Biosynthesis of Cysteine

NICHOLAS M. KREDICH

[SECTION EDITOR: GEORGES COHEN]

Posted January 24, 2008

Departments of Medicine and Biochemistry, Duke University Medical Center, Durham, NC 27710

Mailing address: Departments of Medicine and Biochemistry, Duke University Medical Center, Durham, NC 27710. Phone: 919-684-5318, Fax: 919-684-8358, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

The synthesis of l-cysteine from inorganic sulfur is the predominant mechanism by which reduced sulfur is incorporated into organic compounds. In plants and many microorganisms, including Salmonella enterica serovar Typhimurium and Escherichia coli, inorganic sulfate, the most abundant source of utilizable sulfur in the aerobic biosphere, is taken up and reduced to sulfide, which is then incorporated into l-cysteine in a step that is equivalent to the fixation of ammonia into glutamine or glutamate. l-Cysteine, in turn, is used for protein and glutathione synthesis and serves as the primary source of reduced sulfur in l-methionine, lipoic acid, thiamin, coenzyme A (CoA), molybdopterin, and other organic molecules.

Sulfate reduction by serovar Typhimurium and E. coli is assimilatory, i.e., producing only sufficient sulfur for biosynthetic purposes, and is regulated by a system designated the cysteine regulon. The assimilatory pathway differs in several ways from dissimilatory sulfate reduction, which is carried out by certain anaerobic microorganisms as part of a respiratory pathway that utilizes sulfate as a terminal electron acceptor and produces large quantities of sulfide as an end product (199). Since sulfate uptake and reduction require a large number of transport and enzyme activities, most of the machinery of cysteine biosynthesis is dedicated to sulfide synthesis. However, if sulfide is available in the environment, l-cysteine biosynthesis is a relatively simple two-step process requiring conversion of l-serine to O-acetyl-l-serine, which then reacts with sulfide.

Two additional mechanisms for sulfur fixation have been described in serovar Typhimurium and E. coli. The first occurs through the reaction of thiosulfate with O-acetyl-l-serine to form the thiosulfonate S-sulfocysteine, which is then reduced to l-cysteine (174). This pathway, first described in Aspergillus nidulans (176), is utilized for aerobic growth on thiosulfate and perhaps during anaerobic growth on sulfate. The second mechanism involves the reaction of O-succinyl-l-homoserine with sulfide to form homocysteine in a reaction catalyzed by cystathionine γ-synthase (63, 235). This reaction is probably physiologically insignificant owing to its requirement for high sulfide concentrations, and even when such concentrations are present, the reaction does not totally satisfy cellular sulfur requirements, because serovar Typhimurium and E. coli cannot utilize the sulfur moiety of l-homocysteine for l-cysteine biosynthesis. In yeast, however, the synthesis of l-homocysteine from succinyl-l-homoserine and sulfide represents a major if not the sole mechanism of sulfur fixation (25).

BIOSYNTHETIC PATHWAY AND GENETICS

Sulfur Utilization

Serovar Typhimurium and E. coli can utilize a number of inorganic organic compounds such as sulfate, sulfite, thiosulfate, and sulfide. Organic compounds containing reduced sulfur, including l-cysteine, l-cystine, glutathione, l-djenkolate, lanthionine, isethionate, and l-cysteine sulfinic acid also serve as sole sulfur sources. E. coli has a d-cysteine desulfhydrase, which allows it to grow on d-cysteine (241). Although l-methionine cannot be converted to l-cysteine, it does fulfill 50% of the total sulfur requirement and will support growth of some leaky cys mutants. Liquid medium requires approximately 70 μM sulfur to obtain a density of 10⁹ cells per ml (128). "Sulfur-free" medium E (263), in which MgCl₂ is substituted for MgSO₄, supports growth to 10⁸ cells per ml, implying the presence of about 7 μM sulfate as a contaminant from other components.

In the absence of sulfate and cysteine, certain alkanesulfonates also serve as sulfur sources for cysteine biosynthesis in E. coli but not in serovar Typhimurium (122, 255, 256, 258). Unlike some other gram-negative bacteria, neither can utilize arylsulfonates or sulfate esters as sulfur sources (122, 258), although serovar Typhimurium expresses an arylsulfatase when grown under special conditions in the presence of tyramine (76).

The l-cysteine biosynthetic pathway is shown in Fig. 1A and Fig. 1B, and the genes encoding its activities are listed in Table 1 together with references pertaining to their isolation and characterization. Most cys mutants are defective in one of the steps of sulfate reduction (Fig. 1A) and are in effect sulfide auxotrophs, which can be further distinguished by their responses to forms of reduced inorganic sulfur such as thiosulfate and sulfite (Table 2) (31). cysE mutants cannot synthesize the l-cysteine precursor O-acetyl-l-serine and will not grow on any form of inorganic sulfur unless this product is provided (Fig. 1B). cysB mutants lack a specific transcription activator and are deficient in expression of the entire sulfate reduction pathway, but they do grow on sulfide. Since there are two O-acetylserine (thiol)-lyase isozymes for the reaction of O-acetyl-l-serine and sulfide, mutants lacking only one activity, i.e., cysK or cysM , are still Cys⁺, but cysK cysM strains do not grow on sulfide or on O-acetyl-l-serine. cysM strains cannot utilize thiosulfate because they lack the O-acetylserine (thiol)-lyase isozyme that also catalyzes a reaction between O-acetyl-l-serine and thiosulfate. For unknown reasons, these strains are also cysteine bradytrophs when grown anaerobically on sulfate (61) (see below).

Fig. 1ABiosynthesis of l-cysteine in serovar Typhimurium and E. coli. (A) Sulfate uptake and reduction to sulfide; uptake of taurine and other alkanesulfonates and their reduction to sulfite.

Fig. 1BBiosynthesis of l-cysteine in serovar Typhimurium and E. coli. (B) Activation of l -serine to O-acetylserine and conversion to l-cysteine either directly through reaction with sulfide or indirectly by reacting with thiosulfate to give S-sulfocysteine, which is then converted to l-cysteine.

Table 1Genes of cysteine biosynthesis and metabolism and of anaerobic sulfur metabolism

Table 2Nutritional characteristics of cysteine auxotrophs

Selection for cys Mutants

Chromate resistance provides positive selection for sulfate transport mutants, i.e., cysPUWA (193). Azaserine resistance gives sulfide auxotrophs and cysK mutants and will give cysM mutants in a cysK background (94, 95). Selection for Leu⁺ in a serovar Typhimurium leu-5OO strain can generate Δ(topA cysB) mutants (168). Resistance to 1,2,4-triazole has been used to obtain cysK mutants, constitutive cysB mutants, and a promoter-up cysE mutant (92, 93, 96, 240). cysK mutants can be scored by their appearance as white colonies on bismuth ammonium citrate agar (62). Members of one subclass of serovar Typhimurium 1,2,4-triazole-resistant cysK mutants, originally designated trzB, are unstable and may result from a reversible transposition of cysK from the chromosome to an autogenous plasmid (92). cysB^c mutants have also been selected from strains carrying lac fused to a cys promoter (85). Both cysB and cysE mutants can be obtained by selection for resistance to the antibiotic mecillinam in serovar Typhimurium (182).

Inorganic Sulfur Transport

Sulfate and thiosulfate uptake in E. coli and serovar Typhimurium are achieved through a single periplasmic transport system (50, 119) that utilizes two different but similar periplasmic binding proteins (86, 238). Sulfate uptake by intact cells has a K_m of 36 μM and is inhibited by sulfite, selenate, chromate, and molybdate, which are presumably substrates for sulfate-thiosulfate permease (50). Kinetic studies indicate that selenate and selenite share a single transporter with sulfate (144), but molybdate also has a separate transport system (70, 135, 215, 228). In vivo binding of sulfate to the sulfate binding protein occurs with a K_d of approximately 0.1 μM and is inhibited by chromate but not by thiosulfate or molybdate (192). Inhibition studies suggest that sulfite can be transported by sulfate-thiosulfate permease, but the ability of mutants lacking sulfate and thiosulfate permease to grow on sulfite indicates the existence of some other mechanism as well. Sulfide uptake has not been studied but may occur by diffusion, in a manner similar to that of water uptake.

Components of the sulfate-thiosulfate permease are encoded by the contiguous genes cysP, cysU, cysW, and cysA and by the unlinked gene sbp (75). cysU was originally designated cysT, but the latter is now reserved for a cysteinyl-tRNA gene in E. coli (65, 125, 156). Sulfate uptake in E. coli may also require another gene, cysZ, which is closely linked to cysK (194). An open reading frame, possibly that of cysZ, has been found immediately upstream of cysK (21), but deletion of this region in serovar Typhimurium does not impair sulfate utilization (94).

The cysPUWAM clusters from E. coli (36, 236) and serovar Typhimurium (94) encode two ABC-type periplasmic transport systems (2) in which CysU and cysW are homologous peptides that probably span the membrane and form a channel for the passage of sulfate, thiosulfate, and related anions (236). CysA is presumed to be a membrane-associated peptide, which is homologous to the nucleotide binding peptides characteristic of this class of transport systems. These three genes probably represent the three complementation groups formerly recognized as cysAa, cysAb , and cysAc in serovar Typhimurium (164). sbp and cysP encode the sulfate and thiosulfate periplasmic binding proteins (75, 86). Serovar Typhimurium sulfate binding protein is a 32-kDa monomer that has been purified, sequenced, and characterized by X-ray crystallography (105, 192, 201). The sequence deduced from E. coli sbp predicts an expected 19-amino-acid hydrophobic signal peptide (75). Thiosulfate binding protein sequences deduced from the E. coli and serovar Typhimurium cysP genes are 45% identical to Sbp and include 25-residue signal peptides.

The CysP thiosulfate binding protein appears to have little affinity for sulfate in vitro, but its mutational loss in E. coli results in only a 50% decrease in growth rate at 0.02 mM thiosulfate and a sevenfold reduction in the rate of uptake at 0.01 mM thiosulfate (86). Furthermore, loss of the Sbp sulfate binding protein still allows growth on either sulfate or thiosulfate, implying a functional overlap between Sbp and CysP (238). Only the loss of both binding proteins disrupts utilization of the two sulfur sources.

Organic Sulfur Transport

Little is known about l-cysteine transport in enteric bacteria, but its sensitivity to osmotic shock in E. coli implies the participation of a periplasmic binding protein (45). Glutathione transport in E. coli requires the activity of the ybiK gene, renamed spt for "sulfur peptide transport" (195). Alkanesulfonate uptake will be discussed below.

l-Cystine transport in serovar Typhimurium is mediated by three different systems designated CTS-1, CTS-2, and CTS-3 (5). CTS-1 is a saturable system with a K_m of 2 μM and a V_max of 9 nmol/min/mg of protein in sulfur-limited cells. Loss of this activity in osmotically shocked cells suggests participation of a periplasmic l-cystine binding protein, which has been characterized from E. coli (11, 19) and identified as the product of fliY in both E. coli (173) and serovar Typhimurium (100). Genes for other components of these transport systems have not been identified.

CTS-1 is regulated as part of the cysteine regulon and repressed by growth on l-cystine. Constitutive expression of this activity in a cysB^c regulatory mutant results in l-cystine sensitivity (5), most likely because of the inhibitory effects of l-cysteine on metabolic processes such as homoserine biosynthesis (44, 73). CTS-2 is also a saturable system with a K_m of 0.1 μM and a V_max of only 0.3 nmol/min/mg of protein. It is not known to be part of the cysteine regulon. CTS-3 is an unsaturable system with a capacity of 0.04 nmol/min/mg of protein per μM l-cystine and may represent passive diffusion. Mutational loss of CTS-1 or repression by growth on l-cystine does not impair l-cystine utilization under usual laboratory conditions, which employ 0.1 to 1 mM l-cystine, because the activity of CTS-3 alone can satisfy the sulfur requirement of 3.6 nmol of sulfur per min/mg of protein in cells growing with a doubling time of 50 min (128).

E. colihas two kinetically distinct uptake systems for l-cystine, one of which one is shared with diaminopimelic acid and several l-cystine analogs, while the other is more specific (11). Both are associated with periplasmic binding proteins, one or perhaps both of which are FliY. Although the regulation of these systems has not been studied in detail, loss of the less specific system by growth on rich medium suggests that it may be part of the cysteine regulon.

Sulfate Activation

Enzymatic reduction of sulfate requires its prior activation to a phosphosulfate-mixed anhydride, which in E. coli and serovar Typhimurium occurs through the ATP sulfurylase-catalyzed reaction of sulfate with ATP to give adenosine 5'-phosphosulfate (APS) and PPi (214). A second enzyme, APS kinase, phosphorylates APS with another ATP to give 3'-phosphoadenylsulfate (PAPS) (213), which is then enzymatically reduced by PAPS sulfotransferase (also known as PAPS reductase) to sulfite. ATP sulfurylase is encoded by cysD and cysN, and APS kinase is encoded by cysC with all three genes located in the cysDNC cluster (140, 141, 165).

The extremely low equilibrium constant (10^–8) of the reversible ATP sulfurylase reaction in the forward direction (214) initially led to the assumption that sulfate activation must be driven through the efficient hydrolysis of PPi combined with the conversion of APS to PAPS. Subsequently, the E. coli enzyme was discovered to contain two nonidentical subunits, a 35-kDa catalytic subunit encoded by cysD and a 53-kDa peptide encoded by cysN, which has a deduced sequence homologous to those of GTP binding proteins such as RAS and E. coli EF-Tu (140, 141). The native enzyme is thought to contain four each of the two different subunits (145). GTP stimulates the rate of APS synthesis as much as 116-fold with an apparent K_m of 19 μM and is hydrolyzed in the process (139). When coupled with APS synthesis, GTP hydrolysis lowers the apparent K_eq for the overall reaction by a factor of about 10⁵ (138, 146). The enzyme mechanism has been studied extensively (146, 202, 244, 266) and involves the GTP-dependent formation of an AMP-enzyme reaction intermediate, which then undergoes nucleophilic attack by sulfate to give APS (145).

E. coliAPS kinase has been purified to homogeneity (220, 226) and characterized kinetically (221). It is composed of identical 22-kDa peptides encoded by cysC (140, 141, 219). The peptide subunit is phosphorylated by ATP at Ser-109 (219) and exists as a dimer in the phosphorylated state and mostly as a tetramer in the unphosphorylated state (220). The phosphorylated enzyme probably represents an intermediate on the reaction pathway and can donate its phosphate either to APS to give the product PAPS or to ADP to re-form the substrate ATP.

PAPS Reduction

Although many organisms synthesize PAPS as a donor for organic sulfate ester formation (a process not known to occur in serovar Typhimurium and E. coli), reduction of this compound is confined to plants and microorganisms (39, 48). E. coli PAPS sulfotransferase is a homodimer of 28-kDa subunits (132) encoded by cysH in the cysJIH operon (51, 131, 190). Thioredoxin is believed to serve as the physiologic reductant, but trxA mutants, which lack this factor, are Cys⁺ because they can substitute another reductant, glutaredoxin, which is encoded by grx (250, 251, 253). trxA grx strains are Cys^– and accumulate secondary mutations that prevent sulfate uptake or activation of sulfate to PAPS unless they are maintained on sufficient cysteine to repress these activities (216). A similar phenomenon has been observed in aged cultures of cysH mutants (68) and in cysQ mutants (178) and is thought to be due to PAPS toxicity. CysQ is postulated to participate in PAPS catabolism, and its deduced amino acid sequence suggests it is a monophosphatase (178, 200, 272).

The enzyme mechanism for PAPS sulfotransferase involves transfer of a sulfo moiety from PAPS to one of the two redox sulfhydryl groups of thioredoxin to give an organic thiosulfate, thioredoxin-S-SO₃^– (225, 252). An enzyme-S-SO₃^– intermediate that transfers the sulfo group to thioredoxin (132) has been proposed, and the deduced amino acid sequences of the serovar Typhimurium and E. coli PAPS sulfotransferase show a single cysteine residue, which would have to be the acceptor site in such a mechanism (132, 190). Once formed, thioredoxin-S-SO₃^– rearranges to give free sulfite and oxidized thioredoxin, which is regenerated by thioredoxin reductase. Thioredoxin-S-SO₃^– may also be reduced to a hydrodisulfide (R-S-SH) before it donates a sulfide moiety to O-acetyl-l-serine (252), but such a mechanism would bypass NADPH-sulfite reductase, and it is clear that E. coli and serovar Typhimurium mutants lacking this enzyme do not grow on sulfate.

Sulfite Reductase

During aerobic growth, the reduction of sulfite to sulfide is catalyzed by NADPH-sulfite reductase (SiR) (234), and serovar Typhimurium mutants lacking this enzyme accumulate sulfite from sulfate, implying that sulfite is a normal intermediate in assimilatory sulfate reduction (51). SiR also exhibits nitrite reductase activity, which is of doubtful physiologic significance. Under anaerobic conditions serovar Typhimurium expresses another type of sulfite reductase, which is discussed below.

SiR contains nonidentical subunits, where α is a 66-kDa flavoprotein (SiR-FP) encoded by cysJ, and β is a 64-kDa hemoprotein (SiR-HP) encoded by cysI in the cysJIH operon (183, 190, 231). The stoichiometry of the purified holoenzyme was originally reported as α₈β₄, but more recent studies with holoenzyme reconstituted in vitro from individually purified subunits suggest an α₈β₈ structure (274). The role of SiR-FP is to accept electrons from NADPH and transfer them to SiR-HP, which then reduces sulfite. Although SiR-FP and SiR-HP are tightly associated in the holoenzyme, purified preparations of each have been obtained from E. coli by dissociation of the holoenzyme in urea (231), from expression vectors (59, 274), and from serovar Typhimurium by purifying one component from a mutant unable to express the other (233).

SiR-FP exists as an octamer in solution that contains both flavin dinucleotide (FAD) and flavin mononucleotide (FMN) and has substantial NADPH-cytochrome c reductase and other diaphorase-like activities (231). It transfers electrons in the sequence NADPH → FAD → FMN → SiR-HP (232). The deduced sequence is homologous to that of rat liver cytochrome P450 reductase (183, 203), which also contains FAD and FMN (108), and, indeed, purified SiR-FP has significant cytochrome P450c17 reductase activity (273). The FAD binding domains of both proteins appear to have been derived from the same ancestral gene as ferredoxin-NADPH reductase and NADH cytochrome C₅ reductase; the FMN binding domain is homologous to that of bacterial flavodoxins.

A 60-kDa fragment of E. coli SiR-FP, starting at alanine 52, has been overexpressed and purified as a monomer, suggesting that the N-terminal domain is required for polymerization (275). The purified fragment contains FMN and FAD and is fully functional, combining tightly with monomeric SiR-HP to give a simplified holoenzyme with 20% the activity of native SiR (274). A crystal structure for this fragment has been reported (69). A smaller 18-kDa fragment containing the FMN domain has also been characterized. It forms a stable complex with SiR-HP, and in its fully reduced state can transfer a single electron to the SiR-HP siroheme (24).

Whereas cytochrome P-450 reductase exists as a 77-kDa monomer that contains one molecule each of FMN and FAD, octameric SiR-FP obtained from purified E. coli holoenzyme was initially reported to contain only four FMNs and four FADs, implying that the subunits are arranged in such a manner that binding of one flavin precludes binding of a second to the same subunit (231). However, the stoichiometry found for SiR-FP purified from an expression vector was found to be 1.6 to 1.7 flavins per subunit, suggesting that it contains one FMN and one FAD per subunit (58).

Free SiR-HP is a monomer containing one Fe₄S₄ cluster and one siroheme, an unusual heme found thus far only in sulfite and nitrite reductases (32, 171, 172, 231, 262). The sequence of electron flow is flavoprotein → Fe₄S₄ cluster → siroheme → sulfite, and reduced methylviologen can be substituted for the flavoprotein and NADPH (232). The deduced E. coli and serovar Typhimurium sequences include four cysteine residues arranged in two clusters as Ala-Cys-X₅-Cys separated by 37 residues from Gly-Cys-Pro-Asn-Gly-Cys (190). X-ray crystallographic studies (40, 159) have shown that these four cysteines bind the Fe₄S₄ cluster, and spectroscopic studies indicate that one of these serves as a bridging ligand, which electronically couples the cluster to siroheme (27, 30, 120, 121). This arrangement of cysteines may be regarded as a "siroheme motif" and has been noted with variations in other siroheme enzymes, including an assimilatory sulfite reductase from Desulfovibrio vulgaris (248); the anaerobically expressed sulfite reductase from serovar Typhimurium (91); nitrite reductases of bacteria, fungi, and spinach (4, 123, 198); and a dissimilatory sulfite reductase from an extremely thermophilic archaeal species, Archaeoglobus fulgidus (43). Crystallographic studies have also been reported of SiR-HP in complexes with substrates, products and other ligands, and in different oxidation states (41, 42).

Synthesis of the siroheme component of SiR-HP is catalyzed by siroheme synthase, a multifunctional 52-kDa protein encoded by cysG, which catalyzes four successive reactions beginning with two S-adenosylmethionine-dependent methylations of uroporphyrinogen III to give precorrin-2, followed by the NAD⁺-dependent dehydrogenation to sirohydrochlorin, and finally the insertion of iron to give siroheme (242, 265). cysG is contiguous with nir genes involved in NADPH-nitrite reductase expression (198, 271).

CysG consists of two domains: the C-terminal CysG^A, which as an isolated fragment can carry out the first two methylation reactions of siroheme and cobalamin synthesis; and CysG^B, which contains a putative pyridine dinucleotide binding site and appears to be responsible for the dehydrogenation of precorrin-2 to sirohydrochlorin and its subsequent ferrochelation to siroheme (264, 269). Serovar Typhimurium cbiY encodes an activity that can substitute for CysG^B in a strain lacking CysG^B while still retaining CysG^A (210). In vitro studies suggest that CbiY is used primarily in cobalamin synthesis where it acts as a cobalt chelatase, but it can also serve as a ferrochelatase in siroheme synthesis.

Siroheme synthesis is a limiting factor for overexpression of NADPH-sulfite reductase from a plasmid, but this deficiency can be overcome by including a copy of cysG on the expression vector (271). Since sirohydrochlorin is also an intermediate in cobalamin synthesis, cysG mutants are pleiotropic, deficient not only in sulfite reductase and nitrite reductase, but also in cobalamin, which is synthesized by serovar Typhimurium under anaerobic conditions, but not by E. coli (112).

Alkanesulfonate Utilization

In E. coli taurine is taken up and metabolized by activities encoded by tauABCD, where tauABC specifies an ABC-type transporter (2), and tauD encodes an α-ketoglutarate-dependent dioxygenase, which gives sulfite as a product (54, 261). The ssuABCDE cluster is required for most alkanesulfonates other than taurine and includes another ABC-type transporter encoded by ssuABC and a two-component monooxygenase encoded by ssuD and ssuE (55, 259). Purified SsuD has been found to catalyze the FMNH₂ and oxygen-dependent conversion of an alkanesulfonate to sulfite and the corresponding aldehyde, while SsuE is an FMN reducing enzyme providing SsuD with FMNH₂ (55). A crystal structure of SsuD has been reported (53). The finding that both systems yield sulfite as a direct product is consistent with earlier reports that sulfonate sulfur utilization requires sulfite reductase, serine transacetylase and CysB (255, 256, 257).

Serine Transacetylase and Cysteine Synthase

Serine transacetylase catalyzes the acetylation of l-serine by acetyl-CoA to give O-acetyl-l-serine, the direct precursor of l-cysteine (127, 130). The enzyme is feedback inhibited by l-cysteine, thus providing kinetic regulation of this short branch of the pathway. Furthermore, mutant strains lacking serine transacetylase are defective in expression of the enzymes of sulfate reduction because O-acetyl-l-serine is the precursor of N-acetyl-l-serine, the inducer of the cysteine regulon (see below).

Serine transacetylase resides in a multifunctional complex termed cysteine synthase, which also contains O-acetylserine (thiol)-lyase-A, one of two isozymes catalyzing the synthesis of l-cysteine from O-acetyl-l-serine and sulfide (8, 127). Serine transacetylase and O-acetylserine (thiol)-lyase-A are encoded by cysE and cysK, respectively (62, 97, 113, 130), and have predicted subunit masses of 30 and 35 kDa (21, 47, 137, 239). The complex has a mass of approximately 309 kDa but readily forms aggregates that are two and four times that size (38, 127). The smallest form contains four O-acetylserine (thiol)-lyase-A subunits and six serine transacetylase subunits. Velocity sedimentation, light-scattering, and chemical cross-linking experiments show that serine transacetylase itself is a dimer of trimers (80), a structure also noted in crystallographic structures of the E. coli enzyme and Haemophilus influenzae enzymes, in which l-cysteine was found to bind to the serine substrate site (180, 204).

Ordinarily, serine transacetylase is much less abundant than O-acetylserine (thiol)-lyase-A and is isolated as the cysteine synthetase complex (127). The free enzyme is found, however, in mutants that overproduce serine transacetylase (96, 136) or lack O-acetylserine (thiol)-lyase-A (97). The Cys⁺ phenotype of the latter class of mutants indicates that the complex is not required for in vivo function of serine transacetylase. In vitro, O-acetyl-l-serine dissociates the cysteine synthetase complex into a serine transacetylase hexamer and two O-acetylserine (thiol)-lyase-A dimers, which can be resolved from each other by chromatography and then reassociated into the cysteine synthetase complex (127). The dissociation reaction has a K_0.5 for O-acetyl-l-serine of about 20 μM and is inhibited by sulfide. The physiologic role of the dissociation reaction is not known, but it is interesting that serine transacetylase may be physically associated with O-acetylserine (thiol)-lyase in plants as well (52). In experiments designed to determine whether the cysteine synthase complex releases O-acetyl-l-serine prior to its reaction with sulfide, Cook and Wedding (37) found only a slight decrease in the predicted lag time between O-acetyl-l-serine synthesis and reaction, which was consistent with release of this intermediate from the complex but also suggestive of a small kinetic advantage from increased local concentrations of O-acetyl-l-serine.

The kinetic mechanism for the free enzyme has been reported as bi-bi ping pong, with acetyl-CoA being added first with the release of CoA, and then l-serine being added with the release of O-acetyl-l-serine (136, 163). Others have described a random-order ternary complex reaction mechanism (81). Kinetic studies of the cysteine synthase enzyme are complicated by the formation of aggregates with different affinities and various degrees of positive cooperativity for acetyl-CoA (38). The substrate l-serine and the feedback inhibitor l-cysteine alter apparent K_m values for acetyl-CoA through effects on aggregation. At 0.1 mM acetyl-CoA, l-cysteine inhibits with an apparent K_i of about 10^–6 M (127, 130). Crystallographic studies have shown that l-cysteine competes with l-serine for binding (204), a conclusion also supported by microcalorimetry (79).

Loss of inhibition by l-cysteine has been noted in C-terminal point mutations that lead to cysteine overproduction (47, 247) and in an enzymatically active fragment lacking the C-terminal 20 amino acids (162), indicating an important contribution of this region to feedback control. The C-terminal 10 amino acids have also been found to be essential for cysteine synthase complex formation (161).

O- Acetylserine (Thiol)-Lyase and S-Sulfocysteine Synthase

Synthesis of l-cysteine from O-acetyl-l-serine and sulfide is catalyzed by two distinct O-acetylserine (thiol)-lyase isozymes designated -A and -B and encoded by cysK (21, 62, 97, 137) and cysM (95, 236, 237, 239). The deduced amino acid sequences of the two E. coli isozymes are 43% identical. Mutational loss of either does not appreciably affect aerobic growth on sulfate, but the -A isozyme probably plays the larger role because its activity is at least 10-fold higher under these conditions (95). O-Acetylserine (thiol)-lyase-B is more important for growth on thiosulfate and for anaerobic growth on sulfate (see below).

O-Acetylserine (thiol)-lyase-A is a dimer of 35-kDa subunits that exists both in the free form and complexed with serine transacetylase as cysteine synthase (126, 127). The enzyme exhibits wide substrate specificity and in lieu of sulfide can use several different nucleophiles, including 1,2,4-triazole (128), methyl mercaptan (8), 5-thio(2-nitrobenzoate) (246), cyanide (22), and azide (28, 191). Enzymatically active analogs of O-acetyl-l-serine include O-propionyl-l-serine, O-butyryl-l-serine, β-chloro-l-alanine (36), and azaserine (diazoacetyl-l-serine) (95). Reactivity with azaserine releases the toxic product diazoacetic acid, thus providing a method of positive selection for mutants deficient in either O-acetylserine (thiol)-lyase or sulfide synthesis (see above).

The structure and mechanism of O-acetylserine (thiol)-lyase-A have been extensively characterized by crystallographic, spectroscopic, and kinetic studies, which have been summarized by Rabeh and Cook (207). Each subunit contains one pyridoxal phosphate, which forms a Shiff base with the lysine residue at position 41 (211). The overall kinetic mechanism is bi-bi ping pong, with O-acetyl-l-serine binding first and displacing Lys-41 to form its own Shiff base with pyridoxal phosphate (35, 36, 246). Acetate is then released by β elimination, leaving a pyridoxal phosphate-bound α-aminoacrylate intermediate that undergoes nucleophilic attack by sulfide to form l-cysteine. K_m values for the free enzyme are 1.0 mM for O-acetyl-l-serine and 6 μM for sulfide (246). The enzyme is product inhibited by l-cysteine but only at concentrations that are too high (~3 mM) to be of physiologic significance. Formation of the cysteine synthase complex alters certain kinetic properties of O-acetylserine (thiol)-lyase-A, decreasing the V_max by 50% and increasing the apparent K_m for O-acetyl-l-serine about fourfold (127). Crystal structures have been reported for the enzyme with and without substrates and analogues (17, 167) and with chloride, which binds to an allosteric site at the dimer interface (18, 245), and with sulfate, which binds to the active site.

O-Acetylserine (thiol)-lyase-B is also a dimer of identical subunits, each with a predicted mass of 33 kDa (174). It contains pyridoxal phosphate and has a kinetic mechanism that is identical with that of the -A isozyme, but is not known to form a complex with serine transacetylase (9, 95). K_m values are 0.9 mM for O-acetyl-l-serine and 10 μM for sulfide, and the V_max is approximately 90% that of O-acetylserine (thiol)-lyase-A (246). O-Acetylserine (thiol)-lyase-B differs from the -A isozyme by its ability to use thiosulfate as a nucleophile, giving S-sulfocysteine (R-S-SO₃) as a product (174). The apparent K_m for thiosulfate in this reaction is 2.7 mM. The conversion of S-sulfocysteine to cysteine has not been characterized in serovar Typhimurium and E. coli but could involve hydrolysis to cysteine and sulfate or reduction by glutathione to cysteine and sulfite (270). The advantage of this relatively short pathway is that it obviates the need for sulfate reduction by directly incorporating the sulfane moiety of thiosulfate into an organic form that requires only a one- or two-electron reduction (depending on whether sulfate or sulfite are by-products) to form l-cysteine. The physiologic importance of this activity is questionable during aerobic growth on sulfate, since cysM mutants grow quite well, and thiosulfate is not recognized as an intermediate in sulfate assimilation by serovar Typhimurium and E. coli. cysM mutants do lack the ability to utilize thiosulfate, however (175), and are cysteine bradytrophs during anaerobic growth on sulfate, suggesting that thiosulfate may be an intermediate of sulfate reduction under these conditions (61) (see below).

REGULATION

l-Cysteine biosynthesis in serovar Typhimurium and E. coli ceases almost entirely when cells are grown on l-cysteine or l-cystine, owing to a combination of end product inhibition of serine transacetylase by l-cysteine and a gene regulatory system known as the cysteine regulon, wherein genes for sulfate assimilation and alkanesulfonate utilization are expressed only when sulfur is limiting. These two mechanisms are interdependent, since the inducer for transcription activation, N-acetyl-l-serine, is derived from the serine transacetylase product, O-acetyl-l-serine. This relationship provides communication between branches of the pathway, allowing l-cysteine to regulate genes for sulfate assimilation and thiosulfate uptake negatively by inhibiting serine transacetylase (Fig. 2). Sulfur limitation derepresses these genes by relieving end product inhibition of serine transacetylase with a resultant accumulation of O-acetyl-l-serine and N-acetyl-l-serine.

Fig. 2Interrelationships of control mechanisms for l-cysteine biosynthesis. l-Cysteine feedback inhibits synthesis of O-acetyl-l-serine, which is the precursor for the inducer N-acetyl-l-serine. Sulfide and thiosulfate also exert end product inhibition in their capacities as anti-inducers and by their abilities to react with and deplete O-acetyl-l-serine.

A second type of product regulation results from the ability of sulfide and thiosulfate to act as anti-inducers so that l-cysteine, sulfide, and thiosulfate are all negative regulators of the cysteine regulon (Fig. 2). Other sulfur compounds, such as sulfite and methionine, can partially down-regulate the system but most likely do so indirectly through their abilities to serve as readily utilized sulfur sources and to increase intracellular levels of sulfide and l-cysteine (267). APS also acts as a negative regulator for alkanesulfonate utilization in E. coli (see below).

Kinetic Regulation

End product inhibition of serine transacetylase by l-cysteine represents the major, and probably the only, physiologically significant form of kinetic regulation in the pathway. The efficiency of this mechanism is illustrated by the fact that a sixfold increase in levels of wild-type serine transacetylase does not affect expression of the cysteine regulon (96), but a mutant with a normal level of a feedback-resistant enzyme excretes l-cysteine up to a concentration of about 2.5 mM (47), which is about 36 times the amount required for growth to a cell density of 10⁹ cells per ml (128).

Both O-acetylserine (thiol)-lyase isozymes are inhibited by l-cysteine and sulfide but at concentrations that are too high to be of regulatory significance (246). In vitro studies have uncovered no evidence of significant kinetic inhibition within the sulfate reduction pathway or in thiosulfate uptake (50, 57), and a regulatory mutant that cannot repress the sulfate reduction pathway excretes large amounts of sulfide, implying a lack of kinetic inhibition in vivo (15). The inability to feedback inhibit sulfate uptake and reduction poses a potential problem of synthesizing excess sulfide after a sudden shift from sulfur limitation to sulfur abundance. It may be significant in this regard that ATP sulfurylase and APS kinase activities rapidly decay in cells depleted for O-acetyl-l-serine, thereby shutting off the pathway at an early step (128). This mechanism may be useful in preventing toxicity from PAPS accumulation (68, 178, 216).

Overview of the Cysteine Regulon

The cysteine regulon comprises those genes participating in l-cysteine synthesis and transport systems that either are regulated by sulfur availability or participate in this response. The regulon includes the genes for l-cystine, glutathione, sulfate, and thiosulfate uptake; sulfate activation and reduction to sulfide; both O-acetylserine (thiol)-lyase isozymes; and genes involved in alkanesulfonate utilization (Table 1). All are positively regulated by the LysR-type (77, 223) transcription regulator (LTTR) CysB and the inducer N-acetyl-l-serine, although in the case of ssu, CysB is required only indirectly (see below). Cbl ("CysB-like") encodes a second LTTR, Cbl, which is essential for tau and ssu expression (107, 258, 260). Since cbl expression is activated by CysB, it too is part of the cysteine regulon. Although required for sulfite reduction, cysG is not considered part of the cysteine regulon (271).

cysBand cysE belong to the cysteine regulon because they encode the specific transcription activator CysB and serine transacetylase, which synthesizes the immediate precursor of the inducer N-acetyl-l-serine. cysB itself is negatively autoregulated (12, 109, 189), but cysE is probably not regulated at all. cysB and cysE mutants are pleiotropic and are not derepressed for activities of the cysteine regulon by sulfur limitation (113, 114, 126), but this defect can be overcome in cysE strains by either O-acetyl-l-serine or N-acetyl-l-serine (187). O-Acetyl-l-serine also supports the growth of cysE mutants, but N-acetyl-l-serine does not, because it is not converted to O-acetyl-l-serine, which is an immediate precursor l-cysteine.

Sulfur limitation is a necessary condition for depression of the cysteine regulon (116, 126), not only because l-cysteine inhibits serine transacetylase and inducer synthesis, but also because sulfide and thiosulfate are anti-inducers, which prevent derepression of the pathway by a direct effect on CysB (87, 188). l-Cysteine (or l-cystine) does the same because it is degraded to sulfide by the inducible enzyme cysteine desulfhydrase (129). Maximal derepression occurs during growth on the limiting sulfur sources glutathione and l-djenkolic acid (51). Sulfate, sulfite, and thiosulfate give intermediate levels of derepression, except for the activities of alkanesulfonate utilization, which are repressed. Sulfide, l-cystine, and l-cysteine provide maximum repression (126, 196, 268). In E. coli, this order of sulfur sources correlates directly with intracellular l-cysteine and inversely with ATP sulfurylase and APS kinase levels (267). Wild-type and even cysB strains grow well on sulfide because repressed levels of the two O-acetylserine (thiol)-lyase isozymes are still sufficient for an adequate rate of l-cysteine biosynthesis (126).

Transcription Units

A total of 21 genes have been given the designation "cys." Additional genes that can participate in providing l-cysteine include fliY, sbp, ybiK (spt), dcyD, cbl, tauABCD, ssuEADCB, and those for the l-cystine transport system CTS-l (Table 1). Most encode activities directly required for cysteine biosynthesis, while others function indirectly in l-cysteine and inorganic sulfur metabolism and are not known to be part of the cysteine regulon. The latter group includes cysL, the designation for a class of Cys⁺ selenate-resistant mutants with mutations in or near cysPUWAMK that may have an altered sulfate permease that is selectively defective in selenate uptake (92); cysQ, which encodes a putative monophosphatase that protects against PAPS toxicity (178, 200, 272); cysS, the gene for cysteinyl-tRNA ligase (14, 58, 84); cysT, gene for cysteinyl-tRNA (65, 125); cysX, an open reading frame of unknown significance that overlaps E. coli cysE (249); cysZ encodes an unidentified protein required for sulfate transport in E. coli but not in serovar Typhimurium (21, 194); and cysG, which is required for siroheme synthesis but is transcribed from the nirB promoter as part of a cluster of genes involved in the anaerobic expression of nitrite reductase (153, 197, 198) as well as from a second FNR-independent promoter located within the nirC gene (72). Baseline expression of cysG provides sufficient siroheme for sulfate assimilation but can be limiting when attempting to overexpress active SiR-HP (271). Genes known to be members of the cysteine regulon comprise 13 or 14 different transcription units (Table 3). Unmapped genes for CTS-1 are also regulated as part of the cysteine regulon (5) and presumably constitute one or more additional transcription units.

Table 3Transcription units of the cysteine regulon

Not included in Table 3 are genes that allow serovar Typhimurium to convert several partially reduced inorganic sulfur compounds to sulfide during anaerobic growth (Table 1; see below). These include the asrABC genes for an anaerobic sulfite reductase that is distinct from the aerobic enzyme encoded by cysJ and cysI; the phsABC cluster encoding a thiosulfate reductase; and the ttrRSBCA cluster involved in tetrathionate reduction. These three systems are suspected to play a role in anaerobic respiration and are unlikely to be part of the cysteine regulon. Nonetheless, their expression provides sufficient sulfide under anaerobic conditions to overcome the Cys^– phenotype of certain serovar Typhimurium cys mutants.

cysPUWAMK Region.

Translational coupling of genes in the cysPUWA cluster (236) indicates that they are expressed as a single transcription unit from a promoter just upstream of cysP (86, 87). cysM is separated from cysA by only 174 bp in E. coli and may also be part of this operon, which is transcribed counterclockwise on the chromosome. cysK is separated from cysPUWAM, by about 5 kb, but is transcribed clockwise from its own promoter (21, 166). An open reading frame, which may correspond to E. coli cysZ (194), is located just upstream of cysK, but nothing is known about its product or regulation (21).

cysCND-cysHIJ Region.

The cysDNC cluster comprises a single operon with a promoter immediately preceding cysD (140, 141, 154). The nearby cysJIH cluster is a single operon with a promoter just upstream of cysJ (150, 151, 185, 187) and is separated from cysDNC by approximately 12 kb (98, 104, 106). The gene order is thyA → cysJIH iap → cysDNC in E. coli and thyA → cysCND → cysHIJ in serovar Typhimurium (46, 115, 140, 141, 183, 190), indicating that the cysCND-cysHIJ segment is inverted with respect to other chromosomal loci.

Other Transcription Units.

cysEis found in the order xyl mtl cysE ria-pyrE in both serovar Typhimurium and E. coli (13, 47, 157, 239). cysB is found in the order purB → pyrF → cysB topA trpA → aroD in serovar Typhimurium and purB → trpA-topA cysB pyrF → aroD in E. coli, in which the chromosomal region between 25 and 35 min is inverted (101). fliY, sbp, ybiK (spt), dcyD and cbl are probably regulated as single genes, while tauABCD and ssuEADCB represent multigene operons present in E. coli but not in serovar Typhimurium (258).

Molecular Aspects of the Cysteine Regulon

Transcription start sites for the cysJIH, cysK, cysPUWAM, and cysDNC operons contain −10 regions that conform to the consensus TATAAT (21, 86, 87, 160, 187). As observed with many other positively regulated promoters (208), the −35 regions show little or no resemblance to the consensus TTGACA. In vitro studies with the cysJIH, cysK, and cysP promoters have confirmed that they are inefficient at forming transcription initiation complexes without CysB and N-acetyl-l-serine. Activation of the tauA and ssuE promoters requires Cbl and is considered below.

CysB Protein, Inducer, and Anti-Inducers.

CysB is a homotetramer of 36-kDa subunits encoded by cysB (160, 186) with a predicted N-terminal helix-turn-helix DNA binding motif (16). Highly purified CysB has been characterized by in vitro binding, footprinting, and transcription studies, which have shown that CysB binds as a tetramer (88) just upstream of the –35 regions of the cysP, cysJIH, and cysK promoters and activates transcription in the presence of N-acetyl-l-serine (87, 166, 187). It also binds to the +1 region of the cysB promoter, where it acts as a repressor (189). Binding to these promoters occurs even in the absence of inducer and is qualitatively and quantitatively altered by N-acetyl-l-serine in ways that vary from one cys promoter to another.

Originally, O-acetyl-l-serine was thought to be the inducer of the cysteine regulon (116, 126), but subsequent studies have indicated that the true inducer is N-acetyl-l-serine (187). N-Acetyl-l-serine has been found to bind in vitro to the closely related Klebsiella aerogenes CysB protein with a dissociation constant of about 4 mM and a stoichiometry of 1 molecule per protein subunit (152). It now seems most likely that the activity of O-acetyl-l-serine noted in many experiments was due to its conversion to N-acetyl-l-serine by a nonenzymatic, intramolecular O- to N-acetyl migration, which occurs at a rate of about 1% per minute at pH 7.6 (63). It is not known whether an enzyme catalyzes this reaction in vivo, nor is anything known about N-acetyl-l-serine degradation. Owing to its availability and perhaps more efficient uptake, O-acetyl-l-serine has been used as a source of N-acetyl-l-serine in many studies. For this review the term "acetylserine" will be used unless a distinction between O-acetyl-l-serine and N-acetyl-l-serine is relevant.

The effects of the anti-inducers sulfide and thiosulfate are direct and have been noted in vivo with the cysDNC promoter in a cysK cysM strain unable to convert these sulfur sources to l-cysteine (181). Acetylserine affects in vitro DNA binding and transcription initiation at concentrations of 0.1 to 5 mM (87, 166, 187, 188), and sulfide and thiosulfate reverse these effects at concentrations ranging from 0.5 to 2 mM for sulfide and from 0.025 to 0.25 mM for thiosulfate. These anti-inducers do not affect the DNA binding that occurs in the absence of acetylserine (87, 188). Neither do they affect acetylserine binding to K. aerogenes CysB in vitro, suggesting that they act by preventing or altering conformational changes caused by bound inducer (152).

CysB Binding Sites in the cysJIH,cysK,cysP, and cysB Promoters.

Nine different CysB binding sites have been identified in these four promoters (Fig. 3) and several more in the cbl, tauA, and ssuE promoters (see below). The positively regulated promoters cysJ, cysK, and cysP have multiple sites, including an activation site located just upstream of the –35 region that is required for transcription activation (87, 166, 187, 188). Binding of CysB to these activation sites, designated CBS-J1, CBS-K1, and CBS-P1, is stimulated by acetylserine. The cysB promoter has a single repressor site designated CBS-B, which is centered at the transcription start site and binds with lower affinity in the presence of acetylserine, thereby modulating autoregulation (89, 189). Accessory sites are found both upstream and downstream of activation sites and include CBS-J2 and CBS-J3 in the cysJIH promoter, CBS-K2 in the cysK promoter, and CBS-P2 and CBS-P3 in the cysP promoter (Fig. 3). They are of unknown function but may serve to sequester CysB at cys promoters even when sulfur is replete and gene activation is not required. Acetylserine stimulates binding to CBS-J2 and CBS-P2 but inhibits binding to CBS-J3, CBS-K2, and CBS-P3 (87, 89, 166).

Fig. 3Binding sites for CysB in the positively regulated cysJ, cysK, and cysP promoters and in the negatively regulated cysB promoter. CBS-J1, CBS-K1, and CBS-P1 are activation sites and are required for transcription activation by CysB. CBS-B is a repressor site for cysB negative autoregulation, and the other sites are termed accessory sites. The inducer N-acetyl-l-serine stimulates CysB binding to the shaded sites and inhibits binding to the other sites. The upstream portion of CBS-J3 has not been characterized. Arrows indicate transcription start sites and promoter –35 positions.

Hydroxyl-radical footprints and nucleotide sequence alignments indicate that CysB binding sites are composed of 19-bp half-sites that vary in their spacing and relative orientations in patterns that can be correlated with the effects of acetylserine on CysB binding (89). Activation sites are composed of a pair of convergently oriented half-sites, designated a and b, which are separated by 1 bp in CBS-J1 and CBS-P1, and by 2 bp in CBS-K1 (Fig. 4). This spacing allows CysB to bind to one side of the DNA duplex over a span of about 40 bp. The same topology is found in CBS-J2 and CBS-P2, except that the half-site separation in CBS-P2 is 3 bp. The increased binding affinities of all five sites in the presence of acetylserine are thought to be a consequence of this convergent half-site arrangement.

Fig. 4Correlation between half-site organization, CysB-induced DNA bending, and effects of N-acetyl-l-serine. Half-sites are designated a, b, and, in one case, c, and are oriented as indicated. Half-site pairs, i.e., binding sites, with increased affinities in the presence of N-acetyl-l-serine are shaded. Bend points induced by CysB binding that are sensitive to N-acetyl-l-serine are indicated on the cysK and cysP promoters by vertical arrows.

Half-site arrangements in CBS-B and CBS-K2 are similar to each other but quite different from those of activation sites (89). Here, half-sites are divergently oriented and separated by one or two helical turns, i.e., 21 bp in CBS-B and 11 or 23 bp in CBS-K2. This arrangement is thought to be responsible for decreased binding affinity in the presence of acetylserine. As with activation sites, CysB binds to a single helical face in CBS-B and CBS-K2, but here it bends the DNA to give a larger footprint of approximately 60 bp. CBS-K2 actually contains three half-sites, designated K2a, K2b, and Kc (Fig. 4). Without acetylserine CysB binds to the upstream CBS-K2a half-site and the closer of two downstream half-sites, CBS-K2c, whereas in the presence of inducer, binding shifts one helical turn from CBS-K2c to the more distant downstream half-site CBS-K2b.

Binding to CBS-P3 is also inhibited by acetylserine (87), but in this case the effect is due to an upstream lateral movement of CysB from CBS-P3 to the overlapping CBS-P1 site (Fig. 3). CBS-P3 is actually a combination of the downstream half-site of CBS-P1 (CBS-P1b) and an extra half-site located between CBS-P1 and CBS-P2. This topology is thought to result in DNA bending that is altered by acetylserine.

CysB-Induced DNA Bending.

In the absence of acetylserine, CysB induces a 100° bend just upstream of CBS-K1 in the cysK promoter by binding as a single tetramer to the activation site and the CBS-K2c half-site (88, 166). This interaction is believed to enlist three different CysB subunits, each binding to a different half-site, and is prohibited by acetylserine, which reduces the bend angle to about 50° by causing CysB to bind exclusively to CBS-K1. A similar phenomenon has been noted for the cysP promoter where CysB binds to CBS-P1 and CBS-P3b in the absence of acetylserine inducing a 100° bend just downstream of CBS-PI (87, 88, 89), and acetylserine reduces the bend angle by causing CysB to bind only to CBS-P1. For both promoters, the arrangement of a half-site oriented toward an activation site but separated from it by a single helical turn results in CysB-induced DNA bending that is sensitive to acetylserine. The significance of such bending to transcription regulation is in doubt, however, because mutant promoters lacking CBS-K2 are not bent but appear to function normally in vitro and in vivo (123, 166). As noted above, bending of CBS-K2 and CBS-B involves different half-site combinations and is not sensitive to acetylserine.

Regulation of Alkanesulfonate Utilization.

The E. coli gene cbl encodes a "CysB-like protein" that is 40% identical with CysB and essential for expression of the E. coli tauABCD and ssuEADCB operons (107, 258). cbl is not found in serovar Typhimurium, which also lacks tau and ssu genes. K. aerogenes, however, contains a gene for a protein that is 86% identical with E. coli Cbl, which is located immediately downstream of the nitrogen assimilation gene nac, but not cotranscribed with nac (227). Cbl has been crystallized, but its molecular structure has not been reported (243).

Strains lacking cbl are unable to utilize alkanesulfonates as a sulfur source (260) but grow normally on inorganic sulfur, although OASS-B levels are reduced by about 60%, suggesting that, at least in E. coli, Cbl may participate in cysM regulation (107). cbl is considered part of the cysteine regulon because its expression is repressed by l-cysteine and derepressed by sulfur limitation; it requires cysB and perhaps acetylserine; in vitro studies have shown that binding of CysB to a cbl promoter fragment is stimulated by acetylserine (107).

tauABCDexpression requires both Cbl and CysB; the latter is required not only for cbl activation, but also directly for activation of tauABCD itself. In vitro studies have shown that Cbl binds to the tauA promoter at a single site located between positions –112 and –68 relative to transcription initiation (260). CysB binds to at least two regions located at –224 to –183 and at –120 to –11. The latter larger region may bind multiple CysBs. Acetylserine does not affect Cbl binding but does stimulate CysB binding to a site between –112 and –68.

In vivo expression of ssu genes also requires Cbl and CysB (260), but in this case CysB plays only an indirect role as the activator of cbl. When cbl is expressed from a CysB-independent promoter, ssuEADCB can be activated in a cysB background (20). Since the immediate product of alkanesulfonate reduction is sulfite, CysB is still required for sulfur utilization because sulfite reductase expression depends on CysB. In vitro studies confirm the role of Cbl, showing that it binds to the ssu promoter region between positions –75 and –31 and is sufficient for transcription initiation without CysB or acetylserine (20, 259). It has been suggested that the requirement of CysB for expression of the tau operon but not the ssu operon may be related to the "unfavorable" far upstream position of the tauE promoter Cbl binding site compared with that of ssuE (20).

Although not required for activation, CysB does bind to the ssuE promoter somewhere between positions –35 and +64, and this binding is stimulated by acetylserine and thiosulfate (259). The extension of the CysB binding site downstream of the transcription initiation start site may account for the finding that CysB down-regulates ssu expression both in vivo and in vitro (20, 259).

Whereas growth on sulfate lowers levels of the enzymes of inorganic sulfur assimilation by about 50% (presumably through its conversion to the anti-inducer sulfide), it almost totally eliminates tau and ssu expression, suggesting that sulfate or a metabolite of sulfate acts as a repressor (259). In vitro studies with the ssuEpromoter indicate that this negative coregulator is APS, which has been demonstrated to be a potent inhibitor of Cbl-dependent transcription initiation (20). No effects of sulfate, sulfite, sulfide, or N-acetylserine were noted, but thiosulfate did inhibit somewhat, consistent with the finding that thiosulfate inhibits Cbl binding to the ssuE promoter (259). APS was found not to affect Cbl binding, suggesting that it acts only as an inhibitor of positive control (20). In vivo studies support this role for APS, showing that sulfate does not prevent ssu expression in a cysN mutant unable to synthesize APS and that expression is almost totally eliminated in a cysC mutant unable to metabolize APS to PAPS (20).

An S149M cbl mutant analogous to the cysB constitutive mutant T149M (33, 34) was only partially repressed for ssu expression by sulfate in vivo, implying that this residue is also involved in the inhibitory effect of APS on activation by Cbl (20). In vitro binding by Cbl S149M and transcription initiation were not inhibited by thiosulfate as they are with wild-type Cbl, and inhibition of transcription initiation was only partially inhibited by APS. APS is presumed to mediate negative regulation of tau expression as well. Seiflein and Lawrence (229) have suggested that K. aerogenes also senses sulfate levels through APS and have opined that this might be mediated through a hypothetical regulatory protein, which they termed "MtcR"; Bytowski et al. (20) have further speculated that MtcR may be Cbl.

Mechanism of Transcription Activation by CysB.

CysB residues 19 to 44 include a helix-turn-helix motif common to LTTRs, which is presumed to bind to specific DNA regulatory sites. In accord with this model, single-amino-acid substitutions at E11, S20, T22, I33, S34, E41, and L44 have been found to impair in vivo and in vitro DNA binding (33, 147). Proteins lacking the C-terminal 5 to 30 amino acids also fail to bind, but it's not clear whether this is due to the lack of a specific binding determinant or simply the result of impaired oligomerization or protein instability (147). Studies with other LTTRs, e.g., AmpR (7), NahR (224), and OxyR (133), have also shown binding defects with substitutions in the helix-turn-helix motif and in the C-terminal region.

The turn between the two helices has been shown to be important for positive control, and changes at residues Y27, T28, and S29 prevent cysP promoter activation both in vivo and in vitro while leaving intact DNA binding and the in vitro effects of acetylserine on binding (147, 148). A Q30 mutant protein was similarly affected but to a lesser extent. These residues are believed to define an activating region "AR" for positive control of the cysP promoter and perhaps other cys promoters as well (148). The interhelical region of the LTTR GcvR may also be part of an AR for that regulator (117).

Mutations affecting oligomerization of CysB monomers have been characterized by analyzing "negative trans-dominance" as an indicator of mutant and wild-type subunit interactions. Such studies suggest that oligomerization requires residues in the N-terminal region at positions 41, 44, and 48, and somewhere in the C-terminal 16 amino acids (147). Similar conclusions regarding a role for the C-terminal region in oligomerization have been drawn from studies with OxyR (133) and NahR (224).

For many LTTRs, amino acid substitutions that allow regulatory function in the absence of effector, i.e., a constitutive phenotype, are frequently located in the center of the polypeptide, implying that this region is important in effector recognition and response (7, 23, 134, 158, 223, 224). This is clearly true for CysB. Residue T149 is very important in determining inducer response, and 9 of the 19 possible substitutions at this position are partially or totally constitutive for cysK and cysJIH expression during growth on l-cystine (33). Two highly purified mutant proteins, T149M and T149P, activate transcription from the cysK, cysJIH, and cysP promoters in the absence of inducer and are insensitive to the anti-inducers sulfide and thiosulfate (34). Insertions of as many as 14 amino acids are also tolerated at T149 without loss of activation function, and two such proteins are partially constitutive in vivo, suggesting that this region lies on a surface of the molecule. Other amino acid substitutions giving a constitutive phenotype have been found at residues W166 (34), Y164, Y197, and A227 (147). Substitutions at residues M160, T196, A244, and A247 are not constitutive but do affect response to inducer while retaining DNA binding activity (147).

The crystallographic structure of an amino-terminal-truncated CysB from K. aerogenes extending from residues 88 to 324 provides insights on the role and mechanism of action of inducer and anti-inducers (254). The structure shows a dimer in which each monomer contains two α/β domains connected by two short segments of polypeptide. The two α/β domains fold to form a central cavity lined by polar side chains, which contains a sulfate anion and five water molecules. The general structure is very similar to that of sulfate binding protein Sbp and other periplasmic binding proteins that contain a ligand cavity (206), as well as to the structure reported for similarly truncated OxyR from E. coli (26). N-Acetyl-l-serine can easily be modeled into the CysB cavity where it would displace the sulfate anion and two water molecules. Since thiosulfate and sulfide would also fit, the cavity has been proposed as the site at which inducer and anti-inducers interact with CysB (254). This conclusion is strengthened by the fact that residue T149 and most of the others implicated in inducer response are located on or near the cavity surface.

The CysB crystal structure may also provide an explanation for the unusually large footprints of CysB. The twofold axis of symmetry relating the two monomers is oriented perpendicular to the cofactor binding domains, giving an elongated structure that could place the two DNA binding regions (not included in the explicit structure but estimated by modeling) more than 125 Å apart, close to the 140 Å encompassed by two 19-bp half-sites (254). The crystal structure of another LTTR, CbnR, shows a tetrameric structure that includes the N-terminal DNA binding domains (170). Asymmetric subunit interactions in CbnR allow these domains to line up in a V-shape linear fashion that would permit binding to two adjacent DNA binding sites on the same side of the DNA helix giving a 60-bp footprint,

CysB W166R has been shown to be moderately constitutive in vivo for cysK expression, and low level constitutive for cysJIH expression, but deficient in cysDNC expression, giving a Cys^– phenotype on sulfate (34, 126, 141, 154). This residue is also situated near the proposed ligand cavity. In vitro studies have shown that purified CysB W166R binds to the cysJ, cysK, and cysP promoters, although with less affinity than wild type, and that such binding is unaffected by acetylserine (34). Bending of the cysK and cysP promoters by the mutant protein was also unaffected by inducer and was similar to that found with wild-type CysB in the absence of acetylserine. The mutant protein activated in vitro transcription from both promoters, but only at 2 mM Mg²⁺ and not at the 10 mM concentration commonly used in such studies. Expression was not stimulated by inducer. These findings suggest that in vivo unbound Mg²⁺ concentrations are closer to 2 mM than the reported total concentration of 9 mM (177), probably owing to binding by proteins, nucleic acids, and other phosphates.

At 2 mM Mg²⁺ wild-type CysB can also partially activate the cysK promoter in vitro in the absence of acetylserine at a level (about 25%) close to that found in vivo in a sulfur-limited cysE strain, which cannot synthesize inducer (34). In contrast to CysB W166R, in vitro activation was further stimulated by acetylserine. In vivo constitutive expression of cysK does not occur in a cysE cysB double mutant, indicating its dependence on CysB (125). In vitro cysJ expression with wild-type CysB at 2 mM Mg²⁺ still required acetylserine, just as it does in vivo in a cysE strain, indicating a fundamental difference between the cysK and cysJ promoters. This difference has been correlated with the topology of their activation sites, where half-sites are separated by 1 bp in the CBS-J1 and 2 bp in CBS-K1 (89). In vitro studies have shown that insertion of an extra base pairs between half-sites in CBS-J1 gives a CBS-K1-like phenotype where wild-type CysB can now activate transcription with or without acetylserine (34). Conversely, deletion of 1 bp between CBS-K1 half-sites produced an activation site that was strictly dependent on inducer for activation. These results suggest that one role for acetylserine may be to decrease the distance between the DNA binding domains of two the CysB subunits that interact with an activation site.

CysB Interactions with RNA Polymerase.

The upstream positions of the activation sites CBS-J1, CBS-K1, and CBS-P1 in their respective promoters suggest that CysB functions as a Class I transcription activator (102, 103) and imply that CysB makes contact with the C-terminal domain of the RNA polymerase subunit, αCTD. Such transcription activators are distinguished by the fact that they are inactive with RNA polymerase containing subunits with deletions or certain point mutations in αCTD (89, 277). Supporting this notion is the finding that two strains with such mutations, rpoA341 (αCTD K271E) in E. coli (67) and rpoA155 (αCTD L289F) in serovar Typhimurium (149), are cysteine auxotrophs or bradytrophs. Furthermore, defective induction of adi, the gene for arginine decarboxylase, in rpoA341 has been attributed to the finding that adi expression requires CysB (143, 230).

Interactions between CysB and RNA polymerase have been analyzed in detail by using a library of RNA polymerase α subunit mutants containing single-alanine substitutions in αCTD (148). A number of residues were identified between positions 255 and 325 that were important for in vivo activation of the cysP promoter with most of the severe defects found at positions expected to occur on the surface of αCTD. Residues comprising the "265 determinant" were found to be important for cysP expression and were assumed to be involved in binding to DNA near CBS-P1, as has been reported for other activators (10, 66, 169, 222). An analysis using a Lex-based two-hybrid system (49) for estimating contacts with αCTD in vivo showed that residues K271 and E273, belonging to the "273 determinant" used by the regulatory protein Fis at the rrnB P1 and rrnE P1 promoters (1), are required for CysB-αCTD interaction and also supported the conclusion that the 265 determinant is not (148). Furthermore, the interaction was significantly weakened in a CysB Y27A mutant, implying contact between the CysB AR region and the 273 determinants of αCTD.

The finding that the rpoA341 (K271E) mutation negatively affects cysP promoter activity but not the activity of the cysJ and cysK promoters has been interpreted to mean that perhaps only the cysP promoter is class I (cited as unpublished in reference 148). This distinction is consistent with the fact that CBS-P1 is better situated upstream to be a class I promoter, extending from promoter positions –85 to –45 whereas CBS-J1 and CBS-K1 lie 11 to 12 bp further downstream and slightly overlap the –35 region. Unpublished data (cited as unpublished in reference 148) indicate, however, that the CysB AR region defined in studies with the cysP promoter is also required for cysJ promoter activation.

NADPH-sulfite reductase and O-acetylserine (thiol)-lyase levels in serovar Typhimurium are lowered by the DNA gyrase inhibitors nalidixic acid and novobiocin, suggesting that expression of these activities is sensitive to the state of DNA superhelicity (185). It is not known whether this effect involves cysB expression or the interaction of CysB with the cysK and cysJIH promoters. A number of prokaryote promoters, including those that depend on catabolite repressor protein, are known to be sensitive to DNA superhelicity (217), but there is no evidence to suggest that catabolite repressor protein is involved in expression of the cysteine regulon.

The Cysteine Regulon and Global Control

Given the role of reduced organic sulfur in many different biologic processes, the cysteine regulon can be considered a global control system that might be expected to interact with other metabolic processes. One example of such an interaction is the apparent requirement for CysB in the efficient induction of the E. coli adigene, as noted above (230). In addition, a functional cysPUWAM operon has been reported to be required for the anaerobic induction of aidB, a component of the adaptive response to alkylation damage (155). The significance of this finding is unknown, and mutations in genes required in the synthesis of APS and PAPS have no such effect on aidB expression.

cysBmutations have also been observed to affect utilization of certain carbon sources, giving decreased levels of activities associated with glucitrol and alanine utilization (205). The mechanism is unclear since these effects were partially reversed by both a good sulfur source, l-cysteine, and a limiting sulfur source, djenkolate. cAMP also largely reversed the effect, suggesting that adenylate cyclase may be responsive to sulfur limitation.

The cysteine regulon has been implicated in the development of resistance to the antibiotic mecillinam in serovar Typhimurium (182). Most cysE and cysB mutants show antibiotic resistance, which can be overcome by acetylserine or l-cysteine in cysE strains and by thiosulfate, sulfite, sulfide, or l-cysteine in cysB strains. The role of CysB and acetylserine in this phenomenon is not known. Resistance to the antibiotic novobiocin has also been linked to the cysteine regulon with increased resistance found in cysB and cysE strains (209). In this case CysB has been found to negatively regulate the gene hslJ, which is required for novobiocin resistance, but not for mecillinam resistance (142). CysB appears to act as a repressor of transcription by binding to the hslJ promoter, and binding is stimulated by acetylserine (118).

ANAEROBIC SULFUR METABOLISM

Mechanisms for sulfate uptake, activation, and reduction to sulfite as well as a requirement for O-acetyl-l-serine appear for the most part to be the same for aerobic and anaerobic growth. There are several exceptions including the finding that serovar Typhimurium cysB mutants are Cys⁺ when grown anaerobically, implying the existence of another regulatory gene that is expressed or perhaps is only active under anaerobic conditions (6). Furthermore, although serovar Typhimurium cysM mutants are Cys⁺ on sulfate under aerobic conditions, they are Cys^– when grown anaerobically (61), suggesting that anaerobic sulfate reduction may proceed through thiosulfate, which is then incorporated into S-sulfocysteine through the activity of the cysM product O-acetylserine (thiol)-lyase-B. Finally, serovar Typhimurium cysJ and cysI mutants are Cys⁺ during anaerobic growth owing to the activity of a sulfite reductase that is expressed from the asrABC operon only under anaerobic conditions (71, 90, 91). These genes are not present in E. coli. asrC encodes a peptide that is homologous to the cysI-encoded NADPH-sulfite reductase and has a siroheme binding motif; asrA encodes a peptide with a ferredoxin-like arrangement of cysteine residues; and asrB may encode a flavoprotein analogous to that of the flavoprotein of the aerobic NADPH-sulfite reductase. A partially purified preparation of this activity preferred NADH over NADPH as a reductant and was reversibly inhibited by oxygen (71).

Serovar Typhimurium also differs from E. coli in having anaerobically expressed thiosulfate and tetrathionate reductases that require anaerobiosis and their respective substrates for expression. These enzymes are responsible for conversion of tetrathionate to thiosulfate with subsequent reduction to sulfite, which is then reduced to H₂S as the end product. The membrane-bound thiosulfate reductase (6, 29, 83, 212) is expressed from the phsABC locus where the predicted amino acid sequences indicate that the enzyme is a molybdoprotein oxidoreductase (64, 74). Anaerobic thiosulfate and sulfite reduction require F₀F₁ ATP synthase activity implying that this enzyme plays a role in certain anaerobic oxidation-reduction reactions (218).

Tetrathionate reductase is expressed from the ttrRSBCA locus where DNA sequence analysis (78) suggests that TtrA contains a [4Fe-4S] cluster and, like thiosulfate reductase, a molybdopterin cofactor, which is consistent with an earlier an observation showing that the enzyme contains molybdenum (82). TtrB was predicted to bind four [4Fe-4S] clusters and TtrC appears to be an integral membrane protein containing a quinol oxidation site, implying a periplasmic site for the enzyme. ttrS and ttrR are thought to encode sensor and response elements of a two-component regulatory system for expression of the ttrBCA operon.

It has been proposed that the three serovar Typhimurium anaerobic reductases for sulfite, thiosulfate, and tetrathionate may function primarily in anaerobic respiration (91) with only the incidental property of complementing certain cys mutants under anaerobic conditions. If so, it is unlikely that they are regulated by the cysteine operon.

CYSTEINYL-tRNA SYNTHETASE

The gene for E. coli cysteinyl-tRNA synthetase, cysS (14), encodes a 52-kDa polypeptide with the HIGH and KMSKS peptide motifs that identify it as a class I enzyme (3, 58, 84). Cysteinyl-tRNA synthetase is the smallest monomeric aminoacyl-tRNA synthetase of E. coli, and its sequence is closely related to those of the enzymes for methionine, leucine, isoleucine, and valine. Its relatively small size and regions of homology with seryl-tRNA synthetase, a class II enzyme, have prompted speculation that cysteinyl-tRNA synthetase is closely related to a primordial aminoacyl-tRNA synthetase (3, 58). The enzyme is highly specific for l-cysteine and does not possess an editing function (60). Specificity is achieved through the formation of a thiolate bond between l-cysteine and a zinc ion at the active site of the enzyme, which has been demonstrated by spectroscopic studies () and by solution of the crystal structure (179). The enzyme forms small amounts of cysteine thiolactone from cysteinyl-tRNA^Cys via a nucleophilic attack on the cysteine carboxyl group by the sulfhydryl group (111), but the significance of this is unknown. It has been proposed that this activity may be the remnant of a function that provided an activated form of l-cysteine for protein synthesis in early biotic systems.

A single tRNA^Cys species has been identified in E. coli (156) with its gene, cysT, situated between those for a tRNA^Leu and a tRNA^Gly at centisome 42 on the chromosome (65, 125). The discriminator base, U73, of tRNA^Cys is a major determinant for recognition by cysteinyl-tRNA synthetase, as are elements within the tertiary domain defined by the d stem loop, the TΨC stem loop, and the variable loop (124). Mutations in the anticodon wobble base also impair recognition, although changing the GCA cysteine anticodon to the GAA phenylalanine anticodon does not effect aminoacylation with cysteine.