genes Dev. 1: 565- 572. Trp aporepressor with the indole ring of bound co- BENNETT,G. N., and C. YANOFSKY, 1978 Sequence analysis of repressor is highly specific. All of the changes at Val5' operator constitutive mutants of the tryptophan operonof reduce the activity of Trp aporepressor. Nonetheless, Escherichia coli. J. Mol. Biol. 121: 179-192. a subset of Val5' changes has new specificities of BENSON,N., P. SUGIONO,S. BASS,L. V. MENDELMANand P. YoUD ERIAN,1986 General selection for specific DNA-binding ac- corepressor activation in vivo. These mutant apore- tivities. Genetics 114: 1-14. pressors prefer tryptophan to 5-MT, or prefer 5-MT COHEN,G. N., and F. JACOB, 1959 Sur la repression de la synthese much more strongly than wild-type aporepressor, sug- intervenant dans la formation du tryptophan chez E. coli. C. R. gesting that the side chain of Val5' interacts directly Acad. Sci. Ser. D. 248: 3490-3492. with carbon-5 of the indole ring vivo, as predicted DOERMANN,A. J., 1952 The intracellular growth of bacterio- in phages I. Liberation of intracellular bacteriophage T4 by pre- from the interpretation of the crystal structures of mature lysiswith another phage or with cyanide. J. Gen. Trp repressor (SCHEVITZet al. 1985; LAWSONet al. Physiol. 35: 645-656. 1988; OTWINOWSKIet al. 1988). ELLEDGE,S. J., and R. W. DAVIS,1989 Position and density effects Mutant Aporepressors 35

on repression by stationary and mobile DNA-binding proteins. SCHAGGER,H., and G. VONJAGOW, 1987 Tricine-sodium dodecyl Genes Dev. 3: 185-197. sulfate-polyacrylamidegel electrophoresis for the separation of GARDELLA,T., H. MOYLE and M.M. SUSSKIND,1989 A mutant proteins in the range from 1 to 100 kDa. Anal. Biochem. 166 Escherichia coli u70 subunit of RNA polymerase with altered 368-379. promoter specificity.J. Mol. Biol. 206: 579-590. SCHEVITZ,R. W., Z. OTWINOWSKI,A. JOACHIMIAK, C. L. LAWSON GUNSALUS,R. P., and C. YANOFSKY, 1980 Nucleotide sequence and P. B. SIGLER,1985 The three-dimensional structure of and expression of Escherichia coli trpR, the structural gene for trp repressor. Nature 317: 782-786. the trp aporepressor. Proc. Natl. Acad. Sci. USA 77: 7 1 17- SOMERVILLE,R. L., 1983Tryptophan: biosynthesis, regulation, 7121. and large-scale production, pp. 35 1-378 in Amino Acids:Biosyn- GUNSALUS,R. P., A. G. MIGUEL and G. L. GUNSALUS, thesis and Genetic Regulation, edited by K. M. HERRMANN and 1986 Intracellular Trp repressor levels in Escherichia coli. J. R. L. SOMERVILLE.Addison-Wesley, Reading, Mass. Bacteriol. 167: 272-278. STAACKE,D., B. WALTER,B. KISTERS-WOIKE,B. V. WILKEN-BERG- HARLOW,E., and D. LANE,1988 pp. 471-510 in Antibodies:A MANN and B. MULLER-HILL,1990 How Trp repressor binds LaboratoryManual. Cold Spring Harbor Laboratory Press, to its operator. EMBO J. 9: 1963-1967. Cold Spring Harbor, N.Y. SUSSKIND,M. M., 1980 A new gene of bacteriophage P22 which HE,J., and K. S. MATTHEWS,1990 Effect of amino acid alterations regulates synthesis of antirepressor. J. Mol.Biol. 138: 685- in the tryptophan binding site of the trp repressor. J. Biol. 713. Chem. 265: 731-737. TAYLOR,J. W., J. OTT and F. ECKSTEIN,1985 The rapid gener- JOACHIMIAK, A,, R. L. KELLEY,R. P. GUNSALUS,C. YANOFSKYand ation of oligonucleotide-directed mutations at high frequency P. SIGLER,1983 Purification and characterization of trp apo- using phosphorothiolate-modified DNA.Nucleic Acids Res. repressor. Proc. Natl. Acad. Sci. USA 80 668-672. 13: 8765-8785. KUMAMOTO,A. K., W. G. MILLER and R. P. GUNSALUS, TSANG,V. C., J. M. PERALTAand A. R. SIMONS,1983 Enzyme- 1987 Escherichia coli tryptophan repressor binds multiple sites linked immunoelectrotransfer blot techniques (EITB) for within the aroH and trp operators. Genes Dev. 1: 556-564. studying the specificities of antigens and antibodies separated LAWSON,C. L., R. -G. ZHANG, R. W. SCHEVITZ,Z. OTWINOWSKI, by gel electrophoresis. Methods Enzymol. 92: 377-391. A. J. JOACHIMIAK and P. B. SIGLER, 1988 Flexibility of the WELLS, J. A,, D. B. POWERS,R. R.BOTT, B. A. KATZ,M. H. DNA-binding domains of trp repressor. Proteins 3: 18-3 1. ULTSCH, A. A. KOSSIAKOFF,S. D. POWERS,R. M. ADAMS,H. MARMORSTEIN,R. Q., A. JOACHIMIAK, M. SPRINZand P. SIGLER, H. HEYNEKER,B. C. CUNNINGHAM,V. J.MILLER, T. P. GRAY- 1987 The structural basis forthe interaction between L- CAR and D. A. ESTELL, 1987 Protein engineering of subtilisin, tryptophan and the Escherichia coli trp aporepressor. J. Biol. pp. 279-287, in ProteinEngineering, edited byD. OXENDER Chem. 262: 4922-4927. and F. Fox, Alan R. Liss, New York, N.Y. MENON,K. P., and N. L. LEE, 1990 Activation of ara operons by ZHANG, R. -G., A. JOACHIMIAK, C. L. LAWSON,R. W. SCHEVITZ,Z. a truncated AraC protein does not require inducer. Proc. Natl. OTWINOWSKIand P.B. SIGFER,1987 The crystal structure Acad. Sci. USA 87: 3708-3712. of trp aporepressor at 1.8 A shows how binding tryptophan OTWINOWSKI,Z., R. W. SCHEVITZ,R. -G. ZHANG,C. L. LAWSON, enhances DNA affinity. Nature 327: 591-597. A. JOACHIMIAK, R. Q. MARMORSTEIN,B.F. LUISIand P. B. ZURAWSKI,G., R. P. GUNSALUS,K. D. BROWN andC. YANOFSKY, SIGLER,1988 The crystal structure of the trp repressor/ 1981 Structure and regulation of aroH, the structural gene operator complex at atomic resolution. Nature 335: 321-329. for the tryptophan-repressible 3-deoxy-~-arabino-heptulosonic SANGER,F., S. NICKLENand A. COUISON, 1977 DNA sequencing acid-7-phosphate synthetase of Escherichia coli. J. Mol.Biol. with chain terminating inhibitors. Proc. Natl. Acad. Sci. USA 145: 47-73. 74: 5463-5476. Communicating editor: G. MOSIG