|Home | About | Journals | Submit | Contact Us | Français|
Rhodanese homology domains (RHDs) play important roles in sulfur trafficking mechanisms essential to the biosynthesis of sulfur-containing cofactors and nucleosides. We have now determined the crystal structure at 2.10 Å resolution for the Bacillus anthracis coenzyme A-disulfide reductase isoform (BaCoADR-RHD) containing a C-terminal RHD domain; this is the first structural representative of the multidomain proteins class of the rhodanese superfamily. The catalytic Cys44 of the CoADR module is separated by a distance of 25 Å from the active-site Cys514′ of the RHD domain from the complementary subunit. In stark contrast to the B. anthracis CoADR [Wallen, J.R., Paige, C., Mallett, T.C., Karplus, P.A., and Claiborne, A. (2008) Biochemistry 47, 5182-5193], the BaCoADR-RHD isoform does not catalyze the reduction of coenzyme A-disulfide, although both enzymes conserve the Cys-SSCoA redox center. NADH titrations have been combined with a synchrotron reduction protocol to examine the structural and redox behavior of the Cys44-SSCoA center. The synchrotron-reduced (Cys44 + CoASH) structure reveals ordered binding for the adenosine 3′-phosphate 5′-pyrophosphate moiety of CoASH, but the absence of density for the pantetheine arm indicates that it is flexible within the reduced active site. Steady-state kinetic analyses with the alternate disulfide substrates methyl methanethiolsulfonate (MMTS) and 5, 5′-dithiobis(2-nitrobenzoate) (DTNB), including the appropriate Cys→Ser mutants, demonstrate that MMTS reduction occurs within the CoADR active site. NADH-dependent DTNB reduction, on the other hand, requires communication between Cys44 and Cys514′, and we propose that reduction of the Cys44-SSCoA disulfide promotes the transfer of reducing equivalents to the RHD, with the swinging pantetheine arm serving as a ca. 20 Å bridge.
As reviewed by Mueller (1), nature seems to have settled on protein persulfide groups (R-SSH) rather than toxic bisulfide (HS−) as the dominant source of S2- equivalents for biosynthesis of sulfur-containing cofactors and thionucleosides. In the latter category, the biosynthesis of 4-thiouridine at position 8 in tRNA is catalyzed in Escherichia coli by ThiI (2, 3), a rhodanese homology domain (RHD)1 protein that forms a persulfide derivative of Cys456 (Cys456-SSH) within the RHD domain. Transfer of the terminal sulfur to the O4-adenylated U8-tRNA intermediate is proposed to yield an intramolecular protein disulfide between Cys456 and Cys344 of the ThiI PP-loop domain. A similar sulfur transfer mechanism has been considered for the human molybdopterin synthase sulfurase MOCS3 (4-6); to our knowledge, however, no crystal structures have been reported for either of these RHD proteins. The role of the E. coli ThiI RHD protein in 4-thiouridine synthesis has been contrasted with the biosynthesis of 2-thiouridine at position 34 in specific (Lys, Gln, Glu) tRNAs (7); in the latter case, the five E. coli Tus proteins (TusA-E) provide a Cys—SSH-based sulfur relay system that traffics a sulfur atom from free Cys, through a cascade of protein persulfide intermediates hosted on TusA, TusD, and TusE, to the 2-thiouridine product. Structures have been reported for TusA (8), for the TusBCD complex (9), and for the terminal MnmA thiouridylase-tRNA complex (10). The recent structural and functional analysis of the sulfur oxidation (S0 → SO42-) pathway in the purple sulfur bacterium Allochromatium vinosum (11) has identified key similarities between the E. coli Tus system and the Dsr proteins that comprise a sulfur trafficking portal for delivery of persulfidic S0 from the periplasm to the cytoplasmic sulfur oxidation machinery.
Very recently, the participation of “flavoprotein disulfide reductases” has been demonstrated in sulfur trafficking and/or sulfide metabolism in prokaryotes. Brito et al. (12) have determined the crystal structure of an archaeal sulfide:quinone oxidoreductase; in addition to the active-site Cys178 and Cys350, located proximal to the covalently-bound FAD isoalloxazine on the re face, a chain of three sulfur atoms is identified bridging the two Cys. These and other observations, including the active-site Cys161-Cys337 disulfide identified in the closely-related structure of the A. vinosum flavocytochrome c sulfide dehydrogenase [(13); flavoprotein subunit), have led to a mechanistic proposal (Scheme 1) in which repeated cycles of 1) HS− attack on Cys350 of the disulfide and 2) electron transfer from the nascent Cys178-S-, via the flavin, to quinone, allow extension of the polysulfide chain attached to Cys350. In a final step for which the chemical details are less clear, the Cys178-Cys350 disulfide is reformed and the polysulfide product is released. Although sulfide:quinone oxidoreductase and flavocytochrome c sulfide dehydrogenase are members of the NADH dehydrogenase subgroup of the “two dinucleotide binding domains” flavoproteins superfamily (14) that also includes the flavoprotein disulfide reductases (15), the second dinucleotide binding domain in these two enzymes does not bind NAD(P)H; HS- replaces the pyridine nucleotide as hydride donor. Furthermore, Cys178 and Cys350 do not correspond to any of the active-site Cys motifs identified in the flavoprotein disulfide reductases.
In our report on the crystal structure of coenzyme A-disulfide reductase (CoADR) from Staphylococcus aureus (16), we identified novel bacterial CoADR isoforms containing C-terminal extensions including an RHD. The CoADR-RHD proteins link the CoADR module (ca. 450 residues) with a C-terminal RHD domain (ca. 110 residues); thus far the one CoADR-RHD protein characterized is that from the sulfur-reducing facultative anaerobe Shewanella loihica PV-4 (17). The recombinant Sh. loihica protein catalyzed the NADH-dependent reductions of both polysulfide [Sn2-, where n = 4 and 5 species are predominant at the assay pH of 8.0 (18); kcat = 28 s−1] and DTNB [kcat/Km(DTNB) = 1.7 × 105 M−1s−1]. Although CoADR is a functional flavoprotein disulfide reductase [Group 3 (15)], CoADR-RHD did not catalyze the reduction of CoAS-disulfide (CoAD) as do the S. aureus (19) and Bacillus anthracis (20) CoADRs. The purified Sh. loihica enzyme gave a visible absorption spectrum characteristic of a flavoprotein, but anaerobic addition of five equiv of NADH/FAD reduced only ca. 50% of the flavin. Lukose et al. (17) suggested two possible roles for the RHD domain in Sn2- reduction (Scheme 2), either in 1) delivery of bound Sn2- to the active-site Cys43 of the CoADR module or 2) thiol-polysulfide interchange with a Cys43-SSnH intermediate, yielding a Cys43-SSH persulfide for reduction. While either scheme could in principle lead to full reduction of Sn2- → nHS−, neither restores the Cys43-SSCoA disulfide presumed to exist in the oxidized enzyme.
We have selected B. anthracis as a model organism for investigations into unique CoASH biosynthetic and redox functions, as in this organism CoASH replaces GSH as the major low-molecular-weight thiol (21). While Bacillus subtilis also lacks GSH, Cys has been shown to be the major low-molecular-weight thiol in one report (22), and the four B. subtilis strains represented in the National Center for Biotechnology Information database lack both CoADR and CoADR-RHD isoforms.2 Given the increased expression of the BAS0736 gene encoding CoADR-RHD in B. anthracis Sterne late in sporulation (23), the absence of structural representatives for multidomain enzymes containing the RHD (1, 24), and the well-defined structures and mechanisms for the CoADR enzymes from B. anthracis (20) and S. aureus (16, 19) [another GSH-deficient organism (22)], we have undertaken kinetic and spectroscopic studies of the B. anthracis CoADR-RHD, based on the alternate disulfide substrates MMTS and DTNB. We also present the crystal structures for both oxidized (FAD, Cys44-SSCoA) and synchrotron-reduced (FAD, Cys44-SH + CoASH) forms of CoADR-RHD, along with a mass spectrometric analysis of enzyme inactivation during NADH:MMTS oxidoreductase turnover.
Expression and purification of SaCoADR followed the published protocol (19). The BaCoADR-RHD clone (Ames strain; BA0774) was provided by Dr. Thomas Blank and Dr. Arthur Friedlander, U.S. Army Medical Research Institute of Infectious Diseases. The protein was expressed with an N-terminal His-tag provided by the pTHCm (25) plasmid vector in E. coli B834(DE3) cells. The expression and purification protocol followed that described recently for BaCoADR (20), with the following exceptions: expression cultures contained 50 μg/mL chloramphenicol and were harvested 8-9 h after induction; there were small differences in resuspension and Q-Sepharose loading buffers; and the Ni Sepharose High Performance (GE Healthcare) protocol described for BaCoADR Tyr→Phe mutants was employed. The purified CoADR-RHD protein was buffer-exchanged into 50 mM potassium phosphate, pH 7.0, containing 0.5 mM EDTA, concentrated to 10 mg/mL prior to adding glycerol to 20% (v/v), flash-frozen in liquid nitrogen, and stored in 0.05–0.5-mL aliquots at −80°C. The expression and purification of SeMet CoADR-RHD was essentially identical to that described for the native protein, except that cultures were grown in MOPS medium (EZ Rich Defined Medium, Teknova), with SeMet replacing Met. Mutant SaCoADRs (Y361F, Y419F, and the Y361,419F double mutant) and CoADR-RHDs (C44S and C514S) were generated as described for BaCoADR (20), using the respective wild-type expression plasmids as templates. Yields for native and mutant CoADR-RHDs and for the SeMet enzyme were 8 and 5 mg/L, respectively.
CoADR activities for wild-type and mutant SaCoADRs and CoADR-RHDs were measured as described recently for BaCoADR (20), using a Cary 50 spectrophotometer (Varian) thermostated at 25°C. MMTS-dependent turnover (26) was measured at 340 nm in a total volume of 1 mL containing 50 mM potassium phosphate, pH 7.0, 0.5 mM EDTA, 32 μM NADH [which is at least 30 × Km(NADPH) for SaCoADR reduction of CoAD (19)], and 1 mM MMTS, with a final CoADR-RHD concentration of 20 nM. DTNB-dependent turnover was measured identically, except that 200 μM DTNB replaced MMTS, and 5-thio-2-nitrobenzoate (TNB) product formation was monitored at 412 nm (27). Steady-state kinetic parameters for MMTS-dependent turnover with SaCoADR and CoADR-RHD were determined from initial velocity measurements made with the Applied Photophysics DX.17 MV stopped-flow spectrophotometer, with final enzyme concentrations of 100 nM. MMTS concentrations were varied over the range 0.01-1.5 mM, at a fixed NAD(P)H concentration of 32 μM. Initial rates were linear for at least 4 s (CoADR-RHD) or 6 s (SaCoADR, 33% of the total ΔA340); all measurements were repeated six times and were corrected for NAD(P)H oxidase activities measured in parallel, and kinetic data were analyzed as recently described for BaCoADR (20).
Extinction coefficients for wild-type and C514S CoADR-RHD proteins (ε451 = 12,000 M−1cm−1 and 11,900 M−1cm−1, respectively) were determined by standard methods (19). An extinction coefficient for reduced wild-type BaCoADR-RHD (ε451 = 1,200 M−1cm−1) was assumed, based on the extinction coefficient determined experimentally for dithionite-reduced BaCoADR (20). Anaerobic titrations followed established protocols (20), using a Hewlett-Packard model 8452A diode-array spectrophotometer.
SeMet CoADR-RHD was used for all crystal preparation and data collection. Thawed aliquots of SeMet CoADR-RHD were concentrated to 7 mg/mL in 10 mM sodium N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid, pH 7.2, and 4 mM NAD+ was added. Large, single crystals grew in 4 μL + 4 μL sitting drops over reservoirs of 0.5 mL 8-16% PEG8000, 2% (v/v) 2-methyl-2,4-pentanediol, and 0.2 M potassium acetate, pH 6.0, at 15°C. These crystals were harvested and crushed before streak seeding (28) to fresh drops in plates that had been equilibrated overnight. Crystals grew to full size in 1-2 weeks; prior to data collection, they were soaked in 18-20% PEG8000, 15-20% 2-methyl-2,4-pentanediol, and 0.2 M potassium acetate, pH 6.0, and flash-frozen in a nitrogen stream at 100 K. An oxidized (Cys44-SSCoA) CoADR-RHD data set was collected in-house on a Rigaku Saturn-92 CCD detector using Cu Kα radiation from a MicroMax-007 rotating anode X-ray generator and processed using HKL2000 (29). A three-wavelength data set for MAD phasing was collected with a single SeMet CoADR-RHD crystal at beamline X12C of the National Synchrotron Light Source and indexed, integrated, and scaled using d*TREK (30). For synchrotron reduction experiments, another SeMet CoADR-RHD crystal was used to collect six complete data sets (in the order peak1, remote1, inflection1, peak2, remote2, and inflection2) at beamline X26C of the National Synchrotron Light Source. These data sets were also indexed, integrated, and scaled using d*TREK. Table 1 summarizes data collection statistics for the MAD-phasing, in-house (LAB), and peak1 (SYNC1) and inflection2 (SYNC6) data sets that were used in the structure determinations.
Using the MAD data set, 31 of 34 Se sites per dimer were located using SOLVE (31). These heavy atom positions were refined, and initial phases were calculated using MLPHARE (32). Manual inspection of difference map peaks allowed identification of two more Se sites, and these were included in subsequent SOLVE calculations. Density modification and automatic building with RESOLVE (33, 34) allowed 77% of the dimer to be built, and the remaining residues, FAD, and CoAS- were built manually with O (35). Refinement was carried out at 2.0 Å resolution using CNS (36) with rounds of manual rebuilding, simulated annealing, energy minimization, and individual B-factor refinement. Water molecules were identified using a 3σ difference Fourier cutoff. Multiple rounds of refinement in CNS resulted in an R-factor = 17.6% and R-free = 22.3% for SeMet CoADR-RHD (data not shown).
Because partial reduction of the active-site disulfide was observed, this structure was not used for structural interpretation. This structure was used to determine the oxidized (Cys44-SSCoA) CoADR-RHD structure by molecular replacement using the LAB data set. Initial rounds of refinement were carried out in CNS; final rounds were performed using TLS (37) plus restrained refinement in REFMAC5 (38), with manual rebuilding in COOT (39). Waters (in 2Fo-Fc peaks greater than 1σ) were added to the structure using COOT. The SYNC1 and SYNC6 synchrotron-reduced CoADR-RHD structures were determined using the final oxidized CoADR-RHD structure as the molecular replacement model. The final SYNC1 and SYNC6 structures were refined using TLS plus restrained refinement in REFMAC5. Refinement statistics for all three models (LAB, SYNC1, and SYNC6) are summarized in Table 2.
Structure superpositions and sequence alignments were performed using DALI (40), DALILITE (41), and COOT (39), and CLUSTALW (42), DIALIGN (43), and ESPript (44), respectively. Figures of protein molecules and residues were prepared using PyMOL (45).
ESI-TOF mass spectrometric analysis was carried out with an Agilent MSD TOF system. The operating conditions for mass spectrometric analysis were as recently described by Jönsson et al. (46), except the fragmentor voltage was 360 V. Sample injection and data analysis protocols were also as recently described (46). Reactions (100 μL) containing 60 μM wild-type CoADR-RHD, with varying concentrations of NADH and MMTS, in 50 mM potassium phosphate, pH 7.0, 0.5 mM EDTA, were initiated at room temperature by addition of NADH. NADH and MMTS concentrations were varied (at an [MMTS]/[NADH] ratio of 1.5-2) to provide for 0, 10, and 100 turnovers, with NADH as the limiting substrate. After a ca. 1 min incubation, each sample was applied to a Bio-Gel P6 spin column to remove small molecules and buffer exchange into 50 mM ammonium bicarbonate, pH 8.0.
Reactions (500 μL) containing 5 μM wild-type CoADR-RHD and 15 mM NADH and MMTS (sufficient for 3,000 turnovers), in the pH 7.0 phosphate buffer given above, were initiated by addition of enzyme at room temperature. For analysis by limited proteolysis (47), native oxidized CoADR-RHD (1:50 trypsin:enzyme, w/w) and CoADR-RHD after 3,000 turnovers (1:500 trypsin:enzyme) were incubated in 100 μL of 50 mM Tris-SO42-, pH 8.0, plus 1 mM CaCl2, at 4°C for 1 h and for 30 min, respectively. Reactions were quenched by adding EDTA to a final concentration of 4 mM, and samples were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
Well-diffracting crystals of SeMet-containing BaCoADR-RHD were used to solve the structure by MAD phasing. Since the Cys44-SSCoA disulfide was partially reduced by synchrotron radiation in this structure, we collected data using a laboratory X-ray source to obtain the unperturbed oxidized structure at 2.1 Å resolution. This BaCoADR-RHD model includes residues Ser2-Tyr554 for both chains A and B of the biological dimer (there is one dimer in the asymmetric unit), two FAD cofactors, two covalently bound CoAS-, and 665 solvent waters, all with reasonable geometry (Table 2 and Figure 1). Residues from chain B are designated by a prime symbol. Electron density is also present for the main chain of two residues from the N-terminal His-tag (modeled as Gly0 and Gly1) in chain A and one residue in chain B (Gly1′). There is also weak electron density for the ADP moiety of NAD+ in both subunits, but modeling of these positions results in B-factors above 100 Å2, and so they are not included in the final model. Some uninterpreted electron density is also observed near the active-site Cys514 of the RHD (see below). Although non-crystallographic symmetry restraints were not used in refinement, the A and B chains superimpose well, with a Cα rmsd of 0.5 Å. Each monomer consists of a CoADR module (residues 2-451) linked to a C-terminal RHD (residues 455-554), with the RHD′ domain (chain B) packing against the CoADR module of chain A (Figure 1).
A structural superposition of the BaCoADR-RHD and BaCoADR (20) dimers shows them to be quite similar (Cα rmsd = 1.4 Å over 865 residues; Figure 2A), consistent with their 35% sequence identity. Figure 3A gives a partial structure-based sequence alignment for BaCoADR and the CoADR modules of six bacterial CoADR-RHD proteins, including the B. anthracis enzyme. The domain structures are the same (Figure 3A, legend), and the largest structural difference involves BaCoADR-RHD residues Glu183-Pro190, a loop that in BaCoADR (Glu180-Thr187) changes conformation on NAD(P)H binding (20). In BaCoADR-RHD, these residues are well-ordered and adopt a conformation similar to that observed in the BaCoADR-NAD(P)H complex, suggesting that the conformational change observed in BaCoADR on binding NAD(P)H is not required for CoADR-RHD. In contrast to the NADPH-specific SaCoADR (16) and the dual specificity BaCoADR, the conservation of Gly165 and Glu183 in the CoADR-RHDs (Figure 3A) suggests a marked preference for NADH. We have recently observed (20) that BaCoADR Ala160 replaces the active-site Tyr conserved in all known functional GRs (e.g., Tyr197 in human GR), in NADH peroxidase, and in all known functional NADH oxidases. In human GR the reductive half-reaction was seriously compromised, for example, in the Y197S mutant (48). Figure 3A demonstrates that, while the Pelobacter carbinolicus CoADR-RHD (49) has Leu at the equivalent position, all other enzymes analyzed have Phe. The preference for the aromatic side chain as evidenced in the CoADR modules of these enzymes suggests that this residue may play a similar role in covering the re-face of the flavin in the free enzyme. GR Tyr197 is known to swing out, opening the pocket for the incoming NADPH (50); although bound NAD+ has a limited occupancy in the BaCoADR-RHD structure, with the side chain of Phe163 exhibiting some disorder as well, Phe163 in the refined BaCoADR-RHD model is in the “out” conformation. Figure 3A also demonstrates that the CoADR-RHDs conserve the essential (in BaCoADR and SaCoADR) active-site cysteine (Cys44) as well as one of the two active-site tyrosines (Tyr428′) that play important roles in CoADR catalysis (20). The P. carbinolicus, Clostridium botulinum (51), and Clostridium novyi-NT (52) CoADR-RHDs do not conserve Tyr370 (replaced by Glu, Leu, and Leu, respectively); this is consistent with the earlier observation that while BaCoADR Tyr367′ is not essential for activity, BaCoADR Tyr425′ plays a primary role in protonation of the CoAS-II product during CoAD reduction (20).
In the active-site area, BaCoADR-RHD contains a Cys44-SSCoA mixed disulfide in the oxidized structure; similar to BaCoADR, but in stark contrast to SaCoADR (16), the CoAS-3′-phosphate is buried in the protein interior (Figure 1). A notable difference between the nonflavin redox centers (Cys-SSCoA) in BaCoADR and CoADR-RHD is that the respective CoAS-sulfur atoms point in opposite directions (Figure 2B). This is curious given the conserved orientations of the partnering Cys side chains but is reliably determined by the electron density. In CoADR-RHD this difference facilitates a more favorable interaction between the CoAS-sulfur and Tyr428′-OH (3.2 Å). While BaCoADR-RHD Tyr428′ and BaCoADR Tyr425′ superimpose very favorably, Tyr370′-OH is shifted 2.2 Å compared to Tyr367′-OH. As a result, CoADR-RHD Tyr370′ is positioned more closely to both the Cys44-SSCoA disulfide and to Tyr428′ (Tyr-OH—to—Tyr-OH distance of only 2.6 Å).
A major difference in CoAS-binding compared with BaCoADR is that the adenine end of the cofactor is well buried in BaCoADR-RHD. This occurs, not due to any major change in the interaction within the CoADR module (Figure 2A), but from the position of the RHD′ domain (Figure 1); this will be discussed further below.
BaCoADR-RHD is the first structural representative of the multidomain proteins class of the rhodanese superfamily (24); a DALI search performed on the RHD domain (residues Asp455-Tyr554) identifies the RhdA rhodanese from Azotobacter vinelandii [rmsd = 1.9 Å, 21% identity (53)] as the most closely-related structure among functionally characterized proteins. Figure 3B gives a structure-based sequence alignment for the C-terminal catalytic domain of RhdA (residues 140-271), which contains the active-site Cys230, and the RHD domains of the CoADR-RHD isoforms from Figure 3A. Three segments of RhdA (RhdA residues 176-183, 200-209, and 216-218) are absent in the RHDs (Figures 3B and and4A4A).
The active-site Cys514 of CoADR-RHD is positioned at the protein surface, at a distance of 25 Å from Cys44′ of chain B (Figure 1). Cys514 and the whole CQxxxR active-site loop adopt a conformation very similar to that of the RhdA catalytic domain (Figure 4A and C). Cys514-Sγ adopts two conformations (Figure 4B and C), with the stronger one matching the RhdA active-site structure. In addition, a separate superposition (not shown) indicates that RhdA residues Asp200-Thr209 (see above) would clash with a portion of the FAD-binding domain of the CoADR module. This provides a structural basis for understanding the deletion of this segment in the RHD domain of BaCoADR-RHD. Of special interest in the P. carbinolicus and clostridial CoADR-RHD sequences (Figure 3B) is the replacement of Gln515 in the CQxxxR loop with a Lys or Arg residue. A similar CKxxxR catalytic loop sequence is also seen in the polysulfide-sulfur transferase (Sud) protein from Wolinella succinogenes [(54); see later section]. Krepinsky and Leimkühler (6) proposed that the presence of two basic residues in an RHD catalytic loop promotes recognition of thiosulfate (S=SO32-) as sulfur donor, so this substitution may indicate a shift in preferred sulfur donor for the P. carbinolicus and clostridial proteins.
As noted above, the adenosine 3′-phosphate 5′-pyrophosphate moiety of CoAS- interacts extensively with the RHD′ domain, leading to well-defined binding for the whole Cys44-SSCoA moiety in the oxidized enzyme (Figure 5A). The CoAS- is displaced by as much as 4 Å at the adenine relative to the equivalent position in BaCoADR (Figure 2A). The major interactions of adenine with the RHD′ domain (Figure 5A) involve residues Leu516′, Gly517′, Met518′, and Tyr521′ of the RHD catalytic loop. In addition, the CoAS-3′-phosphate interacts with Arg20, Arg23, Arg310, Asn448′, and Tyr521′, while Arg23 and Lys436′ both interact with the CoAS-pyrophosphate. The interactions with 3′-phosphate and 5′-pyrophosphate elements of CoAS- are similar to those seen in BaCoADR, which binds CoASH tightly enough that it remains in its binding pocket in the reduced BaCoADR-NAD(P)H complexes (20). The additional interactions with the RHD′ domain would be expected to anchor it even more firmly.
In order to obtain the reduced (Cys44-SH + CoASH) structure of CoADR-RHD, NADH soaks were performed, but the crystals dissolved almost immediately. Since partial Cys44-SSCoA reduction was observed in the structure determined by MAD phasing, we decided to use synchrotron reduction as a tool to produce the reduced form, as has been done by others (55, 56). We collected a series of six complete data sets from a single crystal at the National Synchrotron Light Source and used the first (SYNC1) and sixth (SYNC6) to define partially and fully reduced structures. Electron density from the SYNC6 data set shows that disulfide reduction has occurred (Figure 5B); we note that the higher R-factors (Table 2) and loss of electron density for some side chains (notably, some Asp and Glu carboxylates; data not shown) also reflect some general crystal damage during the synchrotron reduction experiment.
The adenosine 3′-phosphate 5′-pyrophosphate moiety of CoAS- remains ordered in the SYNC6 structure, but the absence of electron density for the whole pantetheine arm implies that it is flexible within the active site of the reduced enzyme. Berkholz et al. (56) reported that synchrotron-reduced structures may not accurately represent a chemically-reduced structure, because in the frozen crystal the normal conformational changes associated with disulfide reduction may be inhibited. Thus the changes seen here may be an underestimate of the changes that would occur in solution, but should not be an overestimate. This strengthens our conclusion that when the disulfide is opened, the whole pantetheine arm does become mobile.
This disordering, plus the proximity of the CoAS-adenine to the RHD active site, led us to consider the possibility that the two-electron reduced form of BaCoADR-RHD provides a pantetheine swinging arm that bridges the 25 Å separation between the active-site Cys44 of the CoADR module and Cys514′ of the RHD, with the tight network of electrostatic interactions described above anchoring the reduced CoASH to the protein. As shown in Figure 5B, the conformation of the reduced pantetheine arm can easily be adjusted to a “bent” form that bridges the gap. This bent form has precedent as it is similar to that observed experimentally in the crystal structure of the E. coli pantothenate kinase-CoASH complex [PDB entry 1ESM (57)]. Also, there is precedent for pantetheine to serve as a swinging arm. The 4′-phosphopantetheine arm derived from CoASH provides the carrier proteins and intermediate domains in modular polyketide synthases, vertebrate fatty acid synthases, nonribosomal peptide synthases (58-60), and now, 10-formyltetrahydrofolate dehydrogenase (61), with a ca. 20 Å reach into the active sites of the respective catalytic domains.
The synchrotron-reduced structures also show changes in the RHD active site (Figure 6). In the oxidized form, Cys514′ adopts two distinct conformations for the active-site Cys-Sγ (Figure 6A). The primary conformation matches that reported for RhdA Cys230 in the persulfide (Cys230-SSH) structure [Figure 4C (53)], and the secondary one has an uninterpreted peak of extra density associated with it. While we did very carefully evaluate the possibility that the secondary Cys514′-Sγ position in the oxidized form represents a persulfide sulfur, the distance from the Cys514′ sulfur atom (the primary Cys514′-Sγ conformation) to the closest point of the extra density is well over 3 Å, indicating the absence of any covalently-bonded atom in this position. The density is continuous, but we have not been able to interpret it.3 In contrast, in the SYNC6 structure (Figure 6C) only the primary Cys514′ conformation remains, and the additional uninterpreted density has also disappeared. Other than those changes described previously that are thought to reflect crystal damage in this experiment, the movements of Cys44, CoAS(H), and Cys514′ represent the only major differences in the SYNC6 structure.
Lukose et al. (17) have recently reported that anaerobic titration of Sh. loihica PV-4 CoADR-RHD resulted in ca. 50% flavin reduction with 5 equiv of NADH/FAD. Examination of the spectral course of that titration indicates the presence of an isosbestic point at 350 nm over the titration phase 0-2.1 equiv NADH/FAD, demonstrating the absence of excess NADH. Figure 7 gives the spectral course of an anaerobic NADH titration with BaCoADR-RHD. The first phase of the titration (0-0.8 equiv NADH/FAD; Figure 7A) does not result in flavin reduction, and the difference spectra indicating maxima at 412 and 518 nm reflect the reduction of Cys44-SSCoA and the appearance of a weak Cys44-S−→FAD charge-transfer intermediate (20). Isosbestic points at 362, 398, 450, and 502 nm confirm the oxidation of all NADH and the absence of any E(FADH2) component. Figure 7C represents the first phase as a lag in the absorbance change at both 450 and 502 nm. The second phase of the titration (0.8-2.1 equiv NADH/FAD; Figure 7B) gives a further increase in absorbance over the range 500-800 nm, which accompanies substoichiometric flavin reduction. It is also clear that excess NADH accumulates, beyond 1 equiv/FAD added. Overall, this redox system (Scheme 3) resembles that described earlier for NADH peroxidase (62), another representative of the NADH peroxidase/oxidase and CoADR subgroup within the two dinucleotide binding domains flavoproteins superfamily [(14); also Group 3 of the flavoprotein disulfide reductase family (15)]. The EH2 intermediate exhibits none of the redox asymmetry documented recently for BaCoADR (20) and differs from that described for NADH peroxidase primarily in the intensity of the 520 nm charge-transfer band. The second phase of the CoADR-RHD redox scheme consists of an equilibrium between static EH2(NADH) and EH4(NAD+) complexes, where the latter intermediate is characterized by a decrease in A450 and an increase in absorbance beyond 650 nm. Only partial flavin reduction continues to be observed at up to 14 equiv NADH/FAD, indicating that the EH2(NADH) species is favored thermodynamically in this system. At 2.1 equiv NADH/FAD, only ca. 17% flavin reduction is evident; even at 14 equiv NADH/FAD, only ca. 27% of the flavin appears reduced.
In the standard CoADR assay, BaCoADR-RHD does not catalyze CoAD reduction; the same absence of CoADR activity has been reported for the Sh. loihica enzyme (17). This could be due to an inability of the bulky CoAD substrate to approach Cys44 of the reduced CoADR-RHD in the proper geometry; alternatively, the single CoASH may remain bound so tightly (Figure 5B and Ref 20) that it cannot be displaced by a new CoAD substrate. To test whether the enzyme is capable of carrying out disulfide reductase chemistry we assayed the activities of both SaCoADR and BaCoADR-RHD with MMTS. MMTS [CH3SSO2CH3 (63)] is a small neutral thiol reagent that has been used as an alternate substrate with yeast GR (26) and Plasmodium falciparum thioredoxin reductase (64); both are flavoprotein disulfide reductases within the same superfamily (disulfide reductase subgroup) as CoADR [(14); also, flavoprotein disulfide reductase Groups 1 and 2, respectively (15)]. MMTS reduction yields CH3SH + CH3SO2−; for GR, kcat/Km(MMTS) = 3.7 × 103 M−1s−1 (via the interchange Cys), and turnover is accompanied by a relatively slow inactivation process that has been attributed to the parallel reaction of the charge-transfer Cys (26).
Figure 8A and Table 3 give the steady-state kinetic parameters for SaCoADR with NADPH and MMTS, determined from stopped-flow initial velocity measurements. With SaCoADR, wild-type [kcat = 6.4 ± 0.1 s−1, Km(MMTS) = 78 ± 3 μM], Y361F [kcat = 10.4 ± 0.1 s−1, Km(MMTS) = 47 ± 2 μM], Y419F [kcat = 1.2 ± 0.1 s−1, Km(MMTS) = 575 ± 41 μM], and the Y361,419F double mutant [kcat = 1.4 ± 0.1 s−1, Km(MMTS) = 346 ± 48 μM] were analyzed. The kcat/Km(MMTS) for wild-type SaCoADR of 8.2 × 104 M−1s−1 represents a 22-fold increase over the value for yeast GR (26). Surprisingly, the catalytic efficiency of the SaCoADR Y361F mutant is enhanced 2.7-fold over the wild-type enzyme with this alternate substrate and may reflect a preference for the more hydrophobic Phe side chain in stabilizing apolar elements (e.g., -CH3 groups) in either transition state [Cys43-S−···S(CH3)SO2CH3 or FADH2···Cys43-SSCH3; Scheme 4] during MMTS reduction. kcat/Km(MMTS) values for both Y419F and Y361,419F SaCoADR mutants are 2-5% of the wild-type value; these results support the proposed role for Tyr419 in acid-base catalysis,4 in this case involving protonation of CH3S− [pKa = 10.3 (65)] in the reductive half-reaction [Scheme 4; the pKa of CH3SO2− is 2.3 (66)]. The high activity of the Y361F mutant also reinforces the suspicion that the substitution of Leu, or even Glu, for Tyr370 in the clostridial and P. carbinolicus CoADR-RHD sequences (Figure 3A) need not be debilitating.
Figure 8B and Table 3 show that wild-type [kcat = 29 ± 1 s−1, Km(MMTS) = 128 ± 20 μM] and C514S [kcat = 29 ± 2 s−1, Km(MMTS) = 244 ± 34 μM] BaCoADR-RHDs catalyze MMTS reduction with an efficiency [kcat/Km(MMTS)] ca. twice that for wild-type SaCoADR, showing CoADR-RHDs to be catalytically competent for disulfide reduction. As expected, the C44S mutant has no detectable activity, confirming that all MMTS reduction occurs within the CoADR module. In contrast, however, to the linear initial rates observed with wild-type enzyme in the stopped-flow spectrophotometer, steady-state turnover experiments with the Cary 50 spectrophotometer demonstrated a complete loss of activity over 3 min (Figure 9), similar to but 8-fold faster (at 1 mM MMTS) than the inactivation observed with yeast GR (26). The pseudo-first order rate for inactivation at 1 mM MMTS is 1.6 min−1 (Figure 9), and this rate is dependent on [MMTS]. Parallel analysis of the CoADR-RHD C514S mutant demonstrates that Cys514 of the RHD is required for rapid inactivation. The addition of fresh NADH at this point does not reverse the inactivation, and freshly added wild-type enzyme duplicates the original progress curve, showing that no inhibitory product has accumulated. Also, preincubation of CoADR-RHD with 1 mM MMTS has no effect on the initial rate observed with subsequent NADH addition.
Lukose et al. (17) reported that Sh. loihica CoADR-RHD catalyzes NADH-dependent DTNB reduction [apparent Km = 116 μM, kcat = 19.9 s−1, and kcat/Km(DTNB) = 1.7 × 105 M−1 s−1]; these steady-state parameters predict a turnover number of 12.6 s−1 at 200 μM DTNB and saturating NADH. When wild-type BaCoADR-RHD is assayed in the presence of 32 μM NADH and 200 μM DTNB, a turnover number of 5.2 s−1 is observed (data not shown), suggesting that the kinetic parameters for DTNB reduction by the two CoADR-RHDs are similar. In contrast to the results with MMTS, however, there is no turnover-dependent inactivation with DTNB, and the C514S mutant is only ca. 4% as active as wild-type enzyme. The C44S mutant again has no detectable reductase activity. These results demonstrate that Cys514 is essential for efficient DTNB reduction and suggest that the two Cys-dependent active sites (CoADR and RHD′) communicate during turnover, despite their 25 Å separation in the crystal structures.
The simplest mechanism for this communication is our proposal above, that the tightly bound, reduced CoASH functions as a cofactor (Figure 5B) that provides for communication between Cys44 and Cys514′, via a swinging pantetheine arm (Scheme 5). Beginning with the oxidized (FAD, Cys44-SSCoA) wild-type enzyme (species 1 in Scheme 5), DTNB reacts with Cys514′ to give the mixed disulfide (species 2 in Scheme 5). The catalytic loop of the RHD domain has a net positive charge (Arg in position 6), and Krepinsky and Leimkühler (6) have described how negatively-charged sulfur donors like thiosulfate and 3-mercaptopyruvate interact electrostatically with the respective sulfurtransferase active sites. A similar electrostatic attraction could direct the anionic DTNB substrate to this site, within the RHD domain. In the second step, NADH reduction via the flavin (Scheme 3) yields the EH2 intermediate (Cys44-SH + CoASH), and the nascent pantetheine-SH arm swings into position, approaching the Cys514′-SSTNB disulfide; subsequent thiol-disulfide exchange gives pantetheine-SSTNB and restores Cys514′-SH (species 3 in Scheme 5). In the penultimate catalytic step, the swinging pantetheine-SSTNB arm delivers the mixed disulfide to the active site of the CoADR module (species 4 in Scheme 5); the final thiol-disulfide interchange with Cys44-S− restores the resting, oxidized form of the enzyme (species 1).
The very low (4% of wild-type) DTNB reductase activity measured with the C514S mutant may be attributable to a direct but inefficient reaction between the pantetheine-SH and DTNB, but with no catalytic source of pantetheine-SH (the C44S mutant) there is no activity. Nor do we support any scheme involving direct reaction of Cys44 with DTNB (in either wild-type or C514S mutant); this is consistent with the failure of SaCoADR to catalyze DTNB reduction (67). Finally, the enzyme would be expected to form a pantetheine-SSCys514′ derivative in the event TNB is displaced instead of Cys514′-SH (species 3). In the absence of a disulfide reductant, the enzyme would be trapped in this catalytically inactive form; such a scenario may have relevance to the inactivation of the wild-type enzyme observed during turnover with MMTS.
To summarize, CoADR-RHD Cys44 is essential for both DTNB and MMTS reductase activities; the respective kcat/Km values of 1.7 × 105 M−1s−1 [Sh. loihica CoADR-RHD with DTNB (17)] 2.2 × 105 M−1s−1 (BaCoADR-RHD with MMTS, Table 3) are very similar, both to each other and to the value for SaCoADR with MMTS (8.2 × 104 M−1s−1). Cys514′ of CoADR-RHD is essential for efficient DTNB reduction, but it is not involved in catalytic MMTS reduction. It is required for rapid inactivation during turnover with MMTS, but no inactivation is observed during DTNB reduction.
We have tested the possibility that a pantetheine-SSCys514′ disulfide is responsible for the inactivation observed during MMTS turnover, using ESI-TOF mass spectrometry. The oxidized wild-type CoADR-RHD (species 1 in Scheme 5) presents two distinct m/z species by ESI-TOF [Figure 10, upper panel (black)]: the holoprotein [m/z species c; BaCoADR-RHD Cys44-SSCoA form with noncovalent FAD (m = 66,461 Da)] and the deflavo enzyme without FAD [m/z species b (m = 65,680 Da)]. Scheme 6 provides a chemical and structural view of Cys514′-SSCoA formation and inactivation, as proposed. The cycle begins with the BaCoADR-RHD Cys44-SSCH3 intermediate formed in turnover; Cys514′-SH reacts with a second MMTS in a reaction comparable to that described earlier with DTNB. The resulting Cys44-SSCH3, Cys514′-SSCH3 species is common to the two proposed mechanisms of inactivation. In mechanism B, the reduced pantetheine-SH of bound CoASH swings into position approaching the Cys514′-SSCH3 disulfide. The resulting thiol-disulfide exchange gives the Cys514′-SSCoA disulfide, displacing CH3SH. Formation of this species, during MMTS turnover and inactivation in the absence of a disulfide reductant, would also be expected to give m/z species b and/or c. Figure 10 [middle panel (red), and lower panel (green)] clearly shows that both m/z species decrease with the extent of inactivation. Instead, there is a correlation between the number of catalytic cycles (and increasing enzyme inactivation), reduction in the intensities of m/z species b and c, and the appearance of a new species a (m = 64,921 Da), attributed to the loss of CoAS(H) from the deflavo enzyme. This loss of CoAS(H) can only occur via an initial NADH-reduced (Cys44-SH + CoASH) intermediate, as implied in Scheme 6 and as will be discussed below based on preliminary experiments. While there is no m/z peak corresponding to free CoASH (m = 768 Da), there is a commensurate increase in a new low mass species of 814 Da, in perfect agreement with the mass calculated for CoA-SSCH3. The simple addition of CH3S- (m = 47 Da) to Cys514′ and/or Cys44 should have been detectable, since ESI-TOF instruments routinely achieve measurements with better than 5 ppm accuracy.
We conclude, that while the inactivation of CoADR-RHD during turnover with MMTS is not due to accumulation of a Cys514′-SSCoA form, it is correlated with modification of the tightly bound, reduced CoASH cofactor with MMTS; this most likely occurs via the proposed Cys514′-SSCH3 intermediate (Scheme 6, mechanism A). The molecular linkage between CoASH S-methylthiolation and the loss of MMTS reductase activity within the CoADR module is currently under investigation. Limited proteolysis experiments, for example, have demonstrated that the MMTS-inactivated enzyme is much more sensitive to trypsin digestion (data not shown).
The structural characterization and catalytic properties presented for BaCoADR-RHD in this report represent the first crystallographic analysis for any member of the multidomain proteins class of the rhodanese superfamily (24). Given the well-characterized CoADR enzymes from S. aureus and B. anthracis, this work also describes a new isoform within Group 3 of the flavoprotein disulfide reductase family (15). As such, the contributions of these enzymes to the biologically important topics of sulfur (and persulfide) trafficking and sulfide metabolism in prokaryotes (1, 24, 68) can now be more clearly appreciated. The dimeric BaCoADR-RHD structure demonstrates that the two active-site Cys, Cys44 of the CoADR module and Cys514′ of the RHD′ domain, are separated by 25 Å. In contrast to BaCoADR, the adenine end of bound CoAS- interacts extensively with the RHD′ domain and is well buried, anchoring it firmly in the protein. This likely contributes to the observation that BaCoADR-RHD does not recognize CoAD as a substrate. Reductive titration of the wild-type enzyme with NADH demonstrates flavin-mediated reduction of Cys44-SSCoA; synchrotron reduction gives a two-electron reduced (Cys44-SH + CoASH) enzyme as well, and the adenosine 3′-phosphate 5′-pyrophosphate of bound CoASH remains well ordered in this structure. The pantetheine arm, however, becomes mobile; these observations lead to a swinging pantetheine arm model by which bound CoASH shuttles reducing equivalents from the CoADR active site to the RHD′ domain, bridging the 25 Å gap.
MMTS and DTNB are alternate disulfide substrates for BaCoADR-RHD, but in mechanisms that are very distinct. MMTS reacts, as in the case of SaCoADR, at the CoADR active site; the C44S mutant is inactive, but the C514S mutant retains activity. DTNB, which is not a substrate for SaCoADR (67), reacts at Cys514′ but requires communication (the swinging pantetheine arm) between Cys44 and Cys514′; both C44S and C514S mutants have little-or-no activity. The swinging pantetheine arm and both Cys44 and Cys514′ are essential to the inactivation of BaCoADR-RHD observed in turnover with MMTS, as analyzed by ESI-TOF mass spectrometry.
The CoADR-RHD isoform is found in Sh. loihica, in B. anthracis and several other bacterial pathogens, and in the anti-tumor agent C. novyi-NT. A distinct CoADR-RHD-SirA-DsrE isoform has been identified in the Fe(III)-reducing anaerobe P. carbinolicus (49) and in several clostridia, including C. botulinum (51) and Clostridium beijerinckii (GenBank CP000721), another known Fe(III) reducer (69). In the latter isoform, the ca. 70-residue SirA module appears to be another sulfur-transfer protein as it is 43% identical in sequence to the E. coli TusA protein [(7, 8); see “Introduction”] and conserves the active-site CPxP motif involved directly in Cys—SSH-based sulfur relay. The DsrE domain of ca. 150 residues conserves the active-site Cys of E. coli TusD [(9); Cys78, which is equivalent to Cys781 of the P. carbinolicus multidomain enzyme] that accepts sulfur from the TusA persulfide; this Cys is also conserved in the A. vinosum DsrE protein [(11); see “Introduction”], which is a structural homolog of TusD. These observations support a proposed role in sulfur transfer for the DsrE domain as well, in this third CoADR isoform; the Pcar_0429 gene encoding P. carbinolicus CoADR-RHD-SirA-DsrE was upregulated over 20-fold during sulfur-dependent growth on Fe(III), compared with fermentative growth on acetoin (49). This is consistent with a possible function in reducing S0 → HS−, with HS− serving as the intermediate in Fe(III) reduction at the cell surface.
The Pcar_1579 gene encoding CoADR-RHD was upregulated five-fold during sulfur-dependent growth of P. carbinolicus on Fe(III) (49), and Lukose et al. (17) demonstrated that the Sh. loihica enzyme catalyzed the NADH-dependent reduction of polysulfide: NADH + S42- → S32- + HS− + NAD+ (see “Introduction,” Scheme 2). Sequence comparisons have shown that the polysulfide-sulfur transferase (Sud) protein from W. succinogenes contains a catalytic loop very similar to that of the RhdA rhodanese and the RHDs; an NMR structure is available for the sulfur-loaded form of the Sud protein and includes a calculated model for the Cys89-SS5H polysulfide tail (54). The positively-charged binding pocket consists of Arg46, Lys90, and Arg94 of the catalytic loop and one other active-site Arg residue; Arg46 and Arg94 are conserved in all RHDs, and Lys90 is either conserved or replaced with Arg in the P. carbinolicus and clostridial enzymes (Figure 3B; aligned with BaCoADR-RHD residues 476, 515, and 519). The fourth active-site Arg of the Sud protein appears in the segment equivalent to RhdA residues 200-209 that is absent in the RHD domains. All things considered, we conclude that the Cys514 equivalent in Sh. loihica CoADR-RHD (Cys531) is the primary S4-52- reaction site. While the Sud protein transfers its polysulfide ligand to the active site of polysulfide reductase (18, 54), Sh. loihica CoADR-RHD both binds and reduces polysulfide → HS−, as demonstrated by Lukose et al. (17).
Scheme 2, A (see ”Introduction”) represents one of two mechanistic proposals forwarded to account for polysulfide binding and reduction by the Sh. loihica enzyme. Our structural and kinetic analysis of BaCoADR-RHD (50% identical in sequence) leads to the following points: 1) There is no provision for restoring the Cys-SSCoA disulfide in the resting, oxidized enzyme. 2) The distance between Cys43-Sγ and Cys531′-Sγ is ca. 25 Å. In the model of the Sud protein Cys89-SS5H polysulfide tail, the distance between S1 and S5 is 5.3 Å (54). These considerations make it extremely unlikely that there could be direct chemical interaction between Cys43-Sγ and Cys531′-SS4H. 3) Based on the characterization of the DTNB reaction with BaCoADR-RHD, as presented in Scheme 5, both the initial reaction with S42- at Cys531′ and the requirement for communication with the active site of the CoADR module are supported. But the direct reaction of Cys43 with Cys531′-SS4H is extremely unlikely.
As described in Scheme 7, our mechanistic proposal for polysulfide reduction by CoADR-RHD begins with an enzyme form in which 1) S52- has reacted at Cys514′ (BaCoADR-RHD numbering), and 2) NADH has reduced the Cys44-SSCoA disulfide. The swinging pantetheine arm of tightly bound, reduced CoASH moves into position close to the polysulfide; attack by the pantetheine-S− on the terminal sulfur of Cys514′-SS4H yields the nascent pantetheine-SSH persulfide, reducing the polysulfide tail by one S unit. This structure is analogous to species 3 in Scheme 5. The swinging arm now moves back to its original position, delivering the persulfide for thiol interchange with Cys44-S−. The Cys44-SSCoA disulfide re-forms, restoring the resting, oxidized enzyme (with Cys514′-SS3H, analogous to -species 2 in Scheme 5) and releasing HS−. The cycle is repeated three times to complete the reduction of S52- → 5HS−. The overall reaction can be contrasted with that described for sulfide:quinone oxidoreductase [(12); see “Introduction,” Scheme 1], which converts HS− → polysulfide at the expense of a quinone oxidant; the mechanisms are mirror images, working in opposite directions. CoADR-RHD converts 5NADH + S52- → 5NAD+ + 5HS−; sulfide:quinone oxidoreductase converts 5HS− + 5Q → S52- + 5QH2. Cys514′ is the functional equivalent of sulfide:quinone oxidoreductase Cys350, and the swinging pantetheine arm is essential for connection with Cys44.
The polysulfide reduction catalyzed by Sh. loihica CoADR-RHD may be complemented by reactions with other sulfur donors such as thiosulfate and 3-mercaptopyruvate; the RHD active sites of E. coli ThiI and human MOCS3 (see “Introduction”) accept sulfur directly from free cysteine, as catalyzed by IscS (3) and Nfs1 (70) desulfurases, respectively. Reduction of the Cys-SSCoA disulfide promotes the movement of the swinging pantetheine-SH arm and reaction of the latter with the RHD Cys-SSH. The nascent pantetheine-SSH swings back into position near the Cys of the CoADR module, where HS− is released or transferred. In the mechanisms proposed for E. coli ThiI (3) and human MOCS3 (4), formation or transfer of the HS− equivalent involves an interdomain protein disulfide intermediate, which must be reduced in the catalytic cycle. In CoADR-RHD turnover, the same formation or transfer of the HS− equivalent involves re-formation of the Cys-SSCoA disulfide; NADH serves as the catalytic reductant. For the P. carbinolicus CoADR-RHD, the nascent HS− equivalent has been suggested to serve as the intermediate in Fe(III) reduction at the cell surface (49). In the CoADRRHD-SirA-DsrE isoform, the terminal sulfur could be transferred directly from the Cys529-SSH persulfide (P. carbinolicus numbering) to the active-site Cys591 of the SirA domain, in a cascade leading ultimately to the DsrE domain Cys781. B. anthracis Sterne, while it lacks this third CoADR isoform, does have CoADR-RHD (BAS0736), SirA-RHD (BAS0738; YrkF homolog), SirA (BAS0740; TusA homolog), and DsrE-like (BAS0743; YrkE homolog) protein-encoding loci clustered together.2 In B. anthracis, the ThiI protein is also known to lack the C-terminal RHD found in E. coli ThiI (71); BaCoADR-RHD could function to provide the HS− equivalent needed by ThiI for 4-thiouridine-tRNA synthesis.
There are several indications for sporulation-specific functions involving sulfur metabolism and low-molecular-weight thiols such as Cys and CoASH in B. subtilis and B. anthracis. The BAS0736 gene (Sterne strain) encoding BaCoADR-RHD is expressed at low levels in late log phase (23) but is upregulated ca. fourfold during sporulation.5 The transcript is present at even higher levels (ca. 20—30-fold increase) during the first 60 min after inoculation, which corresponds to germination and outgrowth. The BASΔ0736 deletion mutant6 grows normally in brain heart infusion broth but appears significantly compromised both in germination assays and in resuming exponential growth following exposure to the disulfide stress reagent diamide. Crow et al. (72) have recently reported the crystal structure and functional analysis of the B. subtilis thiol-disulfide oxidoreductase StoA, which is an extracytoplasmic enzyme important for synthesis of the spore core peptidoglycan. The bifunctional enzyme MiaB catalyzes linked thiolation and methylation reactions in the conversion of N6-isopentenyladenosine at position 37 of tRNATyr to N6-isopentenyl-2-thiomethyladenosine (73), and this methylthiolation modification has been suggested to relate to sporulation in B. subtilis (74). The putative MiaB gene product (NP_389583) has been identified in B. subtilis strain 168 (75). The B. subtilis spore photoproduct lyase is a [4Fe-4S] enzyme that relies on the essential Cys141, possibly involving a cysteinyl radical intermediate, in the repair of thymine dimers that result from UV irradiation of B. subtilis spores (76). This enzyme is a major factor in the extreme resistance of sporulating cells to UV radiation. Cysteine is the major low-molecular-weight thiol in B. subtilis (22), and the yubC gene encoding cysteine dioxygenase (Cys-SH + O2 → Cys-SO2−) is significantly upregulated during sporulation (77). A yubC mutant was able to form spores, but with a reduced efficiency. These results suggest that further evaluation of CoADR-RHD, which is found in a number of spore-forming bacterial pathogens (e.g., B. anthracis and C. botulinum), should be pursued as a potential target for the design of therapeutic agents.
†This work was supported by National Institutes of Health Grant GM-35394 (A.C.), by WFUSM Development Funds (C.M.F.), and by National Science Foundation Grant MCB-9982727 (P.A.K.). Data for this study were measured at beamlines X12C and X26C of the National Synchrotron Light Source. Financial support comes principally from the Offices of Biological and Environmental Research and of Basic Energy Sciences of the U.S. Department of Energy and from the National Center for Research Resources of the National Institutes of Health.
‡Coordinates have been deposited with the Protein Data Bank under the file names 3ICR, 3ICS, and 3ICT.
1Abbreviations: RHD, rhodanese homology domain; CoADR, coenzyme A-disulfide reductase; CoADR-RHD, coenzyme A-disulfide reductase isoform with a C-terminal rhodanese homology domain; DTNB, 5, 5′-dithiobis(2-nitrobenzoate); CoAD, coenzyme A-disulfide; MMTS, methyl methanethiolsulfonate; SaCoADR, Staphylococcus aureus coenzyme A-disulfide reductase; BaCoADR-RHD, Bacillus anthracis coenzyme A-disulfide reductase isoform with a C-terminal rhodanese homology domain; BaCoADR, Bacillus anthracis coenzyme A-disulfide reductase; SeMet, selenomethionine; TNB, 2-nitro-5-thiosulfobenzoate; MAD, multiwavelength anomalous dispersion; LAB, data set collected in-house using a rotating anode X-ray source; SYNC1, synchrotron data set (peak1) collected over the first 180 degrees; SYNC6, final synchrotron data set (inflection2); ESI-TOF, electrospray ionization time-of-flight; GR, glutathione reductase; CoAS-II, the assigned half of coenzyme A-disulfide corresponding to the first CoASH product during reduction; EH2, two-electron reduced enzyme; EH4, four-electron reduced enzyme; E, oxidized enzyme.
2A. Claiborne, unpublished observations.
3Brito et al. (Ref 12) reported a similar peak of uninterpreted density near Cys350 in the archaeal sulfide:quinone oxidoreductase structure.
4During anaerobic titration of the SaCoADR Y419F mutant, complete flavin reduction is observed on addition of 1 equiv of NADPH/FAD; this is commensurate with the appearance of a strong charge-transfer band [E(FADH2-NADP+)] centered at 755 nm. This behavior is very distinct from that of wild-type SaCoADR, in which only one FAD/dimer is reduced with NADPH in static titrations, and demonstrates that Tyr419 is paramount to this intrinsic redox asymmetry.
5N. Bergman, personal communication.
6C. Paige, S.D. Reid, P.C. Hanna, and A. Claiborne, unpublished experiments.