Transient Kinetic Studies on the Allosteric Transition of Phosphoglycerate Dehydrogenase*

Stopped flow spectrophotometry was used to investigate the kinetics of the transition of the phosphoglycerate dehy- drogenase (3-phosphoglycerate: NAD oxidoreductase, EC 1.1.1.95) reaction from the active to the inhibited rate upon the addition of the physiological inhibitor serine. The transition was characterized by a single first order rate constant fk,,,,,,,) which was independent of enzyme concentration. At PH 8.5, kc,,,,,, increased in a hyperbolic manner with serine concentration from 2 to 8 s-l. The increase ink,,,,,,, occurred at serine concentrations where the steady state inhibition was virtually complete. These results indicate that serine inhibition is an allosteric process involving a conformational change in the enzyme. A model is presented in which serine at low concentrations binds exclusively to the in- hibited state of the enzyme and shifts the equilibrium to-ward that state; at high serine concentrations, serine binds to the active state, facilitating its conversion to the inhibited state. An alternative model, which we favor, proposes two classes of inhibitor binding sites, The kinetics of the fluorescence quenching of enzyme-bound NADH by serine (Sugimoto, E., and Pizer, L. I. (1968) J. Biol. 2090-2098), inhibition

Stopped flow spectrophotometry was used to investigate the kinetics of the transition of the phosphoglycerate dehydrogenase (3-phosphoglycerate: NAD oxidoreductase, EC 1.1.1.95) reaction from the active to the inhibited rate upon the addition of the physiological inhibitor serine. The transition was characterized by a single first order rate constant fk,,,,,,,) which was independent of enzyme concentration. At PH 8.5, kc,,,,,, increased in a hyperbolic manner with serine concentration from 2 to 8 s-l. The increase ink,,,,,,, occurred at serine concentrations where the steady state inhibition was virtually complete. These results indicate that serine inhibition is an allosteric process involving a conformational change in the enzyme. A model is presented in which serine at low concentrations binds exclusively to the inhibited state of the enzyme and shifts the equilibrium toward that state; at high serine concentrations, serine binds to the active state, facilitating its conversion to the inhibited state. An alternative model, which we favor, proposes two classes of inhibitor binding sites, The kinetics of the fluorescence quenching of enzymebound NADH by serine (Sugimoto, E., and Pizer, L. I. (1968) J. Biol. Chem. 243, 2090-2098, measured by stopped flow fluorimetry, was also characterized by a single first order rate constant (k,,,,,,,,) which was independent of enzyme concentration.
At pH 8.5, k,,,,,, ,,,, ranged from 0.4 ss' at low serine concentrations to 1.1 s-' at high serine concentrations. These results indicate that the fluorescence quenching induced by serine is a manifestation of a structural change in the enzyme.
Enzyme and excess NADH were mixed with substrate and serine in the stopped flow instrument, and enzyme-bound NADH fluorescence was monitored by exciting through the protein at 285 nm. A rapid fluorescence quenching process, which occurred within the mixing time, was followed by a slower fluorescence enhancement process which terminated in a steady state level corresponding to the quenched fluorescence of the enzyme. NADH. serine complex. The rapid quenching was the result of substrate binding (Dubrow, R., and Pizer, L. I. (1977) J. Biol. Chem. 252, 1539-1551. The fluorescence enhancement was characterized by a single first order rate constant whose value for a given serine *  concentration corresponded with kobs,i. This data shows that the quenched state of the enzyme.NADH complex is the state which is directly responsible for the inhibition of enzyme activity. During catalysis the quenched state is achieved from a different initial conformation, and consequently at a different rate, than in the absence of substrate. k,bs.i and k~hs.f.,l. were also measured using glycine, another inhibitor.
The ultraviolet difference spectrum between enzyme and enzyme plus serine was determined and proposed to be the result of the same structural change which is responsible for the fluorescence quenching by serine.
Much effort has gone into obtaining an understanding at the molecular level of the interaction between small molecular weight effecters and enzymes.

MATERIALS AND METHODS
Reagents -Serine and glycine were obtained from Calbiochem (A grade). The glycine was recrystallized twice from hot water to ensure that it was free from traces of serine. The sources of other reagents are provided in the accompanying paper (15). Phosphoglycerate Dehydrogenase -The purification and storage of phosphoglycerate dehydrogenase are described in the accompanying paper (15). For use in an experiment, the purified enzyme was passed through a small Sephadex G-25 column equilibrated with the proper buffer. The enzyme concentration was estimated from the absorbance at 280 nm (8).
Kinetic Measurements -Rapid absorbance or fluorescence changes were measured using a Durrum-Gibson stopped-flow spectrophotometer equipped with a fluorescence attachment. Spectrophotometric measurements were made by following the percentage of transmission at 340 nm using a cuvette of Z-cm path length. Rate-The following terminology will be used. The The steady state level of inhibition of the reverse direction reaction (hydroxypyruvate-P reduction) as a function of serine concentration at the two pH values used in these studies (7.5 and 8.5) was determined (Fig. 1) FIG. 3. Dependence of kobs i on serine concentration. Experiments such as those described in Fig. 2B were performed. At pH 8.5, two different enzyme concentrations (0.5 PM and 1 PM) and three different NADH concentrations CO,25 PM; A, 50 PM; n , 100 jtM) were used in obtaining the data. (The initial experiments were done using 100 PM NADH. The NADH concentration was lowered for subsequent experiments because for a given change in concentration, the change in transmission was greater, the lower the initial concentration. Thus the quality of the experiments was improved.) At pH 7.5, 0.4 PM enzyme and 20 @M NADH were used. All other conditions were the same as those described in Fig. 28 except that the serine concentration was varied. k,,,,i (closed symbols) was evaluated from semilog plots such as those shown in Fig. 2B (inset). The open circles represent the final steady state level of inhibition achieved in these experiments. The steady state inhibited rates were obtained from the same traces as the kebSSi values, and compared to the rate of fully active enzyme, which was determined in a control experiment.
correlate with the kinetics of the transition of the enzyme reaction from the active to the inhibited rate. Sugimoto and Pizer (8) demonstrated in a static experiment that serine quenched the fluorescence of phosphoglycerate dehydrogenase-bound NADH and postulated that the fluorescence quenching was due to a conformational change in the enzyme. The experiments presented in this section were designed to test whether the kinetics of the fluorescence change were consistent with that hypothesis, and whether the kinetics correlated with the kinetics of the loss of enzyme activity. The experiments were performed in the stopped flow instrument using a fluorescence attachment. Phosphoglycerate dehydrogenase and NADH were mixed with serine and the fluorescence of enzyme-bound reduced pyridine nucleotide was observed. The quantity of NADH present after mixing was suffrcient to effectively saturate the coenzyme binding sites as determined by fluorescence titrations. was not being measured. As an additional control, enzyme, NADH, and 1 mM serine contained in one syringe were mixed with buffer. No reaction was seen, and the fluorescence level was the final quenched level observed in the quenching reaction.
The kinetics of fluorescence quenching under all conditions tested was a single first order process. The insets in Fig. 4 show first order plots of the fraction of quenchable fluorescence remaining uersus time derived from representative experiments.
Observed first order rate constants (k,,,,,.,,) were obtained from these plots. When the serine concentration was held constant, k,,,,,, did not vary significantly with enzyme concentration.
The dependence of kObs,,.* on serine concentration is shown in Figs. 5 and 6. The amplitude of the decrease in fluorescence as a function of serine concentration is also shown. At pH 7.5, varying the serine concentration over a 5000-fold range (2 PM to 10 mM) had a minimal effect on kobs,f.p, which increased from 0.55 to 0.87 sl. At pH 8.5, kobs,,,Q, (0.4 s-l) was independent of serine concentration between 4 and 20 pM. k,bS,,,q then

1532
Allosteric Transition of Phosphoglycerate Dehydrogenase the protein at 285 nm, and bound pyridine nucleotide fluorescence was observed through a 400 nm cut-off filter. Excitation was done through the protein, rather than directly, in order to reduce the background change in fluorescence due to oxidation of NADH in the reaction. (Serine quenches the fluorescence of the enzyme.NADH complex when it is excited through the protein in a similar manner to when it is excited directly.) The results show a rapid fluorescence quenching, that occurred within the mixed time (~4 ms), followed by a slower fluorescence enhancement process which terminated in a steady state level of fluorescence lower than the initial fluorescence of the phosphoglycerate dehydrogenase plus NADH (Fig. 7).
The top trace in Fig. 7 was a base-line control in which enzyme and NADH were mixed with buffer to determine the unperturbed fluorescence level of enzyme and NADH. The initial quenching was greater than 95% of the enzyme-bound reduced pyridine nucleotide fluorescence and is the result of hydroxypyruvate-P binding (discussed in accompanying paper (15)).
Two other controls were performed: 1. Serine was preincubated with enzyme. Phosphoglycerate dehydrogenase, NADH, and 1 mM serine were mixed with hydroxypyruvate-P.
No reaction was seen, and the fluorescence level was the level to which the fluorescence was enhanced in the actual experiment (when the loss of fluorescence in the actual experiment due to the oxidation of NADH in the reaction was corrected for (see "Appendix")).
The fluorescence level was also the same level as when enzyme, NADH, and 1 mM serine were mixed with buffer. Thus the fluorescence enhancement was a serine-dependent process, and the  quenched fluorescence of the enzyme. NADH . serine complex was not altered by hydroxypyruvate-P. The final level of fluorescence achieved by the enhancement process was the same as the quenched fluorescence level of the enzyme. NADH . serine complex.
2. The reaction was run without serine (Fig. &I). Enzyme and NADH were mixed with hydroxypyruvate-P.
There was a lag of about 50 ms followed by a zero order decrease in fluorescence. The lag was shown not to be due to an inadequate instrumental time constant, and has been shown to be due to a rapid isomerization of the enzyme during the first turnover (see accompanying paper (1.5)). (A lag in the fluorescence enhancement process can also be seen (Fig. 8B). Its implications will be discussed later.) The zero order decrease in fluorescence was due to the disappearance of NADH in the reaction. Free NADH does fluoresce somewhat when activated at 285 nm. This control again demonstrates that the fluorescence enhancement is a serine-dependent process. The results from these experiments could be analyzed on the basis of a single first order process (see "Appendix"). This is demonstrated by the linearity of the semilog plots of the fraction of the enhanceable fluorescence remaining uersus time (Fig. 7, inset). (The 50-ms lag, which will be explained later, was ignored in deriving these plots.  The kow /I values were evaluated from semilog plots such as those shown in the insets in Fig. 7. The k,,,,,, values were those from Fig. 3 Table I. The magnitudes of the two rate constants correlated well. The fluorescence enhancement experiment was also carried out by excitating at 340 nm. There was an initial rapid quenching of 85% of the enzyme-bound NADH fluorescence, which occurred within the mixing time. Because of the higher background due to the reaction, a net enhancement following the rapid quenching could only be observed at the higher serine concentrations, where the reaction was being inhibited at a faster rate. (At lower serine concentrations Fi > F,,,,,, whereas at higher serine concentrations F,.,,, > F, (see "Appendix").) First order rate constants for the enhancement process, determined at 0.1 and 1 mM serine, were found to be the same as when excitation was at 285 nm.
The results presented in this section resolve the discrepancy between the magnitudes of hoha,, and k,,,.,,,,. A serine-dependent fluorescence enhancement of enzyme-bound NADH, whose rate correlated with the rate of decay of enzyme activity, was observed during the pre-steady state inhibition of the reaction. Fluorescence quenching by serine of the binary enzyme. pyridine nucleotide complex and fluorescence enhancement by serine of the enzyme-bound reduced pyridine nucleotide during turnover resulted in the same final equilibrium level of fluorescence. These results are consistent with the quenched state of the binary complex being the state which is directly responsible for the loss of enzyme activity. However, during turnover it is achieved via a different pathway (from a different initial conformation) than in the absence of hydroxpyruvate-P.
Kinetics ofPre-steady State Inhibition Process and Fluorescence Quenching with Glycine -Phosphoglycerate dehydrogenase activity is inhibited by glycine (7). The steady state level of inhibition of hydroxypyruvate-P reduction was determined as a function of glycine concentration and a sigmoid curve was obtained. The Hill numbers were 2.38 and 2.34 and the I,, values were 0.73 and 0.85 mM at pH 7.5 and 8.5, respectively. Phosphoglycerate dehydrogenase was about 170 times less sensitive to glycine inhibition than to serine inhibition. Glycine quenched enzyme-bound NADH fluorescence to the same extent as did serine. It is not known whether glycine and serine bind at the same or different sites on the enzyme.
In order to study how a change in ligand can influence the rate of an allosteric process, the pre-steady state inhibition and fluorescence quenching kinetics were studied with glycine in place of serine. The results of the glycine concentration dependence of kabs,, and kuhs,,,,, are shown in Figs. 9 and 10. At pH 7.5, the pre-steady state inhibition kinetics were identical for serine and glycine. At pH 8.5, Kobs,, (9 ~~'1 did not vary between glycine concentrations of 2 and 200 mM. This concentration range was equivalent to a serine concentration Experiments such as those described m Fig. 28  The closed circles represent kabs,, as evaluated from semilog plots such as those shown in Fig. 2B, inset; 0, the final steady state inhibited level achieved in these experiments. The steady state inhibited rates were obtained from the same traces as the ha,, values and compared to the rate of fully active enzyme, which was determined in a control experiment, range between 0.01 and 1 mM in terms of the level of final steady state inhibition. The invariant k,,,,, for glycine of 9 s' at pH 8.5 did not differ significantly from the 8 s-l upper limit for kobs,, for serine. At pH 7.5, between 0.75 and 200 mM glycine, kobs.l,q. increased from 1 to 1.4 ss'. These values were about 30 to 50% higher than the k,,,,,, values measured for the corresponding (in terms of degree of fluorescence quenching) serine concentrations. This difference was obtained using the same enzyme preparation on the same day. At PH 8.5, koss.f..s. was invariant at 1.1 s-' between 0.75 and 2 mM glycine, and then increased between 2 and 200 mM glycine to 2 s'. This pattern was similar to the pattern seen with serine, except that the amount of the increase was lower with glycine. The kobs,,.,,. values for the lower glycine concentra-  Photograph of the original chart trace of the ultravlolet difference spectrum between the enzyme. reduced coenzyme complex and the enzyme 'reduced coenzyme complex plus serine at pH 7.5. The spectrum was determined on a Cary 15 spectrophotometer. The enzyme concentration was 10 ELM and the NADH concentration was 5 PM. Matched quartz cuvettes containing 1 ml of identical enzyme solutions were placed in the sample and reference compartments, and a base-line was recorded.
Then 10 ~1 of 10 rnM serine were added to the sample cuvette, 10 ~1 of distilled HZ0 were added to the reference cuvette, and the difference spectrum was recorded.  (Fig.  11, whereas the serine concentration which resulted in a halfmaximal fluorescence quenching of the enzyme. NADH complex by serine was 9.5 @M (Fig. 6). (This titration was repeated in a static equilibrium experiment and a value of 8.8 PM was obtained.') At pH 8.5, the serine concentration required for the half-maximal increase in kohs.i was 55 pM (Fig. 31, whereas the serine concentration required for the half-maximal increase in k,bs,I.r,., was 100 FM (Fig. 6). The major form of the enzyme during the steady state reaction is the ternary complex phosphoglycerate dehydrogenase hydroxypyruvate-P'NADH (15). These observations show that hydroxypyru- Arguments analogous to those made for serine indicate that the inhibition of phosphoglycerate dehydrogenase by glycine and the fluorescence quenching of the enzyme.NADH complex by glycine were also the result of a conformational change in the enzyme.
The ultraviolet difference spectrum between the enzyme 'NADH complex and the enzyme. NADH complex plus serine (Fig. 11) shows that the interaction between serine and the enzyme. NADH complex resulted in a change in the environment of some of the aromatic amino acid residues of the enzyme. The change could be due to a modification of the environment of the serine-binding site upon the binding of serine to the enzyme or to a conformational change induced by serine. Preliminary stopped flow experiments have indicated that the rate of increase in absorbance at 287 nm when serine was mixed with the enzyme. NADH complex was the same as the rate of fluorescence quenching of the enzyme. NADH complex by serine.' Thus the increase in absorbance at 287 nm was probably the result of the same conformational change which resulted in the fluorescence quenching.
Pathway of Allosteric Transition - Koshland et al. (3) have pointed out that there are two pathways by which an allosteric transition may occur. One pathway requires that a spontaneous equilibrium exist between two conformational states of the protein (or monomers). An effector molecule does not directly induce a conformational transition in the protein molecule to which it binds, but preferential binding of the effector to one of the conformational states displaces the conformational equilibrium in the direction of that state (equilibrium displacement transition). If symmetry is maintained among subunits, this pathway corresponds to the model presented by Monod,Wyman,and Changeux (2). Alternatively, the binding of an effector molecule to a protein molecule can directly induce a conformational transition in that protein molecule (ligand-induced transition). The sequential-type allosteric model (3) may proceed via either pathway.
Mechanism 1 (Fig. 12) presents an equilibrium displacement model for the allosteric transition of phosphoglycerate dehydrogenase which is consistent with the serine concentration dependence of kubs,, at pH 8.5. The R and T forms are in a pre-equilibrium, and serine binds preferentially to T (Kser,T K K ser,s). There are two pathways by which T-Ser can be reached when serine is mixed with R. At lower serine concentrations, serine binds exclusively to T and displaces the R-T equilibrium, causing an allosteric transition from R to T (pathway 1). At higher serine concentrations serine binds to R and destabilizes it, so that the R-Ser to T-Ser transition occurs at a faster rate than the R to T transition. Thus at lower serine concentrations the allosteric transition takes place exclusively via pathway 1. As the serine concentration increases and binding to R begins to occur, there is a competition between Pathway 1 is the predominant pathway at low serine concentrations, and pathway 2 is the predominant pathway at high serine concentrations. Further details of this mechanism are presented in the text.
the two pathways, and the predominant pathway is converted to pathway 2 as R becomes saturated with serine.
The basis for the model is the difference between I,,, for the steady state inhibition by serine (5.1 pM) and the serine concentration which resulted in a half-maximal increase in kOb4., (55 PM). The increase in kobS,, began to occur at 20 FM serine, a concentration at which the steady state level of inhibition was already 94% (Fig. 3). Thus a process was beginning to occur at a serine concentration which was already capable of bringing about the allosteric transition. That process, according to Mechanism 1, was the binding of serine to R.
This model is essentially equivalent to the Monod, Wyman, Changeux model with nonexclusive effector binding.
A ligand-induced mechanism is also consistent with the serine data (Mechanism 2). One can postulate the existence of two classes of serine binding sites, a tight (class A) and a loose (class B) class. At low serine concentrations, serine binds to the tight sites only and induces the allosteric transition. As the serine concentration is raised, binding to the loose sites begins to occur, as well. Enzyme molecules which have serine bound to both the tight and loose sites isomerize at a faster rate than those which just have the tight sites occupied. There is no teleological reason for there being a class of loose serine binding sites, since binding to the tight sites would be sufficient to cause virtually complete inhibition of the enzyme. It is possible, that the class B sites are the "glycine" binding sites, and at present there is no evidence as to whether serine and glycine bind to the enzyme at the same or different sites. Another possibility is negative cooperativity in serine binding. The steady state velocity of the catalytic reaction in the direction of phosphoglycerate oxidation as a function of 3-phosphoglycerate concentration yielded a biphasic double reciprocal plot (10, 151, suggesting negative cooperativity in 3-phosphoglycerate binding. In addition, the binding of NADH to the enzyme exhibits negative cooperative behavior (8).
The pH 7.5 data are consistent with Mechanism 1, with the following alternative possibilities: the R-Ser to T-Ser isomerization occurs at about the same rate as the R to T isomerization; or serine does not bind to R at all. The data are also consistent with Mechanism 2 in which binding of serine to the loose sites does not affect the rate of the allosteric transition, or in which serine does not bind to the loose sites at this pH.
The qualitative patterns of the serine concentration dependence ofkobs,f.a. were the same as those of kobs,i. Using the same reasoning applied above, it is apparent that both Mechanisms 1 and 2 can explain the allosteric transition of the enzyme NADH complex.
The glycine concentration dependence of k,b,,j and koba,l,s,