Advertisement

A Stopped Flow Transient Kinetic Analysis of Substrate Binding and Catalysis in Escherichia coli d-3-Phosphoglycerate Dehydrogenase*

  • Rodney L. Burton
    Affiliations
    Department of Developmental Biology, Washington University School of Medicine, St. Louis, Missouri 63110
    Search for articles by this author
  • Jeremiah W. Hanes
    Affiliations
    Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853
    Search for articles by this author
  • Gregory A. Grant
    Correspondence
    To whom correspondence should be addressed: Dept. of Developmental Biology, Box 8103, Washington University School of Medicine, 660 S. Euclid Ave., St. Louis, MO 63110. Tel.: 314-362-3367; Fax: 314-362-4698
    Affiliations
    Department of Developmental Biology, Washington University School of Medicine, St. Louis, Missouri 63110

    Department of Medicine, Washington University School of Medicine, St. Louis, Missouri 63110
    Search for articles by this author
  • Author Footnotes
    * This work was supported, in whole or in part, by National Institutes of Health Grant GM 56676 (to G. A. G.). 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.
    The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. 1–4.
Open AccessPublished:September 06, 2008DOI:https://doi.org/10.1074/jbc.M805180200
      Pre-steady state, stopped flow analysis of Escherichia coli d-3-phosphoglycerate dehydrogenase was performed by following the fluorescence of protein tryptophan and the fluorescence resonance energy transfer from protein tryptophan to bound NADH. The results indicate that binding of substrates is ordered, with coenzyme, NADH, binding first. Furthermore, the analysis indicated that there are two sets of sites on the tetrameric enzyme that can be differentiated by their kinetic behavior. NADH binding was consistent with an initial binding event followed by a slow conformational change for each site. The slow conformational change is responsible for the apparent tight binding of NADH to the apoenzyme but is too slow to participate in the catalytic cycle when the enzyme is rapidly turning over. Subsequent binding of the substrate, α-ketoglutarate, was characterized by a rapid equilibrium binding event followed by a conformational change for each site. Catalysis in the direction of NAD+ reduction showed a distinct burst of activity followed by a slow rate of turnover, indicating that the rate-limiting step is after hydride transfer. Catalysis in the direction of NADH oxidation did not display burst kinetics, indicating that the rate-limiting step is at or before the hydride transfer step. The burst data indicated that the rate of NAD+ reduction (3.8 s–1) is similar to the kcat of the enzyme (2–3 s–1) in that direction. However, analysis of the reaction with deuterated NADH failed to show an effect on the velocity of the reaction with a VH/VD = 1.07 ± 0.06. None of the other rates determined by stopped flow analysis could account for the kcat of the enzyme in either direction (forward kcat = 0.01 s–1, reverse kcat = 2–3 s–1), suggesting that the rate-limiting step in both directions is a conformational change in the enzyme that is not detected optically.
      d-3-Phosphoglycerate dehydrogenase (PGDH)
      The abbreviations used are: PGDH, d-3-phosphoglycerate dehydrogenase; FRET, fluorescence resonance energy transfer.
      2The abbreviations used are: PGDH, d-3-phosphoglycerate dehydrogenase; FRET, fluorescence resonance energy transfer.
      (EC 1.1.1.95) catalyzes the NADH/NAD+-dependent interconversion of phosphoglyceric acid and hydroxypyruvic acid phosphate (
      • Walsh D.A.
      • Sallach H.J.
      ). The enzyme is a tetramer of identical subunits, each consisting of three structurally distinct domains, the nucleotide binding domain, the substrate binding domain, and the regulatory domain (
      • Schuller D.J.
      • Grant G.A.
      • Banaszak L.J.
      ). The binding of l-serine, the end product of the pathway, to the regulatory domain produces inhibition of the catalytic activity of the enzyme. The regulatory domain consists of a βαββαβ motif characteristic of the ACT domain family, a structural motif that binds small molecules and is found in enzymes and transcriptional regulators mostly involved in amino acid metabolism (
      • Aravind L.
      • Koonin E.V.
      ,
      • Grant G.A.
      ). The word “ACT” derives from the first letter of the names of three proteins originally proposed to contain this motif, aspartate kinase, chorismate mutase, and TyrA (
      • Aravind L.
      • Koonin E.V.
      ).
      The subunit contacts in the PGDH tetramer occur between two sets of nucleotide binding domains and two sets of regulatory, or ACT, domains such that the tetramer appears to be an elongated doughnut or toroid (
      • Schuller D.J.
      • Grant G.A.
      • Banaszak L.J.
      ,
      • Bell J.K.
      • Grant G.A.
      • Banaszak L.J.
      ) with the regulatory domains at the ends of the long axis of the tetramer. Serine binds at the interface between the regulatory subunits with two serine-binding sites at each of two regulatory subunit interfaces. Recent work with hybrid tetramers of Escherichia coli PGDH (
      • Grant G.A.
      • Xu X.L.
      • Hu Z.
      ) has shown that the enzyme exhibits half-of-the-sites activity with reference to the catalytic sites but also that only two of the serine binding sites need be occupied, one at each regulatory subunit interface, for full inhibition of catalysis to occur. Thus, the enzyme appears structurally and functionally as a “dimer of dimers.”
      Previous studies provided evidence for at least two NADH-bound forms of the enzyme that bind l-serine in the physiological range (
      • Grant G.A.
      • Hu Z.
      • Xu X.L.
      ). These studies also demonstrated that NADH binding had the effect of decreasing the degree of positive and negative cooperativity observed for serine binding as compared with the enzyme in the absence of NADH. The intrinsic dissociation constants for serine binding determined by equilibrium dialysis in the presence of saturating levels of NADH are 5, 11, 3, and 36 μm (
      • Grant G.A.
      • Hu Z.
      • Xu X.L.
      ).
      In the direction of l-serine biosynthesis (forward reaction), PGDH converts phosphoglyceric acid and NAD+ to hydroxypyruvic acid phosphate and NADH. However, in vitro, the equilibrium of the reaction lies far in the opposite direction, so that the enzyme is usually assayed by following the conversion of NADH to NAD+ (reverse reaction) (
      • Pizer L.I.
      • Sugimoto E.
      ). The equilibrium constant for the forward reaction, under physiological conditions, has been reported to be 10–11 to 10–13 m (
      • Sugimoto E.
      • Pizer L.I.
      ,
      • Merrill D.K.
      • McAlexander J.C.
      • Guynn R.W.
      ). It was originally reported (
      • Sugimoto E.
      • Pizer L.I.
      ) that purified E. coli PGDH retained between 2 and 3 NADH molecules bound per tetramer that could not easily be removed by dialysis. This observation implies very tight binding of NADH, and it was estimated that the dissociation constant for NADH is ∼50 nm (
      • Sugimoto E.
      • Pizer L.I.
      ). Because of this, early transient kinetic analysis of PGDH (
      • Dubrow R.
      • Pizer L.I.
      ) utilized enzyme preparations with either partially or fully bound NADH and did not address the binding of NADH itself. To better understand how the activity of the enzyme is regulated, it is important to understand the kinetics of substrate binding and catalysis at a more detailed level than is provided by typical steady state kinetic analysis. This is particularly true when determining the effects of mutations, where a greater understanding of the consequence of these mutations on the individual rate constants of the reaction pathway is important. Furthermore, steady state kinetic analysis is limited by the very low Km for NADH, which makes it difficult to do steady state analysis with varying NADH concentrations when the signal that is followed is due to the concentration of NADH (absorbance at 340 nm). There are only two reports in the literature from 1977 (
      • Dubrow R.
      • Pizer L.I.
      ,
      • Dubrow R.
      • Pizer L.I.
      ) dealing with a pre-steady state analysis of E. coli PGDH. Although these studies provide a good background, they do not provide sufficient detail of many aspects of the catalytic pathway. The investigations described here are designed to provide the additional necessary information to better understand the catalytic process of this enzyme.

      MATERIALS AND METHODS

      E. coli PGDH was expressed and isolated using affinity chromatography on 5′-AMP-Sepharose, as previously described (
      • Schuller D.J.
      • Fetter C.J.
      • Banaszak L.J.
      • Grant G.A.
      ,
      • Al-Rabiee R.
      • Lee E.J.
      • Grant G.A.
      ). Steady state activity was measured by following the conversion between NADH and NAD+ by monitoring the change in absorbance at 340 nm in the presence of enzyme and α-ketoglutarate. α-Ketoglutarate is a substrate analog for E. coli PGDH (
      • Zhao G.
      • Winkler M.E.
      ) and was utilized in these experiments because it is more easily obtained and more stable than the physiologic substrate, hydroxypyruvic acid phosphate. Enzyme concentration was determined using an E1% of 6.7 (
      • Al-Rabiee R.
      • Lee E.J.
      • Grant G.A.
      ), and the molar extinction coefficient of NADH is 6.22 × 103 m–1 cm–1.
      Removal of NADH from PGDH—It was originally reported that E. coli PGDH was isolated with NADH irreversibly bound to it (
      • Sugimoto E.
      • Pizer L.I.
      ). This observation led to earlier transient kinetic work being done in the presence of prebound NADH (
      • Dubrow R.
      • Pizer L.I.
      ,
      • Dubrow R.
      • Pizer L.I.
      ). E. coli PGDH was isolated by affinity chromatography on 5′-AMP-agarose by elution with NADH. After elution, the protein was applied to a 30 × 2.6-cm column of Sephacryl S-200 in 20 mm potassium phosphate buffer, pH 7.5. The absorbance at 340 nm associated with the protein peak shown in supplemental Fig. 1 indicates that NADH remained bound and eluted with the protein, as has been previously reported (
      • Sugimoto E.
      • Pizer L.I.
      ). Since the Kd for NAD+ is at least 100-fold greater than that for NADH (
      • Sugimoto E.
      • Pizer L.I.
      ), NADH was removed by adding substrate to convert it to NAD+ prior to the Sephacryl S-200 column, which then separates oxidized coenzyme, excess substrate, and product from the PGDH.
      Production of Deuterated NADH (NADD)—A-side NADD was produced by the procedure of Viola et al. (
      • Viola R.E.
      • Cook P.F.
      • Cleland W.W.
      ) using perdeuteroethanal (ethanol-d6) from Cambridge Isotope Laboratories. Yeast alcohol dehydrogenase and aldehyde dehydrogenase were obtained from Sigma. The production of NADD from NAD+ was monitored by absorbance at 340 nm, and the product was purified by ion exchange chromatography as described, just prior to use. Concentrations of NADH and NADD were determined by absorbance at 340 nm, using an extinction coefficient of 6.22 × 103 m–1 cm–1.
      Stopped Flow Fluorescence Spectroscopy—Pre-steady state kinetic analyses were performed using an Applied Photophysics model SX-20 stopped flow spectrometer. NADH binding was monitored by fluorescence resonance energy transfer (FRET) between protein tryptophan and bound NADH by exciting the single tryptophan residue in the enzyme at 295 nm and observing the changing emission from tryptophan at 340 nm and NADH at 420 nm, using 340- and 420-nm band pass emission filters, respectively. α-Ketoglutarate binding also produces a change in the 340- and 420-nm FRET but at lower amplitudes than that for NADH and in the opposite direction. The emission at 340 nm was used to monitor α-ketoglutarate binding, since this leads to a larger signal to noise ratio. Enzyme turnover was directly monitored by following the absorbance of NADH at 340 nm. A total of 10,000 points were collected in each trace, and at least five individual traces were averaged at each set of conditions. For long duration traces, a split time mode was utilized, and 10,000 points were collected in each portion. The reaction and all reagents were thermostated at 25 °C with a circulating water bath.
      The binding transients were analyzed using the Pro-Data Viewer fitting software provided by the manufacturer (Applied Photophysics) or with Kaleidograph (Synergy Corp.). The kinetic schemes and equations used to evaluate the data are presented in the supplemental material and in Refs.
      • Johnson K.A.
      and
      • Johnson K.A.
      .
      Kinetic Simulations—Kinetic simulations were performed with Global Kinetic Explorer Professional from KinTek Corp. The program provides dynamic visualization and global fitting of multiple data sets to a single model. Multiple data sets were fit globally to an appropriate model using initial estimates of rates and amplitudes obtained from the fit of the original data. The continuous simulation feature of the software was employed to judge the degree to which individual rate constants are constrained by the data.

      RESULTS

      Pre-steady State Kinetic Analysis of NADH Binding—Each subunit of the PGDH tetramer has a single NADH binding site. NADH binding is monitored by following the time-dependent increase in the FRET signal at 420 nm when the protein is excited at 295 nm. The resulting transients (Fig. 1) showed a fast initial rise in signal (Fig. 1A) followed by a slow, low amplitude rise (Fig. 1B). At all concentrations of NADH examined, the full time course fit best to four exponentials (supplemental Equation 1). Fig. 2 shows the results of these fits plotted as the observed rate constant (kobs) for each exponential versus the NADH concentration. The kobs values for two of the exponentials increase with NADH concentration, fit well to a straight line, and show no indication of becoming saturated. The kobs for the other two processes are relatively very small and show no apparent change in rate, suggesting that they represent unimolecular processes. The two concentration-dependent processes cannot represent successive steps, since they both fit to a straight line, and the rate at high NADH concentration of the slower kobs exceeds the y intercept of the faster kobs. Based on earlier work that demonstrated at least two conformational forms of enzyme with bound NADH (
      • Grant G.A.
      • Hu Z.
      • Xu X.L.
      ), the most likely interpretation of this data is that they represent two separate sets of sites with two successive steps, each consisting of initial binding followed by a conformational change. This model is later confirmed below by the observation that there were also two observed rates for substrate binding and catalysis (see Fig. 7). Furthermore, all attempts, by simulation, to fit the two slowest kobs as additional binding steps were unsuccessful, since the simulations would consistently plateau to a constant value very early rather than continuing to rise as in Fig. 1B. Global simulation of the two-step model for two sets of sites using the amplitudes and estimated rate constants derived from the initial data showed that a slightly better fit was produced when the conformational change with the largest kobs was coupled to the more rapid binding event. The rate constants, derived from the data and refined by the simulation, are presented in Table 1 and in summary form in Fig. 7. The normalized amplitudes are plotted in supplemental Fig. 2 and demonstrate that the fastest concentration dependent process dominated the binding.
      Figure thumbnail gr1
      FIGURE 1Time course of NADH binding to E. coli PGDH. The transients for NADH binding are shown for two time ranges, 0–1 s (A) and 0–50 s (B). Enzyme (2.4 μm subunit) was mixed with varying concentrations of NADH, and the FRET signal was monitored at 420 nm after excitation at 295 nm. The transients are shown for 20 (•), 30 (▪), 35 (▴), 40 (♦), 45 (○), and 50 μm NADH (□). Every third data point is plotted for the 0–1 s range and every 100th data point is plotted for the 0–50 s range. The solid lines are fits of the data to for four exponentials.
      Figure thumbnail gr2
      FIGURE 2NADH binding data. The kobs values derived from the pre-steady state NADH binding transients are plotted versus the NADH concentration. The solid lines are a fit of the data to the equation for a straight line. The data are designated kobs 1 (•), kobs 2 (▪), kobs 3 (○), and kobs 4 (▴).
      Figure thumbnail gr7
      FIGURE 7Summary of the pre-steady state kinetic data for E. coli PGDH. Based on the kinetic data from stopped flow analysis, two sets of sites in the tetramer appear to be binding substrate and coenzyme PGDH preincubated with NADH undergoes a slow conformational change that also binds substrate and turns over to product. Because of the rate of the conformational change resulting from NADH binding, this form of the enzyme only turns over once, initially. These are designated paths A and D. After this initial turnover, only paths B and C are functional during continuous turnover of substrate to product. Moreover, path B represents the major form of the enzyme during turnover. The estimated kinetic constants are shown for each step. The values designated as kobs in parentheses are from single turnover experiments. Arrows are shown for the direction of the reaction designated “forward” and “reverse,” and the nomenclature for the kinetic constants listed in is shown in the center of the figure.
      TABLE 1Estimates of kinetic constants
      Path APath BPath CPath D
      k12.03 ± 0.03 μm -1 s-10.24 ± 0.01 μm-1 s-1
      k-11.11 ± 0.01 s-11.19 ± 0.04 s-1
      k10.48 ± 0.22 s-1 (0.4 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      0.022 ± 0.007 s-1 (0.02 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      k-10.05 ± 0.10 s-1 (≤0.01 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      ,
      The data do not sufficiently constrain the fitting. The values specify a limit over which the fitting does not significantly change during simulation.
      0.012 ± 0.008 s-1 (≤0.01 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      ,
      The data do not sufficiently constrain the fitting. The values specify a limit over which the fitting does not significantly change during simulation.
      Kd239 ± 5 μm114 ± 42 μm (75)
      Values are estimates from simulation of the data holding the other values constant.
      185 ± 99 μm (200)
      Values are estimates from simulation of the data holding the other values constant.
      13 ± 4 μm
      k387 ± 2 s-15.4 ± .8 s-12.3 ± .6 s-17 ± 1 s-1
      k-31.5 ± 1.9 s-1 (1.6 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      6.8 ± 0.2 s-10.05 ± 0.08 s-1 (≤0.1 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      ,
      The data do not sufficiently constrain the fitting. The values specify a limit over which the fitting does not significantly change during simulation.
      1.1 ± 0.8 s-1 (≤1 s-1)
      Values are estimates from simulation of the data holding the other values constant.
      ,
      The data do not sufficiently constrain the fitting. The values specify a limit over which the fitting does not significantly change during simulation.
      k43.8 ± 0.1 s-10.90 ± 0.05 s-1
      k-42.1 ± 0.04 s-10.35 ± 0.04 s-1
      k4 + k-48.9 ± 0.5 s-14.9 ± 0.3 s-11.0 ± 0.1 s-12.2 ± 0.3 s-1
      a Values are estimates from simulation of the data holding the other values constant.
      b The data do not sufficiently constrain the fitting. The values specify a limit over which the fitting does not significantly change during simulation.
      Pre-steady State Kinetic Analysis of α-Ketoglutarate Binding—When α-ketoglutarate binding was investigated in the absence of NADH, there was no fluorescence or absorbance signal that could be detected. Furthermore, isothermal titration calorimetry also failed to produce a signal for α-ketoglutarate binding in the absence of NADH. On the other hand, when α-ketoglutarate binding was investigated in the presence of saturating levels of NADH, a time-dependent decrease in the 420 nm signal was observed as well as a complementary increase in fluorescence at 340 nm due to the enhancement of tryptophan fluorescence. When the enzyme was first equilibrated with α-ketoglutarate and then NADH was added, a rapid increase in the 420 nm signal due to NADH binding was observed, along with a lower amplitude concentration-dependent decrease in the FRET signal due to the α-ketoglutarate binding. Since preincubation with α-ketoglutarate did not eliminate the signal corresponding to its binding, this indicates that binding is ordered and NADH binds first.
      Since NADH binding was shown to involve a relatively slow conformational change following binding of the coenzyme, α-ketoglutarate binding was investigated with and without preincubation of the enzyme with NADH. In the first case, enzyme preincubated with NADH was rapidly mixed with α-ketoglutarate. In the second case, NADH-free enzyme was rapidly mixed with NADH and α-ketoglutarate. Analysis was carried out using the FRET signal at 340 nm, since it was more intense.
      The transients for α-ketoglutarate binding with preincubated NADH fit best to a double exponential (supplemental Fig. 3). When kobs is plotted versus α-ketoglutarate concentration, both exponentials produce nonlinear plots (Fig. 3). It was not possible to go to higher concentrations of α-ketoglutarate, because an increasingly larger portion of the response took place in the dead time of the instrument, and at high concentrations, α-ketoglutarate has an inner filtering effect on the 340 nm fluorescence. Lower concentrations of α-ketoglutarate were also problematic, because the signal to noise ratio became too low to analyze accurately. The kobs plots were fit to supplemental Equation 4 for a rapid equilibrium binding step, followed by a conformational change. These fits produced Kd values of ∼40 and 13 μm. The forward rates of the conformational change (k3) are ∼87 and 7 s–1. The reverse rates for the conformational change (k–3) are estimated at ∼2 and 1 s–1, respectively. However, the k–3 values are subject to a greater degree of error. Simulation of the data gave upper limits of 1.6 and 1 s–1.
      Figure thumbnail gr3
      FIGURE 3Plot of the data derived from the transients of α-ketoglutarate Binding to E. coli PGDH with prebound NADH. The kobs values derived from the pre-steady state α-ketoglutarate binding transients are plotted versus the α-ketoglutarate concentration. The solid lines are a fit of the data to . Enzyme (2 μm subunit) was preincubated for 30 min with saturating levels of NADH and then rapidly mixed with α-ketoglutarate. The protein fluorescence at 340 nm was measured with an excitation at 295 nm.
      In the case where NADH was not allowed to preincubate with the enzyme, the binding transients (supplemental Fig. 4) also fit to two exponentials, and the kobs plots (Fig. 4) also produced nonlinear curves. This again suggested a rapid equilibrium binding followed by a conformational change. These plots yield Kd values of ∼114 and 185 μm, k3 values of ∼5 and 2 s–1, and k–3 values of ∼7 and 0.05 s–1. Simulation produced a value of ∼0.1 s–1 for the second k–3 value. This should also be considered an upper limit, because the data do not sufficiently constrain the simulation. Since binding is ordered, and a significant portion of the NADH binding takes place in the dead time of the instrument, values for the simultaneous NADH binding were treated as constants, and controls of NADH binding were subtracted before fitting for the α-ketoglutarate binding.
      Figure thumbnail gr4
      FIGURE 4Plot of the data derived from the transients of α-ketoglutarate binding to E. coli PGDH without prebound NADH. The kobs derived from the pre-steady state α-ketoglutarate binding transients are plotted versus the α-ketoglutarate concentration. The solid lines are a fit of the data to . Enzyme (2 μm subunit) was rapidly mixed with NADH and α-ketoglutarate. The protein fluorescence at 340 nm was measured with an excitation at 295 nm.
      The two exponentials produced from each of the two α-ketoglutarate binding experiments (with and without previously bound NADH) further validate the “dimer of dimers” conformation of PGDH with two sets of active sites suggested by previous studies (
      • Grant G.A.
      • Xu X.L.
      • Hu Z.
      ,
      • Grant G.A.
      • Hu Z.
      • Xu X.L.
      ). Computer simulation of these data as a rapid equilibrium binding step followed by a conformational change indicates that if the binding event itself contributes to the observed 340 nm FRET signal, there would be a shift from the base line in the signal within the dead time of the instrument before the slower transient is observed. This would manifest itself as a deviation of the base line determined by the sum of the signal from the individual starting components prior to mixing. This is not observed in these experiments, suggesting that the binding of α-ketoglutarate itself does not contribute to the signal and that the transient is solely due to the subsequent conformational change that results following binding.
      Turnover at Saturating Levels of Substrate and Coenzyme—The turnover of NADH at the active site was monitored by absorbance at 340 nm. The reverse direction reaction was performed by preincubating the enzyme with saturating levels of NADH and then rapidly mixing with saturating levels of α-ketoglutarate. When this was done, the absorbance decreased with time and the response was linear along the whole time course. No burst of turnover was observed prior to the steady state turnover. This observation indicates that the rate-limiting step in the direction of NADH oxidation (reverse reaction) must be at or before the chemical turnover step. When the reaction was monitored in the forward direction by preincubating the enzyme with saturating levels of NAD+ and then rapidly mixing with saturated levels of α-hydroxyglutarate, a very well defined burst was observed, which allowed the determination of the rates of chemical turnover (Fig. 5). Furthermore, the burst transient again fit best to two exponentials, and the amplitude of the pre-steady state burst corresponds to ∼0.36 and 0.28 per enzyme site. The rate constants derived from this analysis are presented in Table 1 (k4 and k–4) and Fig. 7. The linear portion of the reaction following the burst yields a kcat for the forward reaction of ∼0.01 s–1.
      Figure thumbnail gr5
      FIGURE 5Kinetics of the pre-steady state burst in the forward reaction direction. Enzyme (4 μm subunit) was rapidly mixed with saturating levels of NAD+ and α-hydroxyglutarate, and the absorbance change at 340 nm was monitored. The solid line is the fit to the data with for two exponentials followed by a steady state rate. Only every 50th data point is shown for clarity.
      Single Turnover Experiments—Single turnover experiments were performed where the enzyme subunit concentration was 20–30 times that of NADH. The experiment was performed with and without preincubation with NADH. For the former, NADH-free enzyme was incubated with NADH for 30 min and then rapidly mixed with saturating levels of α-ketoglutarate. For the latter, NADH-free enzyme was rapidly mixed with saturating levels of α-ketoglutarate and subsaturating levels of NADH. The turnover of NADH was monitored by following the fluorescence at 420 nm with excitation at 340 nm. The observed rates did not vary with increasing enzyme concentration. In both cases, with or without preincubation with NADH, the observed transients fit best to two exponentials with amplitudes of ∼70 and 30% of the total. In the case of preincubation of the enzyme with NADH, the observed rates were 8.9 ± 0.5 and 2.2 ± 0.3 s–1, respectively. In the absence of preincubation with NADH, the observed rates were 4.9 ± 0.3 and 1.0 ± 0.1 s–1, respectively (Table 1). The rates determined for single turnover when NADH was not preincubated with enzyme were approximately the same as the sum of the on- and off-rates (k4 + k–4) for turnover determined from the burst experiment in the forward direction. In the case where NADH was preincubated with enzyme, the observed rates were about twice that for the case without preincubation.
      Deuterium Isotope Effects—When the enzyme was assayed in the presence of saturating levels of substrate and saturating levels of deuterated NADH (NADD), there appeared to be no effect on V (Fig. 6). The ratio of VH/VD was 1.07 ± 0.06 at pH 7.5 and indicates that hydride transfer is not rate-limiting for the reverse reaction under these conditions. However, there was an apparent small effect on V/K. (V/K)H/(V/K)D was 1.5 ± 0.11 from the fit to the reciprocal plot and 1.52 ± 0.22 from the fit to the hyperbolic plot. Since the varied substrate was α-ketoglutarate, this suggests an apparent effect on a rate term after NADH binding and before NAD+ release. Simulation of the steady state rates as a function of α-ketoglutarate for the model in Scheme 1
      Ek2k1AEAk4k3BEABk6k5E ABk8k7EPQk10k9EPQk11EQk13E


      indicates that the only rate changes that would give a true change in V/K without a change in V are for k3 and k4 in the model, which define the binding of α-ketoglutarate. However, as is the case with double reciprocal plots from experimental data, one must be very careful when comparing closely related slopes or intercepts. Additional investigation will be necessary to explain this effect.
      Figure thumbnail gr6
      FIGURE 6The effect of NADD on the steady state rates of PGDH at varying α-ketoglutarate concentrations. The concentration of α-ketoglutarate is varied in the presence of equal and saturating concentrations of NADH (•) or NADD (▪). Top, α-ketoglutarate concentration is plotted versus the steady state velocity of the reaction. Bottom, the same data are plotted in the form of a double reciprocal plot.

      DISCUSSION

      The original work on the pre-steady state kinetics of E. coli PGDH (
      • Dubrow R.
      • Pizer L.I.
      ,
      • Dubrow R.
      • Pizer L.I.
      ) utilized enzyme preparations that contained various amounts of prebound NADH. This was due to the very tight binding of NADH as a result of a conformational change subsequent to binding that was unrecognized at the time. This investigation, which utilizes both NADH-free PGDH and PGDH with prebound NADH, as well as more modern and facile stopped flow technology, provides a more precise and detailed description of the catalytic pathway of E. coli PGDH than was previously available.
      Fig. 7 and Table 1 summarize the kinetic steps (and their associated rates) that can be detected by stopped flow analysis. This scheme results from the observation that the transient for NADH binding fits to four exponentials but that the transients for the subsequent steps each fit to two exponentials, with unique rates depending on whether NADH is prebound or not (E to E* transition). Thus, the data consistently indicated that two sets of sites are present on the enzyme tetramer. These are indicated by branches A + B for one set, and C + D for the other set. Furthermore, each set of sites has two branches representing whether or not NADH was prebound such that E* was formed as the predominate species. E* is formed by a slow conformational change and can undergo subsequent substrate binding and turnover with its own unique rates. These are represented by branches A and D. Branches B and C represent the two sets of sites during continuous turnover.
      When NADH binds to E. coli PGDH, there is a rapid increase in FRET signal that is essentially complete within 100 ms (The first step in branch B and C). This is followed by a more gradual rise in FRET that is still not complete after 1 min (E to E* transition). The entire time course fits best to four exponentials, with the first two exponentials (whose rates are concentration-dependent) essentially defining the initial rapid rise and the last two exponentials (which are concentration-independent) defining the subsequent gradual rise. Since the concentration-dependent steps cannot represent two successive steps, the simplest model is one where there are two distinct sites that bind NADH and that induce a conformational change after binding. Simulation of this model produces an excellent fit to the data and provides rate constants that are consistent with the apparent tight binding of NADH with a calculated net Kd of ∼0.13 μm for the predominant path.
      During continuous catalytic turnover of the enzyme, only branches B and C would be operative, because the rates for formation of E* are less than the kcat of the reaction, so that E* will not accumulate. Furthermore, branch B is the predominant reaction pathway during turnover, since the amplitude of branch B (or B + A) is ∼70% of the total amplitude (supplemental Fig. 2).
      The kinetics of binding of α-ketoglutarate in the presence of NADH, when both are rapidly mixed with enzyme, indicate that binding is ordered with NADH binding first. The kinetics are consistent with a rapid equilibrium step followed by a conformational change. Therefore, only a Kd value can be determined for this step without knowledge of either of the individual on- or off-rates. The apparent rapid equilibrium binding of α-ketoglutarate was not recognized in the earlier literature (
      • Dubrow R.
      • Pizer L.I.
      ). Rather, the concentration dependence was plotted as a straight line that yielded an off-rate of ∼30 s–1. Reevaluation of the original data confirms that it most likely conforms to a shallow hyperbola, consistent with the present analysis.
      The absence of a burst in the reverse direction indicates that the rate-limiting step is at or before the catalytic turnover step. The rate of ∼3.8 s–1 for the turnover in the reverse direction is in the range of the kcat of 2–3 s–1 for this reaction and suggests that this may be the rate-limiting step. However, determination of the velocity of the reaction with NADH or deuterated NADH (NADD) shows a VH/VD of 1, indicating that at this pH, the hydride transfer step is not rate-limiting. The early literature also determined that VH/VD = 1 at pH 7.5 and concluded that there must be an additional conformational change prior to hydride transfer that was rate-limiting. Since our results do not show any prior steps with rates approximating the kcat, our findings are consistent with this conclusion. Although we were not able to determine an on-rate for α-ketoglutarate binding, it must be sufficiently larger than the kcat to satisfy the requirement for the off-rate of binding to be much larger than the forward rate of the subsequent conformational change. Thus, the optical analysis described here did not unequivocally identify the rate-limiting steps for this reaction in either direction. However, the lack of a burst in the reverse direction and the presence of a burst in the forward direction indicate that the rate-limiting steps in both directions occur prior to hydride transfer in the direction of NADH oxidation. Since none of the other steps that have been elucidated have rates that would be rate-limiting, the original conclusion by Pizer (
      • Dubrow R.
      • Pizer L.I.
      ), that there must be another conformational change that is rate-limiting and that is not detected optically, seems justified. A small, but reproducible isotope effect appears to be present for V/K. Simulations of the steady state reaction suggests that this is most consistent with an effect on the α-ketoglutarate binding step. However, in a two-substrate system, the observation of an isotope effect on the binding of the substrate not containing the isotope is unusual. Additional investigation will be needed to address this.
      Single turnover experiments with or without prebound NADH also each fit to two exponentials. The kobs for enzyme without prebound NADH equals, within error, the sum of the forward and reverse rates for the chemistry. On the other hand, the kobs values for single turnover experiments where NADH is prebound to the enzyme produce rates that are approximately twice as large. This observation suggests that the turnover of the initial NADH bound species (pathway A and D) is incrementally faster due to the NADH-induced conformational change that results from prolonged binding. However, since this turnover occurs only once at the beginning of the reaction, it does not contribute to the overall rate during continuous turnover.
      Previous studies (
      • Grant G.A.
      • Hu Z.
      • Xu X.L.
      ,
      • Sugimoto E.
      • Pizer L.I.
      ,
      • Dubrow R.
      • Pizer L.I.
      ) of NADH binding to NADH-free E. coli PGDH have shown that three to four sites in the tetramer are capable of binding NADH. The consistent observation of multiple exponentials in the pre-steady state studies described here suggest that these four sites are apparently grouped into two sets of sites with two sites per tetramer belonging to each set. Furthermore, the amplitude data for NADH binding suggest that one set of sites accounts for ∼70% of the initial binding. Studies utilizing hybrid tetramers of PGDH (
      • Grant G.A.
      • Xu X.L.
      • Hu Z.
      ), where the tetramer has from zero to four catalytically competent active sites, suggested that the enzyme displays half-of-the-sites reactivity. This conclusion was based on the observation that removing one site only slightly decreased the specific activity (∼85% of the total) and removing two sites halved the specific activity (50% of the total), but removing three sites appeared to only slightly decrease the specific activity below that seen for two sites (40% of the total). Reevaluation of the specific activity data for the hybrid tetramers in light of this pre-steady state analysis shows that it is consistent, within error, for a tetramer where two sites account for 70% of the activity and the other two sites account for 30% of the activity. Thus, when one, two, or three sites are eliminated, the specific activity would theoretically drop ∼15, 50, and 65%, respectively. These values are similar in magnitude to the decrease in specific activities seen for the hybrid tetramers. The data do not allow one to determine if only two subunits are active at a given time.
      From the ratio of koff to kon for the two sets of sites, Kd values of 0.75 and 5 μm can be calculated for the initial binding of NADH. It is not possible to tell from the kinetic data if this represents cooperativity between the sets of sites or intrinsic differences in affinities of each set. It is noted that the asymmetric unit of the crystal structure of native E. coli PGDH consists of four subunits, indicating a different structure for each subunit. This would be consistent with the active sites not being equivalent prior to ligand binding.
      In summary, this investigation has resulted in new insights into the mechanism of PGDH catalysis. The consistent observation of multiple exponential fits of the data leads to the conclusion that there are at least two distinct sets of sites within the tetramer. This includes the finding that the tight binding of NADH to the enzyme is due to a slow conformational change that takes place in the absence of the other substrate. Although this complex is capable of binding the other substrate and producing product, it does not contribute to continuous turnover. Substrate binding has been shown to be a rapid equilibrium process followed by a conformational change. This investigation has also confirmed that hydride transfer is not rate-limiting in the reverse reaction direction at pH 7.5, confirming what was originally reported and supporting the idea that there is an additional conformational change that is not observed optically that is rate-limiting.
      E. coli PGDH is the prototypical ACT domain-containing protein. ACT domain-containing proteins that bind and are regulated by amino acids or other ligands are becoming recognized as quite common, particularly in bacteria. The physiologic inhibitor of this enzyme, l-serine, binds to a site shared by two adjacent ACT domains and results in inhibition of catalytic activity. The mechanism of how ligand binding to ACT domains influences the proteins activity and whether it is common to all ACT domain proteins is not well understood. The investigation described here provides a much more detailed description of the catalytic pathway than was previously available and provides the basis for determining the specific catalytic step(s) at which serine produces its effects on native and mutant enzymes, thus providing valuable insight into the mechanism of regulation by ACT domains.

      Acknowledgments

      We thank Dr. Kenneth Johnson for helpful comments and guidance and for conducting the annual “Dawn of the New Enzymology” kinetics workshop. We also thank Shawei Chen for excellent technical assistance.

      Supplementary Material

      References

        • Walsh D.A.
        • Sallach H.J.
        J. Biol. Chem. 1966; 241: 4068-4076
        • Schuller D.J.
        • Grant G.A.
        • Banaszak L.J.
        Nat. Struct. Biol. 1995; 2: 69-75
        • Aravind L.
        • Koonin E.V.
        J. Mol. Biol. 1999; 287: 1023-1040
        • Grant G.A.
        J. Biol. Chem. 2006; 281: 33825-33829
        • Bell J.K.
        • Grant G.A.
        • Banaszak L.J.
        Biochemistry. 2004; 43: 3450-3458
        • Grant G.A.
        • Xu X.L.
        • Hu Z.
        J. Biol. Chem. 2004; 279: 13452-13460
        • Grant G.A.
        • Hu Z.
        • Xu X.L.
        J. Biol. Chem. 2002; 277: 39548-39553
        • Pizer L.I.
        • Sugimoto E.
        Methods Enzymol. 1971; 17: 325-331
        • Sugimoto E.
        • Pizer L.I.
        J. Biol. Chem. 1968; 243: 2081-2089
        • Merrill D.K.
        • McAlexander J.C.
        • Guynn R.W.
        Arch. Biochem. Biophys. 1981; 212: 717-729
        • Sugimoto E.
        • Pizer L.I.
        J. Biol. Chem. 1968; 243: 2090-2098
        • Dubrow R.
        • Pizer L.I.
        J. Biol. Chem. 1977; 252: 1539-1551
        • Dubrow R.
        • Pizer L.I.
        J. Biol. Chem. 1977; 252: 1527-1538
        • Schuller D.J.
        • Fetter C.J.
        • Banaszak L.J.
        • Grant G.A.
        J. Biol. Chem. 1989; 264: 2645-2648
        • Al-Rabiee R.
        • Lee E.J.
        • Grant G.A.
        J. Biol. Chem. 1996; 271: 13013-13017
        • Zhao G.
        • Winkler M.E.
        J. Bacteriol. 1996; 178: 232-239
        • Viola R.E.
        • Cook P.F.
        • Cleland W.W.
        Anal. Biochem. 1979; 96: 334-340
        • Johnson K.A.
        Johnson K.A. Kinetic Analysis of Macromolecules. Oxford University Press, New York2003: 1-18
        • Johnson K.A.
        Sigman D.S. The Enzymes. 3rd Edition. Academic Press, New York1992: 1-61