Self-Association and Allosteric Properties of Glutamine-dependent Carbamyl Phosphate Synthetase REVERSIBLE DISSOCIATION TO MONOMERIC SPECIES*

Abstract Glutamine-dependent carbamyl phosphate synthetase from Escherichia coli can exist in two different conformations which exhibit, respectively, monomer sedimentation coefficients (s020,w) of 7.3 S and 8.7 S. Both conformations undergo rapid reversible self-association upon addition of potassium ions or ornithine. Formation of the 8.7 S conformation is promoted by the addition of phosphate ions. Low concentrations of urea (0.2 to 2.0 m) or guanidine hydrochloride (0.05 to 0.5 m), both of which inhibit the enzymatic activity, favor dissociation in potassium phosphate buffer, which is reversible. Ammonia has been found to be a potent allosteric activator of the enzyme. The presence of ammonia or ornithine (also an allosteric activator) promotes oligomer formation; the maximum sedimentation coefficient observed (in potassium phosphate buffer containing ornithine) is 14.8 S. Sedimentation equilibrium studies indicate the existence of tetramer or higher species in potassium phosphate buffer containing ornithine. Models are presented which explain the apparent relationship between allosteric regulation of the enzyme and its state of association.

self-association upon addition of potassium ions or ornithine.
Formation of the 8.7 S conformation is promoted by the addition of phosphate ions.
Low concentrations of urea (0.2 to 2.0 M) or guanidine hydrochloride (0.05 to 0.5 M), both of which inhibit the enzymatic activity, favor dissociation in potassium phosphate buffer, which is reversible.
Ammonia has been found to be a potent allosteric activator of the enzyme. The presence of ammonia or ornithine (also an allosteric activator) promotes oligomer formation; the maximum sedimentation coefficient observed (in potassium phosphate buffer containing ornithine) is 14.8 S. Sedimentation equilibrium studies indicate the existence of tetramer or higher species in potassium phosphate buffer containing ornithine. Models are presented which explain the apparent relationship between allosteric regulation of the enzyme and its state of association.
Glutamine-dependent carbamyl phosphate synthetase from Escherichia coli catalyzes the following reaction (1) : PATP + HCOS-+ Hz0 + L-glutamine Mg*+ K+ ) NH&02P03-+ L-glutamate + 2ADP + Pi The enzyme also catalyzes carbamyl phosphate synthesis when glutamine is replaced by ammonia. The glutamine-dependent synthetase activity of the enzyme is allosterically regulated by a number of purine and pyrimidine nucleotides. Thus, purine nucleotides (e.g. IMP) stimulate activity, and pyrimidine nucleotides are either inhibitory (e.g. UMP) or have no effect (cytidine * This work was supported by grants from the National Science Foundation and the National Institutes of Health, Public Health Service. nucleotides) on the activity (2) ; this indicates a mechanism by which a balance can be maintained between the rates of purine and pyrimidine biosynthesis.
IMP is a more potent activator than are XMP, Ghll', GT)P, GTP, and AMP, while UMP inhibits more effectively than do UDP and UTP; in general, the degree of activation or inhibition decreases as the number of steps required to synthesize t,he various nuclcotides from IMP and UMP, respectively, increases (2). The finding that ornithine is an effective allosteric activator arose from the observation by PiBrard (3) that ornithine counteracts the inhibition produced by pyrimidine nucleotides.
Anderson and Marvin (4) subsequently demonstrated that ornithine is a potent allosteric activator even in the absence of UMP.
Since E. coli has only one carbamyl phosphate synthetase which catalyzes carbamyl phosphate formation for both the arginine and pyrimidine biosynthetic pathways (5), the ability of ornithine to stimulate the enzyme seems to offer a mechanism that can provide carbamyl phosphate for arginine biosynthesis even when UMP is present in excess. Studies on the several partial reactions catalyzed by carbamyl phosphate synthetase (6) suggest that the allosteric effecters influence the over-all activity by altering the affinity of the enzyme for the ATP utilized in phosphorylation of enzymebound carbamate, rather than for the ATP that is used for the activat.ion of carbon dioxide (2).
Anderson and Marvin (4, 7) carried out sucrose gradient centrifugation studies of the enzyme in potassium phosphate and reported that the sedimentation coefficient varied with increasing enzyme concentration.
They also found that IMP and ornithine increased the sedimentation coefficient while UMP decreased it, and suggested that a monomer-oligomer equilibrium accounted for the changes in the sedimentation behavior of the enzyme. Independent studies in this laboratory (8), in which the sedimentation properties of the enzyme were examined by analytical ultracentrifugation, showed that the sedimentation coefficient of the enzyme can vary from about 7 S to 15 S depending upon the nature of the solvent.
In an effort to understand these observations we undertook a systematic investigation of the effects of solvent perturbation on the sedimentation velocity behavior of the enzyme as observed in the analytical ultracentrifuge.
The results obtained have led to the finding of appropriate solvent conditions for the reversible production of alternate monomer conformations. We have conducted sedimentation equilibrium studies of the monomeroligomer equilibrium produced in the presence of potassium phosphate and ornithine, and also of the monomer which can be obtained in the absence of potassium phosphate.
These studies have shed light on the apparent parallelism between stimulation of enzymatic activity and self-association of the enzyme.
In the course of this work we discovered that ammonia promotes self-association of the enzyme, and that ammonia is also a positive allosteric effector of glutamine-dependent carbamyl phosphate synthetase activity.
The effect of ammonia, which may have metabolic significance, explains the relative insensitivity of the ammonia-dependent carbamyl phosphate synthetase activity to other allosteric effecters.
EXPERIMENTAL PROCEDURE dlaterials-Glutamine-dependent carbamyl phosphate synthetase was purified from E. coli B (Grain Processing Corp.; grown on minimal medium three-fourths of the way through the log phase) by the procedure of Anderson et al. (9). This method of isolation yields a highly purified enzyme which, however, contains at least two minor impurities which migrate between the light and heavy subunits on sodium dodecyl sulfate acrylamide gel electrophoresis.
The enzyme preparation used in sedimentation equilibrium studies was further processed as follows.
The high molecular weight contaminant was removed in a 0 to 527, ammonium sulfate fractionation at 0"; the enzyme was then precipitated by increasing the ammonium sulfate concentration to 65'+& Low molecular weight contaminants were removed by Sephadex G-200 gel filtration in the presence of L-ornithine (150 mM potassium phosphate, 10 mM L-ornithine, 0.5 mM EDTA, pH 7.8).
L-Ornithine, L-glutamine, and the nucleotides were obtained from Sigma. by the use of Dowex 1-X8, as previously described (1, 6, 10). In determinations of the glutamine-dependent carbamyl phosphate synthetase activity carried out in the presence of ammonia, the amount of L-glutamate formed was quantitated. In these studies, the reaction was stopped by addition of 0.1 ml of 1 N hydrochloric acid; after standing at 0" for 10 min, the solution was neutralized by addition of 0.1 ml of 1 M Tris, and the formation of L-glutamate was then determined by use of the glutamate dehydrogenase reaction (11).
Ultracentrifugal Analyses-Sedimentation velocity determinations were carried out on enzyme preparations which were dialyzed at 4' against the appropriate buffer. The sedimentation studies were carried out in a Spinco model E analytical ult,racentrifuge equipped with RTIC unit, electronic speed control, and photoelectric scanning system. Double sector cells of 3.5, 12, and 30 mm path lengths were used and scanning was carried out at 280 nm. Sedimentation coefficients were measured from the rate of movement of the 50% position of apparent boundaries with time, or, when required by the shape of the boundary for associating systems, by calculation of the weight average sedimentation coefficient. High speed sedimentation equilibrium using interference optics  was conducted essentially as described by Yphantis (12). Attainment of equilibrium was ascertained experimentally. All data were analyzed using the computer program of Roark and Yphantis (see Reference 13) to provide smoothed concentrations and various molecular weight moments throughout the cell.

Alternate Conformations
of Carbamyl Phosphate Synthetase Monomer-The sedimentation coefficient of the enzyme in Verona1 buffer (Fig. 1, Curve G) is concentration independent at 7.3 S. The results of sedimentation equilibrium experiments reported below, taken in conjunction with previously reported estimates of the molecular weights of the two constituent polypeptide chains, confirm that this species is monomeric.
In the presence of sodium phosphate (Fig. 1, Curve D), the sedimentation coefficient shows a slight increase with increasing protein concentration, indicating some association. A mathematical analysis of these data as described in the "Appendix" confirms the suspicion that the data do not represent an association of the 7.3 S monomer observed in Veronal.
For this to be true, nearly complete association to a dimer with a sedimentation coefficient of 9.7 S would be required.
Since the monomer appears to be nearly spherical upon examination under the electron microscope,i no reasonable value for the frictional ratio for the dimer can explain such an extreme deviation from the expected sedimentation coefficient for similarly hydrated spheres that have a 2-fold difference in molecular weight (e.g. SZ = S1(22/3)). The only reasonable interpretation of the gentle increase in sedimentation coefficient observed upon increasing protein concentration is that the enzyme has a grossly different conformation in sodium phosphate from that in Veronal. If the monomer in phosphate were greatly expanded, its sedimentation coefficient would be substantially below 7 S and the enzyme would represent a slightly dissociating dimer.
If the enzyme in phosphate were more compact, then the curvature (Fig. 1, Curve D) would represent a slight. tendency toward dimerization. Both alternatives were tested for reasonable least-square fit of the data by approaching the two local minima with bounded variables as discussed in the "Appendix," and, as expected for a curve that is barely rising, both fit the data adequately.
Covalent cross-linking experiments in sodium phosphate show only minor amounts of bands expected for dimers (15,16). In addition, the presence of significant monomer is required to fit the sedimentation equilibrium data in potassium phosphate and ornithine (see below) ; both results exclude the "mostly dime? model. The least-square results show that the monomer has a sedimentation coefficient of 8.7 S and that values for Kt (see "Appendix") and SZ of 0.07 and 13.4 S, respectively, fit the data well.
Both the compact 8.7 S and expanded 7.3 S conformations exhibit self-association in the presence of certain effecters. The association of the 7.3 S conformation promoted by potassium ions (Fig. 1, Curve F) is well described by a monomer-dimer equilibrium.
The concentration span covered is inadequate to evaluate unambiguously the three unknowns, Si, St, and Kz, but if the relation Sp = S1(229 is used, Si is returned as 7.3 S with Kz = 0.2 with a resultant excellent over-all least square fit of the data. The effect of potassium ions on the 8.7 S conformation (Fig. 1, Curve C) is more pronounced, and analysis requires self-association to include polymers beyond dimer for least square agreement within experimental error of the data. Alternate models that fit the data are presented as examples in the "Appendix." Additional confirmation of the proposed change in monomer conformation is that no reasonable associa- Sedimentation velocity studies. A, after dialysis at 4", the enzyme (0.7 mg per ml) was sedimented in 100 mM potassium phosphate, 50 mM Tris-HCl, 100 mM sodium chloride, 0.5 mM EDTA, pH 7.6; B, the enzyme (0.7 mg per ml) was dialyzed for 5 to 6 hr at 4" against 50 mM Tris-HCl, 100 mM sodium chloride, 0.5 mM EDTA (pH 7.6), and then sedimented at 24" in this buffer; C, to 0.475 ml of the enzyme dialyzed as described in B was added 0.025 ml of a solution containing 2 M potassium phosphate and 0.5 mM EDTA (pH 7.6). Photoelectric scans were made at 43, 53, and 49 min, respectively, inA,B,andC. FIG. 3. Effect of potassium phosphate on the sedimentation coefficient. Potassium phosphate concentration was varied at a constant pII (7.8) in a solution that contained enzyme (0.55 mg per ml), 100 mM sodium chloride, and 0.5 mM ETDA.
Sedimentation was carried out at 24" at 52,000 rpm. tion scheme with a monomer sedimentation coefficient of 7.3 S fits these data, whereas values in the vicinity of 8.7 S are the least square choice.
Both the conformational change and the association are reversible.
For example, the same sedimentation coefficient is obtained in potassium phosphate at a given concentration by dilution of a concentrated sample or by concentrating a dilute sample. Fig. 2 demonstrates both effects in that the velocity pattern in potassium phosphate (A, sedimenting at 10.5 S) is transformed to that of the 7.3 S monomer conformation by dialysis against Tris buffer (B). Subsequent addition of potassium phosphate shows a pattern (C) characteristic of the associated 8.7 S monomer and essentially identical to A. Only low concentrations of potassium phosphate are required to produce major changes in the sedimentation coefficient (Fig. 3). The protein concentration was 3.7 mg per ml, speed, 56,000 rpm, temperature, 24"; B, the enzyme (4.2 pg) was assayed by incubation for 10 min at 37" in a solution (final volume, 0.3 ml) containing 5 mM ATP, 5 mM magnesium chloride, 20 mM sodium bicarbonate, 20 mM L-glutamine, 105 mM potassium chloride, and GO IIIM Tris-HCl (pH 7.8).
The amount of glutamate formed was determined.
involves its precipitation by addition of ammonium sulfate. It was observed that when the enzyme was not dialyzed at this stage in order to remove traces of ammonia, unusually high sedimentation coefficients were obtained. This discovery led to a study of the effect of ammonium ion concentration on the sedimentation coefficient.
As shown in Fig. 4A, increasing the concentration of ammonium ions led to an increase in the sedimentation coefficient which reached a maximum at about 14 S at an ammonium ion concentration of 20 InM. Such an increase in sedimentation coefficient resembles the effect produced by L-ornithine, which is a positive allosteric effector (see Fig. 1, Curves A, B, and.E).
Anderson and Marvin (4,7) have observed a similar associating effect of ornithine in a potassium phosphate buffer in studies in which sucrose gradient centrifugation was employed.
As demonstrated in Fig. 1, Curve B, L-ornithine also promotes a self-association of the enzyme in the absence of potassium and phosphate ions. (Two boundaries were observed at a protein concentration of 4.4 mg per ml, whereas at lower protein concentrations, only a single boundary was found as in the other studies described in Fig. 1.) These observations led us to suspect that ammonium ion might also be a positive allosteric effector. The studies described in Fig. 4B indicate that the increase in sedimentation coefficient produced by increasing the concentration of ammonium ions is accompanied by stimulation of enzymatic activity.
The concentration of ammonium ions required for half-maximal effect on the sedimentation coefficient and on the enzymatic activity is about the same. The activation by ammonium ions (Fig. 5) follows a relationship with ATP concentration that is typical of the other allosteric effecters of the enzyme (2), and it thus seems probable that ammonium ion acts by affecting the affinity of the enzyme for ATP.
It is of interest that the ammonia-dependent carbamyl phosphate synthetase activity is not appreciably sensitive to the allosteric activators IMP and ornithine (Fig. 6); in contrast, glutamine-dependent carbamyl phosphate synthetase activity is markedly activated by these effecters in the absence of phosphate (2, 4). On the other hand, UMP produces substantial inhibition of the ammonia-dependent synthetase activity, although this inhibition is significantly less than that exhibited toward the glutamine-dependent carbamyl phosphate synthetase activity (2). These findings seem to reflect the dual function of ammonia as both substrate and allosteric activator. Thus, we may explain the finding that ornithine and IMP fail to produce significant activation of the ammonia-dependent synthetase activity as indicating that ammonia has already interacted with the enzyme so as to promote the conformational state that exhibits increased affinity for .4TP.
However, UMP can apparently antagonize the activating effect of ammonia by favoring a conformational state of the enzyme which binds ATP less strongly.
Addition of ADP and Mg fi leads to a considerable increase in the sedimentation coefficient, and addition of potassium ion (50 mM) in the presence of ADI' and 11g2+ leads to a small but probably significant further increase in sedimentation coefficient (Fig. 7) ion, ADP, ATP, L-ornithine, and ammonia).
The only exception to this generalization so far noted is phosphate ion, which promotes a sizeable increase in sedimentation coefficient (Fig. l), but actually somewhat inhibits glutamine-dependent synthetase activity (Fig. 8). (Magnesium ions do not reverse t.he inhibition found with phosphate.) These dat.a are consistent with the observat,ion noted under "Alternate Conformations of Carbamyl Phosphate Synthetase Monomer," i.e. that the increase in sedimentation coefficient promoted by the addition of phosphate is related to a large conformational alteration.
In distinction, the increase in sedimentation coefficient promoted by those compounds which stimulate enzymatic activity is probably related mainly to a change in the state of association.
Sedimentation Equilibrium Studies-Sedimentation equilibrium experiments on carbamyl phosphate synthetase in the various buffer systems used for the sedimentation velocity studies reported above were generally not possible because the enzyme was not sufficiently stable for the several days required to reach equilibrium.
Although some variation has been noted with different preparations, certain generalizations regarding stability may be made. The enzyme stability is enhanced by increasing the degree of association.
Thus, enzymatic activity in potassium phosphate buffer is reasonably maintained only at high protein concentrations; positive allosteric effecters also stabilize the enzyme.
In addition, the monomeric form of the enzyme in Tris or Verona1 buffer loses its activity and ability to fully reassociate (e.g. when potassium phosphate is added) within a few days, despite the fact that sedimentation velocity behavior in this case is t,ime-independent.
These time-dependent conformational changes prevented sedimentation equilibrium studies except under the maximum associating conditions (in ornithine and potassium phosphate) or on the inactive monomer.
The enzyme exhibits behavior characteristic of a highly homogeneous monomer when examined by the criterion of the  9. Sedimentation equilibrium studies. The apparent weight average molecular weight (Mw) is plotted as a function of protein concentration (fringe displacement in microns). The data were obtained under the following conditions: A, 150 mM potassium (K) phosphate, 0.5 mM EDTA, 10 mM n-ornithine, pH 7.8, 21.9", 12,000 rpm, initial loading concentration 0.25 mg per ml; A, 150 mM potassium phosphate, 0.5 mM EDTA, 10 mM Lornithine, pH 7.8, 8.4", 8,000 rpm, initial loading concentration 1.17 mg per ml; l , 30 mM Veronal-HCl, 0.5 mM EDTA, 100 mM sodium chloride, pH 7.6, 20.4", 19,978 rprn, initial loading concentration 0.25 mg per ml. sedimentation equilibrium carried out in Veronal-sodium chloride buffer (Fig. 9). The weight average molecular weight under t'hese conditions is 163,000 f 4,000 based on a partial specific volume of 0.734 as calculated from the amino acid composition. Additiou of ornithine to t,he enzyme in potassium phosphate buffer shifts the equilibrium strongly towards oligomeric species. At high protein concentration (1.75 mg per ml), the weight average molecular weight (iv,,.) is essentially three times the monomer molecular weight, and the number average molecular weight is less than JI,, which in turn is less than z-average molecular weight.
Thus, the associatiou can only be interpreted by iucluding at least one species with molecular weight correspontling t'o tetramer or higher.
The observation of a masimum sedimentation coefficient corresponding to 15 S (Fig. 1, Curve il) seems somewhat low considering the existence at the higher concentration levels of considerable quantities of trimer and higher polymers. One explanation of this apparent)ly low value of the sedimentation coefhcient may be that there are compensating effects of the association by a relatively strong scdimeiitation coefficient versus c depemleuce of the individual sedimenting species. Analysis of the molecular weight data by a process identical with that described in the "Appendix" shows several ways in which the sediment,ation equilibrium data can be fit. The simplest is a monomer-dimer-trimer-tetramer system with equilibrium constants of Kz = 7, K3 = 520, and K, = 1750. Using these equilibrium coustants and calculating the sedimentation coefficients by the equation Sj = Si(22/3)/(f/fo)j; where (fjjo)j is calculated for liuear polymers with axial ratios of 1, 2, 3, and 4 for monomer, dimer, trimer, and tetramer, respectively, the calculated curve fits the data well. We wish to emphasize that neither the evaluation of K values from the limited sedimentation equilibrium data, nor the derivation of a model for association from the limited curvature present in the sedimentation coefficient versus c data should be taken literally.
The analysis serves 10. Effect of urea and guanidine hydrochloride on the sedimentation coefficient of the enzyme in the presence and absence of potassium (K) phosphate.
Dilutions were made with the same buffers containing either urea or guanidine hydrochloride to yield the concentrations indicated. Sedimentation was performed at 25" at 52,000 rpm.
only to present one model for which the two sets of data would be consistent.
E$ect of [Trea and Guanidine EIydrochloride on Sedimentation Velocity Behavior and on Enzymatic Activity-111 the course of these studies, we examined the effects of urea and guanidine on the sedimentation behavior of the enzyme. As indicated in Fig. 10, relatively low concentrations n!' urea and guanidine hydrochloride produced a dramatic decrease in the sedimentation coefficient when the determinations were carried out in potassium phosphate buffer. When enzyme preparations containing 2 RI urea or 0.5 guanidine hydrochloride were allowed to stand at 26" for 1 hour and then dialyzed against potassium phosphate buffer (150 mM; pI-1 7.8) containing 0.5 mM EDTA, the original sedimentation coefhcients were restored. As indicated in Fig. 10, sedimentation in Veronal-II('1 buffer led t,o a sedimentation coefficient of about 7.3 S over the same range of urea and guanidine hydrochloride coucentrations. Thus, the association of the enzyme which occurs in the presence of potassium phosphate is weakened by urea or guanidine hydro-by guest on March 24, 2020 http://www.jbc.org/ Downloaded from