Role of the Umo Proteins and the Rcs Phosphorelay in the Swarming Motility of the Wild Type and an O-Antigen (waaL) Mutant of Proteus mirabilis

Abstract

Proteus mirabilis is a Gram-negative bacterium that exists as a short rod when grown in liquid medium, but during growth on surfaces it undergoes a distinct physical and biochemical change that culminates in the formation of a swarmer cell. How P. mirabilis senses a surface is not fully understood; however, the inhibition of flagellar rotation and accumulation of putrescine have been proposed to be sensory mechanisms. Our lab recently isolated a transposon insertion in waaL, encoding O-antigen ligase, that resulted in a loss of swarming but not swimming motility. The waaL mutant failed to activate flhDC, the class 1 activator of the flagellar gene cascade, when grown on solid surfaces. Swarming in the waaL mutant was restored by overexpression of flhDC in trans or by a mutation in the response regulator rcsB. To further investigate the role of the Rcs signal transduction pathway and its possible relationship with O-antigen surface sensing, mutations were made in the rcsC, rcsB, rcsF, umoB (igaA), and umoD genes in wild-type and waaL backgrounds. Comparison of the swarming phenotypes of the single and double mutants and of strains overexpressing combinations of the UmoB, UmoD, and RcsF proteins demonstrated the following: (i) there is a differential effect of RcsF and UmoB on swarming in wild-type and waaL backgrounds, (ii) RcsF inhibits UmoB activity but not UmoD activity in a wild-type background, and (iii) UmoD is able to modulate activity of the Rcs system.

INTRODUCTION

Proteus mirabilis is a Gram-negative bacillus that exhibits a cooperative form of motility termed swarming. In liquid culture, P. mirabilis exists as peritrichously flagellated swimmer cells that are 1 to 2 μm in length. When placed on solid surfaces, the swimmer cells undergo physical and biochemical changes to form swarmer cells that are characterized by the following changes: (i) upregulation of flhDC, encoding the class I master regulator of the flagellar regulon, (ii) hyperflagellation, and (iii) a 20- to 50-fold increase in cell length while remaining multinucleated and aseptate (reviewed in references 22, 36, and 38). Elongated swarmer cells align parallel to each other, entangling their flagella to form a swarming raft (23). As a group, this raft radiates out from the central inoculum to form a ring of swarming. When an unknown signal is sensed, the cells consolidate, or dedifferentiate, back into swimmer cells. This process of swarming and consolidation repeats to form a characteristic bull's-eye pattern on an agar plate (39). How P. mirabilis cells recognize they are on a surface and change their gene expression profile accordingly is just beginning to be understood. Several mechanisms have been proposed, including the inhibition of flagellar rotation on surfaces and the accumulation of putrescine, which may act as a cell-to-cell signaling molecule to regulate gene expression (1, 40). Consistent with a role for flagellar inhibition, the addition of antiflagellar antibodies or agents that increase medium viscosity increase the rate of swarmer cell differentiation (3). Also, mutations in fliL, encoding a component of the flagellar basal body, result in swarmer cell differentiation under noninducing conditions (3). The mechanism by which FliL transmits a signal to regulate gene expression has not been defined (3). More recently, we proposed a role for the O antigen in sensing surfaces (35). We hypothesized that O antigen was needed to sense solid surfaces and to relay this signal to the cell through the Rcs phosphorelay.

Two-component regulatory systems (TCS) are one of the most common ways bacteria control gene expression in response to external signals. The canonical TCS consists of an inner membrane-bound sensor kinase, which dimerizes and autophosphorylates itself on a specific histidine, and a cytoplasmic response regulator, which receives this phosphate at an aspartate residue. Typically, the phosphorylated response regulator then binds DNA with greater affinity to regulate gene expression, although not all response regulators control gene expression (20). While TCS are important in bacterial physiology, their prevalence in individual bacteria can vary, from 0 TCS in Mycoplasma genitalium to 80 in Synechocystis species (33, 34). P. mirabilis is predicted to have 16 TCS, although only the Rcs and Rpp systems have been directly shown to influence motility (2, 10, 26, 37, 44).

The Rcs phosphorelay has been well studied in Escherichia coli and Salmonella enterica serovar Typhimurium (6, 9, 11, 15, 16, 21, 25, 27, 41, 43). The Rcs phosphorelay is more complicated than the canonical two-component system; along with the response regulator (RcsB) and sensor kinase (RcsC), it uses an outer membrane activator protein (RcsF) and a phosphotransfer protein (RcsD) (7, 18, 27, 29). A stimulus can be sensed through one of two pathways depending on the origin of the stimulus. If the signal originates externally, it can go through the outer membrane and RcsF, which relays the signal to RcsC. However, if the signal originates in the periplasm or cytoplasmic membrane, it proceeds directly to RcsC, which upon autophosphorylation of its His and Asp residues transfers the phosphate to a His residue on RcsD. In turn, RcsD then transfers the phosphate to the Asp residue on RcsB (25, 28, 29). The phosphorylated RcsB protein can bind DNA and act as either a repressor or an activator (28). More recently, another input, UmoB (IgaA), has been implicated in controlling the Rcs system (4, 5, 12, 13, 17, 30, 31, 42). The Rcs phosphorelay has been shown to respond to various stresses, such as those caused by perturbations in the cell envelope and peptidoglycan, or by osmotic stress (25, 45). The Rcs phosphorelay is important for motility in a variety of organisms, where the RcsB response regulator acts as a repressor of the master regulator flhDC (2, 10, 19, 27, 43). In P. mirabilis, mutations in the Rcs phosphorelay lead to increased swarming and result in differentiation to swarmer cells under normally nonpermissive conditions, such as liquid growth (2, 10, 26, 44).

Additional proteins that have been shown to regulate flhDC in P. mirabilis are the UmoA to -D proteins (13). UmoA to -D were discovered in a search for suppressors of the swarming defect in a flgN flagellar chaperone mutant using an overexpression library. It was shown that suppression by all four loci was due to upregulation of flhDC, and loss-of-function mutations in these genes caused a decrease in flhDC expression as well as a concomitant lack of swarming to various degrees (13). UmoB and UmoD exhibited the most severe phenotypes in terms of swarming and flhDC regulation (13). The cellular location of UmoD is unknown, but it is likely secreted into the periplasm. UmoB exhibits features of an integral membrane protein and is likely located in the cytoplasmic membrane (12). Independently, while looking for Salmonella mutants that could grow in fibroblast cells, Cano et al. discovered an UmoB homolog, which also regulates flhDC expression, that they termed IgaA (4, 5). Interestingly, loss-of-function mutations in S. Typhimurium igaA are lethal, while loss of umoB function is not (4). The lethality of an igaA loss-of-function mutation has necessitated the use of leaky alleles, such as igaA1, that retain some activity and has enabled suppressor mutations to be found that all map to the Rcs system (4, 5, 31). Using tagged Rcs components, an igaA1 mutation was shown to not have an effect on Rcs protein levels, indicating a posttranslational role for IgaA function (12). How IgaA regulates the Rcs phosphorelay is still not known.

Our previous study indicated a role for O antigen in surface sensing and the Rcs phosphorelay in transmitting a surface contact signal to transcriptional regulation (35). Mutations in the waaL gene (PMI3163, formerly rfaL), encoding O-antigen ligase, and wzz (PMI2183, formerly cld), encoding a chain-length determinant for O antigen, result in the failure to derepress the class I flagellar activator FlhDC on solid surfaces. Here we genetically dissected the Rcs system and showed that signaling upon surface contact is different in wild-type and O-antigen-minus cells. We have shown that, like IgaA, UmoB works through the Rcs system and that O antigen is needed for this. We also propose a role for UmoD in this signaling pathway through interactions with UmoB.

MATERIALS AND METHODS

Strains and media.

For cloning purposes, E. coli strain XL1 was used. For conjugal matings, E. coli strain SM10λpir (32) was used as the donor strain, and either PM7002 (wild type) or PM942 (waaL∷mini-Tn5) P. mirabilis strains were used as the recipients. E. coli and P. mirabilis were both grown in modified Luria-Bertani (LB) broth (10 g tryptone, 5 g yeast extract, 5 g NaCl per liter) with shaking at 37°C or on LB plates kept at 37°C. Swarm assays were performed on 1.5% agar plates with appropriate antibiotics. Antibiotics were used for selection at concentrations of 25 μg/ml for both chloramphenicol and streptomycin, 20 μg/ml of kanamycin, and 100 μg/ml for ampicillin for E. coli. Antibiotic concentrations for the selection of P. mirabilis were 100 μg/ml for chloramphenicol, 35 μg/ml for streptomycin, 20 μg/ml for kanamycin, 15 μg/ml for tetracycline, and 300 μg/ml for ampicillin.

Cloning.

rcsF and umoD were cloned into pACYC184 (8) and pBC (Stratagene) and expressed from the tet and lac promoters, respectively, using BamHI and SalI sites added to the genes through PCR. Each gene has its own ribosome-binding site. umoB and umoD were cloned into the multicloning site of pTrc99A and expressed from the lac promoter using BamHI and SalI sites added to the genes through PCR. Isopropyl-β-d-thiogalactopyranoside (IPTG) was not added for expression of these genes from pTrc99A, since the promoter was leaky and produced a hyperswarming phenotype. When both plasmids were expressed in a cell at the same time, double antibiotic selection with chloramphenical and ampicillin at the concentrations listed above was used. Table 1 contains a list of primers used.

Table 1.

Primers used in this work

Primer	Sequence (5′–3′)	Purpose
rcsB.for	GTACAGTCGACTCACCGACCTATCTATGCCT	Deletion mutant
rcsB.rev	GTACAGTCGACTCACCGACCTATCTATGCCT	Deletion mutant
rcsCint.for	ATCAAGGATCCAGAGCGTTCCATTTTAACACG	Deletion mutant
rcsCint.rev	ATGCTTAGTCGACATGCTTCACGCTTAGAGGAGC	Deletion mutant
rcsF.for	ATCAGGGATCCATTTGCATTAAATTAGGGC	Mutation/overexpression
rcsF.rev	ATGACATGTCGACATTCATTGAGTAATTAATAGTGC	Mutation/overexpression
umoB.for	ATCAGGGATCCATTGTTACTAAGCAACACC	Mutation/overexpression
umoB.rev	ATGACATGTCGACAGTAAACACATTGCCTTCC	Mutation/overexpression
umoD1A.for	ATCAAGGATCCTGGTGATAAAAGAGTGAAATCC	Mutation/overexpression
UmoD.rev	ATGACATGTCGACTATCAGTTATCAGCGTTAATGC	Mutation/overexpression

Construction of mutations.

For allelic replacement, the rcsF and umoB genes were cloned into pBC and digested with BglII and HindIII, respectively. The kanamycin cassette from pUC4KIXX was ligated into the gene at either the BglII or HindIII site after blunting the ends with T4 DNA polymerase. The disrupted genes were subcloned into pKNG101 (24) and maintained in E. coli SM10 λpir. The strains were mated with either PM7002 or PM942 waaL∷Km^r and selected on tetracycline and streptomycin to select for a Campbell-type insertion. It was possible to construct kanamycin insertions in these genes in the PM942 waaL∷Km^r background because pKNG101 encodes streptomycin resistance to select for plasmid integration and the use of different kanamycin resistance genes avoids recombination via these genes. In addition, pKNG101 contains the sacB gene, allowing sucrose selection to select for the second recombination event, which results in excision of the vector and either restoring the wild-type allele or leaving only the mutated allele. Allelic replacement was confirmed by Southern blotting (see below). Table 1 shows the primers used.

Campbell insertions were used to make mutations in rcsB and rcsC. An internal gene fragment was cloned into pKNG101 using restriction ends made by PCR at either the XbaI/BamHI or BamHI/SalI sites (Table 1 shows the primers used). The pKNG101 clones were electroporated into E. coli SM10λpir and then mated with either PM7002 or PM942 waaL∷Km^r. Exconjugants representing Campbell-type integration events that disrupted each gene were selected on LB plates with tetracycline and streptomycin, and mutations were confirmed by Southern blotting.

Southern blot analysis.

To confirm that the matings resulted in the appropriate gene disruptions, chromosomal DNA from the rcsB, rcsC, rscF, and umoB mutants was extracted and separately digested with EcoRI, SalI, HindIII, and EcoRI before being transferred to a nitrocellulose membrane and probed with a gene-specific digoxigenin-labeled probe.

Swarm assays.

To examine the swarming phenotype of the strains used in this study, the strains were grown overnight in LB medium with appropriate antibiotics at 37°C shaking. Fresh LB was used to equilibrate the optical density at 600 nm (OD₆₀₀) of the cultures. One-microliter drops of each culture were spotted on an LB plate without antibiotic in triplicate. Measurements of the swarming diameter were taken every 30 min starting at 2.5 h postinoculation. For all strains, the initial drop of culture before the initiation of swarming was 4 mm. The reported values represent the average diameter of triplicate samples during the first consolidation period with standard deviations. Each experiment was repeated 2 to 3 additional times with similar results.

Northern blot analysis.

Cells were grown overnight in LB. All samples were normalized to the same OD₆₀₀ with fresh LB. One-hundred-fifty-microliter drops were spread on 2% LB plates in parallel to produce cultures that were synchronously differentiating. The cells were collected from each plate 4 h after inoculation with LB medium and spun for 1 min at 12,000 rpm. Total RNA was isolated using the MasterPure RNA purification kit (Epicentre, Madison WI). Equal amounts of RNA were run on a 1.2% formaldehyde agarose gel, and transferred to a nitrocellulose membrane. A DNA probe specific to each gene was labeled with digoxigenin and used to examine transcript levels by chemiluminescence using the CDP-Star substrate (Roche Applied Science).

RESULTS

Differential effects of rcs mutations in wild-type and waaL∷Km^r backgrounds.

P. mirabilis strain PM942 waaL∷Km^r lacks O antigen and is unable to swarm due to the failure to derepress flhDC upon contact with solid surfaces (35). This swarming defect can be restored by a null allele in the rcsB gene, and it was hypothesized that in wild-type cells, O-antigen contact with a surface sends a signal that inactivates the Rcs phosphorelay by an unknown mechanism (38). To test if mutations in additional components of the Rcs phosphorelay could also suppress loss of swarming in the waaL background, mutations were constructed in the rcsC (sensor kinase) and rcsF genes. The RcsF outer membrane lipoprotein is a candidate for relaying the signal between O antigen and the Rcs phosphorelay (7, 14). As seen in Fig. 1A, the rcsB and rcsC mutations restored swarming to PM942 waaL∷Km^r with migration distances of 12 and 9 mm, respectively, compared to 0.3 mm for PM942. The rcsF mutation in the PM942 waaL∷Km^r background also restored swarming (4 mm), but these levels were below that conferred by the rcsB and rcsC mutations (Fig. 1A). The rcsB, rcsC, and rcsF mutations were also constructed in a wild-type PM7002 background for comparison purposes. The rcsB and rcsC mutations conferred a hyperswarming phenotype with migration distances of 26 and 22 mm, respectively, compared to 12 mm for wild-type PM7002 (Fig. 1B). This is consistent with previous studies that also noted a hyperswarming phenotype in these mutants (2, 10, 26). Surprisingly, the rcsF∷Km^r mutation had no effect on swarming in wild-type PM7002 cells. The differential effects of the rcsF mutation in wild-type and waaL mutant backgrounds suggested there was an additional pathway controlling Rcs activity. For example, if this second pathway was O antigen dependent and dominant over the RcsF-dependent pathway with respect to controlling Rcs activity, there would be little or no effect of inactivating rcsF in wild-type cells. However, in PM942 waaL∷Km^r, the lack of O antigen inactivates this dominant pathway, allowing the loss of rcsF to partially suppress the swarming defect caused by the lack of O antigen.

Fig 1.

Differential effects of Rcs mutations on swarming in wild-type and PM942 waaL∷Km^r cells. The swarming phenotypes of the indicated strains were determined by growing cells overnight in LB broth with the appropriate antibiotics and adjusting cultures to the same optical density with fresh LB. (A) PM942 waaL∷Km^r and the corresponding rcs mutants. (B) Swarming phenotype of wild-type P. mirabilis and the corresponding rcs mutants. Each reported value represents the average swarming diameter for three individual 1-μl drops for each strain that was measured during the first consolidation period. The starting diameter of each spot after addition to the LB surface was 4 mm, and this was subtracted from the final diameter.

UmoB and UmoD differentially regulate swarming in wild-type and waaL mutant backgrounds.

The UmoA to -D proteins have been shown to increase flhDC expression when overexpressed and to prevent flhDC expression when mutated (13). The UmoB protein is a homolog of IgaA, a protein in S. enterica that has been shown to inhibit the Rcs phosphorelay (12, 13, 30, 42). The functions of UmoA, UmoC, and UmoD are unknown. Interestingly, UmoB overexpression is able to suppress the swarming defect of an umoD mutant but not vice versa, suggesting these proteins act in the same pathway with UmoD acting upstream (13). This result has been independently confirmed in our lab (data not shown). Because of their roles in flhDC expression and possible connection to the Rcs system, UmoB and UmoD were further explored for their roles in swarming and Rcs activation in wild-type PM7002 and waaL∷Km^r backgrounds. For comparison purposes, we also included an rcsB∷Sm^r strain to demonstrate the swarming phenotype of a strain without a functioning Rcs phosphorelay.

Wild-type PM7002 cells overexpressing UmoB exhibited an increase in swarming (21 mm) relative to results for PM7002 with the pTrc vector alone (14 mm) (Fig. 2A). UmoD overexpression further increased swarming (31 mm), resulting in levels that were comparable to those for PM7002 cells with the rcsB∷Sm^r mutation (Fig. 2A). In cells lacking O antigen (PM942 waaL∷Km^r), UmoB overexpression slightly increased swarming (5 mm), but the percent increase could not be determined since PM942 waaL∷Km^r with the pTrc vector is unable to swarm (Fig. 2B). UmoD overexpression in PM942 waaL∷Km^r resulted in a large increase in swarming motility (14 mm); however, this level of swarming was well below that seen with the waaL∷Km^r rcsB∷Sm^r double mutant (25 mm) (Fig. 2B). This suggested that UmoD was less active in the PM942 waaL∷Km^r mutant lacking O antigen because UmoD overexpression in wild-type cells conferred a hyperswarming phenotype that was statistically equivalent to that with an rcsB∷Sm^r mutation (Fig. 2A).

Fig 2.

Differential activities of UmoB and UmoD in wild-type and waaL mutant backgrounds. (A) Swarming phenotype of cells overexpressing the UmoB and UmoD proteins in wild-type P. mirabilis. For comparison, the swarming phenotype of an rcsB mutant is also shown. (B) Cells overexpressing the UmoB and UmoD proteins in PM942 waaL∷Km^r in comparison to results for a waaL rcsB double mutation. Each reported value represents the average swarming diameter for three individual 1-μl drops for each strain that was measured during the first consolidation period. The starting diameter of each spot after addition to the LB surface was 4 mm, and this was subtracted from the final diameter. An asterisk between two values indicates that a statistically significant difference was present (P value < 0.05).

UmoB and UmoD work through the Rcs phosphorelay.

The UmoB homolog IgaA has been shown to inhibit the Rcs phosphorelay in S. enterica (4, 30, 42). To determine if UmoB in P. mirabilis also regulates activity of the Rcs phosphorelay, we examined the epistatic relationship between umoB and rcsB with respect to swarming. An umoB∷Km^r mutation resulted in a loss of swarming (Fig. 3A), while a rcsB∷Sm^r mutation increased swarming over wild-type levels, 38 mm versus 15 mm (Fig. 3A). The umoB∷Km^r rcsB∷Sm^r double mutant exhibited a swarming phenotype that was essentially identical to that of an rcsB∷Sm^r single mutant, 37 mm versus 38 mm, suggesting they act in the same pathway (Fig. 3A). In a similar manner, the umoD∷Km^r mutant was also unable to swarm, and the umoD∷Km^r rcsB∷Sm^r double mutant exhibited a hyperswarming phenotype that was similar to that of the rcsB∷Sm^r mutant, 39 mm versus 38 mm (Fig. 3A). To further investigate the relationship between UmoB and the Rcs phosphorelay, the swarming phenotype of an rcsB∷Sm^r mutant was compared to that of an rcsB∷Sm^r mutant in which UmoB was overexpressed. The swarming phenotype of UmoB overexpression in the rcsB∷Sm^r mutant was not statistically different from that with rcsB∷Sm^r alone, 27 mm versus 25 mm (Fig. 3B), suggesting that UmoB acts through the Rcs system in P. mirabilis as its homolog IgaA does in S. enterica (4, 30, 42).

Fig 3.

UmoB and UmoD act in the Rcs phosphorelay. The swarming phenotypes of the indicated strains were determined by growing cells overnight in LB broth with the appropriate antibiotics and adjusting cultures to the same optical density with fresh LB. Each reported value in each panel represents the average swarming diameter for at least three individual 1-μl drops for each strain that was measured during the first consolidation period. The starting diameter of each spot after addition to the LB surface was 4 mm, and this was subtracted from the final diameter.

In the PM942 waaL∷Km^r background, the phenotypes of the single umoB∷Km^r and rcsB∷Sm^r mutations and the double umoB∷Km^r rcsB∷Sm^r mutations were also examined. The overall effects of these mutations were similar to that observed in wild-type cells (data not shown). First, the umoB∷Km^r mutation in PM942 waaL∷Km^r did not swarm as expected, since the waaL mutation alone results in a loss of swarming. The rcsB∷Sm^r mutation restored swarming in PM942 waaL∷Km^r, as seen previously in Fig. 1A, and the umoB∷Km^r rcsB∷Sm^r double mutant exhibited a swarming phenotype very similar to that with rcsB∷Sm^r alone (data not shown).

Effects of RcsF overexpression on UmoB and UmoD activity.

The Rcs system contains a predicted outer-membrane-associated protein (RcsF) that modulates Rcs phosphorelay activity when the signal originates outside the periplasm (18, 21, 27, 29). To investigate the possibility that RcsF may influence UmoB and/or UmoD activity, RcsF was coexpressed with either UmoB or UmoD in wild-type PM7002. UmoB was expressed on pTrc99A from the lac promoter; however, IPTG was not added because the leaky expression already resulted in a hyperswarming phenotype and overexpression was toxic (data not shown). The rcsF gene was present on the compatible plasmid pACYC184 and expressed from the tet promoter. Both genes had their native ribosome binding sites. When RcsF was overexpressed by itself, there was only a slight negative effect on swarming (Fig. 4A). As seen previously, the overexpression of UmoB resulted in a hyperswarming phenotype. However, when both genes were overexpressed, RcsF inhibited the hyperswarming phenotype conferred by UmoB overexpression, restoring swarming to near wild-type levels (Fig. 4A). There was no appreciable defect on growth in any of the strains used that would account for the swarming effects (data not shown).

Fig 4.

RcsF overexpression inhibits UmoB but not UmoD activity. The swarming phenotypes of the indicated strains were determined by growing cells overnight in LB broth with the appropriate antibiotics and adjusting cultures to the same optical density with fresh LB. (A) Swarming phenotype of wild-type cells overexpressing umoB, rcsF, or both. (B) Swarming phenotype of PM942 waaL∷Km^r cells overexpressing umoB, rcsF, or both. Each reported value represents the average swarming diameter for three individual 1-μl drops for each strain that was measured during the first consolidation period. The starting diameter of each spot after addition to the LB surface was 4 mm, and this was subtracted from the final diameter. An asterisk between two values indicates that a statistically significant difference was present (P value < 0.05).

The RcsF protein was also overexpressed in combination with UmoD using the same approach described above. However, in this case, RcsF overexpression did not inhibit the activity of UmoD, since the expression of both proteins mediated a hyperswarming phenotype that was not statistically different from that with the overexpression of UmoD alone (Fig. 4A).

To determine if loss of O antigen altered the activities of RcsF, it was overexpressed along with UmoB or UmoD in PM942 waaL∷Km^r. Because PM942 waaL∷Km^r does not swarm, any repressive effect of RcsF overexpression could not be seen (Fig. 4B). However, when RcsF was overexpressed at the same time as UmoB or UmoD in PM942 waaL∷Km^r, RcsF overexpression inhibited swarming in both cases (Fig. 4B). The differential effects of RcsF overexpression on UmoD activity in the wild-type and waaL mutant backgrounds suggested that the loss of O antigen increased the activity of RcsF and/or decreased the activity of UmoD.

UmoB activity is increased in an rcsF mutant.

To determine if the loss of rcsF alters UmoB or UmoD activity, either protein was overexpressed in PM7002 and PM942 waaL∷Km^r backgrounds with and without an rcsF∷Km^r mutation. As seen in Fig. 5A, the rcsF∷Km^r mutation increased the activity of UmoB based on the increased levels of swarming. In contrast, the overexpression of UmoD in wild-type and rcsF mutant backgrounds resulted in an equivalent increase in swarming (Fig. 5A). In PM942 waaL∷Km^r, an rcsF mutation increased the ability of UmoB to increase swarming relative to results for PM942 waaL∷Km^r alone (Fig. 5B). However, the ability of UmoD overexpression to stimulate swarming was similar in wild-type and rcsF∷Km^r backgrounds (Fig. 5B).

Fig 5.

UmoB activity is increased in an rcsF mutant. The swarming phenotypes of the indicated strains were determined by growing cells overnight in LB broth with the appropriate antibiotics and adjusting cultures to the same optical density with fresh LB. (A) Swarming phenotype of wild-type or rcsF∷Km^r cells overexpressing umoB or umoD. (B) Swarming phenotype of PM942 waaL∷Km^r with and without the rcsF∷Km^r mutation and overexpressing umoB or umoD. Each reported value represents the average swarming diameter for three individual 1-μl drops for each strain that was measured during the first consolidation period. The starting diameter of each spot after addition to the LB surface was 4 mm, and this was subtracted from the final diameter. An asterisk between two values indicates that a statistically significant difference was present (P value < 0.05).

The levels of rcsF, umoB, and umoD mRNA are unchanged in PM942 waaL∷Km^r.

To determine if rcsF, umoB, or umoD expression was altered by the loss of O antigen, the accumulation of mRNA from each gene was examined in a wild-type PM7002 and PM942 waaL∷Km^r background by Northern blot analysis of RNA from cells collected 4 h post-surface contact, a time point corresponding to the peak of swarmer cell differentiation. There were no significant differences in mRNA levels for these genes between the wild type and PM942 waaL∷Km^r (data not shown).

DISCUSSION

Previous work from our lab has demonstrated that P. mirabilis mutants lacking O antigen or containing a truncated O antigen do not activate flhDC and the flagellar gene cascade when placed on solid surfaces (35). In this study, we investigated the roles of the Rcs phosphorelay and the UmoB and UmoD proteins in this O-antigen-dependent signaling pathway. We hypothesized that interactions between O antigen and a surface initiated a pathway that inhibited the Rcs phosphorelay. Therefore, mutants with O-antigen defects would retain high levels of phosphorylated RcsB when placed on solid surfaces and fail to derepress flhDC. Consistent with this, mutations in the rcsC and rcsB genes restored swarming to the waaL mutant (Fig. 1). In addition, as seen previously, rcsB and rcsC mutations conferred a hyperswarming phenotype in a wild-type background. However, an rcsF mutation had no effect on swarming in a wild-type background but partially restored swarming to the waaL mutant (Fig. 1). One explanation for the differential effects of the rcsF mutation is that RcsF has only a minor role in activating the Rcs phosphorelay during swarming and that an additional protein(s) regulates the Rcs phosphorelay in an O-antigen-dependent manner. If this second pathway imparted the majority of control on the Rcs phosphorelay, the role of RcsF would be apparent only when this second O-antigen-dependent pathway was inactive, i.e., in a waaL mutant.

The UmoD and UmoB proteins were possible components of this second, O-antigen-dependent pathway that controlled Rcs activity. Both UmoB and UmoD increased swarming when overexpressed in wild-type cells, although UmoD conferred a stronger phenotype, with swarming levels similar to those of cells with a nonfunctional Rcs phosphorelay (rcsB∷Sm^r) (Fig. 2A). UmoB and UmoD overexpression also increased swarming when overexpressed in PM942 waaL∷Km^r. However, in this background UmoD appeared to be less active based on swarming levels that were significantly lower than those of PM942 waaL∷Km^r containing an rcsB∷Sm^r mutation (Fig. 2B). Dufour et al. suggested that UmoD and UmoB act in the same pathway to control flhDC expression, with UmoD acting upstream and required for UmoB activity (13). This was based on the finding that overexpression of UmoB could suppress the effect of an umoD mutation on swarming but not vice versa. Our data are consistent with these results and expand on them by providing data consistent with a role for UmoD and UmoB in activating flhDC by inhibiting the Rcs phosphorelay (Fig. 3). This is supported by the following. (i) umoD and umoB mutants fail to swarm, and this swarming defect can be suppressed by an rcsB mutation (Fig. 3). Moreover, the umoB rcsB and umoD rcsB double mutants swarmed at the same high levels as an rcsB mutant alone, suggesting they act in the same pathway (Fig. 3A). (ii) Overexpression of UmoB in the rscB mutant background did not increase swarming above the already high levels exhibited by the rcsB mutant alone (Fig. 3B). (iii) In S. enterica, an UmoB homolog (IgaA) has been shown to inhibit activity of the Rcs phosphorelay (12, 30, 31, 42).

The role of RcsF in swarming was also investigated by overexpressing it together with UmoB or UmoD in either a wild-type or a waaL mutant background. In both backgrounds, the overexpression of RcsF inhibited UmoB activity (Fig. 4A and B). However, the overexpression of RcsF with UmoD had no effect on UmoD activity in wild-type cells (Fig. 4A). Interestingly, in PM942 waaL∷Km^r, RcsF overexpression strongly inhibited the ability of UmoD to increase swarming when overexpressed (Fig. 4B). These data suggest that in PM942 waaL∷Km^r cells, UmoD activity is lower and/or RcsF activity is higher than in wild-type cells. Further data to support a role for RcsF in the inhibition of UmoB are provided in Fig. 5, where the overexpression of UmoB resulted in a greater increase in swarming in the rcsF∷Km^r background than in the wild type.

One caveat of our data that should be taken into consideration is the possibility that the loss of O antigen in the waaL background creates a cell envelope stress that activates the Rcs phosphorelay. Although this possibility cannot be ruled out, we feel that it is unlikely. First, the waaL mutant exhibits normal swimming motility, indicating that the Rcs phosphorelay is not constitutively active (35). However, this does not rule out the possibility that the potential envelope stress due to loss of O antigen is specific to solid surfaces. Second, a wzz mutant containing O antigen with a reduced chain length that does not accumulate unligated O-antigen precursors in the periplasm is also unable to activate flhDC and flaA on surfaces (35). This proposed O-antigen surface sensing pathway likely functions at least partially independently of the surface sensing pathway mediated by the inhibition of flagellar rotation that was proposed by Belas and Suvanasuthi (3). This is based on our unpublished studies that indicate a motA mutation, which leaves the flagellar filament intact but unable to rotate, does not alter the derepression of flhDC on solid surfaces.

Using the above data, a preliminary model is presented for how P. mirabilis senses surfaces using O antigen, the UmoB/D proteins, and the Rcs phosphorelay (Fig. 6). Upon surface contact, O antigen mediates a pertubation or torsional change in the outer membrane that may arise by the weight of cells pushing on O antigen when in contact with a surface or by interactions of O antigen with surfaces which create a pulling force. This outer membrane change then results in two opposing effects: inhibiting the activity of RcsF and increasing the activity of UmoD. The inhibition of RcsF may occur if the O-antigen-mediated surface contact allows a localized region of the outer and inner membranes to exhibit a transient increase in the distance between these membranes, thereby preventing RcsF in the outer membrane from interacting with UmoB in the inner membrane. A recent study by Farris et al. suggests that outer membrane changes induced by antimicrobial peptides can alter RcsF activity; however, in this case it was proposed that the outer membrane change resulted in a spatial or conformational change that favored RcsF contacts with the RcsC or RcsD protein or IgaA (UmoB) (14). Therefore, it is possible that RcsF can differentially respond to various outer membrane changes. The activation of UmoD may occur by a conformation or localization change that allows it to make more favorable contacts with UmoB (IgaA) and increase its activity by a yet-to-be-determined mechanism. The overexpression of UmoD conferred a greater increase in swarming than UmoB overexpression (Fig. 2). Several possibilities can account for this. First, in addition to activating UmoB, UmoD may block RcsF from inhibiting UmoB. Alternatively, if UmoD is limiting when UmoB is overexpressed, the amount of activated UmoB may be lower than that when UmoD is overexpressed. Currently, studies are in progress to determine if RcsF and/or UmoD is capable of physically interacting with the UmoB protein. UmoB is predicted to have a large periplasmic loop (13), and this is a possible site for the interaction of RcsF and/or UmoD.

Fig 6.

A model for Rcs inhibition by surface contact. Surface interactions with O antigen trigger conformational changes in the outer membrane that result in decreased activity of RcsF and/or increased activity of UmoD. This results in activation of UmoB by two mechanisms, direct activation by UmoD and reduced activity of RcsF, an inhibitor of UmoB. The activated form of UmoB then inhibits the Rcs phosphorelay, resulting in reduced levels of phosphorylated RcsB and derepression of the flhDC operon.