A response regulator promotes Francisella tularensis intramacrophage growth by repressing an anti-virulence factor

Summary

The orphan response regulator PmrA is essential for the intramacrophage growth and survival of Francisella tularensis. PmrA was thought to promote intramacrophage growth by binding directly to promoters on the Francisella Pathogenicity Island (FPI) and positively regulating the expression of FPI genes, which encode a Type VI secretion system required for intramacrophage growth. Using both ChIP-Seq and RNA-Seq we identify those regions of the F. tularensis chromosome occupied by PmrA and those genes that are regulated by PmrA. We find that PmrA associates with 252 distinct regions of the F. tularensis chromosome, but exerts regulatory effects at only a few of these locations. Rather than by functioning directly as an activator of FPI gene expression we present evidence that PmrA promotes intramacrophage growth by repressing the expression of a single target gene we refer to as priM (PmrA-repressed inhibitor of intramacrophage growth). Our findings thus indicate that the role of PmrA in facilitating intracellular growth is to repress a previously unknown anti-virulence factor. PriM is the first bacterially-encoded factor to be described that can interfere with the intramacrophage growth and survival of F. tularensis.

Graphical Abstract

Introduction

Francisella tularensis is a Gram-negative facultative intracellular bacterium and the causative agent of the potentially fatal disease tularemia (Sjöstedt, 2007). The ability of F. tularensis to cause disease is strictly dependent on its ability to survive and replicate within macrophage (Gray et al., 2002; Nano et al., 2004; Sjöstedt, 2007; Barel and Charbit, 2013). A locus comprising the Francisella Pathogenicity Island (FPI) is critical to the ability of F. tularensis to grow within macrophage (Nano et al., 2004; Larsson et al., 2005; Nano and Schmerk, 2007). Recent studies have demonstrated that the FPI encodes a secretion system related to the type VI secretion system found in Pseudomonas, Vibrio, and Bacteroidetes species (Barker et al., 2009; Bröms et al., 2010; Russell et al., 2014; Clemens et al., 2015).

Expression of genes on the FPI is positively regulated by the transcription regulators MglA, SspA, and PigR (Lauriano et al., 2004; Brotcke et al., 2006; Charity et al., 2007; Brotcke and Monack, 2008; Charity et al., 2009; Rohlfing and Dove, 2014; Ramsey et al., 2015). MglA and SspA are homologs of E. coli SspA and form a heteromeric complex that associates with RNA polymerase (RNAP) (Baron and Nano, 1998; Lauriano et al., 2004; Charity et al., 2007). PigR is a putative DNA-binding protein that interacts with the RNAP-associated MglA-SspA complex; interaction between PigR and the MglA-SspA complex is necessary for these three regulators to coordinately control the expression of approximately 100 genes, including those found on the FPI (Brotcke and Monack, 2008; Charity et al., 2009; Rohlfing and Dove, 2014). While MglA, SspA, and PigR are found at essentially all promoters in F. tularensis, they exert positive effects only at those that contain a specific sequence motif referred to as the PigR response element, or PRE (Ramsey et al., 2015). Examples of promoters that contain a PRE include the iglA and pdpA promoters present on the FPI (Ramsey et al., 2015). Cells lacking PigR, MglA, or SspA are avirulent, highlighting the importance of these three transcription regulators in the coordinate control of virulence gene expression (Brotcke and Monack, 2008; Wehrly et al., 2009; Charity et al., 2009).

The ability of F. tularensis and the closely related F. novicida to survive and replicate within macrophage is also dependent on the response regulator PmrA (Mohapatra et al., 2007; Sammons-Jackson et al., 2008). PmrA is considered an orphan response regulator as there is no sensor kinase encoded in the same putative operon as pmrA. Initial studies in both F. novicida and in the F. tularensis live vaccine strain (LVS), indicated that PmrA influences the expression of 64–148 genes, exerting both positive and negative effects (Mohapatra et al., 2007; Sammons-Jackson et al., 2008). Subsequent studies in F. novicida also showed that PmrA (otherwise known as QseB) can positively regulate the expression of genes encoding a diguanylate cyclase and a phosphodiesterase that influence biofilm formation by modulating the concentration of cyclic di-GMP (Durham-Colleran et al., 2010; Zogaj et al., 2012).

Although it is not known which of PmrA’s regulatory activities are required for intramacrophage growth, it has been suggested that the ability of PmrA to positively regulate the expression of genes on the FPI is responsible (Mohapatra et al., 2007; Bell et al., 2010; Dai et al., 2011). Indeed, one model that has been proposed to account for the regulatory effects of both PmrA and the MglA-SspA complex on FPI gene expression specifies that PmrA binds directly to FPI promoters and stimulates transcription by interacting with the RNAP-associated MglA-SspA complex (Mohapatra et al., 2007; Bell et al., 2010; Dai et al., 2011). Although this model does not appear to account for any direct involvement of PigR in the control of FPI gene expression it is supported by the findings that purified PmrA can bind directly to the pdpD promoter from the F. novicida FPI, and that MglA and SspA were found to co-precipitate with purified PmrA (Bell et al., 2010).

Using chromatin immunoprecipitation coupled with high-throughput DNA sequencing (ChIP-Seq), we show that PmrA associates with 252 distinct regions of the LVS chromosome and occupies both intragenic and intergenic regions. Using high-throughput sequencing of RNA (RNA-Seq) we show that PmrA appears to directly influence gene expression only when bound at a few of these locations, where it typically functions as a repressor. We do not detect any association between PmrA and the iglA and pdpA promoters on the FPI. Furthermore, we present evidence that the ability of PmrA to directly repress the expression of a single gene accounts for its role in promoting intramacrophage growth and survival. Our findings explain the molecular basis for the requirement of PmrA in intramacrophage growth, uncover a novel anti-virulence factor that is subject to control by PmrA, and have implications for understanding how the expression of genes on the FPI is controlled.

Results

PmrA is found widely across the genome and may be a nucleoid-associated protein

In order to address the question of whether PmrA influences expression of the FPI genes by directly associating with their promoter regions, we used chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-Seq) to determine the location of PmrA on the F. tularensis LVS chromosome. To facilitate the immunoprecipitation of PmrA we constructed a strain of F. tularensis LVS in which the chromosomal copy of the pmrA gene was modified to encode PmrA with a vesicular stomatitis virus-glycoprotein (VSV-G) epitope tag fused to its C-terminus. This results in cells of LVS that synthesize PmrA with a VSV-G epitope tag (PmrA-V) at native levels (LVS PmrA-V).

ChIP-Seq with cells of the LVS PmrA-V strain revealed that PmrA associates with 252 different regions of the F. tularensis chromosome and is located in both intragenic and intergenic regions (Supporting Information Table S1). The wide distribution of PmrA is illustrated in Figure 1A by a representative 400 kb region of the genome. We have previously identified promoter regions in F. tularensis using ChIP-Seq by determining the locations of the β′ subunit of RNAP in cells treated with rifampicin, and by determining the locations of the σ⁷⁰ and σ³² subunits of RNAP (Ramsey et al., 2015). Figure 1A also includes the location of σ⁷⁰, as determined previously by ChIP-Seq (Ramsey et al., 2015), to indicate the location of promoter regions; the location of HipB, which serves as a control for specificity (Ramsey et al., 2015), is also shown. Of the 252 regions enriched for PmrA, 26.6% (67 regions) are associated with promoter regions identified by detection of σ⁷⁰ (Ramsey et al., 2015).

Figure 1. PmrA is found widely across the genome and is found at promoter and non-promoter locations.

A representative illustration of the density of normalized sequencing reads after ChIP-Seq of PmrA (green), MglA (brown), PigR (blue), σ⁷⁰ (orange), and HipB (dark pink) (A) at across a representative 400 kb region of the F. tularensis LVS chromosome (normalized reads are displayed on a log scale); (B) at the FTL_1279–hemL region, where PmrA associates with the FTL_1279 promoter and is found at the 3′ ends of the convergently-transcribed FTL_1282 and hemL genes (normalized reads are displayed on a linear scale); (C) at the surA–ostA region, where PmrA associates with intragenic regions of surA and ostA but is not found at the ostA promoter (normalized reads are displayed on a linear scale). Gray boxes below the read density plots indicate regions of significantly enriched reads with red lines indicating the point of maximum enrichment. Promoter regions, defined as areas with significant enrichment of RNAP with the chromosome (Ramsey et al., 2015), are indicated by the purple boxes below the gene annotations. PmrA is found at promoter regions and non-promoter regions, at both intra- and intergenic locations. HipB is not significantly enriched in these regions. (D) Chart representing the percentage of PmrA locations that overlap with MglA and/or SspA locations.

Because we define a promoter region as one in which we detect the presence of RNAP, it is possible that the PmrA ChIP-Seq peaks which are not found at promoter regions are indicative of PmrA functioning as a repressor at these locations by occluding RNAP. However, of the PmrA-associated regions not associated with promoters, approximately 47% are found entirely within coding regions, locations which are less likely to contain promoters than intergenic regions. We note that 32.5% of all PmrA ChIP-Seq peaks are within genes, and of these intragenic peaks, 18.2% are associated with promoters.

PmrA is not located at the FPI promoters in LVS

It has been suggested that PmrA positively regulates the expression of genes on the FPI by binding immediately upstream of promoters on the FPI and interacting directly with the RNAP-associated MglA-SspA complex (Mohapatra et al., 2007; Bell et al., 2010; Dai et al., 2011). ChIP-Seq with PmrA revealed that PmrA associates with three regions within the FPI, one immediately upstream of pdpE, one in the middle of pdpC, and one upstream of pdpB (Fig. 2). We do not detect any association of PmrA with the iglA or pdpA promoters (Fig. 2), suggesting that in F. tularensis LVS PmrA does not bind directly to either of the two principal promoters on the FPI.

Figure 2. PmrA is not found at the iglA or pdpA promoters of the Francisella Pathogenicity Island.

A representative illustration of the density of normalized sequencing reads after ChIP-Seq of PmrA (green), MglA (brown), PigR (blue), σ⁷⁰ (orange), β′ +rif (purple), and HipB (dark pink) across the FPI (normalized reads are displayed on a linear scale). MglA, PigR, σ⁷⁰, and the β′ subunit of RNAP (in cells grown in the presence of rifampicin to trap RNAP at promoters [β′ +rif]) are found at the iglA and pdpA promoters, while PmrA is not. PmrA is found at three locations within the FPI.

To identify the locations at which PmrA and the MglA-SspA complex may interact on a genome-wide basis, we compared the locations of PmrA (as determined by ChIP-Seq) with the previously published locations of MglA and SspA, as well as with the locations of PigR, a transcription regulator that has been shown to directly interact with the MglA-SspA complex to control gene expression (Brotcke and Monack, 2008; Charity et al., 2009; Rohlfing and Dove, 2014; Ramsey et al., 2015).

As illustrated in Figure 1A, MglA and PigR are found ubiquitously at essentially all promoters across the genome (Ramsey et al., 2015). PmrA is frequently found at locations which are not identified as promoters (Fig. 1B and 1C). PmrA is rarely found at the same locations as MglA and PigR (Fig. 1B and 1C). Indeed, only at 26% of those regions that are occupied by PmrA do we detect MglA and/or SspA at the same location on the chromosome (Fig. 1B and 1D). Additionally, MglA and PigR are found at the iglA and pdpA promoters but PmrA is not (Fig. 2) (Ramsey et al., 2015). Thus, under the conditions of our experiments, PmrA does not occupy either of the principal promoters on the FPI.

Few PmrA sites are associated with direct regulatory effects

In order to determine if all of the locations at which we detect PmrA by ChIP-Seq represent sites at which PmrA controls gene expression, we performed a transcriptomic analysis on cells lacking PmrA. We generated a strain with an in-frame deletion of pmrA (LVS ΔpmrA) and used RNA sequencing (RNA-Seq) to compare transcript abundance in LVS ΔpmrA mutant cells to wild-type cells. We identified 14 annotated genes that are positively regulated by PmrA by a factor of 2 or more (Table 1), and identified 24 annotated genes that are repressed by PmrA by a factor of 2 or more (Table 2). Combining our ChIP-Seq and RNA-Seq data, we find that of the 252 locations at which we detect PmrA, the transcript abundance of associated genes is affected by the loss of PmrA at only 18 (7%) of these sites. (Genes are considered to be associated with a ChIP-Seq peak if the translation start site is within 600 bp upstream to 1 kb downstream of the position of maximal enrichment.) These findings indicate that at most of the regions that are occupied by PmrA, PmrA is not functioning to directly influence gene expression, at least under our experimental conditions.

Table 1.

Genes positively regulated by PmrA.

Locus number	Gene name	Product	Fold changea	PmrA locationb
FTL_0554	rnc	ribonuclease III	−6.0	-
FTL_0553	lepB	signal peptidase I	−3.9	-
FTL_1190	grpE	heat shock protein GrpE	−3.6	-507
FTL_1264	folB	dihydroneopterin aldolase	−2.6	-
FTL_1697	-	metal ion transporter	−2.1	-
FTL_0531	-	hypothetical protein, pseudogene	−2.0	-
FTL_0560	-	hypothetical protein	−2.0	-
FTL_0633	-	rhodanese-related sulfurtransferase	−2.0	−1
FTL_0634	-	NADH oxidase	−2.0	−544
FTL_0709	lpcC	glycosyl transferases group 1 family protein	−2.0	-
FTL_0777	-	hypothetical protein	−2.0	-
FTL_0997	-	hypothetical protein	−2.0	-
FTL_1200	cobS	cobalamin (vitamin B12) synthesis protein/P47K family protein	−2.0	-
FTL_1682	-	hypothetical protein	−2.0	-

^aFold change represents ratio of gene expression level in LVS ΔpmrA compared to LVS, p < 0.01. Only those genes whose expression changed by a factor of 2 or more are listed.

^bPmrA location is reported if the position of maximal enrichment of a PmrA peak is found within 1000 bp upstream to 600 bp downstream of the translation start site of the gene. Longer distances are indicated only if the gene is part of a putative operon with a gene reported in this table.

Table 2.

Genes negatively regulated by PmrA.

Locus number	Gene name	Product	Fold changea	PmrA locationb
FTL_0702	priM	PmrA-repressed inhibitor of intramacrophage growth	298.0	−58
FTL_0815	-	PRC-barrel protein, gene fragment	69.0	−520
FTL_0816	-	hypothetical protein, gene fragment	40.4	−268
FTL_0814	-	hypothetical protein, gene fragment	16.5	−934
FTL_0701	mreA	FAD binding family protein	7.9	−908
FTL_1401	-	hypothetical protein	7.7	−1059
FTL_0152	-	hypothetical protein	4.8	75
FTL_0810	-	cation transport regulator, gene fragment	4.3	−1339
FTL_0841	-	lipoprotein	3.5	−405
FTL_1635	-	chitinase, fragment	3.3	−31
FTL_1636	-	hypothetical protein	2.7	−2120
FTL_0655	-	hypothetical protein	2.7	-
FTL_0840	-	hypothetical protein	2.5	−73
FTL_1114	-	isochorismatase hydrolase family protein	2.5	9
FTL_0881	-	hypothetical protein	2.3	−270
FTL_0049	dtd	D-tyrosyl-tRNA(Tyr) deacylase	2.2	565
FTL_0943	-	Sodium/hydrogen exchanger family protein	2.2	−157
FTL_0447	-	hypothetical protein	2.2	48
FTL_0470	-	hypothetical protein	2.2	−149
FTL_0857	lytB	hypothetical membrane protein	2.1	21
FTL_1219	-	hypothetical protein	2.1	−2
FTL_0935	-	hypothetical protein	2.1	-
FTL_1402	isftu1	transposase	2.0	−207
FTL_0299	murQ	N-acetylmuramic acid 6-phosphate etherase	2.0	-

^aFold change represents ratio of gene expression level in LVS ΔpmrA compared to LVS, p < 0.01. Only those genes whose expression changed by a factor of 2 or more are listed.

^bPmrA location is reported if the position of maximal enrichment of a PmrA peak is found within 1000 bp upstream to 600 bp downstream of the translation start site of the gene. Longer distances are indicated only if the gene is part of a putative operon with a gene reported in this table.

PmrA mainly functions as a repressor

Of the 24 genes negatively regulated by PmrA, 21 are either associated with a PmrA site or are in a putative operon associated with a PmrA site (Table 2). Indeed, at the majority of sites (16 of 18) where PmrA appears to exert a regulatory effect on the expression of associated genes, PmrA functions as a negative regulator. This indicates that when PmrA directly controls the expression of a particular gene, it generally functions as a repressor. It is notable that two genes repressed by PmrA, FTL_1219 and FTL_0881, are known to be positively regulated by MglA and SspA (Table 2) (Brotcke et al., 2006; Charity et al., 2007). PmrA and the MglA-SspA complex therefore appear to exert opposite effects on the expression of some genes.

PmrA exerts its largest effect as a negative regulator of the hypothetical gene FTL_0702, which we have named priM (PmrA-repressed inhibitor of intramacrophage growth). The loss of PmrA results in an approximately 300-fold increase in priM transcript abundance (Table 2; Fig. 3A). In LVS ΔpmrA mutant cells, ectopic expression of pmrA from a heterologous promoter resulted in repression of priM expression as analyzed by qRT-PCR (Fig. 3B). Furthermore, of the 252 regions found to be occupied by PmrA using ChIP-Seq, the putative promoter of priM is the region with the highest degree of enrichment of PmrA (over 500-fold) (Fig. 3A). Taken together, the significant occupancy of PmrA we observe at the putative priM promoter and the large increase in priM transcript abundance we observe in the absence of PmrA suggests that PmrA directly represses the expression of priM.

Figure 3. PmrA directly represses the expression of priM.

(A) A representative illustration of normalized sequencing reads after ChIP-Seq of HipB (dark pink) and PmrA (green), RNA-Seq of LVS and LVS ΔpmrA (minus strand in gold, plus strand in blue), with all normalized reads displayed on a linear scale. The minus strand RNA-Seq data are truncated at 2000 normalized reads. The location with the most PmrA enrichment is found at the priM promoter and priM transcript levels increase dramatically in the absence of PmrA. Transcript levels of mreA also increase, due to the increase in reads mapping to the 3′ end of mreA. (B) Quantitative RT-PCR analysis shows that in cells lacking PmrA, priM transcript abundance increases approximately 300-fold (LVS ΔpmrA pF), validating RNA-Seq data. Ectopic expression of PmrA using the vector pF-PmrA can restore the repression of priM to wild-type levels (LVS ΔpmrA pF-PmrA compared to LVS pF). The figure depicts data from a representative experiment performed with biological triplicates. Transcripts were normalized to tul4, whose expression is not influenced by PmrA. Error bars represent ± 1 standard deviation from the value (calculated using the mean threshold cycle).

In our transcriptomic analyses, we did not detect changes in FPI gene expression in cells lacking PmrA, indicating that PmrA does not appreciably regulate FPI gene expression under the conditions of our experiment. Of the 14 genes we found to be positively regulated by PmrA, those most strongly influenced are found directly downstream of pmrA (FTL_0552) in a putative operon. Specifically, transcripts of the lepB (FTL_0553) and rnc (FTL_0554) genes exhibit the largest decrease in abundance in LVS ΔpmrA mutant cells when compared to wild-type cells (3.9- and 6-fold, respectively) (Table 1). Although it has been reported that purified PmrA binds the pmrA promoter in vitro (Bell et al., 2010), we did not detect any association between PmrA and the pmrA promoter in vivo by ChIP-Seq (Supporting Information Fig. S1). Moreover, unlike the effects of a pmrA deletion on priM expression, we found that the effects of a pmrA deletion on rnc transcript abundance could not be complemented by ectopic expression of pmrA (Supporting Information Fig. S2). These findings suggest that PmrA does not directly control the expression of the pmrA operon (which includes lepB and rnc), and that the decrease in expression of lepB and rnc that we observe in cells of the ΔpmrA mutant are likely the result of polar effects of the pmrA deletion.

Of the remaining 12 genes which we found to be positively regulated by PmrA, we find only three genes which are associated with regions of PmrA enrichment (Table 1). These findings suggest that in LVS PmrA does not function directly to control expression of the majority of positively-regulated genes.

Repression of priM by PmrA is required for intramacrophage growth

It has been proposed that PmrA is required for intramacrophage growth because it directly positively regulates the expression of genes on the FPI (Mohapatra et al., 2007; Bell et al., 2010; Dai et al., 2011). However, our ChIP-Seq studies indicate that PmrA does not occupy either of the two principal promoters on the FPI, and our RNA-Seq studies indicate that PmrA does not influence the expression of FPI genes. We therefore wondered whether cells lacking PmrA might be unable to replicate within macrophages due to the de-repression of priM expression that occurs in these cells. In order to test this hypothesis, we used a macrophage infection model. The results depicted in Figure 4A show that cells of the LVS ΔpmrA mutant strain are attenuated for intramacrophage growth by about four orders of magnitude compared to wild-type LVS cells in murine macrophage-like J774A.1 cells. Cells of a LVS ΔpmrA ΔpriM double mutant strain were attenuated only by one order of magnitude, indicating that deletion of priM in cells of a ΔpmrA mutant enhances intramacrophage survival by three orders of magnitude (Fig. 4A). The loss of priM alone (i.e. in pmrA+ cells) did not affect the ability of bacteria to survive and replicate within macrophage (Fig. 4A). Our results are not due to differences in bacterial uptake, as equivalent numbers of cells of LVS wild-type, ΔpmrA mutant, ΔpmrA ΔpriM double mutant, and ΔpriM mutant were recovered from infected macrophage after 2 hours of infection (Supporting Information Fig. S3A). Cells of the LVS wild-type, ΔpmrA mutant, ΔpmrA ΔpriM double mutant, and ΔpriM mutant grew indistinguishably from one another in vitro, indicating that any differences in intramacrophage growth could not be attributed to differences in growth in vitro. These findings suggest that the attenuation of cells lacking PmrA is attributable, for the most part, to priM, which we propose is an anti-virulence factor.

Figure 4. Production of PriM is inhibitory to intramacrophage growth and survival.

Growth and survival of wild-type LVS (LVS) and cells of the indicated mutant derivatives within J774A.1 cells. Murine macrophage-like J774A.1 cells were infected with cells of the indicated strains at a multiplicity of infection of approximately 5–10. Cells were lysed and bacteria were plated for enumeration (colony forming units [CFU]) 24 hours post-infection. (A) Deletion of priM (LVS ΔpmrA ΔpriM) rescues the intramacrophage growth defect of ΔpmrA mutant cells (LVS ΔpmrA). Cells lacking only PriM (LVS ΔpriM) grow within macrophage as well as wild-type cells (LVS). (B) Deletion of priM (LVS ΔpigR ΔpriM) does not rescue the intramacrophage growth defect of ΔpigR mutant cells (LVS ΔpigR). (C) Introduction of a stop codon into mreA (LVS ΔpmrA mreA(S)) does not rescue intramacrophage growth defect of ΔpmrA mutant cells (LVS ΔpmrA). Cells with a stop codon in mreA (LVS mreA(S)) grow within macrophage as well as wild-type cells (LVS). (D) Introduction of a stop codon into priM (LVS ΔpmrA priM(S)) rescues the intramacrophage growth defect of ΔpmrA mutant cells (LVS ΔpmrA). Cells with a stop codon in priM (LVS priM(S)) grow within macrophage as well as wild-type cells (LVS). (E) Ectopic expression of priM (LVS pF2-PriM) results in decreased intramacrophage growth, compared to wild-type cells with empty vector (LVS pF2).

We considered the possibility that deletion of priM might confer a non-specific growth advantage to attenuated strains in the macrophage infection model. However, we found that the rescue of intramacrophage growth after deletion of priM is specific to cells of a ΔpmrA mutant; deletion of priM does not result in an increase in the ability of cells of a ΔpigR mutant to replicate within macrophage (Fig. 4B).

Our RNA-Seq studies reveal that expression of mreA (FTL_0701) is increased in cells lacking pmrA (Table 2, Fig. 3A). The gene encoding MreA, which is on the plus strand, is located upstream of priM, which is encoded on the minus strand; the two genes are transcribed convergently (Fig. 3A). The increase in mreA transcript abundance in cells of a ΔpmrA mutant appears to be largely due to reads mapping to its 3′ end (Fig. 3A). To test whether the derepression of priM expression that occurs in the absence of PmrA might influence intramacrophage growth through an effect on the expression of mreA, we generated a mutant of mreA in which we introduced a stop codon at position 32 (LVS mreA(S)), which should prevent production of the full length MreA. Introduction of a stop codon into mreA had no effect on the ability of cells to infect or survive and replicate within macrophage, and did not rescue the attenuation of a ΔpmrA mutant (Fig. 4C; Supporting Information Fig. S3B). These findings suggest that PmrA does not influence intramacrophage growth through MreA.

We next asked if it is the production of PriM, rather than the RNA encoded by priM, that is responsible for the intramacrophage growth defect of ΔpmrA mutant cells. In order to answer this question, we generated a mutant of priM with a stop codon at position 17, which should prevent production of the full-length PriM protein. Cells with the stop codon at position 17 in priM (LVS priM(S)) infect and grow as well in macrophage as wild-type cells (Fig 4D; Supporting Information Fig. S3C), similar to cells of a ΔpriM mutant. However, cells of a ΔpmrA mutant that contain a stop codon at position 17 in priM behave indistinguishably from cells of the LVS ΔpmrA ΔpriM double mutant (Fig. 4D; Supporting Information Fig. S3C). These findings are consistent with the idea that it is the protein product of priM that is inhibitory to intramacrophage growth.

We next asked if production of PriM alone is sufficient to prevent the intramacrophage growth of F. tularensis. Wild-type cells expressing priM from the heterologous promoter on plasmid pF2 infect macrophage as well as wild-type cells, but are attenuated for intramacrophage growth approximately 10-fold in comparison to wild-type cells containing an empty vector (Fig. 4E; Supporting Information Fig. S3D). These wild-type cells in which PriM is synthesized ectopically grow indistinguishably from wild-type cells containing the empty vector in vitro, indicating that the effect of ectopic synthesis of PriM on intramacrophage growth cannot be attributed to any differences in growth rate observed in vitro. These findings indicate that production of PriM, in the absence of any other changes which may occur as a result of deletion of pmrA, is sufficient to inhibit intramacrophage growth.

PriM does not generally interfere with the activity of the Type VI Secretion System

The de-repression of priM that occurs in cells of a ΔpmrA mutant results in a large decrease in intramacrophage growth. One system that is known to be essential for intramacrophage growth is the type VI secretion system (T6SS). We therefore asked whether PriM might inhibit intramacrophage growth by interfering with the function of the type VI secretion system. To do this we modified a recently established protocol (Clemens et al., 2015) for use in LVS to examine the amount of IglC secreted in the presence of 5% potassium chloride by cells lacking PigR, PmrA, PriM, or PmrA and PriM, compared to wild-type (Supporting Information Fig. S4). While cells lacking PigR do not synthesize as much IglC as wild-type cells (as expected), the loss of PmrA and/or PriM appears to have little effect on IglC synthesis or secretion (Supporting Information Fig. S4). From these results we conclude that it is unlikely that PriM inhibits intramacrophage growth by preventing secretion of IglC or interfering with the assembly of the T6SS. We cannot, however, rule out the possibility that PriM might exert its effect by interfering with the secretion of a specific T6SS effector.

Discussion

Using ChIP-Seq we have found that PmrA occupies 252 different regions of the LVS chromosome. We find that the ability of PmrA to function as a repressor at only one of these locations explains why PmrA is required for intramacrophage growth. Specifically, we have obtained evidence that de-repression of a gene we refer to as priM is detrimental to the intramacrophage growth of cells that lack PmrA. Our findings demonstrate that the repression of an anti-virulence factor by PmrA is essential for the intramacrophage growth and survival of F. tularensis.

It has been proposed that the ability of PmrA to positively regulate the expression of genes on the FPI accounts for the intramacrophage growth defect of cells lacking PmrA (Fig. 5A) (Mohapatra et al., 2007; Sammons-Jackson et al., 2008; Bell et al., 2010; Dai et al., 2011). Our results suggest that the major contribution of PmrA to intramacrophage survival is not as a positive regulator of the FPI genes but rather as a negative regulator of priM (FTL_0702) which encodes a novel anti-virulence factor. Based on our results, we propose a model to account for the role of PmrA in intramacrophage growth in which PmrA directly binds the priM promoter and represses gene expression by occluding RNAP (Fig. 5B). Our model specifies that RNAP (together with the RNAP-associated factors MglA, SspA, and PigR), can bind the priM promoter and initiate transcription only when PmrA is not bound to this region (Fig. 5C).

Figure 5. PmrA promotes intramacrophage growth by functioning as a repressor.

(A) Prior model suggesting PmrA directly activates expression of FPI genes to promote intramacrophage growth (Mohapatra et al., 2007; Bell et al., 2010; Dai et al., 2011). PmrA binds to a specific site (in green) at FPI gene promoters and interacts with the RNAP-associated MglA-SspA complex to positively regulate expression of FPI genes. (B) and (C) current model describing how PmrA functions to promote intramacrophage survival. (B) PmrA binds to the priM promoter, occluding RNAP and preventing production of PriM, which allows cells to grow within macrophage. In this model, we have illustrated PmrA binding to a site between the −10 and −35 elements of the priM promoter although the actual position of the binding site is not known. (C) In the absence of PmrA, RNAP (together with PigR and the MglA-SspA complex) can bind the priM promoter, resulting in production of PriM and inhibition of intramacrophage growth.

Using RNA-Seq we identified 38 genes that are subject to control by PmrA, the majority of which, including priM, are repressed by this regulator. In contrast to previous studies in both F. novicida and LVS (Mohapatra et al., 2007; Sammons-Jackson et al., 2008), we did not find that PmrA regulates the expression of FPI genes, at least under the conditions of our experiments (Tables 1 and 2). We note that our study was performed using a mutant of LVS that contains an unmarked in-frame deletion of pmrA, whereas prior studies used mutants in which the pmrA gene was replaced or interrupted with a kanamycin resistance cassette (Mohapatra et al., 2007; Sammons-Jackson et al., 2008). Consistent with the idea that in LVS PmrA does not directly regulate the expression of FPI genes, we did not detect any association between PmrA and the two key FPI promoters in this organism by ChIP-Seq (Fig. 2). However, studies with purified PmrA indicate that PmrA can bind the pdpD promoter from the F. novicida FPI in vitro (Bell et al., 2010). Unlike F. novicida, LVS does not contain the DNA sequence corresponding to the pdpD promoter and so our studies do not address whether PmrA occupies this promoter in vivo. The presence of a PmrA binding site at the pdpD promoter on the F. novicida FPI could explain some of the differences between our study and the prior study in F. novicida (Mohapatra et al., 2007). We note, however, that in agreement with the findings of our study, priM (FTL_0702) was identified previously as being repressed by PmrA in both LVS and in F. novicida (Mohapatra et al., 2007; Sammons-Jackson et al., 2008); priM is duplicated in F. novicida and encoded by the genes FTN_1260 and FTN_1261.

Several different models have been proposed to account for how MglA and SspA contribute directly to the expression of genes on the FPI. One of these specified that the RNAP-associated MglA-SspA complex serves as a contact site for PmrA bound at all of the FPI promoters (Fig. 5A) (Mohapatra et al., 2007; Bell et al., 2010; Dai et al., 2011). Given that we do not detect PmrA at either the iglA or pdpA promoter on the FPIs in LVS, and given that we do not find any evidence that PmrA positively regulates the expression of FPI genes, we think it unlikely that this model can account for the involvement of MglA and SspA in the direct control of these promoters. Our findings are consistent with an alternative model in which the RNAP-associated MglA-SspA complex serves as a contact site for the putative DNA-binding protein PigR at the iglA and pdpA promoters (Charity et al., 2009; Rohlfing and Dove, 2014; Ramsey et al., 2015).

Although PmrA associates with 252 distinct regions of the LVS chromosome, our RNA-Seq analyses indicate that PmrA exerts effects on gene expression directly at few of these locations. It is possible that the occupancy of these sites, including those that are intragenic and not associated with promoters, may play more pronounced regulatory roles at other points of the growth curve, or under growth conditions that are different from the one we tested (i.e. reflect situations where regulation is cryptic). It is also possible that, rather than influence gene expression, the occupancy of these sites may influence chromosome structure and reflect a NAP function for PmrA. Indeed, a growing body of evidence stemming from genome-scale studies suggests that other transcription regulators can occupy many sites but only exert regulatory effects at a few (Shimada et al., 2008; Minch et al., 2015; Visweswariah and Busby, 2015). Additionally, it has been suggested that transcription factor binding may influence the accessibility of DNA to both mutagenic agents and repair enzymes (Warnecke et al., 2012). Thus, transcription factor binding within coding regions by proteins such as PmrA may shape the evolution of a gene (Warnecke et al., 2012).

Although many genes are known to be required for the intramacrophage growth and survival of Francisella, to the best of our knowledge priM is the first example of a gene from F. tularensis whose expression is inhibitory to intramacrophage growth. PriM is annotated as a hypothetical protein of unknown function that is predicted to contain a signal sequence at its N-terminus. Although we do not currently know the mechanism by which PriM inhibits the intramacrophage growth of F. tularensis, there is at least one precedent from a study in Salmonella for an anti-virulence factor that functions by inhibiting intramacrophage growth (Pontes et al., 2015). Moreover, in F. novicida, Cas9 of the CRISPR-Cas system promotes virulence by inhibiting the expression of the FTN_1103 gene encoding a lipoprotein (Sampson et al., 2013). In this situation, FTN_1103 functions as an anti-virulence factor not by inhibiting intramacrophage growth (Weiss et al., 2007), but rather by enhancing the inflammatory response (Jones et al., 2012). Because Cas9 encoded by F. tularensis strains may not be functional it is unclear whether Cas9 promotes the virulence of these strains by repressing the synthesis of bacterial lipoproteins (Sampson et al., 2013). PriM and PmrA are conserved amongst species of F. tularensis and F. novicida, suggesting that inhibition of priM expression by PmrA may be a conserved mechanism for promoting the intramacrophage growth and survival of Francisella species.

Experimental Procedures

Growth conditions

F. tularensis subsp. holarctica LVS and its derivatives were grown aerobically at 37°C in Mueller Hinton (MH) broth (Difco), supplemented with glucose (0.1%), ferric pyrophosphate (0.025%), and Isovitalex (2%), or on cystine heart agar (Difco) medium supplemented with 1% hemoglobin solution (VWR) (CHAH). When appropriate, kanamycin was used for selection at 5 μg ml⁻¹. Plasmid construction was performed using Escherichia coli strain XL1-blue (Stratagene) and, when appropriate, 50 μg ml⁻¹ kanamycin was used to select for resistance. E. coli containing plasmid pF2-PriM were grown at 30°C.

Vector construction

The plasmid pKL13 (Ramsey et al., 2015) was modified to generate the vector for the PmrA VSV-G tagging integration construct. The last 527 bp of pmrA was amplified using a 5′ primer containing a KpnI site and a 3′ primer containing a NotI site, allowing the fragment to be fused with DNA specifying an alanine linker and the VSV-G epitope tag. The fragment was subcloned into KpnI/NotI digested pKL13. The resulting plasmid, pKL15, contains the DNA specifying the 3′ end of pmrA and was used to generate strain LVS PmrA-V.

The vector pEX18Kan was modified to make the in-frame deletion constructs for pEXΔpmrA and pEXΔpriM as described previously (Charity et al., 2007).

Plasmid pEX18Kan was also modified to make mutant genes by allelic replacement. Plasmid pEX-mreA(S) was generated by amplification of regions that flank and overlap base pairs 91–96 of mreA (FTL_0701) by PCR, using primers that integrate a PsiI restriction site (TTATAA; coding for leucine followed by a stop codon) in place of base pairs 91–96 (TTAAAA). Plasmid pEX-priM(S) was generated by amplification of regions that flank and overlap base pairs 46–54 of priM by PCR, using primers that integrate an AflII site (AflII site underlined: ACTTAAGTT; coding for threonine, a stop codon, and valine) in place of base pairs 46–54 (ACTTCAGTT). PCR products were ligated together and cloned into pEX18Kan.

Plasmid pF2-PriM was made by replacing sspA in the plasmid pF2-SspA (Charity et al., 2007), which allows for production of PriM under the control of a weakened groES promoter; the pF2 plasmid (Charity et al., 2007) used as a control does not contain DNA encoding PriM. Plasmid pF-PmrA, which synthesizes PmrA under the control of the full groES promoter, was made by cloning the pmrA gene into pFNLTP7 gro-gfp as described (Charity et al., 2009).

Strain construction

Plasmids were integrated into LVS as previously described (Maier et al., 2004). Cells in which a single homologous recombination event occurred between the integration vector and the chromosome were selected on CHAH containing 5 μg ml⁻¹ kanamycin. Strains containing the modified genetic region of interest were confirmed by colony PCR, Western blot, Southern blot, and/or sequencing PCR products made from genomic DNA. Strain LVS PmrA-V, which synthesizes PmrA with a C-terminal VSV-G tag, was generated by electroporation and subsequent integration of plasmid pKL15 into LVS. Strains containing deletions or mutations were generated by allelic exchange. LVS ΔpmrA, which lacks the pmrA gene, was generated from LVS using plasmid pEXΔpmrA. LVS ΔpriM, which lacks the priM gene, was generated from LVS using plasmid pEXΔpriM. LVS ΔpmrA ΔpriM, which lacks both the pmrA and priM genes, was generated from LVS ΔpriM using plasmid pEXΔpmrA. LVS mreA(S), in which a stop codon is incorporated at position 32 of mreA, was generated from LVS using plasmid pEX-mreA(S). LVS ΔpmrA mre(S), which lacks the pmrA gene and contains a stop codon at position 32 of mreA, was generated from LVS mreA(S) using plasmid pEXΔpmrA. LVS priM(S), in which a stop codon is incorporated at position 17 of priM, was generated from LVS using plasmid pEX-priM(S). LVS ΔpmrA priM(S), which lacks the pmrA gene and contains a stop codon at position 17 of priM, was generated from LVS priM(S) using plasmid pEXΔpmrA.

ChIP-Seq and data analysis

ChIP-Seq was performed in triplicate with LVS PmrA-V cells as previously described (Ramsey et al., 2015). Data analysis was performed as described previously (Ramsey et al., 2015). Sequencing reads have been submitted to the NCBI Sequence Read Archive (SRA, http://www.ncbi.nlm.nih.gov/Traces/sra/) with the accession number SRP074325.

RNA isolation

LVS cells and derivatives were grown in liquid culture until cultures reached an optical density at 600 nm (OD₆₀₀) of approximately 0.35. RNA was isolated as described previously (Rohlfing and Dove, 2014).

RNA sequencing and data analysis

Sequencing libraries were made from RNA samples essentially as described previously (Jorth et al., 2014). Briefly, ribosomal RNAs were depleted from 5 μg of total RNA using the Ribo-Zero Magnetic Kit (Bacteria) (Epicentre) according to the manufacturer’s specifications. The remaining RNA was ethanol-precipitated with 10 μg linear acrylamide and resuspended in RNase-free water. RNA was fragmented using the NEBNext Magnesium Fragmentation Module (New England Biolabs) according to the manufacturer’s protocol and ethanol-precipitated. The resulting RNA was used to generate RNA-Seq libraries using the NEBNext Multiplex Small RNA Library Prep Kit for Illumina (New England Biolabs) according to the manufacturer’s protocol. Libraries were sized-selected by PAGE using a 5% gel with TBE (Biorad), stained with SYBR gold nucleic acid stain (Life Technologies), and visualized using a blue-light transillumination system. Fragments corresponding to ~150–300 bp were gel purified, ethanol-precipitated, and resuspended in TE buffer according to the NEBNext Multiplex Small RNA Library Prep Kit for Illumina (New England Biolabs) protocol. Libraries were sequenced by Elim Biopharmaceuticals, Inc. (Hayward, CA), using an Illumina HiSeq 2500 which generated 50 bp single-end reads. Sequencing reads have been submitted to the NCBI Sequence Read Archive (SRA, http://www.ncbi.nlm.nih.gov/Traces/sra/) with the accession number SRP074325.

Sequencing reads were trimmed with Flexbar (version 2.4) (Dodt et al., 2012) to remove adapter sequences incorporated during cDNA library preparation. Rockhopper (McClure et al., 2013) was used to map reads to the F. tularensis subsp. holarctica LVS genome (NCBI locus AM233362), generate files to visualize read alignment, and perform differential gene expression analysis. We further considered genes which are more than 2-fold altered in transcript abundance and have q-values less than or equal to 0.05. Read alignment tracks were visualized using the Integrative Genomics Viewer (IGV), version 2.3.59 (Thorvaldsdóttir et al., 2013).

cDNA synthesis and qRT-PCR

cDNA synthesis using Superscript III reverse transcriptase (Life Technologies) and qRT-PCR using FastStart Essential DNA Green Master (Roche) with the Roche LightCycler 96 detection system were performed essentially as described (Charity et al., 2007). Abundance of the priM and rnc transcripts were measured relative to the tul4 transcript. Experiments were performed twice in biological triplicate.

Macrophage replication assays

Approximately 2×10⁴ cells of the murine macrophage-like J774A.1 cells were seeded in wells of a 96-well plate in DMEM (ATCC) supplemented with 10% bovine fetal serum (Gemini Bio-Products) and incubated overnight at 37°C with 5% CO₂. Macrophage were infected with LVS or derivative strains at an MOI of 5–10, using 3 wells per strain. After 2 hours, cells were washed twice with PBS and covered with DMEM containing 10 μg mL⁻¹ gentamycin. Cells were incubated at 37°C with 5% CO₂ until enumeration, for either 2 or 24 hours after addition of gentamycin. To identify the number of intracellular bacteria, macrophage were washed twice with PBS, lysed for 5 minutes with 1% saponin in PBS at room temperature, diluted in PBS, and plated on CHAH. Each experiment was performed at least twice.

Type VI Secretion Assay

Assays were performed as described (Clemens et al., 2015) with some modifications. Briefly, LVS or derivative strains were grown in supplemented brain heart infusion broth with cysteine (BHIc) (Mc Gann et al., 2010) with or without 5% KCl. Cells were pelleted by centrifugation and a 0.2 micron filter was used to sterilize the supernatant; resulting samples were analyzed by SDS-PAGE and immunoblotting.

Immunoblots

Proteins were separated by SDS-PAGE on 12% Bis-Tris NuPAGE gels in MOPS running buffer (Life Technologies). The XCell II Blot Module (Life Technologies) was used to transfer proteins to PVDF. Membranes were probed as described (Ramsey et al., 2015), using anti-GroEL (diluted 1:160,000; provided by Karsten Hazlett, Albany Medical College, Albany, New York, United States) or anti-IglC (diluted 1:1,000; BEI resources) and subsequently incubated with polyclonal goat anti-rabbit (diluted 1:5,000; Pierce) or polyclonal goat anti-mouse (1:5,000; Pierce), respectively.