ProQ-dependent activation of Salmonella virulence genes mediated by post-transcriptional control of PhoP synthesis

ABSTRACT

Gastrointestinal disease caused by Salmonella enterica is associated with the pathogen’s ability to replicate within epithelial cells and macrophages. Upon host cell entry, the bacteria express a type-three secretion system encoded within Salmonella pathogenicity island 2, through which host-manipulating effector proteins are secreted to establish a stable intracellular niche. Transcription of this intracellular virulence program is activated by the PhoPQ two-component system that senses the low pH and the reduced magnesium concentration of host cell vacuoles. In addition to transcriptional control, Salmonella commonly employ RNA-binding proteins (RBPs) and small regulatory RNAs (sRNAs) to regulate gene expression at the post-transcriptional level. ProQ is a globally acting RBP in Salmonella that promotes expression of the intracellular virulence program, but its RNA repertoire has previously been characterized only under standard laboratory growth conditions. Here, we provide a high-resolution ProQ interactome during conditions mimicking the environment of the Salmonella-containing vacuole (SCV), revealing hundreds of previously unknown ProQ binding sites in sRNAs and mRNA 3′UTRs. ProQ positively affected both the levels and the stability of many sRNA ligands, some of which were previously shown to associate with the well-studied and infection-relevant RBP Hfq. We further show that ProQ activates the expression of PhoP at the post-transcriptional level, which, in turn, leads to upregulation of the intracellular virulence program.

IMPORTANCE

Salmonella enterica is a major pathogen responsible for foodborne gastroenteritis, and a leading model organism for genetic and molecular studies of bacterial virulence mechanisms. One key trait of this pathogen is the ability to survive within infected host cells. During infection, the bacteria employ a type three secretion system that deliver effector proteins to target and manipulate host cell processes. The transcriptional regulation of this virulence program is well understood. By contrast, the factors and mechanisms operating at the post-transcriptional level to control virulence gene expression are less clear. In this study, we have charted the global RNA ligand repertoire of the RNA-binding protein ProQ during in vitro conditions mimicking the host cell environment. This identified hundreds of binding sites and revealed ProQ-dependent stabilization of intracellular-specific small RNAs. Importantly, we show that ProQ post-transcriptionally activates the expression of PhoP, a master transcriptional activator of intracellular virulence in Salmonella.

INTRODUCTION

During infection of an animal host, bacterial pathogens are challenged by ever-changing environmental conditions, host defense systems, competition with the resident microbiota, and assault by bacteriophages. Consequently, these pathogens are equipped with dedicated systems that promote survival and proliferation during infection. For instance, survival of Salmonella enterica serovar Typhimurium (henceforth: Salmonella) within host cells is highly dependent on virulence genes encoded in the pathogenicity island 2 (SPI2), and SPI2-associated virulence genes encoded elsewhere on the chromosome (1). The expression of these genes is subject to extensive regulation to ensure appropriate spatial and temporal induction. At the top of this regulatory hierarchy is the two-component system PhoPQ. Once Salmonella is located in the Salmonella-containing vacuole (SCV), the sensor kinase PhoQ responds to environmental cues such as low Mg²⁺ levels and acidic pH by phosphorylating the response regulator and master transcriptional activator PhoP. PhoP-P, in turn, activates a large number of genes required for intracellular survival and proliferation (2, 3). These include the ssrAB operon, a SPI2-encoded two-component system that promotes expression of the SPI2 type-three secretion system (T3SS) (4, 5). In addition, PhoP promotes expression of the SPI2-activating transcription factor (TF) SlyA, which, in turn, stimulates transcription of many virulence genes (6–8). Mechanistically, this often occurs by PhoP- and/or SlyA-dependent displacement of the global DNA-binding protein H-NS (9, 10).

In addition to this well-described transcriptional regulation, an intricate layer of post-transcriptional control seems to govern the intracellular stage of Salmonella infection. DualRNA-seq analysis during host cell infection (11), as well as transcriptomic studies in host cell-mimicking growth medium (12, 13), identified many intracellularly expressed small regulatory RNAs (sRNAs). For instance, the sRNA PinT controls the timing of virulence factor expression by targeting mRNAs encoding regulators and effectors important for both the invasion process and intracellular proliferation, respectively (11, 14, 15). Moreover, genetic inactivation of global RNA-binding proteins (RBPs) such as Hfq, CsrA, and ProQ, all of which interact with numerous sRNAs, impairs Salmonella virulence gene expression and infectivity (16–20).

The global RBP ProQ belongs to the ProQ/FinO protein family, members of which are found in many proteobacterial species (21, 22). Global profiling in Salmonella and Escherichia coli (E. coli) has attributed hundreds of RNAs as ProQ ligands (23–26), the majority of which are sRNAs and mRNA 3′UTRs. Salmonella lacking ProQ shows reduced SPI2 gene expression during cell culture infection and displays an invasion defect (20). However, the global ProQ ligand repertoire under SPI2-inducing conditions, as well as the basis for ProQ-dependent virulence gene expression, has remained unknown.

In this study, we provide a global analysis of ProQ binding sites in Salmonella under conditions that mimic the environment of the SCV. Under these conditions, ProQ preferentially binds to sRNAs and 3′ regions of mRNAs. The ProQ RNA ligand repertoire is significantly different between conditions that mimic the extracellular and vacuolar states, respectively, a pattern largely explained by RNA ligand availability. We identify SPI2-induced sRNAs whose expression and/or stability is promoted by ProQ. Some of these are also well-characterized Hfq ligands. Consistent with previous studies, the absence of ProQ entails significantly reduced expression of virulence genes associated with the intracellular stage of infection. As we show here, this is caused by post-transcriptional and ProQ-dependent control of PhoP translation, which, in turn, extensively impacts transcription of downstream virulence genes.

RESULTS

Global identification of ProQ binding sites under SPI2-inducing conditions

Cross-linking and immunoprecipitation followed by sequencing (CLIP-seq) recently identified hundreds of ProQ binding sites in the Salmonella transcriptome under standard laboratory conditions (LB medium, early stationary phase) (24). To assess the global ProQ ligand repertoire associated with the intracellular stage of infection, we performed CLIP-seq on Salmonella grown in “SPI2 medium”, commonly used to mimic the environment of the SCV (27). Cultures of Salmonella carrying a proQ-3xflag allele were grown in LB to early exponential phase, shifted to SPI2 medium until reaching an OD₆₀₀ value of 0.3, and thereafter subjected to CLIP-seq using the same protocol as in our previous studies (18, 24). Successful purification of crosslinked and radioactively labeled ProQ-RNA complexes was verified by autoradiographic analysis (Fig. S1). Deep sequencing and mapping of RNA fragments crosslinked to ProQ resulted in strongly enriched read counts in crosslinked over non-crosslinked samples throughout the Salmonella transcriptome (Fig. 1A). ProQ binding sites were subsequently identified by peak calling. Based on two separate experiments, which showed strong correlation (Fig. 1B), more than 9,000 significant (q ≤ 0.05) peaks were identified (Table S1; Fig. 1C). Increasing the stringency further by adding a cut-off for normalized read count indicated that peaks mapping in sRNAs were more robustly detected than those in, e.g., coding sequences (CDSs) (Fig. 1D). Using a read count cut-off of 1,000, yielding a total peak number comparable to that previously obtained in LB, the majority of peaks mapped to sRNAs, mRNA CDSs, and mRNA 3′UTRs (Fig. 1D). Metagene analysis of peaks mapped to mRNAs revealed a strong peak enrichment in the 3′ region (end of CDS/beginning of 3′UTR) (Fig. 1E), while peaks detected in sRNAs were evenly distributed along the gene body (Fig. 1F).

Fig 1.

Transcriptome-wide mapping of ProQ binding sites during intracellular-like conditions. (A) ProQ CLIP-seq read coverage over the Salmonella transcriptome in cross-linked samples and non-crosslinked controls. (B) Correlation between two independent ProQ CLIP-seq experiments. Each data point represents normalized read count for a significant (q < 0.05) ProQ peak in each of the two experiments. (C) Distribution of significant ProQ peaks over the Salmonella transcriptome. Each data point represents the number of peaks within a 10,000 window along the genome. (D) Number of significant peaks, as well as fraction of peaks per RNA class, as a function of lower normalized read count cut-offs. (E) Normalized distribution of ProQ peaks over coding sequences and associated flanking regions. (F) Normalized distribution of ProQ peaks in sRNAs. CL, crosslinking; Rep, biological replicate.

RNA expression levels impact the ProQ ligand repertoire

A comparison of the ProQ interactomes obtained in LB (24) and in SPI2 medium (this study) showed that almost half of the LB peaks were also detected in SPI2 medium (Fig. 2A). The peaks detected in both conditions were generally positioned in the same region of the respective RNA ligand; >75% of peaks were positioned less than 20 nucleotides apart (Fig. 2B), indicating that the location of ProQ binding sites in specific RNA ligands is generally consistent between conditions. In addition to the peaks detected in both conditions, a large number of peaks were uniquely detected either in LB or in SPI2 medium. This result may either reflect condition-specific expression of the RNA ligand, or condition-specific binding despite the RNA ligand being present in both growth conditions. To investigate this, the ratio of expression in SPI2 to LB [data from reference (20)] was calculated for each RNA ligand associated with a condition-specific peak. This showed that RNA ligands associated with SPI2-specific peaks are generally expressed at a higher level in SPI2 medium than in LB, and vice versa (Fig. 2C). Hence, ProQ ligand repertoires are strongly correlated with RNA ligand expression levels.

Fig 2.

Condition-specific ProQ ligand repertoires. (A) Venn diagram showing the overlap between the ProQ interactomes in SPI2-inducing conditions (q < 0.05, read count >1,000) and LB medium (24). (B) Comparison of ProQ binding site location in ligands commonly found in SPI2-inducing medium and LB. (C) Differential expression of RNA ligands uniquely found in SPI2-inducing medium or LB, respectively. Log₂-transformed expression ratio calculated based on normalized read counts from previously published data (20).

ProQ impacts the stability of sRNA ligands under SPI2-inducing conditions

sRNAs constitute a major class of ProQ ligands in SPI2 medium (Fig. 1), many of which were not identified during growth in LB (24) (Fig. 3A). The SPI2-specific sRNA ligands include both uncharacterized sRNAs such as STnc3750, and well-characterized sRNAs such as PinT (Fig. 3B and C). Using Northern blot analysis, the expression and stability of both uncharacterized sRNAs (STnc150, STnc1710, STnc3110, STnc3170, STnc3750, and STnc4210) and well-characterized sRNAs previously shown to be associated with Hfq (ArcZ, DapZ, GcvB, MicA, MicL, PinT, and RybB) were monitored. The majority of assayed sRNAs were more strongly expressed in SPI2 medium than in LB, showed reduced levels in the absence of ProQ, and could be complemented back to wild-type levels, or even higher, by ProQ complementation from a plasmid (Fig. 3D). In many cases, these effects are consistent with ProQ-dependent changes in sRNA stability (Fig. 3E). Interestingly, during growth in SPI2 medium, levels of the Hfq-associated sRNAs MicA, MicL, PinT, and RybB were strongly reduced in the absence of ProQ, while wild-type levels were observed in a ProQ complementation strain. The absence of Hfq did not decrease but rather increased the abundance of these sRNAs. By contrast, ArcZ, DapZ, and GcvB were strongly reduced in the absence of Hfq but showed less or no ProQ-dependence. These results strongly agree with a reanalysis of previously obtained RNA-seq data (20) (Fig. 3F). Taken together, ProQ promotes the expression levels of many of its sRNA ligands, including both uncharacterized and well-characterized Hfq-associated sRNAs, during growth in SPI2 medium.

Fig 3.

ProQ positively affects the expression and stability of intracellularly induced sRNAs. (A) ProQ sRNA ligands detected under SPI2-inducing conditions (q < 0.05, read count >1,000) or in LB (24). (B) ProQ CLIP-seq read coverage plots for the uncharacterized sRNA STnc3750 and the well-characterized Hfq-dependent sRNA PinT, respectively. Numbers above the plots indicate genome position. (C) Secondary structure representations of STnc3750 and PinT based on computational predictions using RNAfold (28). Nucleotides in red denote the ProQ binding site identified by CLIP-seq. (D) Steady-state levels of the indicated sRNAs as determined by Northern blot analysis in wild-type, ΔproQ, proQ complementation (ΔproQ, pProQ), and Δhfq strains during growth in LB (early stationary phase) or SPI2-inducing medium (at an OD₆₀₀ of 0.3). (E) Northern blot analysis of the indicated sRNAs in wild-type, ΔproQ, and proQ complementation (ΔproQ, pProQ) strains before and after Rifampicin-mediated transcriptional shut-off. (F) Differential expression between ΔproQ and wild-type, or Δhfq and wild-type strains, respectively, during growth in SPI2-inducing conditions (based on previously published data (20)). Plasmid pProQ expresses ProQ from an IPTG-inducible promoter. The wild-type, ΔproQ, and Δhfq strains harbored the empty vector pAR007 (pProQ backbone). FC, log₂-transformed fold-change. P, FDR-adjusted P-value. In (D) and (E), ribosomal RNA 5S served as a loading control.

ProQ affects the expression of virulence mRNAs

In addition to sRNAs, the CLIP-seq data indicated possible ProQ-dependent regulation of many virulence-associated mRNAs. These encode structural proteins of the SPI2-encoded T3SS, secreted effectors, virulence-activating TFs, or other virulence-related proteins (Table S1). ProQ’s impact on the expression of SPI2-encoded genes ssrA and ssaG and other virulence-related genes pagK, pagM, and pipB was monitored by qPCR analysis. All of the tested mRNAs exhibited reduced steady-state levels in a ΔproQ strain, and levels similar to a wild-type strain, or even higher, upon complementation by plasmid-expressed ProQ (Fig. 4A). As ProQ promotes the stability of some mRNA ligands (24), mRNA decay rates were monitored by Rifampicin run-off experiments. A positive control, cspE mRNA, was clearly destabilized in the absence of ProQ and regained stability upon ProQ complementation (Fig. 4B). By contrast, the decay rate of the tested virulence mRNAs appeared unaffected by ProQ (Fig. 4B), although the steady-state levels followed the pattern observed by qPCR analysis (Fig. 4A). This suggests a ProQ-dependent but indirect effect, presumably at the level of transcription initiation. Indeed, deletion of proQ resulted in reduced ssaG and pagC promoter activity (Fig. 4C and D), while ProQ overexpression strongly increased their expression (Fig. 4E and F). This effect was abolished in strains lacking SlyA (Fig. 4E and F), a transcriptional activator of both ssaG and pagC (8, 29). We next monitored ssaG promoter activity during Salmonella infection of human U937 cells by flow cytometry, using the P_ssaG-gfp construct. Deletion of proQ reduced the fraction of GFP-expressing cells (Fig. S2A), as well as the mean GFP signal within this fraction (Fig. S2B). Taken together, the absence of ProQ leads to reduced transcription of genes involved in the intracellular stage of infection. Considering that ProQ is known to act at the post-transcriptional level, this effect is likely indirect.

Fig 4.

ProQ indirectly promotes transcription of intracellularly induced virulence genes. (A) Steady-state levels of the indicated mRNAs in wild-type, ΔproQ, and proQ complementation strains during growth in SPI2 medium as determined by quantitative real-time PCR (qRT-PCR). (B) Northern blot analysis of the indicated mRNAs in wild-type, ΔproQ, and proQ complementation strains before and after Rifampicin-mediated transcriptional shut-off. The cspE mRNA served as a positive control for ProQ-dependent stabilization, while tmRNA served as a loading control. In A and B, proQ ++ indicates a ΔproQ strain expressing proQ from its native promoter on a multi-copy plasmid. The wild-type and ΔproQ strains harbor the respective empty vector pZE12-luc. (C–D) Measurements of GFP expression from transcriptional reporters P_ssaG-gfp (C) and P_pagC-gfp (D) in wild-type and ΔproQ strains during growth in SPI2 medium. (E–F) Measurements of GFP expression from transcriptional reporters P_ssaG-gfp (E) and P_pagC-gfp (F) in ΔproQ and ΔproQΔslyA strains with or without ProQ overexpression from an IPTG-inducible promoter on plasmid pProQ during growth in SPI2 medium. In C–F, the wild-type, ΔproQ, and ΔproQΔslyA strains harbor the empty vector pAR007 (pProQ backbone).

ProQ affects the expression of phoP and slyA

Many intracellularly expressed virulence genes in Salmonella, including ssaG and pagC, are activated by TFs PhoP and SlyA, suggesting that a direct regulatory effect of ProQ, resulting in the observed downstream effects (Fig. 4), may converge at one or both of these regulators. As PhoP and SlyA mutually regulate each other’s expression (7, 30), slyA and phoP transcription was monitored using promoter-gfp fusions. Transcription of both slyA and phoP decreased in the absence of proQ compared to the wild-type strain (Fig. 5A and C), and could be restored by ProQ overexpression (Fig. 5B and D). In accordance, steady-state levels of slyA mRNA and SlyA protein were strongly reduced in the absence of ProQ, and restored by ProQ complementation (Fig. S3A and B). However, this is unlikely through effects on mRNA stability, as the slyA mRNA decay rate, in contrast to the steady-state level, was unaffected in the proQ deletion strain (see Rifampicin experiments; Fig. S3C). Moreover, SlyA protein levels became independent of ProQ when expressed from the heterologous araBAD promoter (Fig. S3D). Hence, ProQ appears to indirectly affect expression of slyA at the transcriptional level, without effects at the post-transcriptional level.

Fig 5.

ProQ promotes transcription of phoP and slyA. Measurements of GFP expression from transcriptional reporters P_slyA-gfp (A, B) and P_phoP-gfp (C, D) in wild-type and ΔproQ strains (A, C), or ΔproQ with or without ProQ overexpression from an IPTG-inducible promoter on plasmid pProQ (B, D), during growth in SPI2 medium. The wild-type and ΔproQ strains harbor the empty vector pAR007 (pProQ backbone).

ProQ regulates PhoP synthesis at the post-transcriptional level

Since ProQ-dependent regulation of SlyA expression appeared to be indirect, we next addressed the possibility of ProQ controlling PhoP expression post-transcriptionally. Western blot analysis showed significantly reduced levels of chromosomally expressed PhoP-3xFLAG upon deletion of proQ, and increased levels beyond that of the wild-type strain upon ProQ overexpression (Fig. 6A). Note that attachment of the 3xFLAG tag did not impair PhoP protein activity (Fig. S4). To specifically monitor effects at the post-transcriptional level, a phoP-gfp translational fusion was constructed. This encompassed the phoP 5′UTR, and the first 10 codons of the phoP ORF inserted between a constitutive promoter and gfp on a plasmid. Strikingly, deletion of proQ resulted in significantly reduced PhoP-GFP translation, while ProQ overexpression increased PhoP-GFP beyond wild-type levels (Fig. 6B). By contrast, a translational slyA-gfp fusion was essentially unaffected by altered ProQ levels (Fig. S3E). We next assayed PhoP-GFP translation in strains expressing ProQ mutants identified in recent genetic screens (31–33). Mutation R80H, which completely impairs the RNA-binding activity of the ProQ N-terminal domain (31), abolished ProQ’s ability to activate PhoP translation (Fig. 6C). The same effect was observed with mutation L34Q, which likely impairs the folding of the N-terminal domain (31, 33). Mutations T200P and G185V, both of which impairs the regulatory activity of the ProQ C-terminal domain (32, 33), also completely abolished ProQ-dependent activation of PhoP translation (Fig. 6C). Importantly, the experiments in (Fig. 6C) were carried out with ProQ-dTomato fusion proteins, which do not become unstable upon mutations in the ProQ polypeptide (32, 33). Finally, ProQ-mediated activation of the PhoP-dependent pagC promoter was completely abolished in a phoP deletion strain (Fig. 6D).

Fig 6.

ProQ positively affects translation of PhoP. (A) Western blot analysis of chromosomally encoded PhoP-3xFLAG in wild-type, ΔproQ, and ΔproQ with ProQ overexpression from an IPTG-inducible promoter on a plasmid. GroEL served as a loading control. Numbers below the image represent average values and standard deviation based on three biological replicates. Statistical significance was determined using a two-tailed t test (**P < 0.01; *P < 0.05). (B) GFP expression from translational phoP-gfp fusions in wild-type, ΔproQ, and ΔproQ with or without ProQ overexpression from an IPTG-inducible promoter on plasmid pProQ during growth in SPI2 medium. Statistical significance was determined using a two-tailed t test (***P < 0.001; **P < 0.01; *P < 0.05). (C) GFP expression from a translational phoP-gfp fusion in a ΔproQ strain with ProQ mutants fused to dTomato and expressed from an IPTG-inducible promoter on a plasmid. Control strains include empty vector control (pEmpty), empty vector with dTomato (pdTomato), and wild-type ProQ fused to dTomato (pdTomato-ProQ). Strains were grown in SPI2 medium. (D) GFP expression from transcriptional P_pagC-gfp reporter in the indicated strains during growth in SPI2 medium. In all panels, the wild-type, ΔproQ, ΔphoP, and ΔphoPΔproQ strains harbor the empty vector pAR007 (pProQ backbone).

Together these results suggest ProQ to act through a post-transcriptional mechanism to control PhoP translation, which indirectly leads to extensive effects on transcription of PhoP-dependent promoters in the Salmonella virulence program.

DISCUSSION

A ProQ interactome linked to the intracellular lifestyle of Salmonella

In this paper, we have used CLIP-seq to chart the ProQ ligand repertoire in Salmonella during conditions that mimic the intracellular environment of eukaryotic host cells. Thousands of ProQ binding sites were detected in different classes of cellular RNAs, primarily in mRNA and sRNA transcripts. Many of these ProQ ligands were not detected in previous studies, emphasizing the importance of conducting global RNA-protein interaction studies across various conditions. By monitoring the steady-state levels and decay rates of sRNA ligands, ProQ binding could be linked to effects on expression for both uncharacterized and previously characterized sRNAs. Among the latter were several well-characterized Hfq-dependent sRNAs. Our analysis of ProQ’s role in the expression of Salmonella virulence genes indicates that ProQ positively affects translation of PhoP, which, in turn, leads to enhanced expression of a large number of PhoP-dependent virulence genes.

ProQ-dependent sRNAs in SPI2-inducing conditions

Performing CLIP-seq under SPI2-inducing conditions identified many sRNAs not previously identified as ProQ ligands, and Northern blot analysis showed that the expression and/or stability of many of these depend on ProQ (Fig. 3). It is important to note that many of these sRNAs are transcriptionally activated by PhoP (13). Thus, ProQ may promote their expression both indirectly through ProQ-dependent activation of PhoP translation and directly by binding and protecting them from degradation.

In line with their strongly elevated expression in SPI2 medium (Fig. 3) (12, 20), these sRNAs are likely to have virulence-related functions. For instance, STnc150 is encoded between the divergently transcribed coding genes icdA and STM1239 and was suggested to promote adhesion and intracellular survival through a so-far unknown mechanism (34). STnc3750 is transcribed from a promoter overlapping the 3′ part of the pgtE gene, encoding an outer membrane protease that confers resistance to antimicrobial peptides (35), and is associated with the increased human serum resistance of strain D234580 of Salmonella ST313, which causes invasive nontyphoidal Salmonellosis (36). Assigning functions to such uncharacterized sRNAs will be important for fully understanding the post-transcriptional layer of virulence gene regulation in Salmonella.

A number of previously verified Hfq-dependent sRNAs crosslinked to ProQ under SPI2-inducing conditions and expression of several of these were dependent on ProQ (Fig. 3). Somewhat unexpected, sRNAs such as MicL, MicA, RybB, and PinT instead showed increased steady-state levels in a strain deleted for hfq. However, this result aligns well with a previous RNAseq-based study using the same Salmonella strain background and a similar growth medium (20). It thus appears that in some conditions, ProQ becomes the superior RNA chaperone for assuring the expression of a subset of well-characterized Hfq-dependent sRNAs. This highlights the importance of studying RBP and sRNA biology in more physiologically relevant conditions, and suggests that the crosstalk between the ProQ and Hfq interactomes is influenced by the bacterium’s surrounding environment.

Mechanistic aspects of ProQ-dependent regulation of phoP translation

Since the discovery of ProQ as a global RNA-binder in Salmonella and E. coli, a substantial amount of data regarding ProQ in vivo ligands, and ProQ-dependent gene expression, has become available (23–25). A deeper understanding about ProQ’s RNA-binding activity has emerged from biochemical, structural, and mutational studies (31–33, 37–39). By contrast, mechanistic understanding linking ProQ binding events to effects on particular RNA ligands is limited to the study of a few examples. ProQ-dependent mRNA stabilization has been attributed to 3′UTR binding and protection from 3′ to 5′-dependent exonuclease activity (24), while ProQ may protect sRNA-sRNA duplexes from attack by RNase III (25). In line with the ProQ-dependent stabilization of sRNAs observed here (Fig. 3), overexpression of ProQ in E. coli resulted in increased RNA stabilization on a global level (25), indicating that protection from ribonuclease attack may be a general function for ProQ. This could be further clarified by systems-level measurements of RNA decay rates with respect to ProQ expression. ProQ also affects sRNA-dependent regulation of mRNA translation: ProQ promotes RaiZ-dependent inhibition of hupA mRNA translation (40) and participates in repression of mgtB translation by the STnc540 (20). However, the mechanistic details of ProQ’s role in sRNA-dependent regulation are so far not entirely clear.

We show here that ProQ activates expression of PhoP at the post-transcriptional level (Fig. 6). In contrast, ProQ-dependent activation of SlyA, PagC, and SsaG expression appears to be strictly transcriptional (Fig. 4 through 6; Fig. S3) and likely mediated directly or indirectly through PhoP (7, 9, 41). How does ProQ enhance PhoP synthesis mechanistically? The lack of a CLIP-seq peak in the phoP 5′UTR (Table S1) argues against a mechanism solely depending on ProQ binding. Tentatively, the phoP 5′UTR may be targeted by a yet to be identified sRNA that similar to RaiZ and STnc540 requires ProQ to maintain cellular stability and/or execute target mRNA binding and regulation. In E. coli, the two sRNAs MicA and GcvB target the phoP mRNA to inhibit translation (42, 43). However, the fact that ProQ enhances the levels of MicA, and has no effect on GcvB levels (Fig. 3D and E), speaks against these sRNAs being responsible for the ProQ-dependent effect on PhoP translation. Global identification of RNA-RNA interactions associated with ProQ using the RIL-seq approach may allow identification of so-far unknown sRNAs targeting the phoP mRNA (25).

ProQ’s positive effect on the expression of SPI2 genes and other members of the PhoP regulon fully aligns with a previous study (20). However, in contrast to what we show here, that study did not observe ProQ-dependent effects on PhoP levels in Western blot assays. We do not have a clear explanation for this discrepancy but conclude that our observations of ProQ-dependent effects not only on PhoP protein levels (Western blot analysis, Fig. 6) but also on translation (phoP-gfp translational fusion, Fig. 6), and transcription (P_phoP-gfp fusion, Fig. 5) strongly argues for ProQ being a bona fide regulator of PhoP expression.

Outlook

The present study strengthens the notion of ProQ as a major factor in control of the intracellular virulence program of Salmonella. In addition, ProQ is known to promote other infection-relevant processes, including flagellar motility and biofilm formation (20, 33, 44). The approach used here lend itself for studying additional ProQ-dependent phenotypes for which the underlying mechanism of regulation is currently unknown.

MATERIALS AND METHODS

Bacterial growth conditions

Bacterial cultures were routinely grown aerobically in LB medium at 37°C with shaking at 220 rpm. To mimic the environment of the SCV, bacteria were grown in a minimal SPI2-inducing medium, as previously described (27). The media were supplemented with antibiotics (ampicillin 100 µg/mL, chloramphenicol 30 µg/mL, tetracycline 12.5 µg/mL, and/or kanamycin 50 µg/mL) when appropriate. IPTG (0.5 mM) or arabinose (0.0016%) was added to the media to induce expression of plasmid-borne ProQ or 3xFLAG-SlyA, respectively.

Strain construction

Bacterial strains and oligonucleotides used in the study are listed in Table S2 and Table S3, respectively. Deletion of proQ and slyA was achieved by P22 transduction with JVS-11364 (23), and C1892 from the McClelland collection (45), as donor strains, respectively, and wild-type SL1344 (JVS-1574) as the recipient strain. Chromosomal antibiotic cassettes were removed using the pCP20 plasmid (46). Lambda red recombination including a pSim5-tet plasmid (47) was used to construct the FLAG-tagged strains EHS-2294 (3xflag-slyA) and EHS-3288 (phoP-3xflag). Briefly, the kan-sacB cassette was amplified (slyA: primers EHO-1363/-1364; phoP: primers EHO-1869/-1870), and inserted after the start codon of slyA or before the stop codon of phoP, respectively. Next, a fragment including the 3xFLAG sequence, either amplified from plasmid pSB002 (EHO-1502/-1503) or generated by Klenow fill-in of annealed partially overlapping primers (EHO-1871/-1872), was used to replace the kan-sacB cassette. Replacement of kan-sacB cassette was verified by growth on sucrose plates, PCR, and DNA sequencing.

Plasmid construction

All plasmids used in this study are listed in Table S4. Transcriptional fusions were constructed by inserting PCR-amplified promoter sequences between BamHI and XhoI in plasmid pUA66. The following primers were used to amplify the insert for the respective plasmids, EHO-1314/-1253 (pYMB005), EHO-1315/-1508 (pSB008), EHO-1603/-1604 (pSB011), and EHO-1623/-1624 (pSB012).

To create plasmid pEH809, the MB1 origin of plasmid pBR322 was amplified with primers EHO-1673/-1674, cut with NsiI and XhoI and ligated to a PCR fragment amplified from pBAD33 using primers EHO-1675/-1676. To create plasmid pEH811, a 3xFLAG-slyA fragment (including 5′ and 3′UTRs) amplified from plasmid pSB001 with primers EHO-1689/-1690 was inserted between the NheI and HindIII sites in pEH809. To create plasmid pEH791, pXG-1 was amplified (EHO-450/-1343) and cut with NsiI and XbaI followed by ligation with a PCR fragment encompassing the slyA gene (EHO-1340/-1342). To create pSB001, the pEH791 plasmid was amplified with EHO-1458/-1459 which inserts a 3xFLAG at the N-terminal of the slyA gene.

To create translational fusions (pEH839 and pEH843), SL1344 genomic DNA was amplified with EHO-1786/-1787 (phoP 5'UTR: from TSS 34 bp upstream start codon to the 10th codon) and EHO-1339/-1790 (slyA 5'UTR: from TSS 198 bp upstream start codon to the 10th codon) and cloned in frame with GFP on pXG10-SF using NsiI and NheI.

CLIP-seq

Salmonella SL1344 carrying a chromosomal proQ-3xflag allele was grown in 20 mL LB medium to an OD₆₀₀ of 2.0, after which the cells were washed twice with PBS, once with SPI2 medium, and thereafter diluted 1:50 into fresh SPI2 medium. When the culture reached an OD₆₀₀ of 0.3, half of the culture was irradiated with UV light (254 nm, 800 mJ/cm²), while the other half was left untreated. Immunoprecipitation, Benzonase treatment, dephosphorylation, radioactive labeling, SDS-PAGE, and RNA purification were carried out as described previously (24). DNA libraries were prepared using the NEBNext Small RNA Library kit (NEB) according to the manufacturer’s instruction and sequenced on an Illumina NextSeq500 instrument at vertis Biotechnologie AG (Freising, Germany).

CLIP-seq data analysis

Adapter trimming and merging of read pairs was performed with SeqPrep (https://github.com/jstjohn/SeqPrep). Pairs were merged if the resulting read was at least 12 nucleotides long with at least 12 bases overlapping. Read mapping was performed against the Salmonella SL1344 chromosome (NC_016810.1) with bowtie 1.2.2 (48) allowing for one mismatch and only reporting the best match for each read. Read coverage was analyzed by converting mapped reads to bigWig format with BEDTools genomecov (49) and bedGraphToBigWig (50). Peak calling was performed with PEAKachu (51). The tool was run in adaptive mode with mad-multiplier (-m) 1.0 and fold change (-f) 1.0 using paired-replicates (-r) of BAM files for the respective pairs of crosslinked and control libraries as input. The maximum fragment size (-M) was set to 25 and annotations in GFF format were used to map overlapping features to called peaks. For normalization, “manual” mode was selected together with size factors calculated as in (24).

Rifampicin run-off experiments

Bacterial cultures were grown in LB overnight and inoculated 1:100 in fresh LB and grown to an OD₆₀₀ of 2.0. Cultures were then diluted 1:100 in SPI2 medium supplemented with 0.5 mM IPTG and appropriate antibiotics. At an OD₆₀₀ of 0.3–0.4, a sample from the culture (0 min sample) was mixed with 0.2 volumes of stop solution (95% ethanol, 5% phenol) and immediately frozen in liquid nitrogen. Rifampicin was added to the remaining culture at a final concentration of 500 µg/mL, and samples were taken at indicated timepoints, mixed with stop solution, and snap-frozen in liquid nitrogen. Samples were stored at −80°C until RNA extraction.

RNA extraction

Samples were thawed on ice and centrifuged for 10 min at 13,000 rpm (2 mL tubes) or at 5,000 rpm (15 mL tubes) at 4°C. The supernatant was discarded, and the pellets were resuspended in TE buffer pH 8.0 with 0.5 mg/mL lysozyme. SDS (final concentration 1%) was added, and the tubes were mixed by inversion and incubated for 2 min at 64°C. After the incubation, 0.1 vol of NaOAc 3M (pH 5.2) and 1 vol of acid phenol was added, and the samples were mixed thoroughly. The samples were incubated at 64°C for 6 min and mixed by inversion 6–10 times during the incubation time. The samples were then placed on ice for 5 min and centrifuged for 10 min at 13,000 rpm at room temperature. The aqueous layer was transferred to a Phase-Lock tube, and 1 vol of chloroform was added to each tube. The samples were mixed vigorously by shaking for 30 s and centrifuged 5 min at 13,000 rpm at room temperature. The aqueous layer was transferred to a clean tube and mixed with 3 volumes of ice-cold 30:1 mix (EtOH: 3 M NaOAc pH 5.2) and incubated 12–16 h at −20°C. The samples were then centrifuged for 30 min at 13,000 rpm and 4°C, and the pellets were washed with 80% ice-cold EtOH. After a second centrifugation for 10 min (13,000 rpm, 4°C), the pellets were air dried, dissolved in sterile water, and incubated for 4 min at 65°C with shaking at 900 rpm. RNA quality was assured by agarose gel electrophoresis and SYBR safe (Invitrogen) staining.

Quantitative real-time PCR

Total RNA samples were DNase-treated with TurboDNase followed by phenol:chlorophorm extraction. DNase-treated total RNA (2 µg) was used for cDNA synthesis using the Maxima H Minus First Strand cDNA Synthesis Kit (#K1652, Thermo Scientific). Synthesized cDNA equivalent to 40 ng total RNA was used in each quantitative real-time PCR (qRT-PCR) with Power SYBR Green PCR Master Mix and analyzed with a Step one Plus real time PCR system (Applied Biosystems). The house-keeping gene recA was used for normalization. Primers used for qRT-PCR are listed in Table S2.

Northern blot

Total RNA samples (5–10 µg) were diluted 1:1 in RNA loading buffer (GLII). The samples were incubated for 3 min at 95°C before loaded on an 6% (vol/vol) polyacrylamide (PAA)/ 8 M urea gel together with a radiolabeled pUC19 MSP1 marker (ThermoFischer). After separation, the RNA was transferred to a nitrocellulose XL membrane (GE Healthcare) or a nitrocellulose Hybond-N+ membrane (Amersham, Cytiva) by wet electroblotting at 4°C overnight. RNA was crosslinked to the membrane by UV light exposure at 1200 mJ. The cross-linked membrane was prehybridized in church buffer (0.5 M sodium phosphate buffer pH 7.2, 1 M EDTA, 7% SDS) for 45 min at 42°C before a 5′-P³²-labeled DNA oligonucleotide was added to the hybridization buffer and the incubation continued for 12–18 h. The membrane was then washed twice in 2 × SSC/0.1% SDS and one time in 0.5 × SSC/0.1% SDS. The membrane was dried, exposed to a phosphor screen, and scanned on a Typhoon phosphorimager (Cytiva).

Western blot

Protein samples were harvested from Salmonella strains grown in SPI2 media to an OD₆₀₀ of 0.3 or 0.4. The samples were centrifuged for 10 min at 13,000 rpm and 4°C, and the pellets were resuspended in loading buffer (2× Laemmli Sample Buffer, Bio-Rad). Before loading, samples were heated for 4 min at 95°C and cooled on ice. The samples and a size marker (PageRuler Prestained Protein Ladder, Thermo Fisher Scientific) were separated on precast Any kD Mini-PROTEAN TGX Stain-Free Protein Gels (Bio-Rad) and transferred to a PVDF membrane using the Trans-Blot Turbo Mini PVDF Transfer Packs (Bio-Rad) in the Trans-Blot Turbo Transfer Starter System (Bio-Rad) at 2.5 A and 25 V for 3 min. The membrane was blocked with 3% BSA (Sigma) in TBS-T buffer (50 mM Tris, 150 mM NaCl, 0.1% Tween 20, pH 7.4–7.6) overnight at 4°C. The membrane was then rinsed with TBS-T for a few seconds and then incubated with TBS-T for 10 min and probed with an anti-GroEL-Peroxidase Conjugate antibody produced in rabbit 1:50,000 (Sigma), a monoclonal anti-FLAG M2-Peroxidase (HRP) antibody produced in mouse 1:10 000 (Sigma), or an anti-ProQ antibody at a 1:10,000 in TBS-T 3% BSA for 1 h at room temperature. The anti-ProQ antibody was followed by incubation with an HRP-conjugated anti-rabbit antibody (Sigma) at 1:100,000 in TBS-T 3% BSA for 1 h. After incubation with antibodies, the membranes were washed in TBS-T for 10 min three times. The membranes were then developed with ECL detection reagent (GE Healthcare Life Sciences) and scanned in a ChemiDoc MP System (Bio-Rad).

Growth in 96-well plate reader

Bacterial cultures were grown in LB overnight at 37°C, diluted 1:100 in SPI2 medium, loaded in a 96-well plate, and grown for 16–20 h at 37°C, shaking for 30 s every 10 min in a plate reader (Tecan infinite 200Pro). GFP and OD₆₀₀ were measured every 10 min.

Monocyte cell line culture

The pro-monocytic cell line U937 (ATCC) was cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco) supplemented with 10% heat inactivated Fetal Bovine Serum (HI-FBS) (Gibco) and Penicillin-Streptomycin (PenStrep) (Gibco) in a T25 flasks (Sarstedt) at 37°C with 5% CO₂. Cells were kept at a density of 100,000–1,500,000 cells/mL and passaged every 2–3 days.

Infection assays

Salmonella strains, carrying the PssaG-GFP reporter plasmid as indicated, were grown on LA plates with appropriate antibiotics and subsequently inoculated in four biological replicates in 3 mL LB supplemented with 0.3 M NaCl and kanamycin for 12 h at 37°C on a rotating wheel. At the day of the infection, U937 cells were seeded in a round-bottom 96-well plate with a concentration of 100,000 cells/well. The overnight Salmonella cultures were sub-cultured in 3 mL LB with 0.3 M NaCl and incubated at 37°C for 4 h. After sub-culturing, the bacteria were spun down at 12,700 rpm and re-suspended in RPMI medium. The Salmonella strains were added to the U937 cells at a multiplicity of infection (MOI) of 16 and incubated at 37°C, 5% CO₂ for 30 min to allow for invasion. After incubation, the plate was centrifuged at 100 g for 1 min, and the cells were resuspended in RPMI supplemented with 100 µg/mL Gentamicin. The cells were incubated at 37°C, 5% CO₂ for 3.5 h to allow Salmonella to undergo intracellular maturation and ssaG-driven GFP expression. After the incubation, the samples were prepared for flow cytometry. The cells were centrifuged at 300 g for 1 min and washed two times with washing solution (PBS with 1% BSA). The samples were then resuspended in 2% PFA and fixed for 20 min in darkness. After fixation, the samples were washed one more time as described before and dissolved in washing solution. The samples were analyzed with flow cytometry using a MACSQuant VYB (Miltenyi Biotec), measuring 30,000 events, and the data were acquired by the MACSQuantify Software (Miltenyi Biotec). The software tool FlowJo (FlowJo, LLC 2006–2021) was used to analyze the data.

DATA AVAILABILITY

The ProQ CLIP-seq data is available at the NCBI Gene Expression Omnibus with the accession number GSE250014.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/msphere.00018-24.

Supplemental Material. msphere.00018-24-s0001.pdf.

Figure S1-S4; Tables S2-S4.

Table S1. msphere.00018-24-s0002.xlsx.

CLIP-seq data.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.