Glycine Synthase of the Purinolytic Bacterium, Clostridium acidiurici PURIFICATION OF THE GLYCINE-COz EXCHANGE SYSTEM*

When the growth medium of Clostridium acidiurici was supplemented with trace metals, glycine synthase and glycine-COz exchange activities in cell-free extracts were found to increase significantly. The gly- cine-COZ exchange system was purified and shown to consist of a heat-labile component and a heat-stable component. By gel filtration, heat-labile component had an estimated native M, = 230,000 and contained two subunits of M, = 65,000 and 58,000 on sodium dodecyl sulfate-polyacrylamide gels, indicating an ad32 tetramer. Heat-stable component had an estimated M, = 20,000 and could not be replaced by lipoic acid in reaction mixtures. Pyridoxal phosphate was not bound to either of the purified components but was essential for glycine-COe exchange. By spectral analysis, heat-labile component was shown to interact with pyridoxal phosphate and that reductant influenced this interaction.

derived from CO, and the proposed role for glycine synthase.
Glycine synthase, also referred to as the glycine cleavage system, is a multicomponent enzyme system and has been characterized in a number of biological systems, including the aerobic bacterium, Arthrobacter globiformis (13-151, the anaerobic glycine fermenter, Peptococcus glycinophilus (16-191, and vertebrate (20-27) and plant (28-30) mitochondria. These glycine synthase systems are composed of four enzymes: (i) a large molecular weight pyridoxal phosphate-containing glycine decarboxylase, designated P-protein or PI; (ii) a small molecular weight, heat-stable lipoyl protein, designated Hprotein or P,; (iii) a lipoyl dehydrogenase, designated Lprotein or P,; and (iv) a transferase, designated T-protein or P4. (See Refs. 5 ,20, and 27 for reviews.) Collectively, these proteins catalyze the enclosed reaction in Fig. 2. Furthermore, components PI and Pz jointly catalyze the following glycine-CO, exchange reaction: H2N-CH,-CO~H + 14C02 $ H2N-CH2-l4COZH + CO,. This exchange reaction is used to assay for the PI and P, components of glycine synthase.
In these previously characterized procaryotic and eucaryotic glycine synthase systems, the enzyme appears to function primarily in the degradation rather than synthesis of glycine. However, the proposed role for glycine synthase in the purinolytic clostridia is for the formation of glycine. Considering this apparent physiological difference with the previously characterized glycine synthases, we initiated the present study on the glycine synthase system of C. acidiurici.

RESULTS
Trace Metals, Growth, and Enzyme Activities-When the medium described by Rabinowitz (31) was supplemented with the trace metals indicated under "Experimental Procedures," cell yields of C. acidiurici increased from 0.5 g (wet weight)/ liter to 1.0 g/liter. Furthermore, as shown in Table I, the specific activities of glycine synthase and the glycine-C02 exchange reaction in cell-free extracts were also enhanced significantly.
Purification of Glycine-C02 Exchange System-When cellfree extracts were fractionated with ammonium sulfate followed by gel filtration, no single fraction was found to catalyze either glycine synthase or glycine-C02 exchange reactions.

FIG. 2.
Role of glycine synthase (enclosed) in the formation of acetate from COZ. Steps involved are: (i) methylenetetrahydrofolate is synthesized from CO1 by formate dehydrogenase and a reductive tetrahydrofolate pathway; (ii) glycine is synthesized by glycine synthase (enclosed reaction); (iii) glycine is converted to acetate by either glycine reductase @ath A ) or the glycine/serine pathway (path B). Thus, it was concluded that glycine synthase of C. acidiurici was likely a multicomponent system. We elected to purify the components required for glycine-CO, exchange reaction as the first step in characterizing the complete glycine synthase system. As in other glycine synthase systems, two components were required for glycine-COa exchange. One was heat-labile (inactivated after 5 min at 90 "C) and designated heat-labile component, while the other was heat-stable (not inactivated after 5 min at 90 "C) and designated heat-stable component. In these regards, HLC and HSC bear similarity to the Pprotein and H-protein, respectively, of other glycine synthases (14, 16, 21, 22). Utilizing crude HSC to assay for HLC, HLC was purified first; HSC was subsequently purified by assay with purified HLC. The procedures for the purification of HLC and HSC are described under "Experimental Procedures,'' and Table I1 outlines the results obtained from a typical purification.
HLC was observed to be stable throughout purification except that the specific activity of HLC did not increase significantly in the final stage of purification. While the reason for this is not known, oxidation of a redox-sensitive site on HLC may be involved. We have noted that when dithioerythritol is omitted from the buffer used during the Bio-Gel-HT step, activity of the purified HLC drops greater than 90% in 48 h at 4 "C. Since Hiraga and Kikuchi (25) earlier observed inhibition of the mitochondrial glycine-COa exchange system by potassium ions, HLC may also be affected by the high concentration of potassium phosphate used during Bio-Gel-HT chromatography. However, dialysis of purified HLC against 50 mM potassium phosphate, pH 7.0, with 1 mM dithioerythritol failed to significantly enhance activity.
As seen in Fig. 3, the HLC from Bio-Gel HT was electrophoretically pure. In contrast, HLC from Whatman DE52 contained five protein bands on nondenaturing polyacrylamide gels. The possibility that one of these proteins was required for maximal HLC activity was addressed by supplementing reaction mixtures of purified HLC and HSC with the other Bio-Gel HT protein factions; no stimulation was observed. However, we cannot exclude the possibility that an essential factor was at least partially lost from HLC during the final purification step. HSC from Sephadex G-50 contained two major and three minor protein bands on nondenaturing polyacrylamide gels (data not shown), and efforts to further purify HSC to electrophoretic purity were unsuccessful. Neither HLC nor HSC was purified using strict anaerobic procedures as preliminary tests indicated that glycine synthase and the glycine-C02 exchange activity were not significantly affected by aeration provided fresh dithioerythritol was added to all purification buffers. If glycine synthase exists as an enzyme complex (20), fractionation of the complex may render its various parts more labile to aeration. Although the mitochondrial and bacterial glycine synthases previously investigated appear fairly stable during purification, the enzyme of C. acidiurici may be more labile to fractionation. Catalytic Requirements and Molecular Weights of the Purified System-We observed that the ratio of HLC and HSC had dramatic effects on the glycine-C02 exchange activity. While optimum activity was observed with a ratio of approximately 1 unit of HLC to 1 unit of HSC, greatly increasing or decreasing the HLC:HSC ratio had negative effects on the glycine-C02 exchange activity. The pH optimum for glycine-C02 exchange was 7.0, and pyridoxal phosphate was essential to catalysis since a greater than 90% loss in activity was observed when it was deleted from assay mixtures. Deletion of EDTA only resulted in 10% loss in activity and was not considered essential.
With the purified glycine-CO2 exchange systems from A. globiformis (15) and chicken liver mitochondria (241, the heatstable, lipoyl-containing component, H-protein, could be replaced with lipoic acid. However, no activity was observed when 120 pmol of lipoic acid were substituted for HSC. Additionally, although active in glycine-C02 exchange, the purified HLC/HSC system did not contain glycine synthase activity, indicating the requirement for additional components. The native molecular weights of HLC and HSC, as determined by gel filtration, were estimated to be 230,000 and 20,000 respectively (Fig. 4). SDS-polyacrylamide gel electrophoresis of HLC revealed two bands of apparent equal intensity (Fig. 3), the molecular weights of which were estimated as 65,000 and 58,000 (Fig. 4). This indicates that HLC is a tetramer of a& composition.
Spectral Analyses of HLC-Pyridoxal phosphate displays an absorption maximum of 390 nm, and when bound to a protein, this maximum shifts to between 420 and 430 nm (44, 45). In other glycine synthase systems, the presence of bound pyridoxal phosphate to P-protein has been indicated by an absorbance maximum around 428-430 nm (17, 18, 24, 26).
With the P-protein purified from P. glycinophilus, the 430 nm peak increases in absorption when supplemental pyridoxal phosphate is added (17). As shown in Fig. 5, HLC did not with HLC concentrations of 0.8 mg/ml. Solid line, freshly purified HLC; reference cuvette contained equilibrating buffer used with the Bio-Gel HT column. An identical spectrum was obtained after dialyzing HLC 10 h at 4 "C against 50 mM potassium phosphate, pH 7.0, with 1 mM dithioerythritol. Broken line, HLC dialyzed as above in buffer containing 0.1 mM pyridoxal phosphate without dithioerythritol; reference cuvette contained dialysis buffer after dialysis. Dotted line, HLC dialyzed against above buffer system which contained both 1 mM dithioerythritol and 0.1 mM pyridoxal phosphate, reference cuvette contained dialysis buffer after dialysis.
display an absorption maxima in the visible region, indicating that pyridoxal phosphate was not bound to the purified protein. However, when dialyzed against buffer containing pyridoxal phosphate, an absorption peak was observed at 401 nm; this peak shifted to 424 nm when dialysis was against buffer containing both pyridoxal phosphate and dithioerythritol (Fig. 5). This absorption peak is evidence of an interaction of HLC with pyridoxal phosphate, and that reductant (dithio-Glycine Synthuse of C. acidiurici erythritol) influences this apoenzyme/cofactor interaction. Since HLC appeared to decrease in activity in the last purification step, it seemed possible that the loss in activity may have been due to loss of pyridoxal phosphate from the enzyme. However, no pyridoxal phosphate absorbance maximum was observed in the HLC obtained from Whatman DE52. These observations indicate that, upon purification, HLC does not contain bound pyridoxal phosphate as do other glycine synthase P-proteins. Additionally, no absorption maxima were observed in the visible spectrum of HSC.

DISCUSSION
In this study, we have purified from C. ucidiurici the two components of glycine synthase which catalyze the glycine-CO, exchange reaction. This is the first such study with a purine-fermenting organism. The two components, HLC and HSC, were found to have both similar and dissimilar properties to the glycine-CO2 exchange systems previously characterized. The most apparent similarities are that two components are required for glycine-COz exchange, and that one component, HLC, has a high molecular weight and is heatlabile (analogous to P-protein), while the other, HSC, is smaller and heat-stable (analogous to H-protein).
HLC is an a& tetramer of M, = 230,000 and does not contain bound pyridoxal phosphate when purified. Chicken liver mitochondrial P-protein has a M, = 200,000 but is an LY* dimer with each subunit containing equimolar bound pyridoxyl phosphate (24). A similar M, was obtained for P-protein from rat liver mitochondria (23). P-protein from P. glycirwphilus has a native M, = 125,000 and contains 2 mol of pyridoxyl phosphate/m01 of P-protein (18); to the best of our knowledge, its subunit composition has not been reported. H-protein has been characterized from various sources, and the following molecular weights have been observed: human liver, 12,000 (dimer of 23,000) (46); chicken liver mitochondria, 14,500 (24); A. globiformis, 20,000 (14); P. glycinophilus, 12,600 (tetramer of 48,000) (19). Lipoic acid has been found to be a constituent of H-protein and can replace H-protein in vitro (15, 24). Significantly, in a nonketotic hyperglycinemic patient, H-protein was devoid of lipoic acid (46). In our study, HSC had a native M, = 20,000, and could not be replaced by lipoic acid. The nature of this component of the glycine synthase of C. acidiurici will require further purification of the protein. Since HLC and HSC did not reconstitute glycine synthase, the additional proteins and/or cofactors of the enzyme must also be purified for complete characterization of the purinolytic glycine synthase. The reaction catalyzed by glycine synthase is reversible, and the direction of catalysis is apparently characteristic of the biological system from which the enzyme is studied. Synthesis of glycine via glycine synthase in vertebrates is not apparent (20, 27). On the contrary, humans with congenital defects in glycine synthase accumulate glycine in the blood, a condition known as hyperglycinemia, and it has been concluded that glycine synthase is the major pathway for the catabolism of glycine in normal individuals (46-48). In contrast the enzyme in C. acidiurici is thought to be operative in the direction of glycine synthesis and, as illustrated in Figs. 1 and 2, plays a fundamental role in the flow of purine-derived metabolites and ultimate energy balance in the cell. By virtue of its higher M, and tetrameric composition, the purinolytic HLC appears more complex than P-proteins of glycine synthases which function in the degradation rather than synthesis of glycine, and such biochemical differences between glytine synthase systems may reflect the different physiological roles of the enzyme.
Recent studies by Diirre and Andreesen (6,8) suggest that with selenium-grown purinolytic clostridia, acetate is formed mostly by glycine reductase rather than the glycine-serine pathway (Fig. 1). Energetically, glycine reductase may be a more favorable route of acetate synthesis. While this would not alter the role of glycine synthase in the formation of glycine from COZ, it would have marked effects in the labeling pattern of acetate formed from [YH2=]methylenetetrahydrofolate and CO,. If the glycine formed by glycine synthase were converted to acetate by glycine reductase (Fig. 2, path  A), only the methyl carbon of acetate would be labeled by [i4CH~]methylenetetrahydrofolate.
Alternatively, if the glytine-serine pathway was operative (Fig. 2, path B), both carbons of acetate would be labeled. Waber and Wood (12) found that C. acidiurici formed double-labeled acetate from [i4CHF]methylenetetrahydrofolate, thus supporting the glytine-serine pathway as the path of carbon flow in this organism. However, in their study, C. acidiurici was not cultivated on a medium enriched with supplemental trace metals, and this may have favored the glycine-serine pathway. We observed significant increases in cell yield, glycine synthase, and glycine-CO* exchange activity when C. acidiurici was cultivated on a medium enriched in trace metals. Although the path of carbon flow to acetate was not determined, it is obvious that trace metals have profound physiological effects on C. acidiurici.
Indeed, C. ucidiurici and C. cyclindrosporum can be differentiated on the basis of their metal requirements for the formation of formate dehydrogenase (49). Additionally, under conditions of selenium starvation, C. purinolyticum degrades uric acid to acetate, formate, glycine, NH3, and CO, by an as yet unknown pathway which apparently does not involve either the glycine-serine pathway or glycine reductase (50). , G. (1973) Mol. Cell. Biochem. 1 , 169- Unless othewlse noted. a l l e n z m p n p a r a t j o n s wen perfanned a t 4%. Twenty g of c e l l wste was suspended i n 20 ml 50 M potassium phosphate. pH 1.0. containing 1 1M dithioerythrstol and 0.1 9 deaxyritmucleese I. The suspension was passed thmugh a French pmss at 7 WO psi and c a n t r i f u w d a t 27 OOo I 15 1. The supernatant flu14 was bcanted and t h e i e l l e t nrurpenbd i n 10 nl bu;fer ?above) a d centrifuged.