Structural and Functional Insights into Bacillus subtilis Sigma Factor Inhibitor, CsfB.

Global changes in bacterial gene expression can be orchestrated by the coordinated activation/deactivation of alternative sigma (σ) factor subunits of RNA polymerase. Sigma factors themselves are regulated in myriad ways, including via anti-sigma factors. Here, we have determined the solution structure of anti-sigma factor CsfB, responsible for inhibition of two alternative sigma factors, σG and σE, during spore formation by Bacillus subtilis. CsfB assembles into a symmetrical homodimer, with each monomer bound to a single Zn2+ ion via a treble-clef zinc finger fold. Directed mutagenesis indicates that dimer formation is critical for CsfB-mediated inhibition of both σG and σE, and we have characterized these interactions in vitro. This work represents an advance in our understanding of how CsfB mediates inhibition of two alternative sigma factors to drive developmental gene expression in a bacterium.

Introduction

Eukaryotic and prokaryotic cells alike possess the ability to alter their phenotypes through global changes in gene expression. In bacteria, these transitions enable survival during stress conditions, drive developmental programs, and promote infection of host organisms. One common mechanism bacteria utilize to effect large-scale changes in gene expression is through alternative sigma (σ) factor subunits of RNA polymerase (RNAP). The dissociable RNAP sigma factor subunit is responsible for recognition of promoter DNA and the subsequent initiation of transcription. Most sigma factors are members of the σ70 superfamily, which is subdivided into four classes based upon conservation and the presence/absence of the conserved sigma domains (σ1.1, σ2, σ3, and σ4) that mediate interactions with RNAP and/or promoter DNA (reviewed in Feklistov et al., 2014, Paget, 2015). All bacteria employ an essential primary sigma factor (class I) that directs transcription of housekeeping genes; many bacteria also possess alternative sigma factors (classes II, III, and IV) that compete for binding to RNAP and redirect it to transcribe sets of genes required for adaptive responses. Hence, the suite of genes expressed in a bacterial cell can be reprogrammed by manipulating the levels, activity, or availability of alternative sigma factors (reviewed in Osterberg et al., 2011).

One prevalent form of post-translational regulation of alternative sigma factors occurs via anti-sigma factors: proteins that bind to and prevent their cognate sigma factor from interacting with RNAP. Unlike sigma factors, which share sequence, structural, and functional conservation, anti-sigma factors are more diverse in their sequences, structures, and/or mode of sigma factor inhibition (reviewed in Paget, 2015). A number of structural and bioinformatics analyses have revealed that anti-sigma factors for the class IV extracytoplasmic function (ECF) sigma factors often share one of two conserved anti-sigma domain structures, despite little sequence conservation (reviewed in Campagne et al., 2015). Less is known, however, of the structural features of anti-sigma factors that antagonize non-class IV alternative sigma factors, given the limited number of structures determined to date (Campbell et al., 2002, Masuda et al., 2004, Sorenson et al., 2004).

Here, we have structurally analyzed CsfB (also called Gin), a small, Zn2+-binding anti-sigma factor that inhibits two class III alternative sigma factors during spore formation by the model bacterium Bacillus subtilis (Figure 1A) (Chary et al., 2007, Decatur and Losick, 1996, Karmazyn-Campelli et al., 2008, Rhayat et al., 2009, Serrano et al., 2011, Serrano et al., 2015). In the forespore cell (the nascent spore), CsfB binds and inhibits the late-acting sigma factor σG, helping to ensure that it does not become active before the early-acting sigma factor σF has completed its program of gene expression (Karmazyn-Campelli et al., 2008, Rhayat et al., 2009). In the mother cell, which helps support the development of the forespore, CsfB binds the early-acting sigma factor σE, helping to inactivate it after the switch to σK (Serrano et al., 2015). Here, we report the structure of CsfB and characterize its interaction with σG and σE.

Figure 1

The Anti-sigma Factor CsfB Helps to Orchestrate the Switch from Early to Late Gene Expression during B. subtilis Sporulation

(A) Cartoon depiction of the role of the dual-specificity anti-sigma factor CsfB in regulating the transition from early to late gene expression during B. subtilis sporulation. Early in sporulation (reviewed in Tan and Ramamurthi, 2014), an asymmetric cell division event produces two cells: a smaller forespore (the nascent spore) and a larger mother cell. Initially, these two cells lie side-by-side; the mother cell then engulfs the forespore in a phagocytic-like process. At early times, σF and σE drive gene expression in the forespore and mother cell, respectively. Among the genes activated by σF and σE are those encoding the late-acting sigma factors, σG and σK, respectively (dashed arrows). The anti-sigma factor CsfB is expressed in both compartments under the control of σF and σK (dashed arrows). In the forespore, CsfB antagonizes σG at early times (barred line). In the mother cell, CsfB antagonizes σE at later times (barred line).

(B) 1H-15N HSQC spectrum of CsfB (orange). Full assignment of the cleaved CsfB version appears in black (CsfB1−48), partial assignment of the residual full-length CsfB in blue and the tag residues in gray; sc denotes side chain resonances. The C-terminal residue from the cleaved version (A48) is highlighted by a green square.

Results

Recombinant CsfB Degrades to a Stable but Nonfunctional Domain

We produced recombinant N-terminally histidine-tagged full-length CsfB (residues 1–64), but the protein consistently degraded to a stable product comprising residues 1–48. The gradual disappearance of the C-terminal 16-amino acid fragment was confirmed by electrospray ionization mass spectrometry (Figure S1) and nuclear magnetic resonance (NMR) backbone assignment indicated that the predominant C-terminal residue was A48 (Figure 1B). We predicted that this shorter form of CsfB (CsfB1−48) was nonfunctional, given the absence of residues required for σG inhibition (Rhayat et al., 2009). To confirm this, we assessed the ability of CsfB1−48 to inhibit σG or σE when the proteins were co-expressed during vegetative growth of B. subtilis, an approach that has been used previously (Karmazyn-Campelli et al., 2008, Rhayat et al., 2009). Whereas wild-type CsfB inhibited >99% of σG activity and ∼77% of σE activity, the CsfB1−48 variant displayed no inhibition of either sigma factor (Figure S2).

Isolation of a Functional, Full-Length CsfB Protein In Vitro

Since CsfB1−48 was unable to inhibit σG and σE, we adopted several approaches to obtain a full-length, stable version of CsfB. Initially, we produced a C-terminally histidine-tagged version of CsfB, which was slower to degrade but still consistently converted to the CsfB1−48 species (Figure S3). We next rationally designed a panel of CsfB variants (Table S1) to identify a functional version of CsfB that remained full-length. Of these, A48E (altered at the known cleavage point) proved the most successful, yielding a stable full-length version of CsfB that remained intact for 4 days as confirmed by mass spectrometry (Figure S1). CsfBA48E inhibited both σG and σE to the same extent as wild-type CsfB in vivo (Figure S2), suggesting that the A48E substitution does not alter protein function. Satisfyingly, the NMR HSQC spectrum of CsfBA48E overlaid precisely with that of CsfB1−48 (truncated wild-type), except for the presence of peaks corresponding to the additional C-terminal residues (Figure S4). Some of these additional peaks could be assigned from triple-resonance experiments and, upon revisiting earlier HSQC spectra of freshly purified wild-type CsfB, a low population of these same peaks was visible from the residual full-length protein that had not yet degraded (Figure 1B). Several peaks within the C-terminal region could not be assigned due to a line-broadening effect (Figure S4B). The new C-terminal peaks, whether assignable or not, displayed little dispersion in the proton dimension, a hallmark of low structural complexity.

Interaction of CsfBA48E with σG and σE

With the functional, full-length CsfBA48E protein in hand, we first analyzed its interactions with its target sigma factors. To this end, we produced recombinant full-length σG (residues 1–260) and a truncated version of σE (residues 17–239) lacking the N-terminal membrane-anchored pro-sequence (Peters et al., 1992). We then carried out NMR chemical shift perturbation (CSP) analysis between unlabeled σG or σE and 15N-labeled CsfBA48E. Titration of unlabeled σG caused the majority of CsfBA48E backbone amide signals to gradually disappear (Figure 2A). This result indicates an interaction between CsfBA48E and σG, although the disappearance of most peaks prevented identification of specific positions on CsfBA48E that mediate contact. As a control, we performed CSP analysis between unlabeled σG and 15N-labeled CsfB1−48, the truncated variant incapable of inhibiting σG in vivo. Consistent with the inability of these proteins to interact, no changes to the CsfB1−48 backbone amide signals were observed.

Figure 2

Interactions of CsfBA48E with σG and σE

(A and B) Overlay of 1H-15N SOFAST HMQC spectra of 15N-labeled CsfBA48E alone (blue), and in presence of 2-fold molar excess of (A) σG (red) or (B) σE (purple). Extra peaks appearing upon titration with σE are highlighted by a green square.

(C and D) ITC thermograms of interaction between CsfBA48E and (C) σG or (D) σE. Raw data (upper panels), binding isotherm (lower panels). Fitted data for CsfBA48E-σE interaction: ΔH = −8.04 ± 0.04 kcal/mol; ΔS = 9.19 ± 0.50 cal/(mol·K); N = 1.01 ± 0.00 sites.

When 15N-labeled CsfBA48E was titrated with unlabeled σE, many CsfBA48E backbone amide signals decreased in intensity and shifted position significantly (Figure 2B), indicating a tight interaction in the nanomolar to low micromolar affinity range. As a result of the slow timescale, it was not possible to reliably assign the peaks in their new positions and, unfortunately, the resulting complex was too large for the triple-resonance experiments required to assign the bound state. Hence, we could not assess the relative contributions of each of the bound residues. However, we noted that addition of σE caused several of the peaks corresponding to CsfBA48E residues 49–64 to shift to the 1H upfield region of the spectrum (Figure 2B), suggesting that the C-terminal region becomes more structured upon interaction with σE. As expected, σE caused no shifts in the spectrum of the 15N-labeled CsfB1−48 truncated variant.

We next carried out isothermal titration calorimetry (ITC) to quantify the interaction between CsfBA48E and its two cognate sigma factors. The CsfBA48E-σE interaction was determined to have a Kd of 12.5 ± 2.4  nM (Figure 2D) and 1:1 stoichiometry. CsfBA48E and σG also showed clear evidence of an interaction (Figure 2C), although a Kd and stoichiometry could not be determined, possibly due to instability or aggregation of our recombinant σG. However, by comparing the ITC data to σE experiments, we can conclude that the CsfBA48E-σG binding affinity is likely within the same order of magnitude. As a control, we verified that no interaction was observed between the truncated variant CsfB1−48 and σG or σE under the same conditions.

NMR Solution Structure of CsfB

Next, we sought to solve the solution structure of CsfB. Despite having isolated a functional, full-length CsfB variant (CsfBA48E), we could only obtain high-quality NMR triple-resonance signals for residues comprising the originally purified, truncated CsfB1−48 variant. Given that the C-terminal 16 residues presented low structural complexity, and the folding of the rest of the protein was conserved, we opted to complete the full NMR assignments (BMRB: 34102) and solve the solution structure for CsfB residues 1–48. It forms a tight symmetrical homodimer (Figures 3A and 3B; PDB: 5N7Y; structural statistics in Table 1), where each monomer consists of a treble-clef zinc finger motif (Grishin, 2001).

Figure 3

NMR Solution Structure of the CsfB1−48 Dimer and Functionality of Dimerization-Deficient CsfB Variants

(A) Orthogonal views of ensemble backbone (left) and cartoon (right) representations for the 20 lowest energy ARIA-calculated structures as deposited in the PDB (PDB: 5N7Y).

(B) Detailed view of the dimer interface; hydrophobic buried residues are depicted using ball and stick representation.

(C) Detailed view of the zinc finger coordination shell showing the cysteine residues and the Sγ(i)-HN(i+2) hydrogen bonds (green dashed lines) in the first and second spheres of coordination.

(D and E) CsfB variants lacking putative dimerization residues V37 and/or I38 are deficient for sigma factor inhibition in vivo. Vegetatively growing B. subtilis cells were induced with IPTG to express (D) σG or (E) σE alone or in combination with wild-type or variant CsfB. Sigma factor activity was monitored by light production (measured in relative light units [RLU]) from σG- or σE-dependent luciferase reporter genes (P-lux or P-lux, respectively). Control strains lacking inducible constructs (“Reporter alone”) are shown for comparison in each graph. Error bars indicate SD. Strains used in this assay are listed in Table S5.

Table 1

NMR and Refinement Statistics for the Final 20 Ensemble Structures of CsfB

NMR Distance and Dihedral Constraints (per Monomer)
Distance constraints
Total unambiguous constraints	1,154
Intra-residue	417
Sequential (|i-j| = 1)	231
Medium-range (1 < |i-j| < 4)	118
Long-range (|i-j| > 5)	201
Intermolecular	187
Ambiguous constraints	116
TALOS-derived dihedral constraints
Total dihedral constraints (Φ+Ψ)	64
Structure Statistics
Violations per structure (mean and SD)
Number of violated distance restraints (>0.25 Å)	0.65 ± 0.63
Max. distance constraint violation	0.28 Å
Number of violated dihedral angle restraints (>5°)	2.3 ± 0.9
Max. dihedral angle violation	9.2°
Ramachandran plot analysisa	Residues 8–40
Residues in most favored regions	89.5% ± 4.0%
Residues in additionally allowed regions	10.5% ± 4.0%
Residues in disallowed regions	0.0% ± 0.0%
Derivation from idealized geometry
Bond length (Å)	0.0079 ± 0.0005
Bond angles (°)	0.83 ± 0.05
Average RMSD to mean structure (range 8–40)
Backbone	0.5 ± 0.2
Heavy	1.1 ± 0.2

Obtained from PDB NMR structure validation report.

The folded domain (residues 8–40) of each CsfB monomer contains two β hairpins separated by a short turn, followed by a C-terminal α helix. The structure clearly indicates that each monomer binds a Zn2+ ion; we confirmed a 1:1 Zn2+:CsfB ratio by ICP-MS. Zinc binding by CsfB involves the coordination of two cysteine residues from the first β-hairpin knuckle (C11 and C14) and two additional cysteines from the first turn of the α helix (C30 and C33) in a tetrahedral conformation (Figure 3C), a classic treble-clef zinc finger fold (Krishna et al., 2003, Kaur and Subramanian, 2016). The chemical shift values for 13Cα (∼59 ppm) and 13Cβ (∼31 ppm) are consistent with zinc-binding character (Kornhaber et al., 2006). The second coordination shell is defined by the formation of two hydrogen bonds between the cysteine sulfur atoms (C11 and C30) and the amide group of the residue at position +2 (I13 and D32, respectively). For all zinc coordination parameters, see Table S2.

The CsfB homodimer interface spans 1,138 Å2 (calculated by PISA; Krissinel and Henrick, 2007) and involves numerous intermolecular contacts between the β hairpins and α helices of each monomer (Figure 3B). Several nonpolar residues (V12, I13, I22, L24, I26, V37, I38) are embedded in the dimer interface, creating a hydrophobic core resembling that of a globular protein. In contrast, the surface of the protein displays hydrophilic side chains that create an intricate network of polar contacts. For example, the ɛ-amino group of K36 from one monomer is surrounded by the carboxylate side chain of D32 from the same chain and the hydroxyl group of Y25 from the other monomer. In addition, the side chains of K27 from one monomer and D15 from the other chain, as well as those from S41 and E34, form clear polar contacts.

Finally, we tested the effect of disrupting CsfB dimer formation in B. subtilis by constructing CsfB variants with substitutions at V37 and/or I38. These two residues in the α helix of one CsfB monomer pack against the same two residues in the α helix of the second monomer; these two positions are almost always occupied by hydrophobic residues in CsfB homologs (Rhayat et al., 2009, Camp et al., 2011). We found that the individual alanine substitutions (V37A and I38A) had only modest effects on CsfB-mediated inhibition of σG or σE in the vegetative co-induction assay (Figures 3D and 3E). In contrast, substitution of these residues with glutamate (V37E and I38E) significantly reduced inhibition of σG from nearly 100% to 0% and 15%, respectively. CsfB-mediated inhibition of σE was also significantly compromised by the V37E substitution (reduced from 77% to 9%), while the I38E had a more modest effect on inhibition (reducing it from 77% to 49%; Figures 3D and 3E). Lastly, we found that simultaneous substitution of these positions for alanine (V37A, I38A), also significantly diminished CsfB-dependent σG and σE inhibition, to only 17% and 32%, respectively. These findings imply that dimer formation by CsfB is required for inhibition of σG and σE.

Discussion

Here, we have solved the solution structure of the folded domain of the anti-sigma factor CsfB, which inhibits two class III sigma factors, σG and σE, during B. subtilis sporulation. The two conserved C-X-X-C motifs of CsfB suggested early on that it was likely to bind Zn2+ (Karmazyn-Campelli et al., 2008), a prediction that was verified biochemically in two studies, albeit with different Zn2+:CsfB ratios reported (Rhayat et al., 2009, Serrano et al., 2011). Genetic analyses further hinted that CsfB might function as a dimer (Rhayat et al., 2009). Our solved structure verifies that CsfB is a symmetric homodimer, with each monomer adopting a treble-clef zinc finger fold coordinating a single Zn2+ ion. We confirmed a 1:1 Zn2+:CsfB ratio by ICP-MS, in line not only with our structure but also with the 1:1 Zn2+:CsfB ratio reported by Serrano et al. (2011). Our CsfB structure also offers an explanation for the finding by Rhayat et al. (2009) that mutating the highly conserved glycine at position 21 to cysteine abolished CsfB function. The two alpha protons in this glycine point snugly into the hydrophobic core of the protein such that any other side chain at this position would likely cause a steric hindrance.

Our CsfB structure is inconsistent, however, with a model proposed by Rhayat et al. (2009) in which CsfB forms an asymmetric dimer that coordinates a single Zn2+ ion between different cysteine pairs on alternative monomers. This model was a sensible interpretation of data from a series of cysteine deletion mutants co-expressed in vivo, as well as their measurement of a 0.5:1 Zn2+:CsfB ratio. Given that our NMR data clearly show the presence of a symmetric dimer (one subset of signals), we suspect that the Zn2+:CsfB ratio reported by these authors may be an artifact of their maltose-binding protein-CsfB fusion, the functionality of which was not reported. Alternatively, it may be that purification of CsfB from sporulating B. subtilis accounts for the altered zinc content in the Rhayat et al. (2009) study; our CsfB protein and that of Serrano et al. (2011) were purified from Escherichia coli. In this case, it would be tempting to speculate that CsfB is subject to Zn2+-dependent regulation during sporulation of B. subtilis, as has been suggested previously (Karmazyn-Campelli et al., 2008).

The CsfB and ClpX N-Terminal Domain Treble-Clef Dimers Are Structurally Similar

We have found here that CsfB adopts a treble-clef zinc finger fold, one of the most common zinc finger arrangements (Kaur and Subramanian, 2016, Krishna et al., 2003). Interestingly, CsfB appears most similar to the N-terminal domain (NTD) of the E. coli AAA + ATPase ClpX (PDB: 2DS5–2DS8) (Park et al., 2007); to our knowledge, CsfB and the ClpX NTD are the only examples of dimerization found in the treble-clef zinc finger fold family (as identified by the Dali server [Holm and Laakso, 2016, Holm and Rosenstrom, 2010]). ClpX is the chaperone/unfoldase component of the ClpXP protease, a barrel-shaped proteolytic machine that degrades target proteins for quality control as well as regulation (Baker and Sauer, 2012). It is proposed that the dimerized ClpX NTDs interact with substrates or cofactors of ClpXP and guide them toward the protease complex for degradation (Thibault et al., 2006).

CsfB and the ClpX NTD (PDB: 2DS6) structurally align with a root-mean-square deviation (RMSD) of 4.4 Å (backbone alignment between CsfB 8–43 and CplX 11–49 residues over 196 atoms) and share a sequence identity of ∼20% (Figures 4A and S5). Both homodimerize via a hydrophobic core of residues derived from their C-terminal α helices and two β hairpins. The proteins have a similar area of dimer interface at 945 Å2 for the ClpX NTD (Donaldson et al., 2003) and 1,138 Å2 for CsfB. Although both dimers are held together with a combination of hydrophobic and electrostatic interactions, the pattern of these interactions is disparate (Figures 4B and 4C) casting doubt on their evolutionary relatedness. This, together with poor conservation of solvent-exposed residues, may indicate that the similarity between the CsfB and ClpX NTD dimers is limited to their structural folds, and does not extend to binding partners and/or function.

Figure 4

Structural and Sequence Comparison of CsfB with ClpX_NTD

(A) Sequence alignment of CsfB and ClpX_NTD from different species. Cartoons above the sequences represent the positions involved in secondary structure formation in CsfB.

(B and C) Structural comparison of dimer interfaces in CsfB (orange and blue, top) and ClpX_NTD (PDB: 2DS6, light yellow and cyan, bottom). Zinc cations shown as gray spheres. Residues involved in (B) antiparallel helices packaging and (C) loop contacts are shown for each.

CsfB as an Anti-sigma Factor Structure

To our knowledge, CsfB is only the third anti-sigma factor for class III alternative sigma factors to be structurally analyzed, with the other two being the Bacillus stereothermophilus σF sporulation anti-sigma factor SpoIIAB (Campbell et al., 2002, Masuda et al., 2004) and the Aquifex aeolicus σ28 flagellum biosynthesis anti-sigma factor FlgM (Sorenson et al., 2004). CsfB is unrelated to both of these proteins, although SpoIIAB also homodimerizes (Campbell and Darst, 2000, Campbell et al., 2002). Aside from the ability to coordinate Zn2+, CsfB also displays no structural similarity to the zinc anti-sigma (ZAS) family of anti-sigma factors that inhibit class IV ECF alternative sigma factors (Campbell et al., 2007, Sineva et al., 2017). Lastly, we had previously noted that CsfB resembles the B. subtilis σF inhibitor Fin at the primary amino acid sequence level (Camp et al., 2011). However, comparison of the CsfB structure with our recently reported Fin structure reveals that, although both proteins bind zinc, they fold into completely different motifs (Wang Erickson et al., 2017).

Sigma Factor Inhibition by CsfB

Our ultimate objective is a complete understanding of the mechanism by which CsfB inhibits σG and σE during B. subtilis sporulation, and the regulation thereof. This study made significant headway toward this goal. We have shown that CsfB forms a tight homodimer that binds tightly to its two target sigma factors. In the case of the CsfB-σE interaction, our ITC analysis indicated a stoichiometry of 1:1, suggesting that a CsfB dimer simultaneously binds two σE molecules. We find the alternative scenario, in which the CsfB dimer dissociates upon binding σE, to be unlikely given that the CsfB HSQC spectrum does not drastically reconfigure upon addition of σE. Whether σG is bound in a similar manner remains an open question, given that we could neither calculate a stoichiometry from our ITC data nor observe the CsfB bound state by NMR.

Our data do not yet allow us to draw precise conclusions regarding the interface of CsfB that mediates contact with σE or σG. That said, it is evident that the C-terminal region of CsfB is required for interaction with both sigma factors in vitro and their inhibition in vivo. Our CSP analysis further indicates that this C-terminal region of CsfB becomes more structured upon interaction with σE. We therefore speculate that, at least for σE, two sigma factors are bound, one apiece, to the two C-terminal “tails” of a CsfB dimer. Interestingly, evidence in the literature suggests that the interaction of CsfB with σG is likely to be dissimilar, at least in its detail. CsfB binds to σG at region 2.1 while it binds to σE at regions 2.2–2.3. Moreover, specific amino acids that help CsfB discriminate between σF and σG play no role in discriminating between σE and σK, and vice versa (Serrano et al., 2011, Serrano et al., 2015). As such, comparing and contrasting the structural basis for σG and σE inhibition by CsfB is an exciting challenge for future work.

Last but not least, it is tempting to speculate that degradation/cleavage of CsfB, which posed a significant challenge in this study, may be physiologically relevant in B. subtilis. For example, cleavage/degradation of CsfB may provide a mechanism by which σG ultimately escapes CsfB inhibition at late times in the forespore.

STAR★Methods

Key Resources Table

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Rivka Isaacson (rivka.isaacson@kcl.ac.uk).

Experimental Model and Subject Details

In vivo functionality tests of CsfB variants were performed in B. subtilis strains isogenic with the laboratory strain PY79 (Youngman et al., 1984). For general propagation, B. subtilis strains were grown at 37°C in liquid LB media or on LB agar plates.

Method Details

Plasmid Construction

Plasmids used in this study are listed in the Key Resources Table and Table S3. The sequences of oligonucleotides used in plasmid construction are given in Table S4. Chromosomal DNA from B. subtilis strain PY79 served as a template for polymerase chain reaction (PCR), unless otherwise noted. Sequences of synthetic gene fragments (gBlocks, Integrated DNA Technologies) used in plasmid construction are available upon request. Plasmids were constructed using traditional cloning techniques, site-directed mutagenesis, ligation-independent cloning (Aslanidis and de Jong, 1990), or isothermal assembly (Gibson, 2009), as indicated. Plasmids were propagated in the E. coli strain NEB 5-alpha grown in the presence of the antibiotics ampicillin (100 μg/mL) or kanamycin (50 μg/mL), when appropriate. Site-directed mutagenesis was performed either with the Q5 Site-Directed Mutagenesis Kit or QuikChange Mutagenesis Kit. Isothermal assembly was performed with either the Gibson Assembly Master Mix or the NEBuilder HiFi DNA Assembly Master Mix. All plasmids were verified by DNA sequencing. Construction details for plasmids not previously published are given below.

To generate plasmids for expression and purification of recombinant CsfB from E. coli, the csfB coding sequence was amplified by PCR using primers listed in Table S4 and inserted by ligation independent cloning into pET-46 (N-terminal hexahistidine tag), pNIC28 (TEV protease-cleavable N-terminal hexahistidine tag), and pLATE31 (C-terminal hexahistidine tag), respectively. Plasmids encoding CsfBA48E and other CsfB variants were obtained by site-directed mutagenesis of the pNIC28-CsfB plasmid using the Q5 Site-Directed Mutagenesis Kit.

The plasmid for expression and purification of recombinant σG from E. coli was generated by ligation independent cloning of a PCR product harboring the full-length sigG coding sequence into pET-46 (N-terminal hexahistidine tag). The plasmid for expression and purification of σE (lacking its N-terminal membrane-anchored pro-sequence) from E. coli was constructed by ligating a BamHI/XhoI-digested PCR product harboring sigE codons 17-239 into BamHI/XhoI-digested pET28-TxrA (gift of Dr J.M. Pérez-Cañadillas, Rocasolano Physical Chemistry Institute, Spain), an E. coli expression plasmid containing a TEV protease cleavable N-terminal hexahistidine and thioredoxin tags.

To construct the plasmid (pJJ46) for induction of σG expression in B. subtilis, the sigG 5’ leader (including the native sigG ribosome binding site [RBS]) and coding sequence was amplified by PCR with primers JJ30 and JJ31 and assembled into SalI/NheI-digested pDR110 (gift of David Rudner, Harvard Medical School). The plasmid (pSFO1) for induction of σE (lacking its N-terminal membrane-anchored pro-sequence) in B. subtilis was constructed by assembling a synthetic gene fragment harboring sigE codons 17-239 preceded by an optimized RBS into SalI/SphI-digested pDR111 (gift of David Rudner, Harvard Medical School). The plasmid (pAH88) for IPTG-induction of CsfB in B. subtilis was generated in two steps. First, the csfB 5’ leader (including the native RBS) and coding sequence were amplified using primers AH41 and AH42, digested with HindIII and NheI, and cloned into HindIII/NheI-digested pDR111 (gift of David Rudner, Harvard Medical School), yielding the intermediate plasmid pAH84. The EcoRI/BamHI fragment containing P-csfB and lacI was then subcloned into the respective sites of pDG1664 (Guérout-Fleury et al., 1996) to generate pAH88. The derivatives of pAH88 encoding CsfB variants A48E, V37A, V37E, I38A, I38E, V37A/I38A, or the CsfB1-48 truncation (pKF70, pKF87-pKF91, and pTK2, respectively) were constructed by individually assembling synthetic gene fragments into the HindIII/SphI-digested pAH88 backbone.

Plasmids encoding luciferase reporter genes were constructed using the plasmid pAH328, which harbors the Photorhabdus luminescens bacterial luciferase operon luxABCDE optimized for expression in B. subtilis and preceded by a multiple cloning site (MCS). This plasmid was constructed from pAH321 (Schmalisch et al., 2010) in two steps. First, the BamHI site upstream of the luxE coding sequence in pAH321 was mutated by site-directed mutagenesis using the QuikChange Mutagenesis Kit and primers AH312 and AH313, yielding pAH325. Second, a DNA fragment harboring an EcoRI-SacI-NotI-SpeI-SalI MCS, generated by annealing oligonucleotides AH310 and AH311, was ligated into the EcoRI/SalI-digested backbone of pAH325, yielding pAH328. The σE- and σG-dependent luciferase reporter plasmids (pAH334 and pAH336, respectively) were constructed by ligating EcoRI/SalI-digested PCR products containing either the σE-dependent spoIID promoter (amplified with AH58 and AH59) or the σG-dependent sspB promoter (amplified with AH60 and AH61) into EcoRI/SalI-digested pAH328.

B. subtilis Strain Construction

The full genotypes of B. subtilis strains used in this study, all of which were derived from the wild type laboratory strain PY79 (Youngman et al., 1984), are listed in the Key Resources Table and Table S5. Strains were constructed by transformation of competent cells, prepared as previously described (Wilson and Bott, 1968), with B. subtilis chromosomal DNA, plasmid DNA, or PCR-amplified DNA. Transformants were selected on media with antibiotics, when appropriate, as follows: chloramphenicol (5 μg/mL), erythromycin plus lincomycin (1 μg/mL and 25 μg/mL, respectively), spectinomycin (100 μg/mL), kanamycin (5 μg/mL), phleomycin (0.4 μg/mL), and tetracycline (10 μg/mL). Insertions into amyE or thrC were confirmed by loss of α-amylase activity on LB agar plates with starch or the failure to grow on minimal media, respectively.

The ΔcsfB::tet (AHB199) and ΔsigG::kan (AHB98) deletions have been described (Camp and Losick, 2008). The ΔsigE::[erm]::phleo deletion was built for this study in two steps. First, a ΔsigE::erm deletion strain (AHB201) was constructed by the long-flanking homology PCR (LFH-PCR) method (Wach, 1996). Primers sets AH43/AH44 and AH45/AH46 were used to amplify sequences flanking sigE, which were then used to amplify the erythromycin resistance cassette (erm) from plasmid pAH52 (Ferguson et al., 2007). Proper integration of the resulting ΔsigE::erm LFH-PCR product was confirmed by PCR. To switch the antibiotic resistance of the ΔsigE deletion, AHB201 was transformed with pEr::Pm (Steinmetz and Richter, 1994), resulting in the erythromycin-sensitive, phleomycin-resistant strain SFB31 (ΔsigE::[erm]::phleo).

All other constructs (IPTG-inducible sigG, sigE, and csfB, as well as the σE- and σG-dependent luciferase reporters) were introduced into B. subtilis strains using plasmids constructed for this study.

Recombinant Protein Production

The plasmids encoding CsfB, σG or σE (see the Key Resources Table and Table S3) were transformed into the BL21(DE3)pLysS E. coli strain. Cells were grown either in LB or minimal media supplemented with 0.7 g/l 15N-NH4Cl, 2 g/l 13C-glucose and 1 g/l 13C-15N-Isogro. Protein expression was induced with 0.5 mM IPTG at OD600 = 0.8 and conducted either at 37°C for 4 hours or at 18°C overnight. ZnCl2 at a final concentration of 10 μM was added to minimal media cell culture before IPTG induction for production of CsfB.

The cell pellet was resuspended in 50 mM HEPES pH 7.5, 300 mM NaCl, 5 mM Imidazole, 5% Glycerol, 1 mM DTT, 1 mg/ml lysozyme, 10 μg/ml Dnase I, 5 mM MgCl2, 3x EDTA-free Complete Protease Inhibitor and 2 mM phenylmethylsulfonyl fluoride (PMSF), then lysed by sonication. Recombinant protein was purified from the soluble fraction of the cell lysate by affinity chromatography using a ready-to-use HisTrap FF crude column pre-packed with Ni Sepharose resin. When required, the N-terminal His-tag was removed by overnight incubation at 4°C with TEV protease at a molar ratio protein:TEV of 40:1. The digested protein was then separated from undigested protein and TEV protease using HisTrap FF crude column pre-packed with Ni Sepharose resin. Purified fractions were subjected to a final step of size exclusion chromatography using a Superdex75 16/600 PG column equilibrated with 50 mM HEPES pH 7.5, 150 mM KCl and 0.5 mM TCEP buffer.

The purity and stability of the proteins were checked by SDS-PAGE and mass spectrometry and the presence and stoichiometry of zinc in CsfB was determined by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) using a PerkinElmer NexION 350D spectrometer.

NMR Spectroscopy

Uniformly 15N, 13C-labelled NMR sample was buffer-exchanged into 50 mM HEPES pH 7.5, 150 mM KCl, 0.5 mM TCEP using a HiLoad 16/600 Superdex 75 pg gel filtration column. NMR experiments were carried out on samples >500 μM at 303 K and recorded on Bruker AVANCE spectrometers operating at 500 MHz, 700 MHz and 950 MHz with TXI cryoprobes controlled by Topspin 3 (Bruker Biospin Ltd). Spectra were processed using NMRPipe/NMRDraw (Delaglio et al., 1995) and analyzed using CcpNMR Analysis 2.2 (Vranken et al., 2005). Backbone resonances were assigned in a standard manner using [1H,15N]-HSQC, HNCA, HNCACB, CBCA(CO)NH, and HNCO experiments (Grzesiek and Bax, 1993). Side-chains resonances assignment was performed using a combination of HCCH-TOCSY (Kay et al., 1993) and HBHA(CO)NH (Grzesiek and Bax, 1993). NOE distance restraints and assignments of aromatics rings were obtained from 15N-edited NOESY-HSQC and 13C-edited NOESY-HSQC spectra with a 120 ms mixing time. An additional set of intermolecular distance restraints was obtained from a 12C-filtered, 13C-edited NOESY-HSQC spectrum (Zwahlen et al., 1997) using a mixed CsfB dimer prepared by mixing 15N, 13C-labelled CsfB and unlabelled CsfB in an equimolar ratio. To allow the exchange of the monomeric subunits, the mixture was heated at 50°C for 10 minutes and then cooled slowly.

Structure Calculation

The solution structure of the CsfB1-48 dimer was solved using ARIA2.3 (Rieping et al., 2007), utilizing distance restraints derived from the four NOESY spectra (NOEs from the filtered NOESY experiment were defined as intermolecular while the NOEs in the other NOESY experiments were treated as ambiguous) and dihedral angle restraints estimated by TALOS+ (Shen et al., 2009). Typical annealing parameters were used for distance (10, 15, 50 and 100 Kcal/mol for high temperature, initial cool1, final cool1 and cool2 force constants) and dihedral restraints (50, 150 and 200 Kcal/mol for high temperature, cool1 and cool2 force constants) and a C2 symmetry was imposed with a non-crystallographic symmetry restraints force constant value of 100 Kcal/mol and packing force constants of 15, 10 and 5 Kcal/mol during high temperature, cool1 and cool2 steps.

In the first rounds of calculation, zinc coordination information was not included and, only after checking that the putative involved cysteine residues appeared at the correct disposition for tetrahedral coordination, were the appropriate restrictions for the zinc fingers added (using ARIA2.3 tools). Twenty structures with the lowest energy values were selected out of 200 and subjected to a water refinement process. The final ensemble of the structure (PDB: 5N7Y) was analyzed and represented using MOLMOL (Koradi et al., 1996) and PyMOL.

NMR Titrations

Chemical shift perturbation assays were carried out at 298K using a Bruker AVANCE spectrometer operating at 950 MHz with a TXI cryoprobe controlled by Topspin 3 (Bruker Biospin Ltd). Spectra were processed using NMRPipe/NMRDraw (Delaglio et al., 1995) and analyzed using CcpNMR Analysis 2.2 (Vranken et al., 2005). 100 μM 15N-labelled CsfBA48E in 50 mM HEPES pH 7.5, 150 mM KCl, 0.5 mM TCEP was titrated with unlabelled σG or σE up to a ratio of 1:2 molar equivalents. 1H–15N SOFAST-HMQC spectra were recorded at each titration point.

ITC

Binding of CsfBA48E to σG or σE was measured by ITC using an ITC200 instrument (Microcal Inc. Malvern). Samples were dialyzed into 50 mM HEPES pH 7.5, 150 mM KCl, 0.5 mM TCEP. Titrations were carried out at 25°C using 19 injections of 2 μl with a delay of 180s between injections. For the CsfBA48E interaction with σG the sample cell contained 110 μM σG and the syringe 1.6 mM CsfBA48E. For the CsfBA48E interaction with σE the sample cell contained 60 μM CsfBA48E and the syringe 395 μM σE.

In Vivo σG and σE Inhibition Assay

To measure CsfB-mediated σG and σE inhibition in vivo, B. subtilis strains harboring σG- or σE-dependent luciferase reporter genes were engineered to induce expression of the corresponding sigma factors either alone or in combination with wild type or mutant CsfB. Equal amounts of vegetatively growing cells (1 OD600⋅mL) were collected and concentrated 5-fold. 30 μl of these cells were applied onto 200 μl LB agar pads containing 100 μM IPTG (for σG activity assays) or 10 μM IPTG (for σE activity assays) in white 96-well plates. Bioluminescence from each well was measured at 37°C every 15 min for 6 hours using a Synergy H1M plate reader (BioTek Instruments). Data is reported as the average of at least two (typically three or more) different experiments, with 2-4 technical replicates performed per experiment. CsfB inhibition was calculated as the percentage reduction in σG or σE activity (with background reporter activity subtracted) after 4 or 3 hours of induction, respectively, relative to the total σG or σE activity (also with background reporter activity subtracted) in a strain lacking inducible csfB.

Quantification and Statistical Analysis

ITC Analysis

The obtained data from ITC titrations were analyzed using MicroCal Origin 7 software. Areas under the peaks were integrated and fitted by least-square procedures assuming a 1:1 stoichiometry.

Data obtained from the in vivo σG and σE inhibition assays was collected using the Gen5 Microplate Reader and Imager Software (BioTek Instruments) and subsequently analyzed using Excel (Microsoft Corporation) and Prism (Graphpad Software). Variation in the data was determined by calculating the standard deviation across separate experiments.

Data and Software Availability

The coordinates of the final ensemble of CsfB structure are deposited at the Protein Data Bank Europe (https://www.ebi.ac.uk/pdbe/) under the accession code 5N7Y. The assigned chemical shifts of the protein are also deposited at the Biological Magnetic Resonance Bank (http://www.bmrb.wisc.edu/) under the accession number 34102.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and Virus Strains
E. coli BL21(DE3)pLysS	New England Biolabs	Cat# C2527I
E. coli NEB5-alpha	New England Biolabs	Cat# C2987I
Chemicals, Peptides, and Recombinant Proteins
Ampicillin	Melford Laboratories, Sigma-Aldrich	Cat# A0104, Cat# A9518
Kanamycin	Melford Laboratories, Sigma-Aldrich	Cat# K0126, Cat# K4000
Chloramphenicol	Sigma Aldrich	Cat# C0378
Erythromycin	Sigma-Aldrich	Cat# E6376
Lincomycin	VWR (Alfa Aesar)	Cat #AAJ61251
Spectinomycin	Sigma-Aldrich	Cat# S9007
Phleomycin	Research Products International	Cat# P20200
Tetracycline	VWR (Alfa Aesar)	Cat# AAB21408
LB Broth	Research Products International	Cat# L24061, Cat# L24065
LB Broth High Salt	Melford Laboratories	Cat# L1704
Agar, Bacteriological Grade	VWR (Hardy Diagnostics)	Cat# 89405-068
IPTG	Sigma-Aldrich, Research Products International	Cat# I6758, Cat# AAJ61251
ZnCl2	Sigma-Aldrich	Cat# 229997
15N-NH4Cl	Sigma-Aldrich	Cat# 299251
13C-glucose	Sigma-Aldrich	Cat# 389374
13C-15N-Isogro	Sigma-Aldrich	Cat# 606839
Deuterium Oxide	Sigma-Aldrich	Cat# 151882
HEPES	Melford Laboratories	Cat# B2001
NaCl	Melford Laboratories	Cat# S0520
MgCl2	Melford Laboratories	Cat# M0535
KCl	Melford Laboratories	Cat# P0515
Imidazole	Melford Laboratories	Cat# B4005
cOmplete mini EDTA-free protease inhibitor tablets	Roche	Cat# 11836170001
TCEP	Alfa Aesar	Cat# J60316.09
DTT	Sigma-Aldrich	Cat# D0632
PMSF	Sigma-Aldrich	Cat# P7626
Glycerol	VWR	Cat# 24388.295
DNAse I grade II	Roche	Cat# 10104159001
Lysozyme	Sigma Aldrich	Cat# L6876
Critical Commercial Assays
Q5 Site-Directed Mutagenesis Kit	New England Biolabs	Cat# E0554S
QuikChange Mutagenesis Kit	Agilent Technologies	Cat# 200517
Gibson Assembly Master Mix	New England Biolabs	Cat# E2611S
NEBuilder HiFi DNA Assembly Master Mix	New England Biolabs	Cat# E2621S
Deposited Data
Solution structure of CsfB 1-48	This study	PDB: 5N7Y
Chemical shift assignment of CsfB 1-48	This study	BMRB: 34102
Experimental Models: Organisms/Strains
B. subtilis: Parent strain PY79	Youngman et al., 1984	N/A
B. subtilis: Strain AHB98 (ΔsigG::kan)	Camp and Losick, 2008	N/A
B. subtilis: Strain AHB199 (ΔcsfB::tet)	Camp and Losick, 2008	N/A
B. subtilis: Strain AHB201 (ΔsigE::erm)	This study	N/A
B. subtilis: Strain SFB31 (ΔsigE::[erm]::phleo)	This study	N/A
B. subtilis: Strains used for in vivo σG and σE inhibition assays, see Table S5	This study	N/A
Oligonucleotides
Primers used for plasmid construction, see Table S4	Integrated DNA Technologies	N/A
Synthetic gene fragments used for plasmid construction	Integrated DNA Technologies	N/A
Recombinant DNA
Plasmid: pET-46	Novagen	Cat# 71335-3
Plasmid: pNIC28	Structural Genomics Consortium	Cat# 26103
Plasmid: pLATE31	Thermo Scientific	Cat# K1261
Plasmid: pET28_TxrA	José Manuel Pérez Cañadillas	N/A
Plasmid: pDR110	David Rudner	N/A
Plasmid: pDR111	David Rudner	N/A
Plasmid: pDG1664	Guérout-Fleury et al., 1996	N/A
Plasmid: pAH321	Schmalisch et al., 2010	N/A
Plasmid: pAH328	This study	N/A
Plasmid: pEr::Pm	Steinmetz and Richter, 1994	N/A
Plasmids constructed for expression of CsfB, σG, or σE in E. coli or B. subtilis, see Table S3	This study	N/A
Plasmids harboring σG- or σE-dependent luciferase reporters, see Table S3	This study	N/A
Software and Algorithms
Topspin 3	Bruker Biospin	https://www.bruker.com/service/support-upgrades/software-downloads/nmr.html
NMRPipe/NMRDraw	Delaglio et al., 1995	http://www.nmrpipe.com/
CcpNMR Analysis 2.2	Vranken et al., 2005	http://www.ccpn.ac.uk/v2-software/software/analysis
ARIA2.3	Rieping et al., 2007	http://aria.pasteur.fr/downloads
TALOS+	Shen et al., 2009	https://spin.niddk.nih.gov/bax/software/TALOS/
MOLMOL	Koradi et al., 1996	http://www.msg.ucsf.edu/local/programs/molmol/manual.html
PyMOL	DeLano Scientific LLC	http://www.pymol.org
MicroCal Origin 7	OriginLab	https://www.originlab.com/
Gen5 Microplate Reader and Imager Software	BioTek Instruments	https://www.biotek.com/
Excel	Microsoft Corporation	https://office.microsoft.com/excel/
Prism	GraphPad Software	https://www.graphpad.com/
Other
HisTrap FF crude column pre-packed with Ni Sepharose resin	GE Healthcare Life Sciences	Cat# 17-5286-01
Superdex75 16/60 PG column	GE Healthcare Life Sciences	Cat# 17-1068-01