Identification of Amino Acid Residues Involved in Feedback

The anthranilate synthase-phosphoribosyl transferase complex, a hetero~trameric enzyme made up of the TrpE and TrpD polypeptides, catalyzes three reactions comprising the first two steps of tryptophan biosynthesis in SaEmonella typhimurium. All three activities of the complex are subject to feedback inhibition by tryptophan, which results from allosteric effects associated with the binding of one molecule of inhibitor to each of the TrpE subunits of the complex. Random in vitro chemical mutagenesis of the trpE gene was used to generate a collection of mutant forms of the complex which displayed varying degrees of resistance to feedback inhibition. Single amino acid substitutions, identified by DNA sequencing, were found at 14 different residues within the TrpE polypeptide. The residues were distributed throughout TrpE, but those that appeared to be most critical for regulation were found in two clusters, one at the extreme aminoterminal end, including residues Glu-39, Ser-40, and Ala-41, and the other in the middle of the polypeptide, including residues Asn-288, Pro-289, Met-293, Phe294, and Gly-305. Kinetic and binding studies of the purified mutant complexes demonstrated that 9 of the 14 had a marked decrease in affinity for tryptophan with little or no change in substrate affinity or catalytic capacity. The remaining five enzymes exhibited more subtle changes, having small decreases in inhibitor affinity coupled with small increases in substrate affinity. Mutant enzymes that were not totally feedback-resistant had a decreased kinetic response to tryptophan binding. All enzymes exhibited alterations in  tryptophan-induced conformational changes as monitored by dye-ligand chromatography.


Identification of Amino Acid Residues Involved in Feedback Regulation of the Anthranilate Synthase Complex from ~a l~o n~l l u t~p h i~u r i u~
The anthranilate synthase-phosphoribosyl transferase complex, a hetero~trameric enzyme made up of the TrpE and TrpD polypeptides, catalyzes three reactions comprising the first two steps of tryptophan biosynthesis in SaEmonella typhimurium. All three activities of the complex are subject to feedback inhibition by tryptophan, which results from allosteric effects associated with the binding of one molecule of inhibitor to each of the TrpE subunits of the complex. Random in vitro chemical mutagenesis of the trpE gene was used to generate a collection of mutant forms of the complex which displayed varying degrees of resistance to feedback inhibition. Single amino acid substitutions, identified by DNA sequencing, were found at 14 different residues within the TrpE polypeptide. The residues were distributed throughout TrpE, but those that appeared to be most critical for regulation were found in two clusters, one at the extreme aminoterminal end, including residues Glu-39, Ser-40, and Ala-41, and the other in the middle of the polypeptide, including residues Asn-288,  Kinetic and binding studies of the purified mutant complexes demonstrated that 9 of the 14 had a marked decrease in affinity for tryptophan with little or no change in substrate affinity or catalytic capacity. The remaining five enzymes exhibited more subtle changes, having small decreases in inhibitor affinity coupled with small increases in substrate affinity. Mutant enzymes that were not totally feedback-resistant had a decreased kinetic response to tryptophan binding. All enzymes exhibited alterations in tryptophan-induced conformational changes as monitored by dye-ligand chromatography.
Carbon flow through the tryptophan biosynthetic pathway in bacteria and fungi is controlled by negative feedback regulation of the first enzyme of the pathway, anthranilate synthase (EC 4.1.3.27). This enzyme catalyzes the formation *This work was supported by Research Grant GM35889 and Training Grant G~0 7 0 8 2 from the National Institutes of Health. 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. This paper is dedicated to the memory of Irving Crawford who made so many outstanding contributions to our knowledge of the genes and enzymes of tryptophan biosynthesis.
The inhibition of anthranilate synthase by the end product of the pathway, L-tryptophan, in combination with analogous feedback regulation of the initial enzymes of the phenylalanine and tyrosine pathways, ensures the appropriate partitioning of chorismate into the three amino acid pathways in response to the needs of the cell (1).
The anthranilate synthase from Salmonella typhirnurium is part of a multifunctional, tetrameric complex made up of two molecules each of the TrpE and TrpD polypeptides, the products of the first two genes of the trp operon ( 2 , 3). In addition to synthesizing anthranilate, the complex also catalyzes the second step of the pathway, the transfer of the phosphoribosyl group of 5-phosphorylribose-1-pyrophosphate (PRPP) to anthranilate (phosphoribosyl transferase activity, EC 2.4.2.18) (Reaction 3). The TrpD subunit is a bifunctional molecule with independent structural domains (4). Its aminoterminal domain functions as a glutamine amidotr~sferase, releasing the amide group of g l u t~i n e for the ina at ion of chorismate by the TrpE subunit, while the carboxyl-terminal domain carries out the phosphoribosyl transferase activity of the complex. Glutamine hydrolysis by the TrpD subunit is greatly enhanced by the binding of chorismate to the TrpE subunit (5). Thus, glutamine amidotransferase activity and glutamine-dependent anthranilate synthase activity are exclusive properties of the complex. In contrast, both the complex and the free TrpE subunit are capable of ammoniadependent anthranilate synthase activity (Reaction 2 ) (6). Similarly, both the complex and the free TrpD subunit possess phosphoribosyl transferase activity (7,8).
The glutamine-dependent and ammonia-dependent anthranilate synthase, glutamine amidotransferase, and phosphoribosyl transferase activities of the complex, as well as the ammonia-dependent anthranilate synthase activity of the TrpE subunit, are all subject to feedback inhibition, resulting from the binding of one molecule of tryptophan to the TrpE subunit (9). In each case, inhibition is competitive with respect to chorismate and noncompetitive with respect to the other substrates. It appears likely that tryptophan and chorismate bind to distinct sites on the TrpE subunit of the complex rather than competing for the same site, since the M$+

M92'
Anthranilate + PRPP + N-phosphoribosyl anthranilate + pyrophosphate enzyme can be desensitized with respect to tryptophan inhibition both in vivo by mutations in the TrpE subunit (5, 10) and in uitro by high concentrations of Mg2+ (11). Thus conformational changes within the enzyme associated with the binding of tryptophan to the TrpE subunit appear to be responsible for inhibition of the various activities of the complex.
In this report, we present the results of a mutational analysis of the feedback site of the TrpE subunit of the complex from S. typhimurium. This has led to the identification of amino acid residues critical for tryptophan binding and has provided evidence for the existence of an amino-terminal regulatory domain in the TrpE subunit.

RESULTS
Isolation of Feedback-resistant trpE Mutants-Mutations leading to the loss of feedback inhibition in the anthranilate synthase of S. typhimurium were sought by isolating strains resistant to 5-methyltryptophan (MTR),2 an effective false feedback inhibitor, after random in uitro mutagenesis of the cloned trpE gene. The vector used for the mutagenesis was plasmid pSTG3, which carries the S. typhimurium trpE gene under the control of its own promoter (Fig. 1). The host used for the selection of mutant plasmids was Escherichia coli strain CB90, which is deleted for trpE but has trpD and the remaining genes of the operon intact. The use of a single copy trpD host was necessitated by the fact that the growth on minimal medium of strains with a multicopy plasmid carrying both trpE and trpD is essentially unaffected by 5-methyltryptophan, even a t concentrations as high as 200 pg/ml. In contrast, the growth of CB9O/pSTG3 is completely suppressed on minimal medium by moderate concentrations of 5-methyltryptophan (50 pg/ml), allowing the selection of MTR derivatives. E. coli CB90 was chosen as host, rather than a s. typhimurium strain of similar genotype, because of the superior transformability of E. coli strains. It has been shown that the S. typhimurium TrpE and the E. coli TrpD subunits assemble in uiuo, forming a heterologous complex with close to normal catalytic and feedback properties (12).
By using the CB9O/pSTG3 host-vector system, 45 prototrophic Amp' MTR strains were isolated after in uitro treatment of trpE DNA with either nitrous acid or hydroxylamine. The sensitivity of the heterologous complexes of the MTR isolates to feedback inhibition by tryptophan was assessed in the standard anthranilate synthase assay using crude extracts of cells grown in minimal medium + ampicillin. Each of the mutant enzymes exhibited a loss of feedback control relative to the wild type enzyme, the magnitude of which varied widely (data not shown). Thus, these preliminary results indicated that the stringency of the selective screen used was appropriate for the recovery of mutants with diverse alterations in feedback regulation.
The nature and location of the mutational change in the trpE gene of each of the 45 MTR plasmids was then determined by DNA sequence analysis. This led to the identification of 14 different residues within the TrpE polypeptide at which amino acid substitutions resulted in the feedback-' Portions of this paper (including "Materials and Methods" and resistant MTR phenotype (Table I). The identified residues were distributed throughout the gene, although two clusters were readily apparent. One was located at the extreme amino terminus of the polypeptide, specified by mutations at residues Glu-39, Ser-40, and Ala-41; the other was found in the middle of the polypeptide, specified by mutations at residues Asn-288, Pro-289, Met-293, Phe-294, and Gly-305. Most of the mutations were recovered multiple times. After discounting multiple isolations of the same mutation within a single mutagenesis experiment, it was concluded that the collection consisted of a minimum of 22 independent clones.
Kinetic Analysis of Wild Type and Mutant Anthranilate Synthases-Plasmids were constructed carrying the 14 mutant trpEMTH genes together with the translationally coupled, wild type S. typhimurium trpD+ gene. These were transferred into E. coli strain CB25, which is deleted for the entire trp operon, thereby creating strains synthesizing homologous S. typhimurium anthranilate synthase complexes. The wild type and the 14 mutant enzymes were overexpressed in these constructs, purified to homogeneity, and characterized by steady state kinetic analysis.
The kinetic properties of the wild type enzyme (Table 11) were essentially as reported in previous studies (5, 11). The apparent K,,, for chorismate (KmCh') was 2.3 p~, the K; for tryptophan (KTri') was 1.3 p~, and the turnover number (kcat) was 12 s-'. Feedback inhibition by tryptophan was competitive with respect to chorismate, and positive cooperativity of chorismate binding was apparent at higher concentrations of tryptophan ( Fig. 2 A ) .
The mutant enzymes displayed a variety of changes in kinetic properties (Table  11). Nine of the 14 had striking increases in KTrp, with little or no change in KmChr. Four of these, E39K, S40F, M293T, and C465Y, were completely insensitive to tryptophan inhibition under the conditions tested, indicating >300-fold increase in K?Ip. The other five, A41V, N288D, P289L, F294L, and G305S, exhibited moderate increases in KTv, ranging from 7 to 20 times that of wild type. It is noteworthy that all of the enzymes with TrpE mutations located in the two clusters mentioned above have robust phenotypes, i.e. either total feedback resistance or moderate increases in KtTrp. The remaining mutant enzymes, R128H, c174Y, R402W, G460D, and H515Y, displayed only subtle changes in apparent ligand affinities, characterized in most cases by a marginal increase in KsTri' and a marginal TGT -+ TAT CAT + TAT G l u -3 9 + L y s S e r -4 0 + Phe A l a -4 1 + V a l A r g -1 2 8 j H i s C y s -l 7 4 + T y r P r o -2 8 9 + Leu Met -2 9 3 + Thr Phe -294 -+ Leu G l y -3 0 5 + S e r Arg-402 -+ T r p C y s -465 + Tyr H i s -5 1 5 j T v r Asn-288-Asp G l y -4 6 0 -+ Asp " Both isolates contained a second amino acid substitution, Val-248 + Ala. The two mutations were separated in uitro using standard molecular cloning methods; the MTR phenotype was found to he associated with the Met-293 + Thr change, while the Val-248 + Ala change was found to be silent.

I1
Kinetic constants of wild type and mutant anthranilate synthase complexes Kinetic analysis of the glutamine-dependent anthranilate synthase activity of the complexes was carried out as described under "Materials and Methods." The values for KmChr, KFrP and kcat are mean values from two or more independent determinations. Deviations from the mean in the individual experiments were less than or equal to k25% for KmChr and KTv and less than or equal to &30% for kcat.
Mutant enzymes are designated by the wild type residue followed by the position of the residue and the mutant residue, using the single amino acid code.  plots of steady state kinetic analysis of wild type and feedback-resistant mutant anthranilate synthase complexes. Conditions were as described under "Materials and Methods." Mutant enzymes are designated as in Table  11. Tryptophan concentrations were as follows: 0, none; 0, 2.5 p~; A, 5.0 p M ; W, 7.5 pM; V, 10 pM; *, 20 pM. The curues were drawn with either first order (C and D ) or second order ( A and E ) regression using the Sigma Plot software package (Jandel Scientific). decrease in KmCh'. The sensitivity of these mutant enzymes to feedback inhibition by 5-methyltryptophan relative to the wild type enzyme was found to parallel their sensitivity to tryptophan, indicating that differential sensitivity to 5-methyltryptophan did not account for their isolation (data not shown). Kinetic data for the wild type and for mutant enzymes representative of marginal, moderate, and complete loss of feedback control (H515Y, A41V, and M293T, respectively) are presented in Figs. 2 and 3. The slopes of the lines from Fig. 2 were plotted uersus the concentration of tryptophan. At concentrations of tryptophan where upward curvature was evident, slopes were approximated from a first order regression line drawn to the data obtained a t chorismate concentrations ranging between 2 and 12.5 p~ where the kinetics were essentially linear. Some data points in this range were omitted from the Lineweaver-Burk graphs for clarity.
The turnover number of most of the mutant enzymes varied only slightly (i.e. no greater than +30%) from that of the wild type enzyme (Table 11). Exceptions were E39K, C174Y, and N288D, which had reductions in kcat ranging from 2-to 3fold. Additional experiments (data not shown) indicated that the K,,, for glutamine, the substrate ligand of the amidotransferase domain of the TrpD subunit of the complex, was not significantly altered in any of the mutant enzymes (data not shown).

Tryptophan Binding in Wild Type and Mutant Enzymes-In order to ascertain whether the increases in
KYrp of the mutant enzymes were correlated with decreases in their affinity for tryptophan, dissociation constants for tryptophan (KdTrP) were determined by direct binding analysis using ultrafiltration. A summary of the results for all the enzymes is presented in Table 111. Langmuir double-reciprocal plots of the binding data obtained with the wild type enzyme and with two representative mutant enzymes that retain measurable binding activity (A41V and H515Y) are shown in Fig. 4. The KdTV of the wild type enzyme was found to be 4 pM, in agreement with the value determined previously by equilibrium dialysis (9). There was no detectable tryptophan binding in the four mutant enzymes with complete loss of feedback inhibition (E39K, S40F, M293T, and C465Y). On the other hand, those enzymes with moderate increases in KY" (A41V, N288D, P289L, and G305S) showed moderate (i.e. 3-16-fold) increases in KdTrP, while those with subtle changes in KYrP and KmChr (namely R128H, C174Y, R402W, G460D, and H515Y) had KdTV constants close to that of wild type. It is noteworthy that, in the partially feedback-resistant enzymes where KiTW constants were measurable, the magnitude of the kinetic effect of tryptophan binding, as indicated by the KdTrp/ KiTW ratio, was not constant and in every case was less than that observed with the wild type enzyme (Table 111). This difference was most pronounced in the enzymes with mutations in the middle region of the TrpE polypeptide.
The interaction coefficients ( n H ) of the wild type and mutant complexes were determined from Hill plots of the binding data and are included in Table 111. The n H of the wild type enzyme was 1.2, indicating positive cooperativity between the tryptophan binding sites of its two TrpE subunits. Several of the mutant enzymes lost cooperativity for tryptophan binding, Tryptophan dissociation constants and interaction coefficients of wild type and mutant anthranilate synthase complexes Tryptophan dissociation constants (KdT") and interaction coefficients (nH) were determined by ultrafiltration as described under "Materials and Methods." The values for KdTW and n H are the mean of two or more independent determinations. Deviations from the mean in the individual experiments were less than or equal to +25% for KdTrP and less than or equal to +lo% for n H . KdT"/K'" ratios were calculated using the KFW values from Table 11. Designations for the mutant enzymes are as in Table 11.  whereas others, in particular P289L (nH = 1.6), had increased cooperativity. Feedback Inhibition of Phosphoribosyl Transferase Activity-The phosphoribosyl transferase activity of the TrpD subunit of the complex is also subject to feedback inhibition by tryptophan. This regulation is the result of conformational effects associated with the binding of tryptophan to the TrpE subunit (13) and thus is a manifestation of communication between the regulatory site on TrpE and the catalytic site of the phosphoribosyl transferase domain of the TrpD subunit. In view of this, the feedback sensitivity of the phosphoribosyl transferase of the wild type and mutant enzymes was assessed by assaying activity under standard conditions in the presence of a range of concentrations of tryptophan.

AS-PRT
As found previously (8, ll), inhibition of the phosphoribosyl transferase activity of the wild type enzyme was partial, reaching a maximum of about 70% a t tryptophan concentrations of 5 ~L M and above; half-maximum inhibition (Trpos) was at 1.4 p~ tryptophan (Table IV). However, the sensitivity of the mutant enzymes varied widely, closely paralleling the affinity for tryptophan (i.e. the KdTV values) of the TrpE subunit of each complex. Specifically, the transferase activity of the E39K, S40F, M293T, and C465Y enzymes was completely resistant to tryptophan inhibition, even at concentrations as high as 100 p~, while the activity of the other enzymes displayed slight to moderate resistance with Trpo, values ranging from 1.6 to 21 ~L M ( Table IV). The extent of inhibition in the latter group did not change significantly, in all cases attaining a maximum of 60-70% (data not shown).
Dye-Ligand Chromatographic Behavior of Mutant Anthranilate Synthase Complexes-Dye-ligand chromatography utilizing Matrex Gel Orange A has previously been shown to be an effective method for the purification of the wild type anthranilate synthase-phosphoribosyl transferase complex (3). While the nature of the binding of the enzyme to the dyeconjugated agarose gel is not fully understood, it is likely to involve both electrostatic and hydrophobic interactions between the enzyme and the procion yellow dye (14). It has been suggested that the tryptophan-dependent elution of the complex from the gel results from a conformational change associated with the binding of this ligand to the TrpE subunit (3). Thus, the tryptophan-induced conversion of the enzyme from

IV
Feedback inhibition of phosphoribosyl transferase activity of wild type and mutant anthranilate synthase complexes Phosphoribosyl transferase activity was measured as described under "Materials and Methods" a t tryptophan concentrations varying between 0 and 100 FM. Trpo.,5 is the concentration of tryptophan necessary to achieve half-maximal inhibition; NI indicates no inhibition. a conformation with high affinity for the Orange A gel to one with low affinity may be a manifestation of the allosteric transition associated with ligand binding to TrpE. The behavior of the feedback-resistant mutant enzymes observed during their purification by dye-ligand chromatography supports this idea.

AS-PRT Complex
Like the wild type enzyme, each of the 14 mutant enzymes was found to bind tightly to the Orange A-agarose, with no more than 5-10% of the applied anthranilate synthase activity emerging in the wash fraction of the columns. However, the enzymes were differentiated into three groups on the basis of the elution profiles obtained upon development of the columns with an isocratic tryptophan gradient (Fig. 5). The enzymes of each group were related to one another by the magnitude of the alterations in their KiTrp and KdT" constants.
The wild type enzyme eluted from the Orange A-agarose in typical fashion as a sharp peak between 25 and 50 PM tryptophan (Fig. 5). Very similar elution patterns were found with the mutant enzymes with subtle changes in KY" and KdT" (i.e. N288D, R402W, G460D, and H515Y). The mutant enzymes with moderately elevated K?" and KdT" constants (i.e. A41V, P289L, F294L, and G305S) were also eluted by tryptophan but not as effectively as the wild type enzyme. The broad elution pattern of the P289L enzyme shown in Fig. 5 is representative of this group. Finally, the mutant enzymes that were devoid of tryptophan inhibition and binding (i.e. E39K, S40F, M293T, and C465Y) were completely refractory to tryptophan elution; however, they were readily recovered in excellent yield by salt elution following the tryptophan gradient as exemplified in Fig. 5 by S40F.
To test further the conclusion that alterations in conformational changes induced by tryptophan binding were the basis of these chromatographic results, equivalent amounts of the purified wild type enzyme and feedback-resistant S40F mutant enzyme were brought to 0.5 mM tryptophan and then applied to Orange A gel columns that had been equilibrated with standard buffer containing 0.5 mM tryptophan. The columns were developed with the same buffer, and fractions were collected and assayed for anthranilate synthase activity. Unlike the tight binding of the enzyme observed in the absence of tryptophan, greater than 80% of the activity of the wild type enzyme emerged in the void volume of the column (data not shown), indicating that presaturation of the enzyme with tryptophan substantially decreased its affinity for the gel. In contrast, the same pretreatment of the enzyme and gel with tryptophan had no effect on the chromatographic behavior of the S40F enzyme. As in the absence of tryptophan, greater than 98% of the S40F activity was bound to the gel and was recovered upon subsequent salt elution.

DISCUSSION
Kinetic characterization of the mutant anthranilate synthase complexes generated in this study has shown that feedback resistance can result either from moderate to large increases in KT" or from a combination of small increases in KT" and small decreases in KmChr (Table 11). Companion tryptophan binding studies with the mutant enzymes revealed increases in KdT" that generally paralleled the increases in K?" (Table 111). The majority of the enzymes were of the former type, with marked decreases in inhibitor affinity; a number of these were totally resistant to inhibition and had no detectable tryptophan binding. The mutational changes in this group are undoubtedly the best indicators of residues important for tryptophan binding and feedback regulation. The clustered arrangement of these mutations indicate that two noncontiguous regions of the TrpE polypeptide, i.e. residues 39-41 and 288-305, are components of the feedback site of the TrpE subunit. On the other hand, the residues identified by mutations having only marginal effects on the kinetic properties of the enzyme most likely contribute only indirectly to the structures involved in feedback regulation. These mutations may have been recovered by virtue of the amplification of their relatively weak effects by the multicopy state of the cloned trpE gene in the mutant strains, aided perhaps by reduced feedback sensitivity of the heterologous anthranilate synthase complexes present in the primary mutant isolates (12).
The properties of mutant enzymes such as E39K, S40F, and M293T, in which tryptophan binding was completely eliminated while the apparent affinity for chorismate remained essentially unaffected, indicate the existence of separate binding sites for substrate and inhibitor. This conclusion is supported by our finding that the feedback site mutations described here and the active site mutations previously characterized (3) are segregated to nonoverlapping regions of the TrpE polypeptide. All but one of the moderate to strong regulatory mutations are found in two clusters within the first 305 residues of the 520-residue polypeptide (Fig. 6). In contrast all of the six identified active site residues are located within the carboxyl-terminal third of the polypeptide, the most proximal one being residue Thr-329. It is significant that no inactivating missense mutations have yet been recovered in the amino-terminal two-thirds of TrpE, in spite of extensive random mutagenesis, and that nonconserv tive substitutions placed throughout this region by geneti 6 suppression of nonsense mutations are without effect on catalytic a~t i v i t y .~ Taken together, these findings provide strong evidence that the feedback site and the active site of TrpE are not only distinct but also reside in separate structural domains. In such a structural model, the regulatory domain would be composed of the first 310-320 residues, while the catalytic domain would be made up of the remaining carboxyl-terminal segment of about 200 residues. It is also reasonable to speculate that the two segments of the regulatory domain encompassing residues 39-41 and 288-305 interact to form the tryptophan binding '' R. Bauerle (15) and later verified in this laboratory by the dideoxy chain-terminating method. The sequence presented includes corrections in the original sequence a t residues 61,70,164,187,348,359,360,368,395,397, and 481. The sequences of 12 TrpE homologues were compared with the S. typhimurium sequence using the alignment of Crawford (16). Boldface capital letters, using the single letter amino acid code, designate identity at that position in at least 11 of the 13 sequences; lightface capital letters designate identity or a conservative replacement in a t least 9 of the 13 sequences; and lower case letters designate nonconserved residues in more than 5 of the 13 sequences. site. This model is consistent with the results of the proteolytic probing of the native TrpE subunit, which revealed the existence of a large amino-terminal domain with molecular mass of about 30 kDa and two smaller carboxyl-terminal domains of 16 and 12 kDa (3). Furthermore, photoaffinity labeling of the native subunit with 6-azidotryptophan, an effective false feedback inhibitor of the enzyme, has localized the tryptophan binding site to the putative amino-terminal regulatory d~m a i n .~ It is interesting to consider these results in view of the unique amino acid sequence conservation that exists among TrpE homologues. The DNA sequences of the trpE genes of more than a dozen bacterial and yeast species have been determined, and the derived amino acid sequences of the polypeptides have been aligned (16). As summarized in Fig.  6, the TrpE polypeptides are highly conserved, as expected. However, the sequence conservation is found predominately in the carboxyl-terminal half of the molecule, where 55% of the residues are invariant or highly conserved in 13 homologues compared. In contrast, only 13% of the residues of the amino-terminal half are conserved, and these are mostly located in a few stretches of limited similarity. Significantly, the two segments of the polypeptide identified here to be J. Hess and R. Bauerle, manuscript in preparation.

s V f d a f r l i q g v v n i p t q e r e a m f f g G l i A Y D l V a g F E a l p h l e a g n n~P D Y c f y L a g~
important for feedback inhibition of the S. typhimurium enzyme, i.e. residues 39-41 and 288-305, are within highly conserved regions that are separated by a very large stretch with little sequence conservation. On the other hand, the active site residues in TrpE are found exclusively at invariant or highly conserved residues within the highly conserved carboxyl terminus of the polypeptide. It is curious that sequence divergence has been so much greater throughout most of the putative amino-terminal regulatory domain of the TrpE polypeptide than in the carboxyl-terminal catalytic domain; nevertheless, all of the homologues are feedback-regulated by tryptophan. Although no detailed mutational analysis of a TrpE homologue from another organism has yet been reported, a single feedback-resistant mutation in the TrpE subunit of Brevibacterium lactofermentum has been identified a t Ser-38, which aligns with Ser-40 of the S. typhimurium enzyme (17).
One apparent discrepancy with respect to the domain model proposed here for TrpE is the strong feedback-resistant mutation at Cys-465 (Table 11). However, several factors suggest that Cys-465 is not likely to be an essential residue within the feedback site. Unlike all the other residues altered in the feedback-resistant enzymes, residue 465 is not highly conserved (Fig. 6), with cysteine being replaced by alanine, glycine, or serine in 5 of the 13 TrpE homologues. Moreover, in chemical modification studies of the TrpE polypeptide from Serratia marcescens with the sulfhydryl reagent, 5,5'dithiobis-2-nitrobenzoic acid (13), there was no indication that this or any other cysteine residue was protected by tryptophan. Lastly, aside from the loss of feedback inhibition, the C465Y enzyme also suffered a significant increase in KmChr (Table 11). Thus it may be that Cys-465 is positioned in the folded structure of TrpE such that the bulky aromatic substitution in the C465Y mutant enzyme perturbed both the feedback site and the active site of the TrpE subunit. Interactions between the putative regulatory and catalytic domains are also suggested by the fact that several of the regulatory mutations (e.g. E39K, C174Y, and N288D) had significant effects on kc,, (Table 11) and by the finding that some inactivating active site mutations simultaneously altered the apparent affinity of the enzyme for tryptophan (as measured by the relative sensitivity of the PRT activity to feedback inhibition).3 There are indications that conformational properties and subunit interactions were also affected in the mutant complexes. Those that were not completely feedback-resistant had a decreased response to tryptophan binding, as indicated by the reduced KdTrP/KFV ratios, and some of these had changes in the apparent cooperativity of tryptophan binding (Table 111). Also, the sensitivity of the phosphoribosyl transferase activity to tryptophan inhibition decreased in all the mutant enzymes in parallel with decreases in the apparent affinity for the inhibitor (Table IV). Similarly, the effectiveness of tryptophan elution of the mutant enzymes from Orange A-agarose declined in parallel with the decline in the avidity of tryptophan binding (Fig. 5).
The conformational properties of the anthranilate synthase complex are presently not well understood. However, the results of dye-ligand chromatography suggest that the native enzyme exists in a conformational state intermediate to the classical T and R states proposed for allosteric proteins (18). Whereas the unliganded enzyme has strong affinity for the Orange A-agarose, both the T state ligand, tryptophan (Fig.  5)