Conserved Proline Residues of Bacillus subtilis Intramembrane Metalloprotease SpoIVFB Are Important for Substrate Interaction and Cleavage

ABSTRACT

Intramembrane metalloproteases (IMMPs) regulate diverse biological processes by cleaving membrane-associated substrates within the membrane or near its surface. SpoIVFB is an intramembrane metalloprotease of Bacillus subtilis that cleaves Pro-σ^K during endosporulation. Intramembrane metalloproteases have a broadly conserved NPDG motif, which in the structure of an archaeal enzyme is located in a short loop that interrupts a transmembrane segment facing the active site. The aspartate residue of the NPDG motif acts as a ligand of the zinc ion involved in catalysis. The functions of other residues in the short loop are less well understood. We found that the predicted short loop of SpoIVFB contains two highly conserved proline residues, P132 of the NPDG motif and P135. Mutational analysis revealed that both proline residues are important for Pro-σ^K cleavage in Escherichia coli engineered to synthesize the proteins. Substitutions for either residue also impaired the Pro-σ^K interaction with SpoIVFB in copurification assays. Disulfide cross-linking experiments showed that the predicted short loop of SpoIVFB is in proximity to the N-terminal pro-sequence region (Proregion) of Pro-σ^K. Alanine substitutions for N129 and P132 of the SpoIVFB NPDG motif reduced cross-linking between its predicted short loop and the Proregion more than a P135A substitution. Conversely, the SpoIVFB P135A substitution reduced Pro-σ^K cleavage more than the N129A and P132A substitutions during sporulation of B. subtilis. We conclude that all three conserved residues of SpoIVFB are important for substrate interaction and cleavage, and we propose that P135 is necessary to position D137 to act as a zinc ligand.

IMPORTANCE Intramembrane metalloproteases (IMMPs) function in numerous signaling pathways. Bacterial IMMPs govern stress responses, including the sporulation of some species, thus enhancing the virulence and persistence of pathogens. Knowledge of IMMP-substrate interactions could aid therapeutic design, but structures of IMMP·substrate complexes are unknown. We examined the interaction of the IMMP SpoIVFB with its substrate Pro-σ^K, whose cleavage is required for Bacillus subtilis endosporulation. We found that conserved proline residues in a short loop predicted to interrupt a SpoIVFB transmembrane segment are important for Pro-σ^K binding and cleavage. The corresponding residues of the Escherichia coli IMMP RseP have also been shown to be important for substrate interaction and cleavage, suggesting that this is a broadly conserved feature of IMMPs, potentially suitable as a therapeutic target.

INTRODUCTION

Regulated intramembrane proteolysis (RIP) controls many biological processes in all three domains of life (1, 2). RIP utilizes intramembrane proteases (IPs), which cleave membrane-associated substrates within a transmembrane segment (TMS) or near the membrane surface. There are four known IP families: aspartyl proteases like presenilin (a component of γ-secretase), serine proteases (also referred to as rhomboids), the glutamyl protease Rce1, and intramembrane metalloproteases (IMMPs) (2, 3). Crystal structures have been solved for representatives of each IP family (3–8). Additionally, structures have been solved for rhomboid·peptide inhibitor complexes (9, 10) and γ-secretase·substrate complexes (11, 12); however, more work is needed in order to determine how IPs recognize and bind their substrates.

IMMPs activate transcription factors in numerous signaling pathways (1, 2). S2P, an IMMP in humans, is involved in the regulation of cholesterol homeostasis and responses to endoplasmic reticulum stress and viral infection (13, 14). Bacterial IMMPs control endosporulation, enhance pathogenicity, regulate stress responses and polar morphogenesis, produce mating signals, and clear signal peptides from the membrane (15–19). During endosporulation of Bacillus and Clostridium species, the IMMP SpoIVFB cleaves Pro-σ^K to make active σ^K, which directs RNA polymerase to transcribe genes necessary for mature spore formation (16, 20). Endospores are dormant and can survive harsh environmental conditions (21), enhancing the persistence of pathogenic species (22, 23). Thus, knowledge about SpoIVFB substrate recognition and binding could lead to new strategies to manipulate endosporulation and other processes involving IMMPs in bacteria and eukaryotes.

During endosporulation, Bacillus subtilis forms two separate compartments, the mother cell (MC) and the forespore (FS) (24) (Fig. 1A). SpoIVFB is held inactive in the outer FS membrane by the inhibitory proteins BofA and SpoIVFA (25–29). A proteolytic cascade, initiated by the FS-secreted serine proteases SpoIVB and CtpB, relieves inhibition of SpoIVFB (30–32). SpoIVB cleaves the C-terminal end of SpoIVFA (33, 34), and CtpB can cleave the C-terminal ends of both SpoIVFA and BofA (34–36). Once inhibition is removed, SpoIVFB cleaves the N-terminal pro-sequence region (Proregion) of Pro-σ^K, releasing σ^K into the MC (37–39). σ^K directs RNA polymerase to transcribe genes whose products form the spore coat and lyse the MC, releasing a mature spore (40, 41).

FIG 1.

Cartoon depictions of SpoIVFB and Pro-σ^K. (A) Membrane topology of SpoIVFB during endosporulation of B. subtilis and when synthesized in E. coli. During endosporulation, the forespore (FS) is engulfed by the mother cell (MC), resulting in two membranes surrounding the FS (left). SpoIVFB (green) and Pro-σ^K (red) are synthesized in the MC and inserted into the outer FS membrane (OFM). When these proteins are produced in the E. coli cytosol (right), they insert into the inner membrane (IM). The expanded view of the membranes (middle) shows SpoIVFB with an extra transmembrane segment (cytTM) at its N terminus, the six predicted transmembrane segments (TMSs), and the interdomain linker that connects the C-terminal cystathionine-β-synthase (CBS) domain with TMS6. The locations of the HEXXH and NPDG motifs of SpoIVFB are indicated. When produced in E. coli, cytTM-SpoIVFB recognizes C-terminally truncated Pro-σ^K (1–127) as the substrate and cleaves the N-terminal 21-residue pro-sequence from Pro-Greek sigma superscript K, which appears to associate peripherally with the outer FS membrane in B. subtilis and the IM in E. coli. See the text for references. IFM, inner forespore membrane; OM, outer membrane. (B) Expanded view of the short loop predicted to interrupt TMS4 of SpoIVFB. Residues comprising the conserved NPDG motif are indicated in yellow.

A structural model of SpoIVFB has been built based on the crystal structure of an archaeal homolog (6) and constraints from cross-linking and coevolutionary analyses (42–44). In the model, SpoIVFB has an N-terminal domain with six TMSs (45, 46), including an HEXXH metalloprotease motif in TMS2 and an NPDG motif in TMS4 (37, 38), and a C-terminal cystathionine-β-synthase (CBS) domain (39, 47) (Fig. 1A). Both the H43 and H47 residues in the HEXXH motif and D137 in the NPDG motif are predicted zinc ligands that form a catalytic core (37, 38, 47–49). Most of the NPDG motif is predicted to reside in a short loop that interrupts TMS4 (42–44) (Fig. 1B) (6). Substitutions for N129, P132, and D137 in the NPDG motif of SpoIVFB impaired RIP of Pro-σ^K during sporulation (37, 38), demonstrating that residues of the NPDG motif are important for substrate cleavage. Disulfide cross-linking experiments showed proximity between SpoIVFB P135 in the predicted short loop and residues near the cleavage site in Pro-σ^K (42).

Cleavage of Pro-σ^K by SpoIVFB can be reproduced by expressing these proteins in Escherichia coli (29) (Fig. 1A). Using this assay and bioinformatics, we found that both P132 and P135 in the predicted SpoIVFB short loop are important for Pro-σ^K cleavage and are highly conserved in SpoIVFB orthologs. Changes to these proline residues impaired the interaction between catalytically inactive SpoIVFB and Pro-σ^K produced in E. coli, as determined by copurification (pulldown) assays. However, the effects of alanine substitutions for P132 and P135 differed, both in disulfide cross-linking experiments between the predicted SpoIVFB short loop and a residue near the cleavage site in Pro-σ^K and in experiments that measured Pro-σ^K cleavage during B. subtilis sporulation. Our results provide new insights into the functions of residues in the predicted SpoIVFB short loop and the NPDG motif that is broadly conserved in IMMPs.

RESULTS

Two conserved proline residues in the predicted short loop of SpoIVFB are important for cleavage of Pro-σ^K in E. coli.

The predicted SpoIVFB short loop located in TMS4 contains most of the broadly conserved NPDG (N129, P132, D137, and G138) motif (also referred to as the LDG motif) in IMMPs (Fig. 1A and B) (37, 47, 50). To identify additional residues that may play a role in substrate interaction and cleavage, an alignment of 136 SpoIVFB orthologs was made (Fig. 2A; see also Fig. S1 in the supplemental material). The predicted short loop region is highly conserved, with little variability in residues and no variability in length. Interestingly, a second Pro residue, P135 in B. subtilis SpoIVFB, is conserved in all orthologs. Previous alignments of diverse IMMPs showed that some contain two or three Pro residues at variable positions within the predicted short loop region, while others contain only one Pro residue (37, 47). Therefore, we tested whether B. subtilis SpoIVFB P135 is important for Pro-σ^K cleavage. We also tested the functional importance of I133 and W134, which are not highly conserved in orthologs (Fig. 2A and Fig. S1). We compared them with two residues of the broadly conserved NPDG motif, N129 and P132, which have previously been shown to be important for RIP of Pro-σ^K during B. subtilis sporulation (37).

FIG 2.

Conserved residues in the predicted short loop region of SpoIVFB and effects of substitutions on the cleavage of Pro-σ^K(1–127) in E. coli. (A) Alignment of the predicted short loop region of B. subtilis SpoIVFB with representative orthologs. Here and in the full alignment (see Fig. S1 in the supplemental material), highlighted residues are 100% conserved (black), at least 90% conserved (dark gray), or at least 70% conserved (light gray). (B) Effects of Ala and Cys substitutions in SpoIVFB on the cleavage of Pro-σ^K(1–127). Plasmids were used to express Pro-σ^K(1–127) and cytTM-SpoIVFB (lane 1) (pFB9) or the indicated cytTM-SpoIVFB variant (lanes 2 to 11) (pEN14, pEN18, pFB10, pFB12 to pFB16, pJV22, or pSO288) in E. coli. Samples collected after 2 h of IPTG induction were subjected to immunoblot analysis with FLAG (top) and penta-His (bottom) antibodies (7- and 30-s exposures). The graph shows quantification of the cleavage ratio {cleavage product/[Pro-σ^K(1–127) + cleavage product]} for three biological replicates (bars, means; error bars, 1 standard deviation; open circles, individual replicates). Results of Student’s two-tailed t tests comparing the cleavage ratios of the N129A variant (lane 2) and cytTM-SpoIVFB (lane 1), the P132A (lane 4) and P135A (lane 10) variants, and the P132C (lane 5) and P135C (lane 11) variants are shown above the error bars for the variant with the lower mean (*, P < 0.05; **, P < 0.01). MW, molecular weight.

To test the cleavage of Pro-σ^K by SpoIVFB easily and in the absence of other B. subtilis proteins, E. coli was used to express variants of SpoIVFB and Pro-σ^K in various combinations. The SpoIVFB variant cytTM-SpoIVFB-FLAG₂ (cytTM-SpoIVFB) (Fig. 1A) contains a cytochrome TMS (cytTM) (51), which improves accumulation (39), perhaps by increasing the efficiency of the recognition and insertion process of the protein translocon (52). The substrate variant Pro-σ^K(1–127)-His₆ [Pro-σ^K(1–127)] lacks the C-terminal half of Pro-σ^K but can be cleaved by SpoIVFB, and the cleavage product can easily be separated from Pro-σ^K(1–127) by SDS-PAGE (53) [note that Pro-σ^K(1–126) was renamed Pro-σ^K(1–127), as explained in reference 43]. When Pro-σ^K(1–127) was produced with cytTM-SpoIVFB, cleavage was abundant (Fig. 2B, lane 1). To quantify cleavage, we measured the amount of the cleavage product and divided that by the amount of Pro-σ^K(1–127) plus the cleavage product, which resulted in a cleavage ratio of 0.88 ± 0.01.

When Ala and Cys substitutions were made for P132 and P135, very little cleavage product was observed (Fig. 2B, bottom immunoblot, lanes 4, 5, 10, and 11), demonstrating that both conserved Pro residues are important for substrate cleavage in E. coli. A longer exposure of the immunoblot revealed small amounts of cleavage product. Interestingly, the N129A variant showed a small but reproducible decrease in the cleavage ratio (lane 2) compared with cytTM-SpoIVFB (lane 1) or the N129C variant (lane 3). The top immunoblot in Fig. 2B shows the accumulation of the cytTM-SpoIVFB variants. The accumulation of the N129A variant appeared to be slightly reduced (lane 2) due to proteolysis since a breakdown species was detected (see Fig. S2 in the supplemental material). The accumulation of the P132 and P135 variants, as well as the W134C variant, also appeared to be reduced slightly, yet the cleavage ratio of the W134C variant (Fig. 2B, lane 9) was similar to that of cytTM-SpoIVFB (lane 1) or the W134A variant (lane 8). Ala and Cys substitutions for I133 also had no measurable effect on the cleavage ratio (lanes 6 and 7). Since substitutions for P132 and P135 strongly impaired Pro-σ^K(1–127) cleavage in E. coli, we focused on these substitutions to test substrate binding as described below.

Substitutions for P132 and P135 of SpoIVFB reduce interaction with Pro-σ^K(1–127).

To test substrate binding, E. coli was used to express Pro-σ^K(1–127) in combination with catalytically inactive cytTM-SpoIVFB E44Q variants (which have a FLAG₂ tag), and pulldown assays were performed by mixing cell extracts with cobalt resin, which binds to the His₆ tag on Pro-σ^K(1–127). As expected, the cytTM-SpoIVFB variant copurified with Pro-σ^K(1–127) (Fig. 3A). Both proteins were observed in the diluted (to match the input concentration) bound sample (lane 6), indicative of a strong interaction, as well as in the concentrated bound sample (lane 7). The cytTM-SpoIVFB variant was undetectable in the bound samples of a negative control with Pro-σ^K(1–127) lacking the His₆ tag (lanes 13 and 14), indicating that the cytTM-SpoIVFB variant does not bind nonspecifically to the resin. As expected, none of the negative-control samples showed a detectable signal with penta-His antibodies (Fig. 3, top, lanes 8 to 14), and with Pro-σ^K antibodies, only the input and unbound negative-control samples showed Pro-σ^K(1–127) (see Fig. S3, middle, lanes 8 to 14, in the supplemental material).

FIG 3.

Effects of substitutions for SpoIVFB P132 or P135 on copurification with Pro-σ^K(1–127). Plasmids were used to express Pro-σ^K(1–127) or a variant lacking the His₆ tag (as a negative control) in combination with catalytically inactive cytTM-SpoIVFB E44Q variants in E. coli. Samples collected after 2 h of IPTG induction were subjected to copurification with cobalt resin. Input, unbound, wash, 1/15 bound (diluted to match the input), and (undiluted) bound samples were subjected to immunoblot analysis with penta-His and FLAG antibodies as indicated. (A) Representative results from two biological replicates. cytTM-SpoIVFB E44Q was produced in combination with Pro-σ^K(1–127) (lanes 1 to 7) (pJV26) or a variant lacking the His₆ tag (negative control) (lanes 8 to 14) (pSO292), or Pro-σ^K(1–127) was produced in combination with the indicated cytTM-SpoIVFB E44Q variant (lanes 15 to 21, pJV25; lanes 22 to 28, pJV29; lanes 29 to 35, pSO291; lanes 36 to 42, pJV23). (B) Quantification of binding. The graph shows quantification of the bound ratios [diluted bound/(unbound + washes + diluted bound)] for two biological replicates of the indicated cytTM-SpoIVFB E44Q variants (bars, mean; error bars, 1 standard deviation). Results of Student’s two-tailed t tests comparing Ala- or Cys-substituted variants with the cytTM-SpoIVFB E44Q control are summarized above the error bars, and the result of comparing the combined bound ratios of the P132 variants with those of the P135 variants is summarized above the bracket (*, P < 0.05; **, P < 0.01).

Ala and Cys substitutions for P132 and P135 of the cytTM-SpoIVFB variant appeared to reduce the interaction with Pro-σ^K(1–127) since less of each variant appeared to be present in the bound samples (Fig. 3A, lanes 20, 21, 27, 28, 34, 35, 41, and 42) than in the control (lanes 6 and 7). To quantify binding, we measured the amount of the cytTM-SpoIVFB variant in the diluted bound sample and divided that by the amounts in the unbound, wash, and diluted bound samples (Fig. 3B). On average, the Ala- or Cys-substituted P132 variants exhibited about 2-fold less binding to Pro-σ^K(1–127) than the cytTM-SpoIVFB E44Q control (P = 0.044 and 0.061 for the P132A and P132C variants, respectively), and the P135 variants exhibited about 4-fold less binding (P = 0.0096 and 0.020 for the P135A and P135C variants, respectively). Pairwise comparisons of P132A and P135A variants or P132C and P135C variants yielded P values of >0.05, but comparison of the combined bound ratios of the P132 variants with those of the P135 variants yielded P values of 0.026 (Fig. 3B). We conclude that both P132 and P135 in the predicted short loop of SpoIVFB are important for the interaction with Pro-σ^K(1–127), and P135 is more important for the substrate interaction than P132 of the NPDG motif.

The predicted short loop of SpoIVFB is in proximity to the Proregion of Pro-σ^K(1–127) when the proteins interact.

SpoIVFB cleaves Pro-σ^K (40) and Pro-σ^K(1–127) (29) between residues S21 and Y22. In previous disulfide cross-linking experiments, Cys substitutions for several residues near the cleavage site of otherwise Cys-less Pro-σ^K(1–127) formed a cross-linked complex with single-Cys (E44C, V70C, or P135C) cytTM-SpoIVFB variants in E. coli (42). However, the P135C variant formed only small amounts of cross-linked complex. To examine proximity between other residues in the predicted SpoIVFB short loop region and the Proregion of Pro-σ^K(1–127), we performed additional disulfide cross-linking experiments.

For comparison as controls, we coproduced single-Cys (F18C, V20C, or S21C) or Cys-less Pro-σ^K(1–127) variants with a single-Cys E44C cytTM-SpoIVFB variant in E. coli. Cells were treated with the oxidant Cu²⁺(phenanthroline)₃ to promote disulfide bond formation. As expected, the formation of a species of the expected size for a cross-linked complex between the two proteins was detected by immunoblotting with anti-FLAG antibodies in some cases (Fig. 4, lanes 1 to 12). The F18C (lane 8) and V20C (lane 2) Pro-σ^K(1–127) variants formed more cross-linked complex than the S21C (lane 11) Pro-σ^K(1–127) variant with the E44C cytTM-SpoIVFB variant, in agreement with previous results (42). Treatment with the reducing agent dithiothreitol (DTT) decreased the abundance of the cross-linked complexes (lanes 3, 9, and 12), indicating that the cross-links are reversible. As expected, a cross-linked complex was not observed with Cys-less Pro-σ^K(1–127) (lane 5). A species of the expected size for a cross-linked dimer of single-Cys cytTM-SpoIVFB was also observed in samples treated with the oxidant. The formation of the apparent dimer varies, as reported previously (42).

FIG 4.

Disulfide cross-linking between the predicted short loop region of SpoIVFB and the Proregion of Pro-σ^K(1–127). Plasmids were used to express catalytically inactive single-Cys cytTM-SpoIVFB variants in combination with Cys-less (as a negative control) or single-Cys Pro-σ^K(1–127) variants in E. coli. Samples collected after 2 h of IPTG induction were treated with Cu²⁺(phenanthroline)₃ (Cu +) for 60 min or with 2-phenanthroline (Cu −) as a control. Samples were treated with tricarboxylic acid (TCA) to precipitate proteins, resuspended in sample buffer with (+) or without (−) DTT, and subjected to immunoblot analysis with FLAG antibodies to visualize the cytTM-SpoIVFB monomer, dimer, and complex with Pro-σ^K(1–127). Representative results from two biological replicates are shown. As controls for cross-linked complex formation, single-Cys E44C cytTM-SpoIVFB was produced in combination with single-Cys V20C (lanes 1 to 3) (pSO169), F18C (lanes 7 to 9) (pSO167), S21C (lanes 10 to 12) (pSO170), or Cys-less (lanes 4 to 6) (pSO79) Pro-σ^K(1–127). Single-Cys N129C cytTM-SpoIVFB E44Q was produced in combination with single-Cys F18C (lanes 13 to 15) (pSO283), S21C (lanes 16 to 18) (pJV12), Y22C (lanes 19 to 21) (pJV1), or N25C (lanes 22 to 24) (pSO277) Pro-σ^K(1–127). Likewise, single-Cys P132C cytTM-SpoIVFB E44Q was produced in combination with the indicated single-Cys Pro-σ^K(1–127) variants (lanes 25 to 36) (pSO284, pSO269, pSO272, and pSO278), single-Cys I133C cytTM-SpoIVFB E44Q was produced in combination with the indicated single-Cys Pro-σ^K(1–127) variants (lanes 37 to 48) (pSO285, pSO270, pSO273, and pSO279), single-Cys W134C cytTM-SpoIVFB E44Q was produced in combination with the indicated single-Cys Pro-σ^K(1–127) variants (lanes 49 to 60) (pSO286, pSO271, pSO274, and pSO280), and single-Cys P135C cytTM-SpoIVFB E44Q was produced in combination with the indicated single-Cys Pro-σ^K(1–127) variants (lanes 61 to 72) (pSO287, pJV28, pJV5, and pSO281).

Single-Cys (F18C, S21C, Y22C, or N25C) Pro-σ^K(1–127) variants were coproduced with catalytically inactive single-Cys (N129C, P132C, I133C, W134C, or P135C) cytTM-SpoIVFB E44Q variants, and disulfide cross-linking experiments were performed as described above. The I133C and W134C cytTM-SpoIVFB variants formed a cross-linked complex with all four Pro-σ^K(1–127) variants (Fig. 4, lanes 37 to 60). The I133C cytTM-SpoIVFB variant formed the most complex with the N25C Pro-σ^K(1–127) variant (lane 47), while the W134C cytTM-SpoIVFB variant showed abundant complex formation with both the S21C and N25C Pro-σ^K(1–127) variants (lanes 53 and 59). The N129C, P132C, and P135C cytTM-SpoIVFB variants formed small amounts of cross-linked complex with the Pro-σ^K(1–127) variants (lanes 13 to 36 and 61 to 72). The results for the P135C cytTM-SpoIVFB variant were consistent with those reported previously (42). We also observed small amounts of cross-linked complex formed between the Pro-σ^K(1–127) variants and cytTM-SpoIVFB variants with a single Cys located at positions predicted to be distal from the active site (Fig. S4). The small amounts of complex formation may reflect transient interactions between the Proregion of Pro-σ^K(1–127) and various parts of cytTM-SpoIVFB (see Discussion). In contrast, an S124C cytTM-SpoIVFB variant formed abundant complex with the F18C and Y22C Pro-σ^K(1–127) variants (Fig. S4D, lanes 38 and 44). The S124 residue of SpoIVFB is predicted to be located in TMS4 near the predicted short loop (Fig. S4A and B). We conclude that S124 and both I133 and W134 in the predicted short loop of SpoIVFB are in proximity to several residues near the cleavage site in Pro-σ^K(1–127) when the two proteins interact.

Alanine substitutions for N129 and P132 of SpoIVFB reduce cross-linking between its predicted short loop and the Proregion of Pro-σ^K(1–127) more than a P135A substitution.

To measure the effects of Ala substitutions for N129, P132, and P135 of SpoIVFB on proximity between the predicted SpoIVFB short loop and the Proregion of Pro-σ^K(1–127), we examined whether the substitutions affect disulfide cross-linking between catalytically inactive single-Cys (I133C or W134C) cytTM-SpoIVFB E44Q variants and the single-Cys N25C Pro-σ^K(1–127) variant in experiments similar to those shown in Fig. 4.

The N129A and P132A substitutions appeared to reduce cross-linking between the I133C cytTM-SpoIVFB and N25C Pro-σ^K(1–127) variants more than the P135A substitution (Fig. 5A, lanes 8, 11, and 14). Control experiments with no Ala substitution in the I133C cytTM-SpoIVFB variant in combination with the N25C Pro-σ^K(1–127) variant (lane 5) or Cys-less Pro-σ^K(1–127) (lane 2) were performed for comparison. The N129A and P132A substitutions also appeared to reduce cross-linking between the W134C cytTM-SpoIVFB and N25C Pro-σ^K(1–127) variants more than the P135A substitution (see Fig. S5, lanes 8, 11, and 14, in the supplemental material). To quantify cross-linking in the samples treated with only Cu²⁺(phenanthroline)₃, we measured the amount of the cytTM-SpoIVFB variant in the cross-linked complex and divided that by the amount in the monomer, dimer, and complex (Fig. 5B). The N129A and P132A substitutions reduced the cross-linked ratio to values similar those for to the Cys-less Pro-σ^K(1–127) negative controls, whereas the P135A substitution reduced the cross-linked ratio by about one-third on average compared with the corresponding I133C or W134C cytTM-SpoIVFB positive control lacking an Ala substitution (P = 0.075 and 0.087 for I133C and W134C, respectively). Comparison of the combined cross-linked ratios of the positive controls with those of the P135A-substituted variants yielded a P value of 0.0061 (Fig. 5B). Pairwise comparisons of the P135A-substituted I133C or W134C cytTM-SpoIVFB variants with the corresponding N129A- or P132A-substituted variants supported that the latter substitutions reduce cross-linking more than the P135A substitution (P < 0.05).

FIG 5.

Effects of alanine substitutions for SpoIVFB N129, P132, and P135 on disulfide cross-linking between the predicted short loop and the Proregion of Pro-σ^K(1–127). Plasmids were used to express Cys-less or single-Cys N25C Pro-σ^K(1–127) variants in combination with catalytically inactive single-Cys I133C cytTM-SpoIVFB E44Q variants in E. coli. Samples collected after 2 h of IPTG induction were treated with Cu²⁺(phenanthroline)₃ (Cu +) for 60 min or with 2-phenanthroline (Cu −) as a control. Samples were treated with TCA to precipitate proteins, resuspended in sample buffer with (+) or without (−) DTT, and subjected to immunoblot analysis with FLAG antibodies to visualize the cytTM-SpoIVFB monomer, dimer, and complex with Pro-σ^K(1–127). (A) Representative results from two biological replicates. As controls, single-Cys I133C cytTM-SpoIVFB E44Q was produced in combination with Cys-less (lanes 1 to 3) (pSO295) or single-Cys N25C (lanes 4 to 6) (pSO279) Pro-σ^K(1–127). Single-Cys I133C cytTM-SpoIVFB E44Q with an Ala substitution at N129 (lanes 7 to 9) (pSO296), P132 (lanes 10 to 12) (pSO297), or P135A (lanes 13 to 15) (pSO298) was produced in combination with single-Cys N25C Pro-σ^K(1–127). (B) Quantification of cross-linking. The graph shows the quantification of the cross-linked ratio [complex/(monomer + dimer + complex)] for two biological replicates of the indicated cytTM-SpoIVFB E44Q variants (bars, mean; error bars, 1 standard deviation). Results of Student’s two-tailed t tests comparing the N129A- or P132A-substituted I133C or W134C cytTM-SpoIVFB E44Q variants pairwise with the corresponding P135A-substituted variants are summarized above the error bars, and the result of comparing the combined cross-linked ratios of the I133C and W134C cytTM-SpoIVFB positive controls lacking an Ala substitution with the combined ratios of the P135A-substituted variants is summarized above the bracket (*, P < 0.05; **, P < 0.01).

We infer that the N129A and P132A substitutions reduce proximity between the predicted SpoIVFB short loop and the Proregion of Pro-σ^K(1–127), perhaps decreasing substrate cleavage (Fig. 2B) and interaction (at least in the case of the P132A substitution) (Fig. 3). In contrast, the P135A substitution appears to have less of an effect on proximity between the predicted SpoIVFB short loop and the Proregion, yet substrate cleavage (Fig. 2B) and interaction (Fig. 3) are strongly impaired, suggesting that P135 plays a different role (see Discussion).

SpoIVFB P135 is crucial for Pro-σ^K cleavage during B. subtilis sporulation.

A B. subtilis spoIVFΔAB::cat deletion mutant provides a genetic background to test the effects of Ala substitutions in SpoIVFB during sporulation (54). The effects of SpoIVFB variants (N129A, P132A, I133A, W134A, and P135A) were tested by ectopically integrating mutant versions of the spoIVFAB operon into the chromosome of the spoIVFΔAB::cat mutant lacking the endogenous spoIVFAB operon. The strains were starved to induce sporulation, and samples were collected at appropriate times to observe Pro-σ^K cleavage as well as the SpoIVFA and SpoIVFB levels by immunoblotting. To quantify the cleavage of Pro-σ^K, we measured the amount of σ^K and divided that by the amounts of Pro-σ^K and σ^K. As a control, in wild-type B. subtilis strain PY79, Pro-σ^K cleavage was observed at 4 and 5 h poststarvation (PS), resulting in cleavage ratios of 0.32 ± 0.04 and 0.57 ± 0.02, respectively (Fig. 6, lanes 1 and 2). For the spoIVFΔAB::cat deletion mutant, no Pro-σ^K cleavage was observed, as expected, since SpoIVFA and SpoIVFB were absent (lanes 3 and 4). When SpoIVFA and wild-type SpoIVFB were produced ectopically in the deletion mutant, Pro-σ^K cleavage as well as the SpoIVFA and SpoIVFB levels were restored (lanes 5 and 6) to normal (lanes 1 and 2).

FIG 6.

Effects of alanine substitutions in SpoIVFB on Pro-σ^K cleavage during B. subtilis sporulation. Wild-type strain PY79 (lanes 1 and 2), a spoIVFΔAB::cat deletion mutant (lanes 3 and 4), and the deletion mutant with either the wild-type (WT) spoIVF operon (lanes 5 and 6) or the indicated mutant version (lanes 7 to 16) integrated ectopically at amyE were starved to induce sporulation. Samples collected at 4 and 5 h poststarvation (PS) were subjected to immunoblot analysis with antibodies against SpoIVFA, SpoIVFB, and Pro-σ^K. The graph shows quantification of the cleavage ratio [cleavage product/(Pro-σ^K + cleavage product)] for three biological replicates (bars, means; error bars, 1 standard deviation; open circles, individual replicates). Student’s two-tailed t tests were performed to compare certain cleavage ratios. *, P ≤ 0.01; **, P = 0.0002 (in comparison with the data for the deletion mutant with the wild-type spoIVF operon at amyE [lane 6]).

The ectopic production of the I133A and W134A SpoIVFB variants in the deletion mutant restored the cleavage of Pro-σ^K (Fig. 6, lanes 11 to 14). Cleavage was comparable to those of wild-type strain PY79 (lanes 1 and 2) and the deletion mutant with ectopic production of SpoIVFA and wild-type SpoIVFB (lanes 5 and 6). The I133A and W134A SpoIVFB variants accumulated normally (Fig. 6; see also Fig. S5 in the supplemental material for a longer exposure of the immunoblot), but less SpoIVFA accumulated at 5 h PS (Fig. 6, lanes 12 and 14).

The production of the N129A and P132A SpoIVFB variants only partially restored Pro-σ^K cleavage in the deletion mutant (Fig. 6, lanes 7 to 10). SpoIVFB N129A accumulated normally, but very little SpoIVFB P132A accumulated (Fig. 6 and Fig. S5), and less SpoIVFA accumulated at 5 h PS in the latter strain (Fig. 6, lane 10). Despite very little accumulation of SpoIVFB P132A, the cleavage of Pro-σ^K was comparable to that observed for SpoIVFB N129A (lanes 7 to 10). We note that the N129A and P132A cytTM-SpoIVFB variants accumulated comparably when expressed in E. coli (albeit the accumulation of both appeared to be reduced slightly in comparison with cytTM-SpoIVFB) (Fig. 2B and Fig. S2), but the P129A variant cleaved Pro-σ^K(1–127) much more efficiently than did the P132A variant, indicating differences in protein accumulation and activity in E. coli versus sporulating B. subtilis (see Discussion).

The P135A substitution in SpoIVFB had the largest impact on Pro-σ^K cleavage in sporulating B. subtilis (Fig. 6, lanes 15 and 16). Very little σ^K was observed at 4 or 5 h PS. SpoIVFB P135A accumulated normally (Fig. 6 and Fig. S5), but less SpoIVFA accumulated at 5 h PS (Fig. 6, lane 16), as observed for several other strains (lanes 10, 12, and 14). Thus, P135 is crucial for Pro-σ^K cleavage during B. subtilis sporulation, as reported previously for the nearby D137 of the NPDG motif (37, 38).

DISCUSSION

Our results provide evidence that residues in the predicted short loop of SpoIVFB are important for substrate interaction and cleavage. Although bioinformatics identified a broadly conserved NPDG motif in IMMPs (37, 47, 49) that is predicted to be located in a short loop near the active site based on the structure of an archaeal enzyme (6), functional analysis of predicted short loop residues has been reported only for E. coli RseP (50). Our results reveal notable differences in function between residues in the predicted short loop regions of SpoIVFB and RseP, yet both studies support the conclusion that residues in the predicted short loop region are important for substrate interaction and cleavage, suggesting that this is a broadly conserved feature of IMMPs.

IMMPs have a variable number of proline residues in their predicted short loop.

Phylogenetic analysis of IMMPs defined four groups, including those that contain a C-terminal CBS domain like SpoIVFB (group III), those that contain one or more PDZ domains such as RseP and human S2P (group I), one devoid of additional domains (group II), and one with a conserved N-terminal extension that may form a soluble domain (group IV) (47). All four groups contain the NPDG motif, but the conservation of additional Pro residues in the predicted short loop varies. SpoIVFB and its orthologs have one additional Pro residue (P135 in B. subtilis) between the Pro and Asp of their NPDG motif (Fig. 2A; see also Fig. S1 in the supplemental material), as do diverse bacterial and archaeal group III IMMPs (47). RseP also has one additional Pro residue (P399) between the Pro and Asp of its NPDG motif (37), as do diverse bacterial and predicted chloroplast group I IMMPs (47), but in these IMMPs, the additional Pro residue is one position closer to the Pro of the NPDG motif than in SpoIVFB. Archaeal group I IMMPs have one additional Pro residue in the same position as that in SpoIVFB, or they lack an additional Pro residue, as do eukaryotic group I IMMPs like human S2P and group IV IMMPs (47). Group II IMMPs are the most variable in terms of additional Pro residues in their predicted short loop, ranging from zero to two additional Pro residues between the Pro and Asp of their NPDG motif (47). Our analysis of B. subtilis YydH suggests that it is a group II IMMP that lacks an additional Pro residue. YydH has been suggested to cleave YydF, resulting in an exported peptide that causes cell envelope stress sensed by the well-characterized LiaRS two-component system (55), so it may be feasible to study the functions of residues in the predicted short loop of YydH. Interestingly, group II IMMPs with two additional Pro residues in their predicted short loop are encoded in genomes of hyperthermophilic archaea and bacteria (47), perhaps suggesting an important role of additional Pro residues under extreme conditions.

Residues in the predicted short loop of SpoIVFB function differently in substrate interaction and cleavage than in RseP.

As noted above, both SpoIVFB and RseP contain one additional Pro residue between the Pro and Asp of their NPDG motif albeit at a slightly different position. The NPDG motif of RseP spans from N394 to G403, interrupting TMS3. The Pro residue of the NPDG motif is P397, and the additional Pro residue is P399. Cys substitutions for N394, P397, and P399 of RseP reduced substrate cleavage (50) but not as dramatically as Cys substitutions for P132 and P135 of SpoIVFB reduced substrate cleavage in E. coli (Fig. 2B). Each of the three Cys substitutions in RseP strongly impaired substrate interactions based on pulldown assays (50), whereas a Cys substitution for P135 of SpoIVFB impaired the substrate interaction more strongly than did a Cys substitution for P132 (Fig. 3). Finally, Cys substitutions for P397 and P399 of RseP resulted in readily detectable disulfide cross-linking with Cys substitutions for substrate residues near the cleavage site, but a Cys substitution for N394 of RseP gave no detectable cross-linking (50). In contrast, Cys substitutions for N129, P132, and P135 of SpoIVFB resulted in small amounts of cross-linking with Cys substitutions near the substrate cleavage site (Fig. 4). However, Cys substitutions for I133 and W134 of SpoIVFB exhibited readily detectable cross-linking with Cys substitutions near the substrate cleavage site (Fig. 4).

The differing cross-linking, pulldown, and substrate cleavage results for predicted short loop residues of RseP and SpoIVFB suggest somewhat different roles in substrate interaction and cleavage, which is not surprising given the complexity of IMMP-substrate interactions so far uncovered for RseP (50, 56–59) and SpoIVFB (42–44, 60, 61). Yet both our results for SpoIVFB reported here and the work by Koide et al. on RseP (50) demonstrate that the Asn and Pro of the NPDG motif as well as the additional Pro in the predicted short loop are important for substrate interaction and cleavage. As explained above, SpoIVFB and RseP are in distinct IMMP groups (47, 49), so the importance of predicted short loop residues in substrate interaction and cleavage may be a broadly conserved feature of IMMPs.

Dynamic interactions between IMMPs and their substrates.

The interpretation of RseP and SpoIVFB disulfide cross-linking results has invoked dynamic interactions with RseA and Pro-σ^K, respectively. The finding that RseP P397C and P399C formed cross-links with Cys substitutions at five adjacent positions immediately N terminally of the RseA cleavage site was proposed to reflect a flexible and/or multistep binding interaction (50). E44C and V70C cytTM-SpoIVFB variants formed abundant cross-links with Cys substitutions for several residues near the Pro-σ^K(1–127) cleavage site, likewise supporting a dynamic interaction (42). A P135C cytTM-SpoIVFB variant formed small amounts of cross-linked complex with Cys substitutions for several residues near the Pro-σ^K(1–127) cleavage site (42) (Fig. 4), and we observed similar results for the N129C and P132C cytTM-SpoIVFB variants (Fig. 4), providing some support for a dynamic interaction between the predicted short loop and the substrate. Stronger support for this notion is provided by our finding that the I133C and W134C cytTM-SpoIVFB variants form more abundant cross-links with Cys substitutions for several residues near the Pro-σ^K(1–127) cleavage site (Fig. 4). We also discovered that an S124C cytTM-SpoIVFB variant forms abundant cross-links with the F18C and Y22C Pro-σ^K(1–127) variants (Fig. S4D), supporting a dynamic binding interaction.

Intriguingly, S124 of SpoIVFB is near N122, a residue highly conserved among SpoIVFB orthologs (Fig. 1A and Fig. S1). Ala and Asp substitutions for N122 impaired Pro-σ^K cleavage during B. subtilis sporulation, but a conservative Gln substitution had no effect (37). SpoIVFB N122 and RseP N389 are predicted to be similarly located in a TMS interrupted by a predicted short loop. Cys, Gly, and Leu substitutions for N389 impaired RseA binding and cleavage, but a Gln substitution had no effect (50). RseP N389C gave no detectable disulfide cross-linking with Cys substitutions near the RseA cleavage site, but Gly and Leu substitutions for N389 in RseP P399C decreased cross-linking to RseA variants with Cys substitutions near the cleavage site, whereas a Gln substitution for N389 had no effect (50). Those authors suggested that Asn or Gln at position 389 of RseP might be important to establish a proper structural arrangement of nearby residues directly involved in substrate binding. We propose that Asn or Gln at position 122 of SpoIVFB likewise facilitates substrate binding. We predict that Gly and Leu substitutions for N122 in the cytTM-SpoIVFB S124C variant would decrease cross-linking to the F18C and Y22C Pro-σ^K(1–127) variants. We note that the polar side chain of S124 in SpoIVFB may not interact with the aromatic side chains of F18 and Y22 in Pro-σ^K, but the side chains of nearby residues may be directly involved in a dynamic binding interaction.

The small amounts of cross-linking observed for the N129C, P132C, and P135C cytTM-SpoIVFB variants with Cys substitutions near the Pro-σ^K(1–127) cleavage site may reflect less overall binding, as observed for the P132C and P135C variants in pulldown assays (Fig. 3), and/or local binding interactions that are more transient than those for the S124C, I133C, and W134C variants. Surprisingly, we found that cytTM-SpoIVFB variants with Cys substitutions predicted to be distal from the active site (S108C, A231C, and E282C) also exhibited small amounts of cross-linking with Cys substitutions near the substrate cleavage site (Fig. S4D). We note that cross-linking appeared to be the least for the S108C variant in which the Cys substitution is located on a predicted loop between TMS3 and TMS4 of SpoIVFB (Fig. S4A and B). The loop is predicted to be in the periplasm upon expression in E. coli, whereas the Cys substitutions in the A231C and E282C variants are in the predicted cytoplasmic CBS domain of SpoIVFB (Fig. S4A), which has been shown to interact with Pro-σ^K(1–127) (39, 43). The Proregion of Pro-σ^K(1–127) may transiently interact with the SpoIVFB CBS domain prior to the proposed dynamic interaction with the active-site cleft that includes some relatively persistent local binding interactions and results in cleavage. The coexpression of both proteins from a single plasmid in this work may have enhanced our ability to detect transient interactions. In previous work, cross-linking was not detected between an E282C cytTM-SpoIVFB variant expressed from one plasmid and a K24C Pro-σ^K(1–127) variant expressed from an incompatible plasmid (42). Although maintenance of both plasmids was selected using different markers, the copy number of the two plasmids and the resulting expression of the two proteins may have been unequal in some cells, decreasing the ability to detect transient interactions.

SpoIVFB P135 functions differently than N129 and P132 in substrate interaction and cleavage.

The effects of an Ala substitution for P135 in the predicted short loop of SpoIVFB differed strikingly from the effects of both N129A and P132A substitutions in two experiments. First, the P135A substitution appeared to have less of an effect on proximity between the predicted SpoIVFB short loop and the Proregion of Pro-σ^K(1–127) (Fig. 5). Second, the P135A substitution impaired the cleavage of Pro-σ^K more strongly during B. subtilis sporulation (Fig. 6). These two observations, taken together with our finding that Ala and Cys substitutions for P135 impaired the interaction with Pro-σ^K(1–127) more strongly than did Ala and Cys substitutions for P132 (Fig. 3), suggest that P135 plays a different role in substrate interaction and cleavage than P132 and N129.

We propose that SpoIVFB P135 positions D137 to act as a zinc ligand. Like the P135A substitution (Fig. 6), substitutions for D137 (to A, E, or N) allowed the SpoIVFB variant to accumulate but strongly impaired Pro-σ^K cleavage in sporulating B. subtilis, as did substitutions for H43 or H47, the other two predicted zinc ligands (37, 38). Improper zinc binding by SpoIVFB P135A could also explain the observed differences in substrate interactions compared with SpoIVFB P132A (Fig. 3 and 5), if P132A primarily affects the local interaction between the predicted short loop and the Proregion, whereas P135A permits the local interaction but impairs several other interactions between SpoIVFB and Pro-σ^K. Like P132A, N129A may primarily affect the local interaction, which could explain their similar effects on substrate interaction (Fig. 5) and Pro-σ^K cleavage during B. subtilis sporulation (Fig. 6). However, N129A had much less of an effect than P132A on Pro-σ^K(1–127) cleavage in E. coli (Fig. 2B). At present, we cannot account for this difference, but we note that differences between heterologous overexpression in growing E. coli and expression in sporulating B. subtilis have been observed in several studies of IMMP activity (43, 61, 62).

The predicted short loop of SpoIVFB may orient Pro-σ^K for cleavage.

Previous disulfide cross-linking studies suggested that side chains of certain residues near the cleavage site in Pro-σ^K adopt preferred orientations in the active site of SpoIVFB (42). The Pro-σ^K S21 side chain was suggested to point toward P135 of the predicted SpoIVFB short loop. In agreement, the S21C Pro-σ^K(1–127) variant formed a small amount of cross-linked complex with the P135C cytTM-SpoIVFB variant, as well as with the N129C and P132C variants, and more abundant cross-links with the I133C and W134C variants (Fig. 4). In contrast, the Pro-σ^K S21 side chain appeared to point away from the catalytic E44 residue of SpoIVFB (42), and we confirmed that the S21C Pro-σ^K(1–127) variant exhibits very little cross-linking to the E44C cytTM-SpoIVFB variant (Fig. 4). During catalysis, E44 within the HEXXH motif is thought to activate a zinc-bound water molecule for hydrolysis of the peptide bond between S21 and Y22 of Pro-σ^K (37, 38, 40). Unlike the S21 side chain, the Pro-σ^K Y22 side chain appears to be in proximity to both E44 (42) and the predicted short loop region of SpoIVFB (Fig. 4).

Based on the apparent preferred orientation of the Pro-σ^K S21 side chain (42) and the preference of SpoIVFB for a residue with a small side chain at position 21 of Pro-σ^K (39, 61), it has been proposed that SpoIVFB has a binding pocket for the residue at position 21 of the substrate, which ensures efficient hydrolysis of the correct peptide bond (42, 61). In a structural model of SpoIVFB in complex with Pro-σ^K, P132 and P135 constrain the predicted SpoIVFB short loop so that the side chains of W134, P135, and L136 form a hydrophobic face of the active-site region (43, 44). Conceivably, the hydrophobic face limits the side chain size at position 21 of the substrate, and the Ser of Pro-σ^K at that position is oriented by interactions with residues of the predicted short loop. Since an Ala substitution for W134 in the predicted short loop of SpoIVFB did not impair Pro-σ^K cleavage in E. coli (Fig. 2B) or sporulating B. subtilis (Fig. 6), perhaps the SpoIVFB W134A variant would cleave Pro-σ^K variants with a residue at position 21 having a larger side chain than tolerated by wild-type SpoIVFB. More work will be needed to test this prediction and more broadly to utilize the knowledge generated in this study to manipulate IMMP activity.

Taken together with the previous investigation of the E. coli RseP predicted short loop (50), our results suggest that a broadly conserved feature of IMMPs is the importance of short loop residues for substrate interaction and cleavage. Yet residues in the predicted short loop regions of SpoIVFB and RseP do not appear to function identically. The differences may provide opportunities to develop selective modulators of IMMPs for therapeutic and other purposes. Interestingly, RseP orthologs in several species confer sensitivity to certain bacteriocins (63, 64). In a well-studied case, the bacteriocin appears to interact with the RseP ortholog near the predicted short loop (65). Recently, a novel hybrid bacteriocin with greater activity against Staphylococcus haemolyticus was shown to use the RseP ortholog as a receptor and, in combination with other bacteriocins, was effective against biofilms (66). A better understanding of the functions of predicted short loop residues in substrate interaction and cleavage may lead to additional therapies that target IMMPs.

MATERIALS AND METHODS

Plasmids, primers, and strains.

Plasmids used in this study are described in Table S1 in the supplemental material. Primers used in plasmid construction are listed in Table S2 in the supplemental material. Plasmids were cloned into E. coli strain DH5α (67). Relevant parts of plasmids were verified by DNA sequencing with primers listed in Table S3 in the supplemental material. B. subtilis strains used in this study are described in Table S4 in the supplemental material.

Sequence analysis of SpoIVFB orthologs.

Orthologs of B. subtilis SpoIVFB were collected from the NCBI and UniProt databases by searching for genes annotated as spoIVFB in endospore-forming bacterial species (20). The protein sequences of SpoIVFB orthologs were aligned using the T-Coffee multiple-sequence alignment package (68). Residues identical in at least 70% of the sequences were considered conserved.

Pro-σ^K(1–127) cleavage in E. coli.

Strain BL21(DE3) (Novagen) was used to express proteins in E. coli. Plasmid transformants were selected on Luria-Bertani (LB) agar supplemented with kanamycin sulfate (50 μg/mL). Transformants (4 to 5 colonies) were grown in LB medium with 50 μg/mL kanamycin sulfate at 37°C with shaking (200 rpm). A culture grown overnight (200 μL) was transferred to 10 mL of LB medium with 50 ug/mL kanamycin sulfate, cultures were grown at 37°C with shaking (250 rpm) to an optical density of 60 to 80 Klett units, and isopropyl β-d-thiogalactopyranoside (IPTG) (0.5 mM) was added to induce protein production for 2 h. Equivalent amounts of cells (based on the optical density in Klett units) were collected (12,000 × g for 1 min), and extracts were prepared (29) and then subjected to immunoblot analysis.

Immunoblot analysis.

Samples were subjected to immunoblot analysis as described previously (69). Briefly, proteins were separated by SDS-PAGE using either 14% or 10% Prosieve polyacrylamide gels (Lonza) and electroblotted onto Immobilon-P membranes (Millipore). Protein migration was monitored using the SeeBlue Plus2 prestained standard (Invitrogen), and blots were blocked with 5% nonfat dry milk (Meijer) in Tris-buffered saline–Tween 20 (TBST) (20 mM Tris-HCl [pH 7.5], 0.5 M NaCl, 0.1% Tween 20) for 1 h at 25°C with shaking. Blots were probed with antibodies against His₆ (catalog number 34460; Qiagen) (1:10,000), FLAG₂ (catalog number A8592; Sigma) (1:10,000), Pro-σ^K (70) (1:3,000), SpoIVFA (69) (1:3,000), and/or SpoIVFB (38, 43) (1:5,000) diluted in TBST with 2% milk overnight at 4°C with shaking. Since the Pro-σ^K, SpoIVFA, and SpoIVFB antibodies were not horseradish peroxidase (HRP) conjugated, they were detected with goat anti-rabbit–HRP antibody (catalog number 170-6515; Bio-Rad) (1:10,000) diluted in TBST with 2% milk for 1 h at 25°C with shaking. Signals were generated using the Western Lightning Plus ECL reagent (PerkinElmer) and detected using a ChemiDoc MP imaging system (Bio-Rad). Unsaturated signals were quantified using the Image Lab 5.1 software (Bio-Rad) lane and bands tool in order to determine the Pro-σ^K(1–127) cleavage ratio.

Cobalt affinity purification.

E. coli BL21(DE3) was transformed with a plasmid, grown in LB medium (1 L), and induced with IPTG as described above. The culture was split, cells were harvested, and cell pellets were stored at −80°C. Each cell pellet was resuspended in 20 mL of lysis buffer containing phosphate-buffered saline (PBS) (pH 7.4), and cell lysates were prepared as described previously (60). Cell lysates were centrifuged (15,000 × g for 15 min at 4°C) to sediment cell debris. The supernatant was treated with 1% n-dodecyl-β-d-maltoside (DDM) (Anatrace) for 1 h at 4°C to solubilize membrane proteins and then centrifuged at 150,000 × g for 1 h at 4°C. The supernatant was designated the input sample, and 15 mL was mixed with imidazole (5 mM) and 0.5 mL of Talon Superflow metal affinity resin (Clontech) that had been equilibrated with buffer (PBS [pH 7.4], 0.1% DDM, 5 mM 2-mercaptoethanol, 10% glycerol). The mixture was rotated for 1 h at 4°C. The cobalt resin was sedimented by centrifugation at 708 × g for 2 min at 4°C, and the supernatant was saved (unbound sample). The resin was washed three times with 5 mL wash buffer (PBS [pH 7.4], 150 mM NaCl, 10% glycerol, 0.1% DDM, 20 mM imidazole), each time with gentle mixing for 1 min on a Vortexer II instrument (VWR) set at 4, and then sedimented as described above. The resin was mixed with 0.5 mL 2× sample buffer and boiled for 3 min (bound sample). A portion of the bound sample was diluted 15-fold (1/15 bound sample) with 1× sample buffer to match the concentration of the input sample. Samples were subjected to immunoblot analysis using 14% Prosieve polyacrylamide gels (Lonza).

Disulfide cross-linking.

As described above for Pro-σ^K(1–127) cleavage, E. coli BL21(DE3) was transformed with a plasmid, grown in LB medium (10 mL), and induced with IPTG, and equivalent amounts of cells were collected. Cells were mixed with chloramphenicol (200 μg/mL) and 2-phenanthroline (3 mM), collected by centrifugation (12,000 × g for 1 min), washed with 10 mM Tris-HCl (pH 8.1) containing 3 mM 2-phenanthroline, and suspended in 10 mM Tris-HCl (pH 8.1). Samples were treated with 1 mM Cu²⁺(phenanthroline)₃ or 3 mM 2-phenanthroline (as a negative control) for 60 min at 37°C, followed by incubation with neocuproine (12.5 mM) for 5 min at 37°C. Cells were lysed, and proteins were precipitated by the addition of trichloroacetic acid (5%) and inversion every 5 min for 30 min on ice. Proteins were sedimented by centrifugation (12,000 × g) for 15 min at 4°C, the supernatant was removed, and the pellet was washed with cold acetone. The pellets were sedimented by centrifugation (12,000 × g) for 5 min at 4°C, and the supernatants were discarded. The pellets were dried for 5 min at 25°C and resuspended in buffer (100 mM Tris-HCl [pH 7.5], 1.5% SDS, 5 mM EDTA, 25 mM N-ethylmaleimide) for 30 min at 25°C. Portions were mixed with an equal volume of sample buffer (25 mM Tris-HCl [pH 6.8], 2% SDS, 10% glycerol, 0.015% bromophenol blue) with or without 100 mM DTT and incubated at 37°C for 10 min prior to immunoblot analysis using 10% Prosieve polyacrylamide gels (Lonza).

B. subtilis sporulation.

Wild-type B. subtilis strain PY79 and its derivative BSL51 in which the spoIVFAB operon was replaced with a chloramphenicol resistance gene served as controls. Plasmids bearing the wild-type spoIVF operon or a mutant version, bordered by regions of homology to B. subtilis amyE, were transformed into strain BSL51. Transformants with a gene replacement at amyE were selected on LB agar with spectinomycin sulfate (100 μg/mL) and identified by the loss of amylase activity (71). Sporulation was induced by growing cells in the absence of antibiotics and resuspending the cells in SM medium (71). At the indicated times PS, equivalent amounts of cells (based on the optical density in Klett units) were centrifuged (12,000 × g for 1 min), supernatants were removed, and cell pellets were stored at −80°C. Whole-cell extracts were prepared as described previously for E. coli (29), except that samples were incubated at 50°C for 3 min instead of boiling for 3 min (43), and proteins were subjected to immunoblot analysis using 14% Prosieve polyacrylamide gels (Lonza).