An IgaA/UmoB Family Protein from Serratia marcescens Regulates Motility, Capsular Polysaccharide Biosynthesis, and Secondary Metabolite Production

ABSTRACT

Secondary metabolites are an important source of pharmaceuticals and key modulators of microbe-microbe interactions. The bacterium Serratia marcescens is part of the Enterobacteriaceae family of eubacteria and produces a number of biologically active secondary metabolites. In this study, we screened for novel regulators of secondary metabolites synthesized by a clinical isolate of S. marcescens and found mutations in a gene for an uncharacterized UmoB/IgaA family member here named gumB. Mutation of gumB conferred a severe loss of the secondary metabolites prodigiosin and serratamolide. The gumB mutation conferred pleiotropic phenotypes, including altered biofilm formation, highly increased capsular polysaccharide production, and loss of swimming and swarming motility. These phenotypes corresponded to transcriptional changes in fimA, wecA, and flhD. Unlike other UmoB/IgaA family members, gumB was found to be not essential for growth in S. marcescens, yet igaA from Salmonella enterica, yrfF from Escherichia coli, and an uncharacterized predicted ortholog from Klebsiella pneumoniae complemented the gumB mutant secondary metabolite defects, suggesting highly conserved function. These data support the idea that UmoB/IgaA family proteins are functionally conserved and extend the known regulatory influence of UmoB/IgaA family proteins to the control of competition-associated secondary metabolites and biofilm formation.

IMPORTANCE IgaA/UmoB family proteins are found in members of the Enterobacteriaceae family of bacteria, which are of environmental and public health importance. IgaA/UmoB family proteins are thought to be inner membrane proteins that report extracellular stresses to intracellular signaling pathways that respond to environmental challenge. This study introduces a new member of the IgaA/UmoB family and demonstrates a high degree of functional similarity between IgaA/UmoB family proteins. Moreover, this study extends the phenomena controlled by IgaA/UmoB family proteins to include the biosynthesis of antimicrobial secondary metabolites.

INTRODUCTION

Microorganism-derived secondary metabolites include important interspecies competitive factors and are a major potential source of therapeutic agents (1–6). The bacterium Serratia marcescens is found in a wide range of ecological niches and is known for the production of antibiotic secondary metabolites, including the peptide antibiotic althiomycin, the red pigment prodigiosin, and the biosurfactant serratamolide, also known as serrawettin W1 (7–11).

Both prodigiosin and serratamolide are reported to have broad-range antimicrobial activity, and serratamolide promotes swarming motility, a behavior that may allow the bacterium to become less susceptible to antibiotics and move into more favorable microenvironments (8, 12–17). Serratamolide from S. marcescens is hemolytic and was demonstrated to enable bacteria of different genera to evade phagocytosis by neutrophils in vitro, together suggesting that it may be a virulence factor and impact polymicrobial infections (18, 19). Regulation of secondary metabolites by S. marcescens occurs at the transcriptional level by several transcription factors, including CopA, cAMP receptor protein (CRP), EepR, HexS, PigP, RssAB, SmaI, and SpnR (19–27).

The goal of this study was to characterize the pleiotropic phenotypes of a newly identified secondary metabolite regulator. The mutations in mutants with reduced prodigiosin and serratamolide in S. marcescens mapped to an uncharacterized gene predicted to code for a protein of the IgaA/UmoB family. IgaA/UmoB family proteins were first described from the highly motile bacterium Proteus mirabilis by the Hughes group in 1998, where mutation of the umoB gene conferred a downregulation of the flhDC flagellar master regulator operon (28). Since then, most studies on this family of proteins have been performed using Salmonella enterica serovar Typhimurium. The igaA gene was found to be a regulator of intracellular growth in S. enterica and was later shown to regulate capsular polysaccharide production and pathogenesis (29–33). In Escherichia coli, mutations in the IgaA/UmoB-encoding gene, yrfF, were found to be enriched following serial coculture with macrophages and were found in a keratitis isolate (34, 35). The igaA and yrfF genes are essential for growth in Salmonella and E. coli bacteria (30), but umoB is not required for Proteus spp. to survive (28). The essential nature of igaA and yrfF allowed identification of suppressor mutations of the lethality phenotype that mapped to the Rcs regulatory system (30, 31, 36). Subsequent studies using bacteria from several genera indicate that IgaA/UmoB controls the Rcs phosphorelay signaling system that controls hundreds of genes (32, 37–39). Data suggest that IgaA/UmoB, together with RcsF, regulates the Rcs system in response to envelope stress and surface interactions by controlling the phosphorylation state of histidine kinase protein RcsC through an unknown mechanism (39–41). S. marcescens has an Rcs phosphorelay system that has been implicated in the regulation of outer membrane vesicle production and could be activated by a mutation of genes involved in biosynthesis of the enterobacterial common antigen outer membrane component (42, 43).

Because igaA and yrfF are essential genes in E. coli and S. enterica, there is incomplete knowledge about the range of biological processes that are mediated through this conserved family of proteins. This study characterized pleiotropic phenotypes of a mutant involving the S. marcescens IgaA/UmoB family gene, gumB, and found novel roles for IgaA/UmoB family proteins in their influence over secondary metabolite prodigiosin and biofilm production.

RESULTS

Isolation of an IgaA/UmoB family gene in S. marcescens.

Transposon (Tn) mutagenesis was used to identify genes that regulate secondary metabolite production in a clinical keratitis isolate of S. marcescens, strain K904 (25). Two mutants with severely reduced red pigmentation and hemolysis were isolated from nonsaturating mutagenesis of ∼6,000 mutants (Fig. 1A and B). The coloration of the mutant colonies ranged from white (24 h) to pink over time (≥48 h); similarly, over time, the colonies became elastic (gummy) and recalcitrant to manipulation with a toothpick and developed a rugose colony morphology (Fig. 1A).

FIG 1.

Isolation of mutants with secondary metabolite biosynthesis defects and a distinct colony morphology. (A) Representative images of the wild-type (WT) strain K904 and the two gummy mutants. The colonies were top-lit and demonstrate a lack of prodigiosin and altered colony morphology. (B) Representative images show a hemolysis zone (white arrow) on blood agar plates around K904 and a lack of hemolysis zones around two gumB transposon mutants and a defined gumB mutant. The plate was lighted from below to highlight the hemolysis zone, obscuring colony color. (C) Genetic map of the gumB gene and surrounding genes. Transposon insertion sites are noted by gray arrow points. (D) Phylogram of amino acid sequences of known and predicted IgaA/UmoB family proteins. The dendrogram was made using the “one-click” mode of the tree rendering program from www.Phylogeny.fr, using default settings. The number of substitutions per site are proportional to the branch length.

The mutations mapped to an uncharacterized gene in the S. marcescens genome, corresponding to open reading frame (ORF) SMDB11_3859 of sequenced strain Db11 (44). The mutations mapped to bp 381 relative to the start codon for gummy mutant 1 and bp 168 for gummy mutant 2 (Fig. 1C). Other mutants defective in both prodigiosin and serratamolide production mapped to the previously identified eepS gene (27) and an uncharacterized transcription factor of the GntR family that will be described elsewhere.

At the amino acid level, the identified gene from strain K904 is predicted to code for a protein that is 99% identical (701/710 amino acids) to the corresponding SMDB11_3859-encoded predicted protein. This protein consists of a Pfam PF07095/IgaA domain and is 42% identical to UmoB from Proteus mirabilis, 56% identical to YrfF of Escherichia coli, and 57% identical to IgaA from Salmonella enterica serovar Typhimurium. The family of proteins is conserved in the Enterobacteriaceae, including a putative protein coded by an uncharacterized ORF in K. pneumoniae predicted to be 54% identical to GumB at the amino acid level (see Fig. S1 in the supplemental material). Phylogenetic analysis suggests that the SMDB11_3859-encoded protein is structurally intermediate in relatedness between UmoB and the highly similar IgaA and YrfF proteins (Fig. 1D). In silico analysis of the GumB primary sequence suggests that it is in the inner membrane with three transmembrane helices (Fig. S1). Published models propose that IgaA/UmoB sits in the inner membrane and responds to RcsF localization or conformation to transmit envelop stress signals, and this is accomplished through IgaA/UmoB control of the Rcs transcription factor system (39–41).

Since umoB was the first gene of the IgaA/UmoB family with experimentally defined function, we here refer to the S. marcescens SMDB11_3859 as the gummy homolog of umoB, gumB.

Unlike igaA and yrfF, the gumB gene is not essential for viability.

The gumB open reading frame appears to be monocistronic (Fig. 1C), although there is a predicted operon 98 bp downstream of the stop codon of gumB. The adjacent operon composed of ORFs SMDB11_3960 and SMDB11_3961 is predicted to code for a GMP/IMP nucleotidase and ribosome-associated heat shock protein Hsp15, respectively. As a first step to test whether the observed mutant phenotypes were due to a loss of GumB activity or a polar effect on an adjacent gene, an in-frame deletion mutation of gumB was introduced into the K904 chromosome by two-step allelic replacement. Colonies of the resulting mutant strain were similar to the transposon mutants with respect to a lack of pigmentation and the development of an elastic and highly textured colony morphology (Fig. 2A and Movie S1 in the supplemental material). The isolation of transposon insertion mutations in gumB and the generation of a gumB deletion mutation suggest that unlike what was observed with igaA from S. enterica and yrfF from E. coli, the gumB gene of S. marcescens is not essential for growth.

FIG 2.

Complementation of gumB mutant defects and growth analysis. (A) The gumB mutant colony and color phenotypes are complemented by wild-type gumB on a plasmid under the control of the P_lac promoter. Representative images are of the strain K904 and the ΔgumB mutant with vector control plasmid pMQ132 and complementation plasmid pgumB (pMQ480). Bacteria were incubated for 48 h at 30°C on LB or blood agar plates and lighted from above. (B and C) Growth curves of the wild-type strain K904 and the ΔgumB mutant in LB (B) or M9 minimal medium supplemented with glucose (C). Means ± standard deviation (SD) are shown, n ≥ 3.

Complementation analysis of the ΔgumB mutation with wild-type gumB on a multicopy plasmid further supports that the observed mutant defects, loss of pigmentation, and rugose phenotypes are due to a mutation of gumB and not another mutation or a polar effect on adjacent genes (Fig. 2A).

Growth analysis in rich medium demonstrates the slightly retarded growth of the ΔgumB mutant in log phase between 4 and 8 h; however, the gumB mutant achieved a final optical density similar to that of the wild type by stationary phase (∼9 h) (Fig. 2B). Given the slight growth reduction of the gumB mutant in rich medium, we tested whether it was defective in primary metabolite production by assessing growth in M9 minimal medium. The ΔgumB mutant grew slightly better than the wild type in M9 medium supplemented with glucose (Fig. 2C). These data indicate that despite the pleiotropic phenotypes imparted by a mutation of gumB, the mutant strain is able to generate all of the primary metabolites necessary for growth in minimal medium.

ΔgumB mutant phenotypes are complemented by the igaA gene from S. enterica, yrfF from E. coli, and kumO from K. pneumoniae.

To genetically test whether the GumB protein shares functionality with the IgaA/UmoB family proteins noted above, genes from E. coli (yrfF), K. pneumoniae (kumO), and S. enterica (igaA) were cloned and introduced into the ΔgumB mutant strain on a multicopy plasmid. The pigment and colony morphology phenotypes of the ΔgumB mutant were restored to those of the wild type by the expression of igaA and yrfF (Fig. 3A). The K. pneumoniae gene complemented the colony morphology phenotype and partially complemented the pigment phenotype that is clear in the larger colonies (Fig. 3B). These results support the idea that the UmoB/IgaA family proteins are highly conserved among opportunistic pathogens from the Enterobacteriaceae.

FIG 3.

Complementation of gumB mutant defects by IgaA/UmoB family genes from several genera. (A) Image of the ΔgumB mutant with vector alone (pMQ132) or pMQ132 with various IgaA/UmoB family genes under the control of P_lac to test for complementation. (B) Zoomed-in view of panel A to highlight the colony morphology complementation of the gumB mutant with a predicted IgaA/UmoB gene from K. pneumoniae (kumO).

Motility defects were conferred by mutation of gumB.

Tests were done to determine whether GumB shared function with IgaA and UmoB in the regulation of flagellum-based motility. The ΔgumB mutant did not produce a swimming zone when measured at 24 h, unlike the wild type (Fig. 4A). GumB was also required for surface swarming motility (Fig. 4A). The gumB mutant swarming defects (Fig. 4B) and swimming defects (Fig. 4C) were complemented by the wild-type gumB gene on a plasmid.

FIG 4.

S. marcescens GumB is necessary for swimming and swarming motility. (A) Representative images of the swimming and swarming defects of the ΔgumB mutant. (B) Representative images of complementation of the ΔgumB swarming defect. (C) Complementation of the ΔgumB swimming phenotype. n ≥ 5 independent plates per genotype from 2 separate days. Asterisks indicate a significant difference from all of the other genotypes (P < 0.001). Mean and SD are shown. Vector, pMQ132; pgumB, pMQ480. (D) Fold change of both flagellum-related gene transcript levels. qRT-PCR of fliC and flhD gene expression, n ≥ 4. Mean and SD are shown. *, P < 0.05; **, P < 0.01.

To test whether expression of the flagellar master regulator operon flhDC was altered in the ΔgumB mutant, quantitative-reverse transcriptase PCR (qRT-PCR) was employed. The transcript level of the flhD gene was reduced (P < 0.05, Mann-Whitney test) by several orders of magnitude in the ΔgumB mutant compared to the wild-type K904 strain when measured at an optical density at 600 nm (OD₆₀₀) of 3 (Fig. 4D). This reduction in flhDC operon expression correlated with an ∼10,000-fold reduction of flagellar gene (fliC) expression when measured by qRT-PCR (Fig. 4D). Transmission electron microscopy (TEM) analysis of the ΔgumB mutant cells for stationary-phase liquid cultures also demonstrated a loss of flagella (Fig. S2), with 71% of the wild-type cells having a flagellum, and <1% of ΔgumB mutant cells with a flagellum (n = 105 cells for each genotype; P < 0.001, Fisher's exact test). Together, these data suggest that GumB plays a role in flagellum-based motility through positive control of the flhDC flagellum master regulator operon.

Secondary metabolite defects of the gumB mutant.

The red pigment prodigiosin is a secondary metabolite made by a subset of S. marcescens strains and is thought to be involved in competition between microbes and to modulate energy levels during stationary phase. IgaA family proteins have not been implicated in the control of secondary metabolites, so the obvious defect in pigmentation of the ΔgumB mutant was quantified. Prodigiosin was extracted and measured from stationary-phase cultures of the wild-type and ΔgumB mutant strains with complementation plasmid, pMQ480, or vector negative control. Prodigiosin levels were significantly lower in the ΔgumB mutant and could be complemented in trans (Fig. 5A).

FIG 5.

Prodigiosin pigmentation is severely reduced in a ΔgumB mutant. (A) The gumB mutant grown in LB medium is defective in prodigiosin production (75-fold reduction compared to strain K904; n = 3, P < 0.01). Extracted prodigiosin was measured by absorbance, and the defect could be complemented by wild-type gumB in trans. The vector control plasmid is pMQ132, and pgumB is pMQ480. (B) qRT-PCR analysis revealed reduced expression from the prodigiosin biosynthetic locus; pigA gene expression in the ΔgumB mutant was down 103-fold compared to the wild type (P = 0.029, n = 4). Means and standard deviations are shown. *, P < 0.05; **, P < 0.01.

This effect could stem from reduced transcriptional expression of the prodigiosin biosynthesis operon. Reverse transcriptase PCR (RT-PCR) analysis was used on stationary-phase cultures, as prodigiosin is produced during stationary phase (OD₆₀₀, 3). A clear ∼100-fold reduction in the pigA transcript level was observed in the ΔgumB mutant, suggesting that the gumB mutant pigment defect is largely at the transcriptional level (Fig. 5B).

The biosurfactant serratamolide (45), also called serrawettin W1 (8), is antimicrobial and required for the hemolytic activity of S. marcescens strain K904 on blood agar plates and for growth inhibition of Staphylococcus aureus and other microbes (14, 19). The ΔgumB mutant was defective in hemolysis and antistaphylococcal inhibition zones, suggesting a serratamolide defect (Fig. 6A and B). The ΔgumB mutant was correspondingly defective in the biosynthesis of serratamolide when relative amounts were measured by mass spectrometry (Fig. 6C) and in expression of the serratamolide biosynthesis gene swrW at an OD₆₀₀ of 3.0 (Fig. 6D). Hemolysis, an indirect measurement of serratamolide production, was used in a complementation analysis, and gumB on a plasmid was able to restore serratamolide production to the ΔgumB mutant (Fig. 6A).

FIG 6.

Serratamolide is severely reduced in a ΔgumB mutant. (A) Image of blood agar plate, lit from below. Serratamolide is responsible for hemolysis zones generated by strain K904. The ΔgumB mutant was defective for hemolysis and could be complemented by wild-type gumB in trans. The vector control plasmid is pMQ132, and pgumB is pMQ480. (B) Representative top-lit image. Inhibition of S. aureus bacterial growth (black arrow) by S. marcescens is serratamolide dependent and is eliminated in the ΔgumB mutant. (C) Mass spectrometry reveals a >50-fold significant reduction of serratamolide in the ΔgumB mutant (P < 0.01, n = 3). (D) qRT-PCR analysis demonstrates reduced expression of the serratamolide biosynthetic gene swrW, which was reduced >1,000-fold in the relative transcript levels for the mutant compared to the wild type (P = 0.029, n = 4). Mean and SD are shown. *, P < 0.05; **, P < 0.01.

The quantitative RT-PCR-generated expression levels of pswP, a gene necessary for both prodigiosin and serratamolide production (46), were similar for the K904 (0.0014 ± 0.0027) and the K904 ΔgumB (0.0012 ± 0.0024) strains (P = 0.84, Student's t test). Similarly, multicopy expression of pswP did not alter the pigment or serratamolide defects of the ΔgumB mutant, suggesting that the gumB secondary metabolite defect was PswP independent.

GumB inhibits exopolysaccharide biosynthesis.

To determine the mechanism by which GumB-deficient mutant colonies become elastic, suppressor mutant analysis was performed. Transposon mutagenesis was performed on the ΔgumB mutant strain, and colonies that had lost the rugose colony morphology were chosen for analysis. Multiple independent transposon mutations in the gumB deletion mutant strain yielded smooth nongummy colonies (Fig. 7A). These transposons mapped to adjacent genes in the formerly described capsular polysaccharide/enterobacterial common antigen biosynthetic locus, which has recently been shown to be important for virulence (47, 48). Five mutations were in the wza polysaccharide exporter gene, and two others were found upstream of the wecA exopolysaccharide biosynthesis gene; these genes correspond to strain Db11 ORFs SMDB11_2151 and SMDB11_2152, respectively. TEM analysis of bacteria grown on plates had staining consistent with excessive extracellular polysaccharide surrounding ΔgumB mutant bacteria that is absent in the ΔgumB wza double mutant (Fig. S2). These data suggested that capsular polysaccharide transcription in the ΔgumB mutant was increased.

FIG 7.

Gummy phenotype of the ΔgumB mutant requires the capsular polysaccharide operon. (A) Suppressor mutants of the rugose colony morphology phenotype of the ΔgumB mutant map to and wza genes of the capsular polysaccharide operon. (B) Extracellular polysaccharide (EPS) was extracted and measured from liquid cultures. Mean and SD are shown, n = 4. Asterisks indicate a significant increase relative to other groups (ANOVA, Tukey's posttest, P < 0.01). (C) qRT-PCR analysis revealed elevated relative wecA gene expression levels in the ΔgumB mutant (P = 0.016, n = 5). Mean and SD are shown. *, P < 0.05. (D) Biofilm formation in LB medium on borosilicate glass tubes grown under high-sheer conditions for 6 and 20 h and stained with crystal violet. The ΔgumB mutant had an elevated biofilm that was absent in a ΔgumB wza mutant. Mean and SD are shown, n = 12 biological replicates.

Total extracellular polysaccharide production was measured from the liquid cultures of the wild-type K904 strain, the ΔgumB mutant, and the ΔgumB wza double mutant. The amount of extracellular polysaccharide extracted from the ΔgumB mutant was ∼2-fold higher than that from the wild type and the ΔgumB wza double mutant (Fig. 7B). Dry weight polysaccharide measurements showed a similar trend, with 3.2 ± 0.8 mg/ml derived from K904, 5.6 ± 0.7 mg/ml from the ΔgumB mutant, and 2.1 ± 0.3 mg/ml from the ΔgumB wza mutant.

Transcriptional analysis of a gene from the capsular polysaccharide biosynthesis operon, wecA, supports the prediction that there is a deregulation of capsular polysaccharide production in the ΔgumB mutant. There was an approximate 100-fold increase in the wecA transcript level in the ΔgumB mutant strain compared to the K904 strain, at an OD₆₀₀ of 3 (P < 0.05; Fig. 7C). These data suggest that the gummy and rugose colony phenotype of the gumB mutant is a result of deregulated capsular polysaccharide production.

Mutation of the capsule gene wza in the ΔgumB mutant did not alter hemolysis, motility, or pigmentation phenotypes, suggesting that the excessive capsular polysaccharide of the ΔgumB mutant is not responsible for most of its phenotypes (Fig. 6A, 7A, and S3).

Because extracellular polysaccharides can promote biofilm formation, we tested whether the gumB mutant had altered biofilm formation. Biofilms were formed on borosilicate glass under high-sheer conditions (rotated at ∼62 rpm) in LB medium and then stained with crystal violet. At 6 h, the ΔgumB mutant had a minor increase in biofilm formation compared to K904, whereas the ΔgumB wza::Tn double mutant had a 2.3-fold reduction in biofilm formation compared to K904 (Fig. 7D). At 20 h, the ΔgumB wza::Tn double mutant was defective (6.5-fold; P < 0.01, analysis of variance [ANOVA] with Tukey's posttest) in biofilm formation compared to the ΔgumB mutant, suggesting that under the tested conditions, the extracellular polysaccharide regulated by GumB contributes to biofilm formation (Fig. 7D).

GumB mutant phenotypes are not strain specific.

To test whether ΔgumB mutant phenotypes were specific to strain K904, the gumB gene was deleted in S. marcescens strains Db11, a nonpigmented insect pathogen, and CHASM, a pigmented environmental isolate. These deletions conferred a similar pigment phenotype in CHASM (Fig. S4A) and similar swarming defects in both CHASM and Db11 (Fig. S4B), indicating that the role of GumB in regulating secondary metabolite production and motility is not strain specific.

DISCUSSION

This study, designed to identify new regulators of secondary metabolism in S. marcescens, reports a new member of the IgaA/UmoB family of proteins, named GumB. Mutation of the gumB gene introduced pleiotropic phenotypes, most notably a severe reduction in secondary metabolite production.

The genetic results presented here indicate that IgaA/UmoB family genes from several genera of Enterobacteriaceae, including an uncharacterized ortholog from K. pneumoniae, are functionally and structurally conserved. Therefore, it was somewhat surprising that the gumB gene is not essential for viability in S. marcescens as yrfF and igaA are in E. coli and S. enterica, respectively. Analysis of the S. marcescens genome suggests that there is not another obvious IgaA/UmoB gene, so the essential requirement for IgaA/UmoB family proteins found in E. coli and S. enterica must be downstream of GumB and absent or redundant in S. marcescens. Similar to the situation in S. marcescens, the umoB gene in P. mirabilis is not necessary for viability (28). While gumB is not essential, there was a slight growth delay in the ΔgumB mutant grown in LB medium compared to the wild type, although the final culture densities were the same. This growth delay may be due to altered expression of cell division genes in Rcs system and IgaA/UmoB gene mutants, as seen in E. coli and S. enterica (36, 49–51).

The polysaccharide phenotype of the gumB mutant is similar to those of S. enterica with partial-function igaA mutations, where increased activation of the Rcs system is required for the highly mucoid phenotype (29–33). This hypercapsule phenotype conferred by mutation of IgaA/UmoB genes is predicted to be the reason why E. coli cells with nonlethal mutations in yrfF had a selective advantage in a model in which they were serially cocultured with a phagocytic cell line (34). The hypercapsule suggests that IgaA/UmoB proteins may negatively regulate virulence; however, studies with S. enterica bearing nonlethal mutations in igaA indicate that IgaA, at least, is required for virulence in a rodent infection model (31). Importantly, data presented indicated that, other than biofilm formation and colony morphology, excess polysaccharide production is not responsible for the other gumB mutant phenotypes.

This study provides evidence for polysaccharides contributing to S. marcescens biofilm formation, whereas previous studies suggest a dominant role for type I fimbriae (52, 53). This is a new role for IgaA/UmoB family proteins, although it has been demonstrated that the Rcs system can control biofilms (54), and it is likely that GumB works through regulation of the Rcs system in S. marcescens. This is predicted because known IgaA/UmoB proteins, such as IgaA and YrfF, regulate the Rcs system, and here, we showed that the igaA and yrfF genes could complement gumB defects, implying structural and functional similarity between GumB, IgaA, and YrfF. In-progress genetic studies further support that gumB mutant phenotypes require the S. marcescens Rcs system genes (N. A. Stella and R. M. Q. Shanks, unpublished data).

Whereas GumB shares function with known IgaA/UmoB family proteins with respect to flagellar and polysaccharide biosynthesis, it was unknown whether this family of proteins mediates secondary metabolite function. Here, we measured a severe reduction in both prodigiosin and serratamolide. Given the ability of these two compounds to prevent the growth of other microbes, promote motility, promote resistance to antimicrobials via swarming motility, provide resistance to phagocytosis, and cause lysis of mammalian cells, a gumB mutant of S. marcescens would be expected to be at a competitive disadvantage in many niches. A loss of secondary metabolite production was observed when the gumB gene was mutated in three different strains. This is important, as a given gene does not necessarily confer the same phenotypes when mutated in different strains, as has been reported for cyclic-AMP-regulated pathways in S. marcescens (27).

IgaA family proteins are thought to regulate the Rcs system in response to cellular stress (39, 42, 55, 56). Based on this model, GumB likely senses extracellular stress and modifies the activity of the RcsC and or RcsD histidine kinases that in turn influence the expression of many genes by the RcsB response regulator. One may thus consider a ΔgumB mutant cell to be in a constitutively “stressed-out” state. The production of excess capsular polysaccharide is thought to relieve outer membrane stress (57) and promote biofilm formation, both of which may enhance survival. In this stressed state, the bacteria increase the expression of capsular polysaccharide genes but stop the production of flagella and competition-promoting secondary metabolites, perhaps to conserve energy and reduce activation of host immune systems through eliminating the production of pathogen-associated molecular patterns. How GumB interacts with other regulators is unknown at this time; however, it is notable that regulation of pigment, serratamolide, and flagellum by GumB follows a directly opposite trend from that of cAMP-CRP (19, 25, 53, 58), so it could be that GumB-associated phenotypes are mediated through control of intracellular cAMP levels and the corresponding activity of the CRP. Indeed, there is evidence in P. mirabilis that the Rcs system regulates CRP expression (54).

In summary, this report describes a new regulator of secondary metabolites, motility, and biofilm formation in the bacterium S. marcescens. The GumB gene is predicted to be highly important for S. marcescens strains to compete in a variety of environments, given the biological functions that it controls. These could include the use of S. marcescens as a biological control agent, where its GumB-regulated behaviors would be necessary, and in the human gut, where successful competition of S. marcescens is associated with Crohn's disease (59). Last, this study extends the known role of the conserved IgaA/GumB family proteins to the control of secondary metabolites and biofilm formation.

MATERIALS AND METHODS

Strains and growth medium used in this study.

Bacteria were grown in lysogeny broth (LB) (60) or M9 minimal medium (61) supplemented with glucose at 0.4% (wt/vol). Saccharomyces cerevisiae was grown in yeast extract-peptone-dextrose (YPD) broth or synthetic complete (SC)-uracil medium (62). Swimming and swarming agars consisted of LB medium supplemented with 0.3% and 0.6% (wt/vol) agar, respectively. Cultures were grown with aeration using a TC-7 tissue culture roller. All incubations were performed at 30°C. The strains used are listed in Table 1.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
Strains
SM10 λpir	E. coli conjugation strain	70
S17-1 λpir	E. coli conjugation strain	70
EC100D	E. coli cloning strain, pir-116 version	Epicentre
K746	E. coli keratitis isolate	71
LT2	S. enterica serovar Typhimurium	72
883	K. pneumoniae, clinical isolate	73
CHASM	S. marcescens, environmental isolate	25
K904	S. marcescens, keratitis isolate	25
Db11	S. marcescens, insect isolate	74
CMS2265	K904 gummy mutant 1, K904 gumB::Tn	This study
CMS4001	K904 ΔgumB	This study
CMS4124	K904 ΔgumB wza::Tn	This study
CMS4225	K904 fimC-kan	75
CMS4619	K904 ΔgumB wecA::Tn	This study
CMS4320	K904 gummy mutant 2, K904 gumB::Tn	This study
CMS4462	Db11 ΔgumB	This study
CMS4662	CHASM ΔgumB	This study
Plasmids
pBT20	Transposon delivery plasmid	63
pSC189	Transposon delivery plasmid	64
pMQ132	pBBR1-replicon shuttle vector, aacC-1	68
pMQ200	oriR6K plasmid with PBAD promoter, nptII	68
pMQ460	Allelic replacement vector	76
pMQ480	pMQ132 plus gumB from S. marcescens strain K904	This study
pMQ507	pMQ460 plus ΔgumB	This study
pMQ530	pMQ132 plus igaA from S. enterica strain LT2	This study
pMQ529	pMQ132 plus kumO from K. pneumoniae strain 883	This study
pMQ531	pMQ132 plus yrfF from E. coli strain K746	This study

Genetic manipulations and plasmid construction.

Transposon mutagenesis was performed as previously described (52) using transposon delivery plasmids pBT20 (63) and pSC189 (64). Insertion sites were mapped by marker rescue (64) or arbitrary PCR (65) followed by sequencing. Mutants of S. marcescens strain K904 were plated on LB or blood agar and were visually screened for a loss of pigmentation and hemolysis zones, as previously described (27).

Allelic replacement of the gumB open reading frame (ORF) was performed as previously described but using plasmid pMQ507 (all plasmids used are listed in Table 1) (66). The mutation deletes 2,044 bp of the 2,133-bp gene, starting at the ninth codon. To make pMQ507, 470 bp of DNA upstream of gumB and 662 bp of DNA downstream of gumB were amplified with primers that have additional sequence such that the amplicons can recombine with each other and with allelic replacement vector pMQ460. This and other plasmids were made by homologous recombination using yeast in vivo recombination (67, 68). The oligonucleotide primers used to make pMQ507 were primers 3419 to 3422 and are listed in Table 2. Plasmids were verified by PCR and sequencing of cloned junctions, and deletion mutations were verified by PCR.

TABLE 2.

Oligonucleotide primers used in this study

Primer no.	Purposea	Sequence (5′ to 3′)b
1055	fimA-RT	ACTACACCCTGCGTTTCGAC
1056	fimA-RT	GCGTTAGAGTTTGCCTGACC
2638	16S-RT	AACTGGAGGAAGGTGGGGAT
2639	16S-RT	AGGAGGTGATCCAACCGCA
2891	pswP-RT	CGTGACATCGTCACCTTCACG
2892	pswP-RT	GCCAAAGAGAGCCTGTTCAAG
2911	pigA-RT	GGAGCGAACTGACCTTCAAC
2912	pigA-RT	CTGTTCCAGACGCAGTTTCA
3417	gumB cloning	ggccagtgccaagcttgcatgcctgcaggtcgactctaTTGAAGCAGCTGTCGTAGTAAC
3418	gumB cloning	tgtgagcggataacaatttcacacaggaaacagctATGAGCACAATAGTGTTGATATTGG
3419	gumB deletion	cggccagtgccaagcttgcatgcctgcaggtcgactctaGATTGCCTCAAAGAGGTTACC
3420	gumB deletion	gaagcagctgtcgtagtaacgctggCCAATATCAACACTATTGTGCTCAT
3421	gumB deletion	atgagcacaatagtgttgatattggCCAGCGTTACTACGACAGCTGCTTC
3422	gumB deletion	attgtgagcggataacaatttcacacaggaaacagctGGTTGAACCCGGTGTGCTGTTGC
3570	wecA-RT	ACCGAGCATCACTTCCTGAT
3571	wecA-RT	GTACCTGACGCTGATGCTGA
3575	flhD-RT	AATGTTTCGCCTGGGTATTG
3576	flhD-RT	ATAGCAAAATGCCGGTATGG
3582	swrW-RT	AATTAGGCGAGATCGAGCAA
3583	swrW-RT	AACAGGACGGCACCATAAAG
3655	yrfF cloning	aaattctgttttatcagaccgcttctgcgttctgatTTATTCGATAAGGCTTTCTGAAGG
3656	yrfF cloning	gtgagcggataacaatttcacacaggaaacagctATGAGCACCATTGTGATTTTTTTAGC
3658	igaA cloning	aattctgttttatcagaccgcttctgcgttctgatTCAGATGAGATTTTCCGGAGAGACG
3659	igaA cloning	gtgagcggataacaatttcacacaggaaacagctATGAGCACCATTCTGATTTTTATAGC
3661	kumO cloning	caaattctgttttatcagaccgcttctgcgttctgatTTAATCTTGCGAGTCAGATGAGG
3662	kumO cloning	aattgtgagcggataacaatttcacacaggaaacagctATGGGCACCTTTCTGATATTCC
3809	fliC-RT	GTATCTCTCTGGCGCAGACC
3810	fliC-RT	ATGGTTTCACCGTCGTTAGC

^a“-RT” indicates that primers are used for qRT-PCR.

^bUppercase letters indicate sequences that prime amplification of the desired DNA, whereas lowercase letters target homologous recombination with a plasmid.

To make complementation plasmids, the igaA-umoB family genes from E. coli (yrfF), Klebsiella pneumoniae (kumO), S. marcescens (gumB), and S. enterica (igaA) were amplified with primers 3655/3656, 3661/3662, 3417/3418, and 3658/3659, respectively. These ORFs were placed under transcriptional control of the E. coli P_lac promoter on pMQ132 (68).

Transmission electron microscopy.

TEM of bacteria was performed using cultures grown overnight in LB medium for 18 to 20 h or taken from agar plates after 48 h of growth, washed in phosphate-buffered saline (PBS), and processed and observed as previously described (53). At least four independent preparations were made for each group.

Secondary metabolite assays.

Prodigiosin was extracted from cultures grown for 18 to 20 h in LB medium using acidified ethanol and the absorbance measured, as previously described (25).

Serratamolide was measured by mass spectrometry from cultures grown for 20 h in LB medium and normalized to an OD₆₀₀ of 2.0, as previously described (27). Zones of growth inhibition for Staphylococcus aureus were measured as previously described (14).

Extracellular polysaccharide quantitation.

S. marcescens polysaccharides were isolated as described by Anderson et al. (48) and Masuko et al. (69). Briefly, bacteria were grown overnight in LB medium for 16 to 20 h and pelleted by centrifugation at 25,000 × g and 15°C for 15 min. The supernatants were discarded, and the bacterial pellets were suspended in 30 ml of PBS and 6 ml of 1% Zwittergent 3-14 (EMD Millipore) in 100 mM citric acid (pH 2) and incubated at 50°C for 20 min. Following incubation, the bacteria were pelleted by centrifugation at 25,000 × g and 15°C for 30 min, and the cell-free supernatants were transferred to 250-ml centrifuge bottles. Four volumes of cold (−20°C) ethanol were added to each and the samples placed at −20°C overnight. After overnight precipitation, the samples were centrifuged at 27,000 × g and 4°C for 45 min, and the supernatants were discarded. The precipitated polysaccharides were allowed to air-dry in a chemical fume hood. The polysaccharides were weighed, and the total carbohydrate content was quantitated using a microplate phenol-sulfuric assay (48, 69).

Biofilm assay.

Biofilm formation on borosilicate glass was performed as previously described (52, 53). Biofilms were stained with crystal violet (0.1% [wt/vol]) and solubilized with glacial acetic acid (33%), and the absorbance at 590 nm was measured using a plate reader (BioTek Synergy 2). Biofilms were generated for 6 and 20 h in LB medium in tubes rotated on a TC-7 tissue culture roller (New Brunswick Scientific).

qRT-PCR and RNA preparation.

RNA preparation from cultures with an OD₆₀₀ of 3.0 and qRT-PCR were performed as previously described (27). Briefly, cultures were treated with RNAprotect reagent (Qiagen), and RNA was purified using an RNeasy kit (Qiagen) and concentrated using a spin column (RNA Clean and Concentrator; Zymo Research). Two rounds of DNase treatment were performed (10 units for 15 min at room temperature on column with Qiagen DNase I, and 1 unit for 30 min at 37°C with Promega RQ1 DNase after purification and prior to the concentration step). Controls for cDNA preparation included no-reverse transcriptase reactions for each RNA sample, included to validate the absence of chromosomal DNA in RNA samples; any samples with detectable chromosomal DNA contamination were excluded prior to experimentation. The coefficient of variance was less than 10% between experiments.

Statistical analysis.

Experiments were done at least twice, with a minimum of three biological replicates. The GraphPad Prism software was used to perform Student's t tests, Mann-Whitney U tests, Fisher's exact test, and one-way ANOVA with Tukey's posttest. Significance was set at a P value of <0.05.

Accession number(s).

The sequences of the gumB gene from S. marcescens K904 and the kumO gene from K. pneumoniae were deposited in GenBank under accession numbers KY098906 and KY098907, respectively.