Subunit Interactions in Aspartate Transcarbamylase THE INTERACTION BETWEEN CATALYTIC AND REGULATORY SUBUNITS AND THE EFFECT OF LIGANDS

SUMMARY The interaction between the catalytic subunit (c3) and the regulatory subunit (rz) of aspartate transcarbamylase from Escherichia coli was studied by measuring the reversible formation of the c3r6 complex as a function of r2 concentration. Conversion to the native enzyme was prevented by using a very low concentration of c8 (40 ng per ml) in the presence of bovine serum albumin. A simple hyperbolic r2 saturation curve was obtained suggesting the presence of only one kind of c:r domain. From the association constant for the formation of cars, the free energy of c:r interaction can be estimated to be about -10 Cal per mole. Neither CTP nor ATP appears to affect the strength of c : r interaction in this complex. Succinate in the presence of carbamyl phosphate promotes tighter binding. is inhibited


SUMMARY
The interaction between the catalytic subunit (c3) and the regulatory subunit (rz) of aspartate transcarbamylase from Escherichia coli was studied by measuring the reversible formation of the c3r6 complex as a function of r2 concentration.
Conversion to the native enzyme was prevented by using a very low concentration of c8 (40 ng per ml) in the presence of bovine serum albumin. A simple hyperbolic r2 saturation curve was obtained suggesting the presence of only one kind of c:r domain.
From the association constant for the formation of cars, the free energy of c:r interaction can be estimated to be about -10 Cal per mole.
Neither CTP nor ATP appears to affect the strength of c : r interaction in this complex.
Succinate in the presence of carbamyl phosphate promotes tighter binding.
At higher concentrations of cg and nonsaturating levels of r2, conversion to the native enzyme (c6r6) takes place. This renaturation process is second order with respect to the concentration of c3 and is virtually irreversible. Renaturation is inhibited by saturating levels of r2 and to some extent by both CTP and ATP.
The effect of ligands on c :r interactions reported here may have significance in the allosteric mechanism of the native enzyme.
In the preceding paper (a), physicochemical evidence was reported indicating that the catalytic subunit (~3)~ of aspartate transcarbamylase interacts reversibly with the regulatory subunits (r2) to form a complex which most probably has a c3rg structure.
,4lthough at nonsaturating levels of rZ, the reversible formation of the complex is accompanied by conversion to the native enzyme (c6rs), these two processes can be studied separately by selecting the appropriate c3 concentration. Thus, in the reversible reaction cg + 3r2 s care, the ratio of c3 to c3rg should be independent of the total concentration of cg and c3r6 * This work was supported by Grant MT-2954 from the Medical Research Council of Canada. A preliminary report of these results has been presented (1). This paper is Part III of a series on subunit interactions in  whereas the conversion of c3 and cars to c6r6 should be highly concentration-dependent.
At extremely low concentrations of cZ, conversion to c6r6 therefore becomes insignificant and the formation and breakdown of the car6 complex should then appear as a simple equilibrium.
The association constant then provides a direct measure of the strength of subunit interactions.
We report here results indicating that at, low c3 concentrations, the interaction between c3 and r2 may indeed be described as an equilibrium involving only one type of binding site. An unexpected finding is that the strength of the c:r subunit interaction so determined is unaffected by CTP and ATP.
Also presented here are the kinetics of conversion of the c3r6 complex to the native enzyme at somewhat higher concentrations of c3 and nonsaturating levels of r2. This process is second order and is inhibited by high levels of rZ and by both CTP and ATP. The implications of these results 011 the allosteric mechanism of aspartate transcarbamylase will be discussed in the following paper (3).

MATEMALS AND METHODS
The preparation of aspartate transcarbamylase, the separation into subunits, and the enzyme activity assays were as described in the preceding paper (2). The chemicals used were also the same except that the L-aspart,ic acid used in the assays was U-"C-la-

RESULTS
Extent of c3r6 formation as junction of r2 concentration-At low aspartate concentrations (I or 2 m&r), the postulated c3r6 complex is more active than the catalytic subunit (c3) because the K, of the complex is smaller (2,4). Using the increase in enzymic activity to measure c3rg, we have studied the extent of c3r6 formation as a function of rZ. In order to minimize the extent of conversion to the native enzyme, very low concentrations of c3 (40 ng per ml) were used in these experiments.
Previous work has shown using the continuous titrimetric assay that the formation of c3rg was so fast that the increase in activity upon the addition of r2 occurred with no apparent time lag (4). It was therefore possible to start the enzymic reaction soon (1 min) after the addition of r2 to ea. By keeping the preincubation time to a minimum and by using comparatively short assay times (5 and 10 min), the possibility of conversion to the native enzyme was further decreased.
If such conversion took place, it would cause a marked reduction in the measured activity because the native enzyme has very little activity at these low aspartate concentrations.
The fact that the duplicate assay for 5 and for 10 min always agreed within experimental error therefore showed that under the above conditions, no significant conversion to the native enzyme occurred during the assay.
As judged by the increase in activity upon adding r2, the formation of the c3rg complex appears to be a simple saturation process and its dependence on r2 concentration follows a hyperbolic curve (Fig. la). These results suggest that only a single binding step is involved.
The simplest, though not the only possible explanation (see "Discussion"), is that the following equilibria exist (Equation 1) and that the three c:r domains in c3r6 are independent of each other. -r2 This explanation further assumes that the formation of each c:r contact affects only the active site located in the particular catalytic polypeptide involved and not the neighboring active sites. If these assumptions are valid, we can simply consider that the catalytic polypeptides can exist in two states (cr and co) depending respectively on whether a regulatory subunit is bound to it or not. The above equilibria can then be represented as a single process, co + r2 s c' for which Equation 2 holds where the square bracket terms indicate the concentration of the species and K,,,,, is the association constant for c : r binding.
13ecause cr and co have different kinetic properties, the change in activity at 1 mM or 2 mM aspartate can be related to the concentration of these two states. Thus the increase in activity, (v -~0) is given by v.at -vo vsnt -vo A plot of l/(v -vo) uersus l/[rz] should therefore be a straight line and when l/(v -vo) = 0, the intercept l/[rz] is equal to -LSOC.
The double reciprocal plot (Fig. lb) shows that the above relationship holds at both 1 InM and 2 mM aspartate con-  For the data in the presence of succinate the line shown is obtained by least squares fit.
For the ATP and CTP data, in order to provide comparison, the line shown here is the same graph as in Fig. 16 for 1 mM aspartate obtained in the absence of either CTP or ATP.
However, for determining the association constants in the CTP and ATP experiments (Table I), a least squares procedure was used.
centrations. The association constants determined from the intercepts (Table I) are not significantly affected by the change in aspartate concentration.
Although it would be desirable to make similar measurements at higher concentrations of aspartate, these experiments could not be performed because it would then be impossible to distinguish between c3r6 and cQ on the basis of activity.
If the observed r2 saturation indeed represents the equilibrium of binding at the c :r domain, then the free energy of interaction (AGf,,) can be calculated from -1tT 111 K,,,,, to be = -10.2 Cal per mol.
E$ect of Ligands on TZ Saturation-It was of considerable interest to examine the possible effects of the allosteric ligands CTP and ATP on the strength of subunit interactions as measured by the r2 saturation method described above. Surprisingly, the data (Fig. 2) obtained in the presence of either 1 mM CTP or 5 rnM ATP are quite similar to those obtained in the absence of the ligand.
The values of K,,,,, computed from a least squares fit of the data also do not differ substantially (Table I). It thus appears that CTl' and AT1 have little or no effect on the c:r interactions in the c3r6 complex.
On the other hand, the sub- strate analog succinate appears to promote tighter binding between c3 and r2 as reflected in the higher association constant.
To gain some insight into the nature of the forces involved in c:r binding, we also determined the effect of urea, ionic strength, and organic solvent on the subunit interactions.
It was found that urea at a concentration of 2 M greatly weakened the binding whereas reduction in the ionic strength resulted in tighter binding ( Fig. 3 and Table I). Addition of 0.2 nr propanol to the lower ionic strength medium was shown to lead to weaker binding. It thus appears that ionic charges, hydrophobic interactions, and perhaps hydrogen bonds may all be involved in the association between the catalytic and the regulatory subunits. Conversion inio Native Enzyme (c6r6)-It has been shown that at higher concentrations of CQ and nonsaturating concentrations of r2, conversion to the native enzyme takes place (2,4). Although it was possible previously to monitor this renaturation process in a continuous manner by observing the decrease in activity with time (4), this method was unreliable for quantitative rate measurements because the depletion of substrates and the accumulation of products might introduce serious errors. Because the renaturation process is essentially irreversible (a), it was decided to follow it by removing samples at intervals and diluting into the assay mixture containing r2. As the conversion of c3rZn (see Footnote 2) to the native enzyme (cgrO) is known to be dependent on carzn concentrations (4), the 5-or IO-fold dilution into the assay mixture will essentially stop further renaturation. The presence of a relatively high concentration of r2 in the assay mixture ensures that practically all of the remaining c3rZn will be converted to c3r6 and this also serves to prevent further renaturation in the withdrawn samples. Previously, it was found that in the presence of substrates, renaturation occurred over approximately 15 min at 0.75 pg of c3 and 2.5 pg of r2 per ml. However, in the absence of substrates, the renaturation appeared to proceed faster, and lower initial c3 concentrations were necessary to permit convenient study of the process. The r2 concentration was also increased to help further reduce the rate of renaturation. The course of renaturation at 0.2 pg of c3 and 5 pg of r2 per ml is shown in Fig. 4 The mixture contained initially 0.2 wg of ca per ml in 0.2 M Tris-acetate, pH 8.5, 20 mM 2-mercaptoethanol, and 50 pg per ml of the albumin at 25' and renaturation was started by adding 5 rg per ml of rz. Samples from the same renaturation mixture were assayed in three different assay systems for 10 and 20 min as described under "Materials and Methods." The substrates and ligands indicated on the graphs denote the conditions in the assay systems.
were assayed under three different conditions in order to provide some information about the nature of the product. Initially no native enzyme was present and all of the carzn was converted in the assay mixture to c3rg which had about the same activity at 20 mM aspartate in the presence of either CT1 (1 mM) or ATP (5 mM). As renaturation proceeded, the activity in the CTI' assay decreased by large amounts and approached a low final level whereas the activity in the ATP assay decreased to a much smaller extent and considerable final activity was maintained. At 1 mM aspartate, the relatively low initial activity due to c3r6 dropped further and finally reached only a barely detectable level. These results are consistent with the expected conversion of carp,, (which is measured in the assay as c3rg characterized by a low K, and insensitivity to CTP and ATl') to the native enzyme which is known to have little activity at low aspartate concentrations and to be sensitive to CTP and ATP. The duplicate lomin and 20.min assays agreed within experimental error indicating that no further renaturation took place once the samples were diluted into the assay mixture.
Because the activity in the 1 mM aspartate assay decreased 20.fold during renaturation, this assay provided the most accurate means of monitoring the renaturation process and was used in all of the subsequent experiments.
The 5-fold decrease in activity in the CTP-containing assay at 20 mM aspartate also gave similar (though always slightJy higher) renaturation rates. The assay containing ATI' was however, less reliable for following renaturation because the maximum decrease in activity amounted to less than 40%.
The kinetic data on renaturation were analyzed by converting the activities to concentrations of c3rzm and c6r6. In the following analysis, vo represents the activity at zero time; ot, the activity at time t; and u,, the activity at the end of renaturation.
If we consider the specific activity of c3r6 and c6r6 in terms of the amount of catalytic polypeptide contained in these species, then we obtain are indicated on the graph. Ten micrograms of rz per ml were added to start renaturation in each case. Other conditions were the same as in Fig. 4. 1)uplicate samples were assayed for 10 and 20 min at 1 mM aspartate. The second order rate constants were determined from the slope of the lines in Figs. 5 and 6 and expressed in terms of native enzyme formed from 2 eq of cQrZn (see Footnote 2) using molar quantities.
Conditions and therefore (ut -uJ/~u~ -v,) is equal to (CX) the fraction of the total catalytic polypeptides existing in the form of cprqn. When l/a is plotted against renaturation time (t), a straight line is obtained (Fig. 5) indicating second order dependence on c3rZn. This result is also confirmed by varying the initial concentration of c3 and the second order rate constants obtained (Table II)  The renaturation rate was found to depend also on the r2 concentration (Fig. 6). The rate constant decreased approximately linearly with the increase in r2 concentration within the range of r2 levels used. These results are in agreement with our earlier qualitative observation (2,4) indicating that reassociation to the native enzyme is inhibited when all of the c3 present is converted t0 c3r6. The implication of this finding on the mechanism of reassociation is discussed in a later section.
The renaturation process was found to be inhibited to some extent by CTI' and by ATY (Fig. 7). In both cases, as the concentration of the nucleotide was increased, the amount of inhibition reached a maximum and then declined.
Maximal inhibition occurred at nucleotide concentrations (0.5 mM CTl' and 1 m&f ATI') which are known to alter significantly the activity of the native enzyme (5). These concentration levels are in the 7. Effect of CTP and ATP on renaturation. Renaturation was conducted with0.2pg of ca and 1Opg of riper ml in the presence of varvina amounts of CTP and ATP and the rate was followed bv taking duplicate samples at 30 and at 60 min. Other conditions were the same as in Fig. 5. The second order rate constants were then plotted as percentages against concentration of ligand. same range as the binding constants determined for CTP and ATI' binding to r2 (6,7). The tendency to produce lower inhibition at high concentrations of either CTP and ATI' is difficult to explain but may bc due to the binding at the weaker set of sites on r2 reported by various workers (8-M).
The effect is not caused by the occupation of the carbamyl phosphate site in the active center because it is not obtained with the competitive inhibitor pyrophosphate at a level equal to five times the Ki.

I)ISCUSSION
If we consider the arrangement of subunits and hence the number of c :r domains in the native enzyme, we can expect the formation of car6 from c3 and r2 to proceed via the intermediates c3r2 and c3r4. Of the various species with the general formula c3rZn (where n = 0, 1, 2, or 3), it was only possible to characterize ~3 and c3r6 because the reversibility of c3ra formation implies that the intermediate species would normally be found only in an equilibrium mixture containing all of the different forms. The properties of c3r2 ad c3r4 can therefore only be obtained by extrapolating from the corresponding properties of c3 and c3r6. 111 a mixture of cs and c3rg, the measurement of the proportion of c3r6 depends on the fact that c3rs has higher activity at low aspartate concentrations.
We must therefore consider the following alternat.ives regarding the activity of c3r2 (and similarly earl) at low levels of aspartate: (a) that it has the same activity as c3; (5) that it has the same activity as c3r6; (c) that car2 has an activity equal to two-thirds the activity of c3 plus one-third the activity of c3r6; and (d) that its activity is not related in a simple manner to that of c3 and c3r6. The observation that the curve relat)ing r2 concentration with the increase in activity is a simple hyperbola is difficult to reconcile with the complex situation postulated under (d) but is not inconsistent with the remaining alternatives.
If the binding of one or two r2 subunits to c3 produces no change at the active sites (alternative a) and the characteristic properties of car6 are brought on only when the third r2 subunit binds, then a simple hyperbolic saturation curve would result. However, from a mechanistic point of view, it is unlikely that such an all-or-none system operates.
Similarly, it would be difficult to envisage a molecular mechanism whereby the binding of the 1st r2 subunit results in all of the properties of c3rg whereas the binding of two subsequent r2 subunits produces no eff cct (alternative b). Thus the most reasonable explanation of the hyperbolic r2 saturation curve is to postulate that as each r2 subunit binds to a catalytic polypcptidc in c3 it affects only the active site on that polypeptidc giving therefore one-third the effect observed in c3rs (alternative c). This ,mechanism, together with the assumption that the binding of each r2 subunit does not affect the binding of other r2 subunits, is able to account for the observed r2 saturation curve in a mechanistically reasonable manner.
The hyperbolic saturation curve then simply reflects the single process of r2 binding to one of three independent domains on c3. This explanation of the r2 saturation results is in agreement with previous evidence indicating that the catalytic polypeptides in c3 behave independently (11). If we assume that the above explanation is correct, then the r2 saturation provides a most convenient measure of the strength of c:r interaction.
Of the three types of subunit interactions present in native aspartate transcarbamylase, the c:r interaction appears to be the weakest. The c : c interaction must be stronger because no dissociation of c3 into an individual c subunit is detectable under nondenaturing conditions (I 1). In the case of r2, spontaneous dissociation into monomers has been observed (6) ; however, the tendency to dissociate is significantly less than in the case of c3r6. Therefore, the r:r interaction should be somewhat intermediate in strength between the c :c and the c:r interactions.
The c : r interaction is however sufficiently strong to maintain the c6r6 structure of the native enzyme because even the most energetically favorable mode of dissociation (i.e. w-6 --cerq + r2) would require the break-up of two c:r contacts. The standard free energy change for this process would therefore be about 20 Cal corresponding to a K,,,,, of 4.5 X lOI hl-I. This estimate of the minimum energy required to dissociate the native enzyme is consistent with the observation that no dissociation can be detected at room temperature even when the aspartate transcarbamylase concentration is in the microgram per ml range (12). Our estimate of the strength of c:r interaction also implies that csr4 should be relatively stable because its dissociation would also involve the breakup of two c : r contacts regardless of which of the following processes occurs: c6r4 S c& + r2; c6r4 $ c3 + c3r4; or c6r4 = 2 c3r2. On the other hand hand cgr2 should dissociate as readily as c3r6 because this process involves only one c : r contact.
These calculations arc supported by the recent reports of the isolation of c6r4 from renaturation mixtures deficient in rz (13,14).
The results on the kinetics of reassociation clearly show that the formation of native enzyme is second order with respect to the concentration of c3r2-. This indicates that a rate-determining bimolecular step is involved but does not provide further information regarding the reacting species. The observation that saturating levels of r2 strongly inhibit the reassociation process shows that 2 molecules of car6 react poorly if at all. This result is to be expected because in this case no c :r contact can be made between the two reacting components.
Apart from c3r6, the most prevalent cQr2n species under our experimental conditions is car4 because the r2 concentrations used (5 to 20 pg per ml) are many times greater than the half-saturation level (1 pg per ml). From a mechanistic point of view, the most favorable reaction would be the combination of c3r4 with c3r2 directly to give cgr6 as this simply requires a "snapping together" of two components. If this would be the main pathway for the formation of cgrg, then the reassociation constant would be proportional to [ l/M3 This result would suggest that renaturation under the above conditions does not proceed predominantly via a combination of c3r4 with cs2 but that the main pathway consists of a rate-determining reaction between c3r., and c3r6 followed by kinetically fast rearrangement steps to give the native enzyme. Much more extensive studies regarding the effect of r2 concentration on the renaturation rate would be required to substantiate this suggested mechanism.
The effects of ligands on r2 saturation and on the renaturation kinetics are important for the elucidation of the allosteric mechanism because they provide insight regarding changes in subunit interactions.
Ideally, it would be desirable to investigate the effect of aspartate over the whole concentration range in which sigmoidal kinetics is observed.
However, this has not been possible because it is only at low aspartate concentrations that cs and cara can be distinguished on the basis of activity. If the effect of aspartate is similar to that of succinate, then the observed increase in strength of c : r interaction induced by succinate may be significant in the allosteric transition between the tight and the relaxed forms of the native enzyme. One must be cautious however, in extrapolating from the succinate experiments because we have shown by inhibition studies that in addition to being a substrate analog, succinate has other effects on in line with the lack of influence of these nucleotides on the activity of c3r6 observed earlier (2,4). It is highly unlikely that this absence of effects by CTP and ATP is due to the inability of these ligands to bind to c3r6. If this were the case, then one would observe a shift of the r2 saturation curve by CTP or ATP because these ligands are known to bind to rz (7) and therefore their presence would favor the dissociation of c3r6. The identical r2 saturation curve obtained in the presence or absence of the nucleotides therefore implies that they bind equally well to r2 and t0 c3r& The fact that CTI' and ATP influence the conversion of c3rZn to c6r6 is also consistent with their ability to bind to w-6.
Although the well known effects of CTP and ATP on the kinetics of the native enzyme might lead one to expect that the two ligands should have opposite effects on c3r6, our results show that both CTP and ATP inhibit the conversion to c6r6. A possible explanation is that in the absence of these ligands, c3r6 has the optimal orientation of subunits for incorporation into the native c6r6 structure (perhaps by having a similar subunit orientation to that in aspartate transcarbamylase).
If this is the case, then CTP and ATP, by changing this orientation of subunits in opposite directions, both would inhibit the process of conversion to c6r6. It is probably significant in this regard that Colman and Markus (15) have observed that both CTP and