Decarboxylation of glycine by serine hydroxymethyltransferase in the presence of lipoic acid.

Serine hydroxymethyltransferase and the glycine cleavage system are both present in liver mitochondria and both bind glycine to form a pyridoxal 5'-phosphate carbanionic quinoid species. Lipoic acid has been shown to have the ability to intercept the carbanionic intermediate formed from the binary complex of serine hydroxymethyltransferase and glycine and form an intermediate adduct which is ultimately processed to yield CO2 and a methylamine adduct. Kinetic studies have shown that the lipoic acid-dependent decarboxylation of glycine catalyzed by serine hydroxymethyltransferase proceeds through a sequential mechanism. This lipoic acid-dependent decarboxylation catalyzed by serine hydroxymethyltransferase is similar to the initial reaction of the glycine cleavage system and to the lipoic acid-dependent decarboxylation of glycine by the P-protein alone suggesting that both enzymes could serve in lieu of each other.

Serine hydroxymethyltransferase and the glycine cleavage system are both present in liver mitochondria and both bind glycine to form a pyridoxal 5"phosphate carbanionic quinoid species. Lipoic acid has been shown to have the ability to intercept the carbanionic intermediate formed from the binary complex of serine hydroxymethyltransferase and glycine and form an intermediate adduct which is ultimately processed to yield COz and a methylamine adduct. Kinetic studies have shown that the lipoic acid-dependent decarboxylation of glycine catalyzed by serine hydroxymethyltransferase proceeds through a sequential mechanism.
This lipoic acid-dependent decarboxylation catalyzed by serine hydroxymethyltransferase is similar to the initial reaction of the glycine cleavage system and to the lipoic acid-dependent decarboxylation of glycine by the P-protein alone suggesting that both enzymes could serve in lieu of each other. ~~~~~~~~~~~ Serine hydroxymethyltransferase (EC 2.1.2.1) is unique among pyridoxal 5"phosphate-dependent enzymes in that it is able to carry out a large number of amino acid transformations catalyzed by other Bs enzymes, e.g. Reactions a-c.
' P and H in the circles represent the P-protein containing pyridoxal 5'-phosphate and the H-protein containing lipoic acid, respectively.
It has been demonstrated that serine hydroxymethyltransferase can bind glycine and catalyze the exchange of the pro-S-a-proton similar to that found with the P-protein (Schirch and Jenkins, 1964). It has also been shown that lipoic acid can replace the H-protein as an acceptor of the carbanion (Hiraga and Kikuchi, 1980). Since serine hydroxymethyltransferase is capable of catalyzing amino acid transformations of various other Bs enzymes, we postulated that serine hydroxymethyltransferase, when in the presence of lipoic acid, could catalyze the decarboxylation of glycine to mimic the glycine cleavage system.
Results presented in this text suggest that serine hydroxymethyltransferase and P-protein both can account for the decarboxylation of glycine in mitochondria. Having this activity associated with serine hydroxymethyltransferase makes it conceivable that the complete degradation of serine to 2 mol of N5,N'o-methylenetetrahydrofolic acid and 1 mol each of C 0 2 and NHe could be initiated by a single mitochondrial pyridoxal 5"phosphate enzyme. lr+i at concentratlonr Indicated below, pH 7.5. 1nW DL-serine, llnn EDTA. 0.lmM PLP, and Zlnn 2-mercaptoethanol. Buffer C 1 s potassium phorpKtte, pH 7.5, a t concentrations lndlcdted below. llnn K -s e r i n e , llnn EOTA, and 2d4 2-mercaptoethanol.

MATERIALS AND METHODS~
Enzyme Preparation cedure of Ulevitch and Kallen (1977) 16 modlfled by Hamen and Davlr (19791. The C Y t O B O l I C Senne hydmxymethyltrdnlferIse war p u n f i e d t o hamgenelty frm sheep liver by the proisoenzyme frm hog l i v e r YIP p u r i f i e d t o honagenelty by a method developed by Cook (Matthers, et. al., 1982).
HmgeneOuI beef l i v e r enzyme was isolated, fop the flrrt tlm. using the following procedure. Frozen l i v e r (600 9) w s sliced ground and horngentzed w t h 2 volumer of buffer Ain a Uaring blender. Centrifugat~on'of thy5 homog;nate f o r 20 mln r m u e d c e l l u l a r debris and connective tissue. and the dark red supernatant "as placed l n a balling water bath and cons t a n t l y s t l r r e d u n t i l it reached 6B'C. Follnlng thls treatment, the suspenrlm was nmnediatel y coaled i n an ice bath to 1O' C. Centrlfugation for IS "in renoved denatured pratelnr: and The dialysate was loaded on a 5 CL I 30 cm phosphocellulore colunn that had prevlourly been equilibrated with 5onH b u f f e r 8 . Uarhing the c o l m n with the mme buffer r m v e d unbound "Materials and Methods" and Tables I and I1 are presented in miniprint as prepared by the authors. The miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 82M-2629, cite the authors, and include a check or money order for $2.00 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.
The following abbreviations are used PLP, pyridoxal 5"phosphate; NADH, nicotinamide adenine dinucleotide (reduced form); Yashing the column u i t h thissame buffer eluted the e n z m , and active fractions were pooled. Table 1 Purification o f Bovlne Senne Hydroxylnethyltranrferase ACtivlty was determined routlnely using the %-phenylr e t i n a assay (Ulevltch and Kallen. 1977). Oata frm a t y p i c a l p w l f i c a t l e n are presented i n Table 1. hepatic tissue using the subsequent pmedure. The tomgeneour enzyme * d l eluted using 0.311 b u f f e r w e d aMve, and stored In t h i s s t a t e a t 4' C u n t i l use. Table 2 depicts values Obtained i n 1 typlcal pwlfication.
The final step involved loading the dialyzed pool onto a latrexm tiel Red A colmn. 1.5

RESULTS
Specificity of the Glycine Carbanion for Lipoic Acid-Glycine reacts with serine hydroxymethyltransferase to form a Schiff base which absorbs a t 428 nm. A small percentage of the pro-S proton dissociates from the complex (Schirch and Jenkins, 1964) forming a carbanionic quinoid species ( X , , , 505 nm). Near complete dissociation of the proton is achieved in the presence of tetrahydrofolic acid or N5-methyltetrahydrofolate (Fig. 1). In order to determine if oxidized lipoic acid could intercept the glycine anion of this ternary complex, spectral changes in the band a t 505 nm were monitored in the presence of lipoic acid. As illustrated in Fig. 1, a significant decrease in the 505-nm band occurs with incremental increases in lipoic acid. This ternary complex was titrated with lipoic acid; from a secondary plot of 1/AA505 versus l/[lipoic acid] (Fig. 2) and from computer a n a l y~i s ,~ dissociation constants for glycine were determined to be 7.3 k 0.08 mM for the porcine mitochondrial isoenzyme and 16.2 f 0.3 mM, 56.2 f 0.3 mM, and 7.5 k 0.2 mM for the cytosolic isoenzymes from porcine, bovine, and lamb hepatic tissues, respectively.
To assess the specificity of the glycine carbanion for disulfides, oxidized dithiothreitol was added, replacing lipoic acid in the reactions. Titration of the ternary complex with oxidized dithiothreitol up to 10 mM had no effect on the spectrum of the carbanionic quinoid.
Serine hydroxymethyltransferase forms anionic quinoid intermediates with both D-and L-amino acids (Ulevitch and Kallen, 1977) as a result of a-proton abstraction from both D-and L-amino acids capable of forming Schiff bases with serine hydroxymethyltransferase. Experiments were conducted with L-phenylalanine to determine if lipoic acid could quench anionic quinoid species for any such serine hydroxymethyltransferase-amino acid binary complex. Titration with lipoic acid, using concentrations as high as 10 mM, did not affect the spectrum of the quinoid species generated by the binary complex of L-phenylalanine and enzyme.
Catalytic Turnover of Lipoic Acid and Glycine by Serine Hydroxymethyltransferase-After it was established that lipoic acid was capable of quenching the quinoid peak, it was then desirable to note if the reaction was catalytic when not in the presence of methyltetrahydrofolic acid. Since oxidized lipoic acid can trap the glycine anion, we postulated that the reaction illustrated in Fig, 3 could occur. The final product will contain a free sulfhydryl group providing an accessible functional group to serve as a probe to assay for enzymatic activity. As a result, a new assay procedure was developed for the glycine decarboxylase activity of serine hydroxymethyltransferase using dithiobis(nitrobenzoic acid) to react with the free sulfhydryl group generated from the decarboxylative addition of glycine to lipoic acid. To establish the validity of  the assay procedure and thus the catalytic nature of the reaction between glycine and lipoic acid, we demonstrated that the rate of reaction obeyed saturation kinetics with respect to concentration of substrates as well as concentration of enzymatically active bovine serine hydroxymethyltransferase.'j The serine hydroxymethyltransferase from beef liver was also used to investigate the kinetic mechanism of the decarboxylation of glycine in the presence of lipoic acid. From the data presented in Fig. 4, a sequential mechanism for the L. R. Zieske and L. Davis, unpublished data.
glycine-lipoic acid oxidoreductase and decarboxylation reaction can be suggested. From the linear replots of the lines in Fig. 4 and computer analysis, the K,,, values were determined to be 27.0 f 0.8 mM for glycine and 12.7 f 0.3 mM for lipoic acid.
In order to determine if bovine serine hydroxymethyltransferase was able to catalyze decarboxylation of the glycinelipoate adduct, the reaction was carried out using [l-"CC] glycine. In the reaction between [1-"Clglycine and lipoic acid in the presence of serine hydroxymethyltransferase a significant amount of ['4C]C02 was released and trapped in KOH and no decarboxylation occurred in the absence of either lipoic acid or enzyme. In the presence of lipoic acid, the decarboxylation reaction proceeded at a rate of 1 nmol/min/ (mg of enzyme) of ["C]CO2 produced. This rate of decarboxylation of glycine agrees with the relative rate of formation of the methylamine-lipoate adduct indicating a stoichiometric relationship among the products.
Inhibition of Reactions Involving a Glycine Anion by Lipoic Acid-After demonstrating that lipoic acid can add to the glycine anion and cause decarboxylation, it was also desirable to determine if lipoic acid could be involved in other reactions catalyzed by bovine serine hydroxymethyltransferase in which a glycine anion is an intermediate. To investigate this possibility, the reaction involving the cleavage of P-phenylserine to benzaldehyde and glycine was studied in the presence of lipoic acid. Lipoic acid was observed to have a competitive inhibitory effect on the cleavage reaction as illustrated in Fig.   5. The inhibition constant, K,,, for lipoic acid was determined to be 0.92 2 0.02 mM from computer analysis.
Reactions of the Glycine Carbanion with Coenzyme A Derivatives-We have now established that serine hydroxymethyltransferase, in addition to its role in Reactions a-c stated previously, can also catalyze the decarboxylation of glycine illustrated in Reaction d.
We found it of interest to investigate the possibility that serine hydroxymethyltransferase could catalyze condensation reactions with acyl coenzyme A derivatives to mimic reactions catalyzed by the following pyridoxal 5"phosphate-dependent enzymes, 6-aminolevulinate synthase and aminoacetone synthase. These reactions involve the condensation of acetyl coenzyme A or succinyl coenzyme A and glycine in a manner similar to lipoic acid and glycine. TO carry out this study, we generated the carbanionic quinoid complex of glycine, serine hydroxymethyltransferase, and N5-methyltetrahydrofolic acid and titrated it with incremental increases in concentration of the acyl coenzyme A derivatives of interest, and the spectral change of the band was monitored at 505 nm. Through such experiments, it was determined that even at concentrations as high as 10 mM of each of the acyl coenzyme A derivatives, no spectral change occurred.
Association of Lipoamide Dehydrogenase Activity with Serine Hydroxymethyltransferase-The evidence presented in this text shows that the reaction between glycine and lipoic acid catalyzed by serine hydroxymethyltransferase mimics those reactions that occur between the P-protein and the Hprotein of the glycine cleavage system. Another feature of the P-protein is a close association of this enzyme with the Lprotein (lipoamide dehydrogenase activity) of the glycine cleavage system. To determine if this association was also observed with the various serine hydroxymethyltransferases, lipoamide dehydrogenase activity was monitored throughout the purifications of the various serine hydroxymethyltransferases. The results indicate that even nearly homogeneous serine hydroxymethyltransferase, in all cases, had some resid- ., 10 mM; 0 , 1 5 mM; A, 20 M. ual lipoamide dehydrogenase activity remaining. During each step of the purification, however, the ratio of serine hydroxymethylase activity/lipoamide dehydrogenase activity did not remain constant and dramatically increased, i.e. from 36.5 prior to affinity chromatography to over 300 upon elution for the cytosolic bovine isoenzyme. This lipoamide dehydrogenase activity, when present in a ratio below 100, competed with serine hydroxymethyltransferase for lipoic acid, thus diminishing the rate of decarboxylation of glycine catalyzed by serine hydroxymethyltransferase. Therefore, homogeneous serine hydroxymethyltransferases were used throughout these studies. and 0,2.8 mM.

DISCUSSION
Putative carbanion intermediates formed during the reaction of pyridoxal 5'-phosphate-dependent enzymes with their substrates have been very difficult to trap. Evidence in this manuscript demonstrates the ability of lipoic acid to intercept glycine carbanion intermediates formed from binary complexes of serine hydroxymethyltransferase and glycine. The only previous probes for such carbanion intermediates have been oxidation-reduction indicators such as hexacyanoferrate (111) (Healy and Christen, 1973;Shylapnikov and Karpeisky, 1969). In addition to trapping the glycine carbanion, lipoic acid forms an adduct which is processed further by the hy-droxymethyltransferase to yield COz and a methylamine lipoic acid adduct. The overall reaction is comparable to the initial reaction of the glycine cleavage system catalyzed by the Pprotein component in which glycine is transferred to the lipoate moiety of the H-protein with subsequent loss of CO,. In addition, the decarboxylation reaction catalyzed by serine hydroxymethyltransferase is identical with the lipoic aciddependent decarboxylation of glycine by the P-protein alone. Therefore, both serine hydroxymethyltransferase and the Pprotein component of the glycine cleavage complex will catalyze lipoic acid-dependent decarboxylation of glycine.
The decarboxylation of glycine catalyzed by the hydroxymethyltransferase proceeds by a pathway which requires prior dissociation of the pro-S-a-proton from glycine. The resulting carbanion then adds reductively to lipoic acid with stoichiometric release of CO, and a methylamine adduct of lipoic acid. In addition to establishing a 1:l stoichiometry for CO, release and methylamine-lipoate production, we were also interested in demonstrating that the rate of the reaction obeyed saturation kinetics with respect to the concentration of substrates and was proportional to the concentration of enzymatically active hydroxymethyltransferase. Both of these criteria were met and the reaction was observed to proceed through a sequential mechanism, with all substrates binding prior to the release of any products.
A sequential mechanism has also been suggested for the decarboxylation of glycine by the Pand H-proteins of the glycine cleavage complex (Hiraga and Kikuchi, 1980;Fujiwara et al., 1979) and other pyridoxal 5"phosphate-dependent decarboxylases which required carbanion addition prior to loss of COP from the amino acid cosubstrates (Zaman et al., 1973;McGilvray and Morris, 1971;Krisnangkura and Sweeley, 1976). Since the addition of the glycine carbanion to lipoic acid in this system is mechanistically similar to the addition of glycine carbanions to acyl-CoA derivatives in other enzyme systems, we tested acetyl-coA and succinyl-CoA as acceptors for the glycine carbanion in the presence of serine hydroxymethyltransferase. These acyl-CoA derivatives were not acceptors for the glycine carbanion in the presence of serine hydroxymethyltransferase. However, the inability to trap the carbanion with these derivatives may be due to the inability of the acyl-CoA to bind to the binary complex or a result of some stereochemical factors. In reactions where an amino acid carbanion is observed to condense with acyl-CoA derivatives, the enzymes selectively remove the pro-R proton or its equivalent (Zaman et al., 1973;Krisnangkura and Sweeley, 1976) whereas serine hydroxymethyltransferase selectively removes the pro-S proton from glycine (Schirch and Jenkins, 1964). Further evidence for carbanion stereoselectivity is provided by experiments which showed lipoic acid to be unable to react with the L-phenylalanine carbanion generated with serine hydroxymethyltransferase. Since the proton abstraction from L-phenylalanine by the hydroxymethyltransferase is equivalent to loss of the pro-R proton from glycine, this observation might suggest that the pro-R carbanion is not accessible for lipoic acid condensation.
Structural specificity for lipoic acid was tested for by substituting oxidized dithiothreitol in its place; however, only lipoic acid was active. The lack of activity may result from dithiothreitol not binding, differences in redox potential, or ring strain. Significantly, the uniqueness of lipoic acid as a substrate in the decarboxylation of glycine would imply a specific binding site for it on the hydroxymethyltransferase. Other evidence for a lipoic acid binding site was obtained from studies which revealed the ability of lipoic acid to intercept glycine carbanion intermediates formed during the cleav-age of (3-threo-phenylserine by the hydroxymethyltransferase. The fact that this reaction is inhibited by lipoic acid would suggest that it is able to complete with protons for the glycine carbanion intermediate. Secondly, the fact that the inhibition is competitive is consistent with lipoic acid and phenylserine competing for the same enzyme form. Both observations previously discussed would suggest that lipoic acid binds at some unique site on serine hydroxymethyltransferase. Previous interpretations of serine and glycine catabolism has implied that separate mitochondrial pyridoxal 5'-phosphate-dependent enzymes are required to carry out the complete degradative pathway of serine to COz, NH3, and methylenetetrahydrofolic acid within the mitochondrion. One enzyme, serine hydroxymethyltransferase, generates glycine and methylenetetrahydrofolic acid from serine; then, it releases the glycine to the P-protein, the second enzyme, which initiates the further breakdown to COz, NH3, and a second mole of methylenetetrahydrofolic acid. This manuscript suggests that this complete mitochondrial catabolic process could be initiated by the serine hydroxymethyltransferase without the transfer of glycine to a different enzyme, thus, eliminating the requirement for two separate pyridoxal 5"phosphatecontaining enzymes to process serine in mitochondria. However, final assessment of the ability of serine hydroxymethyltransferase to function alone or in lieu of the P-protein await experiments coupling the hydroxymethyltransferase with purified H-protein of the glycine cleavage system. This is required since the H-protein containing lipoic acid is 1000-fold more effective in enhancing the glycine decarboxylase activity of the P-protein than lipoic acid alone (Hiraga and Kikuchi, 1980) and is the natural aminomethyl group carrier.