Contact-dependent growth inhibition requires the essential outer membrane protein BamA (YaeT) as the receptor and the inner membrane transport protein AcrB

Abstract

Contact-dependent growth inhibition (CDI) is a phenomenon by which bacterial cell growth is regulated by direct cell-to-cell contact via the CdiA/CdiB two-partner secretion system. Characterization of mutants resistant to CDI allowed us to identify BamA (YaeT) as the outer membrane receptor for CDI and AcrB as a potential downstream target. Notably, both BamA and AcrB are part of distinct multi-component machines. The Bam machine assembles outer membrane beta-barrel proteins into the outer membrane and the Acr machine exports small molecules into the extracellular milieu. We discovered that a mutation that reduces expression of BamA decreased binding of CDI+ inhibitor cells, measured by flow cytometry with fluorescently labelled bacteria. In addition, alpha-BamA antibodies, which recognized extracellular epitopes of BamA based on immunofluorescence, specifically blocked inhibitor-target cells binding and CDI. A second class of CDI-resistant mutants identified carried null mutations in the acrB gene. AcrB is an inner membrane component of a multidrug efflux pump that normally forms a cell envelope-spanning complex with the membrane fusion protein AcrA and the outer membrane protein TolC. Strikingly, the requirement for the BamA and AcrB proteins in CDI is independent of their multi-component machines, and thus their role in the CDI pathway may reflect novel, import-related functions.

Figure 1. Mutations in acrB predominate amongst CDI^R mutants

(A). The transposon locations in ten CDI^R mutants are shown, with six unique insertion sites identified. (B). After transduction into E. coli MC4100, transposon mutants (two of ten are shown) and a ΔacrB mutant (RAM1279) were incubated with DL4608 CDI⁺ inhibitor cells at a 20:1 inhibitor to target ratio for 3 h, and cell numbers were determined by plating on LB-kan medium. Squares and circles represent data from two separate experiments, each with a wild-type control. The time course in Fig. 1B demonstrates that near maximal inhibition is achieved by two hours, therefore our remaining CDI experiments are limited to this time point.

Figure 2. Analysis of the effects of the bamA101 mutation on CDI resistance and BamA expression

(A). Complementation of the bamA101 mutant with bamA. E. coli MC1061 with the indicated genotypes were assessed for CDI resistance as described in Experimental procedures. Cell numbers were quantitated prior to addition of CDI⁺ inhibitor cells DL4577 (0 h), and after 2 h of incubation (2 h). Both the MC1061bamA101 and MC1061bamA⁺ isolates contained plasmid pZS21amp whereas bamA101(bamA⁺) contained plasmid pZS21amp-bamA⁺. (B). Effect of bamA101 on expression of BamA and other OMPs. Expression of BamA and partly degraded BamA (BamA*) (upper panel), and DegP, LamB, OmpA, BamD in E. coli JCM158 and mutant derivatives (lower panel) was determined by immunoblot analysis. Lanes (1), wild-type; (2), Δwzb; (3), bamA101; (4), bamA101 Δwzb; (5), bamD5; (6), bamD5 Δwzb. (C). AcrA and TolC are not required for CDI. MC4100 ΔacrA (RAM1277), ΔacrB (RAM1279), and tolC-(CAG12184) target cells were incubated with CDI⁺ inhibitory E. coli (DL4608) at a 20:1 inhibitor to target cell ratio. Viable target cell counts were determined after 3 h. MC4100 (“wild-type”) was included as a positive control for sensitivity.

Fig. 3. Evidence that BamD, SurA, and BamB are not required for CDI

(A). BamA levels were determined by immunoblot for E. coli samples containing mutations in genes coding for different outer membrane proteins. E. coli JCM158 with the indicated mutations were mixed with CDI⁺ E. coli DL4577 (B) or CDI⁻ DL4956 mock inhibitor cells (C) for 2 h, and target cell number were determined (Experimental procedures). The first bar for each E. coli isolate shows target cell number prior to addition of inhibitor cells and the second bar shows cell number after 2 h incubation.

Fig. 4. The POTRA-3 domain of BamA is not required for CDI

(A). E. coli JCM158 with the indicated mutations (shown above the bars) and plasmids (shown below the x-axis) were mixed with CDI⁺ E. coli DL4577 and target cell numbers were determined after 2 h. The E. coli isolates used in (A) were analyzed in parallel by immunoblotting for BamA (panel B) and other outer membrane proteins (panel C) as described (Experimental procedures). Lane numbers in panels B and C correspond in order from left to right to the samples analyzed in panel A.

Fig. 5. Analysis of the effects of bamA101 on expression of BamA at the cell surface

To detect expression of BamA at the cell surface, E. coli were fixed in formalin, which blocks antibody entry into cells, or methanol, which permeabilizes cells to antibodies. Antibodies to full-length BamA (α–BamA) or to the POTRA domains of BamA (α–POTRA) (indicated at left), were incubated with either formalin- or methanol-fixed cells (indicated at right) followed by an AlexaFluor 488-labeled secondary antibody as described in Experimental procedures. Left panels are phase contrast images, right panels are fluorescence images. (A). JCM158 bamA⁺ control, (B). DL5503 bamA101.

Fig. 6. Analysis of the effects of target cell CDI^R mutations on inhibitor-target cell binding

The effects of target cell acrB and bamA mutations on binding to CDI⁺ inhibitor cells was assessed by mixing GFP-labeled DL4905 CDI⁺ inhibitor cells with DsRed-labeled target cells (1 inhibitor with 4 targets) for 15 min, then scanning 10⁵ bacteria by FACS (Experimental procedures). Boxed areas depicted show gating boundaries for GFP⁺ inhibitory cells (“I”), DsRed-labeled target cells (“T”), and cell aggregates (“A”) containing at least one inhibitory and one target cell. See Table II for data quantitation. (A). JCM158, (B). DL4905 GFP⁺ CDI⁺ inhibitor cells, (C). DL5529 (JCM158 DsRed⁺), (D). DL4095 + DL5529 (bamA wild-type), (E). DL4905 + DL5530 (bamA101), (F). DL4905 + DL5542 (acrB::kan).

Fig. 7. α-BamA antibodies block inhibitor-target cell binding

E. coli target cells (BamA⁺) labeled with DsRed (DL5529) were incubated with purified antibodies from (A) pre-immune, (B) α-BamA POTRA, and (C) α-BamA sera, then mixed for 15 min with GFP-tagged CDI⁺ inhibitor cells (DL4905) and analyzed by FACS (Experimental procedures). Cell gating boundaries are the same as for Fig. 6.

Fig. 8. α-BamA antibodies inhibit CDI

(A). α-BamA antibodies block CDI. Antibodies against whole BamA (α-BamA), the POTRA domains of BamA (α-POTRA), Imp, (α-Imp), and pre-immune serum (Preimmune) were purified and used to determine if they protect E. coli DL5422 from CDI as described in Experimental procedures. A bamA101 strain (DL5430) was included as a resistant control. CDI protection (y-axis) is expressed as the fold protection of DL5422 against CDI measured for each antibody preparation and the bamA101 resistant control compared with a no antibody addition control. (B). α-BamA antibodies do not induce OmpA misfolding. Samples were pre-incubated with either 1 = PBS, 2 = Pre-Imm, 3 = α–POTRA, or 4 = α–BamA for two hours and loaded onto an SDS-PAGE gel either with or without heat denaturation. Western blots were incubated with α–OmpA antiserum. In the unheated samples (25°C), OmpA is in a folded conformation (OmpA-F). Folded OmpA migrates faster than its unfolded conformation (OmpA-U), visible upon heat denaturation (100°C).

Fig. 9. Evidence that the inhibition of CDI by α-BamA is specific

(A). Immunoblot analysis of E. coli DL5432 with α-Imp antibodies. Analysis was performed as described in Fig. S3/Experimental procedures using inner membrane plus soluble protein (lane 1) and outer membrane (lane 2) fractions. α-Imp is reactive with Imp and OmpA (arrows), and an unknown protein (asterisk). (B). Immunofluorescence was carried out using formalin fixed DL5432 cells incubated with α-Imp and α-BamA antibodies under the same conditions as described in Fig. 8A. The control was treated identically except that primary antibody was not added. Left panels are phase contrast images and right panels are fluorescence images.

Figure 10. Effect of the bamA101 allele on CDI⁺ inhibitor cells

E. coli with the indicated genotypes (x-axis), carrying either cosmid vector pWEB::TNC (CDI⁻) or cosmid pDAL660Δ1–39 (CDI⁺) (see Fig. S1), were used as inhibitor cells against E. coli DL4372 targets as described in Experimental procedures. Viable cell target numbers were measured at “0” time and after 2 h incubation for each sample.