The envelope protects the bacterial cell from toxic substances or environmental stress. Extracellular or intracellular signals may cause perturbation of the cell envelope, which activates the Rcs system (Majdalani et al., 2005). In E. coli, these signals include the damage of periplasmic peptidoglycan caused by lysozyme (Callewaert et al., 2009), the transient cell wall-deficient (or L-form) state caused by exposure to β-lactam antibiotics (Hirakawa et al., 2003; Glover et al., 2009), defective lipoprotein sorting by lolA mutation (Tao et al., 2012), loss of the periplasm-spanning protein TolA (Vianney et al., 2005; Morgan et al., 2014), and modification of the peptidoglycan in the absence of certain penicillin-binding proteins PBP4, PBP5, PBP7, or AmpH (Table 2; Evans et al., 2013). Rcs system is also induced by perturbation of the cell envelope in other bacteria. For example, a change in the concentration of osmoregulated periplasmic glucans (OPGs) in Dickeya dadantii (Bontemps-Gallo et al., 2013), the absence of UDP-glucose dehydrogenase in Edwardsiella tarda (Lv et al., 2012), alteration in the enterobacterial common antigen (ECA) structure in Serratia marcescens (Castelli and Vescovi, 2011), and the outer membrane damage caused by oxidative stress in Salmonella can modulate Rcs phosphorelay system (Table 2; Farizano et al., 2014). In summary, the perturbation of the envelope caused by these extracellular or intracellular signals can be sensed by Rcs system, which in turn regulates its target genes to allow cells to adapt to the environmental or genetic changes.

RcsF, a small lipoprotein (14 kDa), is membrane-associated via a lipidated N-terminal membrane-anchored helix at the outer membrane, with a loop and a central four-stranded β-sheet in the periplasmic space (Castanié-Cornet et al., 2006). Acting as a mediator between environmental stimuli and Rcs phosphorelay system, RcsF activates Rcs system signaling in response to different types of envelope stress (Shiba et al., 2012; Laloux and Collet, 2017). The mechanism by which envelope stress induces the Rcs system via RcsF has been recently investigated by Silhavy’s lab and Collet’s lab (Cho et al., 2014; Konovalova et al., 2014). Both of them found that Rcs system interacts with BamA and three β-barrels (OmpA, OmpF, and OmpC) to detect envelope stress. Cho et al. (2014) proposed that the interaction between RcsF and BamA plays a major role in RcsF sensing (Figure 1A). In the absence of envelope stress, RcsF is trapped by the outer membrane β-barrel assembly machinery (Bam) and funneled to β-barrels (Cho et al., 2014; Konovalova et al., 2014). In this situation, RcsF cannot interact with YrfF (IgaA in Salmonella enterica serovar Typhimurium) to activate Rcs system (Figure 1A). In the presence of envelope stress, the interaction between RcsF and BamA is impaired and therefore RcsF cannot be funneled to OmpA/C/F. This allows RcsF to be exposed to the periplasm and free to activate Rcs system (Figure 1A). This model proposed that RcsF senses envelope stress by monitoring the Bam complex (Cho et al., 2014). Konovalova et al. (2014, 2016) proposed another model to explain how RcsF senses the lipopolysaccharide (LPS) defects. In their model, RcsF forms a complex with β-barrels and uses its positively charged, surface-exposed N-terminal domain to directly monitor lateral interactions between LPS molecules, which in turn regulate Rcs system (Konovalova et al., 2014, 2016). These two models might represent two different ways for RcsF to sense different types of stress (as reviewed in Laloux and Collet, 2017). Their findings elucidate the mechanism by which RcsF senses and transmits signals as a sentinel in the transduction signal pathway and are important to dissect the regulatory mechanism of Rcs system.

The functional N-terminal domain of RcsF contains five basic amino acid residues and six prolines (Umekawa et al., 2013). The positively charged residues are important for RcsF to sense alterations in LPS (Konovalova et al., 2016). The prolines might also be important for RcsF to sense environmental signals but their role has not been investigated. RcsF has two non-consecutive disulfide bridges, Cys74–Cys118 and Cys109–Cys124 in the E. coli protein, whose formation is catalyzed by the periplasmic disulfide isomerase DsbC (Leverrier et al., 2011; Rogov et al., 2011). Both of the disulfide bridges Cys74–Cys118 and Cys109–Cys124 are important for the activity of RcsF (Rogov et al., 2011). However, whether the formation of disulfide bridges is involved in sensing the environmental signal also needs to be investigated. Future work should continue to reveal the detailed mechanism of RcsF sensing at the molecular level.

Rcs system can be regulated by the expression level or stability of Rcs system proteins (Majdalani and Gottesman, 2005). For example, expression of rcsA is autoregulated by the RcsA–RcsB heterodimer in E. coli, S. enterica, and Klebsiella pneumoniae (Wang and Harshey, 2009). rcsB is transcribed from two promoters, PrcsDB and PrcsB in S. enterica serovar Typhimurium (Wang and Harshey, 2009; Pescaretti Mde et al., 2010). RcsB represses rcsD gene expression by binding directly to the PrcsDB promoter, negatively autoregulating rcsB in S. enterica serovar Typhimurium (Pescaretti Mde et al., 2010). Furthermore, the rcsA gene is indirectly upregulated by c-di-GMP in E. coli (Sledjeski and Gottesman, 1995; Mendez-Ortiz et al., 2006) and negatively regulated by the quorum-sensing regulator EsaR in Pantoea stewartii (Carlier and von Bodman, 2006).

All auxiliary regulators seem to be repressed by the nucleoid-associated global repressor protein H-NS (Sledjeski and Gottesman, 1995; Krin et al., 2010; Venkatesh et al., 2010; Martinez-Santos et al., 2012; Pannen et al., 2016). In addition, transcription of dctR is activated by the regulator YdeO in response to extracellular acid stress (Yamanaka et al., 2014). The expression of matA in the mat (ecp) operon encoding fimbrial adhesin is induced at acidic pH, low temperature, or high acetate concentration (Lehti et al., 2013). Finally, RcsA is unstable at 37°C and can be degraded by the ClpYQ and Lon proteases (Allen et al., 1987; Stout et al., 1991; Chang et al., 2016). The regulatory roles of RcsB heterodimers are largely controlled by the availability and regulation of the auxiliary regulators.

RcsC is located at the inner membrane, with most of the protein located in the cytoplasm and a small periplasmic domain in the periplasmic space (Löhr et al., 2005). The atomic structure of the C-terminus of E. coli RcsC was determined by NMR, revealing a novel α–β-loop (ABL) domain and a PR domain (Löhr et al., 2005; Rogov et al., 2006). Upon phosphorylation of the RcsC-PR domain, it is recognized by RcsD-HPt and phosphoryl transmission takes place (Löhr et al., 2005). Meanwhile, RcsC-ABL facilitates phosphoryl transfer between RcsC-HK and RcsC-PR, or between RcsC-PR and RcsD-HPt (Rogov et al., 2006). These interactions mediate the signal transduction function of RcsC, while the first identified ABL domain of RcsC revealed a coordination module. Thus, RcsC contributes to the complexity of prokaryotic signaling systems.

RcsC was initially identified as a kinase in the Rcs phosphorelay system. As we know, many kinases are also phosphatases, and subsequent studies showed that RcsC also possesses phosphatase activity (Garcia-Calderon et al., 2005). RcsC functions together with RcsD as a phosphatase to precisely control the low levels of phosphorylated RcsB in the cell in the absence of environmental signal. Residues Ala904, Asp875, and Thr903 are required for the putative phosphatase activity of RcsC-PR (Garcia-Calderon et al., 2005). Recently, Ancona et al. (2015) reported that RcsC also exhibits dual functionality in Erwinia amylovora. Mutation of Thr904 (corresponding to Thr903 in E. coli RcsC) and Ala905 (corresponding to Ala904 in E. coli RcsC) constitutively activates the Rcs system (Ancona et al., 2015). The dual functions of RcsC are essential for its regulatory role to prevent the constitutive activation of the Rcs system. However, the molecular mechanism of its phosphatase activity is not yet clear.

RcsD is an inner membrane protein with a substantial periplasmic domain. Unlike RcsC, RcsD does not show any autophosphorylation activity, due to the lack of a canonical active site (Majdalani and Gottesman, 2005). The structure of the HPt domain of E. coli RcsD has been determined and shown to be similar to ArcB-HPt (Rogov et al., 2004). RcsD-HPt has a recognition area in close vicinity to the phosphoryl site (His842) that is recognized by two cognate PR domains, RcsB-PR and RcsC-PR. The interaction interfaces of the two domains with RcsD-HPt largely overlap. Meanwhile, Schmoe et al. (2011) describe a complex between RcsD-ABL and RcsB. Earlier Rogov et al. (2006) describe a complex between RcsC-ABL and RcsD. These complexes probably enhance the interaction between membrane-bound proteins and soluble components to increase the rate and efficiency of the phosphorelay signaling cascade (Schmoe et al., 2011).

The C-terminal DNA-binding domain of RcsB, belonging to the FixJ/NarL family, forms a helix-turn-helix (HTH) DNA-binding motif (Pristovsek et al., 2003). The N-terminal domain of RcsB, containing the phosphoacceptor site, belongs to the family of two-component receiver domains (Filippova et al., 2016). This receiver domain participates in RcsB dimer formation and also contributes to dimer formation with other transcription factor partners. In response to environmental stimuli, RcsB is phosphorylated by RcsD, and phosphorylation modulates its own DNA-binding affinity at the promoter region of target genes (Majdalani and Gottesman, 2005; Ancona et al., 2015). The phosphorylation status of RcsB is critical for biofilm development in S. enterica serovar Typhimurium (Latasa et al., 2012). Phosphorylated RcsB stimulates the transcription of sRNA rprA, and the accumulation of rprA leads to repression of the transcriptional regulator CsgD, resulting in inhibition of biofilm development (Latasa et al., 2012). In the absence of environmental signals, RcsC and RcsD might act as a phosphatase to dephosphorylate RcsB to maintain a low level of rcsB expression (Clarke, 2010; Hu et al., 2013). In this situation, AcP, the intermediate of the Pta-AckA pathway, might provide the phosphoryl group for RcsB, ensuring a low level of RcsB phosphorylation (Hu et al., 2013). Recently, Filippova et al. (2016) solved the crystal structure of non-phosphorylated receiver domain of RcsB. They found that the RcsB receiver domain can exist as a dimer in solution (Filippova et al., 2016). They suggested the non-phosphorylated and phosphorylated states of RcsB use a similar dimerization interface in the homodimer or heterodimers (Filippova et al., 2016). How phosphorylation affects the DNA binding ability and regulatory roles of RcsB is still not clear.

Another RcsB modification is lysine acetylation. RcsB can be acetylated by AcP and deacetylated by CobB (Thao et al., 2010; Hu et al., 2013). Thao et al. (2010) found that Lys180 of RcsB can be acetylated in vitro, while Hu et al. (2013) reported that several lysines in RcsB can be acetylated in vivo and acetylation of Lys154 negatively regulates RcsB activity. Castano-Cerezo et al. (2014) reported that acetylation of RcsB prevents DNA binding, activates flagella biosynthesis and motility, and compromises acid stress survival. Their finding indicates that acetylation might play an important role as same as phosphorylation in regulation of the activity of RcsB. Although Rcs system is important for enterobacteria, constitutive activation of the Rcs system could reduce the fitness and virulence (Ancona et al., 2015). Thus phosphorylation and acetylation might act together to tightly control the activity of RcsB, which in turn fine-tunes the function of Rcs system to adapt to the differential environment. The inter-conversion between phosphorylation and acetylation of RcsB remains largely unclear and needs to be investigated.

RcsB binds DNA either as a homodimer or heterodimer with auxiliary regulators (Table 1). The auxiliary proteins cooperate with RcsB to modulate transcription by binding to specific DNA motifs. The motifs binding by different dimers have a similarity in one half that is probably bound by RcsB, whereas the other half varies. The interaction of RcsB with auxiliary proteins might alter the specificity or strengthen the ability of DNA binding (Salscheider et al., 2014).

RcsB homodimer regulates targets once RcsB is phosphorylated. The binding sites are usually located upstream of the -35 region of the RNA polymerase-binding site (Majdalani and Gottesman, 2005). RcsB homodimer positively regulates sRNA rprA and the osmoregulated gene osmC in E. coli, as well as other targets in an RcsB phosphorylation-dependent manner (Majdalani and Gottesman, 2005; Clarke, 2010). RcsA stabilizes RcsA–RcsB complex to the RcsAB box and enhances the regulation activity of RcsB (Majdalani and Gottesman, 2005; Clarke, 2010). It is worth noting that RcsA’s function is dependent of phosphorylation of RcsB, whereas the other auxiliary regulators such as MatA, BglJ, and GadE act independently of phosphorylation of RcsB (Castanié-Cornet et al., 2010; Venkatesh et al., 2010; Pannen et al., 2016). For phosphorylation-independent heterodimers, it is not the level of RcsB phosphorylation in the cytoplasm but rather the presence of different auxiliary regulators that is important for its functions (Pannen et al., 2016). In addition, GadE–RcsB heterodimer activates the transcription of glutamate-dependent (Gad) acid resistant genes gadA/BC, but acetylation of RcsB negatively regulates gadA/BC (Castanié-Cornet et al., 2010; Castano-Cerezo et al., 2014). This indicates the function of phosphorylation-independent heterodimers might be affected by the acetylation of RcsB.

Castanié-Cornet et al. (2010) and Venkatesh et al. (2010) reported that BglJ–RcsB and GadE–RcsB heterodimers are active independent of RcsB phosphorylation. Pannen et al. (2016) found that MatA–RcsB heterodimer is also active independent of RcsB phosphorylation. In addition, they investigated conserved or surface-exposed residues and their roles in the activity of RcsB in E. coli. Their results revealed that mutation of residues in the vicinity of the phosphorylation site such as Asp56, Asp11, Thr87, and Lys109 impairs the activity of phosphorylation-dependent dimers such as RcsA–RcsB and RcsB–RcsB. Meanwhile, residue Lys109 also affects the activity of phosphorylation-independent heterodimers such as BglJ–RcsB and MatA–RcsB. One possible explanation is that a phosphorylation-induced structural change releases the DNA-binding domain that is otherwise buried or inhibited by the non-phosphorylated form of the PR domain. For both phosphorylation-dependent and -independent dimers, mutation of surface-exposed residue Ile14 impairs the activity of the dimers, indicating that RcsB and its auxiliary proteins interact via a surface similar to that with which NarL and related proteins interact (Pannen et al., 2016). Their findings suggest that phosphorylation might cause a structure change to stabilize the active form of RcsB–RcsB and RcsA–RcsB, while the heterodimers of BglJ–RcsB and MatA–RcsB are intrinsically active.

The heterodimers formed by the transcriptional regulator RcsB with auxiliary regulatory proteins modulate the DNA binding specificity and expands the regulatory role of Rcs system. In addition, the flexibility of the Rcs system regulatory role is further increased because some of the RcsB heterodimers depend on RcsB phosphorylation but some of them do not depend on RcsB phosphorylation. Multiple auxiliary regulators with varying dependence on the phosphorylation of RcsB endow flexibility to the Rcs system that may allow the bacteria to adapt to diverse environments.

The plague pathogen Y. pestis forms biofilms in the midgut of its flea host to enhance transmission, and cyclic-di-GMP positively regulates biofilm development. In Y. pestis, c-di-GMP is synthesized by diguanylate cyclases HmsT and HmsD, and degraded by HmsP (Sun et al., 2011). We showed that the transcription of hmsT is directly repressed by RcsB in Y. pestis (Sun et al., 2012). RcsB binds to a conserved Rcs-binding motif which overlaps with the transcriptional start site of hmsT, leading to the repression of the transcription of hmsT (Sun et al., 2012). In addition, we also found that RcsB positively and directly regulates the hmsCDE operon (Guo et al., 2015). Furthermore, Fang et al. (2015) found that RcsB also directly represses the transcription of hmsHFRS, which is responsible for the biosynthesis and translocation of biofilm matrix exopolysaccharide. In summary, the Rcs system acts as a master repressor of biofilm formation by regulating the production of cyclic-di-GMP and biofilm matrix exopolysaccharide synthesis to facilitate environmental adaptation in Y. pestis.

Hundreds of genes were found to be regulated by the Rcs system using genome-wide approaches in E. amylovora, E. coli, Y. pestis, and S. enterica (Hinchliffe et al., 2008; Bury-Mone et al., 2009; Mariscotti and Garcia-del Portillo, 2009; Wang et al., 2012). Numerous Rcs-associated phenotypes have been characterized including quorum sensing in E. coli (Shimada et al., 2013), cell division in Proteus mirabilis (Howery et al., 2015), lipoprotein sorting in E. coli (Tao et al., 2012), and biogenesis of outer membrane vesicles in S. marcescens (McMahon et al., 2012). Recent advances on the targets of the Rcs system have focused on acid resistance and virulence.

The Gad system is the most efficient acid resistant system in E. coli (Foster, 2004; De Biase and Pennacchietti, 2012; Lund et al., 2014). RcsB controls the expression of gad genes and is required for acid resistance in E. coli (Castanié-Cornet et al., 2007). Castanié-Cornet et al. (2007, 2010) reported that the basal activity of RcsB is necessary and sufficient for GadE dependent regulation of gad genes, whereas increased RcsB activity through activation of the Rcs phosphorelay system or through overproduction of the protein leads to general repression of the gad genes. This result is consistent with the later findings that the regulatory role of RcsB/GadE is independent of RcsB phosphorylation (Castanié-Cornet et al., 2010), while the regulatory role of RcsB homodimer is dependent of RcsB phosphorylation. Further studies showed that activation of transcription gadA, encoding a glutamate decarboxylase, requires binding of RcsB/GadE to a GAD box, while repression of gadA transcription is caused by binding of RcsB to an RcsB box which lies in the -10 and -35 regions of gadA promoter (Castanié-Cornet et al., 2010). In response to acid stress, RcsB might function together with GadE to precisely regulate the expression of acid resistance-related genes (Castanié-Cornet et al., 2007; Johnson et al., 2011). However, a number of acid resistance-related genes including safA, slp, and gadW are regulated by RcsB but not GadE (Johnson et al., 2011). Johnson et al. (2011) speculated that RcsB might work with the response regulator EvgA as a heterodimer to regulate acid resistance-related genes.

The Rcs system also contributes to bacterial virulence. LcrF, a master transcriptional regulator of virulence genes in Yersinia pseudotuberculosis, is regulated by Rcs system (Li et al., 2015). Rcs system can also modulate the interaction between pathogen and host. For example, RcsB regulates the colonization of E. coli on the mouse intestine (Lasaro et al., 2014). In addition, deletion of cytoskeletal element mreC induces RcsC, resulted in downregulation of flagella systems and the SPI-1 type 3 secretion system (T3SS) and attenuated virulence in Salmonella (Bulmer et al., 2012). In S. marcescens, RcsB also represses the pore-forming toxin ShlA that is responsible for early induction of autophagy in host cells (Di Venanzio et al., 2014). Therefore, the Rcs system can affect the interaction between pathogen and host by regulating the assembly of the secretion system, colonization of the host, and early induction of autophagy in host cells.

The rcsA gene in Y. pestis contains a 30 bp internal duplication, and is therefore nonfunctional, whereas rcsA in the ancestral Y. pseudotuberculosis species is functional and can repress biofilm formation. When Y. pestis is supplemented with the functional Y. pseudotuberculosis RcsA, biofilm formation is repressed and even abolished in fleas. Therefore, the conversion of rcsA to a pseudogene during the evolution of Y. pestis represents a case of positive selection (Sun et al., 2008). Furthermore, the loss of rcsA in Y. pestis following the gain of phospholipase D encoded by ymt, together with the loss of two minor genes in the bacterial progenitor, facilitated the transmission of fleas by stimulating c-di-GMP-mediated biofilm formation in the flea foregut (Sun et al., 2014). The evolutionary history of rcsA in Y. pestis is one example that a few genetic changes such as gene losses can alter the regulation of targets and consequently the infection phenotype, leading to arthropod–borne transmission. Despite an apparent frameshift, RcsD remains functional and acts in concert with RcsB to promote biofilm formation in Y. pestis, but it functions in a way that is opposite to RcsD in Y. pseudotuberculosis (Sun et al., 2008). The effect of mutating rcsD on Rcs system signal transduction and its biological significance requires further experimental investigation.

Escherichia coli O157: H7 (O157) is a frequent cause of foodborne disease in humans, and is transmitted to human through food vehicles. To cause infection, O157 needs to adapt to and survive in diverse conditions. Curli fimbriae play an important role in surface attachment, biofilm formation, and confer cell protection from toxic compounds. The production of curli in E. coli O157: H7 is intrastrain and interstrain variant (Carter et al., 2012). The curli producing (C+) O157 variants are much more acid sensitive, but grow better in nutrient-limited environment than curli-deficient (C-) variants. Deletion of rcsB gene is responsible for the C+ phenotypes in two O157 strains, isolated during the 1993 U.S. hamburger-associated outbreak (Carter et al., 2012). In addition, the C+ O157 variants can also be generated by insertion or single nucleotide change in rcsB gene (Chen et al., 2016; Sharma et al., 2017). RcsB affects bacterial survival in growth-restrictive environments through regulation of genes related to biofilm formation and certain metabolic functions. Inactivation of rcsB in O157 might enhance the persistence and survival ability in nutrient limited environment, but decrease the resistance to heat shock and acid stress. Thus, mutation of rcsB generates a distinct O157 subpopulation that may confer this pathogen with survival advantages in the ever changing environment.