Feedback Inhibition in the PhoQ/PhoP Signaling System by a Membrane Peptide

1Many prokaryotes inhabit multiple niches with disparate environmental conditions and challenges for proliferation. Not surprisingly, they have evolved a plethora of regulatory circuits that enable them to adapt to these environments. One important and extensively studied example is the signaling system controlled by the sensor kinase PhoQ and response regulator PhoP, which is found in enterics such as Escherichia coli, Salmonella typhimurium, and related bacteria. This two-component system is activated by signals such as low Mg2+, low pH, or the presence of antimicrobial peptides, and leads to expression of genes that encode, among others, Mg2+ transporters, enzymes that modify the cell envelope and confer resistance to cationic antimicrobial peptides –, enzymes that alleviate stress associated with low pH –, and other factors that regulate virulence in numerous gram-negative pathogens of humans, insects, and plants –.

2The sensor kinase PhoQ is an integral membrane protein whose periplasmic domain is involved in signal detection. Signal transduction occurs via PhoQ autophosphorylation, phosphotransfer to PhoP, and PhoQ-mediated dephosphorylation of phospho-PhoP. Magnesium sensing appears to be mediated by an acidic patch in the periplasmic domain of PhoQ, which structural data suggests is proximal to the membrane –. This acid pocket may also play a role in antimicrobial peptide sensing via displacement of divalent cations by these positively charged peptides.

3In many cases, adaptive regulatory systems have some form of negative feedback to modulate the cellular response. Indeed, negative feedback is a common theme in cell regulation and its role in maintaining homeostatic control is well-established,. Negative feedback can also play a similar role in reducing cell-to-cell variability within a population and can increase the activation kinetics in some circuits,. More generally, negative feedback, when combined with additional layers of regulation, may produce complex dynamics or process multiple input signals.

4Since the PhoQ/PhoP system functions as a critical stress response circuit for survival under conditions of low magnesium or in the presence of antimicrobial peptides, we hypothesized that there may be sources of negative feedback in this circuit. We would expect that such a negative feedback loop would have at least one component that is regulated by PhoP. Microarray and sequence analysis indicate PhoP influences transcription of a large set of genes,,. However, relatively few genes have been shown to be directly regulated by PhoP,,, and few of these have known functions. From this short list, none appeared to be obvious candidates for mediators of negative feedback. We therefore screened a reporter strain containing deletions of different PhoP-regulated genes for evidence of increased PhoQ/PhoP-dependent transcription. From an analysis of seven deletion strains, we found that mgrB (yobG), which is predicted to encode a 47 amino acid peptide of unknown function, plays a critical role in regulating the PhoQ/PhoP pathway.

5To identify factors that mediate negative feedback in the PhoQ/PhoP circuit, a set of genes that have been shown to be directly regulated by PhoP in E. coli, were individually deleted in a PhoP transcriptional reporter strain. The reporter contained the PhoP-regulated mgrB promoter driving YFP expression (PmgrB-yfp), which was integrated at the phage lambda attachment site. The strains also expressed CFP from a constitutive promoter, which served as an internal control for protein expression and fluorescence intensity. We chose the mgrB transcriptional reporter because it gives the highest level of fluorescence from among a collection of fluorescent reporters of PhoP-regulated transcription. As seen in cells growing on LB agar plates, which are moderate inducing conditions for the E. coli PhoQ/PhoP system, deletion of mgtA, rstA, nagA, slyB, vboR, and yrbL had no visible effect on YFP fluorescence (Figure 1A). Deletion of mgrB, however, resulted in a clear increase in YFP with no visible difference in CFP fluorescence when compared with the wild-type strain (Figure 1A). To obtain a quantitative measure of the effects of the gene deletions on transcription levels, we measured the ratio of YFP to CFP fluorescence in single cells by fluorescence microscopy and image analysis. Cells were grown in medium containing either 100 µM or 10 mM Mg2+. Consistent with the results observed on agar plates, deletions of mgtA, rstA, nagA, slyB, vboR, and yrbL showed no effect on the YFP/CFP fluorescence (the example of slyB is shown in Figure 1B, other examples are shown in Figure S1). Deletion of mgrB, on the other hand, resulted in roughly a 9-fold increase in YFP/CFP fluorescence at 10 mM Mg2+ and a 3-fold increase at 100 µM Mg2+, relative to the corresponding wild-type strain (Figure 1B). The observed increase in fluorescence was strictly dependent on PhoQ (Figure 1B). Deletion of mgrB in fluorescent transcriptional reporter strains for three other PhoP-regulated promoters, PmgtA, PphoPQ, and PhemL, resulted in similar increases in fluorescence (Figure 1C and 1D, and data not shown), suggesting that the effect of the deletion is likely to be at a point upstream in the pathway common to all of these genes, i.e. at some point in PhoQ/PhoP signaling. We also verified that the mgrB deletion could be complemented by a plasmid containing mgrB cloned downstream of the trc promoter (Figure 2). When this same plasmid was put in an mgrB + strain, it resulted in a further repression of reporter gene transcription (Figure 2), presumably due to increased expression of MgrB above the wild-type level. Taken together, the above results suggest that the mgrB gene product acts as a repressive factor in PhoQ/PhoP signaling.

6The mgrB gene consists of a small open reading frame of 141 base pairs. A recent study of small open reading frames in E. coli verified protein expression from the wild-type locus of an epitope-tagged mgrB. The 47 amino acid MgrB peptide could potentially have a type I (secretion) or type II (lipidation) signal sequence,. Alternatively, the N-terminal hydrophobic stretch may function as a transmembrane domain (Figure 3A). To determine the localization of MgrB, we analyzed cell envelope and soluble protein fractions by Western blotting with rabbit antisera raised against a peptide from the C-terminus of the protein. We were able to detect a protein between 4 and 7 kilodaltons in the whole-cell lysate and envelope fraction of cells expressing mgrB, but not in the soluble protein fraction (Figure 3B). The envelope fraction of cells expressing MgrB showed no detectable contamination with cytoplasmic or periplasmic proteins as assessed by Western blots for the cytoplasmic proteins CFP and YFP and the periplasmic protein beta-lactamase (top two panels of Figure 3B).

7To further confirm membrane association, we constructed a fusion of GFP to the N-terminus of MgrB. The fusion is similar to wild-type MgrB in its ability to complement a deletion, as indicated by repression of PhoQ/PhoP signaling (Figure S2). Western blots show no evidence of cleavage or degradation of the fusion protein (data not shown). The fact that GFP is not cleaved from MgrB by a signal peptidase suggests that nascent MgrB is not secreted into the periplasm or lipidated. Fluorescence microscopy of cells expressing GFP-MgrB revealed a halo-like fluorescence at the cell boundary, suggesting localization to the envelope, whereas cells expressing GFP alone showed uniform fluorescence indicative of cytoplasmic localization (Figure 3C). When GFP is targeted to the periplasm through the Sec pathway, either as a secreted protein or as part of a membrane protein, it does not fold properly and fails to fluoresce,. The only reported pathway for producing fluorescent GFP in the periplasm is through the Tat secretion system,. However MgrB is not predicted to have a signal sequence that would target it for secretion through this pathway. Taken together, the above results indicate that MgrB is associated with the inner-membrane with its N-terminus in the cytoplasm. The size and hydrophobicity profile of the peptide further suggest that it spans the membrane with a single transmembrane domain.

8Given the localization of MgrB, we hypothesized that it exerts its effects through PhoQ. We therefore tested the action of MgrB on a chimera in which the periplasmic domain of PhoQ from E. coli was replaced with the highly divergent periplasmic domain of PhoQ from Pseudomonas aeruginosa, a bacterium that does not appear to possess an mgrB ortholog. We measured fluorescence of a phoQ− mgrB− reporter strain for PhoP-dependent transcription that was transformed with a plasmid expressing either E. coli PhoQ or the PhoQ chimera (PhoQchim), or a control plasmid. The strain also contained either a compatible plasmid expressing mgrB or a compatible control plasmid (Figure 4). We note that the fluorescence levels of the MgrB− PhoQ+ chim strain were lower than the corresponding levels for the MgrB− PhoQ+ strain, but they were significantly higher than the fluorescence levels of the PhoQ− strain. As with the complementation experiments above, introduction of the MgrB expression plasmid resulted in decreased fluorescence of PhoQ+ cells. In contrast, no change in fluorescence levels was observed when the plasmid was introduced into the PhoQ+ chim strain. That MgrB does not decrease PhoP-regulated transcription when the periplasmic sensor domain of PhoQ is modified suggests MgrB acts at or upstream of PhoQ in the signaling pathway.

9To look for evidence of an interaction between MgrB and PhoQ in E. coli, we used a bacterial two-hybrid assay based on split adenylyl cyclase. In this system, adenylyl cyclase activity is reconstituted when the T18 and T25 fragments of Bordetella pertussis CyaA are brought into close proximity. The resulting increase in cAMP levels is detected through expression of beta-galactosidase from the lac promoter. We fused the T18 and T25 fragments to the N-termini of MgrB and PhoQ, respectively. Based on the known topology of PhoQ, and the topology of MgrB (discussed above), both CyaA fragments should be in the cytoplasm. A strain expressing T18-MgrB and T25-PhoQ showed a significantly higher level of beta-galactosidase activity when compared with strains expressing the T18 and T25 fragments alone (Figure 5) or expressing the fusions to either MgrB or PhoQ individually (data not shown). These results suggest there is a physical interaction between MgrB and PhoQ. Furthermore, a strain expressing T18-MgrB and the T25- fragment fused to the N-terminus of PhoQchim showed a minimal increase in beta-galactosidase activity relative to the controls. This is unlikely to be due to a defect in PhoQchim expression because we were able to detect PhoQchim–PhoQchim interactions at levels comparable to those for (wild-type) PhoQ-PhoQ interactions, consistent with previous reports that both PhoQ and PhoQchim form functional complexes (Figure S3). Taken together, these results suggest that the periplasmic domain of E. coli PhoQ is important for the interaction with MgrB. Interestingly, co-expression of T18-MgrB and a fusion of T25 to the N-terminus of MgrB also showed significant beta-galactosidase activity relative to the controls (Figure 5), suggesting that MgrB may form dimers or higher order complexes.

10The mgrB gene, also known as yobG, has been identified in Salmonella enterica, where it has been shown to be activated by PhoP in microarray and transcriptional reporter experiments,. Like many other small open reading frames, mgrB is frequently missed in genome annotations. However we have been able to identify putative mgrB homologs in numerous species among several genera of Enterobacteriaceae. Specific examples and their alignments are shown in Figure 6. All of the genomes for which we could identify a candidate mgrB also contained phoP and phoQ orthologs. However, the converse does not appear to be the case: we were unable to find an mgrB homolog in Erwinia or Pseudomonas species, though members of both of these genera possess phoP and phoQ. In addition, although the endosymbiont Sodalis glossinidus has phoP and phoQ genes, its mgrB is unlikely to be functional due to the presence of two internal stop codons (Figure 6).

11To test whether the action of MgrB on PhoQ/PhoP signaling is conserved among several genera, we over-expressed the native MgrB from Salmonella enterica serovar typhimurium and Yesinia pestis in their respective host strains and measured transcription of the promoter for phoN, a PhoP-regulated gene from Salmonella. For both Y. pestis and S. typhimurium, MgrB over-expression resulted in a significant repression of reporter activity (Figure 7). We also verified that the mgrB genes from both of these organisms complement an mgrB deletion in E. coli (data not shown).

12Based on the results presented here, we propose that the peptide MgrB spans the inner-membrane and represses PhoP phosphorylation by inhibiting PhoQ kinase activity, stimulating phosphatase activity, or both (Figure 8). Our results further suggest that an interaction between MgrB and the periplasmic domain of PhoQ plays a critical role. Since mgrB transcription is activated by phosphorylated PhoP, MgrB is part of a negative feedback loop in the PhoQ/PhoP signaling circuit. It is particularly striking that deletion of mgrB results in a strong increase in PhoP-regulated transcription even for growth in 10 mM Mg2+–a condition that strongly represses PhoQ/PhoP signaling in wild-type cells. The four PhoP-regulated promoters that we tested (promoters for mgrB, phoPQ, mgtA, and hemL) show at least some level of magnesium responsiveness in an mgrB − strain (Figure 1B–1D and data not shown). The significant decrease in magnesium sensitivity of mgtA transcription in the absence of MgrB (Figure 1D) most likely indicates that mgtA has reached near-maximal levels of PhoP-activated transcription for the mgrB − strain growing in 10 mM Mg2+. Indeed, previous studies suggest that the E. coli mgtA promoter saturates at lower levels of PhoQ/PhoP stimulation when compared to the other promoters considered here. We have also found that PhoP-regulated transcription remains responsive to the antimicrobial peptide LL37 and to acidic pH in mgrB − strains (Figure S4 and Figure S5, respectively). We note, however, that mgrB deletion affects the fold-change in reporter gene expression for all three stimuli (Mg2+, pH, and LL37). Thus, MgrB modulates the magnitude and sensitivity of PhoQ/PhoP signaling but is not strictly required for magnesium, pH, and antimicrobial peptide responsiveness.

13Negative feedback has been shown to increase activation kinetics and minimize cell-to-cell variability in some regulatory circuits –. Such behavior can be readily understood for simple examples of feedback arising from negative autogenous control. However it is difficult to predict the effects of negative feedback in more complex circuits with additional regulators, which have the potential for time delays and additional sources of cell-to-cell variability. Indeed, deletion of MgrB has relatively little effect on the kinetics of fluorescent protein reporter accumulation (Figure S6) and on the cell-to-cell variability in fluorescent protein expression (Figure S7), despite its strong effect on the magnitude of PhoQ/PhoP signaling. We hypothesize that MgrB may provide a point of control for integrating additional input signals that modulate PhoQ/PhoP activity. For example, such signals could act by regulating the expression level or repressive action of MgrB (Figure 8).

14We have shown that for bacteria from at least three different genera, the native MgrB homolog represses PhoQ/PhoP signaling, suggesting that this mode of negative regulation is broadly conserved. At least two other proteins have been reported to regulate PhoQ/PhoP activity, however the level of functional conservation appears to be more limited for these cases. YneN (B1500) is a small integral-membrane protein that stimulates PhoQ in E. coli. This protein is a member of the EvgS/EvgA regulon and thus mediates cross-regulation between two signaling circuits. This mechanism for PhoQ regulation does not appear to be widely conserved; we have been unable to identify YneN orthologs in sequenced genomes other than those of various E. coli and Shigella isolates. Another protein, SlyB, has been shown to mediate some negative repression of the PhoQ/PhoP system in Salmonella typhimurium. Deletion of slyB resulted in a roughly 1.5-fold increase in transcription of multiple PhoP-regulated genes in this bacterium. However, we did not detect any difference in transcription of PhoP-regulated genes when we compared wild-type and slyB − strains of E. coli (Figure 1B). This may indicate another example of divergence in the PhoQ/PhoP regulatory circuit, even for these closely related species,.

15MgrB-mediated regulation of PhoQ is part of an emerging theme of modulation of membrane proteins by small hydrophobic peptides. It is a striking example of a small, easily overlooked open reading frame that plays a critical role in regulating a signal transduction pathway. Sequence alignments reveal several conserved residues, which may be critical for MgrB function (Figure 6). Further genetic and biochemical work will be required to understand the mechanism by which MgrB represses PhoQ, and to determine whether MgrB has additional roles as part of this important two-component signaling system.

16See Supporting Information for tables of strains (Table S1), plasmids (Table S2), and PCR primers (Table S3). A description of plasmid construction is given in Text S1.

17Deletions of PhoP-regulated genes in E. coli were transduced from strains in the Keio Collection by P1vir transduction. The deletions were confirmed by PCR using primers that flank the gene. When necessary, kanamycin resistance markers were removed with FLP recombinase by transforming with pCP20 and subsequent curing of the plasmid.

18E. coli strains were grown at 37°C, unless otherwise indicated, in Luria-Bertani (LB) medium (Difco - BD, Franklin Lakes, NJ) or minimal A medium supplemented with 0.2% glucose, 0.1% casamino acids (Difco) and with the indicated concentration of MgSO4. Salmonella strains were grown in LB at 37°C. Yersinia pestis strains were grown at 26°C in brain-heart infusion (BHI) broth (Difco). The lac and trc promoters were induced with isopropyl β-D-1-thiogalactopyranoside (IPTG) at a final concentration of 1 mM when indicated. When IPTG was not mentioned in the description of the culture conditions, the basal transcription from the trc promoter was used to drive expression.

19Strains were streaked onto LB Agar plates (Difco) and incubated overnight at 37°C. Images of YFP and CFP fluorescence were acquired with a home-built fluorescence illuminator as previously described.

20Single-cell measurements were performed essentially as described previously. Briefly, overnight cultures, grown in minimal A medium with 1 mM MgSO4, and 50 µg/mL ampicillin with or without 50 µg/mL spectinomycin to maintain plasmids when necessary, were diluted back 1∶1000 in pre-warmed minimal medium containing 100 µM, 1 mM, or 10 mM MgSO4 and grown to an OD600 between 0.2 and 0.3. Cultures were cooled quickly with an ice water slurry and streptomycin was added to a final concentration of 250 µg/mL to inhibit protein synthesis. Images were acquired and analyzed as previously described,.

21Envelopes were prepared with a protocol modified from. An overnight culture in LB medium with 50 µg/mL ampicillin was diluted 1∶1000 into the same medium supplemented with 1mM IPTG and shaken at 37 degrees for 5.5 hours at 250 rpm. The culture was split in half, chilled on ice, and then spun at 3,300 g for 10 minutes at 4°C. One pellet was saved at −20°C for the total lysate. The second pellet was resuspended in a 10 mL of 30mM Tris pH 7.8 and pelleted again. The pellet was then resuspended by vortexing in 200 µL of a cold 20% sucrose/30mM Tris solution. 20 µL of freshly prepared 10 mg/mL lysozyme in 0.1 M EDTA pH 7.0 was added and the solution was mixed by inverting at 4°C for 20 minutes. 3 mL of 3 mM EDTA pH 7.5 was then added, the solution was sonicated on ice, and then centrifuged at 47,000 g at 4°C for 75 minutes. The supernatant, which is the soluble fraction, was removed and saved. The pellet, which is the envelope fraction, was resuspended in 2 mM Tris pH 8.0. To bring the volume of supernatant down to the same concentration of cell equivalents as the envelope fraction, the supernatant was concentrated with a Speed Vac (Thermo Fisher, Waltham, MA).

22Samples were boiled in Tris-Tricine loading buffer with 0.2 M dithiothreitol for 5–7 minutes and loaded on 16.5% Tris tricine gels (ReadyGel - BioRad, Hercules, CA). Proteins were transferred to Immobilon-P PVDF (Millipore, Billerica, MA) followed by western blot analysis. A rabbit polyclonal MgrB antiserum was generated by using a KLH-conjugated synthetic peptide corresponding to residues 27–40 of the predicted MgrB protein sequence (GenScript, Piscataway, NJ). The peptide was synthesized with a C-to-S substitution for the first cysteine residue. Beta-lactamase and CFP/YFP were detected with rabbit polyclonal anti-beta-lactamase (Millipore,) and anti-GFP (A.v. Peptide Antibody – Clontech, Mountain View, CA) antibodies. A horseradish peroxidase-conjugated anti-rabbit antiserum (GE Healthcare, Piscataway, NJ) was used as the secondary antibody.

23Cultures of BTH101 bearing combinations of pKT25- and pUT18-derived plasmids were grown for 9 hours in LB supplemented with 100 µg/mL ampicillin and 50 µg/mL kanamycin. The cultures were then diluted 1∶1000 into medium supplemented with 1 mM IPTG and allowed to grow at 30°C for 14 hours. Cultures were grown at 30°C in order to increase complementation efficiency. Cells were cooled quickly in an ice slurry and kept on ice for 30 minutes. Beta-galactosidase assays were performed as described in using chloroform and SDS for permeabilization.

24Cultures of Salmonella were grown overnight in LB with 50 µg/mL ampicillin and 50 µg/mL spectinomycin. The next day, they were diluted 1∶1000 into the same medium and grown to OD600 0.1–0.3. Cells were permeabilized with chloroform and SDS and assayed as in. Cultures of Yersina were grown and assayed in the same fashion except the culture medium was BHI with 200 µg/mL ampicillin and 50 µg/mL spectinomycin and the overnight culture was diluted 1∶50.