Concerted Feedback Inhibition MODIFIER-INDUCED OLIGOMERIZATION OF THE PXEUDOMONAX

SUMMARY Concerted feedback inhibition of aspartokinase activity from Pseudomonasfluorescens by the combination of L-threo-nine and r.-lysine, or by L-threonine and L-methionine was freely and rapidly reversible. Inhibition by L-threonine, or by the concerted pairs of amino acids was kinetically “mixed” with respect to L-aspartate and noncompetitive with regard to ATP. Stokes radius of the from 44 to a much larger value of A when both L-threonine and L-lysine were added to the enzyme. The same amino acids increased the Stokes radius of the enzyme at 15” in buffer solutions containing ATP, L-aspartate, Mg2+, the the phosphorylation


SUMMARY
Concerted feedback inhibition of aspartokinase activity from Pseudomonasfluorescens by the combination of L-threonine and r.-lysine, or by L-threonine and L-methionine was freely and rapidly reversible.
Inhibition by L-threonine, or by the concerted pairs of amino acids was kinetically "mixed" with respect to L-aspartate and noncompetitive with regard to ATP.
In the presence of L-threonine the Stokes radius of the enzyme increased from 44 to 49 A; a much larger value of about 56 A was estimated when both L-threonine and L-lysine were added to the enzyme.
The same amino acids increased the Stokes radius of the enzyme at 15" in buffer solutions containing ATP, L-aspartate, and Mg2+, the substrates of the phosphorylation reaction, indicating that the larger molecular species were catalytically inactive. Although in phosphate buffer at 25" the combination of L-threonine plus L-methionine did not show an additional increment in the molecular radius over that seen in L-threonine alone, the same combination induced a large change (from 49 to 54 A) in the buffer containing the substrates at 15". L-Threonine, and a combination of L-threonine plus L-lysine also increased the ~20,~ value of the aspartokinase by about 50% ; whereas, a 10% increase in the S value was noted in L-lysine alone.
The cumulative results strongly suggest that the inhibitory amino acids caused oligomerization of the enzyme. Since the inhibition of enzyme activity accompanied the structural alterations of the protein and a good correlation was seen between the degree of inhibition and the extent of enzyme oligomerization under similar experimental conditions, it is proposed that the mechanism of inhibition by L-threonine and by the concerted pairs of amino acid end products may involve modifier-induced formation of inactive enzyme oligomers. * This research was supported by Grant GB 29016 X from the National Science Foundation, and was taken from a dissertation submitted by S. M. D. to the Graduate School of the University of Michigan in partial fulfillment of the requirements for the degree of Doctor of Philosophy.
$ Predoctoral Trainee of the United States Public Health Service, Grant GM 00187. Present address, Department of Biochemistry, University of Miami School of Medicine, Miami, Fla. 33152.
$J To whom inquiries should be addressed.
It is generally accepted that the regulatory interactions between substrates of a given enzyme reaction and the allosteric modifiers that modulate the catalytic activity of that enzyme are indirect.
Thus, binding of these modifiers on the enzyme molecule is believed to cause conformational transitions in the protein structure influencing the rate of conversion of the substrate to the product.
In the previous paper (1) we have described the basic regulatory properties of the aspartokinase (ATP :L-aspartate 4-phosphotransferase, EC 2.7.2.4) from Pseudomonas fluorescens. The activity of this enzyme was inhibited by the concerted action of L-lysine plus L-threonine, and by L-threonine plus L-methionine; individually, r,-threonine was a weak inhibitor, and n-lysine and L-methionine were slightly stimulatory.
This paper reports the structural changes accompanying the binding of these modifiers to the enzyme molecule. Thedata suggest that the mechanism of inhibition by L-threonine, and by the concerted pairs of amino acid end products may involve modifier-induced formation of inactive enzyme oligomers. (See Ref. 2 for a preliminary report on this work.)

Enzyme Preparation
and Assay-Aspartokinase was purified from P. Jluorescens strain P-14 according to the method described (1). The enzyme activity was measured either by the standard hydroxamate assay or by the coupled pyruvate kinase-lactate dehydrogenase system (1).
For assay of enzyme activity at high protein concentrations, a Gibson-Milnes stopped flow spectrophotometer (3) modified to produce linear absorbance' was used. The standard reaction mixture contained 50 mM imidazole-HCI buffer, pH 7.6, 200 rnM KCl, 10 mM MgC12, 2 mM phosphoenolpyruvate, 0.3 mM NADH, 2 mM ATP, 2 mg per ml of rabbit muscle pyruvate kinase (a gift from Dr. M. Flashner, 320 i.u. per mg), 0.25 mg per ml of lactate dehydrogenase (Calbiochem, 6550 i.u. per mg), and 432 pg per ml of aspartokinase previously dialyzed in 20 mM histidine-KOH buffer, pH 7.6 (specific activity 20 pmoles of aspartohydroxamate formed per min per mg of protein), and the amino acid concentrations as described.
Control experiments with 2-fold higher concentrations of the coupling enzymes and the substrates revealed no change in the initial rate of NADH oxidation.
The amino acids also did not influence the activities of the coupling enzymes.
Other velocity experiments have been reported (1). Reagents-The sources of amino acids, marker enzymes, and other reagents were the same as those given in the previous article (1).

Nature of Inhibition
Since L-threonine, and the combination of n-threonine plus Llysine, and L-threonine plus L-methionine inhibited enzyme activity, the effects of these amino acids on the substrate saturation curves were examined. Fig. 1 shows that in the standard hydroxamate assay, inhibition by 0.5 mM each of L-threonine and L-lysine, or by 5 mM each of L-threonine plus n-methionine was not overcome by increasing the aspartate and ATP concentrations up to 50 and 60 InM, respectively.
The same results were obtained when 20 mM n-threonine alone was used to inhibit the enzyme activity.
The double reciprocal plots of the reaction rates versus substrate concentrations ( Fig. 1, B and D) confirmed the "mixed" inhibition kinetics with respect to aspartate and the noncompetitive nature of inhibition with relation to ATP.2 In the coupled assay, inhibition of enzyme activity by L-threonine also appeared to be "mixed" with respect to aspartate and noncompetitive with regard to ATP.
These data suggest that the binding sites for the inhibitors, ATP and aspartate, are separate, and that indirect interactions are to be expected between these ligands when they occupy their respective sites on the enzyme molecule.

Reversibility and Rate of Inhibition
Concerted feedback inhibition by L-threonine and n-lysine was freely and rapidly reversible.
An enzyme solution incubated in the presence of 5 InM each of these amino acids at 25" up to a period of 2 hours showed full activity within 30 s when diluted and assayed in the absence of these modifiers as compared to a 2 High concentrations of ATP were slightly inhibitory; however, increasing the Mg*+ concentrations in the assay mixture abolished ATP inhibition (see Ref. 5).
A and C, effects of 0.5 mM each of L-threonine plus L-lysine (0) and 5 mM each of L-threonine plus L-methionine (a) on enzyme activity as a function of L-aspartate and ATP concentrations, respectively.
The curves drawn through the open circles represent data obtained in the absence of the inhibitors.
The substrate saturation curves obtained with 20 mM L-threonine could be superimposed with those of the concerted feedback pairs of amino acids, and have been left out to insure clarity. B and D, double reciprocal plots of the same data shown in A and C. Only the rates found in the presence of n-threonine plus L-lysine are given in the reciprocal plots. Enzyme activities (8.4 rg) were measured in the standard hydroxamate assay (1). Incubations were for 10 min at 25". Rate (V) = micromole of aspartohydroxamate formed.
For explanation of the substrate inhibition at higher ATP concentrations, see Footnote 2.
control sample not incubated with the amino acids. A reciprocal experiment in which an enzyme solution was freed of the inhibitors by exhaustive dialysis and subsequently assayed in the presence of the inhibitors revealed the expected reduction of activity in less than 15 s. These data show that both the onset of inhibition as well as the reversibility of inhibition was as rapid as could be measured by the conventional spectrophotometric assay (also, see below).

Structural Alterations in Protein
Changes in Stokes Radius-Gel filtration of aspartokinase on calibrated Sephadex G-200 columns revealed that the elution profile of the enzyme was drastically altered in the presence of the amino acid modifiers ( Fig. 2). At 25", the enzyme in 50 mM potassium phosphate buffer, pH 7.5, containing 100 mM KC1 eluted slightly after yeast alcohol dehydrogenase ( Fig. 2A) ; a Stokes radius of about 44 A was calculated from the equations described (6, 7). When the gel filtration buffer was supplemented with 5 mm n-threonine ( Fig. 2B), the peak of enzyme activity shifted to an earlier fraction to correspond with a molecule having a Stokes radius of about 49 A. In the presence of 1 mM each of n-threonine and L-lysine, the Stokes radius increased further to about 56 A as evidenced by its elution profile coinciding with that of pyruvate kinase (Fig. 2C). From a series of similar Sephadex G-200 experiments in 50 mM phosphate buffer, pH 7.5, containing 100 InM KCl, the Stokes radius of the enzyme was determined in the presence of several amino acids either added individually or in certain combinations; the results are summarized under Experiment I in Table I. It is clear that Lmethionine, n-isoleucine, or L-lysine did not alter the Stokes radius significantly; whereas, L-threonine alone, and the combination of n-threonine plus L-lysine increased the molecular radius of the enzyme.
Although inclusion of L-methionine in the phosphate buffer already supplemented with L-threonine did not cause additional change in the Stokes radius at 25", a synergistic effect of these amino acids on the change in the Stokes radius was observed at 15" in histidine buffer in the presence of the substrates (see Experiment II, Table I). The increases in Stokes radii by these modifiers were also ob- Three different G-200 columns (about 85 X 1.4 cm) were equilibrated at 25" in 50 IrIM potassium phosphate buffer, pH 7.5, containing 100 mM KCl, 0.02% sodium azide, and the inhibitor concentrations as follows: A, no amino acids; B, 5 mM n-threonine; C, 1 mM n-threonine plus 1 mM n-lysine.
Purified aspartokinase (420 pg) was mixed with 200 pg of yeast alcohol dehydrogenase (ADH), 15 rg of horseradish peroxidase (PER), and 100 pg of rabbit muscle pyruvate kinase (PK) and applied to the columns (for assay of enzyme activities and calculations of Stokes radii, see Ref. 1). served in buffers containing the substrates and the cofactors required for the phosphorylation reaction. Results given in Table  I, Experiment II, summarize the data obtained at 15" in 20 mM histidine-KOH buffer, pH 8.0, containing 10 InM MgC12, 200 mM KCl, 5 mu L-aspartate, 2 InM ATP, and 0.02% sodium azide in the presence and absence of several amino acid end products. Without any added modifier, the Stokes radius of the enzyme in the presence of the substrates was 41 A. n-Lysine and n-methionine added singly did not alter the molecular radius.
However, when L-threonine, a combination of n-threonine plus Llysine, or n-threonine plus L-methionine was added to the enzyme the Stokes radius increased drastically.
The changes in the Stokes radii seen in this buffer system were in general quite similar to those observed in phosphate buffer at 25" with one exception: in histidine buffer (containing the substrates of the enzyme) there was a large increase in the molecular radius of the enzyme when L-methionine was added along with L-threonine; this additional increase seen at 15" was not detected in phosphate buffer at 25' (cf. Experiment I, Table I). It should be re-emphasized that in either buffer system, the two amino acids, Llysine and n-methionine, which stimulated enzyme activity under certain conditions (1) appear not to alter the protein structure in a way to be detected by gel filtration. 8543   TABLE   I Stokes radii of aspartokinase in presence and absence of various amino acids Aspartokinase (420 pg) was mixed with yeast alcohol dehydrogenase, horseradish peroxidase, and rabbit muscle pyruvate kinase and applied to Sephadex G-200 columns (85 X 1.4 cm) equilibrated in the following buffer solutions with or without the feedback inhibitors.
When the same buffer was supplemented with 5 mm L-threonine the S value increased to 9.58; although the schlieren pattern was somewhat broader, n-lysine alone seemed to have a small effect, and in the presence of Lthreonine plus L-lysine the S value was almost the same as that calculated for n-threonine alone.
Table II also shows the approximate molecular weights of the various oligomers calculated from the S values given above and from the Stokes radii determined from gel filtration experiments (7). Since the sedimentation velocity centrifugation for each of the experiments was carried out at a single protein concentration between 3.33 and 3.96 mg per ml (see legend to Fig. 3) and the values of Stokes radii were obtained at about 0.4 mg per ml (initial concentration applied on Sephadex G-200), these molecular weight values are only approximate; nevertheless, these results show that the molecular weight of aspartokinase in the presence of n-threonine or n-threonine plus n-lysine, was almost twice as that estimated for the enzyme in the absence of these ligands .

Relationship between Inhibition of Activity and Xtructural Alterations in Enzyme Molecule
The results of gel filtration experiments and sedimentation velocity centrifugation studies presented thus far indicated that those amino acid modifiers which inhibited enzyme activity velocity patterns of aspartokinase in the presence and absence of various amino acids. Purified enzyme was dialyzed extensively against 50 mM potassium phosphate buffer, pH 7.5, containing 100 mM KCl, 10 mM p-mercaptoethanol and the amino acids as indicated: A, no amino acid; B, 5 mM L-lysine; C, 5 mM L-threonine; D, 5 mM L-threonine plus 5 mM L-lysine.
Centrifugations were carried out in single sector cells at 20'; photographs were taken at 8-min intervals at a bar angle of 60" after the rotor had reached two-thirds of the top speed of 52,000 rpm. Sedimentation was from left to right. In all cases, the first picture shown was taken at 48 min. The following under assay conditions (2 to 8 pg per ml) also caused drastic structural changes in the molecule at protein concentrations ranging from 400 pg per ml to 4 mg per ml. In view of this, it was necessary to show that the activity of aspartokinase was also subject to concerted feedback inhibition at higher enzyme concentrations.
The data of stopped flow experiments given in Fig.  4 clearly reveal that the aspartokinase activity was inhibited by the combination of L-threonine and L-lysine at a protein concentration of 432 pg per m1.3 Since the experimental conditions for 2 The stopped flow experiments revealed an interesting aspect of the phosphorylation reaction that could be easily missed in the coupled pyruvate kinase-lactate dehydrogenase assay with a protein concentrations were used: A, 3.33 mg per ml; B, 3.88 mg per ml; C and D, 3.96 mg per ml. conventional spectrophotometer. A significant lag in NADH oxidation (about 100 ms) preceded the linear rate; with increasing concentrations of the amino acid inhibitors, this lag was almost For other details, see legend of Fig. 3. Molecular weights were estimated from the appropriate Stokes radii (as determined by gel filtration on Sephadex G-200), and from the sedimentation coefficient values obtained as above using the equation described by Siegel and Monty (7 Table I; A, per cent inhibition of activity at 2.1 fig per ml of enzyme; 0, per cent inhibition of activity at 432 rg per ml of enzyme (calculated from the data given in Fig. 4). See text for other details.
the stopped flow measurements were almost identical with those of the gel filtration studies shown under Experiment II in Table I (with the exceptions that the enzyme activity was assayed at 25" in the imidazole-HCl buffer, at pH 7.6, and the assay mixtures also contained the coupling enzymes and their substrates), we conclude therefore that the enzyme species with larger molecular radii were catalytically inactive. Fig. 5 summarizes the data of the per cent Stokes radius change and per cent inhibition of enzyme activity as a function of increasing equimolar concentrations of L-lysine plus L-threonine. The results show a sham transition from a smaller active to a larger inactive form of the enzyme; the change is about 90% complete within a narrow range of inhibitor concentrations from zero to 0.5 mM each. A good correlation can be seen between the degree of inhibition (at enzyme concentrations of 2 and 432 pg per ml) and the extent of enzyme oligomerization (at protein concentration of 420 pg per ml) as evidenced by the increase in the Stokes radius. DISCUSSION A wide variety of regulatory enzymes is known to undergo drastic structural changes leading to enzyme aggregation in the presence of feedback modifiers. Some examples are : homoserine dehydrogenase of Rhodospirillum rubrum (8)) chorismate mutaseprephenate dehydratase of Salmonella typhimurium (9), and cytosine triphosphate synthetase (lo), ribonucleotide reductase (11)) lysine-sensitive aspartokinase (12)) and threonine-sensitive aspartokinase I-homoserine dehydrogenase I (13, 14) of Eschecompletely abolished. The inset given in Fig. 4 shows that even at an enzyme concentration of 2 pg per ml but in an otherwise identical assay mixture without the feedback inhibitors, the lag was clearly defined; in the presence of the inhibitors this lag was eliminated.
These observations ruled out the nossibilitv that the coupling enzymes were rate-limiting.
Since the enzyme solution for these experiments was freed of the inhibitory amino acids by exhaustive dialysis, one possible explanation for the observed lag might be that the aspartokinase molecules assume a different conformation in the presence of the substrates and that this new conformational form is enzymatically active.
rich& coli. Nevertheless, it has been clearly established that, in many cases, the aggregation phenomenon may not be directly implicated as the physical basis for the regulation of enzyme activity (9,14,15).
The data presented here on the concerted feedback inhibition of aspartokinase from P. Jluorescens indicate that, in this instance, aggregation of the enzyme by the amino acids may be an integral part of the mechanism of feedback control.
The following observations are relevant in this regard.
(a) In the absence of feedback modifiers, the Stokes radius of the enzyme at a protein concentration of 420 pg per ml and in buffer without the substrates of the phosphorylation reaction was about 44 A. At the same protein concentration but in buffer supplemented with substrates, the Stokes radius was about 41 A (Table II) indicating that the substrates did not alter the protein structure to a significant extent.
(b) In the presence of the inhibitory ligands, the molecular radius of the enzyme increased drastically; increases of similar magnitude were detected in buffers with and without substrates and cofactors at a protein concentration of 420 pg per ml (Table II).
The change in the molecular radius was proportional to the concentrations of the ligands present during gel filtration, and almost 90% of the maximum change was complete at inhibitor concentrations of 0.5 mM each (Fig. 5, open circles).
(c) Assay of enzyme activities at 432 pg per ml using stopped flow technique revealed that inhibition of aspartokinase activity by the combination of L-threonine plus L-lysine increased with increasing concentrations of these ligands reaching 90 ye of the maximum inhibition at 0.5 mM each of the amino acids (cf. Figs. 4 and 5, closed circles).
(d) Upon removal of the inhibitory ligands either by dialysis or by gel filtration, full enzyme activity was restored; under these conditions a return from the oligomeric form of the enzyme to the "native" molecular size was shown by a decrease in Stokes radius and a reduction in the sedimentation coefficient (see enzyme purification scheme, Ref. 1). The above findings clearly show that, at least in one instance, i.e. at a protein concentration of about 420 pg per ml and in the ligand concentration range from 0 to 0.5 InM each, a good correlation exists between the inhibition of enzyme activity and ligandmediated oligomerization of the protein.4 It is possible to argue that the changes in the elution profile on gel filtration in the presence of the amino acid modifiers might have been the result of some conformational changes leading to unfolded forms of the enzyme; alternatively, the large molecular radius could have reflected protein aggregation.
A 50 ye increase in the sedimentation coefficient of the enzyme (as well as a large increase in molecular radius) in the presence of the modifiers (Table II) appears to rule out intramolecular conformational changes in the protein molecule; an unfolding of the protein would accompany a decrease in the S value, since a larger molecule of the same mass would sediment more slowly.
In addition, an approximately 30% increase in the Stokes radius is compatible with a doubling of the molecular weight (17).
The approximate molecular weight of 137,000 of the enzyme in the absence of the modifiers (also, cf. 1) and a molecular weight 4 Although the enzyme activity is also severely inhibited by the feedback modifiers when assayed at protein concentrations varying from 2 to 8 rg per ml (see Fig. 5 of 268,000 in the presence of 5 mM each of L-threonine and Laggregated and the unaggregated species in the presence of this ligand (Table I).
One important question remains to be answered in order to demonstrate convincingly a direct relationship between the structural change in the protein and the inhibition of enzyme activity: does the rate of formation of enzyme oligomers occur as rapidly as the regulatory event itself? We have shown that the reversible concerted feedback inhibition of aspartokinase by n-threonine plus n-lysine occurred at least as rapidly as could be measured with the conventional spectrophotometric method and by the stopped flow apparatus.
However, at present we have little information on the rate of formation of enzyme oligomers. Thus, we cannot completely rule out the possibility that a rapid intramolecular conformational change may precede aggregation of the enzyme molecule.
Nevertheless, the cumulative data obtained thus far strongly suggest that modifier-induced formation of inactive enzyme oligomers is an integral part of the mechanism of feedback inhibition control.