Domain organization of NlpD

NlpD is an outer membrane lipoprotein consisting of multiple domains: a lysin motif (LysM) domain [19–21], a dLytM domain [8], and two linker regions, one at the N-terminus and another between the LysM and dLytM domains (Fig 1). The linker regions are predicted to be disordered/unstructured by DISOPRED [22]. The LysM domain is a common motif in envelope proteins and is associated with PG binding activity [20,21]. The dLytM domain has been reported to mediate amidase activation by the other dLytM factor EnvC [6,8]. To begin dissecting the regulation of amidase activation by NlpD, we initiated a structure-function analysis of the protein in order to determine which domains are necessary and/or sufficient for recruitment to the divisome and amidase regulation. The results are summarized in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Structure-function analysis of NlpD.

The domain organization of NlpD is illustrated. Indicated are the signal sequence (SS; yellow), lysin motif (LysM; blue), and the degenerate LytM domain (dLytM; green). Also shown are the NlpD truncations that were expressed under the control of the IPTG-inducible lactose promoter either as an untagged protein or as a C-terminal mCherry fusion. Truncations lacking ^SSNlpD are expressed as soluble periplasmic proteins fused to the DsbA signal peptide. Columns indicate (i) the NlpD residues present in each truncation, (ii) whether the fusion to mCherry accumulated at division sites strongly (+++), poorly (+), or appeared evenly distributed along the periphery of the cell (-), and (iii) whether the untagged truncation could (+) or could not (-) compensate for the loss of endogenous NlpD for proper cell separation. ND, not determined.

https://doi.org/10.1371/journal.pgen.1006888.g001

The LysM domain of NlpD is necessary and sufficient for septal localization

In order to determine which region of NlpD is the localization determinant, we constructed different NlpD variants with the fluorescent protein mCherry fused at the C-terminus (Fig 1). We expressed truncated versions of NlpD lacking either the dLytM domain or both the dLytM and LysM domains. Additionally, we also expressed soluble periplasmic variants containing only the dLytM domain (NlpD^dLytM) or only the LysM domain (NlpD^LysM). These domains were fused to the signal sequence of DsbA for Sec-mediated export into the periplasm [23]. Finally, we expressed mCherry fused to the signal sequence of NlpD plus 4 additional amino acids after the acylated cysteine to assess the effects of lipidation and OM localization on localization pattern. All the constructs were expressed under the control of the IPTG-inducible lactose promoter (P_lac) from a chromosomally integrated plasmid in an nlpD null strain. Additionally, all fusion proteins were found to be largely intact by western blot analysis using an antibody against the mCherry portion of the fusion (S1A Fig).

Because the nlpD gene contains a promoter for the downstream rpoS gene, we did not use a complete nlpD deletion as our null allele. RpoS is an alternative sigma factor and global regulator that controls the expression of a large number of genes in response to a variety of stresses [24]. A complete deletion of nlpD results in defects in rpoS expression [25]. In order to inactivate NlpD with minimal effects on rpoS expression, we constructed an in-frame deletion of the coding sequence for the N-terminal lipoprotein signal sequence of NlpD (^ΔSSnlpD) to prevent its export to the periplasm where it normally functions. Importantly, ^ΔSSnlpD and ΔnlpD mutants displayed the same severe chaining phenotype when combined with a ΔenvC mutation, suggesting that a ^ΔSSnlpD strain behaves like an nlpD null strain with respect to amidase activation. We therefore use ^ΔSSnlpD as our null allele throughout this study.

As expected from previous results [7], the full-length NlpD-mCherry fusion was functional for cell separation and localized to midcell in both constricted and unconstricted cells (Figs 1 and 2A). All constructs containing the LysM domain, whether lipidated or free-floating, also displayed midcell localization (Figs 1, 2B and 2D), indicating that the LysM domain is sufficient for localization. A fusion with just the N-terminal linker also appeared to have some affinity for the midcell, although the enrichment at the division site was much reduced compared to wild-type NlpD (Figs 1 and 2C). This localization was not due to lipidation since a fusion of mCherry to just the NlpD signal sequence did not show any specific midcell localization (Figs 1 and 2G). Finally, the soluble periplasmic NlpD^dLytM variants (NlpD^(250–379)-mCherry and NlpD^(189–379)-mCherry) were dispersed throughout the periplasm, suggesting that although this domain is responsible for amidase activation, it does not promote NlpD recruitment to the division site (Figs 1, 2E and 2F). Taken together, these results suggest that, similar to other LysM-containing division proteins [26–28], the LysM module of NlpD is both necessary and sufficient for proper localization to the septal ring.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Localization of NlpD fusions in ^ΔSSnlpD cells.

Overnight cultures of MT47 (^ΔSSnlpD) harboring the integrated expression constructs (A) attHKNP20 (P_lac::nlpD^WT-mCherry), (B) attHKMT101 (P_lac::nlpD^(1–189)-mCherry), (C) attHKMT103 (P_lac::nlpD^(1–115)-mCherry), (D) attHKMT178 (P_lac::^ssdsbA-nlpD^(102–175)-mCherry), (E) attHKMT180 (Pl_ac::^ssdsbA-nlpD^(250–379)-mCherry), (F) attHKMT182 (P_lac::^ssdsbA-nlpD^(189–379)-mCherry), or (G) attHKMT149 (P_lac::nlpD^(1–30)-mCherry) were diluted in minimal M9-maltose medium supplemented with 25μM (E-F), 50μM (A-B), 100μM (C-D), or 150μM (G) IPTG. Cells were grown at 30°C to an OD₆₀₀ of 0.15–0.25 before they were visualized on 2% agarose pads by phase contrast and fluorescence microscopy. Arrows indicate localization of the protein fusion to division sites. Bar = 4μm.

https://doi.org/10.1371/journal.pgen.1006888.g002

The dLytM domain of NlpD is required for function and causes lysis when periplasmic

The functional importance of the subdomains of NlpD for cell separation was also investigated using the truncations described above, but in this case as untagged proteins in order to avoid potential negative effects on NlpD functionality due to the mCherry tag. To assess their function, the different truncations were each expressed under the control of the P_lac promoter from a chromosomally integrated plasmid in a strain lacking both NlpD and EnvC (Fig 1). Without a complementing construct, the ^ΔSSnlpD ΔenvC strain formed very long chains of cells that failed to separate (Fig 3A). Full correction of the NlpD defect in this strain should restore its division function such that it resembles a single ΔenvC strain, which primarily forms long cell pairs and short chains [29] (Fig 3B). It can be difficult to quantitatively compare the severity of cell chaining phenotypes between mutants using microscopy alone because many cell chains extend outside of the field of view. We therefore used flow cytometry to complement our phenotypic analysis. In this assay, cells with increased cell length, such as chaining cells or smooth filaments, cause a significant increase in the detected side scatter pulse width (SSC-W) relative to normally dividing cells. The ΔenvC strain, which is relatively uniform in size, showed a tight peak in the SSC-W distribution by flow cytometry (Fig 3C). On the other hand, the long and heterogeneously sized cell chains of the ^ΔSSnlpD ΔenvC strain displayed a broad SSC-W distribution. Thus, this assay allowed us to assess on a cell population-level how efficiently the different NlpD truncations rescued the cell separation defect of the ^ΔSSnlpD ΔenvC double mutant (Fig 3D–3H).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Only full-length NlpD properly promotes cell separation.

(A-C) MT50 (^ΔSSnlpD ΔenvC) or TB140 (ΔenvC) cells were grown in minimal M9-maltose medium at 37°C to an OD₆₀₀ of 0.2–0.3 before visualization on 2% agarose pads with DIC optics (A-B) or analyzed by flow cytometry (C). Bar = 10μm. (D-H) Overnight cultures of MT50 harboring the integrated expression constructs (D) attHKMT20 (P_lac::nlpD^WT), (E) attHKMT102 (P_lac::nlpD^(1–189)), (F) attHKMT104 (P_lac::nlpD^(1–115)), (G) attHKMT179 (Pl_ac::^ssdsbA-nlpD^(250–379)), or (H) attHKMT181 (P_lac::^ssdsbA-nlpD^(189–379)) were diluted in minimal M9-maltose medium and grown at 37°C. Mid-log cultures were then backdiluted into M9-maltose medium with or without the inducer IPTG. Cells were further grown at 37°C to an OD₆₀₀ of 0.2–0.3 before flow cytometry analysis. Histograms of cultures that were either uninduced (grey) or induced with 150μM (D, G-H, blue) or 1mM (E-F, red) IPTG were overlayed. (I) Cells of MT122 (^ΔSSnlpD) and MT123 (^ΔSSnlpD ΔamiC) alone or harboring the integrated constructs attHKMT20 (P_lac::nlpD^WT), attHKMT179 (Pl_ac::^ssdsbA-nlpD^(250–379)), or attHKMT181 (P_lac::^ssdsbA-nlpD^(189–379)) were grown in LB at 30°C. Following normalization for cell density (OD₆₀₀ = 0.5), 5 μl of the resulting cultures was spotted on LB agar containing 150μM IPTG and 20 μg/ml CPRG. The plates were incubated at 30°C and photographed after 14 hours.

https://doi.org/10.1371/journal.pgen.1006888.g003

Expression of wild-type NlpD from the integrated plasmid complemented the ^ΔSSnlpD mutation, and converted the broad SSC-W profile of the double mutant to one resembling the tight distribution of a single ΔenvC strain (Figs 1, 3D and S2A Fig). Variants lacking the dLytM domain all failed to suppress the cell chaining phenotype (Figs 1, 3E, 3F and S2B and S2C Fig). At least one of these variants (NlpD^(1–189)) was expressed and remained largely intact as assessed by immunoblot analysis. However, it is possible that the truncated product observed for this variant affects the ability of the NlpD^(1–189) to correct the cell chaining phenotype of ^ΔSSnlpD ΔenvC cells (S1B Fig). The other variant, NlpD^(1–115), was not detected, either due to the lack of robust antigenic epitopes in this truncation or proteolytic degradation, so we could not determine its functionality. Overall, the results indicate that, similar to EnvC, the dLytM domain of NlpD is required for amidase activation and cell separation in vivo.

To test whether the dLytM domain of NlpD is sufficient for NlpD activity, the soluble NlpD^dLytM variants (NlpD^(250–379) and NlpD^(189–379)) were produced in the ^ΔSSnlpD ΔenvC strain. Neither variant suppressed the cell separation defect as observed by microscopy, but they did appear to reduce the total cell length and increase the number of cells with a lower SSC-W when analyzed by flow cytometry (Figs 1, 3G, 3H and S2D and S2E Fig). Because the EnvC dLytM domain is lytic when produced in isolation [6], we suspected that the NlpD^dLytM variants may also promote aberrant amidase activation. To test for lytic activity, we used the β-galactosidase substrate chlorophenyl red-β-D-galactopyranoside (CPRG), which fails to penetrate the E. coli envelope and cannot be processed by cytoplasmic LacZ unless cells lyse and/or become more permeable [30]. Indeed, expression of both NlpD^dLytM variants resulted in a positive red signal on CPRG-containing medium (Fig 3I). This phenotype was not caused by NlpD^WT, and was largely suppressed by inactivation of AmiC, the target amidase of NlpD (Fig 3I). Thus, the lytic activity of the NlpD^dLytM variants is most likely the result of spurious AmiC activation, rather than a non-specific defect resulting from the production of truncated proteins in the periplasm. However, some residual lysis is observed even in AmiC-defective cells and is probably caused by promiscuous activation of the other amidases, AmiA and AmiB, by the NlpD^dLytM variants as has been observed previously for certain AmiB derivatives [5]. We conclude that although the dLytM domain is critical for NlpD activity and sufficient for amidase activation, it is unable to promote proper cell separation on its own and may require additional domains of the NlpD protein for its proper regulation.

Outer membrane localization of NlpD is important for proper function and regulation

The behavior of the NlpD^dLytM truncations indicated that divisome recruitment and/or OM localization is important for the proper control of AmiC activation by NlpD. To investigate these possibilities further, we generated additional full-length NlpD variants predicted to have an altered localization within the envelope. Firstly, the residue after the acylated cysteine (the +2 position) was altered from a serine to aspartate (NlpD^(S27D)) to prevent transport of this variant by the Lol system and force its retention in the inner membrane [31] (Fig 4A). We also constructed a variant (^DsbANlpD^(27–379)) in which the first 26 residues, including the lipidated cysteine, were replaced by the signal sequence of DsbA to export a full-length, mature form of NlpD as a soluble periplasmic protein.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. OM localization of NlpD is required for proper cell separation.

(A) The domain structure of NlpD is illustrated as in Fig 1. Details of the signal sequence are presented with the lipobox in red and the arrow indicating the cleavage site just before the acylated cysteine. The IM-retained variant (NlpD^(S27D)) contains a mutated signal sequence (indicated by the asterisk) with an aspartate at the +2 position after the acylated cysteine (underlined). The soluble periplasmic variant (NlpD^(27–379)) is fused to the DsbA signal peptide (purple) that is cleaved upon export to the periplasm via the Sec system. (B-D) Cytological assay to determine the subcellular localization of the NlpD variants. Overnight cultures of MT47 (^ΔSSnlpD) expressing different NlpD-mCherry fusions from the integrated constructs (B) attHKNP20 (P_lac::nlpD^WT-mCherry), (C) attHKMT21 (P_lac::nlpD ^(S27D)-mCherry), or (D) attHKMT147 (P_lac::^ssdsbA-nlpD ^(27–379)-mCherry) were diluted in minimal M9-maltose medium supplemented with 25 μM (D) or 150 μM (B, C) IPTG. Cells were grown at 30°C to an OD₆₀₀ of 0.4, washed and then osmotically shocked by resuspension in plasmolysis buffer and the plasmolyzed cells were visualized by phase contrast and fluorescence microscopy. Arrows indicate signals that display a smooth OM peripheral signal in (B), track with the inner membrane in (C), or fill the increased periplasmic spaces of plasmolysis bays in (D). Bar = 4μm. (E-G) Overnight cultures of MT50 (^ΔSSnlpD ΔenvC) harboring the integrated expression constructs (E) attHKMT20 (P_lac::nlpD^WT), (F) attHKMT12 (P_lac::nlpD ^(S27D)), or (G) attHKMT121 (P_lac::^ssdsbA-nlpD ^(27–379)) were diluted in minimal M9-maltose medium and grown at 37°C until mid-log. Cultures were then backdiluted into M9-maltose medium without (grey histogram) or with 150 μM (blue histogram) or 1 mM (red histogram) IPTG and grown at 37°C to an OD₆₀₀ of 0.2–0.3 before flow cytometry analysis. (H) Cells of MT122 (^ΔSSnlpD) and MT123 (^ΔSSnlpD ΔamiC) alone or harboring the integrated constructs described above were grown and spotted on CPRG agar as described in Fig 3I. The subcellular localization of each NlpD variant is indicated in the square brackets: OM, outer membrane; IM, inner membrane; peri, periplasm.

https://doi.org/10.1371/journal.pgen.1006888.g004

In order to assess whether the different variants localized as expected, a cytological assay was used [32]. Cells expressing each variant fused to mCherry at the C-terminus were hyperosmotically shocked to induce plasmolysis and subsequently visualized using both phase contrast and fluorescence microscopy. Upon plasmolysis, the cytoplasm contracted, pulling the inner membrane away from the envelope to form plasmolysis bays (Fig 4B–4D). Wild-type NlpD-mCherry displayed a smooth peripheral fluorescence signal in the plasmolyzed cells, consistent with an OM localization (Fig 4B). On the other hand, the fluorescence pattern of NlpD^(S27D)-mCherry followed the retracted inner membrane, indicating that this variant is indeed retained in the IM (Fig 4C). Finally, the fluorescent signal of NlpD^(27–379)-mCherry accumulated in the increased periplasmic space of the plasmolysis bays, which are phase-gray in the phase-contrast images (Fig 4D). Based on these cytological results, we conclude that the different NlpD variants are present in the expected subcellular locations.

Next, we determined whether the NlpD variants with altered localization were functional in amidase activation by assessing their ability to complement the chaining phenotype of ^ΔSSnlpD ΔenvC cells. In this case, untagged variants were produced from the P_lac promoter from a chromosomally integrated plasmid to achieve levels comparable to those produced from natively expressed nlpD (S3A Fig). Although all variants were largely intact, a prominent degradation product was present with the periplasmic ^DsbANlpD^(27–379) variant, and we were concerned about its possible negative effect(s) on the functionality of the full-length protein. To address this concern, we showed that cells lacking envC but expressing wild-type endogenous NlpD are able to divide properly in the presence of either the IM-retained NlpD variant (S4B Fig) or the periplasmic ^DsbANlpD^(27–379) variant (S4C Fig). Hence, neither the NlpD variant nor any degradation product interferes with the ability of wild-type NlpD to activate AmiC and promote cell separation.

As shown above, expression of wild-type NlpD rescued the chaining phenotype of the ^ΔSSnlpD ΔenvC double mutant such that cells displayed the mild division defect of a single ΔenvC mutant and had a similar flow cytometry profile (Fig 4E and S3B and S3C Fig). Expression of the IM-retained NlpD variant did not suppress the chaining phenotype of the double mutant, even when fully induced, as assessed by both microscopy and flow cytometry (Fig 4F and S3D Fig). Since this variant is stable and intact (S3A Fig) and at least partly able to localize to division sites (S4E Fig), this finding suggests that anchoring NlpD to the IM greatly reduced its ability to activate its cognate amidase AmiC. Notably, the soluble periplasmic NlpD variant also failed to fully complement the nlpD null phenotype, even though it displayed proper septal localization (Fig 4G and S4F Fig). Although production of this variant led to a reduction in chain length as assessed by both microscopy and flow cytometry (Fig 4G and S3E Fig), many chaining cells were still observed in the population by microscopy and detected in the flow cytometry histogram as a tailed distribution of SSC-W towards longer cells. Thus, despite retaining the ability to activate AmiC for cell separation, this NlpD variant does so less efficiently than the wild-type protein. Because ^DsbANlpD^(27–379) induced an AmiC-dependent red signal on CPRG-containing medium (Fig 4H), it is difficult to know how much of the reduced cell chaining of the ^ΔSSnlpD ΔenvC double mutant promoted by this variant is due to normal cell separation activity versus the lysis of cell units within the chains. Thus, the observed reduction of cell chaining may overestimate the separation activity of ^DsbANlpD^(27–379). Also, the cell lysis phenotype induced by this variant suggests that NlpD activation of AmiC is poorly regulated when NlpD is no longer bound to the OM. However, the observed degradation product of the periplasmic ^DsbANlpD^(27–379) variant may also contribute to the cell lysis phenotype, especially if it consists of the dLytM domain alone. Overall, we conclude that OM localization of NlpD is required for the efficient splitting of septal PG by its cognate amidase AmiC and possibly also for proper regulation of AmiC activity.

Identification of factors required for NlpD activation using cell sorting

Just as FtsEX lies upstream of EnvC to connect its activation of the amidases to the activity of the divisome [13], regulators governing NlpD activity are also likely to exist. Moreover, given our observations that OM localization is required for proper NlpD function, we suspected that such regulators may be OM proteins. To identify these factors, we took advantage of the severe chaining phenotype caused by the simultaneous inactivation of both the EnvC and NlpD cell separation pathways. Based on SSC-W, long chains of the ^ΔSSnlpD ΔenvC mutant can be effectively sorted from single ΔenvC mutant cells by gating for cells at the high end of the SSC-W distribution (Fig 5A and 5B). We therefore reasoned that mutants in any gene encoding a factor needed for NlpD activation should similarly display a severe chaining phenotype in an EnvC^- background, and that we could enrich for them by cell sorting.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Flow cytometry-based enrichment and screen to identify potential NlpD regulators.

(A-C) Dot plots of (A) a control non-chaining strain, TB140 (ΔenvC), (B) a control chaining strain, MT50 (^ΔSSnlpD ΔenvC), and (C) the transposon (EZ-Tn5 <Kan-2>) mutagenized TB140 library, with side scatter pulse height (SSC-H; y-axis) versus side scatter pulse width (SSC-W; x-axis). The gate used for sorting is shown on the plot, with the number of events present in that gate indicated as a percentage of the total population. (D-J) Overnight cultures of (D) TB28 (wt), TB140 (ΔenvC), (E) MT47 (^ΔSSnlpD), MT50 (ΔenvC ^ΔSSnlpD), (F) MT140 (ΔyraP), MT135 (ΔenvC ΔyraP), (G) MT51 (ΔtolQ-pal), MT55 (ΔenvC ΔtolQ-pal), or (H) MT141 (^ΔSSnlpD ΔyraP), MT53 (^ΔSSnlpD ΔtolQ-pal) were diluted in LB medium. Cells were grown at 30°C to an OD₆₀₀ of 0.3–0.5 before they were visualized on 2% agarose pads with DIC optics (D-H) or analyzed by flow cytometry (I-J). Bar = 10μm. (K-L) Overnight cultures of MT50 (ΔenvC ^ΔSSnlpD) and (K) MT135 (ΔenvC ΔyraP) and MT179 (ΔenvC ΔyraP ^ΔSSnlpD) or (L) MT55 (ΔenvC ΔtolQ-pal) and MT96 (ΔenvC ΔtolQ-pal ^ΔSSnlpD) were diluted in LB medium. Cells were grown at 30°C to an OD₆₀₀ of 0.15–0.25 before analysis by flow cytometry.

https://doi.org/10.1371/journal.pgen.1006888.g005

For the enrichment and screen, an EnvC-depletion strain was transposon-mutagenized. The temperature-sensitive plasmid containing the inducible envC gene was cured during the mutagenesis procedure by growing the transposon mutants at 37°C [33]. The resulting transposon library in a ΔenvC background was then sorted for cells with high SSC-W (gate shown in Fig 5C). Sorted cells were plated on M9-maltose agar and grown at 37°C. A colony morphology screen was then used to distinguish mutants with severe chaining phenotype from non-chaining cells isolated as background. This secondary screen was based on the observation that colonies containing chaining cells are mucoid when grown on this medium at 37°C. Mucoid colonies were purified on LB and their cell chaining phenotype (as opposed to cell filamentation) confirmed microscopically (S5 Fig) before performing further analysis to map the position of the transposon insertion. As an indication that the screen was working properly, many of the transposon mutants had insertions mapping within nlpD or amiC (6 and 10 unique insertions, respectively). In addition to these expected mutants, many isolates had insertions that mapped to tolB (11 unique), pal (3 unique), tolQ (3 unique) and tolA (1 insertion). These genes encode members of the Tol-Pal system, which has been implicated in OM integrity and OM invagination during cell constriction [34–36]. Finally, three unique insertions were found within yraP, which encodes a predicted OM lipoprotein of unknown function [37].

To further investigate the potential connection of YraP and Tol-Pal with NlpD function, we assessed the division phenotypes of ΔyraP or ΔtolQ-pal mutants singly or in combination with ΔenvC or ^ΔSSnlpD mutations. When grown in LB medium at 30°C, the single ΔyraP and ΔtolQ-pal mutants appeared to grow and divide normally (Fig 5F, 5G and 5I). However, when combined with an envC deletion, inactivation of YraP or Tol-Pal resulted in a severe chaining defect similar to the ^ΔSSnlpD ΔenvC combination (Fig 5E–5G, 5J and S5C–S5E Fig). On the other hand, inactivating YraP or Tol-Pal in cells lacking NlpD had either no effect on division or caused only a mild cell separation defect (Fig 5H). The results are thus consistent with these factors being involved in the same pathway as NlpD, rather than EnvC, for cell separation. A triple ^ΔSSnlpD ΔenvC ΔyraP mutant displayed a cell separation phenotype equivalent to the ^ΔSSnlpD ΔenvC strain (Fig 5K). However, adding the ΔtolQ-pal lesion to a strain lacking NlpD and EnvC further exacerbated the already severe chaining defect of the double mutant (Fig 5L). The genetic results, therefore, suggest that YraP and Tol-Pal are required for proper NlpD function, with Tol-Pal likely playing a role that affects the cell separation activities of additional factors (see Discussion). Notably, consistent with their functional connection, YraP is well conserved among the Gammaproteobacteria and Betaproteobacteria, and its occurrence is correlated with the presence of an NlpD homolog (S6 Fig).

NlpD and AmiC localize to the septal ring independently of YraP or the Tol-Pal system

A potential explanation for the YraP and Tol-Pal requirement for cell separation via the NlpD pathway is that one or both sets of factors is required for the recruitment of NlpD or AmiC to the division site. To investigate this possibility, we expressed either NlpD-mCherry or AmiC-GFP in cells lacking the corresponding native protein and inactivated for YraP, the Tol-Pal system, or both factors. Both fusions displayed a normal midcell localization pattern regardless of the genetic background (Fig 6), although the fluorescence signal of the NlpD-mCherry fusion appeared slightly reduced in the absence of both YraP and Tol-Pal (Fig 6D). Another mild effect observed was that the peripheral NlpD-mCherry signal was higher in ΔyraP cells relative to those with functional YraP (Fig 6B), and the midcell AmiC-GFP signal was slightly reduced (Fig 6F). We conclude that YraP and the Tol-Pal system are not required for the recruitment of NlpD and AmiC to the divisome, although YraP may enhance the septal localization of NlpD and AmiC.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. NlpD and AmiC localize to division sites independently of YraP and/or the Tol-Pal system.

(A-D) Overnight cultures of (A) MT47 (^ΔSSnlpD), (B) MT141 (^ΔSSnlpD ΔyraP), (C) MT53 (^ΔSSnlpD ΔtolQ-pal), or (D) MT147 (^ΔSSnlpD ΔtolQ-pal ΔyraP) harboring the integrated expression construct attHKNP20 (P_lac::nlpD-mCherry) were diluted in minimal M9-maltose medium supplemented with 100μM IPTG. (E-H) Overnight cultures of (E) TB143 (ΔamiC), (F) MT150 (ΔamiC ΔyraP), (G) MT149 (ΔamiC ΔtolQ-pal), or (H) MT178 (ΔamiC ΔtolQ-pal ΔyraP) harboring the integrated expression construct attHKNP16 (P_lac-m3::amiC-gfp) were diluted in minimal M9-maltose medium supplemented with 25μM IPTG. For all strains, cells were grown at 30°C to an OD₆₀₀ of 0.2–0.3 before they were visualized on 2% agarose pads by phase contrast and fluorescence microscopy. In all images, arrows indicate the localization of the protein fusion to division sites. Bar = 4μm.

https://doi.org/10.1371/journal.pgen.1006888.g006

OM localization of YraP is important for its function

YraP is predicted to be an outer membrane (OM) lipoprotein consisting of two bacterial OsmY and nodulation (BON) domains [38] (Fig 7A). To experimentally verify this localization, we used the cytological assay [32] employed above. Upon plasmolysis, wild-type YraP-mCherry displayed a smooth peripheral fluorescence signal consistent with an OM localization (S7A Fig). On the other hand, a variant of YraP-mCherry with the +2 and +3 positions changed to aspartate and glutamate to create a Lol-avoidance signal [31] followed the retracted IM in plasmolyzed cells (Fig 7A and S7B Fig). Swapping the normal signal sequence of YraP-mCherry for that of DsbA resulted in a fusion that showed a typical periplasmic localization, with fluorescence accumulating in the phase grey plasmolysis bays of shocked cells (Fig 7A and S7C Fig). These results strongly support the designation of YraP as an OM lipoprotein.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. OM localization of YraP is required for cell separation in the absence of EnvC.

(A) The domain structure of YraP is illustrated. Indicated are the signal sequence (SS; red) and the two bacterial OsmY and nodulation domains (BON1/2; orange). Details of the signal sequence are presented with the lipobox in red and the arrow indicating the cleavage site just before the acylated cysteine. The IM-retained variant (YraP^IM) contains a mutated signal sequence (indicated by the asterisk) with an aspartate and glutamate at the +2 and +3 positions after the acylated cysteine (underlined). The soluble periplasmic variant (YraP^peri) is fused to the DsbA signal peptide (purple) that is cleaved upon export to the periplasm via the Sec system. (B-I) Overnight cultures of (B) TB140 (ΔenvC) or MT135 (ΔenvC ΔyraP) either (C) alone or harboring the integrated construct (D) attλMT196 (P_lac::yraP^WT), (E) attλMT198 (P_lac::yraP^IM), (F) attλMT209 (P_lac::^ssdsbA-yraP^peri), (G) attλMT197 (P_lac::yraP^WT-mCherry), (H) attλMT199 (P_lac::yraP^IM-mCherry), or (I) attλMT210 (P_lac::^ssdsbA-yraP^peri-mCherry) were diluted in minimal M9-maltose medium only (B-C) or supplemented with 10μM (D, F-G, I) or 20μM (E, H) IPTG. Cells were grown at 30°C to an OD₆₀₀ of 0.2–0.25 before they were visualized on 2% agarose pads with DIC optics. Bar = 10μm.

https://doi.org/10.1371/journal.pgen.1006888.g007

To assess the importance of subcellular localization for YraP function, we expressed the YraP variants described above, either untagged or with the C-terminal mCherry fusion, in the separation defective ΔyraP ΔenvC cells (Fig 7C). Both wild-type YraP variants rescued the chaining phenotype of ΔyraP ΔenvC such that cells only displayed the mild EnvC^- separation phenotype (Fig 7B, 7D and 7G). However, the YraP variants with modified subcellular localization (either IM or periplasmic) failed to complement the chaining phenotype of the ΔyraP ΔenvC strain (Fig 7E, 7F, 7H and 7I). We therefore conclude that YraP, like NlpD, needs to be OM-localized for its proper function.

YraP is a new OM-localized component of the divisome

The genetic results implicate the Tol-Pal system and YraP in the pathway that stimulates AmiC activation by NlpD. All of these factors except YraP have been shown to be recruited to the divisome. We therefore investigated the subcellular localization of YraP. In support of the protein playing a direct role in the division process, a functional YraP-mCherry fusion protein displayed a band-like localization pattern at midcell in cells with a visible constriction (Fig 8A). The IM-retained and periplasmic variants of YraP-mCherry both displayed a dispersed peripheral localization signal (Fig 8B and 8C). These fusions were found to be intact and expressed to comparable protein levels to the wild-type fusion protein by immunoblotting (S7D Fig), indicating that OM localization is required for YraP recruitment to the divisome. Recruitment of YraP-mCherry to the division site was independent of NlpD, the Tol-Pal system, and AmiC (Fig 8E–8G). As expected, however, YraP-mCherry did not localize to potential division sites in cells producing the FtsZ antagonist SulA that blocks divisome formation (S8A and S8B Fig). Notably, unlike NlpD-mCherry, which localizes to divisome assemblies in cells blocked for division by the PBP3 (FtsI) inhibitor cephalexin, YraP-mCherry failed to localize in similarly treated cells (Fig 9). The recruitment pathway for all essential divisome components remains functional in cephalexin treated cells [39], but constriction is blocked due to the disruption of septal cell wall biogenesis caused by the drug. Thus, YraP likely requires the initiation of constriction for its localization. A similar requirement was previously demonstrated for AmiC [40,41].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Recruitment of YraP to division sites is dependent on its OM localization but independent of NlpD, the Tol-Pal complex, and AmiC.

(A-C) Overnight cultures of MT140 (ΔyraP) harboring the integrated construct (A) attλMT197 (P_lac::yraP^WT-mCherry), (B) attλMT199 (P_lac::yraP^IM-mCherry), or (C) attλMT210 (P_lac::^ssdsbA-yraP^peri-mCherry) were diluted in minimal M9-maltose medium supplemented with 10μM (A-C) or 20μM (B) IPTG. (D-G) Overnight cultures of (D) MT140 (ΔyraP), (E) MT141 (ΔyraP ^ΔSSnlpD), (F) MT147 (ΔyraP ^ΔSSnlpD ΔtolQ-pal), and (G) AAY22 (ΔyraP ΔamiC) harboring the integrated construct attλMT197 (P_lac::yraP^WT-mCherry) were diluted in minimal M9-maltose medium supplemented with 10μM IPTG. All cultures were grown at 30°C to an OD₆₀₀ of 0.15–0.2 before cells were visualized on 2% agarose pads by phase contrast and fluorescence microscopy. Arrows indicate localization of the protein fusion to division sites. Bar = 4μm.

https://doi.org/10.1371/journal.pgen.1006888.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Localization of YraP or NlpD in cephalexin-treated cells.

Overnight cultures of (A-B) MT140 (ΔyraP) harboring the integrated construct attλMT197 (P_lac::yraP-mCherry) or (C-D) MT47 (^ΔSSnlpD) harboring the integrated expression construct attHKNP20 (P_lac::nlpD-mCherry) were diluted in minimal M9-maltose medium supplemented with either 25μM (A-B) or 100μM (C-D) IPTG and grown at 30°C until mid-log. Cultures were then backdiluted into M9-maltose medium with the indicated IPTG concentration with or without 10μg/ml cephalexin as indicated. Cells were grown at 30°C to an OD₆₀₀ of 0.2 before they were visualized on 2% agarose pads by phase contrast and fluorescence microscopy. Arrows indicate localization of the protein fusion to division sites. Bar = 10μm.

https://doi.org/10.1371/journal.pgen.1006888.g009

Inactivation of NlpD results in OM defects

The observed midcell recruitment of YraP places it within the divisome, along with the Tol-Pal system [36], NlpD [7], and AmiC [41], and suggests that all of these factors are directly participating in the envelope remodeling process at the division site. Although the precise function of the Tol-Pal system and YraP remains unclear, the defects in OM integrity observed upon their inactivation [37,42,43] indicate that they play critical roles in OM physiology during the formation of new daughter cell poles. Our genetic analysis suggests that these divisome components are required for the cell separation activity of NlpD. We therefore wondered whether a defect in NlpD and/or its cognate amidase AmiC might contribute to the OM problems observed when the Tol-Pal system or YraP are crippled. Notably, although the defect was not as severe as that observed with cells lacking YraP or the Tol-Pal system, ^ΔSSnlpD cells were found to be hypersensitive to treatment with a combination of the detergent sodium-dodecyl sulphate (SDS) and the metal-chelating agent ethylenediaminetetraacetic acid (EDTA), a phenotype indicative of OM defects [44] (Fig 10). Moreover, a ΔamiC mutant does not show a similar hypersensitivity phenotype, indicating that the role of NlpD in maintaining the OM permeability barrier is likely distinct from its function in activating amidase activity for septal PG splitting. However, inactivation of AmiC in ^ΔSSnlpD cells further exacerbated their SDS-EDTA sensitivity, suggesting some contribution of AmiC to OM integrity in the absence of NlpD. This observation suggests that AmiC may be cross activated by other dLytM factors and that its deletion in ^ΔSSnlpD cells further reduces cell separation efficiency, defects in which are well known to result in reduced OM barrier function [10–12,45]. Taken together, the multiple connections between NlpD activity and the OM uncovered in this study support a model in which NlpD provides a key link that helps coordinate OM invagination and PG remodeling at the division site.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. NlpD inactivation results in OM integrity defects.

Cells of TB28 (WT), TB143 (ΔamiC), MT47 (^ΔSSnlpD), MT176 (^ΔSSnlpD ΔamiC), MT140 (ΔyraP), and MT51 (ΔtolQ-pal) were grown overnight in LB at 30°C. Following normalization for cell density (OD₆₀₀ = 1), the resulting cultures were serially diluted (10⁻¹ to 10⁻⁶), and 5 μl of each dilution was spotted on LB agar only or supplemented with the indicated concentrations of SDS and EDTA. The plates were incubated at 37°C and photographed after ~24 hours.

https://doi.org/10.1371/journal.pgen.1006888.g010

Media, bacterial strains, and plasmids

Cells were grown in LB (1% tryptone, 0.5% yeast extract, 0.5% NaCl) or minimal M9 medium [53] supplemented with 0.2% casamino acids and 0.2% maltose. Unless otherwise indicated, antibiotics were used at concentrations of 25 μg/ml (chloramphenicol; Cm), 25 μg/ml (kanamycin; Kan), 5 μg/ml (tetracycline; Tet), or 50 μg/ml (ampicillin, Amp).

The bacterial strains used in this study are listed in S1 Table. All E. coli strains used in the reported experiments are derivatives of MG1655 [54]. All deletion alleles were either sourced from the Keio knockout collection [55] or constructed to resemble those in the collection, with the exception of the ΔtolQ-pal and ^ΔSSnlpD alleles. Plasmids used in this study are listed in S2 Table. Plasmid construction details for this study are provided in S1 Text. Plasmids that are derivatives of CRIM vectors [56] were integrated into phage attachment sites (HK022 or λ) using the helper plasmids pTB102 [57] or pINT-ts [56], respectively, as described [56]. PCR was performed using either KOD polymerase (Novagen) or Q5 polymerase (NEB) according to the manufacturer’s instructions. Unless otherwise indicated, MG1655 chromosomal DNA was used as the template. Plasmid DNA and PCR fragments were purified using the Zyppy plasmid miniprep kit (Zymo Research) or the Qiaquick PCR purification kit (Qiagen), respectively. Sequencing reactions were carried out with an ABI3730xl DNA analyzer at the DNA Resource Core of Dana-Farber/Harvard Cancer Center (funded in part by NCI Cancer Center support grant 2P30CA006516-48).

Recombineering

Since the promoter of rpoS is present within the coding sequence of nlpD, a gene knockout of nlpD is polar on rpoS expression. Therefore, in order to make an nlpD null strain, the N-terminal lipoprotein signal sequence of NlpD was deleted (^ΔSSnlpD) to prevent the export of this protein to the periplasm, where it normally functions. The ^ΔSSnlpD::kan^R allele was constructed by replacing the region between the 2nd codon and 27th codon of nlpD with the Kan^R cassette by λ recombineering as described previously [55,58]. The Kan^R cassette was amplified from pKD13 [59] using the primers 5’-TGTCACTGGTTATTAACCAATTTTTCCTGGGGGATAAATGATTCCGGGGATCCGTCGACC-3’ and 5’-TAACGGAGCTGACCGGTGCCGGTGGATTTGAAGTGTCAGATGTAGGCTGGAGCTGCTTCG-3’. The resulting PCR product was electroporated into strain TB10, and recombinants were selected at 30°C on LB with kanamycin, generating a chromosomal deletion as described previously [60]. The Kan^R cassette was removed using FLP recombinase expressed from pCP20 [59], leaving behind the frt site scar to replace the signal sequence of ^ΔSSnlpD.

The ΔtolQ-pal::Kan^R allele was constructed by replacing the region between the 2nd codon of tolQ and the 7th codon from the stop codon of pal with the Kan^R cassette by λ recombineering as described previously [55,58]. The Kan^R cassette was amplified from pKD13 [59] using the primers 5’-GTGCGCTTCCCAAGTCTATTGTCGCGGAGTTTAAGCAGTGATTCCGGGGATCCGTCGACC-3’ and 5’-TACTGCTCATGCAATTCTCTTAGTAAACCAGTACCGCACGTGTAGGCTGGAGCTGCTTCG-3’. The resulting PCR product was purified and electroporated into strain TB28/pKD46 as described previously [29], and recombinants were selected at 30°C on LB plates supplemented with kanamycin to generate the chromosomal deletion.

The ΔyraP::Kan^R allele was constructed by replacing the region between the 2nd codon and the 7th codon from the stop codon of yraP with the Kan^R cassette as described above. The Kan^R cassette was amplified from pKD13 using the primers 5’-GTGCGCTTCCCAAGTCTATTGTCGCGGAGTTTAAGCAGTGATTCCGGGGATCCGTCGACC-3’ and 5’-TACTGCTCATGCAATTCTCTTAGTAAACCAGTACCGCACGTGTAGGCTGGAGCTGCTTCG-3’. The resulting fragment was electroporated into strain TB10 and the recombinants were selected at 30°C on LB plates with kanamycin, generating a chromosomal deletion as described previously [60].

The ΔtolQ-pal::Kan^R and ΔyraP::Kan^R alleles were transferred between strains by P1 transduction. The P1 lysate for ΔtolQ-pal::Kan^R was made on a strain with ΔtolQ-pal::Kan^R and carrying pMT94 (P_lac::nativeRBS_ybgC_tolQRA) to express the tolQRA genes. When necessary, the Kan^R cassette was removed using FLP recombinase expressed from pCP20 [59].

Plasmolysis

Overnight cultures were backdiluted in M9 medium with 0.2% maltose and IPTG and grown at 30°C (see figure legends for specific IPTG concentrations and final culture OD₆₀₀). Cells were harvested by centrifugation, washed in M9-maltose once, resuspended in plasmolysis buffer (15% sucrose, 25mM HEPES, pH 7.5, 20mM sodium azide), and immediately imaged on 1% agarose pads containing 15% sucrose using phase contrast and fluorescence microscopy as described below.

Membrane staining with FM 4-64FX

Cells were grown in LB medium at 30°C to an OD₆₀₀ of 0.2–0.3. To 1ml of culture, the membrane dye FM 4-64FX was added to a final concentration of 5 μg/ml, cells were then incubated for 10 min in the dark, followed by fixation with 2.5% formaldehyde and 0.04% gluteraldehyde in 60mM sodium phosphate buffer, pH 7.4, for 20 min at room temperature. After 3 washes with PBS (137mM NaCl, 2.7mM KCl, 1mM CaCl₂, 0.5mM MgCl₂, 10mM Na₂PO₄, 1.8mM KH₂PO₄, pH 7.4), the cells were immediately imaged on pads made with 2% agarose in LB using phase contrast and fluorescence microscopy as described below.

Microscopy

Cells were imaged on 2% agarose pads containing 1X M9 salts (unless indicated otherwise) using a Nikon TE2000 inverted microscope outfitted with a Nikon Intensilight illuminator, a Coolsnap HQ2 CCD camera from Photometrics, a Nikon CFI Plan Apo VC 100x objective (1.4 NA) for differential interference contrast (DIC) imaging or a CFI Plan Apo DM 100x objective (1.4 NA) for phase contrast imaging. Please see figure legends for growth conditions used for specific experiments. Filter cubes for fluorescence image acquisition were from Chroma. mCherry and GFP fluorescence images were taken using the ET-mCherry filter set (Chroma 49008) and ET-GFP filter set (Chroma 49002), respectively. Images were captured using Nikon Elements software, exported and cropped for figure preparation in MetaMorph (Molecular Devices).

Protein purification

NlpD was overexpressed and purified with a 6xHis-SUMO (H-SUMO) tag fused to its N-terminus [61,62]. The sequence of the affinity tag was MRGSHHHHHHMASG. The SUMO sequence was amplified from the Saccharomyces cerevisiae genome (gene Smt3) [63]. H-SUMO-NlpD(27–379) was purified from BL21(λDE3)/pTU119. Overnight cultures were diluted 1:200 into 1 L of LB supplemented with ampicillin (50 μg/ml) and cells were grown at 30°C to an OD₆₀₀ of 0.5. IPTG was then added to 1 mM, growth was continued for an additional 3.5 hours, and the cells were harvested by centrifugation. Cell pellets were resuspended in 25 ml Buffer A (20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 20 mM imidazole) and stored at -80°C prior to use for protein purification. For purification, the frozen resuspension was thawed, and the cells were disrupted by passing them through a French pressure cell twice at 15,000 psi. Cell debris were pelleted by centrifugation at 17,000 x g for 10 minutes at 4°C. The clarified lysate was incubated with Ni-NTA agarose beads (Qiagen) for 1 hour at 4°C and then loaded on a gravity flow column (BioRad). The column was washed with 10ml Buffer A and the H-SUMO-NlpD(27–379) was eluted with Buffer B (20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 300 mM imidazole). Peak fractions were collected and dialyzed against 1 L of buffer D (50 mM Tris-HCl, pH 7.4, 300 mM NaCl, 10% glycerol). After purification of the H-SUMO fusion protein, the H-SUMO tag was removed using 6xHis-tagged SUMO protease (H-SP) [63]. Cleavage reactions were passed through Ni-NTA resin (Qiagen) to remove free H-SUMO and H-SP, yielding a pure preparation of untagged NlpD without added non-native residues. Aliquots of both tagged and untagged versions of NlpD(27–379) were stored at -80°C prior to use.

Antibody preparation and affinity purification

Rabbit polyclonal antibodies against purified NlpD were prepared by Covance. Tagged NlpD protein prepared as described above was used in a custom antibody production service. The anti-NlpD antibodies were subsequently affinity-purified using purified untagged NlpD conjugated to AminoLink Plus resin (Pierce), according to the manufacturer’s instructions.

Immunoblotting

Strains were grown as described in the figure legends. At the designated times, cells were harvested and whole-cell extracts were prepared as described previously [64]. The protein concentration of each extract was determined using the Non-Interfering protein assay (G-Biosciences) according to the manufacturer’s instructions. Protein concentrations were normalized between extracts and the indicated amount of total protein from each extract was separated on a 4–20% Mini-PROTEAN^® TGX precast protein gel (BioRad) or 12% polyacrylamide gel. Proteins were transferred to a PVDF membrane (Whatman) and the membrane was blocked with 2–5% skim milk in TBS-T (10mM Tris HCl, pH 7.5, 100mM NaCl, 0.1% Tween 20). The primary antibodies were used at the dilutions of either 1:5,000 for anti-mCherry or 1:10,000 for the affinity-purified anti-NlpD antibody. To reduce non-specific binding, the anti-mCherry primary antibody solution was pre-incubated with 100μL of TB28 (wt) cell lysate (OD₆₀₀ of ~15) prior to application to the membrane. The secondary goat anti-rabbit antibody conjugated to horseradish peroxidase (Rockland) was used at the dilution of 1:40,000, and the blot was developed using the Super Signal West Pico system (Pierce) according to the manufacturer’s protocol. Chemiluminescence was detected using a BioRad Chemidoc system.

Transposon mutagenesis

Mutagenesis of a ΔenvC strain was inefficient, most likely due to the cell division defects displayed by these cells. Instead, an EnvC depletion strain TB140/pMT187 [ΔenvC/Plac::envC-LE] was mutagenized with the Ez-Tn5 <Kan-2> transposome (Epicentre) as previous described [29]. Mutants were selected at 37°C on minimal M9 containing 50 μg/ml kanamycin, yielding a library of ~ 50,000 independent transposon insertions. Growth at 37°C resulted in the loss of the pMT187 plasmid (through combined effects of its temperature-sensitive origin of replication and cleavage by I-Sce-I [33]. The resulting ΔenvC mutant library was pooled as a slurry, and aliquots were stored frozen at -80°C.

Flow cytometry analysis

The cell chaining phenotype of different strains was assessed using flow cytometry. Cells were harvested by a low speed centrifugation (2,500 x g for 2 min), resuspended in PBS to an OD₆₀₀ ~ 0.1–0.2 and filtered through a 70 μm filter before analysis. Flow cytometry analysis was carried out on a MoFlo Astrios EQ (Beckman Coulter) using a 70μm nozzle, at the Division of Immunology’s Flow Cytometry Facility at Harvard Medical School. Flow cytometry histograms of cells (~ 100,000 cells for each sample) with increasing side scatter pulse width (SSC-W) were plotted using FlowJo.

Flow cytometry-based enrichment

For the control strains, overnight cultures were diluted in LB medium and grown at 30°C to an OD₆₀₀ of 0.2–0.3. For the transposon mutagenized ΔenvC library, a frozen aliquot was thawed, back-diluted twice in LB and grown at 30°C to an OD₆₀₀ of 0.3. The cells were then harvested by a low spin centrifugation (1,000 x g for 1–2 min), resuspended in PBS to an OD₆₀₀ ~ 0.1–0.2 and filtered through a 70μm filter before analysis and sorting. Cell resuspensions were analyzed and sorted with a 100μm nozzle on the MoFlo Astrios EQ (Beckman Coulter) described above. Gates for sorting were defined on a side scatter pulse height (SSC-H; y-axis) versus side scatter pulse width (SSC-W; x-axis) dot plot. Sorted cells were plated on minimal M9-maltose plates and incubated at 37°C overnight. Mucoid colonies were purified on LB, the chaining phenotype confirmed by phase contrast microscopy, and the sites of the transposon insertions were identified by arbitrarily primed PCR followed by sequencing [29].

Bioinformatic analysis of YraP and NlpD distribution

Primary protein sequences of E. coli YraP and NlpD were used as queries in a BLASTP search against the 201602 update of NCBI nr database. The E-value cutoff for hits was set at 1.00 x 10⁻²⁰. The phylogenetic tree was produced via PhyloT webtool (http://phylot.biobyte.de), using an abridged version of the NCBI Representative Genomes database. The tree and the plots of YraP and NlpD distributions were then rendered with iToL webtool (http:://itol.embl.de).