Kinetic characterization of channel impaired mutants of tryptophan synthase.

Tryptophan synthase, an alpha 2 beta 2 tetrameric complex, is a classic example of an enzyme that is thought to "channel" a metabolic intermediate (indole) from the active site of the alpha subunit to the active site of the beta subunit. The solution of the three-dimensional structure of the enzyme from Salmonella typhimurium provided physical evidence for a 25-A hydrophobic tunnel which connects the alpha and beta active sites (Hyde, C. C., Ahmed, S. A., Padlan, E. A., Miles, E. W., and Davies, D. R. (1988) J. Biol. Chem. 263, 17857-17871). Using rapid reaction kinetics, we have previously established that indole is indeed channeled and have identified three essential kinetic features which govern efficient channeling. In the current study we have probed the necessity of these features by using site-directed mutagenesis to alter these requirements. We now report the kinetic characterization of two mutants which contain substitutions to block or restrict the tunnel (beta C170F and beta C170W). Preliminary kinetic and structural evidence of a restricted tunnel in the beta C170W has been provided (Schlichting, I., Yang, X. W., Miles, E. W., Kim, A. Y., and Anderson, K. S. (1994) J. Biol. Chem. 269, 26591-26593). The rapid kinetic analysis of these mutant proteins shows that these mutations interfere with efficient channeling of the indole metabolite such that indole can be observed in single enzyme turnover of the physiologically relevant alpha beta reaction. In addition, the beta C170W mutant appears to be impaired in alpha beta intersubunit communication.

Tryptophan synthase catalyzes the last two reactions in the biosynthesis of tryptophan (for reviews, see Miles (1979Miles ( , 1991). The ␣ subunit catalyzes the cleavage of indole 3-glycerol phosphate (IGP) 1 to indole and glyceraldehyde 3-phosphate (G3P) (␣ reaction), while the ␤ subunit catalyzes the condensation of indole with serine in a reaction mediated by pyridoxal phosphate (␤ reaction). The physiologically important reaction is termed the ␣␤ reaction and involves the conversion of IGP and serine to tryptophan and water (Demoss, 1962;Creighton, 1970;Matchett, 1974). These reactions are shown in Scheme I. Tryptophan synthase, an ␣ 2 ␤ 2 tetrameric complex, is considered a classic example of an enzyme that is thought to exhibit substrate channeling: a process in which a metabolic intermediate is directly transferred from one active site to another without free diffusion. The solution of the three-dimensional structure of the enzyme from Salmonella typhimurium provides physical evidence for a 25-Å hydrophobic tunnel which connects the ␣ and ␤ active sites (Hyde et al., 1988). In this case, the metabolic intermediate (indole) would be transferred from the active site of the ␣ subunit to the active site of the ␤ subunit through the connecting tunnel. Using rapid reaction kinetics, we have previously established that indole is indeed channeled and have identified three essential kinetic features which govern efficient channeling. Previous rapid kinetic analysis of tryptophan synthase has shown that the product of the ␣ reaction, indole, is transferred to the ␤ site where it reacts with serine to form tryptophan (Anderson et al., 1991;Lane and Kirschner, 1991;Brzovic et al., 1992). The transfer of the indole intermediate is highly efficient such that indole cannot be observed in a single enzyme turnover of the ␣␤ reaction. The efficient channeling of indole is a consequence of three features of the reaction kinetics. 1) The rate of diffusion of indole through the channel is very fast; 2) the reaction of indole to form tryptophan at the ␤ site is fast and largely irreversible; and 3) the reaction of serine at the ␤ site modulates the formation of indole at the ␣ site or in the tunnel such that indole is not produced until serine has reacted with pyridoxal phosphate to form the highly reactive aminoacrylate (Anderson et al., 1991). This intersubunit communication keeps the ␣ and ␤ reactions in phase so that indole does not accumulate at the ␣ site. The combination of these three aspects of the reaction kinetics leads to efficient channeling of indole by maintaining a low concentration of indole bound to the enzyme. Accordingly, this model for channeling of the indole intermediate predicts that blocking or impeding the passage between active sites might allow detection of indole during a single enzyme turnover. One of the residues lining the tunnel is ␤C170. Sitedirected mutagenesis of the ␤C170 to Phe and Trp has been used to construct mutants of tryptophan synthase in which the tunnel is restricted (Schlichting et al., 1994;Ruvinov et al., 1995). In this report we describe the kinetic analysis of the ␤C170F and ␤C170W mutants by rapid reaction kinetics similar to that previously conducted for wild type tryptophan synthase (Anderson et al., 1991).

MATERIALS AND METHODS
Enzyme Purification-Tryptophan synthase ␣ 2 ␤ 2 complex from S. typhimurium was isolated and purified as described . The ␤C170F and ␤C170W mutants of the ␣ 2 ␤ 2 complex tryptophan synthase were produced as described previously (Ruvinov et al., 1995) and purified in a manner analogous to wild type.
Enzyme Assays-The activities of the ␣ 2 ␤ 2 complex in Reactions 1 and 2 (Table I) were measured by spectrophotometric assays at 37°C (Creighton, 1970). Reaction 3 was measured by a direct spectrophoto-metric assay by monitoring the difference in absorption between indole and L-tryptophan at 290 nm . Reaction 4 was measured by a spectrophotometric assay coupled with lactic dehydrogenase (Crawford and Ito, 1964). The original assay for the activity of the ␤ subunit was modified for the ␣ 2 ␤ 2 complex by replacing ammonium chloride with 0.18 M sodium chloride.
Chemicals-Unlabeled and 14 C-labeled IGP were synthesized enzymatically using tryptophan synthase in the reverse ␣ reaction with indole or [ 14 C]indole and G3P as described previously (Kawasaki et al., 1987). Radiolabeled indole (specific activity 50 mCi/mM) was purchased from Research Products International (Mount Prospect, IL). Glyceraldehyde 3-phosphate was prepared from the dimethyl acetal of glyceraldehyde 3-phosphate (Boehringer Manneheim) using the method of Racker et al. (1959). Unlabeled standards of indole and tryptophan and serine were purchased from Aldrich. All buffers and other reagents employed were of highest commercial purity. Millipore ultrapure water was used for all solutions. All experiments were conducted at 37°C in sodium Bicine buffer (50 mM) at pH 7.8 Stopped-flow Experiments-Stopped-flow measurements were performed using a Kintek SF-2001 (Kintek Instruments, State College, PA). This apparatus has a 1.5-ms dead time, a 0.5-cm path length, and a thermostatted observation cell maintained at 37°C. The change in protein conformation upon serine binding was monitored by following protein fluorescence using a monochromator set at 280 nm on the input and monitoring the changes in intrinsic enzyme fluorescence with an output filter at 340 nm. The serine-pyridoxal phosphate complexes were examined by excitation at 405 nm and monitoring fluorescence at 500 nm. In most experiments an average of four runs was used for data analysis and a minimum of a 5-fold excess of the variable substrate over enzyme was used to allow analysis as a pseudo-first order rate constant. Rate constants were obtained by fitting the data to a single or double exponential by non-linear regression.
Rapid Quench Experiments-The rapid quench experiments were performed using a Kintek RFQ-3 Rapid-Chemical Quench (Kintek Instruments). The reaction was initiated by mixing the enzyme solution (15 l) with the radiolabeled substrates (15 l, approximately 20,000 dpms). In all cases, the concentrations of enzyme and substrates cited in the text are those after mixing and during the enzymatic reaction. The reaction mixture was then quenched by mixing with 67 l of 0.6 M KOH to give a final concentration of 0.2 M KOH during mixing in the rapid quench apparatus. The quenched reaction solution was collected in a 1.5-ml Eppendorf tube, vortexed, and analyzed by HPLC. The substrate and products were quantified by HPLC as described below. In order to ensure that the base was quenching the enzymatic reaction, a control was included with each experiment to insure that catalysis was being terminated. This involved adding the enzyme to a premixed solution of base and substrates. Control experiments were also done to establish the stability of the radiolabeled substrates under the quench conditions employed.
HPLC Analysis-The substrates and products were quantified by HPLC using a radioactivity detector. The HPLC separation was performed using a BDS-Hypersil-C18 reverse phase column (250 ϫ 4.6 mm, Keystone Scientific, Bellefonte, PA) with a flow rate of 1 ml/min. The following gradient separation was employed where solvent A is 0.2% sodium bicarbonate and solvent B is methanol. The linear gradient program was as follows: 0 -100% B in 20 min, hold at 100% B for 10 min, recycle (100 -0% B in 15 min), and then re-equilibrate by holding at 100% A for 5 min. The elution times were as follows: IGP, 7 min; tryptophan, 19 min; and indole, 23 min. The HPLC effluent from the column was mixed with liquid scintillation mixture (Mono-flow V, National Diagnostics) at a flow rate of 4 ml/min. Radioactivity was monitored continuously using a Flo-One radioactivity detector (Radiomatic Instruments, Tampa, FL).
Data Analysis-The KINSIM kinetic simulation program (kindly provided by Carl Frieden and Bruce Barshop, Washington University, St. Louis, MO; Barshop et al., 1983) was used to model all of the kinetic data presented in this paper. The program was modified to allow the input of data from the rapid quench experiments as x,y pairs and to calculate the sum square errors in fitting the data (Anderson et al., 1988). The entire data set was fit to a single model and a single set of rate constants. The data were fit by a trial and error process maintaining the constraints of known dissociation constants and K m values for IGP, G3P, indole, tryptophan, and serine Kawasaki et al., 1987;Phillips et al., 1984). Not all of the rate constants are known with equal certainty. In the ␣ reaction the rates of IGP binding in the forward direction and of indole and G3P binding in the reverse reaction are rough estimates based upon the concentration dependence of the single turnover kinetics in each direction. The corresponding dissociation rates for IGP, indole, and G3P are then estimated from the approximate dissociation equilibrium constants. Although these constants are not known with certainty, their values do not greatly affect the fitting process or the interpretation. The remaining rate constants involving the chemical interconversion of IGP with indole and G3P were obtained as fits to the single turnover kinetics in each direction and are therefore known with more certainty. Regression analysis in the final stages of the fitting process place the limits of error on the rate constants of approximately Ϯ20%. In the ␤ reaction, the rates of serine reaction to form the external aldimine (E-Ser) and the aminoacrylate (E ϳ AA) were obtained by direct fitting of stopped-flow fluorescence data as a function of serine concentration and are within Ϯ10%. The rates of indole channeling and the reaction of indole with the aminoacrylate to form tryptophan are set at the lower limit sufficient to account for the accumulation of indole in the single turnover (Figs. 1, 3, and 6). The rate of tryptophan release is determined by the presteady state burst experiments as well as the steady state turnover rate analogous to previous analysis of wild type tryptophan synthase. The reverse rate constants governing rebinding and reaction of tryptophan are set according to K m and k cat values for the reverse ␣ reaction for wild type tryptophan synthase (Weischet and Kirschner, 1976;Ahmed et al., 1986), although they do not affect the interpretation of the data presented in this paper. SCHEME I. a Assays at 37°C as described under "Materials and Methods." b Binding constants were determined from reciprocal plots of the fluorescent increment (emission at 510 nm with excitation at 420 nm) on addition of L-serine (0.15-62 mM).

RESULTS
The rapid kinetic analysis described in this paper leads to the pathways for the ␤C170F mutant and ␤C170W mutant shown in Schemes II and III, respectively. Key rate constants obtained in Schemes II and III are summarized in Table II and compared with those obtained for the wild type ␣ 2 ␤ 2 complex (Anderson et al., 1991). The curves shown in Figs. 1-8 were calculated by numerical integration using one set of rate constants for each mutant as summarized in Schemes II and III, with no simplifying assumptions. The rates of indole channeling and the reaction of indole with the aminoacrylate to form tryptophan are set at the lower limit sufficient to account for the accumulation of indole in the single turnover experiments. In fitting the data, the rate constants for the reaction were further constrained by dissociation constants and K m values for IGP, G3P, serine, and tryptophan as previously determined Kawasaki et al., 1987;Phillips et al., 1984) and in Table I in this work.

Steady State Kinetic Characterization of Mutants
A comparison of steady state kinetic parameters for wild type and mutant tryptophan synthase is shown in Table I. The steady state enzyme turnover rates of the wild type and mutant ␣ 2 ␤ 2 complexes are compared in Reactions 1-4. This characterization was important for providing direction for an in-depth transient kinetic analysis. There were several differences which are noteworthy of mention. First, the rate of the ␣ reaction (Reaction 1) in the absence of serine for the ␤C170F mutant is similar to that observed for wild type (0.16 s Ϫ1 ) while the rate for the ␤C170W mutant is 2-fold slower (0.08 s Ϫ1 ). Second, the K d for serine was similar for wild type and ␤C170F mutant (0.5 mM) while the K d for the ␤C170W mutant was almost 100-fold higher (50 mM). Our previous analysis with wild type tryptophan synthase has shown that the binding of serine stimulates the rate of cleavage of IGP (␣ reaction). This modulation of activity is mediated by an intersubunit communication in which catalysis at the ␣ subunit is triggered by the pyridoxal phosphate activation of serine to form an aminoacrylate species and a concomitant protein conformational change. This process in turn is coupled to the production of indole at the ␣ site and channeling to the ␤ site. The steady state rate of the ␣␤ reaction (Reaction 2) is enhanced 30-fold for wild type (4.7 s Ϫ1 ) and 7.5-fold for ␤C170F mutant (1.2 s Ϫ1 ) in the presence of serine (Table I). On the other hand, a less than 2-fold activation was noted with the ␤C170W mutant (0.14 s Ϫ1 ). The slower rate of enhancement, particularly in the case of the ␤C170W mutant, suggests that these mutants may be impaired in intersubunit communication as well as channeling. The transient kinetic experiments described below were designed to probe the channeling of indole and intersubunit communication in the tryptophan synthase mutants. Finally, in the absence of ␣ site ligands (indole or IGP) tryptophan synthase catalyzes a side reaction, the slow hydrolysis of serine to form pyruvate, ammonia, and water (Table I, Reaction 4). The rates in Reaction 4 of the ␤C170F and ␤C170W mutants are 2-and 4-fold higher, respectively, than that of the wild type ␣ 2 ␤ 2 tryptophan synthase. Since this side reaction may complicate interpretation of results in experiments in which serine is preincubated with enzyme, some of the single turnover experiments were carried out by simultaneously mixing enzyme with a solution of IGP and serine.

Kinetic Characterization of the ␤C170F Mutant
The rapid kinetic analysis of the ␤C170F mutant described below leads to the kinetic pathway summarized in Scheme II and Table II. The rate constants which distinguish wild type and mutant are highlighted in boxes.
Kinetics of a Single Turnover: The ␣␤ Reaction-We have previously found that the examination of a single enzyme turnover of the ␣␤ reaction is an important diagnostic experiment to evaluate channel impairment. One of the essential features of efficient channeling of indole is that the transfer of indole from the ␣ to the ␤ subunit should be rapid such that no indole should accumulate during a single enzyme turnover. If passage of indole is blocked or impeded then it is likely to be observed during this type of experiment. (In this experiment, excess enzyme (20 M) was mixed with a solution of serine (10 mM) containing a limiting concentration of [ 14 C]IGP (8.7 M), and the disappearance of IGP and production [ 14 C]indole and tryptophan were monitored.) The observed production of indole ( Fig. 1) provides the first kinetic evidence that the channeling is impaired with the ␤C170F ␣ 2 ␤ 2 complex.
Kinetics of Reaction of Indole from Solution-In order to establish that the indole is channeled from the ␣ site to the ␤ SCHEME II. SCHEME III.
site, we performed an experiment to assess the alternate pathway involving dissociation of indole from the ␣ site and fast rebinding to the ␤ site. The rate of reaction of indole from solution was measured in a single turnover experiment using excess enzyme (20 M) preincubated with serine (10 mM). The enzyme-serine complex was mixed with a limiting concentration of [ 14 C]indole (6 M) and the production of [ 14 C]tryptophan was monitored. The disappearance of indole and the formation of tryptophan are shown in Fig. 2. The rate of indole reaction from solution is approximately 0.6 s Ϫ1 which is limited by the rate of indole binding to the enzyme as confirmed by experiments performed at different enzyme concentrations (data not shown). These data provided a second-order rate constant of 0.03 M Ϫ1 s Ϫ1 for the binding of indole to the enzyme. As shown in Fig. 2B, the observed rate of reaction of indole from solution (0.6 s Ϫ1 at 20 M enzyme) does not account for the amount of indole we observe in the ␣␤ single turnover. The combination of the two single turnover experiments (Figs. 1 and 2) indicates that indole is channeled from the ␣ site to the ␤ site in the physiologically relevant ␣␤ reaction. Because the reaction of indole from solution to form tryptophan is slow (0.6 s Ϫ1 ) relative to wild type (ϳ40 s Ϫ1 ), the indole observed during the conversion of IGP to tryptophan in the ␣␤ reaction must have been channeled otherwise a smaller amount of indole would have accumulated in solution (see Fig.   2B). However, the rate of channeling from the ␣ to ␤ site must have slowed from Ͼ1000 s Ϫ1 in wild type to 100 s Ϫ1 in order to account for the observation of any indole. Therefore, based upon this computer simulation of the reaction kinetics, we conclude that indole must pass through the channel and it must do so more slowly than in the wild type enzyme.
Kinetics of the ␣␤ Presteady State Burst Reaction-We sought to provide additional information which would discern whether the indole observed in the single turnover experiment  Anderson et al. (1991). b Scheme II, Figs. 1-4. c Scheme III, Figs. 6 -8. was actually trapped in the tunnel or free in solution. Molecular modeling studies based upon the tryptophan synthase crystal structure have suggested that the hydrophobic channel could accommodate up to 4 molecules of indole per enzyme site (Hyde et al., 1988). Therefore we examined the ␣␤ reaction in a presteady state burst experiment in which substrate is in slight excess over enzyme, to determine whether indole which is trapped in the tunnel builds up to a constant level (Fig. 3).
The time course for the forward reaction was determined by mixing a solution containing enzyme and serine (pre-mixed) with a solution containing [ 14 C]IGP to initiate the reaction. The final concentrations after mixing were 20 M enzyme, 10 mM serine, and 200 M [ 14 C]IGP. At various times after mixing, the reaction was quenched by the addition of 0.2 M KOH (final concentration) and the quenched reaction mixture was then analyzed by HPLC as described under "Materials Methods." The time dependent disappearance of IGP and formation of radiolabeled indole and tryptophan are shown in Fig. 3. The results show a presteady burst of product formation with an amplitude of 0.2 per enzyme site and a rate of 30 s Ϫ1 which was followed by a slower rate of 3 s Ϫ1 corresponding to steady state turnover. As anticipated, we observed the accumulation of a relatively constant quantity of indole corresponding to concentration of approximately 15-20 M or ϳ1 indole per enzyme site. If the indole formed was in solution, a very slow accumulation of indole with a lag would be predicted (see Fig. 3B) and as illustrated does not fit the experimental data. Thus this experiment provides further support for the suggestion that the observed indole is in the tunnel rather than in solution.
Kinetics of the ␣ Reaction-Our previous examination of the conversion of IGP to indole in a single turnover (␣ reaction) in the absence of serine with wild type tryptophan synthase showed that there is not a presteady state burst of indole formation. This indicated that the slow rate of IGP cleavage in the absence of serine is not the result of indole that is held very tightly at the active site of the ␣ subunit and released into solution very slowly (Anderson et al., 1991). Similar results were obtained with the ␤C170F mutant (data not shown) as no burst of indole formation is observed; rather the very slow formation of indole occurs at the steady state rate of (0.16 s Ϫ1 ). Since chemistry is not rate-limiting (see below), this experiment suggests that a step preceding chemistry is rate-limiting. Repeating the experiment at a higher enzyme concentration did not give an increase in rate, indicating that the initial binding rate for IGP was not rate-limiting. In addition, the observed rate was equal to the steady state rate at saturating IGP concentration. Taken together, these data suggest that there is a rate-limiting conformational change which must occur after the initial binding of IGP and limits the rate of the chemical reaction. In the presence of serine this conformational change must occur much faster. These results are very similar to those previously obtained with wild type tryptophan synthase. The rate of the chemical reaction observed here is consistent with the rate of IGP cleavage observed in the forward ␣␤ reaction (with serine), suggesting that only the rate of the conformational change which limits the ␣ reaction (IGP-E 7 IGP-E*) is affected by the presence of serine.
Serine Modulates the ␣ Reaction-In the kinetic pathway of wild type tryptophan synthase, it has been shown that a protein conformational change occurs prior to and limits the rate of IGP cleavage. The rate of tryptophan formation (in the presence of serine) is much faster than the cleavage of IGP in the absence of serine, since serine increases the rate of this protein conformational change occurring at the ␣ subunit. According to this model, the enzyme is converted from a less active state to a more active state upon reaction with serine at the ␣ site. Support for this model was provided in earlier studies with wild type enzyme by comparison of a single turnover ␣␤ reaction in which serine is preincubated with enzyme versus the simultaneous addition of IGP and serine to enzyme. A lag in tryptophan production was observed in the experiment in which there was no preincubation, indicating that IGP cleavage at the ␣ site is slow until serine binds to the ␤ site. A similar comparison could not be made with the ␤C170F mutant since the faster rate of serine hydrolysis complicated the ␣␤ single turnover experiment in which enzyme is preincubated with serine. Nonetheless, this serine activation step appears to be present in the kinetic pathway of the ␤C170F mutant as evidenced by the slow rate of IGP cleavage in the absence of serine and the much faster IGP cleavage during the single turnover of the ␣␤ reaction in the presence of serine. Our estimated rate of IGP cleavage in the presence of serine (24 s Ϫ1 in Scheme II and Table II) is 150-fold higher than the rate in the absence of serine (0.16 s Ϫ1 in Scheme II and Table II). This 150-fold activation by serine is similar to that observed with the wild type enzyme (Anderson et al. (1991) and Table II). The magnitude of this activation step is masked in the steady state comparison since product release is the rate-limiting step.
Kinetics of Serine Reaction Measured by Stopped-flow Fluorescence Methods-To further define the activation step we , and tryptophan (f) were monitored. The curves were calculated by the numerical integration using the rate constants summarized in Scheme II. B, the curves were simulated using a rate of indole binding from solution of 0.03 M Ϫ1 s Ϫ1 (0.6 s Ϫ1 at 20 M enzyme). The rate of reaction of indole from solution is too slow to account for the amount of indole observed in the ␣␤ reaction above. examined the kinetics of serine binding. It has previously been shown that there are absorbance and fluorescence changes exhibited by the pyridoxal phosphate upon reaction with serine and chemical nature of the serine-pyridoxal phosphate complexes have been defined (Faeder and Hammes, 1971;York, 1972;Lane and Kirschner, 1983;Lane et al., 1984;Drewe and Dunn, 1985). In the first step, serine displaces the internal aldimine formed between Lys-87 and the pyridoxal phosphate at the active site of the ␤ subunit to form the external aldimine (E-Ser in Schemes II and III), which is highly fluorescent, with emission at 500 nm upon excitation at 405 nm. The ␣ proton of serine is then removed to form a quinonoid species which is subsequently dehydrated to form a reactive aminoacrylate (E ϳ AA in Schemes II and III). The electrophilic aminoacrylate is then poised to react with the incoming nucleophile, indole, in the physiological reaction to form tryptophan. If no indole is present the aminoacrylate is slowly hydrolyzed to form pyridoxal phosphate, pyruvate, and ammonia (Reaction 4 in Table I).
In the current study, we focused on measuring the rate of formation of the external aldimine and of its decay to the aminoacrylate by stopped-flow fluorescence because of the increased sensitivity of these methods. A representative fluorescence trace is shown in Fig. 4A. The data were fit to a double exponential to measure the rates of formation and decay of the external aldimine. The dependence of the rates of the fast and slow phases on serine concentration are shown in Fig. 4, B and C. The data for the faster phase (formation of the external aldimine) were fit to a straight line giving a second-order rate constant of 0.03 M Ϫ1 s Ϫ1 and showed no signs of leveling off at the higher serine concentrations (Fig. 4B). The concentration dependence of the rate of the slower phase (decay to the aminoacrylate) was fit to a hyperbola which reached an observed maximum with a rate of 12 s Ϫ1 . Complete solution for the two-step reaction kinetics (Johnson, 1992) yielded all four rate constants for the reaction as shown below and summarized in Scheme II: with k 1 ϭ 0.03 M Ϫ1 s Ϫ1 , k Ϫ1 ϭ 31 s Ϫ1 , k 2 ϭ 5 s Ϫ1 , and k Ϫ2 ϭ 7 s Ϫ1 . Thus the rate of aminoacrylate formation for the ␤C170F mutant (12 s Ϫ1 ) is approximately 4-fold slower than wild type (45 s Ϫ1 , Table II).

Changes in Protein Conformation Upon Serine
Binding-Our earlier studies have indicated that the activation at the ␣ site is mediated by a change in protein conformation. A change in intrinsic protein fluorescence was observed when wild type tryptophan synthase was mixed with serine. It was found that the rate of the protein conformational change was coincident with the rate of formation of aminoacrylate. These data suggest that the formation of the aminoacrylate at the ␤ site elicits a protein conformational change which is transmitted to the ␣ site where it triggers the cleavage of IGP to indole and G3P (Anderson et al., 1991). Similar results were obtained with the ␤C170F mutant. A representative trace illustrating the protein conformation is shown in Fig. 5A. The concentration dependence of the rate of the fluorescence change (Fig. 5B) was fit to a hyperbola giving a maximum rate of 15 s Ϫ1 . In conclusion, the kinetics of serine binding occur at a slower rate than wild type. The rate of the aminoacrylate formation (12 s Ϫ1 ) is accompanied by a protein conformational change occurring at a similar rate (15 s Ϫ1 ).

Kinetic Characterization of the ␤C170W Mutant
The effects on efficient channeling of indole proved much more pronounced with the ␤C170W mutation. Earlier studies on the ␤C170W mutant provided kinetic and structural evi- FIG. 4. Kinetics of binding of serine of the ␤C170F mutant tryptophan synthase ␣ 2 ␤ 2 complex to the ␤ subunit. A, a representative trace from a stopped-flow experiment shows the time dependence of formation and decay of the highly fluorescent serine-pyridoxal species termed the external aldimine after mixing enzyme (2 M) with serine (5000 M). Fluorescence excitation was at 405 nm and emission was observed at 500 nm. The smooth line shows the fit to a double exponential with rates for the fast and slow phases of 172 s Ϫ1 and 12.9 s Ϫ1 , respectively. B, the concentration dependence of the rate of serine reaction for the fast phase. The rate of the faster phase was fit to a line having a slope of 0.03 M Ϫ1 s Ϫ1 which is the second-order rate constant for the formation of the external aldimine species. C, the concentration dependence of the rate of serine reaction for the slower phase. The rates of the slower phase were fit to a hyperbola having a maximum rate of 12.9 s Ϫ1 which corresponds to the rate of formation of the aminoacrylate species. dence that the channel was impaired (Schlichting et al., 1994). The steady state kinetic data as well as the transient kinetic data presented below indicate that the mutation has not only blocked or impeded the passage of indole from the ␣ to ␤ site but also has affected ␣␤ intersubunit communication. The rapid kinetic analysis of the ␤C170W mutant described below leads to the kinetic pathway summarized in Scheme III and Table II. This in-depth analysis provided results in reasonable agreement with a previous preliminary report (Schlichting et al., 1994). The rate constants which distinguish the wild type from ␤C170W are highlighted in boxes.
Kinetics of a Single Turnover: The ␣␤ Reaction-A single enzyme turnover of the ␣␤ reaction for the ␤C170W mutant is shown in Fig. 6. In this experiment, excess enzyme (20 M) was mixed with a solution of serine (100 mM) containing a limiting concentration of [ 14 C]IGP (6 M), and the disappearance of IGP and production [ 14 C]indole and tryptophan were monitored. As shown in the figure a small amount of indole is observed during the conversion of substrate to product. The rate of IGP cleavage to indole is similar in the presence (0.12 s Ϫ1 ) or absence (0.08 s Ϫ1 ) of serine. When the experiment was repeated at higher enzyme concentration there was no change in the rate of indole or tryptophan formation indicating that the binding of IGP is not limiting the reaction. Since the rate of IGP cleavage is so slow and there is little or no activation of the ␣ reaction by the presence of serine, we do not know if the indole we observe is in solution or trapped in the tunnel.
Kinetics of Reaction of Indole from Solution-The rate of reaction of indole from solution was measured in a single turnover experiment using excess enzyme (20 M) preincubated with serine (100 mM). The enzyme-serine complex was mixed with a limiting concentration of [ 14 C]indole (2.4 M) and the production of [ 14 C]tryptophan was monitored. The disappearance of indole and the formation of tryptophan are shown in Fig. 7. A comparison of the time scale for Figs. 6 and 7 shows that the ␤ reaction is at least 20-fold faster. The observed rate of indole reaction from solution is approximately 0.35 s Ϫ1 . The rate of the ␤ single turnover reaction is limited by the rate of indole binding to the enzyme as confirmed by experiments performed at different enzyme concentrations (data not shown). These data provided a second-order rate constant of 0.03 M Ϫ1 s Ϫ1 for the binding of indole to the enzyme. This was similar to the rate determined for the ␤C170F mutant and approximately 60-fold slower than wild type (Table II). Since the rate of reaction of indole from solution is similar to the rate of IGP cleavage in the single turnover experiment (0.35 s Ϫ1 versus 0.12 s Ϫ1 ) it is difficult to distinguish whether the indole we observed is in solution or is trapped in the tunnel. However, computer simulations suggest that if the channeling step is eliminated no indole would be observed. Therefore, the indole we observe in the single turnover experiment may actually be trapped in the tunnel.
Kinetics of the ␤ Presteady State Burst Reaction-A presteady state burst experiment to examine the ␣␤ reaction for the ␤C170W mutant will not be informative since the reaction will be limited by the rate of IGP cleavage to indole and therefore no burst of product will be observed. On the other hand, a presteady state analysis of the ␤ reaction did show a burst of product formation. The time course for the forward reaction was determined by mixing a solution containing enzyme and serine ( under "Materials and Methods." The time dependent disappearance of indole and formation of radiolabeled tryptophan are shown in Fig. 8. A presteady burst of product formation was observed with an amplitude of 0.4 per enzyme site and a rate of 6 s Ϫ1 (indole binding rate-limiting) followed by a slower rate of 1 s Ϫ1 corresponding to steady state turnover.
Kinetics of the ␣ Reaction-The examination of the ␣ reaction for the ␤C170W mutant revealed a slow rate of IGP cleavage (0.08 s Ϫ1 ) similar to ␤C170F and wild type (data not shown). As anticipated there was no burst of product indole formation and the rate did not increase with higher enzyme concentration indicating the binding to the enzyme was not rate-limiting. Again, these data indicate that there is a rate-limiting conformational change which must occur after the initial binding of IGP and limits the rate of the chemical reaction.
Serine Does Not Modulate the ␣ Reaction-One of the most surprising results with the ␤C170W mutant was the lack of substantial activation of the ␣ reaction in the presence of serine, especially since the ␤C170F mutant behaved very similar to wild type. In order to understand this lack of stimulation by serine we wanted to take a closer look at the interaction of serine with the enzyme. There are several possibilities which may have lead to the observed results. First the aminoacrylate formation may be altered since it is this chemical species which is involved in the activation step. Second, there may be a problem with ␣␤ intersubunit communication due to an alteration in the protein conformational change. Finally, although the ␤C170W is remote from the ␣ active site, it could nonetheless be important in the triggering step. This possibility might be less likely since the ␤C170F does not exhibit the same type impairment.
Kinetics of Serine Reaction Measured by Stopped-flow Fluorescence Methods-The rate of formation of the external aldimine and its decay to the aminoacrylate were measured by stopped-flow fluorescence as described above. We could not observe the aminoacrylate at lower concentrations of serine. However, at higher concentrations of serine (close to the K m ) the decay of the external aldimine to the aminoacrylate was observed similar to that observed for the ␤C107F mutant (data not shown). The data over a range of serine concentrations were fit to a double exponential to measure the rates of formation and decay of the external aldimine. The dependence of the rates of the fast and slow phases on serine concentration was determined. The data for the faster phase (formation of the external aldimine) were fit to a straight line giving a secondorder rate constant of 0.002 M Ϫ1 s Ϫ1 and showed no sign of leveling off at the higher serine concentrations. The concentration dependence of the rate of the slower phase (decay to the aminoacrylate) was fit to a hyperbola which reached an observed maximum with a rate of 12 s Ϫ1 . Complete solution for the two-step reaction kinetics (Johnson, 1992) yielded all four rate constants for the reaction: with k 1 ϭ 0.002 M Ϫ1 s Ϫ1 , k Ϫ1 ϭ 156 s Ϫ1 , k 2 ϭ 9 s Ϫ1 , and k Ϫ2 ϭ 3 s Ϫ1 as summarized in Scheme III. These results indicate that the equilibrium between external aldimine and aminoacrylate is substantially changed such that a very high concentration of serine is required to shift the equilibrium toward aminoacrylate.
Changes in Protein Conformation Upon Serine Binding-When we examined the intrinsic protein fluorescence upon binding serine, a protein conformational change was not observed even at higher serine concentrations. It is possible that the protein conformational change necessary to activate the ␣ reaction has now become uncoupled from the formation of the aminoacrylate. An alternate explanation involves interference or fluorescence quenching due to the proximity between the ␤170 and ␤177 tryptophans. The ␤177 tryptophan is located approximately 7 Å from ␤170 tryptophan 2 and thus may interfere with the detection of a change in protein fluorescence although similar effects might have been expected to occur with the phenylalanine of the ␤C170F mutant.

DISCUSSION
The use of rapid chemical quench-flow methods to provide a direct kinetic analysis of mutant forms of tryptophan synthase has allowed comparison with wild type and has given us an opportunity to test our current model of channeling and the kinetic consequences of the channel mutations. Several surprising results were obtained related to substrate channeling and protein communication.
According to our model for the kinetics of efficient channeling of indole, we have established as predicted that indole can be observed in a single enzyme turnover of the ␣␤ reaction if the 2 E. Gerhardt, K. S. Anderson, and I. Schlichting, unpublished observations. rate of passage of indole and/or the rate of chemistry at the ␤ subunit are decreased (Schemes II and III and Table II). The phenylalanine mutation slows the rate of channeling approximately 10-fold compared with wild type, whereas the tryptophan mutation reduces the rate Ͼ1000-fold. Although this difference in the two mutations may be related to the size of the obstructing residue, it could also be related indirectly to the lack of activation of the ␣ reaction (Scheme III and Table II) by the aminoacrylate in the ␤C170W mutation. We speculate that the activation of IGP cleavage at the ␣ site is prevented by the presence of the indole ring of ␤Trp-170 which may mimic the indole intermediate in the tunnel. These results are supported by our observations that a protein conformational change in intrinsic protein fluorescence is not detected with this mutant. In addition, preliminary structural analysis indicates that the ␤C170W mutation not only restricts the tunnel but also results in the movement of residues which may be important in transmitting the activation or trigger from the ␤ subunit to the ␣ subunit. 2 For instance, in the presence of 5-fluoroindole propanol phosphate (an inhibitor of the ␣ reaction) there is a 10-Å movement of the guanidinium group of ␤Arg-175 from the surface of the protein in the native wild type tryptophan synthase without ligand to the ␣␤ interface in the ␤C170W mutant structure. The combination of kinetic results presented here as well as more complete structural analysis of this mutant may provide important clues concerning ␣␤ intersubunit communication.
In summary, the kinetic analysis of the mutant enzymes presented in this report reaffirms the essential kinetic features required for efficient channeling. Slowing the rate of passage of indole and/or the rate of reaction of indole at the ␤ site (Anderson et al., 1991) leads to the detection of the indole in a single enzyme turnover of the physiological reaction. Even though the ␤C170F mutant intersubunit communication maintains cou-pling of the ␣ and ␤ reactions, the efficiency of channeling is decreased. Efforts to define the structural basis for this conformational coupling by x-ray crystallography, solid-state nmr, and site-directed mutagenesis are currently underway.