The Prosequence of Pro-ς^K Promotes Membrane Association and Inhibits RNA Polymerase Core Binding

Abstract

Pro-ς^K is the inactive precursor of ς^K, a mother cell-specific sigma factor responsible for the transcription of late sporulation genes of Bacillus subtilis. Upon subcellular fractionation, the majority of the pro-ς^K was present in the membrane fraction. The rest of the pro-ς^K was in a large complex that did not contain RNA polymerase core subunits. In contrast, the majority of the ς^K was associated with core RNA polymerase. Virtually identical fractionation properties were observed when pro-ς^E was analyzed. Pro-ς^K was completely solubilized from the membrane fraction and the large complex by Triton X-100 and was partially solubilized from the membrane fraction by NaCl and KSCN. The membrane association of pro-ς^K did not require spoIVF gene products, which appear to be located in the mother cell membrane that surrounds the forespore, and govern pro-ς^K processing in the mother cell. Furthermore, pro-ς^K associated with the membrane when overproduced in vegetative cells. Overproduction of pro-ς^K in sporulating cells resulted in more pro-ς^K in the membrane fraction. In agreement with the results of cell fractionation experiments, immunofluorescence microscopy showed that pro-ς^K was localized to the mother cell membranes that surround the mother cell and the forespore in sporulating wild-type cells and mutant cells that do not process pro-ς^K. Treatment of extracts with 0.6 M KCl appeared to free most of the pro-ς^K and ς^K from other cell constituents. After salt removal, ς^K, but not pro-ς^K, reassociated with exogenous core RNA polymerase to form holoenzyme. These results suggest that the prosequence inhibits RNA polymerase core binding and targets pro-ς^K to the membrane, where it may interact with the processing machinery.

Endospore formation in the gram-positive bacterium Bacillus subtilis involves the formation of two cellular compartments of unequal size. The two compartments, namely, the mother cell and the forespore, are generated by the asymmetric positioning of a septum. The smaller forespore compartment is later engulfed inside the mother cell through a phagocytosis-like process. The mother cell nurtures the forespore during sporulation and is discarded by lysis upon maturation of the endospore. Gene expression in the two compartments is driven by a cascade of ς factors, namely, ς^F, ς^E, ς^G, and ς^K, in order of their activity (8, 11, 20, 24). The forespore-specific program of gene expression is controlled by ς^F and ς^G, while the mother cell program is controlled by ς^E and ς^K. Each sigma factor is initially inactive. ς^F is the first to become active, and this activation occurs only in the forespore. Activation of subsequent sigma factors in the cascade is triggered by signal transduction between the two compartments. The inactive forms of the mother cell-specific sigma factors are precursor proteins called pro-ς^E and pro-ς^K. Each is synthesized about 1 h before it is activated by proteolysis (3, 22, 26).

The processing of mother cell-specific ς factors is controlled by signals from the forespore. The putative processing enzyme for the conversion of pro-ς^E to ς^E is SpoIIGA (17, 35), which receives a signal from a protein, SpoIIR, generated in the forespore under the control of ς^F (14, 19, 23). Conversion of pro-ς^K to ς^K requires SpoIVFB (3, 4, 26), which is either the processing enzyme or its regulator, and is negatively regulated by SpoIVFA and BofA (3, 4, 15, 33). SpoIVFA, SpoIVFB, and BofA appears to be integral membrane proteins (4, 32, 33), and SpoIVFA and SpoIVFB have been shown to be localized at the boundary between the mother cell and the forespore (32). Activation of SpoIVFB for pro-ς^K processing requires the production of SpoIVB under the control of ς^G in the forespore (2, 10). SpoIVB is inferred to be a secreted protein and is presumed to overcome the inhibitory effects of SpoIVFA and BofA (2, 37).

Pro-ς^K has 20 amino acid residues at its N terminus which must be removed to generate active ς^K (21, 26, 36). Two lines of evidence indicate that pro-ς^K is transcriptionally inactive (26). First, expression of ς^K-dependent gene fusions does not begin until processing occurs. Second, when added to core RNA polymerase (RNAP), pro-ς^K fails to direct transcription from ς^K-dependent promoters in vitro. The role of the prosequence in preventing transcription is not clear. One function of the prosequence may be to mask the DNA-binding activity of ς^K, since the affinity binding constant of purified pro-ς^K for promoter DNA is 1 order of magnitude lower than that of ς^K (5). The results presented here suggest additional functions of the prosequence. We show that the majority of pro-ς^K is membrane-associated in cell extracts and is not associated with the core subunits of RNAP. In agreement with this observation, we find that pro-ς^K immunolocalizes to the mother cell membranes that surround the mother cell and the forespore in sporulating cells. Moreover, pro-ς^K fails to bind to core RNAP in vitro under conditions that permit ς^K binding. These results suggest that two more functions of the prosequence of pro-ς^K are to inhibit RNAP core binding and to promote association with the membrane, where processing may occur.

MATERIALS AND METHODS

General methods.

Sporulation was induced by resuspending growing cells in SM medium as described previously (12). The onset of sporulation (T₀) is defined as the time of resuspension. The B. subtilis strains used in this study are PY79 (wild type) (38) and its derivatives PY79/pSL1 (26), BSL51 (spoIVFΔAB::cat) (27), RL87 (spoIVFB152) (3), and RL136 (spoIVFB152 spoIVCBΔ19) (3). BK183 (spoIVA67 trpC2) is an isolate of 67 (9). RL831 (spoIIIGΔ::neo) is an isolate of RS242 (34).

Western blot analysis.

Samples of different fractions equivalent to the same original volume of culture or containing the same amount of protein, as determined by the Bradford method (1), were separated on sodium dodecyl sulfate (SDS)–12% Prosieve polyacrylamide gels (FMC) with Tris-Tricine electrode buffer (0.1 M Tris, 0.1 M Tricine, 0.1% SDS) and electroblotted to Immobilon-P membranes (Millipore). The membrane was probed with either polyclonal anti-pro-ς^K antiserum (26), anti-FtsH antiserum (a gift from S. Cutting and T. Ogura), anti-Escherichia coli core RNAP antiserum (a gift from M. Chamberlin and C. Kane), or monoclonal anti-ς^E antibody (a gift from W. Haldenwang). In some experiments, the membrane was stripped and reprobed with a different antibody. Horseradish peroxidase-conjugated secondary antibody was either goat anti-rabbit immunoglobulin G or goat anti-mouse immunoglobulin G (Bio-Rad). Chemiluminescence detection was performed following the manufacturer’s instructions (ECL; Amersham).

Column chromatography.

Minicolumns (5.5 by 70 mm) were made from Pasteur pipettes with the narrow end cut off and sealed with glass fiber and beads. Three types of gel filtration media were used: Sephacryl S-300, Sephadex G-200, and Sephadex G-100 (Pharmacia). The flow rate was controlled by gravity and ranged from 50 to 80 μl/min. The void volume and fractionation range were determined by passing various combinations of dextran blue, alcohol dehydrogenase (150 kDa), bovine serum albumin (66 kDa), carbonic anhydrase (29 kDa) and cytochrome c (12.4 kDa) through the columns. Usually a 100-μl sample was loaded and eluted with the same buffer, and 120-μl fractions were collected. Salt- or detergent-treated fractions were eluted with buffer adjusted to contain the same concentration of salt or detergent. If necessary, the column fractions were concentrated by trichloroacetic acid precipitation.

Subcellular fractionation.

Figure 1 is a diagram showing the fractionation scheme used in our experiments. Cells were collected by centrifugation (5,000 × g), washed with 1 M NaCl, and stored at −80°C. The cell pellet was resuspended in 7.5% the original volume of lysis buffer (25 mM HEPES-KOH [pH 7.5], 50 mM NaCl, 10 mM MgCl₂, 1 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol, 1 mg of lysozyme per ml, 0.1 mg of DNase I per ml, 20 μg of RNase A per ml, 1 mM phenylmethylsulfonyl fluoride [PMSF]) and incubated for 10 min at 37°C. Cells were then chilled and lysed by passage through a French pressure cell twice at 1,800 lb/in². The crude lysate was incubated at 37°C for 10 min. Cell debris was removed by centrifugation at 12,000 × g for 10 min. No nucleic acids were detected when the supernatant was analyzed by 2% agarose gel electrophoresis. The supernatant was then subjected to high-speed centrifugation (200,000 × g) for 1.5 h at 4°C. The pellet was homogenized in 1/5 the lysate volume of sucrose gradient buffer (25 mM HEPES-KOH [pH 7.5], 50 mM NaCl, 10 mM MgCl₂, 1 mM PMSF) plus 5% sucrose and loaded on top of a sucrose density gradient made with 2 ml of 55% (wt/vol) and 2 ml of 25% (wt/vol) sucrose in buffer in a 5-ml ultracentrifuge tube. After centrifugation at 200,000 × g for 4 h at 4°C, the membrane fraction was recovered at the interface between 25 and 55% sucrose. The supernatant (cytoplasmic fraction) after the initial high-speed centrifugation (100 μl) was loaded onto a gel filtration column and eluted with lysis buffer omitting the lysozyme, DNase I, and RNase A. Fractions of 120 μl were collected and analyzed by Western blotting.

FIG. 1.

Diagram of subcellular fractionation of sporulating B. subtilis cells. See Materials and Methods for details.

In the experiments testing the effects of salt and detergent, the supernatant after low-speed centrifugation was divided into six aliquots. Salt or detergent was added to different final concentrations, and one fraction was left untreated. All aliquots were kept for 20 min at 4°C and then subjected to high-speed centrifugation as noted above. The supernatant and pellet fractions were analyzed by Western blotting.

Immunofluorescence microscopy and image processing.

The affinity-purified rabbit polyclonal anti-ς^K antibodies (32) were a gift of O. Resnekov and were used at a 1:500 dilution. The secondary antibodies (Jackson Immunolabs) were affinity-purified donkey anti-rabbit antibodies conjugated to fluorescein isothiocyanate (FITC) and were used at a 1:100 dilution. DNA was stained with propidium iodide (PI; Molecular Probes) at a final concentration of 10 μg/ml. Cells were harvested 2.5 and 3.5 h after the onset of sporulation. Immunofluorescence experiments were performed as described by Pogliano et al. (31). PI and FITC images of identical fields of cells were recorded with a cooled charge-coupled device camera (Princeton Instruments) and a personal computer with the MetaMorph imaging system (version 3.0; Universal Imaging Corp.). PI images were assigned to the red channel, and FITC images were assigned to the green channel. Adobe Photoshop (version 3.0.5) was used to overlay FITC images on PI images of identical fields.

In vitro reconstitution of RNAP holoenzyme.

B. subtilis core RNAP was partially purified as described previously (21). To isolate pro-ς^K and ς^K, PY79/pSL1 cells were collected at 4.5 h into sporulation without IPTG induction. Cells from 3 ml of culture were pelleted by centrifugation at 5,000 × g for 5 min, resuspended in 100 μl of lysis buffer with KCl instead of NaCl, and incubated at 37°C for 10 min. The KCl concentration was adjusted to 0.6 M, and the lysate was sonicated. After 10 min at 30°C, the lysate was cleared of unlysed cell debris by centrifugation for 10 min in a microcentrifuge. The supernatant (120 μl) was loaded onto a Sephadex G-100 column and eluted with lysis buffer containing 0.6 M KCl without enzymes. Fractions in the molecular weight range of monomeric pro-ς^K were pooled and dialyzed against lysis buffer to remove the salt. The dialyzed sample was divided into two 100-μl aliquots, each containing approximately 5 pmol of pro-ς^K and 1 pmol of ς^K, as determined by Western blotting. Ten microliters of core RNAP in storage buffer, containing approximately 15 pmol of core subunits, as determined by SDS-PAGE and Coomassie blue staining, was added to one aliquot, and 10 μl of storage buffer was added to the second aliquot. The two aliquots were incubated on ice for 1 h. Each aliquot was then fractionated on the same Sephadex G-100 column. Column fractions were precipitated with trichloroacetic acid and analyzed by Western blotting.

RESULTS

The majority of pro-ς^K is membrane associated.

To investigate whether pro-ς^K is associated with core RNAP, we fractionated crude lysates of sporulating wild-type B. subtilis as diagrammed in Fig. 1. To facilitate the comparison of pro-ς^K and ς^K, cells were collected at 3.5 h after the onset of sporulation (T_3.5), when approximately equal amounts of pro-ς^K and ς^K are present in cells. Cells were treated with lysozyme and lysed by passage through a French pressure cell. The crude lysate was cleared of cell debris by low-speed centrifugation (12,000 × g), and the supernatant was then subjected to high-speed centrifugation (200,000 × g). The resulting pellet was further fractionated on a sucrose density gradient. Samples of different fractions were analyzed by Western blotting using anti-pro-ς^K antibodies (26). As shown in Fig. 2A, the majority of pro-ς^K was detected in the high-speed pellet (lane 3), while ς^K was predominantly present in the high-speed supernatant (cytoplasmic fraction) (lane 2). After further fractionation of the high-speed pellet on a sucrose density gradient, pro-ς^K remained in the membrane fraction (lane 4), whereas the small amount of ς^K in the sample formed a pellet at the bottom of the sucrose gradient tube (data not shown), suggesting that it was present in residual cell debris or in a large aggregate of proteins. The cytoplasmic fraction was apparently depleted of membrane vesicles, as FstH, an integral membrane protein, was not detected (lane 2). All the FtsH was found in the initial high-speed pellet (lane 3) and was recovered in the purified membrane fraction (lane 4). The purified membrane fraction was essentially free of core RNAP, as little β and β′ subunits were detected (lane 4). These results show that the majority of the pro-ς^K in the crude lysate, unlike ς^K, is not associated with core RNAP but is membrane associated.

FIG. 2.

Subcellular fractionation of extracts of sporulating wild-type cells. Cell extracts were fractionated as diagrammed in Fig. 1. Proteins in different fractions were subjected to SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting. (A) Samples of different fractions equivalent to the same original volume of wild-type T_3.5 culture were analyzed for pro-ς^K and ς^K, as well as FtsH and the β and β′ subunits of RNAP by Western blotting. Lane 1, supernatant after centrifugation at 12,000 × g; lane 2, supernatant after centrifugation at 200,000 × g; lane 3, pellet after centrifugation at 200,000 × g; lane 4, membrane fraction purified by sucrose density gradient. (B) The supernatant after centrifugation at 200,000 × g was subjected to size fractionation by passage through a Sephacryl S-300 column. Equal volumes of the column fractions were analyzed for pro-ς^K and ς^K and the RNAP β and β′ subunits. Fraction numbers are indicated over the blots. (C) Samples of different fractions equivalent to the same original volume of wild-type T_1.7 culture were analyzed for pro-ς^E and ς^E. Lane contents are the same as for panel A. (D) The blot that had been probed with anti-pro-ς^K antiserum in panel B was stripped and reprobed with anti-ς^E antibody.

To ask whether pro-ς^K in the cytoplasmic fraction was associated with core RNAP, the supernatant after high-speed centrifugation was size fractionated by passage through a Sephacryl S-300 column, which has a fractionation range of 10 to 1,500 kDa. The molecular mass of ς^K RNAP holoenzyme is about 370 kDa, which should render it readily separated from very high molecular mass complexes and from free pro-ς^K (29 kDa) in this column. The column fractions were analyzed by Western blotting using anti-pro-ς^K antibodies or antibodies against E. coli core RNAP. As shown in Fig. 2B, most of the pro-ς^K was eluted at or near the void volume of the column (lane 1), suggesting that it is part of a very large complex (>1,500 kDa). ς^K was eluted later than pro-ς^K and coeluted with the β and β′ subunits of RNAP (lanes 2 and 3), indicating that ς^K was present in the holoenzyme form. Taken together, these results show that pro-ς^K is not associated with core RNAP in the crude extract of sporulating B. subtilis; rather, most of it is associated with membrane, and the rest is present in a large complex of unknown composition.

We next asked whether pro-ς^E fractionates in the same way as pro-ς^K. Pro-ς^E is the inactive precursor of ς^E. Since pro-ς^E is synthesized earlier than pro-ς^K, wild-type cells were collected at 1 h and 40 min after the onset of sporulation, when approximately equal amounts of pro-ς^E and ς^E are present. Cell extracts were prepared and fractionated as described above. Pro-ς^E and ς^E were analyzed by Western blotting using monoclonal anti-ς^E antibody. As shown in Fig. 2C, pro-ς^E fractionated in a pattern similar to that of pro-ς^K. The majority of pro-ς^E was detected in the high-speed pellet (lane 3) and was recovered in the purified membrane fraction (lane 4). There was a small amount of pro-ς^E in the cytoplasmic fraction of cells collected at T_1.7 (lane 2) or T_3.5 (data not shown). To determine whether pro-ς^E in the cytoplasmic fraction was associated with core RNAP, the blot shown in Fig. 2B was stripped of bound antibodies and reprobed with anti-ς^E antibody. Like pro-ς^K (Fig. 2B, lane 1), pro-ς^E was eluted in the void volume of the Sephacryl S-300 column (Fig. 2D, lane 1), suggesting that it is part of a very large complex (>1,500 kDa) of unknown composition. ς^E was found almost exclusively in the cytoplasmic fraction of cells collected at T_1.7 (Fig. 2C, lane 2) or T_3.5 (data not shown). Like ς^K (Fig. 2B, lanes 2 and 3), some of the ς^E (Fig. 2D, lanes 2 and 3) coeluted with core RNAP (Fig. 2B, lanes 2 and 3) from the sizing column, but unlike ς^K (Fig. 2B), much of the ς^E eluted as free ς^E (Fig. 2D, lanes 5 and 6).

Effects of detergent and salt treatment on the membrane association of pro-ς^K.

After the lysate of wild-type cells collected at T_3.5 was cleared of cell debris by low-speed centrifugation (12,000 × g), the supernatant was treated with detergent or salt and then subjected to high-speed centrifugation (200,000 × g). The resulting supernatant and pellet fractions were analyzed by Western blotting to further characterize the membrane association of pro-ς^K. As expected for a protein interacting with membranes, pro-ς^K was solubilized by 1% Triton X-100 treatment (Fig. 3A). In contrast, the small amount of ς^K found in the high-speed pellet remained in the pellet upon detergent treatment (Fig. 3A). This result is consistent with the finding that ς^K in the pellet did not fractionate with membrane in a sucrose gradient and supports the idea that ς^K is not membrane associated. Instead, we speculate that it may be associated with residual cell debris or a large aggregate of proteins.

FIG. 3.

Effects of detergent and salt treatment on fractionation of pro-ς^K and ς^K. (A) Crude cell extract was cleared of cell debris by centrifugation at 12,000 × g. The supernatant (lane S 12,000 g) was divided into six aliquots and treated with either 1% Triton X-100, 0.5 M NaCl, 1 M NaCl, 0.25 M KSCN, 0.5 M KSCN, or left untreated. These aliquots were then subjected to high-speed centrifugation (90 min, 200,000 × g). Samples of the supernatant (S) and pellet (P) fractions equivalent to the same original volume of wild-type T_3.5 culture were analyzed by Western blotting using anti-pro-ς^K antibodies. (B) The supernatant after centrifugation at 12,000 × g was treated with 1% Triton X-100 (lane L) and size fractionated by passage through a Sephadex G-200 column (lanes 1 through 11). Equal volumes of the column fractions were analyzed for pro-ς^K and ς^K by Western blotting. The numbers over the blot are the column fractions. Fractions 1 and 2 contained materials eluted in the void volume of the column.

To determine the sizes of pro-ς^K and ς^K in the Triton X-100-treated supernatant, instead of subjecting it to high-speed centrifugation, it was size fractionated by passage through a Sephadex G-200 column, which has a fractionation range of 5 to 600 kDa. Figure 3B shows that the majority of ς^K was eluted near the void volume of this column, which is the expected result if ς^K was not dissociated from core RNAP by 1% Triton X-100. However, we cannot rule out the possibility that ς^K dissociated from core RNAP and formed a large aggregate. In contrast, pro-ς^K was eluted in the included volume, indicating that pro-ς^K was dissociated from membranes. In addition, the large complex that had remained in the supernatant of extracts not treated with detergent (Fig. 2B, lane 1) appeared to be dissociated by Triton X-100, suggesting that the interactions of pro-ς^K in the large complex are primarily hydrophobic in nature.

A nonchaotropic salt (NaCl) partially solubilized pro-ς^K from the membrane and a chaotropic salt (KSCN) appeared to be slightly more effective at solubilizing pro-ς^K (Fig. 3A). These results show that ionic interactions are involved in the binding of pro-ς^K to the membrane and suggest that hydrophobic interactions may also be involved. The pro-ς^K remaining in the pellet after treatment with high concentrations of salt may be present inside vesicles and therefore incapable of release by salt. In contrast, both 0.5 M NaCl and 0.25 M KCNS completely solubilized the residual ς^K from the pellet.

Membrane association of pro-ς^K does not depend upon sporulation-specific gene products.

The products of the mother cell-expressed spoIVF operon are thought to be intimately involved in the processing of pro-ς^K. SpoIVFB is either the processing enzyme or a regulator of the processing enzyme (3, 4, 25, 26). SpoIVFA negatively regulates the activity of SpoIVFB, and these proteins are thought to form a complex in the mother cell membrane that surrounds the forespore (3, 4, 32). To investigate whether spoIVF gene products are required for the membrane association of pro-ς^K, a lysate from spoIVF null mutant cells was fractionated and analyzed by Western blotting. As shown in Fig. 4A (lanes 1 to 3), the majority of pro-ς^K was present in the pellet after high-speed centrifugation, just as in wild-type cells (Fig. 2A, lanes 1 to 3). Since spoIVF null mutant cells are processing deficient, only pro-ς^K is present. We conclude that the membrane association of pro-ς^K does not depend upon spoIVF gene products. Another mother cell-specific protein, SpoIVA, is located at the forespore surface and controls the assembly of the spore cortex and coat (7). Pro-ς^K processing is impaired in spoIVA mutant cells (26). We found that pro-ς^K associates with the membrane in spoIVA mutant (spoIVA67) cells (data not shown), suggesting that the spoIVA mutation does not impair the processing of pro-ς^K by interfering with its membrane association.

FIG. 4.

Specificity of the membrane association of pro-ς^K. (A) Sporulating (T_3.5) spoIVF null mutant (BSL51) cells and vegetative wild-type cells expressing pro-ς^K from a plasmid (pSL1) were fractionated as diagrammed in Fig. 1. Proteins equivalent to the same original volume of cells were analyzed by Western blotting. Lanes 1 to 3, supernatant after centrifugation at 12,000 × g and supernatant and pellet after centrifugation at 200,000 × g, respectively, of sporulating BSL51 cells. Lanes 4 to 7, supernatant after centrifugation at 12,000 × g, supernatant and pellet after centrifugation at 200,000 × g, and gradient-purified membrane, respectively, of vegetative PY79/pSL1 cells. (B) Western blot analysis of 2 μg of protein from sucrose gradient-purified membrane of sporulating (T_3.5) wild-type (PY79) cells (lane 1) and sporulating (T_3.5) wild-type cells containing plasmid pSL1 after being induced to make pro-ς^K for 10 min (lane 2), 30 min (lane 3), or 3.5 h (lane 4).

To determine whether membrane association of pro-ς^K occurs in the absence of any sporulation-specific gene products, we induced production of pro-ς^K during vegetative growth from a multicopy plasmid, pSL1, which has the intact sigK gene fused to an IPTG-inducible promoter (P_spac) (26). A lysate was prepared from IPTG-induced cells containing pSL1, fractionated, and analyzed by Western blotting. About half of the pro-ς^K was pelleted by high-speed centrifugation, and it remained in the membrane fraction after sucrose gradient purification (Fig. 4A, lanes 4 to 7). Hence, membrane association of pro-ς^K does not require expression of any sporulation-specific genes. In this experiment, about half of the pro-ς^K remained in the supernatant after high-speed centrifugation (Fig. 4A, lane 5), whereas in sporulating cells, only a small amount of pro-ς^K was found in the cytoplasmic fraction (Fig. 2A, lane 2). The difference may be due to overproduction of pro-ς^K in IPTG-induced vegetative cells containing pSL1. Perhaps the amount of pro-ς^K produced exceeds the number of membrane binding sites; however, we cannot rule out the possibility that a portion of the pro-ς^K is incapable of membrane binding when it is overproduced. Like the pro-ς^K in the cytoplasmic fraction of sporulating cells (Fig. 2B, lane 1), the pro-ς^K in the cytoplasmic fraction of the vegetative cells appeared to be present in a large complex (>1,500 kDa) of unknown composition (data not shown).

Pro-ς^K binding sites are not saturated on the membranes of sporulating cells.

To test whether membranes in sporulating cells have the ability to bind more pro-ς^K, we induced the production of pro-ς^K from pSL1 during sporulation. Wild-type cells bearing pSL1 were induced with IPTG for 10 min, 30 min, or 3.5 h before being harvested at T_3.5. Membrane fractions from these cells were purified by sucrose gradients. Two micrograms of protein from each membrane preparation was analyzed by Western blotting. As shown in Fig. 4B, more pro-ς^K was detected in membranes prepared from cells overproducing pro-ς^K than in membranes prepared from wild-type cells. These results indicate that pro-ς^K binding sites are not saturated on the membranes of sporulating cells when pro-ς^K is produced at the wild-type level. The binding sites may become saturated, though, when pro-ς^K is overproduced in sporulating cells, because more pro-ς^K was found in the cytoplasmic fraction of these cells (data not shown), just as in vegetative cells overproducing pro-ς^K (Fig. 4A, lane 5). However, as noted above, we do not know whether all of the pro-ς^K is capable of membrane binding.

Pro-ς^K localizes to the mother cell membranes that surround the forespore and the mother cell of the postengulfment sporangium.

Pro-ς^K and ς^K were immunolocalized in sporulating cells by using affinity-purified anti-ς^K antibodies (32), secondary antibodies coupled to FITC, and fluorescence microscopy. The rabbit polyclonal anti-ς^K antibodies visualize pro-ς^K as well as ς^K. Therefore, we were able to distinguish both forms of the transcription factor only by costaining the nucleoids to determine the stage of sporulation and by analyzing mutants that are either deficient in pro-ς^K processing or are known to synthesize mature ς^K in the absence of processing. Postengulfment sporangia at stages III and IV in sporulation can be readily identified by their DNA staining pattern. Whereas the forespore chromosome of stage III sporangia (Fig. 5A₁, red, arrow) more closely resembles the mother cell chromosome, albeit slightly more condensed, the forespore nucleoid of stage IV sporangia assumes a characteristic toroidal structure (Fig. 5B₁, red, arrow) upon association with the α/β-type SASPs (small acid-soluble proteins) (30).

FIG. 5.

Immunolocalization of pro-ς^K and ς^K in sporulating cells. The sporangia were harvested at T_2.5 in panels A and at T_3.5 in panels B to E and prepared for immunofluorescence microscopy as described in Materials and Methods. The arrows point to the engulfed forespore compartment and are oriented perpendicularly to the long axis of the sporangia. DNA was stained with PI (red) (A₁, B₁, C₁, D₁, and E₁). Immunostaining of pro-ς^K and ς^K is shown in green (A₂, B₂, C₂, D₂, and E₂). Where the red and green fluorophores overlap, as in the doubly exposed images shown in panels B₃ and E₃, a yellow-orange color is visible. (A) Wild-type (WT) sporangia with almost equally condensed mother cell and forespore nucleoids, which is characteristic of cells at stage III in sporulation before pro-ς^K is processed to ς^K. (B) Wild-type sporangia at stage IV in sporulation, when the forespore nucleoid has assumed its toroidal shape and pro-ς^K has been processed to ς^K in the mother cell. (C) Pro-ς^K processing-deficient spoIIIGΔ::neo mutant sporangia of strain RL831. (D) Processing deficient spoIVFB152 mutant sporangia of strain RL87. (E) spoIVFB152 spoIVCBΔ19 doubly mutant sporangia of strain RL136, which synthesize mature ς^K in the absence of a functional protease for pro-ς^K processing.

Wild-type postengulfment sporangia, which could be assigned to stage III in sporulation by virtue of their DNA staining pattern (Fig. 5A₁, red), displayed pro-ς^K/ς^K immunostaining (Fig. 5A₂, green) in the periphery of most of the mother cell and on one side of the engulfed forespore. This pattern was observed in 81% of 200 sporangia examined (4% showed pro-ς^K/ς^K immunostaining throughout the mother cell cytoplasm and 15% showed no staining). The peripheral forespore staining was often evident as a crescent at the interface between the forespore and the larger volume of the mother cell. There was very little overlap between the green immunostaining of pro-ς^K/ς^K and the red nucleoid staining (Fig. 5A₃), indicating that pro-ς^K/ς^K is associated with the mother cell membranes that surround the mother cell and the forespore. After sporulation proceeded to stage IV, as indicated by the toroidal forespore nucleoid (Fig. 5B₁, red, arrow), when pro-ς^K is known to be converted to ς^K in the mother cell (26), the pattern of pro-ς^K/ς^K immunostaining changed to include the cytoplasm of the mother cell (Fig. 5B₂, green). This staining pattern, observed in 96% of 154 sporangia examined (4% showed no staining), was consistent with the previously reported even distribution of ς^K throughout the mother cell (32).

In spoIIIG and spoIVFB mutant sporangia, which are deficient in pro-ς^K processing and do not proceed in development beyond stages III and IV, respectively, pro-ς^K immunostaining was detected in the periphery of the mother cell and forespore (Fig. 5C and D, green). The pattern shown in Fig. 5C₂ (green) was observed in 83% of 173 spoIIIG mutant sporangia examined, and the pattern shown in Fig. 5D₂ (green) was observed in 88% of 181 spoIVFB mutant sporangia examined (in both cases, 3 to 5% showed staining throughout the mother cell cytoplasm and 9 to 12% showed no staining). Because the peripheral staining pattern in these processing defective mutants was similar to the one observed in wild-type stage III sporangia (Fig. 5A, green), we conclude that pro-ς^K is associated with the mother cell membranes that surround the mother cell and the forespore. In sporangia of a spoIVFB mutant that produces mature ς^K without processing due to a deletion (spoIVCBΔ19) in the prosequence-encoding portion of sigK, immunostaining of ς^K was detected throughout the mother cell (Fig. 5E, green). This staining pattern, observed in 84% of 217 sporangia examined (16% showed no staining), is reminiscent of the one observed in wild-type stage IV sporangia (Fig. 5B, green). We infer that after proteolytic activation, ς^K is released from the membrane and becomes soluble in the mother cell cytoplasm. We conclude that the change in pro-ς^K/ς^K immunostaining from stage III (Fig. 5A, green) to stage IV (Fig. 5B, green) during sporulation of wild-type cells resulted from the conversion of membrane-associated pro-ς^K to soluble ς^K, consistent with our subcellular fractionation results (Fig. 2).

Pro-ς^K does not bind to exogenous core RNAP in vitro.

Very little, if any, of the pro-ς^K in lysates is associated with core RNAP (Fig. 2). Is this because pro-ς^K is unable to bind to core RNAP (due either to intrinsic inability to bind or to association of pro-ς^K with other cellular components like membranes) or because core RNAP is not available for binding? To address this question, our strategy was to dissociate both pro-ς^K and ς^K from other components in the cell lysates and incubate them with exogenous core RNAP. To increase the production of pro-ς^K and ς^K, we used wild-type cells containing pSL1. In the absence of IPTG induction, leaky expression from the P_spac promoter in pSL1 allows accumulation of pro-ς^K during sporulation so that when cells are harvested at T_4.5, when more ς^K has accumulated, both pro-ς^K and ς^K are present at a higher level than in wild-type cells at T_3.5.

A crude lysate was prepared from cells harvested at T_4.5, and KCl was added to a final concentration of 0.6 M. The salt-treated lysate was then size fractionated on a Sephadex G-100 column, which has a fractionation range of 4 to 150 kDa. Both pro-ς^K and ς^K in untreated crude lysate were excluded from this column (data not shown). After salt treatment, a portion of the pro-ς^K and ς^K was retained in the column (Fig. 6A), indicating that ς^K was partially dissociated from core RNAP and pro-ς^K was partially dissociated from the membrane and/or the large complex that remained in the supernatant after high-speed centrifugation (Fig. 2B, lane 1). Fractions 5 to 7 containing dissociated pro-ς^K and ς^K were pooled and dialyzed to remove the salt. The dialyzed sample was incubated with either partially purified core RNAP or with the core RNAP storage buffer and then fractionated in separate experiments on the same Sephadex G-100 column. Upon incubation with core RNAP (Fig. 6B), ς^K was eluted in the void volume, suggesting that it had reassociated with core RNAP. Pro-ς^K was eluted in the included volume after incubation with either core RNAP (Fig. 6B) or storage buffer (Fig. 6C). The same results were obtained when the experiment was repeated with a lysate made from wild-type cells harvested at T_3.5 (data not shown). These results indicate that even after pro-ς^K was dissociated from other cellular components, it did not bind to core RNAP under the conditions we used, whereas ς^K readily reassociated with core RNAP.

FIG. 6.

ς^K, but not pro-ς^K, reassociates with core RNAP after being dissociated by salt treatment. The supernatant after centrifugation at 12,000 × g was prepared in the presence of 0.6 M KCl and separated by a Sephadex G-100 column (A). The void volume of this column was fractions 1 and 2, wherein pro-ς^K and ς^K would be eluted if not treated with salt. Fractions 5 to 7 containing dissociated pro-ς^K and ς^K in approximately the monomeric size range were pooled and dialyzed. The dialyzed fractions were incubated with (B) and without (C) exogenous core RNAP. Proteins were then separated by the same Sephadex G-100 column and analyzed by Western blotting with anti-pro-ς^K antibodies. Only ς^K shifted back to the void volume upon incubation with core RNAP (panel B, lanes 1 and 2), indicating formation of the holoenzyme.

DISCUSSION

We have demonstrated that the majority of pro-ς^K in cell lysates is membrane associated and is not bound to core RNAP. In contrast, nearly all of the ς^K in lysates of sporulating cells is present in the cytoplasmic fraction and appears to be bound to core RNAP. In sporulating cells, pro-ς^K appears to associate with the mother cell membranes that surround the mother cell and the forespore, as visualized by immunofluorescence microscopy. Processing releases ς^K into the mother cell cytoplasm. Most of the pro-ς^K and ς^K can be dissociated from large components in the cell extract by 0.6 M KCl. After removal of the salt, ς^K, but not pro-ς^K, could bind to added core RNAP. These results indicate that the prosequence of pro-ς^K promotes membrane association and inhibits RNAP core binding.

The ability of pro-ς^K to associate with a membrane may facilitate its proteolytic processing to active ς^K. SpoIVFB has been proposed to be either the protease that processes pro-ς^K or a regulator of the protease (3, 4, 25, 26). Encoded in the same operon as SpoIVFB is SpoIVFA, which appears to inhibit SpoIVFB activity until a signal is received from the forespore (2–4). SpoIVFB and SpoIVFA have been shown to be localized at the boundary between the mother cell and the forespore (32). As depicted in Fig. 7, these proteins presumably insert into the mother cell membrane that surrounds the forespore, since the spoIVF operon is expressed in the mother cell (4). Likewise, bofA is thought to be expressed in the mother cell (33). Although BofA has not yet been shown to be localized to the mother cell membrane that surrounds the forespore, it has three putative transmembrane segments, and like SpoIVFA, it appears to inhibit SpoIVFB activity (3, 33). The signal that relieves this inhibition and leads to pro-ς^K processing is generated in the forespore by ς^G-dependent expression of spoIVB (2, 10) (Fig. 7). SpoIVB appears to have a signal sequence, so it may cross the forespore membrane in order to accomplish its signaling function (2, 37). If processing of pro-ς^K requires it to directly interact with SpoIVFB, then the ability of pro-ς^K to associate with the mother cell membrane that surrounds the forespore may facilitate processing by promoting this interaction.

FIG. 7.

Model depicting association of pro-ς^K with the mother cell membrane that surrounds the forespore and signal transduction between the forespore and the mother cell leading to the processing of pro-ς^K. The stippled oval represents a possible abundant membrane protein that interacts with pro-ς^K. See the text for details.

Our immunolocalization experiments showed that pro-ς^K interacts not only with the mother cell membrane that surrounds the forespore but also with the membrane that surrounds the mother cell in sporulating wild-type cells, as well as in spoIIIG and spoIVFB mutant cells (Fig. 5). Does the pro-ς^K associated with the membrane that surrounds the mother cell become processed? It appears that most, if not all, of the pro-ς^K produced in sporulating cells is processed to ς^K. First, very little pro-ς^K was detected late during sporulation (26). Second, a pulse-chase experiment demonstrated that the half-life of pro-ς^K is about 30 min. The majority of the ³⁵S label in pro-ς^K at T₃ was found in ς^K by T₄ (data not shown). Therefore, it seems likely that pro-ς^K associated with the membrane that surrounds the mother cell is either processed there or it dissociates and is processed elsewhere (e.g., at the mother cell membrane that surrounds the forespore). However, we cannot rule out the possibility that some of the pro-ς^K that associates with the membrane that surrounds the mother cell is degraded.

The prosequences of both mother cell-specific ς factors appear to promote membrane association. We found that the fractionation properties of pro-ς^E in lysates of sporulating cells were very similar to those of pro-ς^K (Fig. 2). The majority of pro-ς^E was membrane associated and not bound to core RNAP. ς^E, like ς^K, appeared to be associated with core RNAP in the cytoplasmic fraction. The prosequence of pro-ς^E has been proposed to form an amphipathic α helix with a highly charged face (28), which could presumably interact with negatively charged head groups of membrane lipids, but this would not explain the preferential localization of pro-ς^E to the sporulation septum (13, 18). The results of genetic suppression (29) and chemical cross-linking (13) studies suggest that pro-ς^E interacts with its putative processing enzyme, SpoIIGA. However, pro-ς^E may also interact with another protein in the septal membrane, since localization of pro-ς^E (13) and a pro-ς^E::GFP (green fluorescent protein) fusion protein (18) to the septal membrane occurs in spoIIGA mutant cells.

The 20-amino-acid prosequence of pro-ς^K is very hydrophobic, except for two charged residues at positions 13 and 14 from the N terminus (36). The charged residues might prevent the prosequence from inserting into the membrane like a transmembrane domain of a typical integral membrane protein. In support of this prediction, virtually all the pro-ς^K was found in the aqueous phase of a Triton X-114 fractionation experiment (data not shown). We speculate that pro-ς^K is peripherally associated with the membrane, perhaps via binding of the prosequence to an abundant integral membrane protein (Fig. 7), since the membranes in sporulating cells have the capacity to bind much more pro-ς^K when it is overproduced (Fig. 4B). Alternatively, it is possible that the prosequence of pro-ς^K does not interact directly with a membrane component. Removal of the prosequence could induce a conformational change that prevents membrane association and/or uncovers a site that gives ς^K a higher affinity for core RNAP than for the membrane. The interaction of pro-ς^K with membranes does not require spoIVF gene products (Fig. 4A and 5D) or the products of spoIVA (data not shown) or spoIIIG (Fig. 5C). Indeed, the interaction does not require any sporulation-specific gene products, since pro-ς^K produced in vegetative cells was membrane associated (Fig. 4A).

A small portion of the pro-ς^K and pro-ς^E in cell lysates was not membrane associated (Fig. 2A and C). Rather, the pro-ς factors appeared to be present in large complexes (>1,500 kDa) (Fig. 2B and D) of unknown composition. The large complexes could be aggregates of the pro-ς factors alone or in combination with other proteins. Different methods of cell breakage had little effect on the proportion of pro-ς^K that was membrane associated or present in a large complex. We tested sonication and osmotic shock lysis procedures (data not shown), in addition to the French pressure cell lysis method reported here.

In addition to promoting the membrane association of pro-ς^K, the prosequence also appears to inhibit RNAP core binding. The β and β′ subunits of core RNAP were barely detectable in a membrane fraction that contained abundant pro-ς^K (Fig. 2A, lane 4). Also, the pro-ς^K that was not membrane associated appeared to be present in a large complex (>1,500 kDa) containing very little β and β′ (Fig. 2B, lane 1). Moreover, much less pro-ς^K than ς^K bound to core RNAP after both had been released from large cellular components by salt treatment and the salt was removed by dialysis (Fig. 6B). We cannot rule out the possibility that pro-ς^K remained in small complexes with itself or another protein(s) upon treatment with 0.6 M KCl. However, the elution profile of pro-ς^K from a sizing column was similar to that of ς^K both in the presence of 0.6 M KCl (Fig. 6A) and after salt removal when core RNAP was not added (Fig. 6C). It seems unlikely that pro-ς^K was irreversibly denatured by 0.6 M KCl, since ς^K readily associated with core RNAP upon its addition (Fig. 6B). Under these conditions, the prosequence greatly hindered RNAP core binding. The prosequence may be close to the core-binding domain in the three-dimensional structure of pro-ς^K, directly blocking core binding. Alternatively, cleavage of the prosequence may result in a conformational change which activates core binding. In agreement with our findings, Johnson and Dombroski (16) recently demonstrated that purified ς^K competes much more efficiently than purified pro-ς^K for binding to E. coli core RNAP. Removal of only 6 amino acid residues from the N terminus of pro-ς^K restored core binding, and the holoenzyme was transcriptionally active.

ς⁷⁰ of E. coli does not bind to promoter DNA unless its amino-terminal region 1.1 is removed (6). Removal of the prosequence from pro-ς^K results in a 10-fold increase in DNA-binding activity (5). Our results suggest that in addition to modulating DNA-binding activity, the prosequence of pro-ς^K promotes its membrane association, perhaps facilitating processing to ς^K. Removal of the prosequence releases ς^K from the membrane and appears to unmask its RNAP core-binding activity, allowing the functional holoenzyme to form.