Aspartate transcarbamylase of Escherichia coli. Heterogeneity of binding sites for carbamyl phosphate and fluorinated analogs of carbamyl phosphate.

Some preparations of both native aspartate transcarbamylase from Escherichia coli and catalytic subunit have fewer tight binding sites per oligomer for carbamyl-P than the number of catalytic peptide chains. In contrast, the number of sites for the tight-binding inhibitor N-(phosphonacetyl)-L-aspartate does equal the number of catalytic chains in each case. Binding of the labile carbamyl-P was determined using rapid gel filtration, with conversion to stable carbamyl-L-aspartate during collection. Native enzyme (six catalytic chains) obtained from cells grown under the conditions of J.C. Gerhart and H. Holoubek (J. Biol. Chem. (1967) 242, 2886-2892) has 5.4 tight sites for carbamyl-P at pH 8.0 (KD = 9.9 muM), whereas native enzyme from cells grown with higher concentrations of glucose, uracil, and histidine (to yield more enzyme per unit volume of culture) has only 1.9 tight sites at pH 8.0 (KD = 4.6 muM) and only 2.3 tight sites at pH 7.0 (KD = 2.6 muM). At pH 8.0, catalytic subunit (three catalytic chains) obtained from the former native enzyme has 2.2 tight sites for carbamyl-P (KD = 2.4 muM) and the number of sites is 2.3 in the presence of 35 mM succinate, whereas catalytic subunit obtained from the latter native enzyme has 1.8 tight sites (KD = 3.6 muM) in the absence of succinate and 2.3 tight sites in its presence. The number of tight binding sites is also less than the number of subunit peptide chains in 19F nuclear magnetic resonance experiments performed with catalytic subunit and two fluorinated analogs of carbamyl-P at comparable concentrations of analogs and active sites. A model is proposed in which incomplete removal of formylmethionine from the NH2 termini of the enzyme under conditions of extreme depression affects affinity for ligands.

from Escherichia coli and catalytic subunit have fewer tight binding sites per oligomer for carbamyl-P than the number of catalytic peptide chains. In contrast, the number of sites for the tight-binding inhibitor N-(phosphonacetyl)-L-aspartate does equal the number of catalytic chains in each case. Binding of the labile carbamyl-P was determined using rapid gel filtration, with conversion to stable carbamyl-L-aspartate during collection. Native enzyme (six catalytic chains) obtained from cells grown under the conditions of J. C. Gerhart and H. Holoubek (J. Biol. Chen. (1967) 242, 2886-2892) has 5.4 tight sites for carbamyl-P at pH 8.0 (K, = 9.9 PM), whereas native enzyme from cells grown with higher concentrations of glucose, uracil, and histidine (to yield more enzyme per unit volume of culture) has only 1.9 tight sites at pH 8.0 (K, = 4.6 pM) and only 2.3 tight sites at pH 7.0 (K, = 2.6 FM). At pH 8.0, catalytic subunit (three catalytic chains) obtained from the former native enzyme has 2.2 tight sites for carbamyl-P (K. = 2.4 pM) and the number of sites is 2.3 in the presence of 35 mM succinate, whereas catalytic subunit obtained from the latter native enzyme has 1.8 tight sites (K, = 3.6 pM) in the absence of succinate and 2.3 tight sites in its presence.
The number of tight binding sites is also less than the number of subunit peptide chains in 19F nuclear magnetic resonance experiments performed with catalytic subunit and two fluorinated analogs of carbamyl-P at comparable concentrations of analogs and active sites. A model is proposed in which incomplete removal of formylmethionine from the NH, termini of the enzyme under conditions of extreme derepression affects affinity for ligands.
Application of NMR spectroscopy to problems of biological interest has increased rapidly with the advent of Fourier transform methodology and with the availability of instruments capable of observing nuclei other than 'H. "F is an attractive NMR probe for several reasons, and there are now several studies in which substrate analogs labeled with this nucleus have been used (l-4) or in which the protein has been modified covalently with a fluorine-containing probe (5, 6). "F NMR is almost as sensitive as 'H NMR, but the range of chemical shifts is much larger and the shifts are extremely sensitive to changes in environment.
Since the substrate analog usually contains the only 19F in the system, backgrounds are low. Thus, spectra can be obtained at comparable molar concentrations of protein and "F. In contrast, experiments with 'H probes are usually done with ligand concentrations greatly in excess of protein concentrations (for examples with aspartate transcarbamylase, see Refs. 7 and 8 much greater than that of 'H, for which "F is usually substituted. To the extent that this difference is important, substrate analogs labeled with i9F may behave very differently from substrates upon binding to enzymes. In initial experiments, we found that FLAP,' a trifluoro analog of carbamyl-P, binds tightly to substantially fewer than three sites per catalytic subunit trimer, even though the transition state analog PALA binds tightly to all three sites. Although partial saturation of C,R, by carbamyl-P was observed by Rosenbusch and Griffin (9) using equilibrium dialysis, these workers also reported that C, has 2.9 tight binding sites for carbamyl-P. However, there was a good deal of scatter in the experiments with C,, especially at low ligand concentrations, due at least in part to extensive decomposition of carbamyl-P during the dialyses. The apparent disagreement between our results with FLAP and the results of Rosenbusch and Griffin with carbamyl-P led us to study the binding of Binding of Carbamyl Phosphate to Aspartate Transcarbamylase 5967 carbamyl-P to C, under more optimal experimental conditions. With a ligand as labile as carbamyl-P, the ultrafiltration method of Cantley and Hammes (IO) did not provide sufficient precision to distinguish unambiguously between two and three tight binding sites per C, trimer. The gel filtration method of Hummel and Dreyer (11) has an important advantage for work with a labile ligand in that the experiments can be run quite rapidly. By converting the labile carbamyl-P into a stable form as it is collected, we were able to eliminate any effects of decomposition after collection and also to obtain accurate data within each experiment for the rate of decomposition of carbamyl-P on the columns. Precision and reproducibility with this method were excellent. We find that carbamyl-P does bind tightly to substantially fewer than three sites per C, trimer. Furthermore, the number of tight binding sites for carbamyl-P to C,R, varies quite markedly with the source of the enzyme.  Gerhart and Holoubek (12), in medium containing 4.0 g/liter of nglucose, 50 mg/liter each of L-histidine and L-leucine, and 4 mg/liter of uracil. For cell Batch A these quantities were changed, in order to obtain a higher yield of protein/liter of growth medium, to 6.0 g/liter of n-glucose, 100 mg/liter of L-histidine, no L-leucine (the bacterium is not a leucine auxotroph) and 12 mg/liter of uracil. Conditions similar or identical to the latter ones have been used at the New England Enzyme Center for the growth of many lots of cells because the yields of cells are higher. Enzyme Native aspartate transcarbamylase prepared as described by Gerhart and Holoubek (12) was stored at -20" as an ammonium sulfate precipitate.

Growth
The enzyme was dissolved in 40 mM phosphate buffer (pH 7.0), 0.1 mM EDTA, 10 mM 2-mercaptoethanol before use. Catalytic subunit was prepared as described by Gerhart and Holoubek (12). Protein concentrations were determined by quantitative amino acid analysis, with standards run concurrently. There was no significant change from the amino acid compositions determined previously. Preparations of C,R, were at least 95% pure, as measured by gel electrophoresis of the undenatured protein.
Preparations of C, had specific activities greater than 0.7 mol of carbamyl aspartate formed/ min/g of enzyme at 28" and pH 8.0. The number of PALA-binding sites as measured by the method of Collins and Stark (13)  The binding of carbamyl-P to enzyme was measured at 28" according to Hummel and Dreyer (ll), using Sephadex G-25. Columns were equilibrated with buffer and ["Clcarbamyl-P and each experiment was begun by placing a known amount of enzyme in the same solution onto the column. Concentrations of enzyme in stock solutions were determined by amino acid analysis except for the experiments of Fig. 3, where the concentration was determined from the absorbancy at 280 nm and checked by titrating the active sites with PALA as described by Collins and Stark (13 had a quartet corresponding to the trifluoromethyl group centered 35.4 ppm from external tetramethylsilane, J13c--31 r = 118 Hz, and a doub!et corresponding to the methylene group centered at 129.9 ppm, Jzsc--31 p = 23 Hz. Infrared spectra of FLAP (in a KBr pellet or in D,O) showed that it is a hydrate (gem diol) at neutral pH; no carbonyl band was detected.
Reduced FLAP-To 50 mg of FLAP in water at pH 8.0 was added 1.3 mmol of NaBH, The mixture was stirred at room temperature for 2 h and then 1 M HCl was added slowly to decompose any unreacted borohydride.
This mixture was placed onto Dowex l-X8 formate and eluted with a formic acid/HCl gradient as above. The reduction yielded 40 mg of reduced FLAP. The structure was confirmed by NMR analysis. The isF spectrum had a doublet centered 4.31 ppm up field from external trifluoroacetic acid (consistent with the structureCF,-CHOH-) which collapsed to a singlet upon irradiation of proton resonances, Ji%--l~ = 6.5 Hz; the "P spectrum (with proton noise decoupling) had a single broad peak 22.9 ppm downfield from external W'O,.

NMR Experiments
Fourier transform ' ¶F and Y spectra with proton noise decoupling were obtained with a Varian XL-100 spectrometer.
Internal trifluoroacetate was used as a reference for the igF spectra. Titrations were done by adding aliquots of a stock solution of FLAP or reduced FLAP to the enzyme in an NMR tube, using a Hamilton syringe. In the NMR fast exchange limit, the rate of exchange of free and bound species is larger than the difference in resonance frequencies (l/r >> 2aAr). For binding of carbamyl-P to C,, l/r can be estimated to be 1.7 x lo3 s-l (19). This rate of dissociation is a lower estimate for FLAP, which does not bind to C, as tightly as carbamyl-P. If the difference between the chemical shifts of free and bound FLAP is 100 Hz or less, the condition for fast exchange is satisfied. A single resonance is observed with a chemical shift (a,,,) given by Atz Eo Since 6, was found to vary with pH, the pH of each solution was measured before and after each titration.
Values for the saturation of the enzyme are calculated from 6,,, and Ab , and from these one can determine the distribution of bound and free ligand. Linear Scatchard plots of binding data have slopes equal to -l/K, and intercepts equal to n, the number of molecules of ligand bound/ molecule of enzyme.
Where there is cooperativity or more than one class of sites, the plots are no longer linear. Values for K, and n can be determined by fitting the data to a general expression for saturation Data for the binding of FLAP to catalytic subunit under different conditions fit two models: (a) one tight site and two weak sites, or (b) two tight sites and one weak site. For these models, the expression for saturation is tration range of carbamyl-P that could be investigated was limited, so that a third binding site with an affinity 10 times smaller would not have been detected. We have shown previously that C, and C& from cell Batch A have three or six sites, respectively, for the tightly bound inhibitor PALA (13, 21) and this stoichiometry of PALA binding was confirmed with the preparation of C, used in the experiment of Fig. 3. Rosenbusch and Griffin (9) reported previously that the number of binding sites for carbamyl-P to C,R, increased at pH 7 from three to six in the presence of succinate; we used succinate with C, to test whether the third potential binding site for carbamyl-P would be revealed in the presence of this aspartate analog. Such an experiment can be performed  Calculations are made from the data of Figs. 3 to 6. For experiments in the presence of succinate, the observed K, was obtained from Scatchard plots (not shown) of the uncorrected data. Calculated K, has been obtained from the Scatchard plots shown in Figs. 3 and 4. In this case, the assumption was made that succinate binds only to the C,.carbamyl-P complex so that the effect of succinate on the apparent dissociation constant for carbamyl-P would be evident. The concentration of succinate and the value of K, for its dissociation from the ternary complex were used to calculate the relative amounts of C,.carbamyl-P'succinate. Since carbamyl-P dissociates only from the binary complex (21), the effect of succinate is to increase the apparent dissociation constant for carbamyl-P. because, fortuitously under the conditions we have used, the constants for dissociation of succinate from the C,.succinate binary complex and the C,.carbamyl-P.succinate ternary complex are both 3.5 mM (21). Hence, succinate does not perturb the equilibrium Cs + carbamyl-P * Ca.carbamyl:P because it binds equally well to C, and to C,.carbamyl-P.
As shown in Fig. 3 and Table I, 3.5 mM succinate does increase the number of sites for carbamyl-P and a further increase is observed with 35 mM succinate.
Binding of carbamyl-P to C&, was difficult to observe in the presence of succinate at pH 8, because succinate competes more effectively with carbamyl-P for C,R, than it does for C, at this pH. At 0.5 pM carbamyl-P, increasing concentrations of succinate cause a decrease in the amount of carbamyl-P bound to the enzyme, with half-decrease at 4 mM succinate (data not shown). At appreciably higher concentrations of succinate, no binding at all is observed. Using preparations of C,R, from cells grown under conditions similar to the ones we have used (see below) Rosenbusch and Griffin (9) did observe that, at pH 7, 5 mM succinate increased the number of tight binding sites for carbamyl-P from 3.2 f 0.5 to 5.94 + 1.35; the carbamyl-P concentrations used in the presence of succinate were 10 to 160 WM. They also observed that K, for carbamyl-P increased from 14 fiM without succinate to 27 FM with succinate, an apparent paradox now explained by the binding of succinate to free C& (21). Jacobson and Stark (21) also observed that K, for the dissociation of succinate from C,.succinate is only slightly lower at pH 7 than at pH 8, whereas K, for the dissociation of succinate from C,.carbamyl-Pesuccinate decreases much more. Thus the affinities of succinate for C&, and C,R,.carbamyl-P may be nearly equal at pH 7.

Inhibition of Cs by FLAP-FLAP
is competitive with carbamyl-P (Fig. 7) and noncompetitive with L-aspartate (Fig. 8) at pH 7.8. A replot of the slopes of Fig. 7 versus FLAP concentration is a parabola. At the low concentrations of carbamyl-P and L-aspartate employed, the activity observed in the experiment of Fig. 7 is almost exclusively due to binding of carbamyl-P to the tight sites only. The parabolic competitive inhibition observed with FLAP is very reminiscent of the pattern ob-  served with phosphate, interpreted by Porter et al. (22) as evidence for a complex with 2 phosphates per active site. By analogy, it may be possible to form E.FLAP and E. (FLAP), complexes at each active site. By fitting the parabola, K, values of about 0.04 and 0.1 mM can be calculated for the 1st and 2nd molecules. Other possibilities for parabolic competitive inhibition involve interaction among the active sites of C,: the binding of FLAP to a low affinity site of a complex with stoichiometry C,FLAP, may strengthen the binding of FLAP to the high affinity site already occupied, or the binding of FLAP to one active site of C, may decrease the affinity of the unoccupied site or sites for carbamyl-P. Under the conditions of Fig. 7, there can be no significant binding of carbamyl-P and FLAP to the same active site, since noncompetitive rather than competitive inhibition would then be observed.
NMR Studies with FLAP and C,-All experiments were done with Cg from cell Batch A. "F NMR signals for free FLAP (1046.8 Hz upfield of internal trifluoroacetate) and 3.6 mM FLAP in the presence of 1.5 mM C, (0.2 M Tris-acetate buffer, 10 mM 2-mercaptoethanol, 1 mM EDTA, pH 8.0) are shown in Fig. 9. Scatchard plots of the data (Fig. 10) show that FLAP binds to C, with at least two affinities. At pH 7.0, the data are described best by a model in which 1 molecule of FLAP binds very tightly (K,, = 0.05 mM), while 2 others bind much more weakly (K,, = 1.77 mM). At pH 8.0, the relative stoichiometry is less clear and the data can be described by models with one tight site (KLo = 0.05 mM) and two weak sites (KH,= 0.27 mM) or two tight sites (KLo = 0.08 mM) and one weak site (KH, = 0.8 mM).
The error in measuring & (+0.4 Hz) causes a constant error in ~(1) of f 0.10; for 8 degrees of freedom x2 is 12.0 for the model with one tight site and 9.9 for the one with two tight sites. From the maximum likelihood ratio for the two models, there is a slight (2 to 1) preference for two tight sites at pH 8.0.
Approximately two tight sites are also seen in the binding of carbamyl-P to C, at pH 8.0. At pH 7.5, the data are fit best by assuming that there is one tight site and at pH 8.5, a model with two tight sites gives the best fit (Table  II)  values were assumed to be in the range 0.05 to 0.10 mM and a best fit was then determined. "These data can be fit by models with either one or two tight binding sites (see the text). Dissociation constants are calculated for each case.
'From parabolic competitive inhibition uersus carbamyl-P, data of Fig. 7. were observed in D,O as a function of pH, and the pK, values found were compared with pK, values from titration of FLAP with KOH.
As shown in Table  III, the phosphonate protons have pK, values of 1.8 and 6.3 and the hydroxyl of the hdyrated carbonyl group begins to titrate at pH 12. There is no uptake of KOH by FLAP between pH 7.5 and pH 8.0 but there is a marked upfield movement of the chemical shift of the methylene protons and a small movement of the fluorine shift, perhaps indicating that there is a pH-dependent conformational change of FLAP without an ionization.3 Since the change in "F shift with pK, 8 for free FLAP is small (1.5 Hz) compared to the change in shift for FLAP bound to the enzyme between pH 7 and pH 8 (8.6 Hz), and since the titration of the shift is much sharper than predicted for titration of a single group (Fig. ll), the pH dependence must be due at least in part to titration of C, and not solely to a conformational change of FLAP.   and C, (1.8 mg/ml) in low ionic strength buffer (5 mM imidazole acetate, 1 mM 2-mercaptoethanol, 0.1 mM EDTA, pH 7.0) give a difference spectrum with maxima at 281.4 nm and 289.2 nm and a magnitude 78% of that obtained with carbamyl-P under these conditions. The K, for FLAP at pH 7.0 obtained from a spectral titration is 1.2 mM, assuming three equivalent sites. A Scatchard plot of the data is linear. The value 1.2 mM is much greater than the value of K,,, determined in the NMR experiments. Adding 10 mM succinate results in a new difference spectrum just like the one induced by succinate alone (25), indicating that succinate displaces those molecules of bound FLAP responsible for the difference spectrum. The spectral magnitude decreases as the ionic strength of the buffer increases. In 200 mM imidazole acetate with 10 mM FLAP and 2 mg/ml of C,, no difference spectrum at all is observed, even though FLAP still binds tightly to C, as judged by 18F NMR studies. Both this effect and the large difference between KLo from NMR experiments and the K, derived from the difference spectrum titration indicate that tight binding of the first molecule of FLAP to C, is not accompanied by a change in the near ultraviolet absorption spectrum of the enzyme. If it is assumed that the tight site is saturated without formation of a difference spectrum, spectrophotometric titration of the weak sites leads to an estimate of 0.6 mM for K,, in 5 mM buffer, a value to be compared with 1.0 mM in 10 mM buffer determined by NMR.

Competition between FLAP and Other Anions-
Interaction of Reduced FLAP with C,-The binding of FLAP and reduced FLAP to C, were compared to help evaluate the role of the hydrated carbonyl group. The phosphonate protons of reduced FLAP have pK, values of 2.7 and 7.3, by titration with KOH. Both ionic forms are assumed to bind to C, equally well (19). With low concentrations of buffer (10 mM imidazole acetate, pH 7.0, and 10 mM Tris-acetate, pH 8.0) A* = 16 Hz and KHI = 1.8 mM at pH 7 and A* = 10 Hz and KH, = 0.55 mM at pH 8. The change in chemical shifts upon binding and K,, values are similar to those of FLAP. The data were not adequate to determine accurate values for K,, but two classes of binding sites were definitely observed. Succinate and L-aspartate displace reduced FLAP from C,.
FLAP and C,R,-Accurate dissociation constants or estimates of cooperativity could not be derived in this case because the bound chemical shift is about 2 Hz at pH 7.0 and only about 1 Hz at pH 8.0, so that changes are extremely small and difficult to measure accurately. The decreased magnitude of the bound shifts indicates that the electronic environment of FLAP is more like its environment in water when it is bound to C& than when it is bound to C,. The bound linewidth for FLAP at pH 7.0 is larger with C& (26 Hz) than with C, (12 Hz), as expected from the different sizes of the two enzyme species.

DISCUSSION
Relation of Conditions for Cell Growth to the Number of Tight Binding Sites for Carbamyl-P-Using equilibrium dialysis, Rosenbusch and Griffin (9) found 3.2 f 0.5 tight binding sites for carbamyl-P to C&X, at pH 7.0, K, = 14 pM, compared with 2.3 tight sites and K, = 2.6 pM from our experiment done under essentially the same conditions (Fig. 5). The cells used by Rosenbusch and Griffin were grown under conditions identical with those for cell Batch A, except that 50 fig/liter of L-leucine was also present. Their yield of enzyme was about twice that obtained when the cells were grown under the conditions of Gerhart and Holoubek (12): Therefore, C,& from two independent batches of cells grown in the richer medium have substantially fewer than six tight sites for carbamyl-P. Furthermore, C, from the cells grown in Base1 has about two tight binding sites for carbamyl-P at pH 7 or pH 8 without succinate and about three with succinate,' roughly in agreement with our values at pH 8.
The New England Enzyme Center has grown mutant cells for the isolation of aspartate transcarbamylase under the conditions used for cell Batch A since 1969,5 and enzyme from such cells has been used by a number of investigators. Both C&, and C, from cell Batch A have one binding site for PALA per catalytic chain, and Hammes et al. (26) have shown that Cs, also prepared from cell Batch A, has three binding sites for carbamyl-P in the presence of succinate at high ligand concentrations. Since the maximum velocities for most preparations of enzymes are quite close to the values given by Gerhart and Holoubek (12) when the assays are performed at high concentrations of carbamyl-P, we conclude that the sites with low affinity for carbamyl-P can be filled eventually and that these sites are then fully active. Furthermore, Rosenbusch' has found that the allosteric properties of C& prepared from cells grown in the richer medium are reproducible and very similar to the properties of enzymes from cells grown according to Gerhart and Holoubek (12). Again, such measurements are usually done at high concentrations ofcarbamyl-P.
A Possible Cause for Weak Carbamyl-P Binding-Both Weber (27) and Herve and Stark (28) found about two-thirds the number of free NH, termini expected from the number of peptide chains for both C, and regulatory subunit prepared from cells grown according to Gerhart and Holoubek (12). Weber (29) also reported that highly variable amounts of methionine and threonine were found as the NH, termini in different preparations of regulatory subunit. These results imply that, under conditions of derepression, when a large amount of a single protein is being synthesized, the enzymes which ordinarily remove formylmethionine from nascent chains (30) may be overwhelmed, so that in part formylmethionine or methionine remains instead of the NH, termini present in normal cells. Perhaps incomplete removal of formylmethionine is even less efficient in the richer medium, where the yield of enzyme is greater. If C, chains with formylmethionyl or methionyl termini were to bind carbamyl-P less well than chains with NH,-terminal alanine, the low number of tight binding sites for carbamyl-P in both preparations of C, studied here would be explained. In the case of C,R, from cells of Batch A, there are only about two tight binding sites for carbamyl-P, whereas six is the number predicted from the sum of the two C, subunits. Therefore, the nature of the NH* termini of the regulatory chains in C,& may also affect carbamyl-P binding. A correlation between the nature of NH, termini and the properties of C& might also provide an explanation for the formation and properties of the unusual enzyme produced in the presence of 2-thiouracil and studied by Kerbiriou and Herve (31,32). In this enzyme, homotropic cooperativity has been lost. Since 2-thiouracil is converted to 2-thiouridine phosphate in uioo (see Ref. 33 for a recent discussion and references) and since 2-thio-UMP has been reported to be a strong allosteric inhibitor of C&, (34) UMP or P-thio-UTP in high concentration in the cells stabilizes a structure of nascent R chains in which correct processing of the NH, termini is impeded.
Obviously, careful further work will be required to substantiate or disprove the suggestions made above. However, regardless of the detailed explanation, it is important now to determine the number of tight binding sites for carbamyl-P in preparations of enzyme to be used for binding studies of any kind and for some other kinds of experiments as well. As we have discussed, the capacity of the enzyme to bind 1 molecule of PALA per catalytic chain and to exhibit normal maximum velocity and allosteric properties at high carbamyl-P concentrations does not necessarily mean that the enzyme is chemically homogeneous. With new appreciation for the possible effects of heterogeneity, we must view with new caution our work (23,35) and the work of others (for example, Ref. 36) with derivatives of C, and C&R, in which the number of tight binding sites for carbamyl-P or PALA has been altered through chemical modification. We suggest also that studies of nucleoside triphosphate binding (37)(38)(39)(40) and temperature jump studies with nucleoside triphosphates (41)(42)(43) may be complicated by enzyme heterogeneity.
Experiments with FLAP and Reduced FLAP-With C,, the changes in chemical shifts upon binding (A,) for FLAP and reduced FLAP indicate deshielding of the trifluoromethyl group. However, A, is relatively small (10 to 20 Hz) compared with the values 65 to 100 Hz for the CF, groups of substrate analogs bound to chymotrypsin, where there is a hydrogen bond between the imidazole of histidine 57 and the carbonyl oxygens of the analogs (1, 2). The small value of Ab for the binding of FLAP to C,& indicates that the presence of the regulatory subunit alters the environment of the CF, group in the active site. Although counterbalancing effects are possible, it is likely that there is little difference in polarity between the environment in C&, and the one in solution.
Although both FLAP and carbamyl-P bind tightly to about two sites on C, at pH 8, the number of tight sites for FLAP decreases to one at pH 7. In contrast, Rosenbusch' finds two tight sites for the binding of carbamyl-P to C, at pH 7 or pH 8. It can be seen from Fig. 5 that the number of tight sites for binding of carbamyl-P to C&, increases slightly with decreasing pH. The pH dependence of Ab and K, is not caused by titration of the ligands since FLAP and reduced FLAP have very different titration curves in water but the same dependence of Ab and K, on pH. The anomalous pH dependence for FLAP may indicate that the highly electronegative CF, group interacts with the enzyme in a manner that distorts the active site. Such an interaction is also indicated by the properties of an inactive derivative of C, in which the single sulfhydryl group in the active site (23) has been modified with CF,-C(OH),CH,-.
This derivative binds carbamyl-P but not succinate in the presence of carbamyl-P (44), just like the inactive thionitrobenzoate derivative, also modified on the sulfhydryl group (23). In both cases, electronegative parts of the modifying groups may be near enough to positively charged groups of the enzyme to impede their binding to a carboxylate group of succinate.
Pre-existing Asymmetry or Negative Cooperativity in CI-Although CsRs is a classic example of an allosteric enzyme, cooperative binding of ligands to unmodified C, has not been described previously, except for the kinetic experiments of Heyde (45) with ITP as an inhibitor. Jacobson and Stark (23) had shown that only 1 molecule of PALA binds tightly to C, in which the sulfhydryl group in each active site has been modified with a thionitrobenzoate group. Although the possibility of intrinsic asymmetry was considered, a model in which a conformational change and negative cooperativity were involved was favored. Kempe and Stark (46) have shown that Ca modified in the active sites with one or two molecules of pyridoxamine-P produces a large increase in fluorescence upon the binding of PALA, but that the fully modified derivative, with 3 molecules of pyridoxamine-P per trimer, does not, even though it can be shown that PALA still binds to this species. A model involving interaction among the three active sites of the partially modified species was used to explain these findings. At pH 7, unmodified C, has one tight binding site for FLAP and, from Rosenbusch's data,' two tight binding sites for carbamyl-P. If these properties are to be explained by intrinsic asymmetry, three different kinds of sites are required. Negative cooperativity, in which conformational changes upon partial saturation result in decreased affinity at unoccupied sites, is an attractive alternative, although at present we do not have evidence to choose definitively between these models. Binding of Carbamyl Phosphate to Aspartate Transcarbamylase