DivIVA-Mediated Polar Localization of ComN, a Posttranscriptional Regulator of Bacillus subtilis

All B. subtilis strains are listed in Table 1 and were derived from the wild-type strain PY79. B. subtilis was grown in Luria-Bertani (LB) medium at 37°C or 30°C. Preparation of competent cells and transformation of B. subtilis was performed according to the two-step procedure (16). When present, the antibiotic concentrations were as follows: chloramphenicol, 5 μg/ml; erythromycin plus lincomycin, 1 and 25 μg/ml; spectinomycin, 100 μg/ml; kanamycin, 10 μg/ml; and tetracycline, 10 μg/ml. When necessary, various concentrations of isopropyl-β-d-thiogalactopyranoside (IPTG) or xylose were used. The exact concentrations in each experiment are given in the figure legends.

Cloning and DNA manipulations were carried out according to standard methods (33). Taq DNA polymerase and Phusion DNA polymerase (NEB) were used for PCR. Oligonucleotide primers were purchased from IDT (Coralville, IA) and are listed in Table 2. The plasmids and their relevant characteristics are listed in Table 1. The steps involved in plasmid and strain construction are described in detail in the supplemental material. Sequencing of plasmids was carried out by the sequencing service of the Departamento de Bioquímica, IQ-USP.

Microscopy was performed on a Nikon TE300 inverted microscope equipped with a ×100 PlanFluor objective lens (oil, NA 1.3) and filters for green fluorescent protein (GFP; Endow GFP set, 41018; Chroma Technology) and mCherry (Texas Red BrightLine set, TXRED4040-B; Semrock). Images were captured with a Roper CoolSnap HQ camera. Exposure times varied from 0.3 to 1 s. Images were processed and analyzed with MetaMorph version 7.1 (Universal Imaging) and ImageJ (http://rsb.info.nih.gov/ij/) software. Microscopy of ComGA and colocalization between ComN-mCherry and GFP-DivIVA was performed on a Nikon Eclipse TiE microscope, with filters for GFP (GFP BrightLine Filter Set) and mCherry (mCherry BrightLine Filter Set), and a Plan APO VC Nikon ×100 objective lens (oil, NA 1.4). Images were captured with an Andor I-Xon EMCCD camera. The images were processed and analyzed with Nikon NIS Elements software (version 3.07) and ImageJ.

To perform microscopy, the cells were grown to exponential phase and mounted on slides covered with a layer of agarose-solidified minimal medium or phosphate-buffered saline (PBS). The membranes were stained with FM5-95 or FM4-64 (Molecular Probes) at a final concentration of 1 μg/ml. Dyes were added directly to cells in medium 5 min before mounting and imaging. For studying the localization of competence proteins, cells were imaged for 1.5 h following resuspension in SpII medium, using the slides and conditions described above.

Fusion proteins to the GAL4 DNA-binding domain (BD-protein) or GAL4 transcription activation domain (AD-protein) were expressed in the host strain PJ69-4a, which has two reporter genes for two-hybrid interactions integrated into the genome: HIS3 and ADE2. The method of transformation was that of Gietz et al. (9). PJ69-4a was transformed with combinations of two plasmids (pGBDU-divIVA and pGADC2-divIVA, pGADC2-comN and pGBDU-comN, pGADC2-comN and pGBDU, or pGBDU-comN and pGADC2) and plated in minimal medium (SC) lacking leucine and uracil as a first selection. Colonies that grew in the absence of Leu and Ura were transferred to SC plates lacking Ura, Leu, His, and Ade plus 3-aminotriazole (3AT). The amount of 3AT used varied between 5 and 50 mM. The pair pGBDU-divIVA and pGADC2-divIVA was used as a positive control. To test for self-activation, strains containing pGBDU-comN/pGADC2 and pGADC2-comN/pGBDU combinations were subjected to the same procedures.

All recombinant proteins were expressed in Escherichia coli BL21-CodonPlus (DE3) RIL, transformed with either pGEX-divIVA (GST-DivIVA), pGEX (GST), or pFG43-comN (His-ComN). E. coli cells bearing the different plasmids were grown at 37°C to an optical density at 600 nm (OD₆₀₀) of 0.6, when IPTG was added to a final concentration of 0.5 mM, and further incubated at 37°C for 4 h. Pulldown of His-ComN was assayed as follows. Whole-cell extracts from E. coli cells expressing GST, GST-DivIVA, and His-ComN were generated in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄ [pH 7.4]) and 1 mM phenylmethylsulfonyl fluoride (PMSF) after cellular lysis in a French press (three times) and centrifugation for 40 min at 4°C at 15,000 rpm. Then, 500 μl of glutathione-Sepharose beads (Amersham Biosciences) was incubated with the whole extract of GST and GST-DivIVA in shaking tubes for 2 h at 4°C. Next, the beads were washed three times with 12 ml of PBS and incubated with His-ComN extract for 4 h. Subsequently, the beads were washed three times with 12 ml of PBS, and the bound fraction was eluted with 1 ml of 10 mM reduced glutathione, 50 mM Tris-HCl (pH 8), and 1 mM PMSF. The bound fraction was precipitated with trichloroacetic acid (10%), followed by washing in 100% ethanol and resuspension in protein buffer. The proteins were resolved on SDS-PAGE (15%) and transferred to polyvinylidene difluoride membranes for 3 h at 200 mA. To perform Western blotting, the membranes were incubated with anti-polyhistidine serum (1:5,000; Amersham Biosciences) or with anti-GST serum (1:20,000; Sigma, St. Louis, MO). Immunoblots were developed using the ECL system (Amersham Biosciences).

Cells were grown in SpC medium to stationary phase and resuspended in SpII medium, and 1-ml fractions of the culture were collected every half hour for 2 h. For each time point, cell pellets were resuspended in 500 μl of Z buffer (60 mM Na₂HPO₄·7H₂O, 40 mM NaH₂PO₄·H₂O, 10 mM KCl, 1 mM MgSO₄ [pH 7.0]), and their OD₆₀₀s were recorded. Subsequently, 140 μl of each sample was added to the well of a 96 well-plate with 20 μl of Triton X-100 (2%), 3 μl of β-mercaptoethanol, and 40 μl of o-nitrophenyl-β-d-galactoside (4 mg/ml), and the OD₄₂₀s of the reactions were measured every minute for 40 min in a BioTek Epoch plate reader at room temperature. The measured OD₄₂₀ slopes (S) were introduced into the formula: Miller units = 1,000 × S/0.14 × OD₆₀₀.

Constructs for MS2-GFP expression and tagging of the comE mRNA are described in the supplemental material. Fresh colonies were grown in MC competence medium (100 mM potassium phosphate [pH 7.0], 3 mM sodium citrate, 2% glucose, 22 mg of ferric ammonium citrate/ml, 3 mM manganese sulfate, 0.1% casein, 0.2% potassium glutamate) for 2 h, and then IPTG was added to a final concentration of 50 μM to induce MS2d-GFP. Cultures were grown for 2 more hours before being harvested and visualized on the microscope. Membranes were stained with 0.02 mM TMA-DPH (Molecular Probes) and imaged with exposure times of 300 ms. MS2d-GFP fluorescence was captured using 500 ms of exposure time.

ComN (YrzD) was identified as a potential divisome protein because it displayed a pattern of co-occurrence with the division protein ZapA in the STRING database (37), which identifies functional links between proteins by comparing bacterial genomes. We tested this prediction by constructing a GFP-ComN fusion and determining its subcellular distribution. Fluorescence microscopy of a merodiploid strain expressing GFP-ComN from the P_xyl inducible promoter (FG916) revealed the typical localization pattern of division proteins, with clear accumulation of the protein in midcell bands (Fig. 1). GFP-ComN also seemed to be retained at the newly formed cell poles after septum completion (see arrows in Fig. 1), a pattern commonly seen with late divisome proteins. Localization of the GFP-ComN fusion was independent of the wild-type ComN protein because it also occurred in a strain in which the endogenous comN gene is deleted (FG1078). In addition, tagging of ComN with mCherry at its C terminus via single-crossover integration resulted in a strain that had ComN-mCherry expressed under the endogenous P_comN promoter (FG1254). In this strain, localization of ComN-mCherry was identical to that of the N-terminal GFP fusion expressed from the P_xyl promoter (see Fig. 2B). Thus, the observations that ComN localization is the same irrespective of the fluorescent protein tag (GFP or mCherry), topology (N- or C-terminal) or promoter (P_xyl or P_comN) from which the fusion is expressed suggest that the fluorescent protein fusions are faithfully reporting the localization of the untagged ComN protein.

To determine when ComN associates with the division site, we colocalized FtsZ and ComN using a strain simultaneously expressing FtsZ-mCherry and GFP-ComN (FG925). Only 40% of the Z rings in an exponentially growing population was decorated with GFP-ComN (Fig. 2A), suggesting that there is a significant delay between the initial assembly of the Z ring and when ComN becomes incorporated into the complex. The frequency of colocalization for ComN and FtsZ is similar to those measured for late divisome proteins such as DivIVA, YpsB, and MinCD (36; G. L. Meira and F. J. Gueiros-Filho, unpublished observations). Thus, ComN is likely a late recruit to the division complex.

The observations that ComN displays a pattern and kinetics of localization similar to that of DivIVA and the late division proteins that use DivIVA to reach the division site, such as MinCD and MinJ, prompted us to investigate whether ComN localization was dependent on one of these proteins. We initially investigated the pattern of GFP-ComN in a strain depleted of FtsZ and found that the protein was generally delocalized, as expected for a divisomal protein, but intense staining was still associated with the cell poles. This is very much the same pattern seen for DivIVA when FtsZ is depleted (25) and further suggested that ComN could be recruited to the divisome by DivIVA. Indeed, GFP-ComN localization was completely disrupted in a divIVA mutant. In these strains GFP-ComN was evenly distributed in the cytoplasm and, importantly, no sign of polar accumulation remained (Fig. 3). Because elimination of divIVA drastically reduces the frequency of division, we also analyzed the localization of GFP-ComN in a divIVA minD double mutant. Deletion of minD restores division to the divIVA mutant and allows for a clearer assessment of whether ComN requires DivIVA to reach the septum and cell poles. Significantly, GFP-ComN remained completely delocalized in the divIVA minD double mutant, despite the abundant septa made by this mutant. To rule out the possibility that localization of GFP-ComN was not mediated by DivIVA but instead due to the proteins recruited to the division site by DivIVA, we also analyzed the effect of deleting minD or minJ alone or of simultaneously deleting minD and minJ. In all of these cases, GFP-ComN still displayed the septal and polar localization pattern found in the wild-type strain (Fig. 3, panels minD, minJ, and minJ minD). We also tested the localization of GFP-ComN in a strain lacking ypsB (gpsB), a recently characterized paralog of divIVA that associates late with the divisome but independently of DivIVA (4), and in a strain lacking zapA, the division gene that retrieved comN in the STRING database co-occurrence search. However, localization of GFP-ComN was unaffected by either of the mutations (data not shown). Thus, GFP-ComN requires DivIVA, but not other late divisome proteins or ZapA to localize to the division site and cell poles.

We next used the GAL4 yeast two-hybrid system to investigate whether ComN and DivIVA interact directly. Pairing of ComN and DivIVA fusions in either of the two possible configurations (ComN-AD plus DivIVA-BD or DivIVA-AD plus ComN-BD) resulted in robust growth, indicating the existence of a direct interaction between the two proteins (Fig. 4A). The yeast two-hybrid experiments also showed that ComN is capable of interacting with itself (see Fig. S1 in the supplemental material), suggesting that the physiological state of ComN is an oligomer. To confirm the two-hybrid results, we carried out GST-pulldown experiments. DivIVA was fused to GST and ComN to a 6×His tag and overexpressed in E. coli. The total extract of the strain expressing His-ComN was applied onto a glutathione-agarose column previously loaded with GST-DivIVA. Subsequently, the column was washed and eluted with glutathione, the eluted material was electrophoresed, and Western blotting was performed with anti-His antibodies. As a control, a similar experiment was carried out with GST instead of GST-DivIVA. His-ComN was only detected in the eluate of the GST-DivIVA column, indicating that DivIVA and ComN physically interact in a specific way (Fig. 4B), corroborating the two-hybrid data. Thus, ComN likely interacts directly with DivIVA inside B. subtilis cells.

A central role of DivIVA in B. subtilis is targeting of the Min proteins to the cell poles (25). Because ComN interacts with DivIVA, ComN could play a role in the polar localization of the Min proteins. We tested this possibility by analyzing the cell division pattern in a comN deletion mutant and found that deletion of the comN gene had no detectable effect on either the frequency or the location of division septa (Fig. 5). Because alterations in MinCD localization lead to minicell formation and filamentation, the regular division pattern of the comN mutant indicates that ComN does not contribute to MinCD polar targeting. Several division proteins appear to have redundant functions and show little effect when deleted (10, 17, 36, 39). However, combining the comN mutation with mutations in other divisome components (zapA, noc, minD, and minJ) did not result in a cell division phenotype (data not shown), suggesting that ComN is not involved in the polar targeting of Min proteins or with any other aspect of cell division.

We next investigated whether the polar localization of ComN was important for its role in competence by measuring the development of competence in a divIVA mutant, where ComN is no longer polarly localized (Fig. 3). Because, as noted above, a divIVA mutant is severely impaired for division, and this mutation is detrimental for homologous recombination and thus to transformability (30), we opted to work with a divIVA minD double mutant, which divides much better than the divIVA single mutant and should not have the transformability problem caused by filamentation. Measuring the transformation efficiency revealed that, as expected, the comN mutant was 100-fold less competent than our wild-type strain [transformation frequencies of (2.6 ± 1.9) × 10⁻³ for the wild-type strain versus (2.5 ± 2.4) × 10⁻⁵ for the comN::cat strain], a result similar to that found by Ogura and Tanaka (28). In contrast, the transformation efficiency of the divIVA minD mutant was just 3-fold lower than that of the wild-type strain [(0.8 ± 1) × 10⁻³ versus (2.6 ± 1.9) × 10⁻³]. The low competence of the comN mutant can be explained by a drastic reduction in expression of the comE operon (22). Because comK and comE levels are normal in the divIVA minD mutant (see Fig. S2 in the supplemental material), the mild competence reduction of this mutant cannot be explained by alterations in competence gene expression. To determine whether the reduction in competence of the divIVA minD mutant was due to the delocalization of ComN or whether it was related to other factors such as the absence of MinD or the altered division behavior of this strain, we measured the transformation efficiency of a minD mutant, a strain that has the same division behavior as the divIVA minD mutant, but in which ComN still localizes properly (Fig. 3). The minD mutant exhibited a transformation efficiency similar to that of the wild-type [(4.0 ± 4.5) × 10⁻³] and higher than that of the divIVA minD mutant. This suggests that the delocalization of ComN does have a negative impact on competence development, although the large variability of the transformation measurements does not give statistical support to this conclusion. We therefore conclude that the polar localization of ComN is not essential but increases the efficiency of competence development. One potential caveat to this conclusion would be if DivIVA had an additional effect on competence besides localizing ComN. In this case, the difference in competence between the divIVA minD double mutant and the minD single mutant could not be ascribed solely to the delocalization of ComN. Even though we currently cannot rule out this possibility, we find it unlikely that DivIVA would play two distinct roles in competence.

The effect of delocalizing ComN in competence development led us to investigate whether ComN played a role in the polar localization of the competence apparatus. We initially tested whether ComN could be part of the DNA uptake complex by localizing ComN in competent cells and comparing it to the localization of the DNA uptake machinery. We used a ComGA-CFP fusion as a marker for the DNA uptake machinery and combined it with GFP-ComN in strain FG1109. GFP-ComN was imaged with a YFP filter and controls were carried out to show that under our imaging conditions we could separate the signals originating from the ComGA-CFP and GFP-ComN fusions. Fluorescence microscopy of strain FG1109 in competence medium revealed ComGA-CFP foci in about 1% of the cells (Fig. 6A). This is somewhat lower than what has been reported before (12, 19) and may be due to the fact that our wild-type background (PY79) does not develop competence as efficiently as other B. subtilis strains. Among the cells with foci, the majority (∼60%) were polar, but some were localized on the side walls as well. The pattern of localization of GFP-ComN in the same cells was similar to that found for GFP-ComN in growing cells, with clear polar accumulation (Fig. 6A). No septal band of GFP-ComN was found in the competent cells, however, probably because divisome assembly is blocked as a result of competence development (13). Superimposing the ComGA-CFP and GFP-ComN signals (Fig. 6A) revealed that the localization of ComGA and ComN had limited overlap. There were several cases in which ComN staining appeared as bipolar caps, whereas ComGA foci was found in only one pole (Fig. 6A, white arrow). There were also cells in which a ComGA focus was found in a region where little or no ComN staining was present (Fig. 6A, yellow arrow). Furthermore, in the cases where ComGA foci overlapped with ComN polar caps, the staining patterns of the two proteins was dissimilar enough (note the cell marked with the white arrow in Fig. 6A) that they must represent two distinct cellular structures.

We have also investigated the effect of the comN mutation on ComGA foci formation. Figure 6B shows that ComGA still forms foci in a comN mutant. The frequency of total ComGA foci was similar in the comN mutant and in the wild type. In addition, deletion of comN did not affect the polar localization of ComGA foci (Fig. 6C). Similar results were obtained when we used a RecA-YFP fusion as a marker of the DNA uptake complex (see Fig. S3 in the supplemental material). We have also evaluated ComGA localization in the divIVA minD mutant, a situation in which ComN is delocalized and found that ComGA was still capable of forming polar foci (see Fig. S4 in the supplemental material). These results indicate that at least part of the components of the DNA uptake and processing apparatus can assemble and localize properly in the absence of ComN or when ComN is delocalized. Polar localization of the DNA uptake apparatus has been proposed to occur by a diffusion and capture mechanism, in which foci form at different locations along the cell membrane and diffuse freely until they reach the cell poles and recognize a still unknown protein that “anchors” the foci there (11). The lack of an effect of the comN deletion indicates that ComN is not this anchor protein and is consistent with a previous report that the polar localization of foci was not dependent on DivIVA (12).

The role of ComN as a posttranscriptional regulator of ComE expression (28) implies that ComN should interact with the comE mRNA and, potentially, localize this mRNA to the nascent septum and cell poles. To investigate this possibility, we tagged the comE mRNA with binding sites for phage MS2 coat protein and visualized the comE mRNA distribution in a strain expressing the MS2 coat protein fused to GFP (MS2-GFP) (2). We created two tagged versions of the predicted comE transcript, one with the tag immediately downstream of the comEA coding region (plasmid pAB61a) and the other with the tag downstream of the comEC coding region (plasmid pAB61c), and used these constructs to replace the wild-type comE gene, such that the expression of the tagged transcripts would be under the control of the native comE promoter. To avoid well-known artifacts of the MS2 system (26), we carried out our experiments at a low level of MS2-GFP induction (50 μM IPTG), which we found does not lead to mRNA-independent aggregation of MS2-GFP (see Fig. S5 in the supplemental material). Localization of comE mRNA in cultures grown in competence medium revealed cells in which the MS2-GFP fluorescence was brighter at nascent septa and new poles (Fig. 7A, A′, B, and B′, arrows). There was also MS2-GFP fluorescence accumulation at older septa/poles (inferred to be old because of being constricted or hemispherical), but it was weaker and less frequent than the septal and new pole localizations (Fig. 7A, A′, B, and B′, arrowheads). This localization pattern was identical whether we used the construct with the MS2 tag downstream of comEA (Fig. 7A and A′) or comEC (Fig. 7B and B′). Because the localization of MS2-GFP was very similar to that detected for the ComN protein (see Fig. 1), this suggested that ComN is indeed capable of localizing comE mRNA. To confirm this hypothesis, we carried out several controls. First, we verified that localization of MS2-GFP was dependent on the presence of the tagged comE transcript, since MS2-GFP was homogenously distributed in the cytoplasm of strains lacking tagged comE constructs (see Fig. S5 in the supplemental material, top panel). Furthermore, we found that cells with localized MS2-GFP never comprised more than ca. 10% of the population. This is expected because only the minority of cells that are developing competence should express the tagged comE transcript. Finally, and most importantly, septal and polar MS2-GFP localization was strictly dependent on ComN and DivIVA: no localization of MS2-GFP could be observed in a comN mutant (Fig. 7C and C′) and in a divIVA minD mutant (Fig. 7D and D′), a situation in which ComN is still present in the cell but delocalized (Fig. 3). Thus, our results strongly support the conclusion that ComN, via its interaction with DivIVA, is capable of localizing the comE mRNA to septa and poles.

We sought to determine the biological significance of the comE mRNA localization. Targeting of mRNA is often associated with localized translation and asymmetric protein distribution in eukaryotic cells (5, 35). However, we could not demonstrate this to be the case for the proteins encoded by the comE mRNA. ComEA, which is the DNA receptor of the competence apparatus, is distributed roughly evenly in cell membranes (18) and thus would neither require nor benefit from localized translation. On the other hand, ComEC, the integral membrane protein that constitutes the DNA transport pore, is a polarly localized protein (18), and its localization may depend on its mRNA being localized. We attempted to investigate the effect of the comN and divIVA mutations in the polar localization of ComEC, but we found that the ComEC-YFP fusion produced fluorescence that was too weak to generate meaningful data, similarly to what was reported by the Dubnau lab (20). Thus, at this point, we cannot rule out that some of the components of the competence complex, in particular ComEC, will require ComN for proper localization. Other possible roles for septal localization of the comE mRNA would be to enforce equal partitioning of mRNA molecules between daughter cells that are becoming competent or to control the timing of translation by sequestering the mRNA at a subcellular location (in this case, the septum or cell poles), where it would be inaccessible to ribosomes. Developmental programs such as sporulation and competence commence one or more cell cycles before cells commit to differentiation and stop dividing (21, 23). Using the division complex to sequester mRNAs may function as a developmental checkpoint capable of coupling translation of specific messages to the cessation of cell division.

We have demonstrated here that ComN is a protein whose localization is enriched at the septa and poles of B. subtilis cells. Localization of ComN is mediated by a direct interaction with DivIVA, reinforcing the idea that DivIVA is a general pole-marking protein crucial for the spatial organization of a diverse set of processes (division, chromosome segregation, and gene expression and/or protein segregation). One interesting question raised by our observation is how many more proteins find their way to the poles of B. subtilis via DivIVA? Given the variety of processes in which DivIVA is involved in B. subtilis and other Gram-positive bacteria, we suspect that there may be more proteins that exploit DivIVA to reach the cell poles awaiting discovery. Another important question is the physiological significance of ComN localization. ComN localization, and its ability to localize a target mRNA, suggests that production of the proteins controlled by this factor should occur preferentially at the cell poles. Local production of proteins controlled by ComN is not essential for the polar assembly of at least some components of the competence DNA uptake machinery, but it should play a role in competence acquisition because delocalization of ComN leads to a small reduction in competence efficiency. We postulate that this decrease in competence may be related to delocalized production of ComEC, which may prevent its efficient incorporation into polar DNA uptake complexes.

Finally, the demonstration that ComN mediates targeting of the comE mRNA to septa and poles strengthens the notion that mRNA localization and/or localized translation could play a significant role in the organization of the bacterial cell. Two recent reports showed that bacterial mRNAs are present at specific subcellular locations (26, 27). In one case, C. crescentus mRNAs were shown to stay close to their gene, probably because translation of nascent transcripts by the ribosome prevents their diffusion to other locations (26). In a second report, Nevo-Dinur et al. showed that E. coli mRNAs localized to sites where the proteins encoded by these RNAs should function (the membrane, for integral membrane proteins, and the cell pole in the case of the BglG transcription factor) (27). Such mRNA localization was mediated by cis-acting sequences, but the existence of trans-acting factors capable of binding to the mRNAs and anchoring them in specific cellular locations remained elusive (27). We showed here that ComN functions as one of these trans-acting factors, bolstering the idea that mRNA localization by factors other than the ribosome does indeed occur. Because ComN orthologs are absent from Gram-negative bacteria, other proteins should be responsible for targeting mRNAs to the membrane and poles of E. coli. Identifying these factors and their target mRNAs in E. coli, B. subtilis, and other model bacteria should clarify the breadth and impact of mRNA localization to the cellular organization of prokaryotes.