The LhrC sRNAs control expression of T cell-stimulating antigen TcsA in Listeria monocytogenes by decreasing tcsA mRNA stability

ABSTRACT

The bacterial pathogen Listeria monocytogenes encodes seven homologous small regulatory RNAs, named the LhrC family of sRNAs. The LhrCs are highly induced under infection-relevant conditions and are known to inhibit the expression of multiple target mRNAs encoding virulence-associated surface proteins. In all cases studied so far, the LhrCs use their CU-rich regions for base pairing to complementary AG-rich sequences of the ribosomal binding site (RBS) of specific target mRNAs. Consequently, LhrC-mRNA interaction results in inhibition of translation followed by mRNA degradation, corresponding to the canonical model for sRNA-mediated gene regulation in bacteria. Here, we demonstrate that the LhrC sRNAs employ a different regulatory mechanism when acting to down-regulate the expression of tcsA, encoding a T cell-stimulating antigen. In this case, LhrC base pairs to an AG-rich site located well upstream of the RBS in tcsA mRNA. Using an in vitro translation assay, we found that LhrC could not prevent the ribosome from translating the tcsA messenger. Rather, the LhrC sRNAs act to decrease the half-life of tcsA mRNA in vivo. Importantly, LhrC-mediated destabilization of tcsA mRNA relies on an intact LhrC binding site near the 5´-end of the tcsA mRNA and occurs independently of translation. Based on these findings, we propose an alternative mechanism for LhrC-mediated control in L. monocytogenes that relies solely on sRNA-induced degradation of a target mRNA.

Introduction

Small regulatory RNAs (sRNAs) in bacteria are known to control important biological processes, such as stress tolerance, metabolism and virulence (reviewed in [1]). A major group of sRNAs serve as trans-acting regulators of gene expression at the post-transcriptional level; they regulate translation and/or mRNA stability, either positively or negatively, by imperfect base pairing to partner mRNAs [1–3]. The trans-acting sRNAs often rely on an RNA chaperone, such as Hfq, to promote formation of sRNA-mRNA complexes and to protect the sRNA from degradation by RNases [2,3]. In Gram-negative bacteria, Hfq is known to play a critical role in sRNA-mediated control, whereas in Gram-positive bacteria, Hfq-dependent sRNAs seem to be extremely rare [4]. Indeed, only a single well-characterized example is known in Gram-positive bacteria: the Hfq-dependent sRNA LhrA of the human pathogen Listeria monocytogenes (reviewed in [5]). L. monocytogenes is a facultative intracellular pathogen and the causative agent of the life-threatening foodborne disease listeriosis [6,7]. Early studies on sRNA-mediated control in L. monocytogenes resulted in the identification of three Hfq-binding sRNAs: LhrA, LhrB and LhrC [8]. Later, LhrA was shown to rely on Hfq for stability as well as regulatory activities when L. monocytogenes enters the stationary growth phase [9,10]. More specifically, LhrA is quickly degraded in an Hfq-deficient strain and inhibits translation of specific mRNAs by an Hfq-dependent base pairing mechanism [9]. In line with the classical mode of action defined from studies of Hfq-dependent sRNAs in Gram-negative bacteria, LhrA inhibits translation by imperfect base pairing to the ribosomal binding site (RBS) of target mRNAs [9,10].

Additional studies in L. monocytogenes showed that the Hfq-binding sRNA LhrC also acts by direct base pairing to the RBS of partner mRNAs, but no role for Hfq has been found so far in controlling the stability or regulatory activities of LhrC [8,11–13]. LhrC is a multicopy sRNA and the LhrC family comprises seven siblings: LhrC1-5, Rli22 and Rli33-1 (reviewed in [5]). Members of this family are highly induced under infection-relevant conditions, such as intracellular replication in macrophage cells (LhrC1-5, Rli33-1), whole human blood (all seven members), the intestinal lumen of mice (Rli22), and upon exposure to cell wall-acting antibiotics such as cefuroxime (LhrC1-5, Rli22) [11,14,15]. For LhrC1-5 and Rli22, induction relies on the two-component system LisRK, whereas the general stress sigma factor SigmaB controls expression of Rli33-1 [11,13]. The LhrC family of sRNAs has been shown to target mRNAs encoding surface-exposed virulence-associated proteins, including the virulence adhesin LapB, the oligo-peptide binding protein OppA, and the CD4⁺ T cell-stimulating antigen TcsA [11,12]. In all three cases, the sRNAs exert a negative effect on translation, and for lapB and oppA, the LhrCs are known to act by direct base pairing to the RBS, leading to inhibition of translation followed by mRNA degradation [11,12]. For tcsA, however, LhrC-meditated control appears not to follow the canonical pathway. So far, it has been shown that tcsA expression is reduced under LhrC-inducing conditions at both protein and RNA levels [12]. Additionally, LhrC regulation was demonstrated to occur at the post-transcriptional level but, surprisingly, the predicted binding site for LhrC sRNAs was found to be located far upstream from the RBS in the 5´-untranslated region (5´-UTR) of tcsA [12].

In this study, we aimed to uncover the regulatory mechanism underlying the inhibitory effect of LhrC on tcsA expression. Using LhrC4 as a representative member of the LhrC family, we first mapped the interactions between LhrC4 and tcsA mRNA, showing that LhrC4 base pairs in vitro with two upstream sites in the 5´-UTR of this mRNA. Then, we showed that one of these upstream LhrC-binding sites is indeed crucial for LhrC-dependent regulation of tcsA at the post-transcriptional level in vivo. Further experiments demonstrated that LhrC4 does not prevent in vitro translation of tcsA mRNA; conversely, we found that the LhrC sRNAs act to decrease the half-life of tcsA mRNA in vivo. Importantly, LhrC-mediated control of tcsA occurs independently of translation and relies strictly on an intact LhrC-binding site in the 5´-UTR of tcsA mRNA. Collectively, these findings demonstrate that, for inhibition of tcsA, the LhrCs do not follow the canonical mode of action for sRNA-mediated repression (i.e. sRNA base pairing to the RBS of its target mRNAs). Instead, the LhrCs act to destabilize the tcsA mRNA by base pairing to a specific AG-rich site located far upstream of the RBS.

Results

LhrC4 binds to a region located far upstream of the RBS in tcsA mRNA

Members of the LhrC family are known to down-regulate the expression of tcsA at the post-transcriptional level; however, the mechanism underlying this effect remains to be clarified [12,13]. The LhrC family member LhrC4 was shown previously to bind a tcsA RNA fragment in vitro, covering the entire 133 nucleotide (nt) 5ʹ-UTR and part of the coding region of tcsA, suggesting that the LhrC sRNAs down-regulate tcsA expression by direct base pairing [12]. Interestingly, in silico predictions identified a putative LhrC binding site located approximately 90 nt upstream of the start codon in tcsA mRNA, indicating that post-transcriptional regulation of tcsA follows a non-canonical pathway [12]. To further investigate the in silico predicted sRNA-mRNA binding interaction, we performed additional in vitro binding experiments with 5ʹ-end labeled LhrC4 and either of two unlabeled tcsA fragments: tcsA-up, containing the predicted binding site for LhrC4, denoted the P site (predicted) but excluding the RBS; and tcsA-down, which covers the RBS but excludes the P site (Figure 1(a); Supplementary Figure S1(a)). This assay clearly demonstrated the ability of LhrC4 to bind tcsA-up, whereas there was no binding to tcsA-down (Figure 1(b)). To force binding between LhrC4 and tcsA-down, the two RNAs were mixed, heated and cooled together, yet no interaction could be observed (Supplementary Figure S1(b)). To further address the importance of the P site, a mutant tcsA-up variant was designed (tcsA-mutP) (Figure 1(a)). Clearly, mutations within position −90, −88 and −87 of the P site resulted in decreased binding to LhrC4; however, the residual interaction between tcsA-mutP and LhrC4 suggests that the 5´-UTR of tcsA might contain additional binding sites for LhrC4 (Figure 1(b)). These binding studies suggest that the LhrC sRNAs do not target the RBS region of tcsA mRNA; they rather bind to the P site and, possibly, one or more binding sites in the 5´-UTR far upstream from the RBS.

Figure 1.

The 5´-end of tcsA mRNA contains a putative binding site for LhrC. (a) Schematic overview of the tcsA locus in L. monocytogenes. The regions employed for in vitro binding experiments are illustrated below the tcsA locus. The tcsA-up region corresponds to the 5´-end of the tcsA mRNA and contains the predicted LhrC binding site, named site P. The tcsA-down region comprises the SD and start codon of tcsA. The mutant variant tcsA-mutP contains the indicated substitutions within site P. (b) EMSA of sRNA-mRNA interaction. Radiolabeled LhrC4 was incubated with increasing concentrations of non-labeled RNA fragments corresponding to tcsA-up, tcsA-mutP or tcsA-down. The fraction of unbound LhrC4 is shown below each lane. The experiment was repeated three times with similar results.

In vitro binding studies reveal multiple sites of interaction between LhrC4 and the 5´-end of tcsA mRNA

The LhrCs usually interact with target mRNAs through one or more UCCC-motifs located within single-stranded CU-rich regions [5]. The secondary structure of the LhrC family member LhrC4 is shown in Supplementary Figure S2; it comprises a 5ʹ terminal single-stranded stretch (SS1) and two stem-loop structures (stem-loop A and the terminator stem-loop, respectively) separated by another single-stranded stretch (SS2). The CU-rich regions of SS2 and/or the terminator loop were previously predicted to interact with the P site in tcsA [12] (see Figure 2(a, b)); in LhrC4, these putative binding sites were named site S (single stranded stretch) and site T (terminator stem-loop), respectively (Supplementary Figure S2). Since disruption of site P in tcsA did not eliminate the interaction with LhrC4 (Figure 1(b)), we looked for additional sites of interaction between tcsA and LhrC4. Curiously, manual inspection of the RNA sequences revealed another possible interaction site, named the M site (manual), involving nt −117 to −109 of tcsA and nt 18–26 of LhrC4 (i.e part of SS1 and stem A), which may extend further to include base pairing between nt −129 to −122 of tcsA and nt 31–38 of LhrC4 (Figure 2(c); Supplementary Figures S1(a) and S2). This putative binding site is conserved in the other four highly homologous LhrC species (Supplementary Figure S3).

Figure 2.

Mutational analysis of LhrC4-tcsA interactions using EMSA. Predicted interaction between tcsA site P and LhrC site S (a) or LhrC site T (b). Predicted interaction between tcsA site M and LhrC4 site M (c). Mutations were introduced in tcsA site P (blue) or site M (red), and complementary mutations were introduced in LhrC4 site S (blue), T (blue) or M (red). (d) Radiolabeled LhrC4-wt, or the indicated mutant variants of LhrC4, were incubated with increasing concentrations of non-labeled wild-type tcsA RNA (upper panel), or the mutant variant tcsA-mutMP, carrying mutations in sites M and P (lower panel). The fraction of unbound LhrC4 is shown below each lane. The experiment was repeated three times with similar results.

To assess the predicted sRNA-mRNA interactions we performed in vitro binding experiments using wt and complementary mutant variants of LhrC4 and tcsA-up (Figure 2(a-c)). Using gel shift assays, we tested the binding of 5ʹ-end labeled LhrC4-wt and seven LhrC mutant variants to unlabeled tcsA-wt or tcsA-mutMP (Figure 2(d)). Importantly, the M and P mutations in tcsA-mutMP are complementary to the M, S and T mutations in LhrC4 mutants, as illustrated in Figure 2(a-c). First, the binding of labeled LhrC4 variants to unlabeled tcsA-wt was tested (Figure 2(d), top panel). We noticed that mutations within site S in LhrC4 had no effect on sRNA-mRNA complex formation, whereas LhrC4 carrying mutations within site M or T bound less efficiently to tcsA-wt. When combining two of the three mutant sites (i.e. LhrC4-mutST, LhrC4-mutMS and LhrC4-mutMT), binding to tcsA-wt was clearly diminished, whereas the most prominent effect was seen for LhrC4 carrying mutations within all three sites (LhrC-mutMST). These results indicate that all three sites in LhrC4 contribute to sRNA-mRNA complex formation, but sites M and T play the most important roles. Next, the binding of labeled LhrC4 variants to unlabeled tcsA-mutMP was assessed. As expected, LhrC4-wt was unable to bind tcsA-mutMP (Figure 2(d), lower panel); however, when introducing complementary mutations within site M and T of LhrC4, the binding to tcsA-mutMP was partially restored. Notably, the affinity of LhrC4-mutMT and LhrC4-mutMST to tcsA-mutMP was comparable (i.e. half of the LhrC4 was complexed with tcsA), thus confirming that site S does not contribute substantially to the interaction between LhrC4 and tcsA. Overall, these findings support that site M and site T in LhrC4 interact with tcsA site M and site P, respectively.

To further explore the binding interactions between tcsA and LhrC4, structure probing experiments were performed using RNases A (cleaves after unpaired C or U residues), T1 (cleaves after unpaired G residues) and III (cleaves in long double stranded regions), as well as the chemical probe lead(II) (cleaves single-stranded regions). The results from the structure probing experiments are shown in Figure 3 and schematic overviews are presented in Supplementary Figure S4. The sRNA-mRNA interactions were probed using 5´-end labeled tcsA-wt (Figure 3(a)) or LhrC4-wt (Figure 3(b)), in the absence (-) or presence of unlabeled partner RNA; both wild-type and mutant variants of the partner RNA were tested. Regarding the P site in 5´-end labeled tcsA (Figure 3(a)), we noticed that this region became less single-stranded in the presence of unlabeled LhrC4-wt or LhrC4-mutM, but remained single-stranded upon addition of LhrC-mutST. These results support the predicted interaction at site P in tcsA, involving LhrC4 site S and/or site T. When looking at the structure probing analysis for 5ʹ-end labeled LhrC4 (Figure 3(b)), the results for site S were rather inconclusive, whereas for site T, a decreased RNase A reactivity of U88 in the terminator loop was seen in the presence of tcsA-wt and tcsA-mutM, whereas no protection was observed in the presence of tcsA-mutP (Figure 3(b)). Thus, these data support that site T in the terminator loop of LhrC4 interacts with site P in tcsA. Regarding the manually identified site M in tcsA, Figure 3(a) shows that the presence of unlabeled LhrC4-wt or LhrC4-mutST gave rise to increased RNase III reactivity, which indicates extensive base pairing in the M region of tscA. In contrast, no increase in RNase III reactivity in tcsA site M was observed in the presence of LhrC-mutM, supporting that this mutant LhrC4 is not capable of binding to site M in tcsA. When looking at the structural changes in labeled LhrC4 in the presence of unlabeled tcsA, interactions at LhrC4´s site M were also evident (Figure 3(b)). In the presence of unlabeled tcsA-wt or tcsA-mutP, increased RNase III cleavage was observed at U24 and U33 in LhrC4, being consistent with M site interactions. In contrast, tcsA-mutM did not induce a similar increase in RNase III cleavage at LhrC4 site M. Finally, we noticed that upon tcsA-wt binding, distinct structural changes were seen in the distal part of LhrC4 stem A, which shifted from double stranded to single stranded (Figure 3(b)). The tcsA M- and P-mutations had a somewhat ambiguous effect on the stem-loop A region of LhrC4 (Figure 3(b)). Here, RNase A and T1 cleavages supported a structural change of stem A similar to that caused by tcsA-wt (i.e. distal part of stem A becomes single stranded), whereas the RNase III pattern supported a closed stem-loop A similar to naked LhrC4 and, finally, lead(II) protection was intermediary between an open and a closed structure. These findings suggest that the stem-loop A structure is partially destabilized upon interaction with both mutant variants of tcsA.

Figure 3.

Probing the interaction between tcsA mRNA and LhrC. (a) 5´-end labeled tcsA RNA was partially digested with RNase A, RNase T1, RNase III or lead(II) (Pb), in the absence (-) or presence of non-labeled LhrC4-wt (WT), LhrC4-mutM (M) or LhrC4-mutST (ST). As controls, an alkaline ladder (Alk), RNase T1 ladder (G), and untreated sRNA-mRNA samples (Control) were included. Brackets mark the tcsA regions corresponding to Site P (blue) and Site M (red). (b) 5´-end labeled LhrC4 was partially digested with RNase A, RNase T1, RNase III or lead(II) (Pb), in the absence (-) or presence of non-labeled tcsA-wt (WT), tcsA-mutM (M) or tcsA-mutP (P). As controls, we included an alkaline ladder (Alk), RNase T1 ladder (G), and untreated sRNA-mRNA samples (Control). The brackets mark Site T (blue), Site S (blue), the distal part of Stem A (black), and Site M (red), respectively. The experiment was repeated three times with similar results.

Altogether, the results from the in vitro binding studies provide experimental evidence for the M site interaction between tcsA and LhrC4, as well as the interaction between the P site of tcsA and T site of LhrC4´s terminator loop. Furthermore, either type of interaction seems to destabilize the stem-loop A structure of LhrC4.

The P site is important for LhrC-mediated post-transcriptional control of tcsA in vivo

The in vitro binding studies suggested that LhrC4 targets two sites in tcsA; the M and P sites. To assess the role of sites M and P in vivo, a reporter gene strategy was employed that allows detection of regulatory effects on the post-transcriptional level of tcsA [12]. A DNA sequence encoding the 5′-end of the tcsA transcript, and additional 53 bp of the tcsA coding region, was fused downstream of a moderate promoter which is not affected by the LhrCs [12]. This sequence was inserted in-frame to lacZ in vector pCK-lac, resulting in the translational reporter plasmid pC-tcsA-lacZ. We showed previously that after 2 hours of cefuroxime stress, the β-galactosidase activity in a wild-type strain was reduced by 2.5-fold relative to a mutant strain lacking LhrC1-5 [12]. To evaluate the importance of sites M and P for LhrC-mediated control of tcsA, two mutant variants of pC-tcsA-lacZ were constructed: pC-tcsA-mutM and pC-tcsA-mutP. These plasmids carry the mutations in tcsA site M and P, respectively, as shown in Figure 2(a-c). The wild-type and mutant plasmids were introduced into a L. monocytogenes wild-type strain (LO28-wt), as well as a mutant strain lacking all seven members of the LhrC family (LO28-Δ7). The resulting strains were tested in a β-galactosidase assay; the results are shown in Figure 4. Under control conditions (i.e. no stress applied), the β-galactosidase activity of the wild-type and LhrC mutant strains were comparable for each of the three constructs. However, after two hours of cefuroxime stress, the β-galactosidase activity was found to be 2.3-fold higher in the LhrC mutant strain relative to the wild-type strain containing the wild-type pC-tcsA-lacZ construct. Likewise, a 2.0-fold difference was observed when comparing the LhrC mutant and wild-type strains containing pC-tcsA-mutM; thus, introducing mutations in site M has no major effect on LhrC-mediated control of tcsA. In contrast, mutations in the P site abolished regulation by LhrC on tcsA; the β-galactosidase activity in the LO28-wt and LO28-Δ7 strains containing pC-tcsA-mutP was comparable (1.1-fold difference) (Figure 4). In addition to the β-galactosidase experiments, the tcsA-lacZ mRNA level in cefuroxime-stressed cells was assessed by primer extension (PE) analysis (Figure 4). For strains containing pC-tcsA-lacZ or pC-tcsA-MutM, we observed that the mRNA levels were at least 4-fold higher in the LhrC mutant relative to the wild-type, whereas for strains containing pC-tcsA-mutP, the levels of tcsA-lacZ mRNA were comparable (Figure 4). Overall, the results of the β-galactosidase assay and PE analysis demonstrate that, although LhrC4 interacts with both sites M and P in vitro, only site P is important for LhrC-mediated control of tcsA in vivo.

Figure 4.

Assessing the role of Site M and Site P in tcsA for LhrC regulation. Plasmids containing tcsA wild-type (wt) or mutant variants of tcsA (mutM and mutP) fused in-frame to lacZ, were introduced into L. monocytogenes LO28-wt (LO28) and LO28-Δ7 (Δ7). The β-galactosidase activity of the resulting strains was measured after 2 hours of cefuroxime stress. Non-stressed cells were included as controls. The result shown is the average of three biological replicates, each performed in technical duplicates. After cefuroxime stress, a significant difference was observed between LO28-wt and LO28-Δ7 strains containing tcsA-wt fused to lacZ (2.3-fold); likewise, a significant difference was seen for the LO28-wt and LO28-Δ7 strains containing tcsA-mutM fused to lacZ (2.0-fold difference) (*p < 0.05; **p < 0.001). For the cefuroxime-stressed cells, a primer extension analysis was performed as well (shown below). The relative difference in band intensity between wt and Δ7 strains was 4.5-fold for cells containing pC-tcsA-lacZ; 4.0-fold for pC-tcsA-mutM and 1.3-fold for pC-tcsA-mutP.

The LhrC sRNAs affect the stability of tcsA mRNA independently of translation

The experiments performed so far demonstrate that the LhrC sRNAs down-regulate tcsA expression at the post-transcriptional level by direct base pairing to the upstream site P in the 5´-UTR of tcsA mRNA. However, it remains to be determined if the LhrCs act to prevent translation and/or affect the stability of tcsA mRNA. To further investigate the mechanism underlying the repressive effect of the LhrCs, an in vitro translation experiment was performed, where translation of tcsA mRNA was assessed in the absence or presence of increasing levels of LhrC4 (Figure 5). Interestingly, the production of TcsA protein remained very high, even in the presence of a 50-fold excess of LhrC4 sRNA relative to tcsA mRNA (45% inhibition; see Figure 5). As a control, the LhrC target mRNA oppA was included; for this target, the LhrC sRNAs are known to inhibit translation initiation by direct binding to the Shine-Dalgarno region [12,13]. For oppA, translation was prevented upon addition of only a 2-fold excess of LhrC4 (97% inhibition; see Figure 5). These results demonstrate that LhrC-mediated inhibition of translation occurs very inefficiently for tcsA compared to oppA, suggesting that the LhrCs most likely exert their regulatory effect on tcsA expression at the level of mRNA stability.

Figure 5.

Assessing the regulatory effect of LhrC on in vitro translation of tcsA mRNA and oppA mRNA. In vitro-transcribed tcsA and oppA mRNAs were used as templates for synthesis of TcsA and OppA proteins using a reconstituted protein synthesis system. Increasing concentrations of LhrC sRNA were added to protein synthesis reactions containing either tcsA mRNA or oppA mRNA as template. The first lane (C) contains a control reaction with no template mRNA or LhrC4 added. The levels of TcsA and OppA protein were measured and used for calculating the effect of LhrC4 on protein synthesis (% inhibition). The experiment was performed three times with similar results.

To further explore the mechanism underlying the post-transcriptional regulatory effect of LhrC on tcsA, we determined the in vivo half-life of tcsA mRNA in the presence or absence of LhrC. First, exponentially growing cells were exposed to LhrC-inducing conditions for 1 hour (i.e. cefuroxime exposure). Then, rifampicin was added (time = 0 min) and samples were drawn at various time points for RNA purification. The level of tcsA mRNA was determined by northern blot analysis and the results were used for half-life calculations (Supplementary Figure S5). The resulting half-life of the tcsA mRNA in L. monocytogenes LO28-wt and LO28-Δ7 is presented in Figure 6. Interestingly, the half-life of tcsA mRNA was 3.2-fold higher in the LO28-Δ7 mutant relative to the LO28-wt strain, demonstrating that tcsA mRNA is destabilized in the presence of LhrC sRNAs (Figure 6). To investigate if LhrC-mediated destabilization of tcsA mRNA occurs independently of translation, the start codon of tcsA was disrupted (Figure 6; Supplementary Figure S6). We observed that disruption of the start codon resulted in a decreased half-life when compared to tcsA-wt mRNA. More specifically, the half-life of the non-translated tcsA mRNA was 3.6 min in the LhrC mutant strain (Figure 6), whereas half-life of wild-type tcsA mRNA was longer (7.8 min) (Figure 6). Similarly, in the presence of LhrC, the half-life of the start-codon mutant tcsA mRNA was lower than wild-type tcsA (1.5 min and 2.4 min, respectively). Thus, by mutating the start codon of tcsA, the transcript stability was generally lowered, but most importantly, LhrC-mediated destabilization was still observed, thereby demonstrating that LhrC affects tcsA mRNA stability independently of translation.

Figure 6.

Testing the roles of LhrC and translation on tcsA mRNA stability. L. monocytogenes LO28-wt (LO28) and LO28-Δ7 (Δ7) strains containing chromosomal mutations in Site M (mutM), Site P (mutP) or the start codon of tcsA (mut Start) were used for testing the half-life of tcsA mRNA under LhrC-inducing conditions. The parental strains (i.e. LO28-wt and LO28-Δ7), containing the wild-type tcsA locus (wt), were included as controls. The half-life of tcsA mRNA variants was calculated from the northern blots shown in Supplementary Figure S5. The half-life of tcsA mRNA in LhrC-producing and LhrC-deficient strains was compared to assess the role of LhrC regulation on tcsA mRNA stability; statistically significant differences were found for tcsA-wt (3.2-fold), tcsA-mutM (2.9-fold) and tcsA-mut-Start (2.4-fold) (*p < 0.05; **p < 0.01; ***p < 0.005).

To further analyze the LhrC-mediated destabilization of tcsA mRNA, the M and P site mutations were introduced into the genomic region upstream from the tcsA gene in LO28-wt and LO28-Δ7. The half-life of tcsA mRNA in the resulting strains was determined under LhrC-inducing conditions. Interestingly, when comparing the half-life of tcsA-mutM (1.4 min) with the half-life of tcsA-wt (2.4 min) in the presence of LhrCs, we observed that the M site mutation decreased tcsA mRNA stability (Figure 6); a similar trend was observed in the LhrC mutant background (3.9 min versus 7.8 min; compare LO28-Δ7_tcsA-mutM and LO28-Δ7_tcsA-wt). Curiously, the relative half-life of tcsA-mutM in the absence and presence of LhrC was comparable to the relative half-life of tcsA-wt (2.9-fold versus 3.2-fold; see Figure 6). Thus, although the M mutation renders tcsA mRNA less stable, the LhrC-mediated regulatory effect is preserved. As for the P mutation, the tcsA-mutP mRNA was also less stable in the LhrC mutant (4.2 min) relative to wild-type tcsA mRNA in absence of LhrC (7.8 min) (Figure 6; compare LO28-Δ7_tcsA-mutP to LO28-Δ7_tcsA-wt). Strikingly, the half-life of tcsA-mutP mRNA was comparable in the presence and absence of LhrC (4.0 min and 4.2 min, respectively) (Figure 6; compare LO28-wt_tcsA-mutP and LO28-Δ7_tcsA-mutP). Thus, LhrC-mediated control was clearly abolished by mutating site P in the upstream region of tcsA. Altogether, the results of the half-life experiments show that the LhrCs act through site P to destabilize the tcsA mRNA. In contrast, site M is not important for the LhrC-mediated effect on tcsA mRNA stability.

Discussion

Canonical regulation by sRNAs refers to a direct base pairing interaction overlapping the RBS, leading to ribosome occlusion and translational repression, followed by degradation of the naked mRNA [1–3,5]. For instance, all members of the LhrC family of sRNAs (LhrC1-5, Rli22, and Rli33-1) canonically repress OppA expression by directly base pairing to the RBS, leading to decreased translation and stability of the oppA mRNA [12,13] (illustrated in Figure 7). In contrast, we show in this work that the LhrC sRNAs post-transcriptionally repress TcsA expression by base pairing far upstream of the RBS, suggesting a non-canonical regulatory mechanism (Figure 7). We show by in vitro translation assay that, compared to oppA, the tcsA mRNA is translated efficiently even in the presence of a large excess of LhrC4 (Figure 5), suggesting that translation is not the major point at which LhrC regulates TcsA expression. In contrast, the half-life of tcsA mRNA was increased 3.2-fold in cells harboring a deletion of all seven LhrC family members (LO28-Δ7), suggesting that LhrC promotes tcsA mRNA turnover (Figure 6). A start codon mutation rendered tcsA less stable relative to wt tcsA (Figure 6), likely reflecting increased ribonucleolytic degradation in the absence of translating ribosomes [3]. Importantly, deletion of the LhrC family still led to a 2.4-fold stabilization of tcsA encoding the start codon mutation (Figure 6), confirming that LhrC destabilizes the tcsA mRNA independently of translation. Finally, mutation of site P in tcsA abolished the effect of LhrC-deletion on tcsA mRNA half-life (Figure 6), confirming that base pairing at site P, far upstream of the RBS, leads to destabilization of the tcsA mRNA by the LhrC sRNAs. Together, the data support a mechanism wherein the LhrC sRNAs base pair to the P site to promote turnover of the tcsA mRNA (Figure 7).

Figure 7.

Model of LhrC regulation of oppA (left) and tcsA (right). The LhrC family of sRNAs repress OppA expression by directly base paring to the RBS, leading to ribosome occlusion and translational repression (left). Two CU-rich regions are capable of base pairing to the RBS of oppA mRNA. In contrast, the LhrC sRNAs repress TcsA expression by base pairing to Site P, far upstream of the RBS, which promotes degradation of tcsA mRNA without directly affecting translation (right). The CU-rich region of the terminator loop is involved in base pairing to Site P. Black box, RBS; grey box, Site P; grey ovals, ribosome; black packman, RNase; base pairing is indicated by thin lines.

In Gram-negative bacteria, the base pairing of Hfq-binding sRNAs to the RBS often actively recruit the endoribonuclease RNase E to the target mRNA [2]. The Hfq-binding sRNA LhrA in L. monocytogenes might act in a similar manner, although the RNase(s) involved remains to be identified [3,9,10]. In such cases, inhibition of translation initiation is viewed as the main event in sRNA-mediated gene silencing, whereas mRNA degradation makes the regulation irreversible [2]. Other examples from Gram-negative bacteria illustrate that in some situations, the primary regulatory event of an sRNA might be to direct degradation of its target mRNA. In the case of MicC in Salmonella typhimurium, this sRNA was reported to inhibit expression of OmpD by base pairing to the coding sequence (CDS) of ompD mRNA [16]. When base pairing with the CDS, MicC does not inhibit translation initiation; rather, it stimulates cleavage of ompD mRNA by a controlled process that involves Hfq and RNase E [17]. Recently, the sRNA McaS in Escherichia coli was shown to promote degradation of its target mRNA csgD by base pairing to the 5´-UTR [18]. More specifically, McaS base pairs to an AU-rich region far upstream from the RBS, which stimulates Hfq- and RNase E-dependent cleavage of csgD mRNA [18]. So far, only a few sRNAs in Gram-positive bacteria have been shown to modulate mRNA degradation without affecting translation (e.g. ska mRNA stabilization by FasX sRNA in Streptococcus pyogenes [19] (reviewed in [3])). Although Hfq is not required [12] and RNase E generally is absent in Gram-positive bacteria [3,20], our findings clearly support that the LhrCs in L. monocytogenes act to promote degradation of tcsA mRNA without intervening in the translation process.

The specific ribonuclease(s) acting together with LhrC to destabilize tcsA mRNA remains to be identified. We observed clear alterations in the secondary structure of both LhrC4 and tcsA mRNA upon their intermolecular base-pairing (Figure 3 and Supplementary Figure S4). For instance, the distal portion of LhrC4 stem-loop A was generally destabilized upon tcsA binding, while several regions of tcsA mRNA became more single stranded upon LhrC4 binding (Supplementary Figure S4; nt −101 to −97, −66 to −64, and −47 to −42). It is possible that this refolding of tcsA reveals one or more cleavage site(s) for one or more ribonuclease(s). So far, knowledge on the activities of specific RNases in L. monocytogenes is scarce, but based on studies in related bacteria, in particular Bacillus subtilis and Staphylococcus aureus, several RNases might be implicated [3,20]. For example, the endonuclease RNase Y, which might play an analogous role to the RNase E of Gram-negative bacteria [3,20], prefers AU-rich single-stranded regions, and four uracils (in the −101 to −97 stretch of tcsA) are more exposed upon LhrC binding (Supplementary Figure S4). In addition to RNase Y, L. monocytogenes encodes several other interesting candidates, such as the double strand-specific endoribonuclease RNase III, the exoribonucleases RNase J1 and J2, and an ortholog of the catalytic domain of RNase E (RNase E/G) [3,20]. Indeed, the RNase III cleavage pattern of the tcsA 5´-UTR was altered in several regions upon binding of LhrC4-wt (Figure 3 and Supplementary Figure S4; e.g. increased cleavage of U-55). These observations support that, due to structural changes, tcsA might become more prone to cleavage by one or more ribonuclease(s) upon binding of LhrC sRNAs. To follow up on these findings, we made several attempts to delete some of the abovementioned RNase-encoding genes in L. monocytogenes, but so far, we only succeeded in creating an in-frame deletion of RNase J2. This RNase was clearly dispensable for LhrC-mediated degradation of tcsA mRNA (Supplementary Figure S7). Future studies should focus on defining the ribonuclease(s) involved in sRNA-mediated regulation in L. monocytogenes.

In addition to the in silico-predicted interaction between LhrC4 and tcsA (site P), we found a base pairing interaction even further upstream, dubbed site M (Figures 2 and 3). We were surprised to find that base pairing at site M was dispensable for LhrC-mediated regulation in vivo (Figures 4 and 6), given its near-perfect conservation amongst LhrC1-5 (Supplementary Figure S3). It is possible that these bases are conserved due to their roles in the LhrC secondary structure (Supplementary Figure S2). Notably, C35 is unique to LhrC4 (Supplementary Figure S3); it is possible that this is important for a more extended base-pairing interaction between LhrC4 and tcsA, which might be absent in the other LhrCs. Nevertheless, tcsA mutM, which clearly disrupted base pairing with LhrC4 site M in vitro, was efficiently regulated by the LhrC family in vivo (Figures 5 and 6). These data suggest that site P plays the active regulatory role in vivo, while site M explains the residual base-pairing interaction observed in vitro upon mutation of site P.

Using a phylogenomics approach to detect coevolution between sRNAs and protein-coding genes amongst the 79 complete genomes of the Listeria genus, Cerutti et al. [21] identified 52 sRNAs, most of which were present in the Listeria common ancestor and lost during Listeria evolution. Significant coevolution was detected between the genes encoding 23 sRNAs and 52 protein-coding genes, enabling the characterization of a ‘main hub’ of 12 sRNAs that coevolved with genes encoding cell wall proteins and virulence factors [21]. Interestingly, the LhrCs were among the list of 60 core sRNAs that were present and conserved in all Listeria genomes [21], suggesting an important role for the LhrC sRNAs in Listeria species. Notably, the LhrC siblings are induced under various infection-relevant conditions, and LhrC1-5 and Rli33-1 contribute to virulence in L. monocytoegenes [12,15], likely reflecting their importance in fine-tuning the regulation of OppA, LapB, and TcsA [11–13] – all of which are cell envelope-associated proteins required for full Listerial virulence [22–24]. LapB is a virulence adhesin required for entry into eukaryotic cells [22] and OppA an oligo-peptide binding protein required for growth of L. monocytogenes at low temperature and involved in intracellular survival [23]. TcsA is a lipoprotein and CD4 + T cell-stimulating antigen; although its precise function in L. monocytogenes is still unknown, its disruption leads to a 10-fold reduction in bacterial recovery from the liver and spleen in a murine infection model [24]. Taken together, the evidence suggests that a fine-tuning of the levels of these cell envelope-associated proteins is beneficial to L. monocytogenes under a variety of infection-relevant conditions, such as in blood. As whole human blood contains components integral to the host’s immune response, and surface-exposed proteins are readily recognized by the immune system, the LhrC-mediated down-regulation of OppA, LapB, and TcsA might represent a mechanism by which L. monocytogenes evades host immunity [12]. More recently, we have shown that LhrC1-5 are induced in response to excess heme, the core component of hemoglobin in blood [25]. We demonstrated that LhrC1-5 contribute to the adaptation of L. monocytogenes to excess heme, and that LhrC1-5 down-regulate the expression of genes involved in heme uptake and utilization, including lmo2186, lmo2185 (which encode the heme-binding proteins Hbp1 and Hbp2, respectively) and lmo0484 (which encodes a heme oxygenase-like protein) [25]. As previously demonstrated for oppA and lapB, we found that LhrC4 interacts with lmo2186, lmo2185, and lmo0484 mRNAs and, for lmo0484, LhrC4 uses a CU-rich loop for base-pairing to the AG-rich RBS of the mRNA [25]. Thus, for all LhrC targets characterized so far, the LhrC family members follow the canonical model of sRNA regulation.

In this work – which comprises the first detailed study of the mechanism by which the LhrC sRNAs regulate TcsA expression – we show that, like other LhrC targets, tcsA is repressed post-transcriptionally. In contrast, however, LhrC base pairs at an AG-rich region of tcsA well upstream of the RBS and destabilizes the tcsA mRNA, without directly affecting tcsA translation. This is a novel mechanism for the LhrC family of sRNAs, suggesting that their regulatory repertoire extends well beyond canonical regulation. Moreover, this is, to our knowledge, the first example of an sRNA in L. monocytogenes that primarily regulates mRNA turnover without affecting translation initiation. Therefore, future screens for putative sRNA targets in L. monocytogenes, and studies of the mechanisms utilized by these sRNAs, should not be limited to those sRNAs that are predicted to bind near the RBS of an mRNA.

Experimental procedures

Bacterial strains and growth conditions

The wild-type strains used in this study were Listeria monocytogenes LO28 [26] and Listeria monocytogenes EGD [27]. The mutant strain LO28-Δ7, which carries deletions of all seven LhrC-family members, was constructed in a previous study [13]. Strains carrying genomic substitutions in site M upstream from tcsA (LO28-wt_tcsA-mutM, LO28-Δ7_tcsA-mutM), site P upstream from tcsA (LO28-wt_tcsA-mutP, LO28-Δ7_tcsA-mutP) or the start codon of tcsA (LO28-wt_tcsA-mut-Start, LO28-Δ7_tcsA-mut-Start) were constructed using the temperature-sensitive shuttle vector pAUL-A as described previously [28]. Positive genomic mutants were identified by touch-down PCR and confirmed by sequencing. The mutant strain carrying an in-frame deletion of the RNase J2-coding gene, lmo1434, was constructed as described previously [28]. All fragments for genomic substitution or deletion were amplified from chromosomal DNA in a two-step PCR procedure with primers listed in Supplementary Table S1. L. monocytogenes was routinely grown in Brain Heart Infusion medium (BHI, Oxoid) at 37°C with shaking. When appropriate, antibiotics were added to the medium (50 µg/ml kanamycin, 5 µg/ml erythromycin or 4 µg/ml cefuroxime). Escherichia coli TOP10 (Invitrogen) was used for cloning purposes. E. coli TOP10 was grown in Luria-Bertani medium (LB, oxoid) at 37°C with shaking; when appropriate, 50 µg/ml kanamycin or 150 µg/ml erythromycin was added to the medium. For the Northern blot shown in Supplementary Figure S7, cells were grown to OD₆₀₀ = 0.2 in BHI medium. Then, the culture was split and half of the culture was subjected to 4 µg/ml cefuroxime, whereas the other half was left untreated (control). Samples were harvested after 30 and 60 min and total RNA was purified as described below.

In vitro transcription

DNA templates carrying a T7 RNA polymerase promoter at their 5ʹ-end were generated by PCR (overlapping primers or genomic DNA as template) using primers listed in Supplementary Table S1. In vitro transcription, RNA purification, de-phosphorylation and labeling of the transcripts was performed as described previously [29].

Electrophoretic mobility shift assay (EMSA)

The EMSA experiments were designed and performed as described previously [29]. For the EMSA shown in Supplementary Figure S1, the RNA samples were heated for 1 min at 95°C and then left on ice for 2 min, before incubating the samples at 37°C for 1 hour.

In vitro structure probing

T7-transcribed tcsA-up or LhrC4 was dephosphorylated, 5ʹ-end labeled and purified as described above. Labeled RNA (0.1 pmol radiolabeled RNA and 1 μg yeast tRNA in 1X RNA Structure Buffer (Invitrogen, AM2283)) and 2.5 pmol non-labeled RNA was heated separately for 1 min at 95°C, iced for 2 min and incubated 5 min at 37°C. Labeled and non-labeled RNA was mixed (total volume of 9 μl) and allowed to base-pair by incubation for 1 hour at 37°C. For enzymatic digestion, 1 μl RNase A (2 pg, Invitrogen), 1 μl RNase T1 (0.1 U, Invitrogen) or 0.5 μl 20 mM DTT + 1 μl ShortCut RNase III (0.2 U, New England Biolabs Inc.) was added to the reactions and incubation was continued at 37°C for 5 min. Similarly, for the RNase T1 ladder, 0.1 pmol radiolabeled RNA and 1 μg yeast tRNA in 1X Sequencing Buffer (Invitrogen) was heated, iced, and treated with 1 μl RNase T1 (0.1 U) for 5 min at 37°C. For chemical probing, 1 μl 25 mM lead(II) acetate was added to the RNAs and reactions were incubated for 2.5 min at 37°C. Alkaline hydrolysis was performed in a total volume of 10 μl with 0.2 pmol radiolabeled RNA, 1 μg yeast tRNA and 1X Alkaline Hydrolysis Buffer (Invitrogen) by incubation at 95°C for 5 min. All reactions were terminated by adding 10 μl ice cold Loading Buffer II (Invitrogen). Reactions were separated on an 8% polyacrylamide gel at 45 W for 70 min. The gel was dried and exposed on a PhosphorImager screen. The resulting bands were visualized using a Typhoon FLA9500 (GE Healthcare) and analyzed with IQTL 8.0 quantification software (GE Healthcare).

LacZ-fusions and β-galactosidase assays

Post-transcriptional regulation of tcsA was monitored using in-frame translational lacZ fusions of tcsA in the vector pCK-lac [9]. Four plasmids were constructed: pCK-tcsA-lacZ, which contains the wt sequence of tcsA fused to lacZ, and the mutant variants pCK-tcsA-mutM, pCK-tcsA-mutP and pC-tcsA-mut-Start, where site M, site P or the start codon of tcsA was disrupted. Fusions in pCK-lac were designed and constructed as previously described [11]. In brief, DNA fragments containing a moderate promoter [11] as well as tcsA sequence ranging from the transcription start site into the first codons, were acquired by PCR using the primers listed in Supplementary Table S1. The PCR fragments were digested with EcoRI and BamHI and inserted into pCK-lac. For β-galactosidase assays, cells containing the pCK-plasmids were grown to OD₆₀₀ = 0.2. Then, the cultures were split and one half was treated with 4 µg/ml cefuroxime for 2 hours, whereas the other half of the culture was left untreated (control). Samples (1 ml) were harvested and the β-galactosidase assays conducted as previously described [28]. The β-galactosidase activities were analyzed using a two-tailed Student’s t-test (i.e., wild-type, stressed vs. mutant, stressed). Only differences with at least 95% confidence were reported as statistically significant.

In vitro translation assay

Translation reactions were performed using the PURExpress In Vitro Protein Synthesis kit (New England Biolabs Inc.) in a total volume of 10 μl. Reactions containing 5 pmol in vitro transcribed full-length mRNA (tcsA or oppA) ± excess LhrC4 were heated 2 min at 70°C and slow cooled to 37°C. Then, 8 U RNasin Plus (Promega) and 12 nmol [³⁵S]-methionine (>1000 Ci/mmol, Perkin Elmer) were added together with solution A and B according to the manufacturer’s guidelines. Reactions were incubated for 3 hours at 37°C and analyzed on NuPAGE 4–12% Bis-Tris gels (Invitrogen). Prior to gel loading, samples were mixed with 2 volumes of SDS loading buffer (3 mM EDTA, 30% glycerol, 6% SDS, 180 mM Tris-HCl [pH 7.5], 225 mM DTT, BPB) and boiled for 3 min. PageRuler Prestained Protein Ladder (10 to 180 kDa, ThermoFisher Scientific) was loaded as size marker. Gels were run at 200 V for 40 min in MES SDS running buffer. Dried gels were exposed overnight to PhosphorImager screens for imaging and band intensities were quantified using ImageQuant software (GE Healthcare). To correct for differences in loading, TcsA and OppA band intensities were normalized to the uppermost background band on the autoradiographs.

RNA extraction

For northern blot and primer extension assays, total RNA was extracted as previously described [25]. Briefly, bacterial cells were harvested by centrifugation for 3 min at 5000g at 4°C and snap-frozen in liquid nitrogen. Cells were disrupted by using a FastPrep instrument (Bio101, Thermo Scientific Corporation) and total RNA was extracted using TRIzol Reagent (Ambion) as described previously [9]. The integrity, concentration and purity of the RNA were confirmed by agarose gel electrophoresis and DeNovix DS-11 Fx+.

Northern blot assays

Ten micrograms of total RNA was separated on a 4.5% polyacrylamide gel (300 V, 2.5 hours) and electroblotted for 2 hours at 400 mA, as described previously [9]. Alternatively, 20 μg of total RNA was separated on a formaldehyde agarose gel for 3 hours and 15 min prior to capillarity blotting on a Zeta-Probe membrane (Bio-Rad) [30]. The membranes were hybridized with 5ʹ-end labeled probes for tcsA or 16S, listed in Supplementary Table S1. The resulting bands were visualized as stated in the in vitro structure probing section.

Primer extension analysis

Primer extension analysis was performed as described previously [28]. For each sample, we used 10 µg total RNA purified from the strains subjected to β-galactosidase analysis (see above). As described for the β-galactosidase analysis, the strains were treated with 4 µg/ml of cefuroxime for 1 hour before harvesting and RNA purification. The −40 primer, shown in Supplementary Table S1, was used for the primer extension experiment. The resulting bands were visualized as described above for the in vitro structure probing experiment.

RNA stability

Strains grown until OD₆₀₀ = 0.2 were treated with cefuroxime for 1 hour (37°C, shaking) to induce LhrC expression. A sample (time 0) was withdrawn, snap cooled in liquid N₂ and harvested by centrifugation (1 min, 12,000 x g, 2°C); the pellet was frozen in N₂ and stored at −80°C. The remaining culture was treated with 10 μg/ml rifampicin (10 mg/ml stock solution in DMSO) at 37°C with shaking and samples were harvested in a similar fashion at 2, 4, 8, 12 and 16 min after rifampicin addition. Total RNA was extracted and 10 μl was used for northern blotting as described above. Intensities of tcsA bands were normalized to the 16S rRNA band intensity, and the half-life was determined based on an exponential decay curve fitted to the experimental data. Experiments were conducted in biological triplicates.

In silico predictions

Interactions between LhrC sRNAs and tcsA mRNA were predicted using the IntaRNA software [31–33]. Secondary structure predictions of wild-type and mutant variants of LhrC sRNAs and tcsA mRNA were obtained using the Mfold web server [34]. Multiple sequence alignment of LhrC1-5 was performed using Clustal W version 2.0 [35].

Funding Statement

This work was supported by Novo Nordisk Fonden [NNF17OC0028528]; Villum Fonden [341/300-123012].

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplemental data for this article can be accessed here