Ligand-promoted Strengthening of Interchain Bonding Domains in Catalytic Subunits of Aspartate Transcarbamoylase*

Attempts to account for the allosteric behavior of enzymes in terms of a ligand-promoted conformational change from a constrained, low affinity state to a re- laxed form of higher affinity for substrates require measurements of the changes in interchain interactions accompanying the allosteric transition. With aspartate transcarbamoylase of Escherichia coli this is a difficult task because of the multiplicity of interactions; these include six bonding domains among the catalytic chains in the two catalytic trimers, six bonding domains linking the catalytic and regulatory chains, and three bonds between the pairs of regulatory chains in the three regulatory dimers. Because of the difficulty in meas- uring independently the effect of active site ligands on the various types of interchain bonding domains, we have utilized isolated catalytic subunits to determine whether the same ligands which promote the allosteric transition of the intact enzyme cause a change in the strength of the bonds linking the catalytic chains in the nonallosteric catalytic trimers. Native catalytic sub- units are extremely stable and exhibit little tendency to dissociate into single chains in neutral solutions. Hence hybridization experiments with native and suc- cinylated subunits were used as a sensitive technique to measure the rate of dissociation of the catalytic trimers. The half-time for dissociation at 0 “C was about 75 h and the rate was even slower at 25 “C. a marked strengthening of the bonds between chains in the results on the bonds between catalytic chains differ strikingly from those for the bonds between catalytic and regulatory chains which are weakened by increasing the temper-

Attempts to account for the allosteric behavior of enzymes in terms of a ligand-promoted conformational change from a constrained, low affinity state to a relaxed form of higher affinity for substrates require measurements of the changes in interchain interactions accompanying the allosteric transition. With aspartate transcarbamoylase of Escherichia coli this is a difficult task because of the multiplicity of interactions; these include six bonding domains among the catalytic chains in the two catalytic trimers, six bonding domains linking the catalytic and regulatory chains, and three bonds between the pairs of regulatory chains in the three regulatory dimers. Because of the difficulty in measuring independently the effect of active site ligands on the various types of interchain bonding domains, we have utilized isolated catalytic subunits to determine whether the same ligands which promote the allosteric transition of the intact enzyme cause a change in the strength of the bonds linking the catalytic chains in the nonallosteric catalytic trimers. Native catalytic subunits are extremely stable and exhibit little tendency to dissociate into single chains in neutral solutions. Hence hybridization experiments with native and succinylated subunits were used as a sensitive technique to measure the rate of dissociation of the catalytic trimers. The half-time for dissociation at 0 "C was about 75 h and the rate was even slower at 25 "C. Moreover, the substrate carbamoyl phosphate and the bisubstrate analog, N-(phosphonacety1)-L-aspartate, caused a marked strengthening of the bonds between catalytic chains in the subunits. These results on the bonds between catalytic chains differ strikingly from those for the bonds between catalytic and regulatory chains which are weakened by increasing the temperature from 0 to 25 "C and by the addition of active site ligands. Although the conformation of free catalytic subunits is doubtless different from that of subunits within the intact enzyme, the results showing the ligand-promoted strengthening of the bonds between catalytic chains are likely to be important in accounting for the cooperativity of aspartate transcarbamoylase in terms of interchain interactions. Grant GM 12159 from the National Institute of General Medical * This work was supported by Public Health Service Research Sciences and by National Science Foundation Research Grant PCM76-23308. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Biochemistry at the University of California, Berkeley. Present address, Laboratory of Cellular Metabolism, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD 20205. 5 Author to whom requests for reprints should be addressed.
The homotropic and heterotropic effects exhibited by some allosteric proteins can be attributed to ligand-promoted conformational changes whereby the proteins are converted from a constrained or low affinity state to a relaxed form having a higher affinity for substrates (1). Because these conformational changes are mediated by alterations in subunit interactions (2), it is important to determine the strengths of the interchain bonding domains and the changes in interaction energies caused by ligands. Only rarely has this been achieved (3-6), and in the case of the regulatory enzyme, aspartate transcarbamoylase (carbamoy1phosphate:L-aspartate carbamoyltransferase, EC 2.1.3.2) from Escherichia coli (7), the evaluation of changes in subunit interactions is particularly difficult because of the large number and different types of noncovalent interchain bonding domains (8). In this paper we demonstrate that the same active site ligand which promotes a weakening in the "bonds" between the catalytic and regulatory chains (9) causes a marked strengthening in the bonding domains between the polypeptide chains in isolated catalytic subunits ( 10).
ATCase' is composed of two catalytic trimers (C) and three regulatory dimers (R) with each catalytic polypeptide chain (c) in one C subunit linked noncovalently to a c chain in the other C subunit via an R subunit (11)(12)(13)(14)(15). In this structure, designated CzR3 or C g 6 , there are six c:c bonding domains within the two C trimers, three r:r bonding domains linking the r chains within the R dimers, and six c:r bonding domains connecting the c and r chains (8,15). Although there is a large aqueous cavity between the two C subunits in the center of the molecule (16), recent evidence from x-ray diffraction studies indicates that some atoms in each of the C subunits are in direct contact (17). Because of the multiplicity of interchain interactions and their intrinsic strengths, ATCase, unlike many other oligomeric enzymes, exhibits little tendency to dissociate into subunits even at concentrations as low as 10"' M in buffers at neutral pH (18). Moreover, very little exchange of either C or R subunits was detected when radioactive subunits were incubated with native or reconstituted ATCase (9,19). Hence, experiments on native ATCase, with one ex-' The abbreviations used are: ATCase, aspartate transcarbamoylase; C, catalytic subunit; R, regulatory subunit; c, catalytic polypeptide chain; r, regulatory polypeptide chain; CzR3, ATCase molecule in terms of its subunit structure composed of two C and three R subunits; cg6, ATCase molecule designated by its constituent six c and six r polypeptide chains; C2R2. ATCase-like molecules lacking one R subunit; CN or C-, native C subunit; CS or C, , succinylated subunit in which all three polypeptide chains are modified by reaction with succinic anhydride; C, , hybrid subunit containing two native and one succinylated polypeptide chains, C, , analogous hybrid composed of one native and two succinylated chains; C, and C,, monomeric native chain and monomeric succinylated chain, respectively; C, , dimer composed of two native chains; PALA, N-(phosphonace-ty1)-L-aspartate. ception (20), have not provided quantitative information regarding changes in interchain interactions when the enzyme is converted from the taut (T) state to the relaxed (R) conformation.
An estimate of the ligand-promoted change in the c:r bond strength has been determined with ATCase-like molecules lacking one R subunit (9). This allosteric, R-deficient species, C,R, (21-24), which contains only four c:r bonds, is less stable than the native enzyme and, under certain conditions, disproportionates to yield ATCase and free C subunits (9, 25). The rate-limiting reaction in this disproportionation process has been shown to involve the rupture of two c:r bonds (25) and the addition of the bisubstrate ligand, PALA, which binds strongly to the active sites (26) causes a 300-fold increase in rate (9,25). Unfortunately, C2R2, like native ATCase, has not been useful for evaluating possible changes in the strength of the c:c bonds. Therefore, studies were conducted on isolated, nonallosteric, C subunits in an effort to determine whether interchain interactions are affected by the same active site ligand, PALA, which promoted the T + R transition in intact ATCase and C2R2. Although C subunits are very stable in neutral solutions and show no evident dissociation in sedimentation velocity experiments at 3 p g / d (27), hybrids could be detected when native (CN) and succinylated (CS) subunits were incubated together for prolonged periods of time. Thus, the formation of hybrids was used as a measure of the dissociation of the trimers which in turn was an indication of the strength of the c:c bonding domains.

MATERIALS AND METHODS
ATCase was prepared according to the procedure of Gerhart and Holoubek (28) and the subunits produced by treating ATCase with neohydrin were separated by chromatography on DEAE-cellulose (29). C subunits were succinylated (13) at a 2.5 molar ratio of succinic anhydride/lysine residue to yield a derivative, CS, containing about four succinyl groups/c chain. 'Z51-labeled C subunits were prepared by the procedure of Syvanen et al. (30) and the specific activity of the modified protein was about 5 X IO5 cpm/pg. Succinic anhydride was purchased from Eastman, neohydrin from K & K Laboratories, and carrier-free Iz5I as the sodium salt in NaOH was obtained from Amersham Corp. PALA was kindly provided by Dr. G. R. Stark of Stanford University.
Hybridization was performed by incubating CN (C,,) and CS (C,) in 40 nm Tris-HC1 at pH 8.0 containing 2 mM 2-mercaptoethanol and 0.2 nm EDTA for various lengths of time. Both protein concentration and temperature were varied and the indicated pH values were those measured at the temperature of the experiment (see "Results"). The four members in the hybrid set, C, , , C, , C, , and C, , were separated electrophoretically in 7.5% polyacrylamide gels (6 cm with a 2cm stacking gel). The discontinuous Tris-barbital system described by Gabriel (31) was employed and gels were stained with Coomassie brilliant blue G-250 (32). In experiments with radioactively labeled proteins, the gels were sliced and the radioactivity measurements on the fractions were made with a Nuclear-Chicago y-counter.

c:c Bonding Domains Are Strong as Indicated by Slow
Rate of Hybrid Formation between C, , and C,,,-Prolonged incubation of C,,, subunits with C, subunits is required for the formation of significant amounts of the hybrids, C, and C, .
The rate of hybridization depends on the pH, ionic strength, and specific anions and, as yet, these effects have not been investigated systematically. As seen in Fig. 1, the formation of significant amounts of C,, and C, at 0 "C and pH 8.0 required 94 h. In this experiment the initial concentrations of C,, and C, were the same and the hybrids were produced in approximately equal amounts.
Quantitative measurements of the rate of dissociation of C, , trimers were performed by incubating mixtures of 1251labeled C, , subunits and unlabeled C, trimers in a 1:lO ratio.  Any native monomers (C,) produced by dissociation of the native subunits preferentially form complexes with succinylated monomers (C.) to yield C, hybrids. Similarly C,, dimers which are only marginally stable (33) would combine rapidly with C. monomers to form C, , hybrids. Thus, the amount of C,,, trimers remaining as a function of time provides a direct measure of the rate of rupture of c:c bonds in the native subunits. Fig. 2a shows that the half-time for dissociation of the C,,, subunits was about 75 h at 0 "C. The rate of dissociation at 25 "C was much slower; so little dissociation occurred in 150 h that the half-time could not be determined accurately?
For comparative purposes we have included in Fig. 2b the results of Subramani et al. (9) on the disproportionation of C2R2 to form C2R3 and free C subunits. As seen in Fig. 26, the The effect of temperature on the rate of hybridization varied with different buffers. When the experiment was performed with Tris-HCl (pH 7.5) or potassium phosphate (pH 7.0) as the buffer, hybridization occurred significantly more rapidly at 0 "C than at 25 "C as seen in Fig. 2a. However, when 4-(2-hydroxyethyl)-2-piperazinepropanesulfonic acid was used as the buffer, the rate of hybridization was virtually the Same at the two temperatures. rate of disappearance of C2&, which depends on the rupture of c:r bonds (25), is greater at 25 "C than at 0 "C. Thus, the c:r bonds in C2& are weaker at 25 "C than at 0 "C, whereas the c:c bonds in isolated C subunits are stronger at 25 "C than at 0 "C.

Strength of Interchain Bonds in Catalytic Trimers
Active Site Ligands Increase the Strength of c:c Bonding Domains- Fig. 3 shows the effect of the substrate, carbamoyl phosphate, on the extent of hybrid formation when C,,, and C, were incubated at 0 "C for 100 h at pH 8.0. Significant amounts of both C,,, and C, were formed in the control experiment containing no ligand. However, very little of either hybrid could be detected when the incubation mixture of C,,, and C, contained the substrate, carbamoyl phosphate, at 4 m~. Analogous hybridization experiments with mixtures of the two proteins in the presence of the bisubstrate analog, PALA, also showed virtually no hybrid formation during the same time interval. In contrast, succinate, which is an analog of the substrate, aspartate, had virtually no effect on the rate of hybrid formation; the electrophoretic pattern was virtually identical to that obtained when no ligands were present. When both carbamoyl phosphate and succinate were present together, the pattern was similar to that observed for the incubation mixtures containing either carbamoyl phosphate alone or PALA.
The hybridization experiments demonstrate clearly that ligands which bind at the active sites of the enzyme cause a marked decrease in the rate of dissociation of the subunits, which can be attributed to a strengthening of the c:c bonding domains.
Kinetic measurements of the rate of hybrid formation were made with mixtures of IZ5I-labeled C,,, and unlabeled C, in both the absence and presence of PALA. The dramatic effect of PALA in decreasing the rate of disappearance of C,,, is shown in Fig. 4a. This experiment was performed with the same preparations of l2'1-labeled C,,, and unlabeled C, as those used for Fig. 2. However, the ratio of the two proteins in Fig. 4a was 1:l as contrasted to 1:lO for the experiment in  Fig. 2a can be attributed to the accumulation in the former of significant amounts of C,,, and C, which in turn tend to disproportionate to form some C,,, and C, . Fig. 4b shows the effect of PALA on the rate of disproportionation of C2R2 (9). The c:r bonds are weakened to a very large extent due to the binding of PALA at the active sites (9). Thus, as was observed for the temperature dependence,

DISCUSSION
Although it has been demonstrated that ATCase undergoes a gross conformational change upon the addition of active site ligands (18, 34-38) and that the alteration in quaternary structure is linked to changes in the tertiary structures of both the c and r chains (39-42), our knowledge of the structural differences between the Tand R-states is still meager. Hence it is not yet possible to correlate thermodynamic parameters for the T + R transition evaluated from enzyme kinetics (18) or physicochemical measurements (18, 43) with structural differences between the two conformations. An intermediate goal in this overall effort to describe the transition in molecular detail is the evaluation of the changes in the interchain interactions. Even this limited objective is difficult to achieve because of the multiplicity of bonding domains and their intrinsic strengths.
Each r chain in ATCase is linked noncovalently to one other r chain and to one c chain (8). Similarly each c chain is linked to two other c chains in the same C subunit and to one r chain (8). In addition, there is some slight contact between each c chain in one subunit and another c chain in the other subunit (17). Hence, it is not surprising that no significant exchange of polypeptide chains has yet been demonstrated when ATCase is incubated with free C or R subunits in neutral solutions. Because of this stability of ATCase, the less stable, R-deficient species, C2&, has been utilized for preliminary assessments of the weakening of the c:r bonding domains accompanying the T + R transition (9). These studies indicated that each c:r bond is weakened by about 1.7 kcal/mol in the conversion of the molecules from the Tto the R-state (9). If this value were to be applied to ATCase, the total decrease in the c:r bond strengths would be 10 kcal/mol for the T + R transition. In contrast, the free energy difference evaluated from enzyme kinetics3 is only -3.3 kcal/mol for the It should be noted that the enzyme kinetics was measured in solutions of much larger ionic strength than those which yielded the value, 1.7 kcal/rnol, for the weakening of a c:r bond caused by the ligand-promoted conformational change (18). Thus, it is important to determine whether there are any compensating changes in the c:c bonding domains upon the binding of active site ligands. Since the effect of these ligands on the c:c bonding domains cannot be determined directly with either ATCase or C2R2, we have used isolated C subunits for this purpose.
As seen in Figs. 1 and 2, dissociation of the trimers is very slow and the c:c bonds are stronger at 25 "C than at 0 "C. Moreover, these bonding domains are strengthened markedly by the addition of ligands which bind at the active sites (Fig.  3). These results on the c:c bonds are in marked contrast to those for the c:r bonding domains in CZRZ (9) which are weakened at 25 "C (relative to 0 "C) and in the presence of the bisubstrate ligand, PALA, which promotes the T + R conversion (Figs. 2b and 4b).
Quantitative estimates are not yet available for the magnitude of the strengthening of the c:c bonding domains upon the binding of active site ligands. The extent of hybridization under the conditions used here is so slight that accurate rates of dissociation of the trimers both in the absence and presence of substrate analogs cannot be determined. Although the rate can be enhanced markedly by the presence of specific anions4 so that the effect of ligands in strengthening the c:c bonds could be measured quantitatively, it is not clear that the data obtained from such experiments could be applied to an analysis of the ligand-promoted changes in interchain interactions in intact ATCase. How much change in the c:c bonding domains occurs when the subunits are incorporated into ATCase is not known. It seems reasonable, nonetheless, to assume that some strengthening of the c:c bonds observed with isolated C subunits upon the addition of active site ligands may alsa occur in the intact enzyme.
Two independent sets of observations suggest that the c:c bonding domains are implicated in an important way in mediating the cooperativity exhibited by ATCase. First, the c chains which are not bonded to r chains in CzRz exhibit the kinetic properties of ATCase and not of free C subunits ( of succinate to C subunits is very weak (47). Carbamoyl phosphate, however, causes a marked increase in the strength of the c:c bonds (Fig. 3), a slight increase in the sedimentation coefficient of the C subunits (35), and a significant perturbation of tyrosine and tryptophan residues leading to a small difference spectrum with sharp peaks at 281.2 and 288.6 nm and a low, broad peak at 300 nm (46).
When both carbamoyl phosphate and succinate are added together to the C subunits, the c:c bonds are strengthened, and there is a significant increase in the sedimentation coefficient of the protein (35,45) as well as a marked perturbation of the near ultraviolet absorption spectrum which is considerably greater than that calculated from the sum of the spectra caused by the individual ligands separately (46). Moreover, the binding of succinate is enhanced considerably when carbamoyl phosphate is also present (46,47). The sedimentation and spectral studies indicate that the addition of the two ligands together causes a significant conformational change in the C subunits which has been interpreted as a contraction or compression of the protein (35,45, 46). This combined effect of the two ligands is observed as well when the bisubstrate ligand, PALA, is added alone. The c:c bonds are strengthened (Fig. 4), the sedimentation coefficient is increased (35), and the spectrum of both tyrosine and tryptophan residues is perturbed (26). Moreover, differential scanning calorimetric measurements have shown that the binding of PALA causes a significant increase in the temperature at which C subunits are denatured (48).
The strengthening of the c:c bonds by ligands which bind at the active sites is particularly interesting in the light of the suggestion from x-ray diffraction studies (17) that the active sites in ATCase are shared between adjacent c chains within the C trimers. This proposed joint participation of pairs of chains in the formation of an active site has been invoked recently in speculations aimed at accounting for the effects of various ligands on the rate of hydrogen exchange in tritiated C subunits (49). In these studies the much larger effect of PALA, compared to carbamoyl phosphate, in decreasing the rate of exchange was interpreted as the result of the bisubstrate ligand acting as a bridge between the chains. Carbamoyl phosphate was assumed to be unable to serve effectively in this way. Our findings show that carbamoyl phosphate, like PALA, causes a marked strenghening of the c:c bonds.5 It should be noted, however, that as yet no quantitative data of their relative effectiveness are available. Thus far, definitive evidence has not been presented which permits the identification of the amino acids that contribute to the binding of substrates or their analogs. If PALA is bound through contacts with two chains, then strengthening of the c:c bonds would be expected. Those contacts would have to account as well for the significant increase in bond strength caused by the much smaller ligand, carbamoyl phosphate. When the complete amino acid sequence of the c chains is determined and a more refined structure of the C subunits is available (17, 51), it should be possible to interpret the findings presented here. and assistance.

Acknowledgment-We thank Ying R. Yang for valuable advice
In the experiments involving prolonged incubation of C,,, and C, in the presence of carbamoyl phosphate there may be some carbamylation of the protein due to cyanate produced by the breakdown of carbamoyl phosphate (50). Since succinylation of C,, (under conditions which inactivate the protein and cause significant modification of lysine residues) weakens c:c bonds, it seems unlikely that nonspecific, slight carbamylation of the protein could cause a strengthening of the c:c bonds. Our conclusion that active-site ligands cause a strenghening of the c:c bonds is reinforced, of course, by the results with PALA which does not cause covalent modification of the protein.