Role of calcium as an inhibitor of rat liver carbamylphosphate synthetase I.

The mechanism of Ca2+ inhibition of carbamylphosphate synthetase I has been investigated using purified enzyme obtained from livers of rats fed a high protein diet. Binding of Mn2+ to the enzyme was measured by EPR techniques at pH 7.8, and Scatchard plots of the data indicated one Mn2+-binding site with a K'd of 13 microM. From competition studies between Mn2+ and Ca2+ or Mg2+ binding, values of 180 microM were obtained for K'd (Mg) and 193 microM for K'd (Ca). A nonlinear least squares curve fitting program was used to calculate the K'm for MgATP2- at the metal-nucleotide binding sites using a simplified rate equation of the enzyme reaction mechanism. Values of 140 and 2420 microM were obtained for K'm (MgATP) at the first and second sites, respectively, at pH 7.8, with a free Mg2+ of 1 mM and other substrates and activators present at saturating concentrations. Variations of the bicarbonate, N-acetylglutamate, and ammonia concentrations in the absence and presence of different amounts of total calcium, from which free Ca2+, free Mg2+, MgATP2-, and CaATP2- concentrations were calculated, permitted values for K'i (CaATP) to be obtained by graphic procedures. Mean values of 375 and 120 microM were obtained for K'i (CaATP) at the first and second sites, respectively. Using the above kinetic constants, a computer model of the enzyme reaction was constructed and tested using two further sets of kinetic data obtained by varying the concentrations of Mg2+, Ca2+, MgATP2-, and CaATP2-. Poor fits were obtained unless the formation of a mixed complex involving CaATP2- competition with MgATP2- at the second metal-nucleotide-binding site was incorporated into the rate equation. Nonlinear least squares curve fitting of both sets of experimental data gave a well determined value of 124 microM for this final CaATP2- inhibitory constant. Sensitivity tests for variation of the primary kinetic constants with the computer model showed that the inhibitory effect of free Ca2+ was weak and that the observed calcium inhibition of carbamylphosphate synthetase can be accounted for primarily by competitive interaction of CaATP2- at the second MgATP2- binding site. With 1 mM free Mg2+ and 5 mM MgATP2-, half-maximal inhibition of enzyme activity was obtained with 0.2 mM CaATP2-.

squares curve fitting program was used to calculate the Kk for MgATP2-at the metal-nucleotide binding sites using a simplified rate equation of the enzyme reaction mechanism. Values of 140 and 2420 p~ were obtained for KL (MgATP) at the first and second sites, respectively, at pH 7.8, with a free Mg2+ of 1 mM and other substrates and activators present at saturating concentrations. Variations of the bicarbonate, N-acetylglutamate, and ammonia concentrations in the absence and presence of different amounts of total calcium, from which free Ca2+, free Mg2+, MgATPZ-, and CaATP2-concentrations were calculated, permitted values for K,! (CaATP) to be obtained by graphic procedures. Mean values of 375 and 120 p~ were obtained for K; (CaATP) at the first and second sites, respectively. Using the above kinetic constants, a computer model of the enzyme reaction was constructed and tested using two further sets of kinetic data obtained by varying the concentrations of Mg", Ca2+, MgATP2-, and CaATP2-. Poor fits were obtained unless the formation of a mixed complex involving CaATP2-competition with MgATPZ-at the second metal-nucleotidebinding site was incorporated into the rate equation. Nonlinear least squares curve fitting of both sets of experimental data gave a well determined value of 124 p~ for this final CaATP2-inhibitory constant. Sensitivity tests for variation of the primary kinetic constants with the computer model showed that the inhibitory effect of free Ca2+ was weak and that the observed calcium inhibition of carbamylphosphate synthetase can be accounted for primarily by competitive interaction of CaATP2-at the second MgATPZ-binding site. With 1 mM free Mg2+ and 5 mM MgATP2-, half-maximal inhibition of enzyme activity was obtained with 0.2 mM CaATP2-.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. It accounts for 20-30% of the protein of rat liver mitochondria (1) and enzyme activity is strictly dependent on the presence of Mi2+ and N-acetylglutamate as obligatory activators (2,3). An understanding of its regulation is important since the formation of carbamylphosphate is generally considered to be the primary site for regulation of urea synthesis (4, 5).
In a previous study, it was shown that addition of Ca2+ to either lysed or intact mitochondria caused an inhibition of carbamylphosphate synthetase activity, which could be reversed by the addition of M e (6). Although it was concluded that Ca?+ interfered with the activation of carbamylphosphate synthetase by M$+, it could not be ascertained whether this effect was exerted by competition of free Ca2+ at the Mg"binding site or by CaATP'-at either or both of the MgATP2-binding sites. The present work was initiated to obtain more information on the mechanism 6f Ca2+ inhibition, using purified rat liver carbamylphosphate synthetase, with the view of assessing the possible physiological significance of the effect. This involved an examination of metal-binding constants at the free Mg2"binding site and the kinetic constants and metal-nucleotide specificity for the two MgATP2--binding sites.
Direct binding studies with the purified enzyme showed that although Ca2+ binding was competitive with M$+ or Mn'+ at the free metal-binding site, the binding constant for Ca2+ was approximately 15-fold higher than the value of 10 PM estimated from previous studies for the free Ca2+ concentration in the matrix space of rat liver mitochondria (7). In contrast, the major inhibitory effect of calcium on the isolated enzyme could be accounted for by CaATP2-competition for MgATP" at the metal-nucleotide-binding sites. Analysis of the nonlinear kinetic plots produced as a consequence of binding of MgATP2-at the two metal-nucleotide binding sites, as well as competitive interaction of CaATP2-at both sites, was greatly aided by the development of a kinetic model, based in part on previous studies of the enzyme mechanism (8)(9)(10)(11). This model incorporated values of the kinetic constants for Ca'+ and M e interaction at the metal-binding site and the apparent K , values for MgATP'-at each of the two I study, and was used to evaluate inhibition of enzyme activity attributable to Cat+ and CaATP" in the presence of saturating concentrations of bicarbonate, ammonia, and N-acetylglutamate. A preliminary account of this work has been published in abstract form (12).

EXPERIMENTAL PROCEDURES
Purification of Carbamylphosphate Synthetase from Rat Liuer-Carbamylphosphate synthetase I used for the metal-binding studies was purified to homogeneity from livers of rats fed a high protein diet as described elsewhere (13). For the kinetic studies, the purification scheme was abbreviated as described by Powers (14). The enzyme was stored a t -70 "C in buffer containing 50 mM HEPES' pH 8.0, 2 M ammonium sulfate, 0.1 M sodium acetate, 10 mM Mg acetate and 2 mM dithioerythritol in 20% glycerol. When assayed in buffer containing 50 mM HEPES, pH 7.8, 50 mM KHC03, 20 mM ammonium acetate, 20 mM Mg acetate, 10 mM ATP, 15 mM Nacetylglutamate, 2 mM dithioerythritol, and 10 pM EGTA, the enzyme had a specific activity of 2-3 pmol of carbamylphosphate/mg of protein. min as previously reported (13). The enzyme preparation was desalted using the technique described by Penefsky (15).
Prior to any binding studies, 1 ml of the enzyme preparation (6 mg/ml) was dialyzed for 2 h against 100 ml of buffer containing 50 mM HEPES, pH 7.8, 0.1 M NaCl, 2 mM dithioerythritol, and 0.5 mM EDTA, followed by three changes with 250 ml of a similar buffer but with no EDTA. The protein solution was then passed through a Sephadex G-25 column (0.5 X 20 cm) pre-equilibrated with the same EDTA-free buffer to remove any remaining EDTA contamination and concentrated using Amicon filters. The dialyzed enzyme preparation retained its original specific activity and contained no detectable Mn", M e , or Ca" when assayed by atomic absorption spectroscopy.
Determination of Carbarnylphosphate Synthetase Actiuity-Enzyme activity was determined by radiometric or colorimetric techniques. The colorimetric method of Ceriotty and Gazzaniga (16) was used as described by Lust,y (13) for the experiments presented in Table I. The radiometric assay was used for the kinetic studies and involved measurement of [I4C]bicarbonate incorporation into citrulline in the presence of excess ornithine and ornithine transcarbamylase. The assay buffer (final volume, 0.5 ml) consisted of 50 mM HEPES, pH 7.8, 50 mM Na acetate, 50 mM KHCOs, Na14C03H (1-2 pCi/ml), 10 mM ammonium acetate, 5 mM ATP, 6 mM Mg acetate, 10 mM hi-acetylglutamate, 2 mM dithioerythritol, 10 p~ EGTA, 10 mM ornithine, 10 units of ornithine transcarbamylase, and 10-30 pg of carbamylphosphate synthetase. The reaction was started by the addition of enzyme and 0.1-ml aliquots were removed after 5, 10, and 15 min and added to 0.05 ml of 30% (w/v) perchloric acid in 1-ml polypropylene tubes. After all the "CO, had evolved, the reaction mixture was transferred to 10-ml plastic scintillation vials and 5 ml of scintillant (ASC 11, Amersham Corp.) was added. The radioactivity was determined using an Intertechnique liquid scintillation counter (In/Us Service Corp.). Under these conditions, the reaction was found to be linear for at least 20 min. Small corrections to the linear reaction rates were made for blanks in which either carbamylphosphate synthetase or N-acetylglutamate were omitted. When the bicarbonate concentration was varied, the amount of sodium acetate in the incubation medium was modified to maintain a constant ionic strength.
Metal-binding Studies-Binding of Mn2+ to rat liver carbamylphosphate synthetase I was determined as described by Cohn and Townsend (17) using a Varian EPR-E12 spectrometer operating a t 9 GHz. Binding titrations were performed at 37 "C in 0.05-ml final volume of buffer containing 50 mM HEPES, pH 7.8, 100 mM NaC1, and 2 mM dithioerythritol a t a fixed enzyme concentration in the range from 30 to 60 p~ with the Mn2+ concentration varied to provide from 20 to 80% saturation of the enzyme. The protein concentration was calculated from its absorbance a t 280 nm (13). The number and Kd value of the Mn2+-binding sites were determined from linear regression analysis of Scatchard plots. The dissociation constants for Preparation of ATPPS Diasteroisomers-The two diasteroisomers of ATPBS were prepared enzymatically from ADPBS essentially as described by Ekstein and Goody (19) with the modifications as outlined by Jaffe and Cohn (20). Both isomer stock solutions were extensively extracted with 8-hydroxyquinoline in chloroform to prevent metal contamination.
Calculation of Free and A TP-liganded Metal Concentrations-Since the kinetic experiments required the addition of different mixtures of Mg", Mn", or Ca2+ and ATP, it was necessary to calculate the concentrations of free metal and metal-ATP complexes present in the incubation medium. This problem was considerably simplified since experiments were carried out at pH 7.8 and the contributions of the protonated forms of ATP" both as significant metal-binding contributors and as kinetically competent species can safely be neglected (21). Calculations were performed using an iterative program to solve equations for the binding of two metals with one ligand (22). Kinetic Model of Calcium Inhibition of Carbamylphosphate Synthetase Actiuity-The model used was based on the ordered reaction sequence for metal and metal-nucleotide binding to carbamylphosphate synthetase from previous kinetic studies of Elliott and Tipton (8,9), the pulse-chase experiments of Rubio et al. (10) and Britton et al. (11), and the equilibrium and kinetic studies of Fahien and Cohen (25) and Fahien et al. (26). The model accounts for the presence of one metal site and two separate metal-nucleotide sites and assumes. on the basis of the above studies, that binding of the metal and metalnucleotides occurs in rapid equilibrium so that their apparent K,,, values are numerically equal to their respective dissociation constants. The multiple equilibria that describe the reaction sequence in the presence of Ca2+ are depicted in Fig. 1.
The model describes the situation in which free Ca2+ competes with free Mg2+ for the metal binding site and CaATP2-competes with MgATP2-for its binding at either or both metal-nucleotide sites.
It is assumed that whenever Ca2+ binds to the metal site it produces an abortive E.Ca complex which precludes any further binding of substrates and activators. Additionally, CaATP2-can bind to one or both metal-nucleotide sites to form the respective dead-end complexes. KO and Ki represent the dissociation constants for Mg2+ and Ca2+, respectively. In the absence of Ca'+, and with saturating concentrations of bicabonate, ammonia, and N-acetylglutamate, the first molecule of MgATP2-binds to the E . Mg complex with a dissociation constant termed K,, while the second molecule of MgATP2-binds a t the second metal-nucleotide site with a dissociation constant of aK,, so that (Y represents the ratio of dissociation constants for MgATP2at the two sites. When calcium is added, CaATP2-can bind to either or both metalnucleotide sites to produce in each case a dead-end complex. K-, refers to the dissociation constant for binding of CaATP2-at site 1, and 8K,, is the dissociation constant for CaATP2-binding at site 2. Thus, p represents the ratio of the dissociation constant for CaATP2-a t site 2 relative to that at site 1, when both are occupied by CaATP2-. In addition to the homologous complexes which contain the same metal nucleotide species a t both sites, the possibility of formation of heterologous complexes containing both MgATPZ-and CaATP2-a t either of the two sites must also be considered. CaATPZ-might bind to site 2 when site 1 is occupied by MgATP2-with a dissociation constant defined as yK,,,, or MgATP2-might bind to site 2 when site The rate equation that describes this multiple equilibrium: derived as described by Segel (27), is: In the absence of Ca", Equation 2 reduces to: Computer Simulations and Statistical Fittings-A Fortran program was developed which allowed the simulation of titration curves according to Equation 2 using a PDP-11/23 computer interfaced with a matrix printer and a Tektronix graphics plotter. The program calculated the distribution of free and nucleotide-bound metal species from specified totals and solved for Equation 2 or its reciprocal using a predetermined set of kinetic parameters. The simulation was performed by mathematically incrementing total calcium concentrations at constant total ATP and Mgz+ concentrations, and the results were expressed in the form of either Dixon or double reciprocal plots. A nonlinear least squares regression analysis of Equation 3 was carried out using the BMDPAR routine (a derivative-free nonlinear least squares regression program) from the BMDP statistical package of the University of California as modified for the DEC-10 computer.
Materials-Sodium ['%]bicarbonate (specific activity, 56.1 mCi/ mmol) was obtained from Amersham Corporation. Bio-Gel A-0.5m and Chelex 100 were obtained from Bio-Rad Laboratories. Ultrogel AcA34 was purchased from LKB. Quaternary Sephadex A-50, Sephadex G-25, and Sephadex G50 were obtained from Pharmacia Fine Chemicals. Ammonium sulfate was obtained from Schwarz/Mann. Ornithine transcarbamylase (from Streptococcus faecalis), ATP, ornithine, and dithioerythritol were purchased from Sigma Chemical Company. ADPOS, pyruvate kinase, and lactate dehydrogenase were purchased from Boehringer Mannheim. Amicon filters (A-25) were obtained from Amicon Corporation (Danvers, MA). Metals were * The derivation of the rate equation assumes that the free enzyme form contains no bound metal and that the concentrations of metals and substrates greatly exceeded the concentration of enzyme. obtained as acetate salts from either Merck Suprapur (Mn, Zn, Mg) or Aldrich Gold Label (Cd, Zn, Ca). All other reagents were of the highest purity available commercially.

Calcium Inhibition
of Carbamylphosphate Synthetase-When carbamylphosphate synthetase is assayed under standard conditions with saturating concentrations of bicarbonate, ammonia, and N-acetylglutamate in buffer containing 6 mM Mg" and 5 mM ATP the activity of the enzyme is inhibited by the addition of Ca2+. Fig. 2. 4 (solid line) shows that the enzyme is inhibited by about 90% when the total calcium concentration reaches 5 mM. From the Dixon plot of these data ( Fig. 2 A , inset) a n apparent K, for total calcium of 0.7 mM is obtained. This value was relatively independent of ionic strength but was highly dependent on the concentrations of total Mg2+ and ATP (data not shown). Calcium competes with M$+ for chelation with ATP, and evaluation of the mechanism by which Ca2+ inhibits carbamylphosphate synthetase activity requires a knowledge of the kinetic constants for interactions at both the free metalbinding site and the metal-nucleotide-binding sites.
Competition of Ca" at the Metal-binding Site-Equilibrium binding studies of Mn2+ to enzymes, as monitored by EPR techniques, have been widely used to determine the number and affinity of divalent metal-binding sites to a variety of enzymes, including Escherichiu coli carbamylphosphate synt,hetase (17, 28-30). Fig. 3 (curue A) shows a Scatchard plot of Mn2+ binding to highly purified rat liver carbamylphosphate synthetase. Extrapolation of the line to the abscissa confirms that there is only one free metal-binding site, while the reciprocal of the slope provides a value of 13 p~ for the apparent Kd for Mn'+ binding. When the Mn"-binding titration was performed in the presence of 0.5 mM Ca'+, the Scatchard plot deviated from linearity a t low ratios of bound Mn2+ (Fig. 3,  This finding strongly suggests that both the kinetic and equilibrium binding measurements refer to the same site and that the free metal is in rapid equilibrium with the metalbinding site. Therefore, the K,' for free Ca2+ inhibition a t this site is expected to be equal to the K$ for Ca" binding as determined above, namely about 190 gM.
It is evident from the above considerations that, for the buffer system used for the standard assay conditions, the KL :' S. Cerdan and C. J. Lusty, unpublished observations. for free Mg'+ is approximately equal to the K: for free Ca'+, both being in the range 170-200 p~. Using values for these parameters calculated from the metal-binding studies, and assuming saturating concentrations of all other reactants and activators, the reaction velocity can be calculated on the basis of a single site competition between Ca'+ and Mg2+ using the kinetic equation for simple competitive inhibition. Values for free Ca'+ and free Mg'+ were taken from the curves presented in Fig. 2B, and the calculated enzyme velocity is shown by the dotted line in Fig, 2A. The inhibition attributed to Ca2+ inhibition calculated on this basis was considerably less than the observed inhibition of enzyme activity. A parallel series of experiments using Mn2+ as the activator metal were also performed with similar results. It is clear from these results that the overall inhibition of carbamylphosphate synthetase by calcium cannot be accounted for solely in terms of free Ca'+ competition with the metal ion activator at the metalbinding site of the enzyme.
Specificity and Affinity of the Metal-Nucleotide-binding Sites-In order to investigate the specificity of rat liver carbamylphosphate synthetase for different metal nucleotides, titrations of enzyme activity were performed at pH 7.4 using saturating concentrations of bicarbonate, ammonia, and Nacetylglutamate, with varied metal nucleotide concentrations in the range from 0.1 to 10 mM using either Mg", Mn", Co2+, Zn'+, Cd", or Ca" as the metal ion. The free metal concentration was kept constant during the titration at 10 mM for Mg2+, 5 mM for Co'+, 0.2 mM for Mn2+, and 0.1 mM for zn2+ and Cd". Concentrations of free Mn" higher than about 0.2 mM were found to be inhibitory. The two diasteroisomers of ATPpS (19, 20, 31) have also been used in an attempt to probe the stereospecificity of the metal nucleotide complex. Table I summarizes the results of these experiments using Mgr+, M d + , or Co2+ as the metal ion. No detectable enzyme activity could be obtained with Ca", Zn2+, or Cd2+ as the only metal ion present. These results are in agreement with a previous report on the inhibitory effects of certain heavy metal ions on rat liver carbamylphosphate synthetase I activity (14). Table I    It is evident, therefore, that substitution of the metal ion or substitution of ATP with ATPPS greatly alters the activity of the enzyme, but has a small effect on the apparent binding affinity of the metal-nucleotide to the enzyme. The loss of stereospecificity observed with MnATPPS and CoATPPS can be explained by the known ability of Mn2+ and Co'+ to bind to both oxygen and sulfur while Mg'+ binds preferentially to oxygen (20,32).
In order to distinguish between interactions of metal-nucleotides at the two different metal-nucleotide-binding sites, a different kinetic approach had to be used. It is known from earlier studies that the overall apparent K,,, of carbamylphosphate synthetase for MgATP2-is decreased with an increase of the free Mg2+ concentration (8,13,33). Consequently, the effect of varying the MgATP" concentration on carbamylphosphate synthetase activity was studied in more detail under conditions when the free MgZ' concentration was maintained constant a t a value of 1 mM, which is thought to represent the free Mg'+ concentration in the matrix of normal rat liver mitochondria (34,35). This value is five times higher than the apparent K, for MgL+ at the metal-binding site, so that formation of the E.Mg complex (see Fig. 1) should not be rate-limiting. Under these conditions, with saturating concentrations of bicarbonate, ammonia, and N-acetylglutamate, double reciprocal plots of initial rate versus MgATP'-over the range from 0.2 to 10 mM MgATP" were nonlinear (Fig.  4). Extrapolation of the curve to the abscissa provided a value of 2.5 mM for the K& of MgATP2-. However, since carbamylphosphate synthetase contains two metal-nucleotide-binding sites, the actual meaning of this value is unclear in relation to the relative affinities of MgATP2-at the two sites. This problem was approached by curve fitting the data to the rate Equation 3 which defines the K& for the first MgATP" binding site as K, and the second as aK,. Iterative nonlinear least squares fitting gave the curve shown in Fig. 4 and provided values for K, of 0.14 f 0.01 mM and 17.3 f 1.1 for CY, indicating that the first molecule of MgATP" is bound to the enzyme with considerably higher affinity than the second.
The KA for MgATP" at the second site is thus 2.4 mM, which is similar to the single value obtained by extrapolating the linear portion of the double reciprocal plot of Fig. 4 to the abscissa. The present finding that the first metal-nucleotide site has a higher affinity for MgATP2-than the second site is in agreement with previous observations with the frog liver enzyme (25, 26). These authors reported values of 0.05 and 0.5 mM for the apparent dissociation constants of MgATP" at the first and second sites, respectively. However, our findings contrast with those of Rubio et al. (10,11) who obtained the reverse relative affinities (0.2 and 0.01 mM) using pulsechase experiments with rat liver carbamylphosphate synthetase. Higher values in the region of 1-2 mM are normally reported for the apparent K , or rat liver carbamylphosphate synthetase for MgATP2-with both the isolated enzyme and rat liver mitochondria (13, 33, 36), which can now be interpreted as reflecting mainly interaction of MgATP2-at the second metal-nucleotide-binding site.
Competition of CaATP" at the Metal-Nucleotide-binding Sites--In order to define the inhibitory effects of CaATP" on carbamylphosphate synthetase activity, it was necessary to design kinetic experiments that would allow a discrimination of CaATP" inhibition between the two MgATP2-substrate sites so that the respective CaATP'-inhibitory constants could be calculated. Fig. 5 depicts the binding sequence of substrates and activators suggested by Elliott and Tipton (8) from kinetic studies with bovine liver carbamylphosphate synthetase. Mg2+ is bound first, followed by a random order addition of MgATP" (at site 1) or N-acetylglutamate. This is followed by an ordered addition of bicarbonate, MgATP" (at site 2), and finally ammonia. Since the affinities of the I enzyme for binding of MgATP2-(at site 1) and N-acetylglutamate are similar, either the upper or lower branch of the reaction pathway is expected to be favored by the relative concentrations of N-acetylglutamate or MgATP2-. If the lower branch of the reaction pathway is followed as suggested from other work (37), MgATP2-binds to the first metalnucleotide site before N-acetylglutamate, followed by bicarbonate, the second molecule of MgATP2-, and ammonia. Accordingly, it should be possible to monitor the CaATP2inhibition at each of the metal-nucleotide sites by selecting the appropriate conditions (38). Thus, using saturating concentrations of bicarbonate and ammonia and a high concentration of MgATP2-relative to N-acetylglutamate, titrations of N-acetylglutamate concentration a t a series of fixed calcium concentrations should allow CaATP2-inhibition to be monitored at the first metal-nucleotide site. Similarly, using saturating concentrations of N-acetylglutamate and ammonia and limiting concentrations of bicarbonate in the absence and presence of different calcium concentrations should also monitor CaATP" inhibition at the first site. Finally, using saturating concentrations of bicarbonate and N-acetylglutamate, and limiting ammonia in the absence and presence of calcium should allow an analysis of CaATP2-inhibition at the second metal-nucleotide site. Fig. 6 shows the results of titrations of carbamylphosphate synthetase activity with variation of the N-acetylglutamate Concentration in the range from 0.05 or 1 mM total calcium. The experiment was conducted using 50 mM bicarbonate, 10 mM ammonia, 6 mM total M e , and 5 mM total ATP. The Lineweaver-Burk plot produced straight lines, which intersected at a single point on the abscissa, indicating a noncompetitive inhibition pattern with an apparent K , for N-acetylglutamate of 114 PM. This is similar to the values obtained by other workers (13, 25). Furthermore, it is apparent that Ca2+ or CaATP2-do not affect the apparent K , for Nacetylglutamate. The concentration of CaATP2-in the reaction mixture after addition of either 0.5 or 1 mM total calcium was calculated as in Fig. 2, and replots of the intercepts or the slopes of the lines of Fig. 6 versus [CaATP2-] are shown as insets. Both replots intersected the abscissa at a value equivalent to 0.35 mM CaATP2-, which may be interpreted as an apparent K, for CaATPz-at the first metal nucleotide-binding site. This value was confirmed in further experiments (data not shown) in which seven different calcium concentrations up to 5 mM were used, each with N-acetylglutamate concen- trations of 0.1,0.5, 1, or 5 mM. A Dixon plot of the data uersw total calcium gave straight lines, except with the lowest Nacetylglutamate concentration, which intersected the Ca-ATP2-abscissa at a value of 0.35 mM. These data show that the K: for CaATP2-at the first site is independent of the Nacetylglutamate concentration. The relative concentrations of free Mg'+ and free Ca2+ were such that the inhibitory effect of Caz+ at the metal-binding site was negligible. Fig. 7 shows a Lineweaver-Burk plot with variation of the bicarbonate concentration from 1 to 20 mM with no calcium added and with 0.5 and 1 mM total calcium in the presence of 10 mM N-acetylglutamate, 5 mM total ATP, and 6 mM total Mg+. As in Fig. 6, the inhibition pattern caused by calcium was noncompetitive, and the apparent K , for bicarbonate was 4 mM (cf. Refs. 8 and 13). Replots of the intercepts and the slope of the lines (see insets against [CaATP2-]) were linear, and gave a value of 0.4 mM for the apparent K, for CaATP2inhibition at the first metal-nucleotide site. Thus, irrespective of whether the reaction pathway for mammalian carbamylphosphate synthetase I follows the upper or lower branch of In fact, the observed kinetic behavior favors a reaction mechanism consistent with the lower branch in which N-acetylglutamate binds after the first molecule of MgATP" (37).
In contrast, when the ammonia concentration was varied under the standard assay conditions with 50 mM bicarbonate, 10 mM N-acetylglutamate, 6 mM total Mg", and 5 mM total ATP, apparently parallel lines were produced with Lineweaver-Burk plots at total calcium concentrations of 0, 0.5, and 1 mM (Fig. 8). However, linear regression analysis showed that they were in fact weakly convergent. The apparent K , for ammonia was 1.7 mM (cf. Refs. 8 and 13). Earlier studies by Elliott and Tipton (8), which included ammonia titrations over a range of MgATP" concentrations, showed that at low concentrations of MgATP2-(below 0.5 mM) the lines became convergent. These authors concluded, therefore, that ammonia binds to the enzyme after the second molecule of Mg-ATP2-. The inset of Fig. 8 shows a replot of the intercept of the lines on the ordinate versus the calculated CaATP2concentrations after addition of 0.5 and 1 mM total calcium. The points on the replot fell on a straight line which intersected the abscissa to give an apparent K , for CaATP2of 0.12 mM. This value may be interpreted as representing the inhib- itory interaction of CaATP" at the second metal-nucleotidebinding site.
In summary, and with reference to Fig. 1, the kinetic experiments described above using CaATP2-as an inhibitor of carbamylphosphate synthetase have allowed an assessment of the relative inhibitory constants at the two metal-nucleotide-binding sites. At the first MgATP"-binding site, the ratio of the apparent K,,, for MgATP2-to the apparent K, for CaATP" (Kn/KnJ is 0.37, while at the second MgATP2binding site the ratio of the constants ((uKn/pKn,) is 19.5.
Thus, the overall inhibitory effect of CaATP" on the reaction velocity is expected to be exerted mainly at the second metalnucleotide site.
Kinetic Model to Describe Calcium Inhibition of Carbamylphosphate Synthetase-From kinetic and binding studies with purified rat liver carbamylphosphate synthetase I, values have been obtained for the primary kinetic constants as follows: Kil for free M$' ( K O ) , K: for CaATP2-a t site 2 (PK,,,), 124 @I. Initially, a computer program was developed to fit the experimental data employing the above kinetic constants and a modification of Equation 2 in which the formation of mixed complexes (E.Mg.Mg-ATP2-.CaATP") were neglected. The simulated curve for data such as that of Fig. 2 gave a very poor fit, similar to the dotted line shown in Fig. 2 A . It became obvious, therefore, that formation of mixed metal-nucleotide abortive complexes needed to be incorporated into the kinetic model. This required knowledge of two additional kinetic constants, namely yKn, and 6K, (see Fig. 1).
These constants cannot readily be obtained from primary or secondary kinetic plots; hence, an iterative curve-fitting procedure was used to estimate the remaining unknown parameters. For this purpose, two additional sets of experimental data were used. In the first, the buffer contained 50 mM bicarbonate, 10 mM ammonia, 10 mM N-acetylglutamate as standard additions and four different concentrations of total ATP were used as follows: 0.5, 1, 2, and 5 mM. In each case, total Mg'+ was added to give 1 mM excess free Mg'+ concentration. To each set, different amounts of total calcium were added up to 5 mM (cf. Fig. 2). These conditions were chosen to provide a balanced distribution of data points for a Dixon plot of the reciprocal of velocity uersus the calculated Ca-ATP" values for each of the four sets of MgATPZ-concentrations. In the second experiment, similar conditions were used except that the MgATP2-concentration was varied with a constant free Mg'+ concentration of 1 mM (cf. Fig. .4), and four sets of curves were generated by addition of 0, 0.5, 1, and 2 mM total calcium. These conditions allowed a balanced distribution of points for a double reciprocal plot of velocity uersus MgATP". In order to simplify the nonlinear least squares curve-fitting procedure using the BMDP program, the assumption was initially made that mixed complexes were formed primarily by addition of CaATPY-to the second metalnucleotide site when the first site was occupied by MgATP2-(ie. that 6 >> y; cf. Fig. 1). Consequently, the data from both sets of experiments were used to optimize for y using a modification of rate Equation 2.
The Vn,ax of the enzyme could not be precisely determined graphically from the Dixon plot data; hence, y was calculated for different values of V,,,, until a minimum value was obtained for the sum of the least squares. In addition, 3 of the 28 data points gave very high nonrandom residuals and were eliminated from the analysis. The value for y was then very was used to determine 6. The BMDP program gave large and very poorly determined values for 6; hence, it was assumed that the contribution of the E.Mg.CaATP, .MgATP2 complex (see Fig. 1) to the overall reaction velocity was negligible. Consequently, the kinetic constant for formation of mixed metal-nucleotide complexes (yK,,,) was considered to be 116 PM. Fig. 9 shows simulated plots of the first experiment referred to above expressed in the form of velocity uersus [CaATP2-] (Fig. 9A) or reciprocal velocity uersus [CaATP"] (Fig. 9B), using a modification of rate Equation 2 with the term including 6 eliminated and values for the kinetic constants determined as described above. Fig. 10 shows similar simulations of the second experiment using the same kinetic constants, except for V,,,,,, with the data expressed in the form of velocity uersus [MgATP"]. An excellent simulation of the nonlinear plots is obtained by the kinetic model, indicating that it provides a valid representation of the kinetic mechanisms for calcium inhibition of carbamylphosphate synthetase I.
In summary, the major effect of calcium in regulating the enzyme activity is by CaATP2-inhibition primarily at the second metal-nucleotide-binding site. Under conditions thought to mimic the ionic environment of the matrix space the inhibition by free Ca'+ at the metal binding site is relatively weak. No evidence was obtained for Ca2+ or CaATP2-interaction at the bicarbonate-, ammonia-, or N-acetylglutamatebinding sites.

DISCUSSION
A number of enzymes involved in phosphoryl transfer are known to be inhibited by calcium (39). Among those that have been extensively studied are pyruvate kinase (40), pyruvate carboxylase (41, 42), and bacterial carbamylphosphate synthetase (30). The present study shows that binding constants of free Mg'+ and free Ca2+ to the rat liver enzyme are considerably smaller than those for the bacterial enzyme (30), which is also inhibited by Ca2+ and shares the same stereospecificity for the A diastereoisomer of ATPpS (31,43). Since most enzymes that have a MgATP2--binding site also have an essential requirement for a free metal (usually Mg2+ or Mn"), the question arises whether an observed inhibition of enzyme activity by calcium is caused by interaction of free Ca'+ at the metal-binding site or by interaction of CaATP2at the metal-nucleotide-binding site. Despite the importance of the question, very few studies have addressed this issue, with the exception of a kinetic study of inorganic pyrophosphatase (44).
Despite extensive kinetic studies with mammalian carbamylphosphate synthetase (8,9,33), the nonlinear nature of kinetic plots involving variations of MgATP2-concentration has made it difficult to obtain Michaelis constants by purely graphic procedures. However, from a knowledge of the kinetic mechanism, it is possible to generate a simplified rate equation by using concentrations of reactants and activators, other than MgATP", well above their respective Michaelis constants. This rate equation can then be used to obtain solutions for the unknown kinetic constants by nonlinear least squares curve fitting of the experimental data in order to optimize for the apparent K,,, values of the two MgATP2-sites. Using this procedure in the present study, the results clearly show that the affinity of the first metal-nucleotide-binding site for MgATP2-is much greater than that for the second site. The apparent K,,, for addition of the second molecule of MgATP2to the enzyme in fact is similar to values for the overall KA for MgATP" reported by other workers using isolated car-bamylphosphate synthetase (13), toluene-permeabilized rat liver mitochondria (45), or intact mitochondria (46). The fact that the affinity of the mammalian enzyme for Mg2+ is high relative to that for the second molecule of MgATP2-, and the finding that the ratio of KA (Mg)/K: (Ca) is about unity compared with a value of about 20 for the ratio of KA (MgATP)/K, (CaATP) at the second metal-nucleotide-binding site (aK,/yK,; in Fig. l ) , largely accounts for the observation that the direct contribution of free Ca2+ to overall inhibition of enzyme activity is relatively small. Thus, the interpretation of kinetic studies based on variations of Mg-ATP2-and CaATP2-concentrations is aided by use of free Mg" concentrations higher than the KA (Mg) range, so that despite inevitable variations of free Mg2+ or free Ca2+ concentrations due to their different binding constants to ATP, inhibitory interactions at the metal-binding site contribute little to the observed changes of enzyme reaction rate.
Addition of calcium to purified carbamylphosphate synthetase has, in fact, proved to be a useful inhibitory probe to extend previous studies related to the enzyme mechanism. Separate titrations of N-acetylglutamate, bicarbonate, and ammonia concentrations in the absence and presence of different total calcium concentrations gave linear kinetic plots consistent with binding of the first molecule of MgATP2prior to addition of N-acetylglutamate and bicarbonate (37), and binding of the second molecule of MgATP2-prior to the addition of ammonia (8). These experiments also showed that Caz+ or CaATP" did not affect the apparent K , for Nacetylglutamate (e/ Ref. 45), and allowed a quantitative evaluation of the inhibitory interactions of CaATP2at both metal-nucleotide-binding sites. In addition, the present study provides excellent circumstantial evidence for the formation of mixed MgATP .CaATP complexes and an evaluation of their importance in accounting for the observed inhibition of enzyme activity by calcium. The value of combining purely kinetic studies with nonlinear curve fitting to computer models using different forms of the rate equation to describe the reaction velocity, when marked deviations of the kinetic plots from linearity are obtained, is well illustrated by the present study. Computer simulations are not able to prove reaction mechanisms except in exceptional cases, but they are useful in delineating a minimal mechanism consistent with the experimental data. The fact that in the present instance the model could provide fits for two independent series of kinetic experiments with the same kinetic constants gives an added measure of confidence to the usefulness of the model, and may allow it to be utilized to study the regulation of carbamylphosphate synthetase activity in isolated mitochondria.
The experimental conditions chosen to evaluate the kinetic model of rat liver carbamylphosphate synthetase were designed to resemble the ionic conditions thought to be present in rat liver mitochondria, where rates of citrulline production have been studied as a function of the mitochondrial energy state or calcium content (6,35,36,47). Thus, the free M$+ concentration was maintained at about 1 mM and the Mg-ATP2-concentration was varied up to 10 mM. Rat liver mitochondria as normally prepared in the absence of excess EGTA in the isolation medium contain 10-15 nmol of calcium/mg of protein (48), and the mitochondrial calcium content in the intact liver appears to be in the same range (49). Approximately 99.9% of this calcium is bound, and the free Ca2+ concentration has been estimated to be in the region of 10 pM (7). The nature of the calcium-binding ligands has not been ascertained, although inorganic phosphate (501, phospholipid head groups (7) and proteins are reasonable candidates. Carbamylphosphate synthetase itself is present in liver mitochondria a t a concentration of about 1 mM, and hence will contribute significantly to the overall binding sites for both metals and metal-nucleotides in the mitochondrial matrix. The possibility of significant concentrations of CaATP2in the mitochondrial matrix has not been considered in most studies concerning metabolic regulation. However, the present work with purified carbamylphosphate synthetase suggests that the previously observed inhibition of citrulline production in isolated rat liver mitochondria with an increase of total calcium content (6) can best be explained on the basis of an inhibition of carbamylphosphate synthetase by Ca-ATP", competitive with MgATP2-. If the apparent kinetic constants measured for the isolated enzyme are applicable to the situation present in intact mitochondria, an intramitochondrial CaATP2-concentration of about 0.2 mM would be required to give a 50% inhibition of citrulline production under optimal assay conditions similar to those used in previous work (6,35). However, further knowledge of the number and dissociation constants of the Ca2+-and Mg2c-binding sites in the mitochondria is needed before more definitive conclusions can be reached concerning the physiological regulation of intramitochondrial enzymes by CaATP2-.
One of the expectations on embarking on the present study (see Ref. 51) was that carbamylphosphate synthetase would prove to be a useful Ca2+-sensitive indicator enzyme more in the range of the observed values for the matrix-free Ca" concentration of rat liver mitochondria than other mitochondrial Ca"-sensitive enzymes such as isocitrate dehydrogenase, a-ketoglutarate dehydrogenase, and pyruvate dehydrogenase phosphatase, which exhibit half-maximal stimulation by Ca2+ at about 1 p~ (52). However, although carbamylphosphate synthetase appears to have little usefulness as an enzyme indicator of the mitochondrial free Ca2+ concentration, it is possible that measurements of citrulline production by rat liver mitochondria under defined conditions may be used as an intramitochondrial CaATP2-indicator.