|Home | About | Journals | Submit | Contact Us | Français|
Sigma (σ) factors direct gene transcription by binding to and determining the promoter recognition specificity of RNA polymerase (RNAP) in bacteria. Genes transcribed under the control of alternative sigma factors allow cells to respond to stress and undergo developmental processes, such as sporulation in Bacillus subtilis, in which gene expression is controlled by a cascade of alternative sigma factors. Binding of sigma factors to RNA polymerase depends on the coiled-coil (or clamp helices) motif of the β′ subunit. We have identified an amino acid substitution (L257P) in the coiled coil that markedly inhibits the function of σH, the earliest-acting alternative sigma factor in the sporulation cascade. Cells with this mutant RNAP exhibited an early and severe block in sporulation but not in growth. The mutant was strongly impaired in σH-directed gene expression but not in the activity of the stress-response sigma factor σB. Pulldown experiments showed that the mutant RNAP was defective in associating with σH but could still associate with σA and σB. The differential effects of the L257P substitution on sigma factor binding to RNAP are likely due to a conformational change in the β′ coiled coil that is specifically detrimental for interaction with σH. This is the first example, to our knowledge, of an amino acid substitution in RNAP that exhibits a strong differential effect on a particular alternative sigma factor.
IMPORTANCE In bacteria, all transcription is mediated by a single multisubunit RNA polymerase (RNAP) enzyme. However, promoter-specific transcription initiation necessitates that RNAP associates with a σ factor. Bacteria contain a primary σ factor that directs transcription of housekeeping genes and alternative σ factors that direct transcription in response to environmental or developmental cues. We identified an amino acid substitution (L257P) in the B. subtilis β′ subunit whereby RNAPL257P associates with some σ factors (σA and σB) and enables vegetative cell growth but is defective in utilization of σH and is consequently blocked for sporulation. To our knowledge, this is the first identification of an amino acid substitution within the core enzyme that affects utilization of a specific sigma factor.
Promoter recognition in bacteria is governed by the sigma (σ) subunit of RNA polymerase (RNAP), which directly contacts −10 and −35 sequence elements upstream of the transcription start site. The primary or housekeeping sigma factor is called σ70 in Escherichia coli and σA in Bacillus subtilis. A wide variety of alternative bacterial and phage sigma factors can replace the housekeeping sigma factor to direct the recognition of alternative classes of promoters. Many of these alternative sigma factors are related to housekeeping sigma factors (1), but some, such as the ECF family, are distantly related.
The RNAP core consists of five principal subunits: β, β′, ω, and two α subunits. Sigma factors interact with RNAP in part via the highly conserved α helical region of sigma known as 2.2 and a coiled-coil motif in the β′ subunit (also called the β′ clamp helices). This interaction is essential for holoenzyme formation and is also necessary for sigma region 2.4 to be in the proper orientation to bind to the −10 promoter element and for promoter melting to occur during transcription initiation (2,–5). Here, we report the identification of an amino acid substitution in the coiled coil that selectively blocks the binding and function of an alternative sigma factor in B. subtilis, known as σH, that governs entry into the process of sporulation.
Sporulation in B. subtilis is controlled by a cascade of five alternative sigma factors that appear in a hierarchical regulatory cascade (σH→σF→σE→σG→σK). The σH factor (the product of sigH, also known as spo0H), the earliest-acting factor in the cascade, is needed for the earliest events of spore formation, including the formation of an asymmetrically positioned septum and for the production of the remaining four sigma factors (reviewed in references 6, 7, and 8).
The sigH gene is transcribed during vegetative growth and stationary phase under the control of σA, and cells lacking σH exhibit a severe sporulation defect (9,–12). Transcription of sigH increases early in sporulation due to a positive feedback loop involving σH and the master regulator for sporulation, the response regulator Spo0A. σH drives the expression of spo0A from its Ps promoter. Spo0A in its phosphorylated form (Spo0A~P) inhibits synthesis of the repressor AbrB, which, in turn, represses sigH; accordingly, Spo0A~P indirectly enhances σH synthesis (13,–16). σH also antagonizes the function of the stress response sigma factor σB in that a mutant lacking σH exhibits elevated levels of σB-directed transcription during stationary phase (17,–19).
Here, we report that replacement of leucine 257 with proline in the β′ coiled coil results in a severe block in σH-directed gene expression and, hence, entry into sporulation without impairing vegetative growth or the activity of at least one other alternative sigma factor.
During the course of generating amino acid replacements in RNA polymerase (RNAP), we identified a substitution (L257P) in the coiled-coil region of the β′ subunit that severely impaired sporulation (reducing spore formation by more than 107-fold, forming ~0 to 10 spores/ml) while having little effect on growth (the generation times of the mutant and the wild type in LB medium were almost identical) (Fig. 1A). The L257P substitution was made by a markerless strategy that left no alteration in the rpoC region of the chromosome other than the replacement of leucine at codon 257 with proline (see Materials and Methods). Microscopy revealed that the mutant cells were blocked in sporulation prior to the stage of asymmetric division (Fig. 2). Wild-type cells formed polar septa and underwent engulfment by 2.5 h after resuspension in sporulation medium and formed phase-bright spores by 6.5 h (Fig. 2A to toC).C). In contrast, few if any mutant cells could be detected that exhibited a polar septum even by 6.5 h (Fig. 2D to toFF).
Taking advantage of the strong sporulation phenotype, we selected for suppressor mutations that restored sporulation. The only suppressor mutation recovered proved to be a revertant that replaced the proline codon with a synonymous leucine codon (CTG) versus the original, wild-type leucine codon (CTT) at position 257. We interpret these results as reinforcing the conclusion that the strong sporulation defect was due specifically to the L257P substitution.
Because σH is required for asymmetric septation, we hypothesized that the L257P mutant was defective in σH activity. To measure σH activity, we used a PspoVG42-lacZ reporter gene whose expression is dependent on σH and not dependent on spo0A (20). spoVG is under the control of σH but is repressed by AbrB, which is in turn inhibited by Spo0A. PspoVG42 includes a single-nucleotide substitution in the promoter region that disrupts the AbrB binding site, resulting in a promoter that is solely dependent on σH (20). The L257P mutant exhibited little or no σH activity under sporulation-inducing conditions (Fig. 1B). σH is also known to be active in growing cells, and indeed there was no detectable σH activity in the mutant even at the time of resuspension (time zero [T0]), although wild-type cells already showed some σH activity at T0 (Fig. 1B).
We next asked whether the L257P substitution would impair the activity of another alternative sigma factor, σB. This stress response regulatory protein is activated in response to ethanol (21). The results show that σB activity was turned on normally during growth in response to 2% ethanol in the mutant and indeed reached markedly higher levels than those observed for the wild type during stationary phase. This increased level of σB-directed gene transcription in stationary phase has in fact been observed previously for a mutant lacking σH and is confirmed here (Fig. 1C) (18, 19). We conclude that the effect of the L257P substitution is specific to σH and partially phenocopies a σH mutant in that it causes high levels of σB activity.
Because σB activity is elevated during stationary phase in an L257P mutant, we considered the possibility that the severe L257P phenotype was due to increased σB competition with σH. If σB competition were a major contributor to the L257P phenotype, then a deletion of the gene for σB from L257P cells should rescue σH function. To test this, we measured σH activity and asymmetric septum formation in cells carrying both the rpoCL257P and ΔsigB mutations under sporulation conditions. The rpoCL257P ΔsigB double mutant phenocopied the rpoCL257P single mutant in that the double mutant also exhibited no σH activity (see Fig. S1 in the supplemental material) and did not form asymmetric septa even by 6.5 h into sporulation. Furthermore, σH activity was slightly elevated at early times (0.5 to 1 h) during sporulation in ΔsigB cells with wild-type RNAP (Fig. S1). We interpret these results as indicating that competition with σB does not explain the severe L257P mutant phenotype.
Because L257 is located in the coiled-coil region of β′ (RpoC), which is critical for interaction with σ factors, we hypothesized that the L257P substitution impairs binding to σH. To investigate this hypothesis, we performed pulldown experiments using lysates from stationary-phase cells producing His-tagged wild-type RpoC or His-tagged RpoCL257P. His-tagged RpoCL257P was associated with less (~3- to 4-fold) σH than was His-tagged RpoC, as determined by Western blot analysis carried out with antibodies against σH (Fig. 3A). This effect was specific for σH in that His-tagged RpoCL257P was not defective in its association with σB (Fig. 3).
In fact, the mutant RNAP was associated with slightly (~1.6-fold) more σB than was the wild-type enzyme. Likewise, RNAP from mutant cells lacking σH (ΔsigH) was associated with slightly (~1.6-fold) more σB than was RNAP from wild-type cells (Fig. 3B). Conceivably, σH and σB compete for binding to RNAP, and the inability of σH to bind to the mutant RNAP or the absence of σH from the sigH mutant cells allows somewhat more σB binding than would otherwise be the case. We noted that σB protein levels were higher (~3-fold) in the L257P cell lysate load than in WT and ΔsigH cell lysates. However, the amount of σB associated with RNAP was the same as that in ΔsigH cells (Fig. 3B). A possible interpretation is that there is more free σB in L257P cells. If so, the basis for this is unknown.
We attempted to test for a specific defect in the binding of the coiled coil containing L257P to region 2 of σH using a bacterial two-hybrid system (22). This effort was unsuccessful, as the σHreg2 fusion protein (tested as either λCI-σH or α-σH) proved to be unstable in E. coli.
RpoCL257P is also capable of binding σA, although, curiously, more σA (~3- to 4-fold) was pulled down from L257P mutant cells than from wild-type or ΔsigH cells (Fig. S2). Nevertheless, that RpoCL257P can associate with σA in addition to σB further supports the conclusion that RpoCL257P is specifically defective in its association with σH.
Leucine at position 257 is highly conserved (99% identical in an alignment of nearly 1,000 bacterial sequences) (S. Darst, personal communication). Inspection of an E. coli RNAP holoenzyme crystal structure (23) revealed that L268 (corresponding to B. subtilis L257) is not solvent exposed and is covered in part by another highly conserved leucine at position 324 (corresponding to B. subtilis L313) (S. Darst, personal communication). Replacing L257 with a variety of other amino acids (isoleucine, tyrosine, histidine, phenylalanine, glycine, and tryptophan) resulted in moderate, if any, defects in sporulation efficiency (Fig. S3). We also made a proline substitution at L295, a residue predicted to be opposite L257 that lies on the other helix of the β′ coiled coil, which resulted in only an ~2-fold sporulation defect (Fig. S3). Thus, the dramatic phenotype of L257P mutant cells seems specific to a proline substitution at that position.
Because L257 is buried, we speculate that L257P does not disrupt a direct contact site with σH but rather that the substitution changes the conformation of the coiled coil in a way that is specifically detrimental for interaction with σH. To investigate the structural effects of the L257P substitution, we ran fully atomistic molecular dynamics simulations of the WT and L257P β′ coiled coils using residues 262 to 335 (B. subtilis residues 251 to 324) from the homologous E. coli RNAP crystal structure (Protein Data Bank [PDB] entry 4LJZ) under explicit water (TIP4P-EW) and ion conditions for two independent simulations lasting 100 ns each. Although the simulations were run on the isolated β′ coiled-coil fragment and consequently do not factor in conformational constraints in the context of RNAP, local angle measurements (represented as heat maps in Fig. 4A) revealed that the L257P β′ coiled coil exhibited significantly increased bending in the region around the substitution over time, whereas the WT exhibited only moderate intrinsic flexibility in that region (Fig. 4A). Increased bending in a helix of the L257P mutant could also be seen in the simulation trajectory frames of the cartoon representation of the L257P β′ coiled coil. At the beginning of the simulation (0 ns), there was little bending, as illustrated by the colored tube that indicates the local helical angle (Fig. 4B). A bend then became increasingly obvious by the middle of the trajectory at 50 ns and continued to become enhanced during the latter part of the trajectory. Despite the increased local bending in the helix carrying the L257P substitution, the coiled coil still maintained its alpha-helical structure throughout the simulation (Fig. S4).
Promoter recognition by RNAP is governed by a housekeeping sigma factor as well as by multiple alternative sigma factors. This raises the question of how core enzyme can interact productively with a variety of alternative sigma factors, some of which are not closely related to the primary sigma factor. Here, we have identified an amino acid substitution, L257P, in the coiled coil of the β′ subunit that severely discriminates against the sporulation sigma factor σH but not against the housekeeping sigma factor σA or the stress response sigma factor σB. The coiled coil is a critical docking site for sigma factors on core enzyme (3, 4). Evidently, then, the replacement of leucine with proline perturbs the coiled coil in a manner that specifically blocks the function of the sporulation sigma factor. The results of molecular dynamics simulations raise the appealing possibility that a proline-induced bend in the coiled coil is responsible for the specific effect on σH. Perhaps σA and σB can tolerate the bend when binding to the coiled coil but σH cannot. As pointed out to us by S. Darst (personal communication), L257 appears to interact with L313, another highly conserved leucine, which lies in the so-called rudder domain in the E. coli RNAP structure. We were unable to observe any effect of the L257P substitution on the rudder, since the rudder was similarly highly mobile in the simulation of both the WT and mutant β′ coiled coil. Nonetheless, we cannot exclude that part of the effect of the L257P substitution may be mediated via its interaction with the rudder domain. Lastly, it should be noted that our results do not exclude an alternative possibility in which the reduced association of σH with RNAP in B. subtilis cells harboring the L257P substitution and the block in sporulation in these cells are a consequence of altered competition between the various σ factors rather than a reduction in affinity for σH. In particular, more σA- or σB-associated RNAPL257P could lead to a redistribution of holoenzyme activity in a manner that lowers the levels of σH-containing RNAP and prevents sporulation.
Interestingly, a BLASTp search using a short B. subtilis β′ sequence including the L257P substitution (DLNEPYRRVI) yielded matches with β′ sequences in some chloroplasts and in the marine cyanobacterium Prochlorococcus. A sequence alignment revealed that they all naturally have a proline at the position corresponding to L257 in B. subtilis (Fig. S5). This raises the possibility that this proline is part of an ancient RNAP structure that was present in cyanobacteria, maintained in some chloroplast RNAPs derived from cyanobacteria, but became modified (the coiled coil perhaps becoming rigidified) in contemporary bacteria.
We are left with the question of the basis for the distinctive behavior of σH. A possible clue comes from a comparison of conserved residues in region 2 of σH and other σ70 family members. Region 2 binds the β′ coiled coil (3, 4, 24) and is the most highly conserved region in the σ70 family (1). Figure 5 shows that conserved residues in the core-binding subregions 2.1 and 2.2 of σH among members of the Bacillus and Clostridia genera differ conspicuously at several positions from those in primary sigma factors. For example, residue 45 in σH region 2.1 is a conserved phenylalanine, whereas the corresponding residue in σB, σA, and E. coli σ70 (and other homologous primary sigma factors ) is a highly conserved leucine (Fig. 5). Furthermore, residue 72 in σH region 2.2 is usually a tyrosine instead of the conserved methionine, leucine, or isoleucine in other sigma factors (Fig. 5). A methionine-to-threonine substitution in E. coli σ70 at the position corresponding to σH residue 72 had been previously shown to result in a greater than 15-fold defect in binding to the RNAP core (25). Other conserved σH residues that differ from both σB and primary sigma factors are highlighted in yellow (Fig. 5). Conceivably, these residues represent distinctive features of σH that are responsible for, or contribute to, the specific effect of the L257P substitution in the coiled coil on the function of the sporulation sigma factor.
Escherichia coli strain XL1-Blue was used for propagating plasmids and grown and transformed using standard procedures (26). E. coli BL21(DE3) was used for the expression and purification of recombinant proteins. B. subtilis strains used in this work are listed in Table 1. Transformation of Bacillus was done as previously described (27).
Bacterial strains were propagated in Luria-Bertani medium. When appropriate, antibiotics were included at the following concentrations: chloramphenicol (5 μg/ml for B. subtilis or 25 μg/ml for E. coli), erythromycin plus lincomycin (MLS) (1 μg/ml and 25 μg/ml, respectively), spectinomycin (100 μg/ml), kanamycin (5 μg/ml for B. subtilis or 50 μg/ml for E. coli), phleomycin (0.4 μg/ml), and ampicillin (100 μg/ml).
To measure sporulation efficiency, cells were induced to sporulate by nutrient exhaustion for 25 h at 37°C in Difco (Schaeffer's) sporulation medium (DSM) (28, 29). The number of CFU that survived heat treatment (80°C for 20 min) was determined and normalized to the number of heat-resistant CFU obtained in parallel from the wild-type strain. The suppressor screen to find suppressor mutations that restored sporulation was done by subjecting RpoCL257P cells (strain AWB416 or identical isolate AWB415) to rounds of nutrient deprivation in DSM, heat treatment, and back dilution into DSM until cultures were identified that exhibited wild-type levels of sporulation efficiency. rpoC from individual heat-resistant suppressor colonies was then sequenced.
For all other experiments, sporulation was induced at 37°C by the Sterlini-Mandelstam resuspension method (28, 30) with the modification of using 25% LB instead of CH medium. β-Galactosidase activity was measured as previously described in a Synergy 2 plate reader (BioTek) (31). β-Galactosidase activity is reported in arbitrary units (AU) as the rate of 2-nitrophenyl β-d-galactopyranoside (ONPG) hydrolysis (i.e., maximum rate of reaction [Vmax], with units of optical density at 420 nm [OD420] per minute) divided by the OD600 of the culture at the time of collection.
B. subtilis strains used in this study were derived by transformation of the prototrophic laboratory strain PY79 (32) or derivatives thereof with chromosomal DNA, plasmids, or PCR products. The genes utilized to confer resistance of B. subtilis to antibiotics were cat (chloramphenicol), erm (erythromycin plus lincomycin), spc (spectinomycin), and kan (kanamycin). Competent B. subtilis cells were prepared as previously described (27). Unless otherwise noted, PY79 chromosomal DNA served as a template for PCR amplification. Plasmids were cloned and propagated in the E. coli strain XL1-Blue. Plasmid mutagenesis was performed with the QuikChange II XL site-directed mutagenesis kit (Stratagene).
Strains with markerless deletions were constructed using derivatives of the plasmid pMiniMad2 according to methods adapted from those previously described (33, 34). Briefly, a recipient B. subtilis strain was transformed with a derivative of pMiniMad2 harboring sequences homologous to regions in the chromosome flanking the intended site of deletion or mutation. Cells that integrated pMiniMad2 by single crossover were selected on LB-MLS agar. Approximately 10 MLS-resistant colonies were then picked and grown together in LB at 25°C for 1 to 3 days to allow the plasmid to loop out. Cells were then grown for ~1 day at 37°C in LB to cure the cells of the plasmid completely and then plated on LB agar. Single colonies were then screened for the desired deletion or mutation and verified by sequencing. MLS sensitivity was also checked by patching onto LB-MLS plates.
Plasmid construction and cloning were done by either traditional restriction enzyme methods or isothermal assembly (35). Plasmid digests were done with restriction enzymes (NEB or Thermo) for ~2 to 3 h at the appropriate temperature and also treated with calf intestinal phosphatase (NEB) for 30 min at 37°C.
The genotypes, features, and sources of strains and plasmids used in this study are listed in Table 1. The sequences of primers used in strain and plasmid construction are provided in Table S1 in the supplemental material. Strain and plasmid construction details are in the supplemental material.
Two hundred to 250 ml of B. subtilis culture was pelleted at 5,000 × g for 10 min and was stored at −80°C. To lyse, pellets were incubated in a 1/10 culture volume of lysis buffer (200 mM NaCl, 50 mM Tris, pH 8, 2 mM β-mercaptoethanol [β-ME], 5 mM imidazole, 1 mg/ml lysozyme, 1 cOmplete ULTRA mini EDTA-free protease inhibitor tablet per 10 ml buffer [Roche]) for 1 to 2 h at room temperature with rocking. The lysate was then sonicated at 4°C for up to a 5-min total sonication time, alternating in 0.5- to 1-min intervals between sonication and rest on ice. Cell debris was cleared by centrifugation at 14,000 rpm for 30 min at 4°C in an SS-34 or F21S-8x50y rotor and stored overnight at 4°C. The supernatant was passed through a 0.2-μm filter before use. Clarified lysate protein concentration was measured by Bradford assay (Bio-Rad) (36). Lysates that were compared to each other in pulldown assays had protein concentrations adjusted to match, using filtered lysis buffer if necessary.
Fifty microliters of His tag Dynabeads (Life Technologies) was incubated with 700 μl clarified lysate for at least 1 h at 4°C with rotating to allow binding. Manipulation of the beads was done using a magnetic tube rack half submerged in ice water to keep the protein samples cold. Wash steps were done according to the manufacturer's protocols with wash buffer (200 mM NaCl, 50 mM Tris, pH 8, 5 mM imidazole, 2 mM β-ME, 0.01% Tween 20). Samples were eluted off the His beads by incubation in 100 μl elution buffer (200 mM NaCl, 50 mM Tris, pH 8, 400 mM imidazole, 2 mM β-ME, 0.01% Tween 20) and gentle agitation by hand for 5 min at room temperature, followed by 10 min of incubation on ice. One microliter of load and flowthrough fractions and 9 μl of elution fractions were mixed with protein sample buffer (Amresco) and water (if necessary) to a final volume of 12 μl per sample, run on SDS-PAGE Next Gels (Amresco), transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore), and probed with rabbit polyclonal primary antibodies raised against RNAP (37), σH (laboratory stock), σA (laboratory stock), σB (generous gift from Ulf Gerth), and the secondary antibody goat anti-rabbit-horseradish peroxidase (Bio-Rad). Images were developed using film or the Azure imaging system (Azure Biosystems).
The coordinates of the coiled-coil domain (β′ residues 262 to 335) from the E. coli RNAP crystal structure (PDB entry 4LJZ) were selected as starting coordinates. In one of the coordinate sets, L268 (B. subtilis L257) was mutated to proline. The resulting structures were parameterized with Amber14SB, solvated in TIP4P-EW water with a 15-Å minimal distance between the solute and the border of the water box. The ionic concentration was adjusted to 150 mM with Na+ and Cl− ions at neutrality. After minimization, both wild-type and mutant structures were simulated fully atomistically in two independent simulations with pmemd.cuda (part of the Amber 16 simulation package) (38), each lasting 100 ns for the wild-type and mutant structures. The resulting trajectories were analyzed using Visual Molecular Dynamics (39).
This work was supported by the National Institutes of Health (GM18658 to R.L. and GM044025 to A.H.) and in part by a National Science Foundation graduate research fellowship to A.F.W.E.
We thank Ulf Gerth for the generous gift of the anti-σB antibody and Renate Hellmiss for assistance in preparing figures.
Supplemental material for this article may be found at https://doi.org/10.1128/JB.00277-17.