Production of the neuromodulator H2S by cystathionine beta-synthase via the condensation of cysteine and homocysteine.

Hydrogen sulfide (H2S) has been observed in relatively high concentrations in the mammalian brain and has been shown to act as a neuromodulator. However, there is confusion in the literature regarding the actual source of H2S production. Reactions catalyzed by the cystathionine beta-synthase enzyme (CBS) are one possible source for the production of H2S. Here we show that the CBS enzyme can efficiently produce H2S via a beta-replacement reaction in which cysteine is condensed with homocysteine to form cystathionine and H2S. The production of H2S by this reaction is at least 50 times more efficient than that produced by hydrolysis of cysteine alone via beta-elimination. Kinetic studies demonstrate that the Km and Kcat for cysteine is 3-fold higher and 2-fold lower, respectively, than that for serine. Consistent with these data, in vitro reconstitution studies show that at physiologically relevant concentrations of serine, homocysteine, and cysteine, about 5% of the cystathionine formed is from cysteine. We also show that AdoMet stimulates this H2S producing reaction but that there is no evidence for stimulation by calcium and calmodulin as reported previously. In summary, these results confirm the ability of CBS to produce H2S, but show in contrast to prior reports that the major mechanism is via beta-replacement and not cysteine hydrolysis. In addition, these studies provide a biochemical explanation for the previously inexplicable homocysteine-lowering effects of N-acetylcysteine treatments in humans.

Recently, there has been increased interest in endogenously produced hydrogen sulfide (H 2 S) as a physiologically important molecule. Relatively high concentrations of H 2 S have been observed in the brains of rats, humans, and cows (1)(2)(3). At physiological concentrations it has been shown that H 2 S enhances N-methyl-D-asparate receptor-mediated response and can modify long term potentiation (4 -6). H 2 S also inhibits smooth muscle cell proliferation via the mitogen-activated protein kinase pathway and protects neurons against oxidative stress (7). H 2 S also appears to have an effect on the cardiovascular system, acting as a vasorelaxant by increasing potassium-ATP channel currents (8). Taken together, these observations sug-gest that endogenously produced H 2 S is an important regulatory molecule in humans.
How is endogenous H 2 S produced? Potential sources are alternative reactions catalyzed by the enzyme cystathionine ␤-synthase (CBS) 1 (6). The normal cellular function of CBS is to catalyze the condensation of serine with homocysteine to form cystathionine and water, a key reaction in the transsulfuration pathway. CBS uses pyridoxal phosphate (PLP) as a co-factor and is a member of the ␤-family or fold type II of PLP containing enzymes. Enzymes in this family characteristically have the ability to catalyze ␤-replacement and ␤-elimination reactions from a variety of different substrates (9).
There are two potential mechanisms through which CBS could produce H 2 S. First, CBS could catalyze the production of H 2 S from cysteine by a ␤-elimination or an ␣,␤-elimination reaction (Fig. 1, Alternate Reactions 3 and 4, respectively). This type of reaction has been reported to occur with CBS isolated from mouse brain and from CBS present in rat liver and kidney extracts (6,10). An alternative source for H 2 S production would involve a ␤-replacement reaction. Using this mechanism CBS can produce H 2 S from the reaction of L-cysteine and 2-mercaptoethanol to form S-hydroxyethyl-L-cysteine and H 2 S (Fig. 1, Alternate Reaction 2) (11). While this reaction would not be expected to occur in vivo, a similar ␤-replacement reaction could occur by the condensation of homocysteine with cysteine ( Fig. 1, Alternate Reaction 1). This potential reaction is interesting because it would also be an alternative method for metabolizing homocysteine. Elevated plasma homocysteine levels have been linked to a variety of human diseases, including heart attack, stroke, Alzheimer disease, and osteoporosis (12)(13)(14)(15). Since CBS is a key regulator of homocysteine, it is possible that this alternative reaction may have clinical relevance.
In this paper we report the characterization of the biochemical and kinetic properties of human CBS in catalyzing various H 2 S-producing reactions. We find that human CBS can efficiently catalyze the formation of H 2 S via the condensation of homocysteine with cysteine and that this reaction is likely to occur in vivo.

MATERIALS AND METHODS
CBS Expression Systems-Two expression systems were used to produce human CBS. For the yeast system we used a yeast strain (WY218) that was deleted for endogenous yeast CBS (CYS4) and deleted for yeast O-acetylserine/O-acetylhomoserine sulfhydrylase. Extracts from WY218 exhibit no CBS activity and have no ability to form H 2 S (see Fig. 2, lane 1). Into this strain was transformed a plasmid expressing either wild-type human CBS or a truncated human CBS (amino acids 1-409) (16). Total yeast extracts were made as described previously (17). * This work was supported by National Institutes of Health Grants HL57299 and CA06927 and by an appropriation from the Commonwealth of Pennsylvania. 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.
§ These authors contributed equally to this work. For bacterial expression, Escherichia coli BL21 (DE3) containing pGEX-CBS (16) were grown to an A 600 of 0.6 in LB medium at 37°C. isopropyl 1-thio-␤-D-galactopyranoside was added to a final concentration of 0.05 mM to induce the expression of fusion protein at 20°C. The cells were resuspended in PBS containing 10 mM dithiothreitol, 100 mM MgCl 2 , 0.5 mg/ml lysozyme, 2 units/ml DNase, and 0.86 mg/ml protease inhibitor mixture (Sigma) for 1 h at 4°C and then lysed by freezethawing two times. The lysates were incubated at 4°C for an additional 30 min, briefly sonicated on ice to reduce viscosity, and centrifuged, and the clarified supernatants containing fusion proteins were filtered (0.2-m filter) and applied to a GSTrap column (Amersham Biosciences) that was connected to a Bio-Rad Biologic Duo Flow FPLC equilibrated with PBS (pH 7.3) and 5 mM dithiothreitol. After being washed with 20 column volumes of the same buffer, the column was filled with thrombin protease, sealed, and incubated at room temperature for 16 h. Cleaved protein was eluted using 20 ml of PBS, and bound GST and uncleaved GST fusion proteins were eluted with 20 mM reduced glutathione in Tris-HCl (pH 8.0). Analysis of the purified human CBS protein by SDS-PAGE and Coomassie Brilliant Blue staining indicated that the protein was Ͼ95% pure.
CBS Enzyme Activity Assays-We used three different assays to assess CBS enzyme activity. Assays involving crude extracts were first dialyzed overnight against 50 mM Na/Bicine (pH 8.6) buffer containing 50 M PLP in a Slide-A-Lyzer mini-dialysis unit (Pierce).
For the native gel assays, H 2 S production was assayed by reaction with Pb-acetate using a modified previously described procedure (11,18). Sixty or 100 g of yeast extract was loaded on the 8% native Tris-glycine gels (Novex) at 4°C. After gel electrophoresis was finished, active protein bands in native gels were detected by soaking the gel (8.5 ϫ 7.5 cm) in 50 ml of the reaction assay mixtures for several hours to overnight at room temperature. The reaction mixture contained 200 mM Na/Bicine (pH 8.6), 50 M PLP, 0.25 mg/ml bovine serum albumin, 0.4 mM lead acetate, and substrates: reaction 1 substrates (10 mM L-cysteine, 10 mM L-homocysteine), reaction 2 substrates (10 mM Lcysteine, 10 mM 2-mercaptoethanol), reaction 3 and 4 substrate (10 mM L-cysteine), respectively.
For production of H 2 S from purified CBS, a spectrophotometric assay was used. The reaction of H 2 S with lead acetate to form lead sulfide was monitored continuously by the increase in absorbance at 390 nm in a Hewlett Packard 8453 diode array spectrophotometer thermostatted at 37°C. Reaction mixture (1 ml) contained 200 ng of purified CBS using the reaction conditions described above. Quantitation of H 2 S production was performed using a calibration curve obtained by comparing A 390 with cystathionine production using a Biochrom 30 amino acid analyzer.
For kinetic and other studies, enzyme activities were measured using standard reaction containing 100 mM Na/Bicine (pH 8.6), 200 M PLP, 1 mM tris(2-carboxyethyl)phosphine, 100 M AdoMet, 0.25 mg/ml bovine serum albumin, and 10 mM L-homocysteine. Total reaction volume was 50 l. The concentrations of co-substrates serine or cysteine were between 0 and 40 mM. Reactions were carried out at 37°C for 1 h, and the cystathionine produced was measured by an amino acid analyzer (Biochrom 30). The kinetic parameters were determined using Enz-Kinetics software.
Determination of Mouse Liver Amino Acid Concentrations-Livers from C57BL6 animals maintained on standard mouse chow (LabDiet, 5013) were harvested, weighed, then extracted as described previously (19). The volume of the total extract was then carefully measured, and 50 ml were analyzed using a Biochrom 30 amino acid analyzer.
Amino Acid Analysis-Samples for amino acid analysis were first processed by addition of dithiothreitol to a final concentration of 1.2% and incubation on ice for 10 min. This was followed by addition of sulfocylic acid to 5% and centrifugation at 12,000 ϫ g for 10 min. The samples were then loaded on the Biochrom 30 using an autoloader. Quantitation was performed by calibrating the peak heights to a standard in which the amounts of cysteine, serine, and homocysteine were known. The program EZ Chrom Elite was used to analyze the data.

H 2 S Production by Human CBS in Saccharomyces cerevi-
siae-To identify whether human CBS had H 2 S-forming activity, we measured H 2 S formation using a gel activity assay. In this method, yeast extracts are separated on native gels, exposed to various substrates, and then assessed for H 2 S formation in situ (see "Materials and Methods"). We examined H 2 S formation from either 10 mM L-cysteine alone, 10 mM L-cysteine with 10 mM L-homocysteine, or 10 mM L-cysteine and 10 mM 2-mercaptoethanol. The yeast strain we used (WY218) was deleted for endogenous yeast CBS and contained either a control plasmid, a plasmid expressing wild-type human CBS (aa 1-551), or one expressing a truncated human CBS (aa 1-409) lacking the C-terminal regulatory domain. The truncated form of CBS has been shown to be hyperactive and not responsive to allosteric regulation by AdoMet (16). As shown in Fig. 2, both the full-length and truncated form of the enzyme have significant H 2 S forming ability when cysteine is combined with either homocysteine or ␤-mercaptoethanol. However, neither fulllength nor truncated CBS has significant H 2 S forming ability when only cysteine is present. These results show that human CBS is much more active at producing H 2 S by a ␤-replacement reaction then by a ␤-elimination reaction.
AdoMet is an allosteric effector of CBS that stimulates CBS activity by relieving the inhibition of the C-terminal domain (17). As expected, we found that addition of AdoMet stimulated H 2 S formation from the wild-type enzyme (Fig. 2) but not the truncated enzyme. This result suggests that the regulation of the H 2 S forming ␤-replacement reactions is similar to that of the canonical reaction.
We also examined the H 2 S forming ability of recombinant human CBS purified from E. coli. In this experiment, reactions were carried out in solution, and lead sulfate accumulation was determined using a spectrophotometer. As shown in Fig. 2B, the purified enzyme produces significant levels of H 2 S when either 10 mM L-homocysteine or 10 mM ␤-mercaptoethanol were combined with 10 mM L-cysteine. However, when 10 mM Lcysteine was incubated in the absence of a co-substrate we saw no detectable levels of H 2 S formation. These results confirm that CBS produces H 2 S through condensation of cysteine with homocysteine (or some other substrate) via ␤-replacement

FIG. 1. Potential H 2 S-producing reactions catalyzed by CBS.
The first reaction shown is the standard CBS reaction which does not produce H 2 S. The four alternative reactions are all capable of producing H 2 S. The biochemical mechanism by which each reaction proceeds is indicated. rather than through cysteine hydrolysis and an elimination reaction.
Kinetic Studies of Purified Human CBS-We next compared the relative efficiency of utilization of each of these substrates. We determined the K m and V max for both L-serine and L-cysteine under conditions in which L-homocysteine was present at 10 mM in the reaction mixture. We found that the K m of cysteine compared with serine was 3.5-fold higher (Table I) and the V max 2.3-fold reduced. The ratio of K cat /K m is about 8-fold decreased, suggesting that under physiological conditions serine would be utilized in preference to cysteine.
Competition Studies-We next examined the utilization of either cysteine or serine in the formation of cystathionine in an in vitro competition assay. Either serine, cysteine, or both were added to 200 ng of purified recombinant CBS and incubated for 1 h at 37°C. Subsequently, cystathionine formation and serine/ cysteine usage were measured using an amino acid analyzer ( Fig. 3). At high concentrations (10 mM) of serine, cysteine, and homocysteine, cysteine was used to make 44% of the cystathionine, while serine was used to make 56%. As we decreased the concentration of substrates to 1 mM, the amount of cystathionine formed from cysteine decreased to 35%. At the lowest concentration tested (0.1 mM) only 23% of the cystathionine came from cysteine. These data are consistent with the kinetic data (a 3-fold difference in K m ) and show that the ratio of cysteine to serine used in cystathionine formation depends greatly on the concentration of the substrates present in a tissue.
In examination of the literature we found widely varying references for the concentrations of cysteine, serine, and homocysteine in mammalian livers (20,21). Therefore, we determined the concentrations of these amino acids in mouse liver ourselves. Mouse livers were weighed, extracted, and then analyzed for free amino acid content using an amino acid analyzer. This allowed us to determine the free amino acid content per milligram of liver. To estimate how much water was in the tissue, we determined the weight of mouse livers before and after desiccation. The difference in weight indicates the amount of water in the sample, and this number was then used as the denominator in our calculations. Using this procedure, we estimated that in mouse liver the concentrations of serine, cysteine, and homocysteine are 0.72, 0.47, and 0.58 mM, respectively. In an in vitro reaction using these concentrations of substrates, we determined that about 5% of the cystathionine formed in a mouse liver was derived from cysteine (Fig. 3,  right-hand panel).
No Effects of Calcium and Calmodulin on H 2 S-producing Enzyme Activity-It has been previously reported that CBS contains a calmodulin (CaM) binding motif and that the hydrolysis of cysteine is regulated by calcium and CaM (6). We failed to see any effect of CaM and or calcium addition on CBS activity from mouse liver extract, purified recombinant human CBS produced in E. coli, or from yeast extracts expressing human CBS (data not shown). Since the experiments reported by Kimura were done on mouse brain, we also examined mouse brain extracts for stimulation by Ca 2ϩ and CaM. Again, we failed to see any stimulation of CBS activity in the presence of Ca 2ϩ /CaM (data not shown). These studies show that CBS is not regulated by Ca 2ϩ /calmodulin. DISCUSSION The goal of this work was to clarify the role that CBS may have in the endogenous production of H 2 S in vivo. Work in the field of molecular neurology has clearly established that H 2 S can modulate neuronal signals by modulating signaling of the N-methyl-D-asparate receptor (4 -6). However, there has existed confusion as to the source of H 2 S in vivo. Most of the literature on endogenous H 2 S suggests that H 2 S is formed primarily from the hydrolysis of cysteine by the action of cystathionine ␤-synthase. The data presented here do support the idea that H 2 S is produced by CBS. However, the mechanism for its production is not by the hydrolysis of cysteine but rather by a ␤-replacement reaction utilizing homocysteine and cysteine. This reaction is essentially identical to the endogenous ␤-replacement reaction involving homocysteine and serine, except that cysteine is substituted for serine, resulting in the formation of H 2 S instead of H 2 O. This reaction, like the canonical reaction, is stimulated by addition of AdoMet. Unlike previous investigators, we did not observed any evidence for the stimulation of CBS by Ca 2ϩ /calmodulin (6).
Although our data are in conflict with some recent work, it is actually quite consistent with work from the 1960s examining H 2 S formation from partially purified serine sulfhydrase derived from chicken liver. Braunstein  The gels on the left were developed using cysteine ϩ homocysteine (alternate reaction 1), the middle gels were developed using 2-mercaptoethanol (2-ME) ϩ homocysteine (alternate reaction 2), and the gels on the right were developed only in the presence of cysteine (alternate reactions 3 and 4). The molecular mass markers at the top left are indicated: ferritin (440 kDa), catalase (232 kDa), lactate dehydrogenase (140 kDa), and bovine serum albumin (67 kDa). Markers are shown after staining with Coomassie Blue. B, 200 ng of purified recombinant CBS was assessed for H 2 S production as described under "Materials and Methods." All reactions were done in triplicate, and standard deviation is as shown. The units for enzyme activity are nmol of H 2 S produced per mg of protein per minute. There was no detectable H 2 S produced in the reaction containing cysteine alone.
showed that chicken liver serine sulfhydrase had a 30-fold increase in H 2 S-producing activity when cysteine and homocysteine were added together, compared with cysteine by itself. Later work by this same group showed that the enzyme identified as chicken liver serine sulfhydrase was in fact enzymatically quite similar to partially purified rat liver cystathionine ␤-synthase, and it was proposed that these were in fact the same enzyme (23). The data presented here clearly support this interpretation.
There have been other reports in the literature suggesting that another transsulfuration enzyme, cystathionine ␥-lyase (CGL), may be important in H 2 S production in vivo (10, 24), but our data do not support that idea. We did not observe any evidence of significant H 2 S formation in yeast extracts using our gel overlay assay when only cysteine was added, even though yeast has significant levels of CGL activity (24). Fur-thermore, we failed to observe any detectable H 2 S formation in the extracts of mouse livers of animals deleted for endogenous CBS. 2 The evidence that CGL can produce H 2 S comes from work in which propargylglycine was shown to inhibit H 2 S production in crude liver extracts (10). One possible explanation for this apparent contradiction is that in the context of a crude rat liver extract the propargylglycine may be affecting other enzymes, resulting in decreased production or increased metabolism of H 2 S.
Our data suggest that the production of H 2 S from cysteine and homocysteine does occur in vivo. We found that wild-type human CBS has a K m value for L-cysteine of 6 mM, about 3-fold higher than the K m for serine. We also found that in three substrate reactions (cysteine, homocysteine, and serine) a small but significant portion of the cystathionine produced came from cysteine. When the substrates were added at physiological concentrations (as determined in mouse liver) we found that about 5% of the cystathionine produced came from cysteine. This is consistent with the observation that H 2 S levels are at least an order of magnitude less than cysteine, homocysteine, or S-adenosylmethionine levels in the brain (25).
Our findings may also have clinical relevance. H 2 S levels have been shown to be severely decreased in the brains of Alzheimer disease patients and in the aorta of hypertensive rats (25). If these low levels of H 2 S are found to be pathogenic in human disease, it may be possible to increase H 2 S and lower homocysteine levels by increasing the concentration of cysteine in tissues. In fact, it has been shown that pharmacologic doses of N-acetylcysteine can lower plasma homocysteine levels in humans (26 -28). The likely reason for this effect is that Nacetylcysteine is converted to cysteine inside cells, thereby increasing the concentration of cysteine and thus driving the conversion of homocysteine to cystathionine. This treatment would also be expected to increase the production of H 2 S.
In summary, the findings presented here help clarify the potential role of CBS in the production of H 2 S and the reduction of homocysteine and support the view that CBS may play a key role in neurobiology.  3. Relative usage of serine and cysteine in cystathionine production. A, trace of the 440 nm absorbance of an in vitro reaction generated by the BioChrom 30 amino acid analyzer. In this reaction, physiologic concentrations of serine (ser) (0.72 mM), cysteine (cys) (0.47 mM), and homocysteine (hcy) (0.52 mM) were combined with 200 ng of purified human CBS. The solid line shows the reaction incubated on ice for 1 h, while the dotted line shows the reaction at 37°C. B, reactions were set up using homocysteine plus serine or cysteine or both at the concentrations indicated. Reactions were performed as described under "Materials and Methods." The columns show the amount of cystathionine formed with the darker texture showing the amount of cystathionine derived from serine, while the lighter texture shows that derived from cysteine. The numbers on the bottom row show the ratio of serine versus cysteine usage in a competitive reaction using both serine and cysteine with homocysteine.