A novel L-cysteine/cystine C-S-lyase directing [2Fe-2S] cluster formation of Synechocystis ferredoxin.

Iron-sulfur proteins acquire their clusters by posttranslational assembly. To identify components involved in this process an in vitro assay for holoprotein formation was established using the [2Fe-2S] ferredoxin of the cyanobacterium Synechocystis as a model. Conversion of apoferredoxin to the holo- form was observed in an anaerobic reaction medium containing Fe(NH4)2(SO4)2, L-cysteine, glutathione, and catalytic amounts of Synechocystis extract, specifically depleted of endogeneous ferredoxin. An approximate 2500-fold purification of the converter activity yielded a monomeric, 43-kDa, pyridoxal phosphate-containing enzyme, which catalyzed the breakdown of L-cysteine to yield sulfide (assembled in ferredoxin), pyruvate, and ammonia; 1 mol of [2Fe-2S] ferredoxin was formed per 2 mol of cysteine utilized. The purified enzyme also catalyzed the beta-elimination reaction with cysteine in the absence of apoferredoxin. An increased reactivity was found with cystine instead of cysteine, which should yield cysteine persulfide as the primary product. These results provide a function-based identification of a cysteine/cystine C-S-lyase as a participant in ferredoxin Fe-S cluster formation. A substrate-derived cysteine persulfide could be involved in this reaction.

Iron-sulfur (Fe-S) clusters are found as functional units in numerous electron-transferring proteins. They are essential in basic biological processes such as oxidative phosphorylation, photosynthesis, and nitrogen fixation. More recently, they have also been identified as components of enzymes not concerned with redox reactions (1). Lately they have been implicated in gene regulation (2).
Despite these important and diverse functions and thorough biochemical investigations of Fe-S proteins, the biosynthetic steps leading to Fe-S cluster assembly have not been established. While it seems likely that enzyme-catalyzed reactions are involved in the formation of Fe-S clusters in vivo, the chemical reconstitution of apoproteins with free ferrous ion and sulfide in the presence of thiols (3) has been most intensively exploited for cluster build-up in vitro. Nevertheless, several proteins have been suggested as candidates in the biosynthetic process. Rhodanese (4) and 3-mercaptopyruvate sulfurtransferase (5) were effective in purified in vitro systems, comprising apoprotein, thiol, ferrous ion, and the respective sulfur-containing substrate, thiosulfate, or mercaptopyruvate. Analysis of yeast respiration-deficient mutants suggested a role for the BCS1 gene product (6) in the synthesis of functional Rieske protein. Studies on nif mutants of Azotobacter vinelandii (7) identified nifS as essential for nitrogenase activity. More recently, nifS was cloned and expressed in Escherichia coli; the gene product (NifS) was characterized as a pyridoxal phosphate-containing L-cysteine desulfurase, which yielded alanine and elemental sulfur as products (8); with dithiothreitol present in the reaction mixture, sulfide was produced instead of sulfur. Further work showed that the apo-form of the nitrogenase iron protein component could be activated by NifS-catalyzed reassembly of its [4Fe-4S] cluster in vitro (9). NifS was also successfully used for the in vitro reconstitution of the [4Fe-4S] cluster of a mutant FNR protein (10) as well as of the [2Fe-2S] cluster of SoxR (11), both E. coli proteins. Most recently, a NifS-like protein was isolated from E. coli and found to reactivate apodihydroxy acid dehydratase in vitro (12).
The significance of these various findings depends on the nature of the in vivo sulfur source for Fe-S clusters. Feeding experiments with E. coli assigned this role to cysteine (13). Cysteine sulfur was also incorporated into the [2Fe-2S] cluster of ferredoxin in isolated, intact chloroplasts (14). In the chloroplast system, ferredoxin cluster formation was found to be a stroma-located process, consisting of two separate steps: first, the NADPH-stimulated liberation of sulfide from cysteine followed by the ATP-dependent sulfide incorporation into ferredoxin (15).
With the cyanobacterium Synechocystis as a procaryotic model for the chloroplast organelle we now established an apoto holoferredoxin conversion assay. The procaryotic system was chosen because of the relative ease of obtaining large quantities of cell material without the problem of compartmentalization. The endogeneous [2Fe-2S] ferredoxin was adopted as the Fe-S protein studied because of its abundance and the well characterized chemical procedures of reversible cluster removal (16). Moreover, the physiological existence of apoferredoxin has been demonstrated for the chloroplast system, where cluster assembly occurs in the stroma after proteolytic processing of preferredoxin that is imported from the cytosol (17).
We here report on the set up of the assay, the purification and characterization of the converter protein (provisionally named C-DES because of its cysteine desulfhydrase activity), and the stoichiometry of the cluster formation reaction.

EXPERIMENTAL PROCEDURES
Materials-L-Cysteine and L-propargylglycine were from Fluka; Lallylglycine, L-vinylglycine, and pyridoxal phosphate were from Sigma; D-cysteine was from Novabiochem; and L-cystine was from SERVA. Glutathione and glutathione reductase were purchased from Boehringer Mannheim. Other chemicals were of the highest purity commercially available. L-[U-14 C]Cystine (302.2 mCi/mmol) was obtained from Du Pont NEN. Sephadex G-25 and G-75, DEAE-Sephadex A-25, Q-Sepharose FF, Mono P HR5/5, and Polybuffer 74 were from Pharmacia; Ultrogel AcA 44 and Dowex 50 WX8 were from SERVA; and DE52 cellulose was from Whatman. Ferredoxin-NADP reductase was purified * 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.
Unless otherwise stated, proteins were concentrated by ultrafiltration using PM 10 membranes or Centricon 10 units, both of which were supplied by Amicon. Buffer exchanges were performed using a Sephadex G-25 column.
Growth of Synechocystis-Synechocystis PCC 6714 (ATCC 27178) was grown at 30°C in BG-11 medium (19) supplemented with 0.9 g/liter KHCO 3 in an 18-liter fermenter at pH 7.5-8.5. Cultures were illuminated with fluorescent tubes and gassed continuously with 5% CO 2 . After 4 days of growth, typically 15 g of cell paste was harvested by centrifugation, shock-frozen, and stored at Ϫ70°C.
Isolation of C-DES and Ferredoxin-30 g of frozen cells were thawed, resuspended in 130 ml of 50 mM Mops 1 /NaOH, pH 8.0, containing 5 mM dithiothreitol and 0.5 mM phenylmethanesulfonyl fluoride, and disrupted by sonication. After centrifugation (10 5 ϫ g, 90 min) the extract (1500 mg of protein) was concentrated, transferred into 0.15 M Tris/Cl, pH 7.8, containing 0.3 M NaCl (final volume 80 ml), and loaded onto a DEAE-Sephadex A-25 column (20 cm 2 ϫ 5 cm, 10 cm/h). The column was washed with 80 ml of starting buffer; all unadsorbed protein (Ն1450 mg) was collected and reduced for 30 min at 30°C with 10 mM dithiothreitol. Then the buffer was exchanged for 50 mM Mops/NaOH, pH 7.6, and this protein preparation was used for experiments requiring deferredoxinized extract.
Bound For purification of C-DES, deferredoxinized extract (80 ml) was gelfiltered on Ultrogel AcA 44 (20 cm 2 ϫ 85 cm; 5 cm/h) in 50 mM Mops/ NaOH, pH 7.6, containing 20 M EDTA. C-DES was collected in fractions corresponding to K av values between 0.525 and 0.575. These fractions were diluted with 4 volumes of 50 mM Tris/Cl, pH 7.8, containing 1 mM dithiothreitol and 20 M EDTA and loaded onto DE52 cellulose (20 cm 2 ϫ 3.5 cm; 8 cm/h) equilibrated with this buffer. A linear gradient (560 ml) to 160 mM Tris/Cl, pH 7.8, containing 320 mM NaCl, 1 mM dithiothreitol, and 20 M EDTA was applied. C-DES was eluted in the trailing shoulder of the major protein peak. This eluate fraction was diluted with 7 volumes of 50 mM Tris/Cl, pH 7.4, containing 1 mM dithiothreitol and 20 M EDTA and pumped onto Q-Sepharose FF (5 cm 2 ϫ 1.5 cm; 20 cm/h) equilibrated with this buffer. A linear gradient (100 ml) to 150 mM Tris/Cl, pH 7.4, containing 300 mM NaCl, 1 mM dithiothreitol, and 20 M EDTA was applied. C-DES was eluted just in front of the main protein peak. This fraction was transferred into 25 mM 1-methylpiperazine/iminodiacetic acid, pH 5.7, and applied to a Mono P HR 5/5 column (5 cm/min). Homogeneous C-DES was eluted at a pH of 4.4 using Polybuffer 74 (diluted 1:29 (v/v) and adjusted to pH 4.0 with iminodiacetic acid) as eluant. Separation of C-DES from Polybuffer and transfer into 50 mM Mops/NaOH, pH 7.6, was achieved by gel filtration on Sephadex G-75.
All operations were carried out at 4°C except for chromatofocusing, which was performed at room temperature. The final C-DES preparation was stored at Ϫ70°C; its activity proved to be stable for several months.
Apoferredoxin to Holoferredoxin Conversion Assay-The standard assay for holoferredoxin formation, run at 30°C under argon, contained 10 M apoferredoxin, which was incubated with samples of deferredoxinized Synechocystis extract (or purified fractions thereof) in 55 mM Mops/KOH, pH 7.5, 9 mM glutathione, 0.15 mM pyridoxal phosphate, and 90 M L-cysteine. The reaction was started by the addition of Fe(NH 4 ) 2 (SO 4 ) 2 to a final concentration of 0.4 mM (final volume 30 l to 0.35 ml).
The reaction was stopped, routinely after 1 h, by the addition of 1.7 mM EDTA and 10% glycerol. The holoferredoxin content of the sample was analyzed by native PAGE (20% resolving gel, Laemmli buffer system (24) with SDS omitted); the same result was obtained whether the sample was analyzed immediately or after freezing. The gels were stained with the carbocyanine dye "stains all" (4, 5, 4Ј, 5Ј-dibenzo-3, 3Ј-diethyl-9-methylthiacarbocyanine bromide) (25); the amount of holoferredoxin (0.1-1 g/lane) was estimated by comparison with known standards of ferredoxin.
Exact quantities of holoferredoxin were determined (after transfer into 50 mM Mops/NaOH, pH 7.6) by activity measurements made according to Ref. 26; the rate of cytochrome c reduction by NADPH via ferredoxin, which is mediated by ferredoxin-NADP reductase, was measured, and 1 g of ferredoxin in 0.5 ml gave a ⌬A 550 of 0.2 min Ϫ1 cm Ϫ1 under the conditions employed.
Where specified, substrate concentrations in the standard assay were varied as follows. Free sulfide (up to 0.4 mM), mercaptopyruvate (up to 5 mM), coenzyme A (0.5 mM), or thiosulfate (0.5 mM) was used instead of cysteine; the concentration of cysteine employed was varied from 20 M to 2 mM, and the concentration of Fe(NH 4 ) 2 (SO 4 ) 2 was varied from 20 M to 0.5 mM. ATP and NADPH (up to 5 mM each) were added to some reactions.
Quantification of Pyruvate Formation from Cysteine-The standard assay composition was modified as follows. 45 M [ 14 C]cystine (adjusted to 11,200 dpm/nmol), reduced in situ prior to the addition of apoferredoxin by a 100-fold excess of glutathione to yield [ 14 C]cysteine (5600 dpm/nmol) was employed. [ 14 C]Pyruvate formed during the reaction was quantified both by Dowex 50 WX8 passage and phenylhydrazine derivatization. Following the addition of 2 N HCl to a final concentration of 285 mM, a 30-l aliquot (13,000 dpm) of the assay mix was diluted to 1 ml with H 2 O and applied to a 1-ml column of Dowex 50 WX8 (H ϩ form), equilibrated with 5 mM HCl. The 14 C content of the total eluate, including a 1.5-ml wash with 5 mM HCl, was determined and valued as pyruvate. A further 30-l aliquot was mixed with 10 mol of unlabeled pyruvate; after the addition of HClO 4 (final concentration 75 mM), 50 mol of phenylhydrazine were added to give a final volume of 1 ml. The precipitated pyruvate phenylhydrazone was collected, and its specific radioactivity was determined. Each nmol of [ 14 C]pyruvate (in the sample) raised the value obtained by 560 dpm/mol.
Measurement of the Glutathione Content of Synechocystis Cells-About 50 mg (wet weight) of freshly grown Synechocystis cells were suspended in 0.5 ml of 0.1 M HClO 4 containing 10 mM EDTA and disrupted by vigorous shaking (7 min) with 500 mg of glass beads (125-200 m). The suspension was neutralized with 0.1 M KOH and centrifuged. Glutathione contained in the supernatant was determined using the kinetic glutathione reductase assay coupled to the reduction of 5,5Ј-dithiobis(2-nitrobenzoate) (27).
Electrophoresis-Native separations at pH 7-8 were performed using 8% resolving gels and the buffer system of Williams and Reisfeld (28). After electrophoresis (1.3 watts, 3 h) the gel was sliced in horizontal strips of 2 mm ϫ 7 cm, and proteins were eluted with gentle shaking overnight in 2 ml of 50 mM Mops/NaOH, pH 7.6, containing 1 mM dithiothreitol and 20 M EDTA.

Studies with Crude Extract: Resolving the Holoferredoxin
Formation Assay from the Complete Protein Set-To perform the apo-to holoferredoxin conversion assay, it was necessary to resolve the crude extract of Synechocystis wild type cells into two fractions: one containing specifically the abundant endogenous (holo)ferredoxin and the other comprising all remaining proteins. The strategy pursued is outlined in Fig. 1. The extract proteins were transferred into a high salt buffer (0.43 M chloride) and then applied to a DEAE-Sephadex A-25 column. The stringent buffer conditions and the small pore size of the gel matrix (exclusion limit 30 kDa) permitted essentially all proteins to pass through the column (Table I, Fig. 2). However, a very small amount of protein was retained, namely all of the ferredoxin and a few other polypeptides that could be detected by SDS-PAGE after elution from the column with 0.63 M chloride (Fig. 2, lane C). The ferredoxin in this fraction was purified to homogeneity and converted to the apo-form via acidic pre-cipitation. This preparation was used for studying holoferredoxin formation mediated by deferredoxinized extract.
Apo-to holoferredoxin conversion was analyzed routinely by native PAGE (Fig. 3) making use of the dramatic increase in electrophoretic mobility of ferredoxin upon acquisition of its [2Fe-2S] cluster. Deferredoxinized extract (up to 150 g of protein/lane) contained no other protein that could have interfered by migrating in a similar position (Fig. 3, lane 0). This method also displayed the redox status of residual apoprotein; the intact species, with its 6 cysteinyl residues, which was the only one that could be reconstituted, had the lowest mobility compared with the oxidized species, containing internal disulfide bonds (Fig. 3).
The assay composition was optimized for holoferredoxin formation with deferredoxinized Synechocystis extract as enzyme source. Basic features of the system were the absolute requirement for extract proteins and free ferrous ion for conversion to occur. All assays were performed under argon and with glutathione present as reductant to protect apoferredoxin from any oxidation. The addition of any further sulfur compound (cysteine, sulfide, mercaptopyruvate, coenzyme A, thiosulfate; see "Experimental Procedures") was not needed. However, later in purification L-cysteine was required (Table I), and sulfide (favored initially because of the findings reported in Ref. 15) as well as the other sulfur compounds tested was ineffective. With sulfide-supplemented assays, the order of the additions was critical; our standard assay protocol, with ferrous ion added last, did not result in any spontaneous chemical reconstitution.
The addition of pyridoxal phosphate, ATP, or NADPH had no significant effect. Substitution of deferredoxinized extract by a corresponding amount of the resuspended membranous pellet did not support apo-to holoferredoxin conversion.
It should be noted that the occurrence of glutathione (reported for several cyanobacteria; Ref. 29) in Synechocystis cells was verified; per gram wet cells, 1.8 mol of glutathione were found, corresponding to an intracellular concentration of about 4 mM.
Purification of C-DES from Deferredoxinized Extract-The purification procedure is outlined in Fig. 1 and detailed in Table I; included in the table is the dependence of holoferredoxin formation, mediated by the various protein fractions, on the presence of L-cysteine.
An initial approach to estimate the number and size of proteins involved in conversion was to perform a frontal analysis experiment; deferredoxinized extract was continuously applied to an Ultrogel AcA 44 column. Effluent fractions were assayed; holoferredoxin formation started from a K av value of 0.5. About the same K av value was determined from a zonal gel filtration run (Fig. 4); only protein(s) with a molecular mass of 40 Ϯ 5 kDa should therefore be involved.
Further purification employed ion exchange chromatography on DE 52 cellulose. However, recovery of conversion activity was rather poor (about 10%) without an extra sulfur compound added. Therefore, again the assay mixture was supplemented with the various sulfur compounds (see above). Only L-cysteine was found to restore holoferredoxin formation (recovery 60 -70%), whereby a concentration of 90 M proved to be sufficient; pyridoxal phosphate stimulated slightly, whereas ATP did not exert any effect. These properties were retained after chromatography on Q-Sepharose FF. Thus, by using ion exchangepurified C-DES preparations, the standard assay conditions with 90 M L-cysteine, 0.15 mM pyridoxal phosphate, and 0.4 mM Fe(NH 4 ) 2 (SO 4 ) 2 could be established. Nevertheless, the active fractions were still rather impure.
Therefore, Q-Sepharose fractions were carried through native PAGE or chromatofocusing. Proteins eluted from slices of native gels were tested for conversion and analyzed by SDS-PAGE, which revealed a 43-kDa polypeptide contained in the active eluates (data not shown). Detection of the same single polypeptide correlated with conversion activity recovered after separation of Q-Sepharose fractions by chromatofocusing on a Mono P column (Fig. 2, lane D; Table I). Since both strategies independently identified the same 43-kDa polypeptide, this protein was considered as the active component.  tion (Table I), implicating cysteine as the relevant sulfur source for this enzyme. The omission of pyridoxal phosphate from the standard assay mixture approximately halved the reaction rate, suggesting that C-DES is pyridoxal phosphate-dependent. From the literature, the following product patterns are known to result from cysteine cleavage by pyridoxal phosphate-dependent enzymes: sulfide, pyruvate, and ammonia (31) or sulfur and alanine (8).

Stoichiometry of the Apo-to Holoferredoxin Conversion Reaction Catalyzed by C-DES-With
For every 2 mol of cysteine utilized, approximately 1 mol of holoferredoxin and 2 mol of pyruvate were produced (Fig. 5); residual cysteine was quantified after derivatization with vinylpyridine (data not shown), and ammonia was not measured. To obtain these quantitative data, holoferredoxin activity was determined by the cytochrome c reduction assay. The fate of cysteine was traced by employing the 14 C-labeled compound. [ 14 C]Pyruvate formed was distinguished from alanine by passage through Dowex 50 at pH 2 and further identified by coderivatization with unlabeled pyruvate using phenylhydrazine (Fig. 5). During the time course of the reaction, 37 nmol of ferredoxin were formed per nmol of C-DES employed (Fig. 5).
Therefore, C-DES serves as catalyst that mediates the following overall reaction.
Reactivity of C-DES with L-Cysteine and L-Cystine as Alternative Substrates-Having established the stoichiometry of the complete reaction system (Reaction 1), we investigated whether cysteine was processed by C-DES in the absence of apoferredoxin. In fact, pyruvate formation occurred at the same rate as before (28 nmol min Ϫ1 mg Ϫ1 with 90 M cysteine; see Figs. 5 and 6). In this context we also examined reactions with L-cystine included instead of cysteine (glutathione omitted; Fig.  6). Surprisingly, pyruvate formation occurred readily; with 45 M cystine instead of 90 M cysteine an approximately 15-fold higher rate was observed. When the reaction was run until substrate depletion, finally 2 mol of pyruvate were yielded per mol of cystine employed. The sulfur-containing product, possibly H 2 S 2 , was not characterized. In light of these findings, C-DES should more properly be considered as a cysteine/cystine C-S-lyase.
The question of whether cystine is also a better substrate in the holoferredoxin formation assay could not be investigated, since in the presence of excess glutathione (required for apoferredoxin protection) cystine is predominantly reduced to cysteine.
Structural Properties of C-DES-The molecular mass of C-DES was determined as 40 Ϯ 5 kDa by gel filtration (Fig. 4) and as 43 Ϯ 2 kDa by SDS-PAGE (Fig. 2). These data identified the enzyme as a monomer.
Stimulation of C-DES activity by the addition of pyridoxal phosphate in context with the ␤-elimination reaction observed with either cysteine or cystine as substrate was taken as evidence for C-DES being a pyridoxal phosphate-dependent enzyme whereby about 50% of the molecules retain their coenzyme after purification. Given the small amount of protein available, this assumption could not be directly tested by chemical or spectroscopic analyses. However, NaBH 4 treatment of C-DES (10 mM NaBH 4 for 1 h at 30°C, pH 7.5-7.2) resulted in complete inactivation of the pyridoxal phosphate-loaded enzyme, indicating susceptibility to Schiff base reduction (32).
NifS protein (producing sulfur and alanine from cysteine) has been reported to be particularly sensitive to inactivation by various thiol-alkylating agents including unsaturated amino acid derivatives (8,33). We therefore tested the susceptibility of C-DES toward N-ethylmaleimide (0.1 mM), iodoacetamide (1 mM), L-allylglycine (5 mM), and L-vinylglycine (5 mM). After an incubation period of 30 min at 30°C, C-DES was separated from excess reagent and assayed for apo-to holoferredoxin conversion. Its activity proved to be virtually unimpaired. A further experiment was conducted employing 5 mM L-propargylglycine, which is an irreversible inhibitor of ␥-cystathionase (34). Again C-DES activity remained stable. DISCUSSION Based on a self-contained assay system derived from the cyanobacterium Synechocystis we pursued a systematic search for a protein capable of directing [2Fe-2S] cluster assembly of ferredoxin. Purification identified a monomeric, 43-kDa, pyridoxal phosphate containing lyase named C-DES, which specif-ically and efficiently used L-cysteine as precursor for cluster sulfide.
From our purification results, the cellular concentration of ferredoxin is estimated to be 60 -120 M, and the concentration of C-DES is estimated to be 0.15-1 M. The resulting molar ratio of about 150:1 is of the same order as that employed in our assays (e.g. 120:1 for the experiment outlined in Fig. 5). Taking into account the doubling time of about 10 h under the growth conditions employed, Synechocystis cells synthesize approximately 1 mg of ferredoxin/30 g of wet cells/h. The endogenous ferredoxin [2Fe-2S] cluster-forming activity detected by our assay was more than sufficient to meet the cells' requirements (Table I).
In the initial purification stages, the addition of L-cysteine was not required for in vitro cluster formation. Obviously, some enzyme component(s) of the crude extract supplied the sulfur source, probably by utilizing glutathione. Free sulfide did not exert any effect in our system at any stage of purification, which is in contrast to the chloroplast system (15). A further difference is the lack of any requirement for ATP.
The product pattern resulting from L-cysteine is identical to that of cysteine desulfhydrase of Salmonella typhimurium (31). However, in contrast to C-DES the S. typhimurium protein is tetrameric and is a more typical catabolic enzyme, with its high specific activity (450 mol min Ϫ1 mg Ϫ1 with 2 mM cysteine as substrate; Ref. 35); furthermore, this protein was reported not to attack cystine (31). From a functional point of view, isolation of a NifS-like protein might have been expected, especially when considering the widespread occurrence of NifS-type sequences (8,12), which has recently been extended to Synechocystis: analysis of the total genome sequence of strain PCC 6803 identified three homologues (36). Indeed, C-DES and NifS are both pyridoxal phosphate-containing enzymes, degrading L-cysteine with similar rates (89 nmol min Ϫ1 mg Ϫ1 for NifS with 0.5 mM cysteine (8), 28 nmol min Ϫ1 mg Ϫ1 for C-DES with 90 M cysteine). Both catalyze cysteine cleavage in the absence as well as in the presence of apoprotein; therefore, the release of sulfur from cysteine and its insertion into the Fe-S cluster do not seem to be necessarily coupled processes with these enzymes. However, it should be emphasized that cysteine sulfur released by C-DES was quantitatively found assembled in the [2Fe-2S] cluster of ferredoxin (see Fig. 5), indicating interaction between C-DES and the apoprotein substrate. Distinct from NifS, C-DES is monomeric, produces pyruvate, ammonia, and sulfide instead of alanine and sulfur, and cannot be inhibited by thiol-alkylating reagents.
A striking feature of C-DES is its reactivity with cystine, which finally yields two equivalents of pyruvate. This reaction is known as a secondary activity of certain ␤and ␥-cystathionases (37,38), which are all homooligomeric enzymes. Including cysteine in the presence of excess glutathione in our reaction medium, we initially thought of this reduced compound as the primary sulfur donor species for [2Fe-2S] ferredoxin formation. By the increased reactivity of C-DES with cystine, we became aware of the possibility that the equilibrium concentration of cystine during conversion might be crucial. ␤-Elimination starting with cystine should yield cysteine persulfide as substrate-derived sulfane carrier compound. Whether this persulfidic intermediate is formed during catalysis by C-DES and is involved in sulfur transfer to apoferredoxin awaits further studies. Work on such mechanistic problems will be greatly aided by the availability of substantial amounts of purified C-DES. Cloning, sequencing of the gene, and overexpression of C-DES in E. coli are in progress. Based on the sequence information, targeted inactivation should reveal the physiological consequences of a defective C-DES gene in Synechocystis. Us- FIG. 6. Cyst(e)ine dependence of the ␤-elimination reaction catalyzed by C-DES. Reactions were run in the absence of apoferredoxin, and the [ 14 C]cystine concentration was varied as indicated. The data set for cysteine (q) was obtained using cystine and 9 mM glutathione in the reaction mixtures; the data for cystine (E) were obtained by omission of glutathione from otherwise identical reactions. The rate of [ 14 C]pyruvate formation was analyzed by Dowex 50 passage (see "Experimental Procedures"). Less than 15% of the cyst(e)ine was processed when the samples were stopped for analysis.
ing this approach we hope to examine the in vivo role of C-DES in ferredoxin Fe-S cluster formation.