Small non-coding RNAs that modulate protein activity are non-base-pairing molecules and can interact directly with proteins to influence their activity by sequestration (Michaux et al., 2014). The system understood in this context is the global carbon storage regulator in the CsrA/RsmA family (Carbon storage regulator/Regulator of secondary metabolism; Lenz et al., 2005).

CsrA/RsmA proteins are global regulators of gene expression at a post-transcription level and play an important role in the expression of virulence factors in several proteobacterial pathogens (Babitzke and Romeo, 2007; Vakulskas et al., 2015). CsrA binds to mRNA, commonly inhibiting translation by blocking the ribosome binding site (RBS). This leads to the rapid degradation of their target mRNA and is fundamental to the regulation of some specific virulence systems required for host infection (Vakulskas et al., 2015). The activity of CsrA is regulated by sRNAs of the CsrB family (100 to more than 400 in length). These sRNA remove this protein from the target mRNAs by binding to different sites (Romeo et al., 2013). These sRNAs indirectly activate genes suppressed by CsrA, antagonizing the activity of this protein, and can be found in multiple copies per genome, performing redundant functions (Papenfort and Vanderpool, 2015).

Small non-coding RNAs that act in trans (sRNA-trans) are between 50 and 300 nt in length; they are imperfectly complementary to one or more regions of the target mRNA, which is coded in another region of the genome. sRNA-trans participate in the response of the cell to changes in environmental conditions and are also involved in the regulation of virulence in pathogenic species (Malecka et al., 2015). The general mechanism of sRNA-trans is to occupy the RBS of a target mRNA by base pairing with the Shine-Dalgarno (SD) sequence or the start codon, although they can also interact with the mRNA coding sequence (Feng et al., 2015). As they are only partially complementary with their target, sRNA-trans generally require a chaperone, Hfq, to facilitate RNA-RNA interactions due to limited complementarity between the sRNA-target mRNA and to strengthen its regulation (Wagner, 2013; Van Puyvelde et al., 2015). Hfq belongs to the group of Sm-like (LSm) proteins which are also found in eukaryotes and archaea (Vogel and Luisi, 2011; Papenfort and Vanderpool, 2015). It may actively remodel the RNAs to melt inhibitory secondary structures and also may serve passively as a platform to allow sRNAs and mRNAs to sample potential complementarity (Waters and Storz, 2009). Besides, Hfq is frequently required to protect the sRNAs from cellular ribonucleases such as RNase E (Papenfort and Vanderpool, 2015), and for many intracellular and extracellular pathogens it is known that Hfq deficiency dramatically affects virulence and fitness (Escherichia coli, Kulesus et al., 2008; V. alginolyticus, Deng et al., 2016; Klebsiella pneumoniae, Chiang et al., 2011; Pseudomonas aeruginosa, Sonnleitner et al., 2003; Bordetella Pertussis, Bibova et al., 2013).

In case of sRNA-trans, a single mRNA can be regulated by more than one sRNA (Bardill and Hammer, 2012; Duval et al., 2014; Chang et al., 2015). This type of sRNA has been more widely studied and is the most abundant type of sRNA discovered to date. An example of this group are the sRNAs called Qrr (Quorum Sensing Regulatory RNA) that are involved in the regulation of bacterial quorum sensing (QS; Silveira et al., 2010). These Qrr sRNAs regulate target mRNAs either positively or negatively affecting translation, stability and/or processing of the target mRNA (Bardill and Hammer, 2012; Feng et al., 2015; Van Puyvelde et al., 2015). In the case of negative regulation, the sRNA-trans is paired close to the RBS of the target mRNA, blocking translation. In many cases this generates recruitment of RNase E and subsequent degradation of the mRNA. In positive regulation, the coded sRNA-trans performs an anti-antisense base-pairing in the 5′ mRNA region. The ribosome binding site is thus revealed, promoting the stabilization of the target mRNA and hence translation. In this case, Hfq binds with both sRNA-trans and its corresponding mRNA target mediates the interaction (Fröhlich and Vogel, 2009; Feng et al., 2015). This can decrease the constant of apparent dissociation between the sRNA and its target mRNA and stabilize the sRNAs, protecting them from degradation by the RNase E (Bardill and Hammer, 2012; Duval et al., 2014).

The regulator elements of 5′UTR (untranslated regions) are sequences that act in cis in the mRNA and have gained special attention in recent years. This group of elements is associated with the detection of a wide range of signals, including temperature (so-called thermoswitches; Nguyen and Jacq, 2014) and riboswitches (Smith and Strobel, 2011). The latter control gene expression by direct binding to ions or small molecules such as metabolites, co-enzymes, amino acids and nucleotide bases, causing changes in the secondary structure of the transcript (Cressina et al., 2011; Furukawa et al., 2012).

Small non-coding RNA-cis are complementary to their target mRNA, and thus can interact autonomously. Their size also varies from approximately 100 to more than 300 nt, and may overlap with the 5′ or 3′ ends or be in a middle zone of the gene coded opposite the sRNA (Chang et al., 2015). These sRNAs are located in the antisense strand of their target RNA regions and their mechanism of action leads to the post-transcriptional down-regulation of the target gene through perfect extended duplex (Oliva et al., 2015).

In terms of their mechanisms of action, it has been proposed that sRNAs-cis may have several advantages over their homologs that act in trans. Full and lengthy complementarity between the sRNA and its target, observed in sRNA-cis, can lead to very stable duplexes, allowing regulation that is independent of proteins such as Hfq (Chang et al., 2015). Also, the cis position has the effect that sRNA and target are transcribed in close proximity, which can facilitate and improve interactions between the regulator molecules and their objective (Georg and Hess, 2011; Chang et al., 2015).

CRISPR/Cas systems provide an adaptive immune mechanism in bacteria and archaea (Barrangou et al., 2007; Westra et al., 2014) which recognizes and degrades exogenous nucleic acid belonging to viruses and plasmids that invade the cell (Rath et al., 2015). The cleavage of foreign nucleic acid occurs through Cas proteins specifically guided by sRNA of approximately 30 nt (Makarova et al., 2011). These guide sRNAs, called CRISPR RNAs (crRNAs), are found in a specific locus where each DNA sequence is flanked by a palindromic repeated sequence that participates in its genetic expression (Makarova et al., 2011; Rath et al., 2015). The immunological memory against future encounters with foreign nucleic acid is due to the incorporation of fragments of exogenous genetic material into the genome of the bacterium (Plagens et al., 2015; Rath et al., 2015). These systems are very diverse in terms of number, genetic structure and mechanism of action of Cas proteins (Makarova et al., 2011). Makarova et al. (2011) proposed a new classification of CRISPR/Cas systems of five types (types I–V) divided into two major classes. Class 1 systems have multi-subunit crRNA–effector complexes, whereas Class 2 system functions are carried out by a single subunit crRNA–effector module (Makarova et al., 2011).

Small non-coding RNAs can modulate the expression of genes associated with survival and virulence by pairing with untranslated regions of genes in response to cell density, through the QS pathway (Wen et al., 2016). QS is a process of cell communication by which a bacterial population expresses a set of genes in a coordinated way in response to the cell density in the population (Hammer and Bassler, 2007; Shao and Bassler, 2012, 2014; Van Kessel et al., 2015) through the production, release, detection and response to extracellular signal molecules called autoinducers (Tu and Bassler, 2007; Rutherford et al., 2011).

QS systems are widely distributed in Vibrio species (Zhang et al., 2012); the Qrr sRNA regulate multiple mRNA targets that code for QS regulatory components (Feng et al., 2015) through base-pairing with target mRNAs (Lenz et al., 2004; Svenningsen et al., 2008; Shao et al., 2013). These Qrr sRNAs appear in more than one copy per genome (Papenfort and Vanderpool, 2015). Qrr sRNAs activate the translation of AphA, a key regulator of low cell density (LCD), and suppress the translation of LuxR (V. harveyi)/ HapR (V. cholerae)/ SmcR (V. vulnificus)/ LitR (V. fischeri)/ ValR (V. alginolyticus) and OpaR (V. parahaemolyticus), which is the main regulator of high cell density (HCD; Rutherford et al., 2011; Shao et al., 2013; Van Kessel et al., 2013b). Qrrs are highly conserved at the nucleotide level among Vibrio species but can vary in number and mechanism. For example, the five Qrr (Qrr1-5) of V. harveyi are additive because all of them are required to maintain wild-type-like repression of its master transcriptional regulator. Conversely, the four Qrr (Qrr1-4) of V. cholerae sRNAs are functionally redundant because any of its Qrr is sufficient to repress its master transcriptional regulator (Svenningsen et al., 2008). V. vulnificus has Qrr1-5 sRNAs that function additively, not redundantly, to repress SmcR (Wen et al., 2016). The only Qrr described in V. fischeri represses expression of the master regulator LitR (Kimbrough and Stabb, 2016) while in V. parahaemolyticus just three Qrr (Qrr2-4) sRNAs are known (Zhang et al., 2012). However, this does not imply that other Qrr could exist in these species. New Qrrs could be still discovered because of the large number of sRNA described each year.

Qrr sRNAs also give feedback to suppress genes that code one of the QS synthase-receptor pairs, LuxMN, and the gene that codes the LuxO transcription factor (Zhang et al., 2012; Feng et al., 2015). Phosphorylated LuxO activates the transcription of the Qrr sRNAs, which bind and destabilize the mRNA hapR/luxR/smcR/opaR as a function of Hfq (Tsou et al., 2011). However, the role of Qrr sRNA is not limited to QS, since they also regulate genes outside the QS circuit post-transcriptionally (Feng et al., 2015).

Although V. harveyi is not a human pathogen, is the species for which the QS process has been best characterized and has been considered as a general model for other Vibrio species, including human pathogens. In V. harveyi has been shown that Qrr3 controls its target mRNA through four different mechanisms, showing that a single sRNA molecule can play an important role in the regulation of its different targets by different mechanisms within the same bacterium, highlighting the versatility of these regulatory systems. In fact, Qrr3 sRNA suppresses luxR by catalytic degradation and LuxO by sequestering, in a process that does not lead to the degradation of Qrr. It also suppresses luxM by coupled degradation and activates aphA by revealing the ribosome binding site, while at the same time the sRNA is degraded (Feng et al., 2015). The activation of the translation of AphA (Rutherford et al., 2011; Van Kessel et al., 2013a) indirectly activates the transcription of virulence and biofilm genes, which promotes bacterial colonization and infection (Wang et al., 2014), so an important function of sRNAs in Vibrios is the link between QS and virulence (Carignan et al., 2016). Also, HapR/LuxR/SmcR/OpaR regulates the expression of genes involved in the synthesis and degradation of c-di-GMP reducing the levels of this second messenger that controls the switch between biofilm formation and motility and other fundamental bacterial behavior including fimbrial synthesis, T3SS and RNA modulation (Srivastava and Waters, 2012; Wang et al., 2014; Sengupta et al., 2016).

.

In V. cholerae, sRNAs are critical components of the QS system and they constitute the first described example of sRNAs that regulate the expression of virulence genes in this pathogen (Tsou et al., 2011). There are sRNAs that participate directly in the expression of ToxT or are in turn controlled by ToxT, which is the main regulatory protein that controls the transcription of the genes ctxAB and tcpA-F, considered the main virulence factor genes of this pathogen (Carignan et al., 2016) (Figure 1). sRNAs that are members of the Qrr regulon indirectly control the production of the virulence factors TCP and CT through regulation of AphA and HapR (Bardill and Hammer, 2012; Shao and Bassler, 2014). HapR negatively regulates the expression of aphA and directly suppresses biofilm formation by binding to the promoter of the gene vpsT, which codes a regulator of the exopolysaccharide production cluster vps, and indirectly through aphA, since AphA is a positive regulator of VpsT (Teschler et al., 2015; Carignan et al., 2016) (Figure 1). HapR also suppresses the transcription of vpsR, which codes another transcription factor that also activates genes involved in biofilm formation and integrates signals from at least three different QS systems, which converge to the regulators of the LuxO response (Wang et al., 2014; Vakulskas et al., 2015) (Figure 1).

Notably, strains of V. cholerae modified to not produce Qrr sRNAs do not produce TCP or CT, and their capacity to colonize a lactating mouse model is severely lessened. HapR also positively regulates the transcription of hapA, a gene that codes the protease hemaglutinin (HA/protease), which is a virulence factor involved in the late stage of infection by V. cholerae, and the gene prtV, which codes a metaloprotease that is both a virulence factor against the human host and a protection factor against predators in the environment (Lutz et al., 2013). This system is reinforced by a mutual suppression in which HapR suppresses the transcription of aphA and AphA suppresses the transcription of hapR. Therefore, HCD induce the QS systems to reprogram gene expression, which stops the production of virulence factors and increases the production of transmission factors such as motility (Vakulskas et al., 2015).

Qrr sRNAs (Qrr1-4) act redundantly, and therefore the suppression of all four is required to avoid fully the repression of QS mediated by Hfq of genes downstream from HapR (Tsou et al., 2011). Moreover, in the absence of one or several Qrrs the others are overexpressed, compensating for the regulatory function. Svenningsen et al. (2008) observed that in V. cholerae HapR can activate the transcription of qrr genes, creating a negative feedback loop in the QS circuit. Therefore, HapR directly and indirectly suppresses its own production. Direct suppression of the promoter hapR by HapR only occurs at HCD, and it avoids overproduction of HapR at LCD. On the other hand, post-transcriptional suppression of hapR by the Qrr sRNA feedback loop requires the presence of LuxO-P and HapR, and therefore it only occurs in the transition from HCD to LCD. The latter feedback loop drastically accelerates the transition of V. cholerae cells from the social mode to the individual cell mode (Svenningsen et al., 2008).

Another case is the sRNA VqmR, which was described by Papenfort et al. (2015) in V. cholerae through a differential RNA-seq. VqmR regulates biofilm formation by suppression of vpsT, which is controlled by QS through HapR, revealing that VqmR represents a previously unknown link between biofilm formation and QS. VqmR suppresses the expression of multiple mRNAs, including those that code for the toxin Rtx, and vpsT, which is required for biofilm formation, as demonstrated by the researchers through mutagenesis and microarray analysis (Papenfort et al., 2015). Shao and Bassler (2014) showed that LuxO functions through the sRNAs involved in QS to suppress T6SS, a new virulence factor in V. cholerae (Shao and Bassler, 2014; Cheng et al., 2015), using two mechanisms: the Qrr sRNAs suppress the production of T6SS-activator HapR, thus decreasing the expression; independently of HapR, the Qrr sRNAs suppress T6SS by direct base-pairing with the mRNA that codes the long T6SS cluster. Regulation of T6SS Qrr sRNA is also conserved in pandemic and non-pandemic strains of V. cholerae (Shao and Bassler, 2014).

There is also a relation between virulence and QS in other human pathogen Vibrios. The expression of virulence factors in V. parahaemolyticus is also modulated by QS, allowing differential genetic regulation under LCD and HCD conditions. OpaR controls opacity, biofilm formation and motility, represses T3SS1 regulons and oppositely, regulates the two T6SS in this bacterium (Gode-Potratz and McCarter, 2011; Burke et al., 2015). This QS-dependent regulation of T6SS is crucial as this system plays an important role in pandemic and non-pandemic strains of V. parahaemolyticus (Ceccarelli et al., 2013). Similarly, Qrr sRNAs affect the expression of pathogenicity genes in V. vulnificus by modulating the expression of the smcR (Figure 2). The quorum-sensing pathway is similar to V. cholerae, and SmcR is responsible for regulating the expression of diverse virulence factors (Wen et al., 2016). smcR mutants show attenuated cytotoxicity compared to the wild type (Lee et al., 2007). Wen et al. (2016) showed that Qrr sRNAs in V. vulnificus are regulated via LuxO and modulate the expression of virulence factors via SmcR. Interestingly, the authors also showed that Qrr sRNAs are responsive to iron concentration, and that iron and QS converge on Qrrs to control the expression of virulence factors (Wen et al., 2016) (Figure 2).

The CsrA/RsmA pathway has been identified in several γ-proteobacterial species, in which it regulates a wide range of cellular physiology: regulation of carbon metabolism, motility, biofilm formation, production of secondary metabolites and QS, thus affecting virulence (Babitzke and Romeo, 2007).

Maturation and dispersal of biofilms in V. cholerae, as well as CT and TCP production, are controlled by multiple systems independent of QS whose effects converge on a single regulator circuit (Rutherford and Bassler, 2012; Mey et al., 2015). CsrA influences the expression of virulence factors and QS through this circuit, connecting the response to autoinducers CAI-1 and AI-2 (Mey et al., 2015) (Figure 1). VarS/VarA (Virulence Associated Regulator) regulate QS indirectly through CsrA and three CsrA-inhibiting sRNAs: the aforementioned CsrB and also CsrC and CsrD (Lenz et al., 2005) (Figure 1). The response regulator VarA has been identified as a virulence regulator in this bacterium (Tsou et al., 2011). VarS/VarA has also been found to be involved in the control of HapR expression through LuxO and in the control of the CqsS and LuxPQ systems (Figure 1). Lenz et al. (2004) observed that Hfq is required for the regulation of QS due to the dependent chain of LuxO and that hfq mutation affects the stability of the hapR messenger. Therefore, the VarS/VarA-CsrA/CsrBCD system converges with the QS system in V. cholerae to regulate the expression of the Qrr sRNAs and thus the entire QS regulon (Lenz et al., 2004). All these pathways converge to HapR, which thus represents a central node in this regulation network (Figure 1).

A fundamental element of homeostasis in microorganisms is iron, which is a cofactor of a large number of enzymes involved in essential biological processes (such as photosynthesis, respiration, DNA biosynthesis, etc.; Mey et al., 2005b; Repoila and Darfeuille, 2009; Porcheron and Dozois, 2015). However, excess free intracellular iron will give rise to reactions that generate toxic oxygen-reactive species (Nguyen and Jacq, 2014). Therefore internal iron levels must be rigorously controlled in accordance with physiological requirements, thus avoiding toxic effects.

Bacteria respond to fluctuations in iron concentration by coordinating the expression of genes that code for proteins involved in iron transport and storage, and genes that code for iron-dependent enzymes (Repoila and Darfeuille, 2009; Nguyen and Jacq, 2014). In Gram-negative bacteria, the main repressor of iron uptake systems is the ferric uptake regulation (Fur) protein. Fur complexed with Fe²⁺ binds to DNA (Bagg and Neilands, 1987), turning off the transcription of the iron uptake genes and limiting the entry of excess iron into the cell. When iron is limited in the cell, Fur is inactivated by the release of the iron cofactor, and the iron uptake genes are transcribed. Fur also regulates other genes involved in general metabolism and pathogenicity, thus the availability of iron in the host constitutes a key signal which coordinately regulates the expression of those genes (Tanabe et al., 2013).

Uncommonly, E. coli fur mutants have significantly reduced levels of total cellular iron despite the constitutive expression of iron uptake pathways (Abdul-Tehrani et al., 1999). This apparent paradox was explained by the presence of a sRNA, RyhB, which regulates the repression of numerous iron-containing proteins (Massé and Gottesman, 2002). In the absence of Fur, the RyhB sRNA have a key role in the control of iron homeostasis, repressing the synthesis of succinate dehydrogenase, aconitase, fumarase and iron superoxide dismutase. Total cellular iron is decreased as a result, despite a significant increase in cytosolic iron. Fur represses the synthesis of the RyhB, and RyhB in turn negatively regulates the synthesis of proteins that bind iron in the cell (Massé and Gottesman, 2002).

The mechanism of regulation by RyhB is post-transcriptional, requires Hfq and involves base pairing between RyhB and its mRNA target and subsequent RNase E-mediated degradation of the RyhB-mRNA duplex (Massé et al., 2003). Porcheron and Dozois (2015) distinguished three direct and indirect regulation mechanisms of virulence determinants by Fur and RyhB in response to environmental iron concentration. Fur and RyhB can act directly, regulating the expression of a gene, as is the case with genes involved in iron acquisition that are directly suppressed by Fur. The second mechanism involves other transcriptional regulators that determine invasion, acid resistance and toxin production; hence Fur/RyhB would indirectly regulate virulence. The last mechanism is related to the regulation of the locus of enterocyte effacement genes in E. coli, which occurs through the transcriptional regulator Ler; however, neither Fur nor RyhB directly regulates the expression of Ler (Porcheron and Dozois, 2015).

RyhB has a conserved region between E. coli and V. cholerae. The sRNA is longer in the latter, reaching lengths of ∼60 nt on each side of the conserved region (Porcheron et al., 2014). Despite the conserved region, the two sRNAs differ in the degree and function of their regulons, since various mRNAs encoding iron-containing proteins in E. coli are similarly regulated by RyhB in V. cholerae, but other transcripts regulated by RyhB in E. coli showed no RyhB-dependent regulation in V. cholerae and vice versa (Updegrove et al., 2015). Thus in both species RyhB has a core function for iron homeostasis, but in V. cholerae the sRNA has acquired new physiological roles (Updegrove et al., 2015).

RyhB is regulated by Fur in V. cholerae, and interacts with the protein Hfq, required for the suppression of genes that code for the SodB superoxide dismutase and several enzymes in the tricarboxylic acid cycle, including succinate dehydrogenase SdhCDAB and fumarate reductase FumA (Figure 2). RyhB also modulates the expression of several genes that control motility, chemotaxis and biofilm formation in V. cholerae, since several genes involved in flagella biosynthesis and chemotaxis are negatively regulated in a ryhB mutant of V. cholerae, and as mentioned above these are important processes in the virulence of the microorganism (Davis et al., 2005; Mey et al., 2005a) (Figure 2). Both Hfq and RyhB can be co-purified from V. cholerae and high levels of RyhB are only detectable in the presence of Hfq, suggesting that as with E. coli, Hfq stabilizes this sRNA (Davis et al., 2005). Using microarray analysis, Mey et al. (2005b) showed that numerous genes that are involved in iron acquisition, including fhuAC, feoAB and irgA, were increased when RyhB was expressed. In addition, a fur mutant of V. cholerae was unable to use pyruvate, succinate or fumarate as carbon sources (Litwin and Calderwood, 1994; Mey et al., 2005a). This suggests that V. cholerae may use a system analogous to the E. coli RyhB mechanism for regulating genes encoding iron-containing proteins and those involved in iron metabolism.

Homologs of V. cholerae RyhB have been identified in other Vibrio species including V. vulnificus and V. parahaemolyticus (Mey et al., 2005a; Porcheron et al., 2014). In V. vulnificus iron is necessary for growth and increased host mortality in vivo (Wen et al., 2016). Alice et al. (2008) carried out competitive experiments in an “iron-overloaded mouse” model with the wild type strain of V. vulnificus, which showed that the mutant ryhB is less virulent. While the fur mutant (lethal dose) LD₅₀ is the same as the wild type strain, the LD₅₀ of the RyhB mutant is more than 30 times greater than that of the wild type strain (Alice et al., 2008; Porcheron and Dozois, 2015). The researchers attributed the attenuated virulent phenotype of the V. vulnificus ryhB mutant to a defect in the growth observed in this mutant under limited iron conditions. However, in a double mutant ΔryhB fur::pDM4 ΔlacZ growth was recovered under limited iron conditions, although the virulence remained attenuated, possibly implying that other genes related to the virulence of V. vulnificus may be directly or indirectly under the control of RyhB (Alice et al., 2008).

Vibrio parahaemolyticus can use several siderophores, including an analog of vibrioferrin, exogenous aerobactin, ferrichrome and enterobactin to acquire iron under low iron conditions (Tanabe et al., 2013). It has been observed that the production of vibrioferrin decreased in a ryhB deletion mutant of V. parahaemolyticus compared to the parent wild type strain. This suggests that RyhB positively regulates the production of vibrioferrin, and the evidence indicates that this probably occurs through Hfq-dependent stabilization of the operon pvsOpm which is involved in its biosynthesis. This was the first report to state that RyhB positively regulated a polycistronic mRNA (Tanabe et al., 2013).