Rli51 Attenuates Transcription of the Listeria Pathogenicity Island 1 Gene mpl and Functions as a Trans-Acting sRNA in Intracellular Bacteria

Abstract

Listeria pathogenicity island 1 (LIPI-1) is a genetic region containing a cluster of genes essential for virulence of the bacterial pathogen Listeria monocytogenes. Main virulence factors in LIPI-1 include long 5′ untranslated regions (5′UTRs), among which is Rli51, a small RNA (sRNA) in the 5′UTR of the Zn-metalloprotease-coding mpl. So far, Rli51 function and molecular mechanisms have remained obscure. Here, we show that Rli51 exhibits a dual mechanism of regulation, functioning as a cis- and as a trans-acting sRNA. Under nutrient-rich conditions, rli51-mpl transcription is prematurely terminated, releasing a short 121-nucleotide-long sRNA. Rli51 is predicted to function as a transcription attenuator that can fold into either a terminator or a thermodynamically more stable antiterminator. We show that the sRNA Rli21/RliI binds to a single-stranded RNA loop in Rli51, which is essential to mediate premature transcription termination, suggesting that sRNA binding could stabilize the terminator fold. During intracellular infection, rli51 transcription is increased, which generates a higher abundance of the short Rli51 sRNA and allows for transcriptional read-through into mpl. Comparative intracellular bacterial transcriptomics in rli51-null mutants and the wild-type reference strain EGD-e suggests that Rli51 upregulates iron-scavenging proteins and downregulates virulence factors from LIPI-1. MS2 affinity purification confirmed that Rli51 binds transcripts of the heme-binding protein Lmo2186 and Lmo0937 in vivo. These results prove that Rli51 functions as a trans-acting sRNA in intracellular bacteria. Our research shows a growth condition-dependent mechanism of regulation for Rli51, preventing unintended mpl transcription in extracellular bacteria and regulating genes important for virulence in intracellular bacteria.

1. Introduction

Small non-coding RNAs (sRNAs) are bacterial post-transcriptional regulators that control gene expression through varied molecular mechanisms affecting target mRNA/s availability, stability, and/or translation [1]. sRNAs are classically divided into two main categories: cis-acting and trans-acting, depending on whether the sRNA is encoded within the same locus as its target mRNA(s) or a different locus, respectively. Cis-acting sRNAs are either regulatory segments included in untranslated regions (UTRs) or encoded on the opposite strand of target mRNAs (e.g., antisense RNAs and excludons) and regulate exclusively that particular single gene. By contrast, trans-acting sRNAs can regulate multiple targets through imperfect binding to the mRNAs [2]. However, cumulative evidence in the last few years revealed a somehow overlooked mechanistic diversity in sRNA-based regulation, which suggests that this basic classification might be limited [3].

sRNA-based regulation has been associated with virulence in most model pathogenic bacteria [4,5]. Hints linking sRNA function with virulence have most often relied on two facts. First, the sRNA locus belongs to a genetic segment carrying virulence-associated genes, which is absent in non-pathogenic strains (i.e., a pathogenicity island or PAI) [6,7], and second, the sRNA is induced during the course of the infection of eukaryotic cells [8,9,10,11]. Subsequently, the definite proof was that null mutants for the sRNA showed reduced virulence in in vitro and/or in vivo models of infection. However, the number of sRNAs whose deletion leads to an unequivocal attenuation phenotype is still exceedingly low compared to the number of sRNAs located in PAIs and/or induced in infection conditions. For instance, comparative genome-wide analyses showed that around one-fifth of all identified sRNAs (29 out of 150) in the foodborne pathogen Listeria monocytogenes (L. monocytogenes) are specifically expressed in intracellular bacteria infecting macrophages. On the other hand, a similar number of sRNA loci were exclusively encoded in pathogenic L. monocytogenes strains and were absent in related non-pathogenic Listeria species [9]. However, to our knowledge, a solid functional association of L. monocytogenes sRNAs with virulence has been established only for a handful of them [8,9,12]. Thus, at present, the precise regulatory role of the majority of sRNAs induced in infection conditions and/or located in PAIs has not been described yet.

L. monocytogenes is a Gram-positive bacterial pathogen that causes listeriosis, one of the most serious foodborne diseases in humans and animals [13]. The most relevant virulence factors are clustered in the genome making up PAIs, from which the Listeria pathogenicity island 1 (LIPI-1) is present in most isolates of pathogenic species L. monocytogenes and L. ivanovii [13]. LIPI-1 is highly induced in intracellular bacteria and comprises seven well-characterized protein-coding genes that are critical for intracellular survival and bacterial dissemination: prfA, plcA, hly, mpl, actA, plcB, and orfX [14]. In addition, there are two annotated sRNAs in LIPI-1. Rli74 and Rli51 are located in the 5′-UTR of the virulence factor genes actA and mpl, respectively [7,8,9]. A single promoter drives the intracellular expression of actA, which encodes a cell surface protein essential for intracellular motility and the invasion of neighboring cells. This promoter generates an mRNA with a long UTR that includes Rli74. Deletion of rli74 halted ActA-mediated actin recruitment and cell-to-cell spread, indicating that Rli74 is a cis-acting sRNA essential to achieve a sufficient ActA level to drive bacterial motility within cell host cytoplasm [15]. Similarly, the transcription start site of mpl, which codes for a metalloprotease that activates other important virulence factors, such as the phospholipase PlcB, and is essential for cell-to-cell spread, is located 150 bps upstream of the start codon, suggesting that Rli51 represents a long 5′UTR of the mpl transcript [16,17]. Intriguingly, there are no further molecular or functional analyses investigating the role of Rli51 as a cis-encoded sRNA. Thus, currently, it is unknown whether Rli51 regulates the expression of the metalloprotease Mpl and the relevance that such cis regulation would have for the virulence of L. monocytogenes.

Evidence retrieved from different transcriptomic analyses of L. monocytogenes challenges the view of Rli51 as a 5′-UTR of mpl transcript. Specifically, tiling arrays suggested that Rli51 could be a short 121-nucleotide-long RNA induced in blood and the intestine of infected mice [8]. Consistently, Term-seq, a genome-wide approach aimed to identify the 3′ end of expressed transcripts, identified a transcription termination for Rli51 upstream of the mpl start codon [18]. In fact, the occurrence of Rli51 as a short transcript has been recently verified by northern blot in L. monocytogenes grown in PrfA-inducing conditions [19]. Collectively, these observations suggest that, at least in some conditions, the transcription of Rli51 might not be coupled to mpl mRNA. Therefore, it is not yet clear if Rli51 sRNA is embedded within the 5′-UTR of mpl, is an independent transcript, or undergoes a condition-specific premature transcriptional termination (PTT). Furthermore, we do not know the molecular mechanisms that regulate this process and the functional role of Rli51.

In this work, we aimed to investigate the function of Rli51, an sRNA encoded in the LIPI-1 of L. monocytogenes. Analysis of the transcriptional landscape of the rli51-mpl locus under nutrient-rich laboratory conditions and in intracellular bacteria indicated the occurrence of a conditional regulation of Rli51 transcriptional read-through into mpl. Then, using translational fusions with egfp, we identified the sequence elements and potential trans-acting factors relevant to Rli51 transcription termination regulation. Finally, we explored the functional role of Rli51 in infection-relevant conditions through a transcriptomic analysis of rli51 deletion mutants collected from infected epithelial JEG-3 cells. Our work shows that Rli51 does not participate in mpl upregulation during the infection of eukaryotic cells. Instead, we found that the LIPI-1-encoded sRNA Rli51 functions as a trans-acting sRNA in intracellular bacteria.

2. Results

2.1. Rli51 Does Not Play a Relevant Role in the Upregulation of mpl during Infection

The 5′ UTR regions of the LIPI-1 genes hly and actA are essential for achieving the highest expression in infection-relevant conditions [15,20]. Thus, we first hypothesized a role for Rli51 as a cis-acting element that contributes to mpl upregulation in infection conditions. To test this hypothesis, we analyzed the impact of rli51 deletion on mpl transcript levels. We generated an rli51-null mutant in the L. monocytogenes 1/2a EGD-e strain genomic background and monitored mpl mRNA abundance by reverse transcription coupled to quantitative PCR (RT-qPCR). The transcription of LIPI-1 is induced in intracellular bacteria. Consistently, mpl mRNA abundance in bacteria collected from infected cells was increased by more than 100 folds compared to the extracellular condition (Figure 1A). However, deletion of chromosomal rli51 did not impair mpl upregulation in intracellular bacteria; on the contrary, although not significant, in the Δrli51 strain mpl expression in intracellular bacteria was higher than in the the wild-type strain (Figure 1A). These results indicate that Rli51 does not play a significant role in maximizing mpl expression during infection.

Induction of mpl might result from the activation of a cryptic promoter located downstream of the rli51 promoter, which, in turn, would drive the transcription of an mpl transcript without Rli51. Such a situation would make the infection-specific induction of mpl independent of Rli51. In such a scenario, Rli51 might still have a cis-acting regulatory activity over mpl in other conditions. To ascertain the occurrence of such a yet unidentified promoter element upstream of mpl, we analyzed mpl transcription using translational fusions in which the rli51 promoter had been deleted (Figure 1B) [16,17,21]. In constructs devoid of the rli51 promoter, mpl-egfp expression showed a 10- and 100-fold decrease when grown in BHI and intracellular bacteria, respectively (Figure 1B). Thus, in the absence of the rli51 promoter, there was not any induction of mpl in intracellular bacteria that could be attributed to an additional cryptic promoter. This result indicates that mpl upregulation during infection results exclusively from the activation of the rli51 promoter. Therefore, both in LIPI-1-inducing and in LIPI-1-non-inducing conditions, Mpl protein synthesis originates from a unique mpl mRNA, which includes Rli51 as a 5′UTR.

In addition to transcriptional activation, cellular trans-acting factors could concomitantly act upon Rli51 post-transcriptionally to support mpl upregulation in intracellular bacteria. To analyze the influence of Rli51 on mpl mRNA abundance independently of these potential infection-specific factors, we sought to induce the expression of translational fusions of mpl-egfp in bacteria grown in BHI (Figure 1C). To accomplish this aim, we used an inducible tet promoter that allowed us to achieve a level of mpl-egfp mRNA in extracellular conditions comparable to that observed with the rli51-mpl promoter in intracellular bacteria (Figure 1B,C). Deletion of rli51 in the construct did not result in a decreased mpl-egfp transcript abundance, irrespective of the level of induction; instead, the absence of rli51 even seemed to have a positive effect on mpl-egfp expression at higher levels of transcript induction (Figure 1C). These results recapitulate the effect that chromosomal rli51 deletion had on endogenous mpl expression in intracellular bacteria (Figure 1A) and confirm that mpl upregulation during infection results solely from the induction of the rli51-mpl promoter.

Thus far, we have analyzed the potential effects of Rli51 over the mpl transcript. However, Rli51 might be a binding site for post-transcriptional regulators that control Mpl protein expression without affecting its mRNA abundance. To rule out the participation of such regulators, we analyzed the phenotype of the rli51 deletion mutant in in vitro infection models (Figure 1D). We reasoned that, similar to the previously demonstrated Rli74-mediated cis-acting regulation over ActA [15], disrupting a key regulator of the virulence factor Mpl—an essential protease for the activation of the phospholipase PlcB and cell-to-cell spread [22,23]—would significantly affect the infection process. Therefore, we analyzed the bacterial burden at 6 h after infection, which is enough time for L. monocytogenes to escape from the vacuole and spread from cell to cell [23]. rli51 knockout strains showed no significant differences in the intracellular proliferation index (quotient of the number of intracellular bacteria at 6 h and at 2 h after infection) compared to EGD-e infecting the human syncytiotrophoblast (JEG3) or rat fibroblast (NRK-49F) cell lines (Figure 1D). These results suggest that Mpl protein expression levels are not altered in the absence of Rli51 sRNA.

Altogether, these results demonstrate that Rli51 does not participate in the upregulation of mpl in intracellular bacteria. Conversely, mpl expression seems to be higher in the Δrli51 strain, both in extracellular and intracellular bacteria, which suggests that Rli51 might be a cis-acting negative regulatory element of the mpl transcript.

2.2. Condition-Specific Premature Termination of Rli51-mpl Transcription

The results from previous reports suggest that Rli51 occurred as a short 121-nucleotide-long RNA [8,18,19], which goes against the view of Rli51 as the 5′UTR of the mpl transcript. To investigate whether Rli51 transcription is decoupled from mpl transcription, we quantified endogenous Rli51- and Rli51-mpl-containing transcripts by RT-qPCR (Figure 2A, upper panel). With this approach, we verified that Rli51 occurs both as a short transcript (Rli51 from now on) and as the 5′UTR of mpl (Rli51-mpl) (Figure 2A). Both RNA molecules—Rli51 and Rli51-mpl—are expressed in bacteria grown in BHI and in bacteria collected from infected JEG3 cells; however, the proportion of each species was different (Figure 2A). Rli51-mpl represented less than 2% of all Rli51 in bacteria grown in BHI, while this proportion scaled up to 35% in intracellular bacteria (Figure 2A, right axis). Thus, in exponentially growing bacteria, almost all Rli51 is represented by a short transcript, while in intracellular bacteria, there is a transcriptional read-through that allows for Rli51-mpl transcription. These results suggest that Rli51-mpl transcription may be subjected to condition-specific premature transcription termination (PTT).

We sought to verify if Rli51 prompts Rli51-mpl transcript premature termination. We utilized mpl::egfp translational fusions with the rli51 promoter in which the rli51 locus had been replaced by the 5′UTR of the listeriolysin-encoding hly gene. The 5′UTR of hly is similar to rli51 in length (100 nt). In addition, it positively regulates the expression of listeriolysin during intracellular infection while having only a moderate effect when grown in broth [20]. Constructs including the hly 5′UTR seemed to have an increased mpl-egfp transcript abundance in intracellular bacteria (Figure 2B), suggesting that Rli51 could halt transcriptional read-through. rli51 and mpl loci are separated by a 30 nt long sequence that includes the ribosome-binding site. This sequence was, however, irrelevant for PTT since swapping it by the equivalent segment in hly had no effect compared to the rli51-mpl::egfp control construct (Rli51 * vs. Rli51 in Figure 2B). These results suggest that Rli51 is necessary and sufficient to mediate the PTT of Rli51-mpl RNA.

In intracellular bacteria, we observed an increase in both Rli51 and Rli51-mpl transcripts, which entailed an asymmetric increase in the abundance ratio of Rli51-mpl relative to total Rli51 (Figure 2A). As PrfA-dependent promoters are induced during intracellular infection, we hypothesized that the activation of the rli51-mpl promoter is the sole driving force allowing for the transcriptional read-through of Rli51-mpl in this condition. To test this hypothesis, we used an inducible tet promoter to control the transcription of rli51-mpl-egfp translational fusions. Induction of the tet promoter in exponentially growing bacteria in BHI to levels comparable to those achieved with the rli51 promoter in intracellular bacteria led to a consistent and proportional increase in the transcriptional read-through of Rli51-mpl (Figure 2C, right axis). Replacement of rli51 by the 5′UTR of hly relieved Rli51-mediated PTT, allowing for transcriptional read-through and mpl-egfp transcript production (Figure 2C). The difference in the relative levels of the mpl transcript between Rli51 and constructs containing the 5′UTR of hly was greater at lower levels of induction (Figure 2C). This result indicates that Rli51-mediated PTT is more efficient at a lower transcriptional activity of the promoter. Therefore, Rli51-mediated transcription termination may be alleviated by increasing the transcriptional activity, which would contribute to impeding mpl transcription when bacteria are not under LIPI-1-inducing conditions.

2.3. Trans-Acting Regulation of Rli51-Mediated Transcriptional Termination

We next aimed to decipher the molecular mechanism by which Rli51 mediates condition-specific PTT. Conditional PTT represents quite a pervasive post-transcriptional control mechanism in Gram-positive bacteria, where a particular cis-acting sRNA located in a 5′UTR can fold into two mutually exclusive conformations: a transcriptional terminator that mediates PTT and an antiterminator that allows for reading through into the downstream gene [18,24]. Computational analysis using a recently developed algorithm (Pasific) [25] predicted that Rli51 could indeed fold either as terminator or antiterminator conformations (Figure 3A). The terminator fold included three stem loops, and the last one is the actual terminator. In the antiterminator fold, this terminator stem loop is distorted by being involved in the formation of a 20-nucleotide-long stem together with the first stem loop at the 5′end (Figure 3A). The antiterminator is much more stable than the terminator fold (ΔG = −22.9 Kcal/mol and −10.4 Kcal/mol, respectively), which suggests that the antiterminator should be the predominant conformation (Figure 3A). Considering that Rli51 mediates PTT in L. monocytogenes during exponential growth in BHI broth (Figure 2), these findings suggest that Rli51 adopts a terminator fold under these growth conditions.

Given that the terminator represents a less energetically favorable conformation than the antiterminator, we aimed to investigate the mechanism by which the terminator fold is stabilized under active growth in nutrient-rich conditions. The transition between two alternative RNA conformations (or conformational switching) can be triggered by the binding of a trans-acting factor (i.e., a small molecule, an sRNA or a protein) to a single-stranded RNA (ssRNA) segment [24,26,27]. Therefore, we first sought to determine if any of the ssRNA loops of Rli51 were essential to bring about PTT (Figure 3A, green and red font type). For this purpose, we created a series of translational fusions of rli51 and mpl-egfp under the control of the rli51 promoter. The constructs included rli51 mutant versions with deletions in each of these ssRNA loops (Figure 3B, top panel). Elimination of the predicted terminator of Rli51 (Figure 3A, blue font) allowed for a complete Rli51-mpl transcriptional read-through (Figure 3B, Δter). The fact that deleting the predicted Rli51 terminator abolished PTT experimentally endorses the notion that Rli51 actually folds as a terminator when growing in BHI. In contrast, deletion of the 30-nucleotide-long segment that links Rli51 to mpl had no effect on Rli51 read-through into mpl (Figure 3B, Rli51 *), which is consistent with our previous observation (Figure 2B). Deletion of ssRNA loops 1 and 2 resulted in a 50 and 100% transcriptional read-through, respectively (Figure 3B, ΔL1 and ΔL2), indicating that these regions are important to mediate PTT in bacteria grown in BHI. Interestingly, the loss of PTT could not be attributed to a polar effect on Rli51 conformation resulting from L1 and L2 deletions because, in both constructs, the predicted terminator was not distorted (Supplementary Figure S1). These results are consistent with the participation of ssRNA regions L1 and L2 as binding sites of cellular trans-acting factors to regulate PTT in exponentially growing bacteria.

We hypothesized that Rli51-mediated PTT could be triggered by the binding of regulatory sRNAs to L1 and/or L2 regions. To test this hypothesis, we first searched for sRNA regulators that potentially bind L1 or L2. We used IntaRNA [28] to predict interactions between Rli51 sRNA and all L. monocyotgenes EGD-e genes annotated as sRNAs in the Listeriomics database [29]. We carried out this analysis with the wild-type 121-nucleotide-long rli51 (WT) and the ΔL1 and ΔL2 mutant sequences. From the output results, we first performed a selection of the most reliable predicted interactions. To do so, we selected the top 25% predicted interactions obtained with Rli51 wild-type sequence—based on the interaction energy (ΔG)—that mapped to Rli51 regions L1 or L2. Then, we computed the difference in the ΔG in the ΔL1 and ΔL2 mutant sequences in this group of selected predicted interactions and ranked them based on this difference (Figure 3C; Supplementary Table S1). Using this approach, we identified that Rli54, Rli46, and Rli21 potentially bind the L2 region, Rli113, and Rli126 L1. Additionally, Rli123 was identified to have binding sites in both the L1 and L2 regions (Figure 3C).

To experimentally verify if the identified sRNAs interact with Rli51 in vivo, we performed an MS2 affinity purification (MAP) of Rli51-binding RNAs (Figure 3D, left panel). MAP is an RNA pull-down approach in living bacterial cells that relies on the high-affinity and specific interaction between the coat protein (CP) of the MS2 phage and a hairpin aptamer (MS2) from a viral RNA genome [30]. To accomplish this aim, we generated a chimeric construct consisting of two copies of the MS2 aptamer fused to rli51, which was cloned downstream with an inducible tet promoter and transformed into a Δrli51 background. The hybrid RNA (MS2-Rli51) was induced with anhydrotetracycline, and the bacterial lysates were subsequently incubated with a protein composed of the CP fused to the maltose-binding protein (MBP::MS2-CP). The MBP serves as an affinity tag to purify MS2-Rli51 together with its binders (Figure 3D, left panel) [31]. To assess whether the candidate sRNAs identified in silico actually bind Rli51, we analyzed their abundances in the pull-down fractions obtained from the strain expressing MS2-Rli51 and from a control without the MS2 hairpins (Rli51) (Figure 3D). Rli51 sRNA showed more than a 2-log₁₀ enrichment in MS2-Rli51 pull-down fractions compared to those obtained with the Rli51-expressing (without MS2 hairpins) control strain (Figure 3D, left Y-axis). Among the predicted Rli51-binding partners, Rli21, which was predicted to bind to L2 in Rli51 (Figure 3C; Supplementary Table S1), showed a six-fold enrichment in the MS2-Rli51 pull down compared to the control Rli51 pull down (Figure 3D, right Y-axis). Surprisingly, although Rli123 was predicted to have an extensive binding, spanning from L1 to L2, it did not show a clear enrichment in MS2-Rli51 pull down (Figure 3D, right axis). These results experimentally verified that Rli21 binds to Rli51 in exponentially growing L. monocytogenes. Overall, these results suggest that rli51-mpl PTT under active growth in laboratory conditions results from the stabilization of a transcriptional terminator triggered by the binding of a cellular trans-acting factor, such as Rli21.

2.4. Rli51 Is a Trans-Acting sRNA in Bacteria Infecting Epithelial Cells

We have observed that the induction of the rli51 promoter in intracellular L. monocytogenes releases Rli51-mediated PTT, allowing for transcriptional read-through into mpl (Figure 2A,C). Nevertheless, more than 50% of total Rli51 is still represented by the short 121-nucleotide-long sRNA version (Figure 2A). We wondered about the physiological role of Rli51 sRNA and hypothesized that it could function as a trans-acting sRNA in intracellular bacteria. To test this hypothesis, we sought to compare the transcriptome of wild-type EGD-e and Δrli51 in intracellular bacteria. To accomplish this aim, we extracted total RNA from bacteria collected from infected human epithelial JEG3 cells and performed RNAseq (Figure 4A). A total of 3447 features were detected in both EGD-e and ∆rli51 samples, from which 225 were significantly upregulated and 143 downregulated in intracellular ∆rli51 (p < 0.05; Supplementary Figure S2A,B and Supplementary Table S2). Principal component analysis allowed us to discriminate between wild-type and ∆rli51 samples on the basis of one single dimension, which captured 87.8% of the total variation (Supplementary Figure S2C).

Functional enrichment analysis identified GO terms related to nucleotide binding and metabolism for upregulated genes, while for the downregulated genes, the most significant terms were associated with translation, ribosomes, and response to iron (Supplementary Figure S3A and Supplementary Table S3). Consistently, analysis of the protein–protein interaction networks of the differentially expressed genes (DEGs) using the STRING database showed clustering with high confidence (STRING score higher than 0.700) (Supplementary Figure S3B and Supplementary Table S4). Thus, the loss of rli51 sRNA expression in intracellular bacteria results in a broad reshaping of the L. monocytogenes transcriptome, which suggests that the short sRNA version of Rli51 functions as trans-acting a regulator in this condition.

Comparative transcriptomics of intracellular ∆rli51 and EGD-e showed a consistent impact on gene expression; however, this alone is not enough to prove that Rli51 functions as a trans-acting post-transcriptional regulator. Direct regulation by trans-acting sRNAs entails imperfect base pairing with target mRNAs; therefore, we next sought to demonstrate that Rli51 interacts with target candidates. We used intaRNA to predict binding between Rli51 and the top DEGs (FDR < 0.05) in intracellular Δrli51 compared to EGD-e (Figure 4A,B; Supplementary Table S5). The analysis predicted extended interactions for the three most downregulated genes in intracellular Δrli51 (rli38, lmo2186, and lmo0937), and for the upregulated lmo0207 (orfZ). In addition, the predicted binding sites in Rli51 for these targets lay in critical ssRNA loops L1, L2, or both (Figure 4B; Supplementary Table S5). Interaction energies were quite varied, with the strongest interactions found in the L1 and/or L2 regions of Rli51 (Supplementary Table S5), suggesting a role for these regions also in trans-mediated regulation. Interestingly, according to the prediction, Rli51 could bind lmo0937 and lmo2186 on more than one binding site, suggesting that the net interaction energy might be even more favorable (Supplementary Table S5).

Finally, we carried out an in vivo Rli51 pull-down assay to verify the predicted interactions of Rli51 with target candidates experimentally. We selected the top three target candidates based on the magnitude of the fold expression change in the comparative intracellular transcriptomics of Δrli51 and EGD-e: Rli38, lmo0937, and lmo2186 (Figure 4A). MS2 affinity purification (MAP) followed by qPCR quantification indicated a specific enrichment of lmo0937 and lmo2186 RNAs in MS2-Rli51 pull downs compared to Rli51 control samples, while Rli38 showed a negligible change (Figure 4C). These results indicate that Rli51 interacts with lmo0937 and lmo2186 in vivo. The difference in the enrichment between lmo0937 and lmo2186 (Figure 4C) might be an indication of the stability of the interaction, which depends on multiple factors other than the predicted interaction energy. Overall, these results verify that Rli51 binds cellular RNAs whose expression is altered in the absence of Rli51, which proves that the short version of Rli51 sRNA functions as a trans-acting sRNA in intracellular bacteria.

3. Discussion

Rli51 is an sRNA specific from pathogenic Listeria species induced under infection-relevant conditions [7,8,9,23,32,33]. However, both the mechanism of action and the physiological role of Rli51 have remained largely unexplored. Here, we have found that although Rli51 belongs to the 5′UTR of mpl mRNA, it does not play any role in the upregulation of mpl under infection conditions. Instead, Rli51 mediates a premature termination of transcription (PTT) in bacteria actively growing in a rich medium at 37 °C, which results in the production of a short 121-nucleotide-long sRNA (Figure 5). Under these conditions, the transcriptional activity of the promoter is very low, which could allow Rli51 to fold as a transcriptional terminator with the assistance of trans-acting sRNAs, such as Rli21/RliI (Figure 5). By contrast, PrfA-dependent induction of rli51 in intracellular bacteria leads to an increased abundance of short Rli51 while, at the same time, enables a transcriptional read-through into mpl (Figure 5). Consequently, under infection-relevant conditions, Rli51 occurs in two similarly abundant RNA species: a 121-nucleotide-long sRNA and the 5′UTR of mpl mRNA. The accumulation of short Rli51 sRNA allows it to regulate target transcripts in intracellular bacteria, such as Lmo2186 and Lmo0937 (Figure 5). Thus, Rli51 can regulate gene expression through two different mechanisms: (i) as a cis-acting sRNA, where Rli51 would contribute to preventing leaky mpl transcription under non-infection conditions by mediating a conditional PTT, and (ii) as a trans-acting sRNA, where increased Rli51 levels in intracellular bacteria would contribute to fine tuning a pathogen’s response to the environment encountered within eukaryotic cell compartments.

RNA-mediated post-transcriptional control of genes involved in virulence is a common regulatory mechanism in bacterial pathogens [4,34], and in L. monocytogenes in particular [35]. Essential virulence factors in L. monocytogenes are post-transcriptionally regulated through their 5′UTRs [17]. For instance, the 5′UTR of hly, actA, and inlA is necessary to achieve maximal expression levels under infection-relevant conditions [15,20,36]. By contrast, in the case of the 5′UTR of mpl, i.e., Rli51, it functions as a transcriptional attenuator and does not have a relevant role in the induction of mpl in intracellular bacteria. The 5′UTR of prfA is also a negative regulator that acts as a temperature-sensing regulator, which allows for PrfA translation at 37 °C (the temperature of the host) while preventing it at lower temperatures [37]. In addition, the S-adenosylmethionine riboswitches in L. monocytogenes can release the sRNAs SreA and SreB, which, in turn, regulate prfA by binding to its 5’UTR, suggesting a link between nutrient availability and virulence [26]. Similarly, it would be interesting to identify the physiological signals that ultimately control Rli51 function and, thus, Mpl.

Transcriptional attenuation refers to a blockade imposed by a specific conformation of an RNA segment (referred to as terminator) that functions conditionally. In a particular condition, binding of trans-acting factors (such as the ribosome, a small molecule, an RNA-binding protein, or an sRNA) can stabilize the formation of a terminator hairpin that impedes transcriptional read-through into the downstream gene [38]. In addition, in Gram-negative bacteria, the Rho factor serves as a global attenuator that terminates transcription of specific genes conditionally; sRNA binding within 5′ UTR of those genes disrupts Rho-mediated termination, thus leading to transcriptional read-through [39]. Therefore, sRNA-mediated antitermination is a common mechanism of bacterial gene regulation. We show that Rli51 is predicted to fold as a transcriptional terminator and as an antiterminator; the latter being the most stable conformation. We experimentally verified that the sRNA Rli21 (RliI) binds Rli51, and the predicted binding site lies in loop L2, a single-stranded RNA region that is critical for Rli51 to mediate PTT. This suggests that Rli21 binding stabilizes the formation of the terminator fold in Rli51. Interestingly, Rli21/RliI expression does not change under infection-relevant conditions compared to growth in nutrient-rich broth [8]. Therefore, in conditions of low transcriptional activity of the Rli51 promoter, Rli21/RliI might saturate Rli51 molecules, while under Rli51-inducing conditions, the excess free Rli51 would fold as an antiterminator (Figure 5). Such a regulatory scheme exerted by Rli21, and/or additional trans-acting factors, could explain the co-existence of both Rli51 and Rli51-mpl in intracellular bacteria.

Rli51 mediates condition-dependent PTT, producing a 121 nt long sRNA that acts as a trans-acting regulator for other genes. This conclusion is supported by three observations: (i) the absence of Rli51 in intracellular Δrli51 bacteria altered the expression of a set of genes; (ii) we predicted binding of some of these candidate target transcripts to Rli51 within ssRNA loops L1 and L2; and (iii) we experimentally confirmed they bind Rli51 in vivo. Specifically, we verified that Rli51 binds and positively regulates the expression of Lmo2186. lmo2186 is transcribed as a bicistronic mRNA together with lmo2185. lmo2185 mRNA levels (Supplementary Table S2) are decreased in intracellular Δrli51 compared to EGD-e, suggesting that lmo2185 and lmo2186 are co-regulated by Rli51.

Lmo2185 and Lmo2186 were functionally annotated with the enriched GO term “response to iron” in our comparative transcriptomics (Supplementary Table S3). lmo2185 and lmo2186 encode the fur-regulated hemin-binding proteins Hbp1 and Hbp2, respectively, which are homologs of IsdA and IsdC in Staphylococcus aureus and function as a hemophore that scavenges ferric heme from the environment [40]. The IsdG family heme monooxygenase Lmo0484, which breaks down heme to release free iron [41], was also downregulated in intracellular Δrli51. Additionally, lmo0541 (fhud) and lmo1960 (fhuc), encoding two subunits of the iron siderophore ABC transporter FhuBCDG, were also downregulated in Δrli51 (Supplementary Table S2). These findings suggest that Rli51 plays a role in upregulating iron scavenging systems under iron-limiting conditions, such as those found in intracellular compartments during infection. Interestingly, envelope stress induced by excess heme causes the LhrC4 sRNA to bind to lmo2185, lmo2186, and lmo0484 transcripts, and, at least for lmo0484, this resulted in target transcript downregulation [42]. These results suggest that genes involved in heme uptake and degradation are post-transcriptionally regulated in response to both the excess and limitation of heme.

Our comparative transcriptomics revealed a consistent upregulation of genes belonging to LIPI-1: hly, mpl, actA, plcB, orfX, and orfZ (see lmo0202 to lmo0207 in Supplementary Table S2). mpl, actA, and plcB belong to the lecithinase operon and can be transcribed from the rli51 promoter as a single mRNA [14]. Consistently, the deletion of rli51 resulted in upregulated endogenous mpl or mpl::gfp translational fusions in intracellular bacteria, as determined by targeted transcript quantification by RT-qPCR (Figure 1A,C). Together, these results also suggest a cis-acting mechanism of regulation by Rli51 in intracellular bacteria. The transcriptional activity of the promoter is very high, so it seems that the impact of this negative regulation mechanism is limited during intracellular infection. On the other hand, hly, orfX, and orfZ are transcribed as monocistrons, which suggests that Rli51 could also regulate the expression of other virulence factors through a trans-acting mechanism. In fact, orfZ was the target candidate with the highest predicted interaction energy (Figure 4A and Supplementary Table S5). Future studies will help to ascertain if Rli51 has any role as a general repressor of the virulence program of L. monocytogenes.

Our study sheds light on the function and mechanism of regulation mediated by an sRNA from LIPI-1 but entails limitations. For example, although we have identified Rli21/RliI as an sRNA that binds Rli51 at loop L2, our results do not preclude the involvement of additional sRNAs or other trans-acting factors (such as a small molecule) in the regulation of Rli51-mediated PTT. On the other hand, the deletion of a single sRNA typically results in hard-to-detect subtle phenotypes with modest expression changes in DEGs [43]. Our intracellular comparative transcriptomics was not meant to analyze comprehensively Rli51 regulon in intracellular bacteria though, for which other approaches might have shown clearer expression changes in putative target transcripts [5,44]. The aim of [5,44] was to identify physiologically relevant Rli51 targets in intracellular bacteria so as to prove a trans-acting mechanism. Further studies will help to determine a more detailed view of the function of Rli51 in virulence.

Our work provides compelling evidence of the dual regulatory mechanism of Rli51, acting both in cis and in trans. Similarly, other sRNAs exhibit cis- and trans-acting regulatory mechanisms, influencing virulence [26,45], metabolism [46,47,48], and nutrient uptake [49], which suggests that this dual mechanism might be more common than previously anticipated [50]. In fact, global mapping of 3′ ends in different bacterial species revealed that pervasive conditional PTT populates the transcriptome with 5′UTR-derived sRNAs that might have regulatory functions [18,50,51]. As conditional PTT most often relies on riboswitches, these findings open a yet unexplored level of regulation that links the concentration of particular intracellular metabolites with the release of sRNA regulators [26], suggesting an exciting intracellular sensing and regulatory crosstalk to control bacterial homeostasis.

4. Materials and Methods

4.1. Bacterial Strains and Growth Conditions

The Listeria monocytogenes (L. monocytogenes) strain used here was isogenic to wild-type strain EGD-e (serotype 1/2a) [52]. The rli51-null mutant was generated by allelic replacement. Homology regions flanking the rli51 locus were generated by PCR and cloned one after another into the thermo-sensitive suicide integrative vector pMAD between the BamHI and KpnI restriction sites. Primer pairs used for the amplifications were pMAD-BamHI-AB-Fw with AB-StuI-CD-Rv and AB-StuI-CD-Fw with CD-KpnI-pMAD-Rv. Integration, excision, and screening steps were performed as described [53]. rli51 deletion was confirmed by Sanger sequencing.

Unless otherwise stated, L. monocytogenes strains were grown at 37 °C in brain heart infusion (BHI) broth without shaking. To prepare the cultures in the exponential phase of growth in gene expression analysis experiments, bacteria were allowed to reach an OD₆₀₀ = 0.2 and then diluted two consecutive times. Specifically, a single colony was inoculated in 10 mL of BHI and grown at 37 °C in standing conditions. When it reached an OD₆₀₀ = 0.2, bacteria were diluted with fresh pre-warmed BHI and grown for at least 10 generations such that an OD₆₀₀ = 0.2 is not exceeded. For inducible expression experiments, strains carrying constructs cloned in the tetracycline-inducible expression vector pRMC2 [54] were grown to the exponential phase (OD₆₀₀ = 0.2) (see above) and then exposed to anhydrotetracycline (final concentration: 0.05 µM or 0.1 µM) for 40 min. To transform plasmids into L. monocytogenes, 1 µg of plasmidic DNA was electroporated at 2.4 kV, 200 ohms, and 25 μF in electrocompetent EGD-e and selected on plates supplemented with 10 µg/mL chloramphenicol (pRMC2 derivatives) or 5 µg/mL erythromycin (pMAD derivatives).

4.2. Cloning Procedures

Cloning details for each specific construct made for this work are included in Supplementary Table S6. Cloning procedures were performed in the DH5α E. coli strain. Bacteria were routinely grown in Luria Bertani (LB) broth at 37 °C with shaking at 180 rpms, and strains bearing pRMC2 or pMAD derivatives were selected with 100 µg/mL ampicillin. Ligation reactions and pure plasmids were transformed into chemically compentent bacteria [53]. Q5 DNA polymerase (M0491S), T4 DNA ligase (M0202S), and restriction enzymes (KpnI: R3142S, SalI: R3138S, EcoRI: R3101S, BglII: R0144S, EcoRV: R0195S, DpnI: R0176S) were supplied by New England Biolabs. Kits for extracting plasmidic DNA (NucleoSpin Plasmid: 740588.50) and DNA from agarose gels (NucleoSpin Gel and PCR Clean-up: 740609.50) were supplied by Macherey Nagel. Rli51and MS2::Rli51 constructs and GFP translational fusions were cloned into pRMC2 plasmid [54]. The GFP coding sequence was obtained from pAD1-cGFP [55]. All constructs were assessed by Sanger sequencing.

4.3. Determination of L. monocytogenes Intracellular Proliferation (Gentamicin-Resistance Assay)

Human placental choriocarcinoma JEG-3 epithelial cells (ATCC HTB-36) and NRK-49F-9F (ATCC CRL-1570) rat fibroblasts were propagated in MEM and DMEM, respectively, supplemented with 10% FBS, 4 mM L-glutamine, and nonessential amino acids. Infections were carried on cells at an 80% confluence seeded on 24-well culture dishes. Prior to infection, L. monocytogenes strains were grown in BHI medium for 14 h at 37 °C in standing conditions. Bacteria were collected by centrifugation at 800× g for 1 min at room temperature and then resuspended in the corresponding cell culture medium. Bacterial suspension was added to the eukaryotic cells medium at a multiplicity of infection (MOI) of 10:1 (bacteria/eukaryotic cell). After a 60 min long co-incubation, non-internalized bacteria were removed by three consecutive washing steps with pre-warmed Hank’s Balanced Salt Solution (HBSS) (Gibco). To kill the remaining extracellular bacteria, a fresh medium containing 10 µg/mL gentamicin was added to infected cells. After incubation for 2 or 6 h in these conditions, the eukaryotic cells were lysed with 100 µL of a solution containing 1%TritonX-100–0.1%SDS for 5 min at room temperature. Serial dilutions of this suspension were spread onto BHI agar plates, and the colony-forming units (CFUs) were counted after overnight incubation at 37 °C. The intracellular proliferation rate was calculated by dividing the number of CFUs at 6 h post-infection by that at 2 h post-infection.

4.4. RNA Isolation from Cultured and Intracellular Bacteria

To prepare the samples for RNA extraction from bacteria grown in BHI broth (supplemented with 10 µg/mL chloramphenicol in pRMC2 plasmid-bearing strains), 0.2 volumes of an ice-cold solution containing 5% phenol and 95% ethanol were added to the culture and then kept for 30 min in ice to quench RNA metabolism. The suspension was scrapped off the plate and the bacteria collected by centrifugation at 9300× g at 4 °C for 10 min. Afterwards, the bacterial pellet was washed with 1 mL of an ice-cold solution containing 1% phenol and 19% ethanol diluted in 0.1% DEPC-treated water, transferred into a 1.5 mL tube, re-centrifuged at 9600× g for 5 min, and stored at −80 °C until used.

To obtain intracellular bacteria, human epithelial JEG-3 cells (ATCC HTB-36) were cultured with MEM supplemented with 10% FBS, 4 mM L-glutamine, and nonessential amino acids on 500- cm²-BioDish-XL plates (351040, BD Biosciences, Franklin Lakes, NJ, USA) to an ∼80% confluence (∼5.6 × 10⁷ cells per plate). A single colony was inoculated in 1.5 mL of BHI (supplemented with 10 µg/mL chloramphenicol in pRMC2 plasmid-bearing strains) in a 10 mL glass tube with a vented cap and grown overnight at 37 °C without shaking. This culture was spun down for 1 min at 800× g at room temperature and resuspended in a serum-free culture medium at the same cell density. For each intracellular bacterial sample, two 500 cm² plates were infected for 30 min at an MOI of 10:1 (bacteria/eukaryotic cell). Non-internalized bacteria were washed away with pre-warmed HBSS and killed by adding a fresh culture medium containing 100 µg/mL gentamicin. After a 2-h incubation, the infected cell culture was washed twice with pre-warmed HBSS, and a fresh medium supplemented with 10 µg/mL gentamycin was added. At 6 h after infection, the eukaryotic cell culture was washed five times with HBSS and lysed in a chilled solution containing 0.1% SDS, 1% acidic phenol, and 19% ethanol diluted in 0.1% DEPC-treated water for 30 min at 4 °C. The eukaryotic cell lysate was scrapped off the plate, and the intracellular bacteria were collected by centrifugation (27,500× g, 4 °C, 30 min). Bacteria were washed with the same lysis solution and re-pelleted (15,500× g, 4 °C, 10 min), and the pellet was stored at −80 °C.

Frozen pellets from bacteria cultured in BHI (total amount 3 ODs: 15 mL at an OD₆₀₀ = 0.2) and from intracellular bacteria (obtained from two 500 cm² plates of infected JEG-3 cells) were thawed, resuspended in a freshly prepared solution containing 10% glucose, 12.5 mM Tris-HCl pH 7.6, and 65 mM EDTA. Resuspended cells were transferred to a tube containing 1 volume of acidic phenol. This cell suspension was beaten with acid-washed glass beads (0.4 g ≈ ca. 0.2 mL) twice for 45-s-long cycles at 6 m/s in a FastPrep 24 (MP Biomedicals). As a centrifugation step to pellet unsoluble material and separate the phenolic phase (14,000× g, 4 °C, 10 min), RNA was extracted from the aqueous phase with 1 mL of NZYzol (MB18501, Nzytech, Lisboa, Portugal) followed by two consecutive extraction steps with 0.5 mL of chloroform and then precipitated with isopropanol for 30 min at room temperature. RNA yield was quantified by spectrophotometry with a nanodrop ND2000 (Thermo Fisher Sci., Waltham, MA, USA), and 5 µg of RNA was treated with 3 U of DNase I (Turbo DNA-free kit: AM1907, Thermo Fisher Sci.) for 60 min at 37 °C. RNA integrity was assessed by running a 200 ng sample of the extracted RNA through agarose TAE electrophoresis. Genomic DNA carryover contamination-free RNA was verified by a negative result in a 30-cycle PCR using rpoB-F and rpoB-R primers (Supplementary Table S7) and 20 ng of each RNA sample.

4.5. Reverse Transcription and Real-Time Quantitative PCR (qPCR)

cDNA samples were prepared from 1 µg of total DNA-free RNA using the Applied Biosystems High Capacity cDNA Reverse Transcription Kit (4368814, Thermo Fisher Sci.) including a mix of random hexamers. The RT-PCR protocol was 10 min at 25 °C, 120 min at 37 °C, and 5 min at 85 °C. qPCR was performed in a 10 µL final reaction volume, including 1 ng of cDNA as a template, 500 nM of each target gene-specific primer (Supplementary Table S7), and 5 µL of the Applied Biosystems Power SYBR Green PCR Master Mix (4367659, Thermo Fisher Sci.). qPCR runs were performed in an Applied Biosystems 7500 Real-Time PCR System instrument (Thermo Fisher Sci.) using a standard amplification protocol (10 min at 95 °C; 45 cycles of 15 s at 95 °C and 1 min at 60 °C; dissociation curve of 15 s at 95 °C, 1 min at 60 °C, and a progressive temperature increase until 95 °C). Target-specific primers (listed in Supplementary Table S7) were designed using Applied Biosystems Primer Express 3.0 (Thermo Fisher Sci.). As a quality control of each primer pair, the amplification efficiency of the corresponding qPCR assay was determined by running a standard curve comprising 12.5 ng of a pool of cDNA samples and five consecutive 1/5 serial dilution points. For each sample, a reverse transcription-minus control including 80 ng of RNA was analyzed with the same primer set to assess the extent of genomic DNA contamination.

For absolute RNA quantification, every plate included a standard curve made up of 5 points represented by serial dilutions of an EGD-e genomic DNA sample (12.5, 2.5, 0.5, 0.1, and 0.02 ng of gDNA). To calculate the number of molecules of target genes in each point of the standard curve, we considered a molecular weight of EGD-e gDNA of 1.82 × 10⁹ g/mol (gDNA length 2.94 × 10⁶ bps, average MW per nucleotide in dsDNA 619.96 g/mol) and the Avogadro No. 6.023 × 10²³ molecules/mol. The fluorescence threshold was set such that it crossed the amplification curves (cycle vs. ΔRn) of all concentration points in the standard curve within the linear region of the amplification plot (exponential phase). Each experimental cDNA sample was run in triplicate, and the corresponding Ct was calculated as the mean of the three determinations. Individual technical replicates that deviated more than 0.75 cycles from the other two ones were not considered to calculate the mean Ct value. The No. of molecules of target mRNA transcripts in each experimental sample was calculated by interpolation of the corresponding mean-Ct values in the standard curve plot (Log₁₀ [No. of molecules] vs. Ct). Absolute quantification of each target mRNA transcript was expressed as the No. of molecules per nanogram of total extracted RNA in each sample. The relative quantification of target mRNAs was calculated by dividing its absolute quantification by that of the rpoB transcript. Enrichment ratios in RNA pull-down experiments were calculated by dividing the relative quantification of each specific RNA in MS2-Rli51 eluate by that in the eluates of Rli51 control pull downs. rpoB was used as a housekeeping gene. Every qPCR assay included a non-template control (NTC), which included the same volume of H₂O instead of the cDNA template.

4.6. Production of Recombinant His-MBP-CP

A 500 mL culture of a BL21 E. coli strain with pMAL-C2::His-MBP-CP [56] was grown to OD₆₀₀ = 0.6 in LB supplemented with 10 mM glucose and 100 µg/mL ampicillin (final volume = 500 mL). His-MBP-CP ectopic expression was induced for 2 h after the addition of IPTG (final concentration: 0.3 mM). Afterwards, bacteria were collected by centrifugation (4000× g, 4 °C, 20 min), and the pellet was frozen for 24 h at −20 °C. Bacteria were thawed, resuspended in 10 mL of a lysis buffer (50 mM Na₂HPO₄-NaH₂PO₄ pH = 8, 300 mM NaCl, 1% triton X-100, 10 mM β-mercaptoethanol, 10% glycerol, and 1× Roche EDTA-free complete protease inhibitor cocktail [04693116001, Merck, Rahway, NJ, USA]) per gram of bacterial wet weight and run through four consecutive freeze and thaw cycles in a dry ice/ethanol bath. The suspension was then sonicated (15 pulses of 15 s with 50% amplitude with 1-min-long breaks in ice in between) in a Q500 sonicator with a CL334 converter (QSonica, Newtown, CT, USA), and the lysate was clarified by centrifugation (9000× g, 4 °C, 30 min). To purify His-MBP-CP, 3 mL (bed volume) of TALON^® Metal Affinity Resin (635503, Takara, San Jose, CA, USA) were equilibrated by incubating the resin in batch with 6 volumes of washing buffer (lysis buffer without glycerol and protease inhibitors) for 3 times. The equilibrated resin was incubated with the clarified lysate at a volume ratio of 1/10 (resin/lysate) on a wheel for 2 h at 4 °C, washed 6 times in batches with 6 volumes of washing buffer, and then eluted with washing buffer supplemented with 220 mM imidazole (12 fractions, 250 µL each). Protein concentration was determined with the Pierce BCA reagent (23225, Thermo Fisher Sci.), and the protein-containing fractions were pooled (ca. total volume: 1 mL) and dialyzed against 600 mL of RLN buffer (50 mM Na₂HPO₄-NaH₂PO₄ pH = 7.4, 140 mM NaCl, 1.5 mM MgCl₂, 1 mM DTT) at 4 °C for 16 h.

4.7. MS2-Affinity Purification of Rli51-Binding RNAs (RNA Pull Down)

To prepare bacterial lysates, L. monocytogenes EGD-e strains Δrli51:pRMC2-rli51-MS2 and Δrli51:pRMC2-Rli51 were grown to exponential phases (final volume of 600 mL, OD₆₀₀ = 0.12) in BHI supplemented with 10 µg/mL chloramphenicol. Then, Rli51 and MS2-Rli51 expression was induced by adding anhydrotetracycline (ATc; final concentration: 0.2 µM). When the cultures reached an OD₆₀₀ = 0.2 (ca. after a 40-min-long incubation with ATc), bacteria were collected by centrifugation (8500× g, 4 °C, 10 min), washed with column buffer (20 mM Tris-HCl pH 7.4, 200 mM NaCl, 1 mM EDTA, 1 mM DTT), and frozen at −80 °C for 24 h. The pellet was resuspended in 1 mL of column buffer, subjected to three consecutive freeze and thaw cycles in a dry ice/ethanol bath, and beaten with 0.4 g (ca. 0.2 mL) of acid-washed glass beads for three 20-s-long cycles at 2 m/s in a FastPrep 24 (MP Biomedicals, Auckland, New Zealand). After removing the beads, 4 µL of the Ambion™ RNase Inhibitor (cloned, 40 U/µL; AM2684, Thermo Fisher Sci.) were added to the lysate and then clarified by centrifugation (17,500× g, 4 °C, 30 min).

To prepare the affinity resin for MS2-tagged RNA pull-down experiments, the recombinant purified His-MBP-CP protein was first bound to amylose resin. A total of 500 µL (bed volume) of amylose resin (E8021S, New England Biolabs, Ipswich, MA, USA) was equilibrated in column buffer (6 resin volumes, 5 min, 3 times). This equilibrated resin sample was incubated in a batch with 10 volumes of column buffer containing 0.75 mg/mL of His-MBP-CP on the wheel (4 °C, 2 h). The resin was washed 4 times with 6 resin volumes of column buffer and stored at 4 °C until used (≤5 days).

To conduct the RNA pull down, a 100 µL aliquot (bed volume) of the amylose resin bound to His-MBP-CP (ca. 0.715 mg of His-MBP-CP) was mixed with 1 mL of the bacterial lysate and let to incubate for 2 h on the wheel at 4 °C. The resin was washed 4 times with 6 volumes of column buffer and subsequently transferred to a disposable empty Poly-Prep^® chromatography column (7311550, Bio-Rad, Hercules, CA, USA). Elution was performed with 3 volumes of the column buffer supplemented with 10 mM maltose. The fractions were pooled (total volume: 300 µL), and the RNA was extracted with NZYol (see above).

4.8. PASIFIC Structural Analysis of Rli51

In silico analysis to predict the occurrence of alternative terminator/antiterminator RNA conformation compatible with a riboswitch or an attenuator was carried out using the PASIFIC webserver (https://www.weizmann.ac.il/molgen/Sorek/PASIFIC/; accessed on 16 March 2022) [25]. A 128-nucleotide-long sequence starting from the transcriptional start site of rli51 was selected as the input. This included the 121 nucleotides annotated as the Rli51 transcript plus a U-rich 7-nucleotide-long downstream sequence (AGUUUUA). Default parameters optimized to detect attenuators were used.

4.9. sRNA-sRNA Interaction Prediction

In silico prediction of sRNA-Rli51 pairwise interactions was performed by conducting multiple one-to-one interaction analyses of each sRNA with Rli51 using the command-line version of IntaRNA 2.0 [28,57]. A batch analysis was run using a manually created L. monocytogenes sRNAs list file in FASTA format as the query and the Rli51 sequence as the target. sRNA sequences were obtained from the Listeriomics database [29].

4.10. RNA Sequencing: Library Construction and Next-Generation DNA Sequencing

Library construction and DNA sequencing were performed by Novogene. Prokaryotic RNA samples were depleted of ribosomal RNA and purified by ethanol precipitation. After fragmentation, the first strand of cDNA was synthesized using random hexamer primers. During the second strand cDNA synthesis, dUTPs were replaced with dTTPs in the reaction buffer. The directional library was ready after end repair, A-tailing, adapter ligation, size selection, USER enzyme digestion, amplification, and purification. The library was checked with Qubit and real-time PCR for quantification and a bioanalyzer for size distribution detection. Quantified libraries will be pooled and sequenced on the Illumina platform Novaseq6000, according to effective library concentration and data amount required.

4.11. RNA Sequencing Data Analysis

For data analysis, raw reads were curated by trimming the adapters and removing poly-N sequences and low-quality reads. Only high-quality data were used for downstream analysis. Bowtie2 software [58] was used for indexing the reference genome (NC_003210.1) and mapping the reads. Reads mapping to each gene were counted by FeatureCounts and used to calculate the FPKM for estimating the expression levels of each gene [59]. Principal component analysis (PCA) using these expression levels was conducted using the FactoMineR R package to determine the overall reproducibility of our RNAseq biological replicates. Differential expression analysis between EDG-e (control) and ∆rli51 sample groups was performed by the DeSEQ2 R package based on the negative binomial distribution, and p-values were adjusted by the Benjamini–Hochberg correction to control the false discovery rate. Only those features with a p-value < 0.05 were considered significant. A heatmap of the normalized expression levels of each significant gene was visualized using Morpheus from the Broad Institute (https://software.broadinstitute.org/morpheus; accessed 20 November 2023). For normalization, each gene FPKM was normalized by the average of the FPKM of that gene in all samples and then transformed by log2. The ClusterProfiler R package was used to perform statistical enrichment of differential expression genes based on GO terms and known regulons of L. monocytogenes available in curated databases: PRODORIC (http://www.prodoric.de/; accessed 5 February 2021), RegPrecise (https://regprecise.lbl.gov/index.jsp; accessed 5 February 2021), and CollectTF (http://www.collectf.org/browse/home/; accessed 5 February 2021). Protein–protein interaction (ppi) networks were built by searching the differential expression genes in the STRING database (http://string-db.org/; accessed 20 November 2023).

5. Conclusions

• Rli51 is an sRNA located in the 5′UTR of the LIPI-1 virulence factor mpl that functions both as a cis- and trans-acting sRNA.
• Under nutrient-rich laboratory conditions, the transcription of rli51-mpl is prematurely terminated, producing a 121-nucleotide-long sRNA at very low levels.
• The sRNA Rli21/RliI binds to a single-stranded RNA loop in Rli51, which is crucial for premature transcription termination.
• Under intracellular infection conditions, rli51 is induced, leading to a higher abundance of the short Rli51 sRNA and allowing for transcriptional read-through into mpl.
• Transcriptional read-through allows Mpl virulence factor expression in intracellular bacteria and the short Rli51 sRNA regulates in trans transcripts encoding iron scavenging and cell surface proteins.