H2S Biogenesis by Human Cystathionine γ-Lyase Leads to the Novel Sulfur Metabolites Lanthionine and Homolanthionine and Is Responsive to the Grade of Hyperhomocysteinemia*

Although there is a growing recognition of the significance of hydrogen sulfide (H2S) as a biological signaling molecule involved in vascular and nervous system functions, its biogenesis and regulation are poorly understood. It is widely assumed that desulfhydration of cysteine is the major source of H2S in mammals and is catalyzed by the transsulfuration pathway enzymes, cystathionine β-synthase and cystathionine γ-lyase (CSE). In this study, we demonstrate that the profligacy of human CSE results in a variety of reactions that generate H2S from cysteine and homocysteine. The γ-replacement reaction, which condenses two molecules of homocysteine, yields H2S and a novel biomarker, homolanthionine, which has been reported in urine of homocystinuric patients, whereas a β-replacement reaction, which condenses two molecules of cysteine, generates lanthionine. Kinetic simulations at physiologically relevant concentrations of cysteine and homocysteine, reveal that the α,β-elimination of cysteine accounts for ∼70% of H2S generation. However, the relative importance of homocysteine-derived H2S increases progressively with the grade of hyperhomocysteinemia, and under conditions of severely elevated homocysteine (200 μm), the α,γ-elimination and γ-replacement reactions of homocysteine together are predicted to account for ∼90% of H2S generation by CSE. Excessive H2S production in hyperhomocysteinemia may contribute to the associated cardiovascular pathology.

H 2 S is the newest member of a growing list of gaseous signaling molecules that modulate physiological functions (1)(2)(3). Concentrations of H 2 S ranging from 50 to 160 M have been reported in the brain (4), where it appears to function as a neuromodulator by potentiating the activity of the N-methyl-Daspartate receptor and by altering induction of long term potentiation in the hippocampus, important for memory and learning (5). H 2 S levels in human plasma are reported to be ϳ50 M, and in vitro studies suggest that it functions as a vasodilator by opening K ATP channels in vascular smooth muscle cells (6).
A recent in vivo study has demonstrated the efficacy of H 2 S in attenuating myocardial ischemia-reperfusion injury by protecting mitochondrial function (7). The role of H 2 S in inflammation is suggested by several studies (8 -11); however, the underlying mechanism is unknown. Remarkably, H 2 S can also induce a state of suspended animation in mice by decreasing the metabolic rate and the core body temperature presumably by inhibiting cytochrome c oxidase in the respiratory chain (12).
Endogenous H 2 S is presumed to be generated primarily by desulfhydration of cysteine catalyzed by the two pyridoxal phosphate (PLP) 3 -dependent enzymes in the transsulfuration pathway: cystathionine ␤-synthase (CBS) and cystathionine ␥-lyase (CSE) (13,14). In fact, it is widely assumed, based on the reported absences of CSE in the brain (15) and of H 2 S in the brain of CBS knock-out mice (16), that CBS is the primary source of H 2 S in this organ, whereas CSE plays the equivalent role in the peripheral vasculature (3). However, recent studies have demonstrated that CSE is both present and active in the brain (17,18) and that H 2 S is in fact detected in the brains of transgenic mice lacking CBS (19). The major role of CSE in H 2 S biogenesis in the peripheral system has been convincingly demonstrated in CSE knock-out mice, which exhibit significantly reduced H 2 S levels in the serum and lower H 2 S production rates in aorta and heart (20). The CSE knock-out mice exhibit hypertension and reduced endothelium-dependent vasorelaxation.
CSE belongs to the ␥-family of PLP-dependent enzymes and catalyzes ␣,␥-elimination of cystathionine to give cysteine, ␣-ketobutyrate, and ammonia ( Fig. 1, reaction 1) (21). In principle, a variety of CSE-catalyzed reactions leading to H 2 S formation can be considered, including cysteine-dependent ␤-reactions (Fig. 1, reactions 2, 3, and 6) and homocysteine-dependent ␥-reactions (reactions 4 and 5). An alternative route to H 2 S synthesis from cysteine catalyzed by CSE has been proposed to involve ␤-elimination of cystine, leading to the intermediate formation of thiocysteine (reaction 7), which decomposes to H 2 S in a nonenzymatic reaction with other thiols (13,22,23). However, the significance of cystine as a source of H 2 S, in the reducing intracellular environment is uncertain.
In this study, we have elucidated the kinetics of H 2 S biosynthesis from cysteine and homocysteine catalyzed by recombinant human CSE. The kinetic data have been utilized to simu-late the rate of H 2 S production by CSE at physiologically relevant concentrations of substrates and at three concentrations of homocysteine, to mimic normal, mild, and severe hyperhomocysteinemia, and the simulated data have been validated experimentally. The simulations predict that the relative contribution of homocysteine versus cysteine to H 2 S biogenesis by CSE increases with the grade of hyperhomocysteinemia. Our studies have led to the identification of two novel sulfur metabolites generated as byproducts of H 2 S synthesis by CSE, lanthionine and homolanthionine. The latter could serve as a biomarker for H 2 S production under hyperhomocysteinemic conditions.

Purification of Human CSE
Recombinant human CSE (polymorphic variant S403) was expressed in the Escherichia coli strain BL21(DE3) using an expression plasmid generously provided by Dr. Marcus Wahl (Max Planck Institute, Mantinsried, Germany). The protein was purified as described previously (24) with the following modification. After the Superdex S-200 (Sigma) size exclusion column, the active fractions were pooled, concentrated, and dialyzed against 100 mM Hepes buffer, pH 7.4, before being stored at Ϫ80°C. The concentration of CSE was determined using the Bradford reagent (Bio-Rad) with bovine serum albumin as a standard.

Enzyme Activity Assays
The following assays were employed to assess CSE activity. In all assays, the concentration of the variable substrate ranged from 0.2 ϫ K m1 to 30 ϫ K m1 . One unit of activity is defined as the amount of enzyme needed to form 1 mol of product min Ϫ1 .
Detection of Cysteine-The DTNB assay was used to measure cysteine produced by CSE from cystathionine, as described previously (25). Briefly, 970 l of Hepes buffer (100 mM, pH 7.4) containing various amounts of a diasteromeric mix of cystathionine was mixed with 10 l of 0.1 M DTNB (in ethanol) and incubated at 37°C for 3 min. PLP was omitted from the reaction mixture, since its addition consistently leads to a slight inhibition of the enzymatic activity. Enzyme (20 l of 1 mg/ml protein) was added to initiate the reaction, and an increase in absorption at 412 nm due to formation of the nitrobenzene thiolate anion was monitored for 1 min in a Cary100 UV-visible spectrophotometer thermostatted at 37°C. Control experiments lacking CSE or substrate yielded the background rates for the reaction of DTNB with the free thiols of CSE or the impurities contained in cystathionine (ϳ90% purity) and were subtracted from the enzyme assay data. A molar extinction coefficient of 13,600 M Ϫ1 cm Ϫ1 was used to estimate the concentration of cysteine generated.
Detection of H 2 S-H 2 S generation was measured in one of two ways. For in-gel assays, H 2 S production was assayed by reaction with lead acetate using a modification of a previously described method (26,27). Purified CSE (40 g/lane) was loaded into wells of a native 4 -15% gradient Tris-glycine gel (Bio-Rad). Immediately after gel electrophoresis (at 4°C), the gel was cut between the lanes, and the strips were soaked for 6 h at room temperature in 40 ml of the reaction mixture (100 mM Hepes buffer (pH 7.4), 0.4 mM lead acetate, and substrates: reaction 1 (30 mM L-homocysteine), reaction 2 (10 mM L-cysteine, 30 mM L-homocysteine), or reaction 3 (10 mM L-cysteine)). Bands producing H 2 S developed a dark brown color that was analyzed using the Gel Doc 2000 gel documentation system (Bio-Rad).
Production of H 2 S by CSE from different substrates was measured in a spectrophotometric assay in which the reaction of H 2 S with lead acetate to form lead sulfide was monitored continuously by the increase in absorption at 390 nm. After the reaction mixture (980 l) containing 100 mM Hepes buffer (pH 7.4), 0.4 mM lead acetate, and varying concentrations of substrate (homocysteine, cysteine, or both) was preincubated at 37°C for 4 min, 20 g of CSE was added to the assay mixture to initiate the reaction, which was monitored at 37°C for 3 min. Lead acetate (0.4 mM) did not inhibit CSE, as determined in the DTNB assay described above. The molar extinction coefficient for lead sulfide under these con- ditions was determined to be 5,500 M Ϫ1 cm Ϫ1 using NaHS as a standard.
Detection of ␣-Ketoacid Products-Determination of ␣-ketoacids generated in the CSE assay was performed as described (28,29). Briefly, 1 ml of the assay mixture containing 100 mM Hepes buffer (pH 7.4) and varying concentrations of homocysteine or cysteine was preincubated for 5 min at 37°C, and the reaction was initiated by adding 20 -50 g of CSE. At the desired time points, 200-l aliquots of the reaction mixture were quenched by adding 200 l of 10% trichloroacetic acid. The precipitated proteins were removed by centrifugation at 14,000 ϫ g for 10 min, and 200 l of the supernatant was mixed with 500 l of 0.5 M sodium acetate buffer, pH 5.0, and 200 l of 0.1% 3-methyl-2-benzothiazolinone hydrazone hydrochloride and then incubated at 50°C for at least 30 min. The control experiment lacking substrate was performed in parallel. After the mixture cooled down to room temperature, the absorbance at 316 nm was read. The concentration of ␣-ketobutyrate in the reaction mixture was calculated using a standard curve generated with known concentrations of ␣-ketobutyrate. The concentration of pyruvate in the reaction mixture was calculated similarly by measuring the absorbance at 324 nm and using the appropriate standard curve.
HPLC Analysis of Cystathionine-The HPLC method was used to estimate the rate of cystathionine formation (reaction 6) at 10 mM cysteine and varying concentrations of homocysteine. The concentration of cystathionine was determined following o-phthaldialdehyde derivatization, essentially as described previously (30,31). Briefly, the enzymatic reaction was stopped by the addition of an equal volume of 10% trichloroacetic acid, and the precipitated protein was removed by centrifugation. The supernatant was neutralized to pH 7-8 with a small amount of saturated K 2 CO 3 and diluted 1:4 with borate buffer (0.2 M, pH 9.6). A 50-l aliquot of each sample was removed and derivatized with 25 l of o-phthaldialdehyde solution (15 mM o-phthaldialdehyde, 30 mM 2-mercaptoethanol, and 10% methanol in 0.2 M sodium borate buffer, pH 9.6) in an autosampler (Agilent 1100 series) for 1 min at 10°C. A 10-l aliquot of the derivatized sample was then injected into the HPLC column (ZORBAX Eclipse XDB-C18 (5-m) analytical column 4.6 ϫ 150 mm) and eluted at a flow rate of 1 ml/min with buffers A (80% 0.1 M sodium acetate and 20% methanol, pH 4.75) and B (20% 0.1 M sodium acetate and 80% methanol, pH 4.75). The following increasing gradient of buffer B was used for elution: 0 -10 min, 30 -60%; 10 -15 min, 60 -100%; 15-20 min, 100%; 20 -22 min, 100 -30%; 22-30 min, 30%, with a corresponding decrease in the percentage of buffer A. The detector was set at 340-nm excitation and 450-nm emission wavelengths. Under these conditions, cystathionine eluted with a retention time of 14.06 min. The concentration of cystathionine was determined using calibration coefficients obtained with the standard. The same HPLC method was used to analyze whether serine and homoserine were produced by the CSEcatalyzed ␣,␤-elimination (reaction 2) and ␣,␥-elimination (reaction 4) reactions, respectively. The retention times for these compounds were 4.17 min (serine) and 5.26 min for homoserine.

H 2 S Production at Physiologically Relevant Concentrations of
Substrate-H 2 S formation was detected using the lead acetate assay described above with the following exception. The reaction mixture (1-ml final volume) contained 100 mM Hepes buffer (pH 7.4), 0.4 mM lead acetate, 5 M cystathionine, 100 M cysteine, and either 10, 40, or 200 M homocysteine. Following incubation at 37°C for 4 min, the reaction was initiated by the addition of 100 g of CSE (corresponding to 2.2 M active sites), and the reaction was monitored at 390 nm for 3 min. The higher concentration of protein was necessary for monitoring the slow reaction rates at these low substrate concentrations. We note that the total substrate concentration for H 2 S generation (i.e. cysteine and homocysteine) varied from 50-to 140fold excess over the concentration of active sites, and ϳ2-10 turnovers were completed during the 3-min time course of the assay.

Mass Spectrometric (MS) Analysis of Reaction Products
For the qualitative analysis of other products in H 2 S generation reactions, a Q TRAP TM mass spectrometer (Applied Biosystems) equipped with a Turbo ion spray source operated in the positive ion mode was employed. Data acquisition was conducted using Analysis software (Applied Biosystems) with a built-in information-dependent acquisition scan function. The supernatant from the assay mixture obtained after protein precipitation by trichloroacetic acid was injected into the mass spectrometer. Control reaction mixtures from which CSE was omitted were run separately.

Analysis of Kinetic Data
Cystathionine, the substrate for CSE, is a condensation product of two amino acids, serine and homocysteine. The active site pocket therefore has binding determinants for two amino acids. In the H 2 S-generating reactions catalyzed by CSE (reactions 2-6), either one (reaction 2 and 4) or both (reactions 3, 5, and 6) amino acid binding pockets are occupied. We refer to the kinetic parameters associated with the single substrate reaction (i.e. ignoring H 2 O) as K m1 and V max1 . The parameters K m2 and V max2 then refer to substrate binding at the second site and the reaction velocity of the bimolecular reaction involving two amino acids, respectively.
Cysteine Production from Cystathionine-The K m and V max values for reaction 1 were determined directly from Michaelis-Menten kinetic analysis using the DTNB assay described above and Equation 1.
We note that commercially available cystathionine is a mixture of diastereomers of which only one, the L,L-isomer, is expected to serve as substrate for CSE. Hence, the substrate concentration was divided by a factor of 4 to obtain the value for K m that is reported.
Pyruvate or ␣-Ketobutyrate Generation-The K m and V max values for CSE-catalyzed pyruvate (reaction 2) or ␣-ketobutyrate (reaction 4) production from cysteine or homocysteine, respectively, were determined using the ␣-ketoacid assay described above and Equations 2 and 3.
To account for the observed substrate inhibition, an inhibition constant (i.e. the K i term) was included in Equations 2 and 3. H 2 S Production from Cysteine or Homocysteine-In principle, the reactions for H 2 S production by CSE can follow either a binary (ping-pong) or ternary or sequential (random or ordered) mechanism. Hence, the experimental data for H 2 S production from cysteine (i.e. reactions 2 ϩ 3) or from homocysteine (i.e. reactions 4 ϩ 5) monitored by the continuous lead acetate assay described above were fitted using Equations 4 -6.
Binding of the second mole of cysteine or homocysteine in reactions 3 and 5, respectively, will affect the V max1 values for H 2 S formation in the unimolecular reactions 2 and 4. To account for this, an inhibition term where K i ϭ K m2 was introduced as shown in Equations 4 -6.
Equations 4 -6 describe a random sequential, ordered sequential, and ping-pong mechanism, respectively. A Hill coefficient was included in these equations to account for cooperativity of binding for the second substrate, which was indicated by the kinetic data and our fitting attempts. Equations 4 -6, as written, describe v H 2 S from cysteine (i.e. reactions 2 ϩ 3) for the alternative mechanisms.
For analysis of v H 2 S from homocysteine (i.e. reactions 4 ϩ 5), Equations 4 -6 were also employed, making the corresponding substitutions (i.e. [Cys] for [HCys], etc.). Unlike cysteine, the dependence of the reaction velocity for H 2 S generation on homocysteine concentration did not show two well separated phases. Hence, the values for K m1(Hcys) , V max1 , and K i for ␣-ketobutyrate generation (obtained from Equation 3) were used as input parameters in Equations 4 -6. The quality of fits obtained for the ordered sequential mechanism was significantly worse than for the other two mechanisms (Tables S1 and S2). In contrast, the quality of fits for the ping-pong versus the random sequential mechanism was indistinguishable.
Cystathionine Production from Cysteine Plus Homocysteine-The K m and V max values for reaction 6 were determined using the HPLC assay for cystathionine formation, as described above, and Equation 7.
In principle, reaction 6 can follow either a binary or ternary mechanism with either cysteine or homocysteine binding first. However, a reasonable fit was only obtained for the ping-pong mechanism where cysteine binds first (Table S3). Equation 7 describes a ping-pong mechanism, in which competitive inhibition terms for the binding of the first and second substrates were included, since the simultaneous presence of both substrates leads to competition at each binding site by the other substrate (K i(Hcys) ϭ K m1(HCys) and K i(Cys) ϭ K m2(Cys) ). H 2 S Production from Cysteine and Homocysteine-Next, the goodness of the kinetic parameters obtained for reactions 2-6 was assessed by fitting the experimental data for H 2 S formation obtained in the presence of 10 mM cysteine and varying concentrations of homocysteine. In this set of experiments, the observed rate of H 2 S production represents the sum of reactions 2-6, as described by Equation 8. The values of v 2 -v 6 corresponding to the reaction velocities for 2-6 were computed using Equations 9 -13 for the ping-pong mechanism, where n, h, and k represent the Hill coefficient for the binding of the second substrate in reactions 3, 5, and 6, respectively, and the kinetic parameters were obtained as described above. A term for competitive inhibition (K m (1 ϩ [I]/K i )) was introduced in Equations 9 -13, since the simultaneous presence of both substrates leads to competition for each binding site by the other substrate.
Determination of the Rates of H 2 S Production at Physiological Substrate Concentrations-The contributions of the various CSE-catalyzed reactions to total H 2 S production were computed at normal, medium, and high homocysteine concentrations ( The K i for cystathionine was ignored, since the concentration of cystathionine used to simulate physiological conditions is low (5 M), whereas the K i for cystathionine for H 2 S production is relatively high (0.78 Ϯ 0.1 mM). For instance, the inclusion of the K i term for cystathionine affected the value for v 2 by Ͻ2%. The values for v 1 corresponding to reaction 1, at varying homocysteine concentrations, were computed using Equation 14, where cysteine and homocysteine act as competitive inhibitors for binding of cystathionine. K i (Cys) and K i (HCys) were assumed to be equal to K m1(Cys) and K m1(HCys) respectively. The resulting reaction rates (v 1 -v 6 ) for reactions 1-6 were then used to calculate the turnover numbers (i.e. v/[E]) for each reaction at the substrate concentrations described above and are expressed per mole of CSE active site.

RESULTS
Purification and Biophysical Characterization of CSE-Purification of recombinant human CSE was accomplished in three chromatographic steps, and the purified protein was judged to be ϳ95% pure by gel electrophoresis (Fig. 2, inset). The typical yield was ϳ20 mg of pure protein/liter of culture. The specific activity of as-purified CSE is 3.1 Ϯ 0.1 units/mg in the DTNB assay with cystathionine as substrate and is similar to the value published previously (2.5 units/mg) (25). As expected, the absorption spectrum of purified CSE is typical of a PLP-dependent enzyme with a maximum at 428 nm and a 280:428 nm ratio of ϳ1:6 (Fig. 2). H 2 S Production by Human CSE-The ability of human CSE to generate H 2 S was first assessed by an in-gel activity assay. For this experiment, native gel strips containing equal amounts of purified  The reaction of H 2 S with lead acetate to form lead sulfide was monitored by the increase in absorbance at 390 nm under quasi-steady-state conditions, as described under "Experimental Procedures," using as substrates 30 mM homocysteine (a), 30 mM homocysteine plus 10 mM cysteine (b), and 10 mM cysteine (c). Inset, in-gel activity staining of CSE. 40 g of CSE was loaded in each lane and separated on a 4 -15% native gradient gel, and H 2 S-producing activity was detected as described under "Experimental Procedures." Although the major band corresponds to the native tetrameric form of CSE, a small proportion appear as high order oligomers. The molecular weight markers (M) were stained with Coomassie Blue. CSE were exposed to the following conditions (30 mM homocysteine, 30 mM homocysteine plus 10 mM L-cysteine, or 10 mM L-cysteine). H 2 S generation was revealed by the appearance of a dark lead sulfide-containing band on the gel. As shown in Fig. 3 (inset), the most intense bands were seen in the presence of homocysteine (lane a). When both homocysteine and cysteine were present in the reaction mixture, lower H 2 S production was observed (lane b), whereas cysteine alone supported the lowest level of H 2 S generation (lane c). These results indicate that at saturating concentrations, homocysteine rather than cysteine is the more effective substrate for H 2 S generation by CSE. We note that CSE migrates as three bands on the native gel, indicating that although the major population is a tetramer (fastest migrating band), a minor proportion exists as higher order oligomers.
The kinetics of H 2 S generation by CSE were further characterized using a continuous spectrophotometric assay. The specific activities under V max conditions for H 2 S formation are 6.6 Ϯ 0.5 units/mg from homocysteine and 1.2 Ϯ 0.3 units/mg from cysteine ( Table 1). As also seen in Fig. 3, the rate of H 2 S formation from homocysteine is higher than from cysteine or from homocysteine plus cysteine. The decrease in the initial velocity of H 2 S formation when both substrates are present in comparison with the rate observed with homocysteine alone results from the occupancy of a portion of the enzyme active sites by the slower substrate, cysteine. This has the net effect of decreased total H 2 S flux generation. Conversely, the apparent activation of H 2 S production when both substrates are present in comparison with cysteine alone results from the fraction of the enzyme that is catalyzing homocysteine-dependent H 2 S production, which occurs at a faster rate than from cysteine. Propargylglycine, a suicide inhibitor of CSE (32), completely blocked H 2 S formation (not shown). Unlike rat CSE that reportedly uses cystine (Fig. 1, reaction 7) rather than cysteine to generate H 2 S (23), H 2 S formation from cystine was not observed with human CSE (data not shown).
In the presence of homocysteine alone, not only was ␣-ketobutyrate detected, consistent with an ␣,␥-elimination (reaction 4), but a new metabolite, homolanthionine (m/z ϭ 237; Fig. 4), was seen, indicating the ␥-replacement of one molecule of homocysteine by another (reaction 5). Serine and homoserine, the products of ␤and ␥-elimination reactions, respectively (reactions 2 and 4), were not detected by HPLC, but their downstream products, pyruvate and ␣-ketobutyrate, respectively, were observed. Lanthionine and homolanthionine are structural homologs of cystathionine that differ by the absence or presence of an extra methylene group, respectively. The identity of homolanthionine was confirmed by MS/MS analysis in which two daughter ion peaks were assigned with m/z ϭ 102 (corresponding to HOOCCH(NH 2 )CH 2 CH 2 )) and 134 (corresponding to SCH 2 CH 2 CH(NH 2 )COOH) that were 14 atomic mass units heavier than the corresponding peaks seen in the MS/MS spectrum of cystathionine (not shown). The identity of lanthionine was confirmed by MS/MS analysis, which was identical to that of a commercial sample of lanthionine in which two daughter ion peaks were assigned with m/z ϭ 120 (corresponding to SCH 2 CH(NH 2 )COOH) and m/z ϭ 192 (corresponding to loss of NH 3 from lanthionine) (not shown). These data establish that homolanthionine and lanthionine produced by CSE are derived from homocysteine and cysteine, respectively.

Effect of Nitric Oxide (NO) on H 2 S-producing Activity of CSE-
Previously, it has been reported that the NO donor, sodium nitroprusside, increases the endogenous levels of H 2 S in vascular tissues (6). The mechanism of this increase was proposed to involve either an NOinduced increase in CSE activity or NO-dependent up-regulation of CSE expression (6). However, we observed no effect of sodium nitroprusside on H 2 S production by CSE (data not shown), indicating that the effect of NO is not at the level of CSE activity.
Kinetics of H 2 S Generation by CSE-Product analyses provided direct evidence for five of the six possible CSE-dependent H 2 S-generating reactions described in Fig.  1 (i.e. reactions 2-6). The kinetics of pyruvate (reaction 2) and ␣-ketobutyrate (reaction 4) formation from cysteine and homocysteine, respectively, and the kinetics of H 2 S formation from the same substrates (i.e. reactions 2 ϩ 3 or reactions 4 ϩ 5) are shown in Fig. 5. The kinetic data were fitted to alternative mechanisms (i.e. binary versus ternary), and the data are presented in Tables S1 and S2. The values of the kinetic parameters obtained from fits to the ping-pong mechanism allowed deconvolution of the K m and V max values associated with each of the four reactions ( Table 1). The dependence of the rate of H 2 S formation on cysteine concentration is markedly biphasic (Fig. 5B). CSE exhibits a considerably higher affinity for cysteine binding to site 1 (3.7 Ϯ 1.1 mM) than to site 2 (33 Ϯ 8 mM), and cooperativity for binding of the second mole of cysteine was seen (n ϭ 3 Ϯ 1).
Deconvolution of the two phases contributing to the rate of H 2 S formation from homocysteine (Fig. 5D) reveals that the K m for site 1 is 2-fold lower than for site 2 (2.7 Ϯ 1.4 and 5.9 Ϯ 1.2 mM, respectively). The kinetics of reaction 6 (i.e. the condensation of homocysteine and cysteine) were monitored by the rate of cystathionine formation. The kinetic data could only be fit of at least three independent experiments. The data were analyzed as described under "Experimental Procedures," and the kinetic parameters obtained from these plots are shown in Table 1.
with a ping-pong mechanism in which cysteine is the first substrate to bind (Table S3). The relative catalytic efficiencies (i.e. k cat /K m ) for the five H 2 S-generating reactions follow the order 5 Ͼ 4 Ͼ 2 Ͼ 3 Ͼ 6 ( Table 1).
The kinetic parameters obtained for reactions 2-6 were then employed to simulate the kinetics of H 2 S formation in the presence of cysteine and homocysteine (Fig. 6A). The excellent correspondence between the simulated and experimental data supports the validity of the kinetic parameters reported in Table 1.
Relative Contributions of the CSE-catalyzed Reactions to H 2 S Generation-Since the cleavage of cystathionine (reaction 1) represents the primary function of CSE in the transsulfuration pathway, it is pertinent to compare the catalytic efficiency of this reaction with those of the side reactions leading to H 2 S generation (Table 1). Under V max conditions, the most efficient H 2 S-generating reaction (i.e. ␥-replacement of homocysteine (reaction 5)) exhibits a k cat /K m value that is ϳ10-fold lower than that for the ␣,␥-elimination of cystathionine. Furthermore, the K m for cystathionine (0.28 Ϯ 0.03 mM) is significantly lower than for homocysteine.
In the cell, the substrate concentrations are low compared with their K m values (i.e. [S] Ͻ Ͻ K m ). Under these conditions, most of the enzyme active sites are unoccupied, and the partitioning of CSE into the various H 2 S-generating reactions is governed by the rate of each reaction (i.e. v ϭ V max [S]/K m ). This is distinct from the situation under in vitro steady-state assays conducted at high concentrations of substrate, where the k cat /K m ratio determines the enzyme specificity for competing substrates. Thus, in the cell, substrate availability will play a crucial role in determining the partitioning of CSE between competing reaction paths, and regulatory mechanisms are likely to exist that lead to enhanced or diminished production of H 2 S and to the diversion of CSE from its role in the transsulfuration pathway.
Using the kinetic parameters described in Table 1, we simulated the relative contributions of each of the reactions to total H 2 S production at three concentrations of homocysteine, representing normal (10 M) versus moderate (40 M) and severe (200 M) hyperhomocysteinemia (Tables 2 and 3 and Fig. 6, B and C). According to our simulations, under normal conditions, ␣,␤-elimination of cysteine (reaction 2) is predicted to be the major source of CSEderived H 2 S, accounting for ϳ70% of the total (Table 3 and Fig.  6, B and C). The ␣,␥-elimination of homocysteine (reaction 4) is the next significant contributor (ϳ29%), whereas the ␤and ␥-replacement reactions (reactions 3, 5, and 6) are of negligible importance. The balance between the reaction shifts, however, with increasing concentrations of homocysteine such that the ␣,␥-elimination of homocysteine (reaction 4) becomes a significant source of H 2 S at moderate and the principal source of H 2 S at severely elevated homocysteine concentrations (Fig. 6, B and  C). The condensation reaction between 2 mol of cysteine (reaction 3) is a minor contributor to the net H 2 S pool. Since the rate FIGURE 6. The relative contributions of reactions 2-6 to H 2 S production at varying homocysteine concentrations. A, the rate of H 2 S production (E) observed in the presence of 10 mM cysteine and varying concentration of homocysteine in 0.1 M Hepes buffer, pH 7.4, at 37°C. Each data point represents the mean Ϯ S.D. of three independent experiments. The relative contributions of the individual reactions (v 2 -v 6 ) to the net rate of H 2 S production (solid line) were simulated using the kinetic parameters reported in Table 1 Table 2). The reaction numbers are indicated above the bar graphs on the left. C, the relative proportions of CSE-derived H 2 S from cysteine versus homocysteine at three concentrations of homocysteine and physiological concentrations of cysteine and cystathionine (Table 3).
of reaction 5 has a square dependence on the concentration of homocysteine, it exhibits the greatest sensitivity to increasing homocysteine concentrations, changing ϳ230-fold between 10 and 200 M homocysteine (Table 3). Homolanthionine production could therefore be a useful biomarker for H 2 S production at high homocysteine concentrations. Generation of H 2 S by the ␥-replacement of homocysteine (reaction 5) accounts for ϳ13%) of total H 2 S generation by CSE under conditions of severe hyperhomocysteinemia. Cystathionine formation (reaction 6) is also predicted to rise with increasing homocysteine (Table 3), but it is unlikely to build up, since it is an efficient substrate for CSE. Under conditions of cystinuria, lanthionine production via reaction 3 would be expected to increase.
Comparison of Experimental versus Simulated Kinetic Data at Physiological Concentrations of Substrates-To test the validity of the simulations described above, the kinetics of the CSE-catalyzed production of H 2 S at substrate concentrations chosen to mimic their physiological levels were determined. At low homocysteine concentrations (10 M), 70% of H 2 S is predicted to result from the ␣,␤-elimination of cysteine (v/[E] ϭ 0.0081 s Ϫ1 ) and 29% from the ␣,␥-elimination of homocysteine (v/[E] ϭ 0.00335 s Ϫ1 ) ( Table 2). The experimentally observed turnover number for H 2 S formation under these conditions was 0.012 Ϯ 0.001 s Ϫ1 and similar to the calculated value of 0.0115 s Ϫ1 . As the concentration of homocysteine increases, the net rate of H 2 S production is expected to increase. In addition, the proportion of H 2 S that is derived from homocysteine increases from 29 to 63 to 90% as homocysteine increases from 10 to 40 to 200 M, respectively (Table 3). In contrast, the rate of H 2 S production from cysteine is virtually unchanged, whereas the proportion of cysteine-derived H 2 S decreases from 70 to 37 to 10%.

DISCUSSION
The nonenzymatic liberation of H 2 S from organic polysulfides in garlic bulbs has been reported recently and provides a mechanistic explanation for the vasoactivity of dietary garlic (33). However, despite the growing interest in H 2 S biology and the therapeutic potential of H 2 S-releasing compounds (2), surprisingly little is known about the enzymatic production of this gas and how it may be influenced by changes in sulfur amino acid levels in disease states.
Since the enzymes in the transsulfuration pathway, CBS and CSE, catalyze elimination/addition reactions at the ␤and ␥-positions of sulfur-containing amino acids, respectively, they are logical candidates for the generation of H 2 S. However, conflicting reports in the literature ascribe the generation of H 2 S by CBS and CSE to different substrates. For example, cystine was proposed to be the preferred substrate for H 2 S by CSE (23), whereas the ␤-replacement of cysteine by homocysteine is reported to be the preferred route for H 2 S generation by CBS (27). In this study, we have investigated the various reactions catalyzed by CSE that result in H 2 S biogenesis and, as side products, the novel amino acids, lanthionine and homolanthionine. The multitude of H 2 S-generating reactions (Scheme 1) and the relatively high K m s for homocysteine and cysteine exhibited by CSE versus the intracellular concentrations of these amino acids makes kinetic analysis complex and necessitates the use of simulations to deconvolute the contributions of different substrates to the overall H 2 S pool.
The reaction catalyzed by CSE in the transsulfuration pathway involves elimination at the ␥-carbon of cystathionine. We find that the catalytic efficiency (k cat /K m ) of the canonical cysteine elimination reaction from cystathionine is ϳ20and ϳ30fold higher than for H 2 S elimination from homocysteine and cysteine, respectively (Table 1). At physiologically relevant concentrations of homocysteine (10 M), cysteine (100 M), and cystathionine (5 M), the turnover number for cystathionine cleavage (0.039 s Ϫ1 ) is still 5-fold greater than for cysteine a The reaction numbers correspond to those shown in Fig. 1. b One unit corresponds to 1 mol of product formed min Ϫ1 . The K m and V max values were determined as described under "Experimental Procedures" and reported in Table 1. c In reactions involving two substrates, the order of the K m values reflects the substrate order in the first column. d The values for the turnover numbers at varying concentrations of homocysteine and physiological concentrations of cystathionine (5 M) and cysteine (100 M) were obtained as described under ЉExperimental ProceduresЉ considering a ping-pong mechanism for the bimolecular reaction and the Hill coefficients (n) reported in Table 1. e -Fold change refers to the change in v/͓E͔ with respect to normal conditions (i.e. 10 M homocysteine, which is assigned a value of 1 for each reaction).

TABLE 3 The relative contributions of H 2 S-generating reactions at varying concentrations of homocysteine as predicted from kinetic data analyses
The reaction numbers correspond to those shown in Fig. 1. Values shown are the percentage contribution of each reaction to net H 2 S production at each concentration of homocysteine in the presence of 5 M cystathionine and 100 M cysteine.

Substrate
Reaction