A LysR-Type Transcriptional Regulator, RovM, Senses Nutritional Cues Suggesting that It Is Involved in Metabolic Adaptation of Yersinia pestis to the Flea Gut

Yersinia pestis has evolved as a clonal variant of Yersinia pseudotuberculosis to cause flea-borne biofilm–mediated transmission of the bubonic plague. The LysR-type transcriptional regulator, RovM, is highly induced only during Y. pestis infection of the flea host. RovM homologs in other pathogens regulate biofilm formation, nutrient sensing, and virulence; including in Y. pseudotuberculosis, where RovM represses the major virulence factor, RovA. Here the role that RovM plays during flea infection was investigated using a Y. pestis KIM6+ strain deleted of rovM, ΔrovM. The ΔrovM mutant strain was not affected in characteristic biofilm gut blockage, growth, or survival during single infection of fleas. Nonetheless, during a co-infection of fleas, the ΔrovM mutant exhibited a significant competitive fitness defect relative to the wild type strain. This competitive fitness defect was restored as a fitness advantage relative to the wild type in a ΔrovM mutant complemented in trans to over-express rovM. Consistent with this, Y. pestis strains, producing elevated transcriptional levels of rovM, displayed higher growth rates, and differential ability to form biofilm in response to specific nutrients in comparison to the wild type. In addition, we demonstrated that rovA was not repressed by RovM in fleas, but that elevated transcriptional levels of rovM in vitro correlated with repression of rovA under specific nutritional conditions. Collectively, these findings suggest that RovM likely senses specific nutrient cues in the flea gut environment, and accordingly directs metabolic adaptation to enhance flea gut colonization by Y. pestis.


Introduction
The bubonic plague causing bacillus, Yersinia pestis can be efficiently transmitted to a mammalian host through the formation of a robust biofilm in the foregut proventriculus of its flea vector. To understand the molecular mechanisms that direct biofilm blockage in the flea, the medium, which we refer to as Sebbane minimal media (SMM) [18]. Both of the Y. pestis defined media serve as a base medium to which a novel carbon or nitrogen source can be added. The TMH medium promotes biofilm formation in Y. pestis and is supplemented with 0.2% galactose and 1μg/mL hemin [19]. TMH media not containing arginine (arg) was also used. To simplify reference to the TMH medium supplemented with 0.2% galactose and 1μg/ mL hemin containing or lacking arg, we refer to them as TMH-gal + or TMH-gal + arg -.
The SMM medium contains the minimal basic metabolites that are necessary for Y. pestis growth, including the five amino acids for which it is auxotrophic (isoleucine, valine, phenylalanine, methionine and glycine) [18]. For growth studies to determine the effect of the novel nutrient on Y. pestis growth, the SMM medium, was supplemented with the sugars, glucose, galactose, maltose, arabinose and N,N'-diacetylchitobiose (Sigma) at a concentration of 0.2% (w/v). Alternately, the amino acids, glutamine (gln), arg and histidine (his) that are not normally contained in SMM were supplemented into SMM at concentrations at which they are found in TMH medium, which are 100mM, 24mM and 2mM respectively. The growth studies were performed for 3-4 independent biological replicates. Growth rate was calculated using linear regression analysis of the exponential phase of growth.

Biofilm assays
To determine ability to form biofilm in LB medium, bacteria were grown in LB broth supplemented with 4 mM CaCl 2 and 4 mM MgCl 2 for 18 h at room temperature and diluted to A 600nm 0.02 in the same medium. To determine ability to form biofilm in TMH-gal + or TMHgal + argmedium, bacteria were grown in TMH supplemented with 0.2% galactose only for 18h at room temperature. These cultures were then first diluted 1:10 in phosphate-buffered-saline (PBS) and subsequently to 1:400 in either TMH-gal + or TMH-gal + argmedium. Aliquots of 100 μl were added to wells of 96-well polystyrene dishes, and incubated with shaking at 250 rpm for 48h at ambient room temperature. Wells were washed three times with water to remove media and planktonic cells and the adherent biofilm was stained with 200 μl of 0.05% safranin for 20 min. Wells were washed three times with water and plates were left to dry for 48h. The bound safranin dye was solubilized with 200 μl of 30% acetic acid for 30mins and the A 450nm was measured immediately. The assays were performed using 3-4 independent biological replicates with three technical replicates each.

Y. pestis mutagenesis and complementation
The Y. pestis ΔrovM mutant strain was created using the pKOBEG method [20] in which the entire target gene was replaced by a Tn703 kanamycin cassette from pUC4K (Amersham). The primer pairs, lxmarovMf/lxbarovMr and rxbarovMf/rhindrovMr (Table 1) were used to generate the left and right flanking regions of the rovM gene from Y. pestis KIM6+ genomic DNA by PCR. These two PCR fragments were cloned into the XmaI/XbaI and XbaI/HindIII sites of pUC19 respectively. The kanamycin cassette was amplified using the kanxbaf and kanxbar primers (Table 1) and cloned into the XbaI site of the pUC19 plasmid containing the rovM flanking regions. The new plasmid pUC19rovM500flanking was used as the template to generate the linear PCR fragment using primers lxmarovMf and rhindrovMr, for mutation using the pKOBEG method [20]. Mutations were verified by DNA sequencing. The ΔrovM::kan mutant is referred to as just ΔrovM here. Primers lxmarovMf and rhindrovMr were used to generate a fragment containing the rovM gene and native promoter region from Y. pestis KIM6+ genomic DNA and this fragment was cloned into a high copy number plasmid TOPO-pCR2.1 (pCR_rovM). The fragment containing the rovM gene and native promoter region was then isolated after digesting the pCR_rovM plasmid with BamHI and XhoI and was cloned into the corresponding sites on the low copy number plasmid pACYC177 (pACrovM). Constructs were verified by DNA sequencing. The ΔrovM mutant was transformed by electroporation, independently with either pACrovM or pCR_rovM to achieve in trans complementation of the rovM mutation. The Y. pestis ΔrovM mutant and wild type strains were transformed with an empty pACYC177 plasmid and a wild type strain was transformed with pCR_rovM construct to use in comparative rovM over-expression experiments. All experiments were subsequently performed with the strains harboring plasmids, unless otherwise indicated. All strains and primers used in this study are listed in Table 1.

Flea infections
Y. pestis KIM6+ (pACYC177), ΔrovM mutant (pACYC177) and ΔrovM (pACrovM) strains used for flea infections were grown in Brain Heart Infusion (BHI) broth overnight twice in succession, first at 28°C and then at 37°C without aeration. Bacterial cultures were centrifuged, and the bacterial pellet was resuspended in 0.5 ml PBS and added to 5 ml of fresh heparinized mouse blood at a concentration of approximately 1X10 9 cells/mL. Xenopsylla cheopis fleas were then allowed to feed on the infected blood using a previously described artificial feeding chamber [21,22]. Fleas that took a blood meal were maintained at 21°C and 75% relative humidity, fed twice weekly on uninfected mice and monitored for proventricular blockage for 28 days as previously described [21]. The infection rate was determined by cfu count of the bacterial load in samples of 20 infected fleas collected immediately after the infectious blood meal, 7 days and 28 days post-infection [21,23]. Co-infections were performed similarly to single infections as previously reported [19,24]. The fleas were allowed to feed on blood containing an approximate 1:1 ratio of either the ΔrovM mutant or the ΔrovM (pACrovM) with the wild-type strain. Bacterial loads of 16-20 infected fleas were determined at 0 and 28 days post-infection, by plating on Yersinia selective agar base (Thermo) supplemented with 1μg/μL irgasan only (YSAB-irg) to determine total cfuY. pestis per flea. Simultaneously, plating was carried out on YSAB-irg plus 50μg/mL kanamycin, or 100μg/mL carbenicillin, to select for the ΔrovM mutant or ΔrovM (pACrovM) respectively. Two independent co-infection assays were performed.

Ethics statement
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol (#04001) was approved by the Committee on the Ethics of Animal Experiments of Washington State University.

RT-qPCR
RNA was isolated from three independent exponential phase cultures and from fleas using the RNeasy RNA isolation kit (Qiagen) and was treated with rDnase I (Ambion) to remove contaminating genomic DNA. For RNA isolation from fleas, triplicate independent pools of~35 flea guts from fleas infected with the ΔrovM or wild-type strain was processed two weeks postinfection as previously described [1]. RNA quality was assessed using an Agilent Bioanalyzer 2100 (Agilent Technologies) and only samples with RIN values of 8 and A 260 /A 280 ratios of 2.0 were used. To confirm that the samples were free of genomic DNA real time PCR of RNA was performed. cDNA was synthesized from 2-5 μgs total RNA using the Superscript III reverse transcriptase (Invitrogen) as per manufacturer's instructions. Each 25μL quantitative PCR reaction was carried out using 20ngs of cDNA per sample, in triplicate and the Taqman Universal PCR Master Mix (Life Technologies) on an ABI Prism 7900 sequence detection system (Applied Biosystems) using the following conditions: 95°C for 10 min, followed by 40 cycles of 95°C for 15s and 60°C for 1min. The Taqman primer and probes sets were designed using Primer Express version 2.0 software (Life Technologies) and are listed in Table 1. Primers and probes sets for all three genes, were used at 500nM and 250nM final concentration, respectively. For each primer-probe set assay, a standard curve was prepared using known concentrations of Y. pestis KIM6+ genomic DNA. Standard curves were used to transform C T values into relative DNA quantity. The quantity of cDNA for each experimental gene was normalized relative to the quantity of the reference gene crr (y1485), whose expression is not affected by in vivo or in vitro growth conditions [1,25,26]. The S1 Dataset contains raw data from the RT-qPCR using SMM medium conditions for which RT-qPCR has not previously been performed using the crr normalization control gene. Each assay was performed in triplicate for the three independent biological samples. To calculate fold change we used a ratio of the normalized values calculated for the gene of interest/crr between two strains e.g fold change in gene of interest transcript between strain A and strain B will be calculated by deriving a ratio from (gene of interest /crr) strain A / (gene of interest /crr) strain B .

Statistical analysis
The GraphPad Prism 5 software was used to statistically analyze the data. A One Way ANOVA with a Tukey's multiple comparison post-test was used to determine any significant differences in biofilm formation, growth rates and transcript levels. For flea co-infection assays, a Student's t-test was used to determine significant difference between the mean percentage infection per strain at days 0 and 28, post-infection. Linear regression analysis of the exponential phase of growth was performed to determine the growth rate (μ) of each strain.

Results
Y. pestis strains harboring multiple plasmid-encoded copies of rovM show differential biofilm formation in response to nutritional cues Homologs of RovM in E. coli (LrhA) and Erwinia chrysanthemi (PecT) have been implicated in the regulation of genes required for aggregation and biofilm production [3,5]. Over-expression associated with the rovM homolog cloned on a multicopy plasmid have been implicated in aberrant phenotypes, while the wild type and mutant strains exhibit equivalent phenotypes e.g over-expression of lrhA from a low copy number plasmid resulted in loss of motility in E. coli while the ΔlrhA mutant and wild type remained motile. [3].
Therefore the ability of a ΔrovM mutant strain to successfully produce a biofilm on polystyrene microtiter plates in comparison to the wild type, and wild type and ΔrovM mutant strains transformed with a plasmid harboring the rovM gene (wild type (pCR_rovM), ΔrovM (pACrovM), and ΔrovM (pCR_rovM)) was assessed (Fig 1). The ΔrovM mutant, produced similar quantities of adherent biofilm relative to the wild type strains containing either high or low copy number empty plasmids, in LB (4mM MgCl 2 , 4mM CaCl 2 ) medium ( Fig 1A). Interestingly, both wild type and ΔrovM strains, transformed with pCR_rovM or pACrovM, such that they contained multiple copies of the rovM gene, appeared to produce equally reduced biofilm, but this was due to less adherent biofilm easily washing away. Growth analyses in LB medium and TMH medium (supplemented with 0.2% galactose), demonstrated that wild type (pCR_rovM), ΔrovM (pCR_rovM) and ΔrovM (pACrovM) grew at significantly faster rates than wild type and the rovM mutant (Fig 2).
When cultured in TMH-gal + medium, biofilm was produced in larger amounts by the wild type (pCR_rovM) and ΔrovM (pACrovM) strains relative to the comparable biofilm formed by the wild type and ΔrovM mutant strains containing empty plasmids ( Fig 2B). Dissimilarly, less adherent biofilm was not observed to be the reason for this difference in TMH-gal + medium.
A ΔrovM mutant has a competitive fitness disadvantage in the flea The RovM homologs LrhA and HexA of the insect pathogens, P. temperata and X. nematophila, regulate expression of virulence factors, immuno-suppression and metabolic usage during infection of their respective insect hosts [6,8]. We observed differential biofilm formation in rovM over-expressing strains in different nutritional environments. We therefore predicted that if Y. pestis RovM maintained similar roles during flea infection, then its absence may impact growth, survival, and/or biofilm-mediated gut blockage.
Therefore, the ability of a ΔrovM mutant strain of Y. pestis to successfully infect and produce biofilm blockage in fleas was assessed (Fig 3). During flea infection the wild type, ΔrovM mutant and complemented mutant ΔrovM (pACrovM) strain displayed comparable biofilm blockage (Fig 3A), infection rates (Fig 3B), average bacterial loads per flea (Fig 3C), and flea mortality (data not shown).
If Y. pestis is deleted of a gene whose product is important for adaptation in the flea gut, a fitness defect could result from its competitive interactions, leading to a less transmissible infection. Therefore to investigate if any potential liability was imposed by a mutation of RovM, a competitive co-infection of this mutant with its isogenic wild type strains was performed. A cohort of fleas was co-infected in a 1:1 ratio with the wild type and ΔrovM mutant Y. pestis strains with multiple plasmid-encoded copies of rovM show differential ability to form biofilm in varying nutritional conditions. Biofilm formation of Y. pestis KIM6+ strains in LB medium supplemented with 4mM MgCl 2 and 4mM CaCl 2 (A) and TMH-gal + medium (B). Error bars are the mean±SD of four independent biological replicates. A One Way ANOVA combined with a Tukey's test was used to test significant differences in the biofilm formation between strains. The overall P-value for the ANOVA was P<0.0001. *** represents P<0.001 and indicates a significant difference in biofilm formed between the strains. Horizontal lines mark a groups of bars that are being compared. or ΔrovM (pACrovM) strain (Fig 4). During co-infection with the wild type a significant decrease in the ΔrovM mutant flea infection load was observed at day 28 post-infection, despite its apparent significantly increased infection load at day 0 post-infection. In converse, during a co-infection with the wild type strain, the ΔrovM pACrovM strain showed a significantly increased infection load at 28 days co-infection. Y. pestis strains with multiple plasmid-encoded copies of rovM grow at a faster rate. Growth of Y. pestis strains in LB medium (A) and TMH supplemented with 0.2% galactose (B). Growth curves are the mean±SD of four independent biological replicates. The mean mean±SD of the growth rate (μ) is given in parentheses. All closed symbol growth curves represent strains containing multiple copies of rovM. A One Way ANOVA combined with a Tukey's test was used to test significant differences in the growth rate between strains. The overall P-value for the ANOVA was P<0.0001. Significant differences in μ between strains are indicated by *** for P<0.001, ** for P<0.01 and * for P<0.05. Curly brackets group together strains that do not have significant differences. Capped bars indicate strains that are being compared. RovM regulates growth in response to nutrients LrhA mutants of X. nematophila exhibit metabolic defects that affect its ability to infect its insect host, foretelling that this regulator senses nutritional composition of its host environment [27].
To determine if Y. pestis RovM is similarly involved in metabolic processes, growth analysis of  wild type, ΔrovM mutant and complemented mutant ΔrovM (pACrovM) in a very minimal medium, SMM, was performed following supplementation with sugars and amino acids that are not normally contained within SMM. Supplementing the media with glucose and galactose, resulted in no observable difference in growth rate among the strains (Fig 5A), however, diacetylchitobiose supplementation resulted in significantly higher growth rate for the ΔrovM (pACrovM) strain as compared to the rovM mutant and the wild type strains. Maltose and arabinose supplementation revealed no growth rate differences between strains (data not shown).
Predominant usage of the L-glutamate group of amino acids (pro, his, arg, gln) is predicted to support Y. pestis growth in the flea gut, therefore his, arg, and gln were tested for their ability to support improved growth of Y. pestis in SMM (Fig 5B). Interestingly, while the ΔrovM mutant and wild type grew similarly in gln and arg, the ΔrovM (pACrovM) strain appeared to have a significantly enhanced growth rate when supplemented with these amino acids. Histidine addition however supported equivalent growth rates for all the strains.

Specific nutritional environments direct induction of rovM and RovMassociated repression of rovA
In Y. pseudotuberculosis, the RovM regulator, represses rovA specifically in a minimal nutritional environment [10], and in Y. pestis, rovM is highly induced, while rovA is highly Adaptation of Yersinia pestis to the Flea Gut repressed in fleas [1]. We hypothesized that in Y. pestis similar and specific nutritional environments would cause repression of rovA by RovM, and that this included the flea gut. However, gene expression analysis of flea infections of a ΔrovM mutant strain, showed no significant difference in rovA expression between the wild type and ΔrovM mutant (Fig 6A  and 6B). Following this finding, in vitro transcriptional levels of Y. pestis rovA and rovM in response to differing nutritional environments was also determined. This was performed to establish the inducing environments of rovM and rovA, and whether a reciprocal modulatory association between the 2 regulators in certain nutritional environments does indeed occur. In Y. pestis, the rovM gene had lower transcriptional levels in minimal SMM media versus rich LB in the wild type strain (Fig 6A). The rovA transcript levels were~3-fold higher than rovM and no difference in rovA levels was present between the wild type and ΔrovM mutant in both media (Fig 6B).
However, the ΔrovM (pACrovM) strain exhibited~49 and~22-fold increases in rovM gene expression in LB and SMM glucose, respectively, relative to the wild type growing in the same medium. Simultaneously, expression of rovA was repressed~3-fold in the rovM over-expressing strain, ΔrovM (pACrovM) in LB medium but no difference in expression of rovA was noted in SMM glucose. The levels of rovA was however not affected by rovM in minimal SMM medium supplemented with glucose. Adaptation of Yersinia pestis to the Flea Gut Arginine is sensed by RovM to regulate growth fitness and biofilm production Supplementation of SMM with arg as the novel nitrogen source led to increased growth rate of the ΔrovM (pACrovM) strain (Fig 5B). Arg is predicted to be catabolized by Y. pestis in the flea gut, therefore the ability of the wild type, ΔrovM mutant and complemented mutant, ΔrovM (pACrovM) to respond to arg in the context of biofilm production was tested (Fig 7A, S1 Dataset). Lack of arg in TMH-gal + argmedium, resulted in significant decreases in biofilm formation in strains with multiple copies of rovM, ΔrovM (pACrovM) and wild type (pCR_rovM) (Fig 7A). However, the wild type and ΔrovM mutant strains were able to make comparable amounts of biofilm independent of the presence of arg in TMH medium (Figs 1B and 7A). The growth rate of strains harboring multiple copies of rovM was higher than the wildtype and ΔrovM mutant in TMH-gal + argmedium (Fig 7B), similar to growth in TMH-gal + (Fig 2B). To assess if any differences in rovM transcription occurs in the presence of arg, the rovM expression levels in SMM medium supplemented with arg was determined. Arg addition did not result in induction of rovM expression as noted by the similar expression levels of rovM in the wild type strain in SMM plus glucose and SMM plus arg (Figs 6A and 8). However, in SMM arg, over-expression of rovM occurs in the ΔrovM (pACrovM) and wild type (pCR_rovM) strains with concomitant rovA repression in keeping with what occurs in LB medium. Arginine is sensed by RovM to control biofilm production. (A) Biofilm formation of Y. pestis KIM6+ strains in TMH-gal + arg -. Error bars are the mean±SD of four independent biological replicates. A One Way ANOVA combined with a Tukey's test was used to test significant differences in biofilm formation between strains. The overall P-value for the ANOVA was P<0.0001. Significant differences in biofilm formation are indicated by * for P<0.05 and *** for P<0.001. (B) Growth of Y. pestis strains in TMH-gal + arg -. Error bars are the mean±SD of four independent biological replicates. The mean mean±SD of the growth rate (μ) is given in parentheses. All closed symbol growth curves represent strains containing multiple copies of rovM. A One Way ANOVA combined with a Tukey's test was used to test significant differences in the growth rate between strains. The overall P-value was P<0.0001. Significant difference in growth rates is indicated by *** for a P-value of 0.0001. Horizontal lines group a set of bars together. Capped lines mark the two bars (or groups of bars) that are being compared. doi:10.1371/journal.pone.0137508.g007

Discussion
To understand which Y. pestis genes were essential for flea gut colonization and adaptation to produce a biofilm blockage-mediated transmissible infection we defined the unique transcriptome of Y. pestis in blocked fleas relative to LB in vitro growth conditions previously [1]. Genes that were uniquely highly induced in the flea gut environment were assumed to be prioritized for flea adaptation and transmission, urging a functional understanding of their roles during Y. pestis flea infection. This study aimed to characterize the functional role of one such gene, rovM that encodes a LTTR.
Our data from this study suggests that rovM may be induced in Y. pestis during flea infection by available carbons and nitrogen sources or perhaps other environmental factors. The competitive fitness defect exhibited by the rovM mutant during co-infection with the wild type suggests that lack of RovM leads to a deficiency at acquiring available nutrients in the flea gut. We previously showed that Y. pestis has a distinctive metabolism in the flea gut which is characterized mainly by predominant usage of peptides and amino acids along with lipid and some pentose carbohydrate utilization [1]. Specifically induction of genes involved in the uptake and catabolism of diacetylchitobiose (chbBC, chbF), arg (artQ), and gln (glnQPH), occurred alongside induced rovM expression in fleas [1]. In this study we demonstrated that these three metabolites support a growth fitness advantage of Y. pestis overexpressing rovM in vitro. Alternately, catabolism of hexose sugars is not thought to be significant in the flea gut [1,28] and we observe that rovM over-expression does not impact growth fitness of Y. pestis in the presence of the hexose sugars, glucose, galactose or maltose. The flea blood meal and its catabolic by-products produced from digestion by the flea and its inherent gut microbiota serve as the nutritional source for Y. pestis survival and growth in the flea gut [28][29][30][31]. Therefore Y. pestis may not only be in competition with the flea and its resident microbiota for the blood meal, but it may need to be capable of utilizing by-products of bloodmeal digestion produced by the flea and its microbiota. The transcriptional regulator, RovM likely directs more efficient utilization of some available nutrients in the flea gut to support Y. pestis adaptation to this environment. Certainly the improved fitness during co-infection with the wild type displayed by the, ΔrovM pACrovM strain that over-expresses rovM, supports this notion. This ability to respond to nutritional cues is a function that has been attributed to homologs of RovM. For example, LrhA in E. coli, negatively regulates galactose and aspartate chemoreceptors thereby controlling bacterial motility in response to acquisition of these nutrients [3]; Y. pseudotuberculosis RovM is induced when cultured in minimal medium, [10]; LrhA of X. nematophila supports propagation of this insect pathogen in a minimal medium containing specific amino acids [27], whereas an ΔlrhA mutant shows no growth differences relative to the wild type strain in LB medium. Richards and Goodrich-Blair [27] suggest that metabolites can act as transition signals that might enable pathogen switching from virulence to metabolic activities, or that consumption of specific nutrients during infection may serve as a signal to express genes required for transmission.
Our results indicate that augmented expression of rovM in a complex rich LB medium, leads to less adherent biofilm formation, whereas in the nutritionally defined TMH medium containing arg, more biofilm is produced. Furthermore, rovM over-expression correlates with a faster growth rate in the presence of arg in minimal medium. Together this implies that RovM influences biofilm formation via integration of nutritional cues and that arg may be central to this signaling. This is in line with the idea that metabolites may act as transition signals. Formation of biofilm in Y. pestis occurs in the flea host only, and is important for biofilmmediated transmission of plague and to maintain persistent colonization of the flea gut [21,22]. The hms group of genes, comprising the hmsHFRS gene locus and single genes, hmsT and hmsP are responsible for synthesis and regulation of the extracellular polysaccharide matrix (ECM) that is fundamental to biofilm formation [22,32]. The ECM facilitates biofilm-mediated transmission and persistence in the flea gut [21,22]. Polyamines are also necessary for formation of Y. pestis biofilm through their modulation of HmsR and HmsT protein translation. Interestingly arg is required as a precursor in the first step of polyamine biosynthesis where it is converted by the arginine decarboxylase to agmatine.
Only elevated transcript levels of rovM, occurring through in trans over-expression of the rovM gene (e.g. pACrovM and pCR_rovM here) in ΔrovM mutant and wild type bacteria resulted in notable alterations in biofilm, flea infection and growth fitness. This is consistent with studies focused on Y. pseudotuberculosis [10] and E. coli [3] RovM homologs wherein only elevated levels of RovM could be correlated with notably distinct aberrant phenotypes when no difference occurred in the wild type and RovM-homolog mutant strains. For instance elevated levels of RovM in Y. pseudotuberculosis resulted in hypermotility and virulence attenuation in mice while there were no apparent differences between the wild type and ΔrovM mutant strains [10]. Furthermore, in the case of plasmid-mediated over-expression of lrhA in E. coli, motility was decreased when compared to motility in an lrhA mutant carrying an empty plasmid [3]. The crystallographic structure of Y. pseudotuberculosis RovM reveals a potential requirement for a metabolic inducer molecule to mediate its folding [33] into an active tetrameric unit. Formation of the tetrameric unit is thus essential for RovM regulatory function, which is consistent with the function of other LTTRs. It can then be assumed that in trans over-expression of rovM will result in more functionally active units of RovM, as compared to the wild type and ΔrovM which synthesize few to no transcripts of rovM. Another hallmark of LTTRs is autoregulation [10,34], and explains that induced RovM synthesis leads to further induction of rovM expression. Between 2-15 copies of the rovM gene is expected to be present in the ΔrovM (pACrovM) strain, based on the copy number of the pACYC177 plasmid [35]. Induction of rovM expression in pACrovM transformed strains, may therefore lead to further induction of rovM, accounting for the strident differences in rovM expression between these strains and the wild type and ΔrovM strain in LB and SMM media (Fig 6).
In conditions when rovM is induced, or over-expressed, as observed in the pACrovM and pCR_rovM transformed strains, or naturally during flea infection, more functional RovM units are present to mediate repression of rovA or regulation of metabolic processes. The nutritional specificity for RovM induction and for RovM-dependent repression of rovA are different in Y. pestis and Y. pseudotuberculosis. The induction of rovM and its repression of rovA during growth of Y. pestis in the flea matches what occurs with in vitro grown Y. pseudotuberculosis in minimal medium at low temperatures [10]. However, a Y. pestis ΔrovM mutant sustains low rovA transcription while growing in the insect host, which is in contrast to Y. pseudotuberculosis RovM mediated repression of rovA [10]. It appears that Y. pestis rovA regulation in the flea gut and in SMM glucose is not mediated through RovM, and probably requires other regulatory factors, similar to temperature and growth phase dependent rovA expression in Y. pseudotuberculosis [10].
Y. pestis has diverged from Y. pseudotuberculosis to lose motility, flagella synthesis [36] and production of invasin, which are all RovM-regulated virulence factors that enhance the invasiveness of Y. pseudotuberculosis. The role of RovM in regulating motility independent of RovA [10] is a conserved regulatory function of RovM homologs across a variety of pathogenic bacterial species e.g., LrhA of the nematode pathogen, X. nematophila [8], LrhA of E. coli [3] and HexA of the phytopathogen, E. carotovora [37]. These functions are absent in Y. pestis and thus no longer constitute the Y. pestis RovM regulon. During the clonal evolution of Y. pestis from Y. pseudotuberculosis, besides gains and losses of gene function through horizontal gene transfer and inactivation [36,38], remodeling of gene regulation and global regulatory networks [39][40][41] has occurred. This has likely been a contributing aspect to adaptation of Y. pestis to its ability to colonize and be transmitted by fleas.
We conclude that the RovM protein may play a role in nutritional sensing and regulating physiological processes for advantageous adaptation and colonization of Y. pestis to the flea gut. This role of RovM may be particularly significant for flea-borne transmission during natural plague cycles when competitive colonization of the flea gut is more prevalent.