IgaA negatively regulates the Rcs Phosphorelay via contact with the RcsD Phosphotransfer Protein

Abstract

Two-component systems and phosphorelays play central roles in the ability of bacteria to rapidly respond to changing environments. In E. coli and related enterobacteria, the complex Rcs phosphorelay is a critical player in the bacterial response to antimicrobial peptides, beta-lactam antibiotics, and other disruptions at the cell surface. The Rcs system is unusual in that an inner membrane protein, IgaA, is essential due to its negative regulation of the RcsC/RcsD/RcsB phosphorelay. While it is known that IgaA transduces signals from the outer membrane lipoprotein RcsF, how it interacts with the phosphorelay has remained unknown. Here we performed in vivo interaction assays and genetic dissection of the critical proteins and found that IgaA interacts with the phosphorelay protein RcsD, and that this interaction is necessary for regulation. Interactions between IgaA and RcsD within their respective periplasmic domains of these two proteins anchor repression of signaling. However, the signaling response depends on a second interaction between cytoplasmic loop 1 of IgaA and a truncated Per-Arndt-Sim (PAS-like) domain in RcsD. A single point mutation in the PAS-like domain increased interactions between the two proteins and blocked induction of the phosphorelay. IgaA may regulate RcsC, the histidine kinase that initiates phosphotransfer through the phosphorelay, indirectly, via its contacts with RcsD. Unlike RcsD, and unlike many other histidine kinases, the periplasmic domain of RcsC is dispensable for the response to signals that induce the Rcs phosphorelay system. The multiple contacts between IgaA and RcsD constitute a poised sensing system, preventing potentially toxic over-activation of this phosphorelay while enabling it to rapidly and quantitatively respond to signals.

Fig 1. Signaling via the Rcs Phosphorelay.

A. The six proteins of the Rcs Phosphorelay are shown schematically (not to scale; described in detail in [1]). RcsF (orange) is positioned in the outer membrane, associated with outer membrane porins (OMPs; grey). Most described treatments that induce the phosphorelay require RcsF for activation and thus it is shown as a key sensor for both outer membrane stress (represented by a gold lightning bolt) and periplasmic or peptidoglycan stress (dark blue lightning bolt). IgaA (blue) is a five-pass inner membrane protein that serves as a brake on the phosphorelay; it communicates with RcsF across the periplasm. Current models suggest that upon stress signaling, RcsF increases or changes contacts with IgaA, leading to de-repression of the phosphorelay. RcsC (gold) is induced to autophosphorylate and pass phosphate from its active site His 479 to its REC domain Asp 875. The phosphate is then passed to His 842 on the RcsD (dark grey) histidine phosphotransfer domain, and from there it passes to the RcsB (crimson) REC domain Asp 56. Phosphorylated RcsB forms homodimers or heterodimers with RcsA (purple) to regulate many genes. Shown here are induction of capsule synthesis by the RcsB/RcsA heterodimer and induction of the sRNA RprA by the RcsB homodimer. The red highlight around rprA indicates that an rprA promoter fusion to mCherry (P_rprA-mCherry) is used throughout this work to evaluate activation of the phosphorelay. Note that as with many phosphorelays of this family, phosphate can also flow in reverse from RcsB towards RcsC. IgaA is shown closest to RcsD, based on data presented in this study. In this schematic, RcsC and RcsD are depicted as homodimers, but their state is not currently known. B. The promoter of the sRNA RprA was fused to mCherry to create a reporter for Rcs activation (P_rprA-mCherry), that demonstrates sensitivity and a wide dynamic range. Activity of wild type cells (black, EAW8) was compared to rcsC::Tn10 (blue, EAW18), rcsD541 (green, EAW19), rcsB::kan (red, EAW31) and ΔrcsF::cat (orange, EAW32). All strains were also tested with polymyxin B nonapeptide (PMBN) at 20 μg/ml. Cells were grown in MOPS minimal glucose for the P_rprA-mCherry assay; signal shown is for cells at a density of OD₆₀₀ 0.4. Details of the assay and cell growth are shown in S1A, S1B and S1C Fig and described in Materials and methods.

Fig 2. Interaction of IgaA with RcsD.

A. Beta-galactosidase activity was measured in cyaA deficient cells (strain BTH101) containing a dual plasmid system encoding the T18 and T25 domains of adenylate cyclase fused to proteins of interest and the expression measured compared to background. Each protein fusion plasmid paired with its cognate vector (V) produces very little activity; the three controls were averaged and used as “background” for normalization. Error bars (some too small to be visible) represent standard deviation of three assays. Fusions present are IgaA-T25, RcsD-T18, and RcsC-T18. Beta-galactosidase activities, measured in Miller units, results obtained with the fusions in the opposite orientation (IgaA-T18, RcsD-T25, RcsC-T25), and in strains mutant for rcs genes to test roles of other Rcs proteins on the interaction, are shown in S2A–S2D Fig. B. Relative ratio of RcsD fragment binding to IgaA was determined by comparing the interaction of RcsD truncations to the interaction between full length RcsD and IgaA, normalized to 1 (top black bar). Extent of RcsD present is shown schematically next to the bar graph. The dotted line at y = 0.2 represents the threshold set here for reliable interaction detection, 4x over background signal. This data is compiled from separate sets of assays, each normalized relative to the IgaA/RcsD signal in that experiment. In most cases the IgaA/RcsD interaction is 20x over background, usually 1000 Miller units compared to 50 Miller units for the background control. All measurements were carried out in strain BTH101. Plasmids used were pEAW1 (IgaA-T18), pEAW8 (RcsD-T25), pEAW8b (RcsD_1-683-T25), pEAW8α (RcsD_1-522-T25), pEAW8m2 (RcsD_1-461-T25), pEAW8m (RcsD_1-383-T25), pEAW8peri (RcsD_Δ45-304-T25), and pEAW8s (RcsD_326-C-T25). C. Interactions of IgaA with RcsD derivatives expressing the PAS-like domain. A Lactose MacConkey plate with Ampicillin and Kanamycin was streaked with BTH101 co-transformed with T18 and T25 plasmids and incubated for two days at 30°C. RcsD-T25 plasmids: RcsD (pEAW8), RcsD_1-461 (pEAW8m2), RcsD_Δ45–304 (pEAW8peri), RcsD_{1-461, Δ45–304} (pEAW8m2peri); IgaA-T18 plasmid (pEAW1). Expression of the RcsD-T25 proteins is shown in S2J Fig. Positive interactions are red.

Fig 3. Activity of truncated RcsD proteins.

A. RcsD wild type (upper bar graph, EAW8) and rcsD541 mutant (lower bar graph, EAW19) cells carrying the P_rprA-mCherry fusion were transformed with pBAD24-derived plasmids encoding RcsD or C-terminally truncated pieces of RcsD, grown in MOPS 0.2% glucose with ampicillin (-arabinose) or MOPS glycerol with ampicillin with 0.02% arabinose (+ arabinose). Relative fluorescence values for each strain are shown at OD₆₀₀ 0.4, compared to the WT strain with vector. RcsD truncations used are shown at the top of the figure, with color-coding: black: V (vector, pBAD24); blue: RcsD (pEAW11); brown: RcsD_1-383 (pEAW11m); green: RcsD_1-461 (pEAW11m2). Note that the first two bars in the lower graph, to the left of the vertical dotted line, are in the rcsD⁺ host, not rcsD541, to allow comparison of the rcsD⁺ and rcsD541 strains with the vector. Fluorescence as a function of OD₆₀₀ and additional related plasmids in the same strains are shown in S3A Fig; results in other strain backgrounds are shown in S3B and S3C Fig. B. Experiments as in panel A, but with plasmids carrying truncations of the N-terminus of RcsD, in rcsD⁺ (Rcs wild type; EAW8) and rcsD541 (EAW19) hosts. The constructs are color-coded as follows: black: vector (pBAD24), blue: full length RcsD (pEAW11), cyan: RcsD_Δ45–304 (pEAW11peri), green: RcsD_326-C (pEAW11s), orange: RcsD_686-C (pEAW11c), purple: RcsD_792-C (pEAW11d). Note that for the rcsD541 cells carrying RcsD_Δ45–304, the value shown is at a low OD, the total achieved within 6 hours. C. Cultures of EAW8 (Rcs wild type, carrying the P_rprA-mCherry fusion) transformed with three of the plasmids tested in panels A and B, RcsD (blue), RcsD_Δ45–304 (cyan), and RcsD_1-461 (green), as well as RcsD_{1-461, Δ45–304} (pEAW11m2peri, red) were grown under four conditions, with and without arabinose (as for A and B) and with and without PMBN (40μg/ml). Cultures grown without arabinose were grown in 0.2% glucose. D. Chart summarizing tests of RcsD fragments, their interaction with IgaA and ability to stimulate expression of the P_rprA-mCherry reporter in the absence of an inducing signal. Results from Figs 2B and 3A–3C and S3A Fig. nt: Not tested.

Fig 4. An RcsD mutation that blocks Rcs induction by increasing IgaA interaction.

A. RcsDT411A does not respond to Rcs stimuli. Schematic shows domains of RcsD and position of T411A, within the PAS-like domain. Both wild type and rcsDT411A strains (EAW8 and EAW121) were treated with nothing (-), 20 μg/ml polymyxin B nonapeptide (P₂₀), 5μg/ml MreB inhibitor A22 (A₅) or 0.3μg/ml Mecillinam (M_0.3). Both A22 and Mecillinam give a smaller dynamic range of wild type signaling than PMBN. B. RcsD PAS-like domain mutation T411A interaction with IgaA missing the periplasmic domain. IgaA schematic includes yellow transmembrane domains (TM), amino acid numbering, and loops labeled with their localization. BACTH results are shown as ratios relative to the wild type IgaA/RcsD interaction, which gave 1743 Miller units in this experiment. Plasmids used: IgaA-T18 (pEAW1); IgaA_Δ36-181-T18 (pEAWcyt1); IgaA_Δ263-330-T18 (pEAW1cyt2); IgaA_Δ384–649 (pEAW1peri); RcsD-T25 WT (pEAW8) and RcsD T411A (pEAW8T). Background controls and fold above background values are shown in S5B Fig. C. RcsD and RcsD T411A plasmids in cells carrying the rcsD541 mutation and chromosomal igaA deletions. Strains were plated directly after transformation on MOPS ampicillin glucose plates and incubated overnight at 37°C. Rcs⁺ strains devoid of IgaA activity are unstable and thus cannot be purified or assayed in liquid culture. The left plate contains control strains (clockwise from top left quadrant) rcsD541 with pBAD vector, showing moderate level of fluorescence as expected for a rcsD mutant (EAW19 with pBAD24), rcsD541 with RcsD⁺ on a plasmid, showing low level of fluorescence for a complemented (wild-type Rcs) strain (EAW19 with pEAW11), a rcsD541 ΔigaA strain with the RcsD⁺ plasmid, showing very mucoid growth associated with loss of IgaA; these cells generally will not form colonies on restreaking (EAW95 with pEAW11) and rcsD541 ΔigaA with pBAD vector, showing the same moderate fluorescence in the absence of RcsD (EAW95 with pBAD24). Note that mucoidy scatters the mCherry fluorescence, making it appear lower than the actual output. Right panel and inset, strains carrying indicated igaA deletions in the chromosome in an rcsD541 background, transformed with plasmids expressing either RcsD T411A (pEAW11T) or RcsD⁺ (pEAW11). The inset shows bright streaks within EAW95+pEAW11; this mucoid primary transformant spontaneously generates non-mucoid rcs mutants. Many of these mutants are not nulls, and the loss of mucoidy increases the apparent fluorescence. Therefore, these show up as more brightly fluorescent spots within the mucoidy. The brightly fluorescent sectors on this plate are bright because they are non-mucoid, reflecting a significant decrease in Rcs signaling but are still signaling at a level that is clearly much higher than in the dark, non-mucoid controls on the left-hand plate. Strains used, clockwise from top, two sectors for each strain, first with RcsDT411A, second with RcsD⁺: rcsD541 ΔigaA (EAW95); rcsD541 igaAΔperi (EAW98); rcsD541 igaAΔcyt2 (EAW97); rcsD541 igaAΔcyt1 (EAW96).

Fig 5. The RcsC periplasmic region is dispensable for polymyxin B nonapeptide (PMBN), A22, and mecillinam-induced signaling.

The top panel shows a schematic of RcsC with domains, topology and active site residues noted. A. PMBN induction in various rcsC mutations. Strains included are (L to R) WT: EAW8, rcsB::kan: EAW31, ΔrcsC: EAW91, rcsCH479A, mutant in the active site histidine: EAW92, rcsC_326-C: EAW56, rcsC_Δ48–314: EAW70 and rcsF::cat rcsC_Δ48–314: EAW85. B. The effect of three Rcs stimulating drugs, PMBN, A22 and mecillinam (P₂₀, A₅, M_0.3) on WT and RcsC_Δ48–314. The RcsC periplasmic deletion strain has a lower basal level of signal than WT here; this was also seen with other RcsC periplasmic deletions (S6B Fig).

Fig 6. Proposed interactions of IgaA and RcsD.

In this model, extensive interactions in the periplasm and in the cytoplasm are shown between IgaA and RcsD, consistent with genetic data indicating that signals pass from one compartment to the other via IgaA. RcsF is shown as in Fig 1A, poised to sense outer membrane disturbance and available to contact IgaA to change the course of signaling. Anchoring interactions between RcsD and IgaA in the periplasm contribute to the BACTH interaction signal and are required for IgaA repression of signaling. Interaction of the IgaA cytoplasmic loop 1 (Cyt1) and the PAS-like domain of RcsD (shown in lighter grey) are suggested to comprise the signal-switching interaction, tightened in rcsDT411A, an allele that blocks induction. Some aspects of this model are not known. It is unknown if RcsD acts as a dimer as shown; it is also unknown if RcsF and RcsD interact with separate (as shown) or overlapping parts of the periplasmic loop of IgaA.