Interaction of the Streptomyces Wbl protein WhiD with the principal sigma factor σHrdB depends on the WhiD [4Fe-4S] cluster

The bacterial protein WhiD belongs to the Wbl family of iron-sulfur [Fe-S] proteins present only in the actinomycetes. In Streptomyces coelicolor, it is required for the late stages of sporulation, but precisely how it functions is unknown. Here, we report results from in vitro and in vivo experiments with WhiD from Streptomyces venezuelae (SvWhiD), which differs from S. coelicolor WhiD (ScWhiD) only at the C terminus. We observed that, like ScWhiD and other Wbl proteins, SvWhiD binds a [4Fe-4S] cluster that is moderately sensitive to O2 and highly sensitive to nitric oxide (NO). However, although all previous studies have reported that Wbl proteins are monomers, we found that SvWhiD exists in a monomer-dimer equilibrium associated with its unusual C-terminal extension. Several Wbl proteins of Mycobacterium tuberculosis are known to interact with its principal sigma factor SigA. Using bacterial two-hybrid, gel filtration, and MS analyses, we demonstrate that SvWhiD interacts with domain 4 of the principal sigma factor of Streptomyces, σHrdB (σHrdB4). Using MS, we determined the dissociation constant (Kd) for the SvWhiD-σHrdB4 complex as ∼0.7 μm, consistent with a relatively tight binding interaction. We found that complex formation was cluster dependent and that a reaction with NO, which was complete at 8–10 NO molecules per cluster, resulted in dissociation into the separate proteins. The SvWhiD [4Fe-4S] cluster was significantly less sensitive to reaction with O2 and NO when SvWhiD was bound to σHrdB4, consistent with protection of the cluster in the complex.

in response to hypoxia and NO exposure in activated macrophages (17,18). Importantly, the interaction between WhiB1 and A 4 was found to be unaffected by the presence of O 2 , but highly sensitive to NO (12). Reaction with NO led to dissociation of the complex and a form of WhiB1 that can bind its own promoter (12,23).
Here, we report studies, using in vivo and in vitro approaches, of S. venezuelae WhiD (SvWhiD), showing that, like its Mycobacterial homologues, it forms a complex with domain 4 of the principal sigma factor HrdB , which depends on the [4Fe-4S] cluster. Reaction of [4Fe-4S] WhiD with NO results in nitrosylation of the cluster and dissociation of the complex.

Characterization of SvWhiD
Anerobically purified His-tagged SvWhiD was straw yellow in appearance with an absorbance spectrum typical of a [4Fe-4S] cluster protein, with a maximum located around 410 nm (Fig. 1A). Cluster loading was typically in the range ϳ90%. CD spectroscopy detects cluster optical transitions and is particularly sensitive to the cluster environment. The spectrum in Fig.  1B is very similar to that previously reported for S. coelicolor WhiD (2), consistent with the presence of a [4Fe-4S] cluster in a similar environment. The LC-MS spectrum of purified SvWhiD contained a major peak at 15,923 Da (predicted mass 15,924 Da) (Fig. S1), corresponding to the cluster-free (apo) protein with an N-terminal methionine cleavage. Two lower intensity peaks were also observed at 13,390 and 12,805 Da, corresponding to truncated forms of the protein, resulting from cleavage of 26 and 33 C-terminal residues, respectively, from the full-length protein. From LC-MS and SDS-PAGE (Fig. S1), the truncated forms were estimated to account for ϳ10% of the total protein.
Nondenaturing ESI-MS, in which noncovalently bound cofactors are retained upon ionization, has been shown to be extremely useful for determining the nature of the cluster in [Fe-S] proteins (22,(24)(25)(26)(27). Application here gave a deconvoluted spectrum containing a small peak at 15,919 Da corresponding to the apoprotein with all four cysteines in disulfide bridges. The major peak at 16,273 Da corresponded to [4Fe-4S] WhiD (predicted mass 16,273 Da) ( Fig. 1C and 1D). The spectrum also contained a peak at 13,154 Da, corresponding to the [4Fe-4S]-bound form of one of the truncated forms of SvWhiD (12,805 Da) (see Fig. S1), indicating that the 33 C-terminal residues are not important for cluster binding/stability.
Purification of WhiD under aerobic conditions also resulted in weakly colored fractions, consistent with previous findings that the S. coelicolor protein cluster is moderately resistant to A, the black line spectrum corresponds to that of as isolated SvWhiD, whereas exposure of purified SvWhiD to O 2 results in the red line spectrum, recorded after centrifugation of the apoprotein sample, which was prone to precipitation. B, the CD spectrum of as isolated SvWhiD is shown as a red line. Also plotted is the spectrum of a 2:1 mixture of HrdB 4 -WhiD, show in black demonstrating that the SvWhiD cluster environment is not significantly affected by binding to HrdB 4 . The additional intensity below 350 nm is because of the aromatic residues of HrdB 4 . C, deconvoluted ESI-MS spectrum of SvWhiD under nondenaturing conditions, containing peaks because of monomeric and dimeric forms. D, the monomeric and dimeric [4Fe-4S] SvWhiD peaks plotted on an expanded mass scale. For absorbance and CD spectroscopy experiments, SvWhiD was in 50 mM Tris, 300 mM NaCl, pH 7.2; for ESI-MS experiments, SvWhiD (10 M) was in 250 mM ammonium acetate, pH 7.2.

Iron-sulfur cluster-dependent binding of WhiD to HrdB
O 2 -mediated degradation. The sample lost color entirely 30 min post purification, resulting in apoprotein, as indicated by the complete disappearance of the 410-nm absorbance band (Fig. 1A).

SvWhiD exists in both monomeric and dimeric forms
SvWhiD was found to elute from an analytical gel filtration column ( Fig. 2A) as a broad peak with an elution volume indicative of a mass of ϳ27 kDa, midway between the masses of monomeric and dimeric forms. A low-intensity shoulder on the low mass side of the peak was observed, representing the truncated forms of SvWhiD. SDS-PAGE showed the full-length protein was present across the full elution profile, whereas the truncated forms were observed only in the lower mass fractions, indicating that they exist only as a monomer (Fig. 2B). ApoWhiD was also found to behave as a monomer-dimer mixture, but with some further higher mass species at lower elution volumes (Fig. 2, A and B). Nondenaturing ESI-MS revealed the presence of both monomer and dimer forms of [4Fe-4S] SvWhiD ( Fig. 1C and D). Together, the data indicate that the presence or absence of the cluster does not affect the association state of SvWhiD. The absence of the cluster promotes the formation of higher mass forms of WhiD, which may arise from oxidation of Cys residues, resulting in crosslinking of SvWhiD monomers.

The C-terminal extension of SvWhiD is crucial for protein dimerization
The role of the C-terminal extension in SvWhiD dimerization was explored further by generating a truncated form of WhiD which lacked 33 residues at the C terminus. Truncated SvWhiD was found to elute from an analytical gel filtration column at a volume indicative of a monomer (Fig. S2), and SDS-PAGE confirmed the presence of truncated SvWhiD across the elution profile. UV-visible absorbance and CD spectra revealed no significant differences between truncated and full-length SvWhiD (Fig. S3), indicating that the absence of the C-terminal extension does not affect the cluster environment.
Nondenaturing ESI-MS revealed the presence of only mono- SvWhiD was not observed using parameters optimized to observe dimeric full-length SvWhiD (Fig. S3). Apo-truncated SvWhiD was not observed because of high cluster load (Ͼ90%). Thus, we conclude that the C-terminal part of SvWhiD is essential for dimerization.
A general conclusion from previous characterizations of Wbl proteins is that they are monomeric proteins (2,12,28), and so these findings for SvWhiD were unexpected. We note that SvWhiD has a C-terminal extension of 18 residues compared with ScWhiD, and, although the proteins exhibit overall 78% sequence identity, only 1 of the last 33 residues of SvWhiD is conserved in ScWhiD (the proteins are 100% identical from residues 1-95) (Fig. S4). Therefore, the difference in the C-terminal part is likely to be functionally important. BLAST analysis revealed comparable putative WhiD sequences (sharing Ն60% identity with the SvWhiD C-terminal extension) to be present in some Streptomyces species. Secondary structure analysis of SvWhiD, using Phyre2 (29) suggested that the core of SvWhiD resembles that recently reported for WhiB1 (12,15), whereas the C-terminal extension is predicted to constitute an additional helix (SvWhiD residues 104 -121) (see Fig. S4).

SvWhiD [4Fe-4S] cluster reacts with NO
Wbl proteins from a variety of species have been shown to react slowly with O 2 but rapidly with ϳ8 -10 NO in a concerted manner, resulting in protein-bound iron-nitrosyl species (19 -23). Therefore, the reaction of SvWhiD with NO was investigated. Stepwise titration of [4Fe-4S] WhiD with PROLI-NONOate, to give 0 -20 NO per cluster, resulted in mild precipitation, suggesting that the nitrosylation intermediates/products of SvWhiD may be less stable than those of ScWhiD (22). Scattering because of precipitation distorted the spectral changes upon nitrosylation (Fig. S5), but a shoulder at 362 nm was observed that is indicative of the formation of iron-nitrosyl species. Spectral changes could be more readily visualized by plotting the difference between absorbance at 410 and 362 nm as a function of NO per cluster (Fig. 3A). This shows that the reaction was complete at ϳ9 NO per cluster, demonstrating the reaction of multiple NO

Iron-sulfur cluster-dependent binding of WhiD to HrdB
Iron-sulfur cluster-dependent binding of WhiD to HrdB molecules with each cluster, consistent with data previously reported for S. coelicolor WhiD (19).
The reaction was also investigated by nondenaturing ESI-MS. Deconvoluted mass spectra measured at increasing concentrations of NO (Fig. 3B) revealed increasing abundance of peaks because of apoWhiD, and one and two sulfur adducts of apoWhiD (ϩ32 and ϩ64 Da), along with the eventual loss of [4Fe-4S] WhiD. Absolute ion counts for these species were then used to determine percentage abundance of the [4Fe-4S] form as a function of NO per cluster (Fig. 3C). The plot is similar to that of absorbance data in Fig. 3A in that it indicates that ϳ9 -10 NO molecules are required for full reaction of the SvWhiD [4Fe-4S] cluster. However, the plot is nonlinear, indicating that the reaction is not fully concerted such that the reaction of one cluster with NO does not go entirely to completion before the reaction at another cluster begins.
Interestingly, no nitrosylated forms of the WhiD [4Fe-4S] cluster, nor any iron-nitrosyl products, were detected by nondenaturing MS. Such species were recently detected by MS for ScWhiD (22), suggesting that NO complexes of SvWhiD are less stable under the conditions of the MS experiments than those of ScWhiD.

SvWhiD interacts specifically with the essential principal sigma factor HrdB
To gain insight into SvWhiD function, we screened a bacterial adenylate cyclase two-hybrid (BACTH) (30) shotgun S. venezuelae sonicated DNA genomic library using whiD as bait, to look for interacting proteins. Five of the 13 positive clones isolated carried in-frame fusions to the same gene, hrdB, encoding the essential principal sigma factor, HrdB (31)(32)(33). Further-more, the five positive clones carried only the 3Ј of the hrdB gene (Fig. S6). Among these five clones, the most C-terminal fusion started at Asp-488, corresponding to the beginning of domain 4 of HrdB (Fig. S6), which is responsible for binding to the Ϫ35 element of target promoters. This suggested that interaction with WhiD was principally mediated by this domain. To confirm and extend this analysis, we used the BACTH system to measure the interaction of WhiD with (i) full-length HrdB , (ii) domain 4 alone, and (iii) HrdB lacking domain 4. This analysis showed strong interaction of WhiD with full-length HrdB and with domain 4 alone, but none with HrdB lacking domain 4 (Fig. 4). These results confirmed that the interaction with SvWhiD is principally mediated by domain 4 of HrdB . To determine whether this interaction is specific to the principal sigma factor, we also tested the interaction between WhiD and the closely related sigma factor HrdD (32,33). WhiD and HrdD did not interact, suggesting that WhiD interaction is indeed specific for HrdB (Fig. 4).
To complement the in vivo interaction data, domain 4 of the S. venezuelae sigma factor HrdB ( HrdB 4 ), harboring the HTHmotif that binds to the Ϫ35 promoter element, was expressed and purified, resulting in a His-tagged protein of ϳ11 kDa (Fig.  S7). The lack of Trp and Tyr residues in this HrdB domain resulted in a low extinction coefficient at 280 nm (⑀ ϭ 2850 M Ϫ1 cm Ϫ1 ), and so a 5-fold excess of HrdB 4 was mixed with SvWhiD to enable detection of the protein through its absorbance upon elution from an anaerobic gel filtration column. The elution profile of the SvWhiD/ HrdB 4 mixture contained peaks corresponding to the individual proteins, as judged from elution profiles of the individual proteins run down the column separately   (Fig. 5A). However, also present was a peak at higher mass that was not observed in the elution profiles of the separate proteins. This peak corresponded to a mass of ϳ35 kDa, too low to indicate a (WhiD) 2 -HrdB 4 complex, but higher than that predicted for a monomeric WhiD-HrdB 4 complex. Thus, it is apparent that the SvWhiD monomer-dimer equilibrium described above remains a feature upon complex formation with HrdB 4 . Consistent with this, SDS-PAGE analysis of the gel filtration elution fractions demonstrated the presence of HrdB 4 (and WhiD) across the high-mass fractions (Fig. 5B); equivalent fractions were devoid of HrdB 4 for the separately run protein. SDS-PAGE also suggested the presence of a component of aggregated WhiD/ HrdB 4 at Ͼ50 kDa (Fig. 5B). The specificity of complex formation with HrdB 4 was tested by equivalent experiments with a protein containing HrdD 4 . SvWhiD and HrdD 4 were mixed in a 1:10 ratio (extinction coefficient for HrdD 4 is ⑀ 280 nm ϭ 1490 M Ϫ1 cm Ϫ1 , and so a higher concentration than that employed for HrdB 4 experiments was used) and analyzed by gel filtration and SDS-PAGE. No evidence of an interaction between the proteins was observed; the elution profile of the WhiD/ HrdD 4 mixture was a superposition of the profiles of the individual proteins at ϳ27 kDa and ϳ11 kDa, with no higher mass complex (Fig. 5, C and D). Thus, SvWhiD does not interact with domain 4 of HrdD .

Iron-sulfur cluster-dependent binding of WhiD to HrdB
Nondenaturing MS was used to investigate the interaction between HrdB 4 and SvWhiD. The deconvoluted spectra of a 2:1 mixture of HrdB 4 and SvWhiD revealed the individual proteins, as well as a major peak corresponding to the mass of a WhiD [4Fe-4S]-HrdB 4 complex, along with peaks because of oxygen and/or sulfur adducts (Fig. 6). Two minor species were also observed, at 13,154 Da and 24,619 Da, corresponding to the [4Fe-4S]-bound form of 33-residue truncated SvWhiD alone and in complex with HrdB 4 . This indicates that the C-terminal part of WhiD is not important for interaction with the sigma factor. CD spectroscopy of a 2:1 WhiD/ HrdB 4 mixture showed that complex formation has no significant effect on the cluster environment (Fig. 1B).
The dissociation constant for the interaction between SvWhiD and HrdB 4 was determined using ESI-MS. Multiple samples were prepared containing an increasing concentration

Iron-sulfur cluster-dependent binding of WhiD to HrdB
of HrdB 4 . To assist with quantification, each sample contained a fixed concentration of a protein standard, I151A FNR, 2 which has a mass of 29,123 Da, close to that of the SvWhiD-HrdB 4 complex. m/z spectra (Fig. S8) were deconvoluted and the mass regions of the complex and protein standard plotted (Fig. 7A). Absolute ion counts for the complex and FNR standard were used to determine the extent of complex formation (fractional saturation) and these data were plotted as a function of free HrdB 4 concentration (Fig. 7B) and fitted using a simple binding isotherm. This gave K d ϭ 7.4 (Ϯ1.4) ϫ 10 Ϫ7 M, which indicates a relatively tight binding between the two proteins.

The SvWhiD-HrdB 4 complex depends on the [4Fe-4S] cluster and stabilizes it against O 2 -mediated degradation
ApoSvWhiD was mixed with HrdB 4 under the same conditions as [4Fe-4S] SvWhiD at a 1:5 ratio and analyzed by gel filtration (Fig.  S9). No evidence for complex formation was observed, indicating that the cluster is required for the SvWhiD-HrdB 4 interaction. This is consistent with the data reported for M. tuberculosis WhiB1-SigA (12,15), and with the proposal that the G[V/I]WGG motif is important for mediating protein-protein interactions (11). Indeed, the recent M. tuberculosis WhiB1-70 4 complex structure showed that a number of residues, including Val-59 and Trp-60 of the conserved motif, as well as Trp-3, Phe-17, and Phe-18 (also conserved in SvWhiD), participate in complex-stabilizing hydrophobic/hydrogen bonding interactions close to the [4Fe-4S] clus-

Iron-sulfur cluster-dependent binding of WhiD to HrdB
ter (15). Absence of the cluster would thus be expected to disrupt these interactions and lead to loss of the complex, as found for WhiB1 (12).
Exposure of SvWhiD [4Fe-4S] to aerobic buffer with and without HrdB 4 revealed that the presence of HrdB 4 protected the WhiD [4Fe-4S] cluster from oxidative degradation. In the absence of HrdB 4 , SvWhiD [4Fe-4S] was destabilized after 15 min, resulting in precipitation causing increased scattering (Fig. S10). In the presence of HrdB 4 , a minor decrease in A 410 nm was observed but the cluster absorbance remained largely stable for more than 1 h. Given the lack of change in the CD upon complex formation (Fig. 1B), the protective effect is likely to arise from increased stability of the protein and/or limited accessibility of O 2 to the cluster. We note that the M. tuberculosis WhiB1-70 4 complex structure revealed a solvent-inaccessible [4Fe-4S] cluster (15).

NO-mediated dissociation of the SvWhiD-HrdB 4 complex
Previous studies of Wbl proteins have demonstrated the sensitivity of the SvWbl [4Fe-4S] cluster to NO, suggesting that some Wbl proteins may function as NO sensors (17,19,(21)(22)(23). Reaction of the SvWhiD-HrdB 4 complex with 20 NO per cluster resulted in the nitrosylation of the [4Fe-4S] cluster (Fig. 8A) and dissociation of the complex, as evidenced by gel filtration and SDS-PAGE analysis of the elution fractions (Fig. 8, B  and C).
The effect of NO on the SvWhiD-HrdB 4 complex was also investigated using nondenaturing MS (12,22). Fig. 9A shows deconvoluted mass spectra in the region corresponding to the complex. The data show clearly the loss of the complex as NO was added, and a plot of intensity as a function of NO per cluster (Fig. 9B) indicates that the complex was lost entirely after addition of ϳ8 NO molecules per cluster. The form of the plot is similar to that observed for the loss of [4Fe-4S] SvWhiD upon reaction with NO (Fig. 3), suggesting that the same process (i.e. reaction of NO with the cluster) is controlling both the loss of the cluster and dissociation of the complex. Again, the plot is not linear, suggesting that the reaction is not fully concerted.
Stopped-flow kinetic measurements were performed to determine whether the rate and mechanism of the reaction of the SvWhiD cluster is affected by the presence of HrdB 4 . Kinetic measurements of reaction of SvWhiD with NO in the absence of HrdB 4 were carried out first to determine whether this Wbl protein behaves similarly to those previously characterized: ScWhiD and M. tuberculosis WhiB1 (19). For these proteins, kinetic data were consistent with a four-step mechanism of reaction of the Wbl [4Fe-4S] cluster with NO, with three of these steps detectable at 360 nm (19). For SvWhiD, it was immediately apparent that a full kinetic analysis would not be possible because of the instability of the protein during the nitrosylation reaction, with precipitation occurring, as noted above. However, this did not begin to occur until after 500 ms, enabling measurement of the early part of the reaction, Fig.  10A. The data revealed the presence of two early phases, which have a similar form to those reported for other Wbl proteins. Plots of the observed apparent first order rate constants (k obs ) against NO concentration were linear for the two phases detected (Fig. S11), indicating two sequential NO reactions that are each first order with respect to NO. The derived rate constants are consistent with those reported for the equivalent phases of nitrosylation of ScWhiD and M. tuberculosis WhiB1 (kinetic data are summarized in Table 1). Thus, we conclude that nitrosylation of SvWhiD most likely occurs via a mechanism that is similar to that of other Wbl proteins.
Stopped-flow measurements were then carried out for the SvWhiD-HrdB 4 complex (with a 4-fold excess of HrdB 4 ). Some significant differences were observed. The first phase observed for SvWhiD alone was found to correspond to two consecutive phases in the reaction of the complex with NO, which together occurred over a longer time period (Fig. 10B), indicating that [4Fe-4S] SvWhiD is protected to some degree from reaction with NO. Subsequent to this, a phase resembling the second phase of the reaction of SvWhiD alone was observed but kinetic analysis of it was precluded because of precipitation that caused a rising absorbance baseline. Thus, kinetic analyses were focused on the initial phases.
Experiments in which the concentration of NO was varied revealed that the rate of these two initial phases was essentially

Iron-sulfur cluster-dependent binding of WhiD to HrdB
independent of the NO concentration (Fig. 10C), demonstrating that the rate-limiting step in the reaction of the SvWhiD-HrdB 4 complex with NO does not involve NO. This is in direct contrast to the reaction of [4Fe-4S] SvWhiD alone with NO, for which a first-order relationship was observed. One possible explanation is that reaction of the cluster with NO cannot occur in the complex in its principal conformation and is thus dependent on a reversible conformational change or dissociation event in order to occur. The data indicate that this initial conformational change/dissociation occurs with a rate constant of ϳ100 s Ϫ1 and is the rate-limiting step of the reaction with NO. The fact that the second phase also appears to be independent of NO concentration suggests that a second conformational change, that could be dependent on the first reaction with NO, is necessary in order for further reaction with NO to occur, and that this is rate-limiting for the second phase of the nitrosylation reaction. These data are consistent with the buried nature of the [4Fe-4S] cluster of WhiB1 in complex with 70 4 (15), which suggests that an opening up of the structure may be needed for reaction to occur.

Conclusions
WhiD from S. venezuelae, like other Wbl proteins, binds a [4Fe-4S] cluster. Unusually, however, it exists in a monomerdimer equilibrium, with the monomer-monomer interaction dependent on the C-terminal part of the protein, which is the only part of the protein that is different from the previously characterized monomeric S. coelicolor WhiD protein. The significance of dimerization is unclear, but it is interesting to note that the C-terminal extension, which is predicted to form a helix, is conserved in a range of other WhiD homologues.
SvWhiD forms a tight (K d Ͻ 1 M) and specific complex with the principal sigma factor in Streptomyces, HrdB . Although interactions between M. tuberculosis Wbl proteins and the principal sigma factor in Mycobacterium (SigA) have been widely reported, this is the first demonstration that a Wbl pro-

Iron-sulfur cluster-dependent binding of WhiD to HrdB
tein in Streptomyces functions similarly. Thus, the ability to interact with the principal sigma factor is likely to be a conserved feature of Wbl function throughout the actinomycetes. In Streptomyces and Corynebacteria the ability of the Wbl pro-tein WhiB to function as a transcription factor depends upon a direct interaction with the unrelated transcription factor WhiA. Given our growing understanding of Wbl-sigma interactions, it is possible that a tripartite complex of HrdB , WhiB and WhiA regulates gene expression. One possible explanation for such a requirement could be the lack of a distinct DNAbinding motif in Wbl proteins. WhiB, like most Wbl family members, only carries a series of C-terminal, positively charged amino acid residues that may increase affinity for DNA (the exception is WblC/WhiB7, which binds DNA via a C-terminal AT-hook motif). Thus, the ability of WhiB to function as a transcription factor is mediated via WhiA, which carries an HTH motif. The interaction between the Wbl-family member WhiB and another transcription factor, WhiA, raises the possibility that other Wbl proteins function via interactions with proteins other than the principal sigma factor. Work toward resolving this question is ongoing.
The interaction between Wbls and the principal sigma factor depends on the [4Fe-4S] cluster, shown both here in Streptomyces between WhiD and HrdB and in Mycobacteria between WhiB1 and SigA. Thus, reaction with O 2 or NO, which results in cluster degradation, leads to disassembly of the complex, illustrating a likely mechanism by which environmental signals could be transduced to regulatory responses. In the case of reaction with NO, this is a multistep process as previously described for other Wbl proteins. Complex formation significantly protects the cluster from O 2 -mediated degradation.
The [4Fe-4S] cluster is also protected to some degree in the complex from reaction with NO, and it is reasonable to suggest that the accessibility of NO to the cluster is impaired because of the interaction of WhiD with HrdB . However, reaction still occurs, most likely because of conformational flexibility or reversible dissociation that occurs more slowly than the initial reaction of the unhindered cluster with NO. We note that the overall effect, however, is that the reactions with O 2 and NO are kinetically even more distinct than for [4Fe-4S] SvWhiD alone, such that the SvWhiD-HrdB complex is optimally arranged to distinguish between O 2 and NO.

Overexpression and purification of SvWhiD
Bacterial strains, plasmids and oligonucleotides used in this study are listed in Table S1. A codon-optimized gene was synthesized (GenScript) and subsequently ligated into pET28a using Ndel and HindIII sites, generating pMSW1, for the expression of N-terminally His 6 -tagged S. venezuelae WhiD in Escherichia coli. The protein was overproduced in 5 liters (8 ϫ 1 liter flasks) aerobically grown E. coli BL21 (DE3) cultures in LB containing 50 g/ml kanamycin. Cultures were grown at 37°C with shaking at 200 rpm until A 600 reached 0.6 -0.8, at  Unless stated, all purification steps were performed in an anaerobic cabinet (O 2 Ͻ 2 parts/million). Cell pellets were resuspended in buffer A (50 mM Tris, 300 mM NaCl, 25 mM imidazole, pH 7.5) with the addition of lysozyme (300 g/ml) and PMSF (300 g/ml). Resuspended cells were removed from the anaerobic cabinet and lysed by sonication on ice under N 2 , twice for 8 min 20 s, 0.2-s bursts, 50% power and immediately returned to the anaerobic cabinet. Lysed cells were centrifuged in air-tight centrifuge tubes outside of the anaerobic cabinet at 40,000 ϫ g for 45 min at 1°C and returned into the anaerobic cabinet.
The supernatant was loaded onto a HiTrap Ni-affinity column, washed with buffer A until A 280 nm was Ͻ0.1. Bound proteins were eluted with buffer B (50 mM Tris, 300 mM NaCl, 500 mM imidazole, pH 7.5) from 0 to 100% (v/v) over a linear gradient of 10-ml fractions containing WhiD were pooled and desalted using a HiTrap desalting column into buffer C (50 mM Tris, 300 mM NaCl, pH 7.5) and stored in an anaerobic freezer until needed. Total protein concentration was determined using the Bradford (Bio-Rad) (34) or Rose Bengal (35) assays, with BSA as calibration standard. Purity of the protein was checked using SDS-PAGE gel electrophoresis and LC-MS. Cluster concentration was determined by reference to a calibration curve generated from Fe 3ϩ solutions prepared from SpectrosoL standard iron solution (36), or by using an absorbance extinction coefficient at 410 nm of 17,500 M Ϫ1 CM Ϫ1 .
A codon-optimized gene encoding a truncated form of WhiD lacking the 33 C-terminal residues was synthesized (GenScript) and subsequently ligated into pET28a using Ndel and HindIII sites, for the expression of N-terminally His 6tagged S. venezuelae truncated WhiD in E. coli. The protein was overexpressed and purified as described for full-length SvWhiD. Protein and cluster concentrations were determined as above.

Overexpression and purification of domain 4 of HrdB and HrdD
Codon-optimized genes for the expression of N-terminally His 6 -tagged HrdB 4 and HrdD 4 were also synthesized (Gen-Script) and ligated into pET15b using Ndel and BamHI sites, generating pMSW2 and pMSW3. S. venezuelae HrdB 4 and HrdD 4 proteins were overproduced in 5 liters (8 ϫ 1 liter flasks) aerobically grown E. coli BL21 (DE3) cultures in LB containing 100 g/ml ampicillin. Cultures were grown at 37°C, 200 rpm until A 600 reached 0.6 -0.8, at which point overexpression of proteins was induced by the addition of 0.5 mM IPTG. Cultures were incubated further at 37°C, 200 rpm for 4 h. Cells were harvested by centrifugation at 8000 rpm at 4°C for 10 min and stored at Ϫ80°C until required. S. venezuelae HrdB 4 and HrdD 4 proteins were purified as described above for WhiD, except that all steps were carried out under aerobic conditions.

Overexpression and purification of I151A FNR
Aerobic cultures of E. coli BL21DE3 containing pGS2252 (encoding I151A GST-FNR) were grown and protein isolated as described previously (37), except that aerobic conditions were employed to generate the cluster-free form of I151A GST-FNR. I151A FNR was cleaved from the fusion protein using thrombin, as described previously (38).

Bacterial two-hybrid (BACTH) genomic library construction, screening, and analysis
Construction of genomic BACTH libraries was performed by BIO S&T (Saint-Laurent, Québec, Canada) as described previously (39). To construct the "bait" vector, the whiD gene was amplified using the whiD_BACTH_F and whiD_BACTH_R primers and cloned into the pKT25 plasmid digested with XbaI and BamHI to create the plasmid pIJ10921 encoding the T25 domain of adenylate cyclase fused to the N terminus of WhiD (T25-WhiD). The E. coli BTH101 strain was transformed with pIJ10921 before electroporation of ϳ0.125 g of the T18C genomic library. Transformations were recovered in SOC medium, washed twice with M63 and then plated onto M63 minimal medium, supplemented with 0.3% lactose, 50 g/ml Carb, 25 g/ml Kan, 0.5 mM IPTG, and 40 g/ml X-gal. Plates were incubated for 5-10 days at 30°C and colonies were then restreaked onto MacConkey agar supplemented with 1% maltose, 0.5 mM IPTG, 100 g/ml Carb, and 50 g/ml Kan, incubating for 2 days at 30°C. Strongly interacting clones (cyaAϩ), identified by their red color were grown in liquid culture overnight, selecting only for the pUT18C "prey" plasmid (100 g/ml Carb). DNA was isolated by Miniprep (Qiagen) and sequenced using the pUT18C_F primer.
BACTH vectors encoding the T18 domain of adenylate cyclase fused to the N terminus of HrdB and HrdD (T18-HrdB and T18-HrdD ) were constructed. Full-length hrdB and hrdD genes were amplified using the primer pairs hrdB_BACTH_F/ hrdB_BATCH_R and hrdD_BACTH_F/hrdD_BACTH_R, respectively, and cloned into the pUT18C plasmid digested with XbaI and KpnI to create plasmids pIJ10922 and pIJ10923. Sequences encoding HrdB region 4 only and HrdB lacking region 4 were amplified using the primer pairs hrdB4_ BACTH_F/hrdB4_BACTH_R and hrdBd4_BACTH_F/hrdBd4_ BACTH_R, respectively, and similarly cloned to generate plasmids pIJ10925 and pIJ10926. To test interactions between proteins, E. coli BTH101 was co-transformed with the appropriate pKT25 and pUT18C fusion plasmids. ␤-gal assays were conducted as in previous studies (e.g. (40,41)), according to standard methodology (42). Cultures were additionally spotted (7.5 l) onto M63 minimal medium, supplemented with 0.3% lactose, 50 g/ml Carb, 25 g/ml Kan, 0.5 mM IPTG, and 40 g/ml X-gal.

Analytical gel filtration
Gel filtration was performed aerobically using a pre-calibrated Superdex 75 10/300 GL column, at room temperature in buffer D (50 mM Tris, 300 mM NaCl, pH 8.0) with a flow rate of 0.5 ml/min. Buffers were purged with N 2 and kept in an anaerobic glovebox overnight until needed. Mass of proteins was estimated by reference to a calibration curve generated using Iron-sulfur cluster-dependent binding of WhiD to HrdB BSA, bovine erythrocyte carbonic anhydrase, and horse heart cytochrome c at 10 mg/ml, 3 mg/ml, and 2 mg/ml, respectively, dissolved in buffer D. Protein fractions were then analyzed by SDS-PAGE gel and visualized by silver stain using standard protocols (ProteoSilver TM , Sigma-Aldrich).

Spectroscopic measurements
UV-visible absorbance measurements were made using a Jasco V550 spectrometer. CD spectra were measured with a Jasco J810 spectropolarimeter. Samples for spectroscopy were in buffer C (50 mM Tris, 300 mM NaCl, pH 7.5).

Mass spectrometry
SvWhiD and HrdB 4 were buffer exchanged into 250 mM ammonium acetate, pH 7.2, inside an anaerobic cabinet using mini-PD10 (GE Healthcare) desalting columns. The concentration of [4Fe-4S] WhiD was determined via absorbance at 410 nm and the concentration of HrdB 4 in ammonium acetate was determined using a Bio-Rad dye kit, as described above.
Proteins or protein mixtures were infused directly into the ESI source of a Bruker micrOTOF-QIII mass spectrometer operating in positive ion mode using at 0.3 ml/h using a syringe pump. MS data were acquired over the m/z range of 1000 -3500 continuously for 5 min using Bruker oTOF Control software, with parameters as follows: dry gas flow of 4 liters/min, nebulizer gas pressure 0.8 Bar, dry gas 180°C, capillary voltage 4500 V, offset 500 V, ion energy 5 eV, collision RF 650 Vpp, collision cell energy 10 eV. Processing and analysis of MS experimental data were carried out using Compass DataAnalysis version 4.1 (Bruker Daltonik, Bremen, Germany), and neutral mass spectra were generated using the ESI Compass version 1.3 Maximum Entropy deconvolution algorithm. Exact masses (Ϯ 1 Da) are reported from peak centroids representing the isotope average neutral mass. For apoproteins, exact masses were derived from m/z spectra, for which peaks correspond to [M ϩ nH] nϩ /n. For [4Fe-4S] 2ϩ WhiD, peaks corresponded to [M ϩ [4Fe-4S] 2ϩ ϩ (n-2)H] nϩ /n, where M is the molecular mass of the protein, [4Fe-4S] is the mass of the iron-sulfur cluster of 2ϩ charge, H is the mass of the proton, and n is the total charge (22,24,43).
For NONOate experiments, the reaction syringe was maintained at 25°C. The mass spectrometer was calibrated with ESIlow concentration tuning mix in positive ion mode (Agilent Technologies). For ESI-MS NO titration experiments, MS intensity data were processed to generate relative abundance plots of ion counts for the [4Fe-4S] form as a fraction of the total ion count because of [4Fe-4S]-bound and apo forms. This permitted changes in relative abundance to be followed without distortions because of variations in ionization efficiency that normally occur across a data collection run. For LC-MS, proteins were diluted in an aqueous mixture of 2% (v/v) acetonitrile and 0.1% (v/v) formic acid and analyzed as described previously (21).
For the determination of the dissociation constant for the WhiD-HrdB 4 complex, FNR I151A (37) was used as an internal standard. Solutions of equimolar WhiD and FNR (3 M final concentrations), pre-exchanged into 250 mM ammonium acetate, pH 7.2, were mixed with increasing concentrations of HrdB 4 in the same buffer to give molar ratios of HrdB 4 -WhiD between 0 and 7. Mixed samples were incubated for 5 min before loading into a 500 l gastight syringe and infused directly into the mass spectrometer operating in positive mode. Parameters were as described above for nondenaturing experiments. Each data point was collected over 2 min average (three replicas) and deconvoluted over the range of 10 to 30 kDa. Ion counts for the WhiD-HrdB 4 complex were compared with the combined ion counts for FNR and complex. Data were then expressed as fractional saturation and fitted using a simple binding isotherm from which a K d for the complex was obtained.

Nitric oxide reactivity experiments
For nondenaturing ESI-MS experiments, NO donor DEA-NONOate (Sigma-Aldrich) solutions were prepared immediately before use, in 4°C ammonium acetate buffer and quantified by absorbance at 250 nm (⑀ ϭ 6500 M Ϫ1 cm Ϫ1 ). The half-life of DEA-NONOate at 25°C in 250 mM ammonium acetate buffer, pH 7.2, was determined as t1 ⁄ 2 ϳ7 min, yielding overall 1.5 NO molecules per NONOate (22). DEA-NONOate was added directly to WhiD samples to give a specific ratio of NO to [4Fe-4S] cluster of 20 over the course of 50 min. Spectra were averaged over 2.5 min to obtain data at intervals of 1.1 NO per cluster. For UV-visible absorbance experiments, PROLI-NONOate (Cayman Chemicals) solutions were prepared in 50 mM NaOH and quantified by absorbance at 252 nm (⑀ ϭ 8400 M Ϫ1 cm Ϫ1 ). PROLI-NONOate was titrated into the sample and incubated for 5 min at ambient temperature prior to measurement, to allow full NO release from NONOate (t1 ⁄ 2 ϳ2 s).
For UV-visible stopped-flow experiments, a Pro-Dataupgraded Applied Photophysics Bio-Sequential DX.17 MV spectrophotometer was used, with a 1-cm path length cell. Absorption changes were detected at 360 nm. Experiments were carried out in 50 mM Tris 300 mM NaCl, pH 7.2) using gastight syringes (Hamilton) at 25°C. Prior to use, the stoppedflow instrument was flushed with ϳ50 ml of anaerobic buffer. Solutions of SvWhiD- HrdB 4 complex were prepared at a 4 to 1 excess of HrdB 4 to SvWhiD. All solutions used for stopped-flow experiments were stored and manipulated inside an anaerobic cabinet (Belle Technology). NO solutions were prepared using PROLI-NONOate, as described above. Final traces are averages of 10 individual traces. Fitting of kinetic data were performed using OriginPro8 (OriginLabs).

Data availability
All of the data are contained within the main paper and supporting information. In addition, x,y data for figure plots and original gel images are available at Open Science Framework, doi: 10.17605/OSF.IO/HVC9R.