Polysialic and colanic acids metabolism in Escherichia coli K92 is regulated by RcsA and RcsB

We have shown previously that Escherichia coli K92 produces two different capsular polymers known as CA (colanic acid) and PA (polysialic acid) in a thermoregulated manner. The complex Rcs phosphorelay is largely related to the regulation of CA synthesis. Through deletion of rscA and rscB genes, we show that the Rcs system is involved in the regulation of both CA and PA synthesis in E. coli K92. Deletion of either rcsA or rcsB genes resulted in decreased expression of cps (CA biosynthesis cluster) at 19°C and 37°C, but only CA production was reduced at 19°C. Concerning PA, both deletions enhanced its synthesis at 37°C, which does not correlate with the reduced kps (PA biosynthesis cluster) expression observed in the rcsB mutant. Under this condition, expression of the nan operon responsible for PA catabolism was greatly reduced. Although RcsA and RcsB acted as negative regulators of PA synthesis at 37°C, their absence did not reestablish PA expression at low temperatures, despite the deletion of rcsB resulting in enhanced kps expression. Finally, our results revealed that RcsB controlled the expression of several genes (dsrA, rfaH, h-ns and slyA) involved in the thermoregulation of CA and PA synthesis, indicating that RcsB is part of a complex regulatory mechanism governing the surface appearance in E. coli.


INTRODUCTION
Over 80 distinct capsular or K antigens have been described in Escherichia coli, which are classified into four groups [1]. Group II K antigens exhibit capsule expression at 37 • C but not at low temperatures (18 • C) and their regulation is temperature-dependent [1,2]. The group II capsule gene cluster (kps) consists of a central serotype-specific region 2, encoding proteins for synthesis and polymerization of the specific K antigen that is flanked by conserved regions 1 and 3 ( Figure 1A) [1,3]. Transcription of the kps cluster is driven by two convergent temperature-regulated promoters located upstream of regions 1 and 3 [4]. Transcription of regions 2 and 3 is driven by the PR3 promoter [4] and is dependent on RfaH for transcription elongation [5]. In addition, H-NS .. Abbreviations used: CA, colanic acid; C T , threshold cycle value; Glc-Pro, glucose-proline; LA, LB supplemented with 2 % w/v agar; LB, Luria-Bertani; MM, minimal media; PA, polysialic acid; qRT-PCR, quantitative real-time PCR; RT-PCR, reverse transcriptase-PCR; Xyl-Asn, xylose-asparagine. 1 Correspondence may be addressed to either of these authors (email hmarb@unileon.es or leandro.rodriguez@unileon.es).
plays an unusual dual role, not only being required for maximal transcription at 37 • C but also contributing to transcriptional repression at low temperatures ( 20 • C) [4,6]. Transcription of region 1 is driven by promoter PR1, and its maximal expression at 37 • C requires both H-NS and SlyA, whereas a reduced SlyA expression at 20 • C results in repressed transcription from PR1 [4,6,7]. E. coli K92 synthesizes a type II capsule known as PA (polysialic acid), which is responsible for bacterial virulence [8,9]. As with the group II K antigens, PA is mainly generated at 37 • C and is negligible at low temperatures (below 20 • C) [10]. We previously showed that E. coli K92 is able to synthesize a different capsular polysaccharide known as CA (colanic acid) [11]. CA is associated with bacterial protection against desiccation, extreme temperatures and acidic environmental conditions [12,13], as well as against osmotic and oxidative stress [14]. In contrast to PA, CA is predominantly synthesized at low temperatures [11,15] and does not seem to play a role in bacterial virulence [1,16]. A complex signal transduction pathway, namely the Rcs phosphorelay, controls the expression of the cps operon ( Figure 1B) responsible for CA production. This pathway can be activated at low temperatures and involves a histidine kinase (RcsC), a response regulator (RcsB), a phospho-transfer protein (RcsD), a signal transductor (RcsF) and an auxiliary activator protein (RcsA) [15,17]. The transcriptional activator RcsB is the principal regulator of this system and it forms homodimers which activate cps operon transcription. In addition, RcsB may form heterodimers with the auxiliary activator RcsA, enhancing cps operon transcription. It has been described that RcsB is required for cps gene expression, whereas the absence of RcsA only decreases it [15]. Under normal conditions, the amount of RcsA protein is limited by its low synthesis level and because it is rapidly degraded by a Lon protease. However, synthesis of RcsA is increased at low temperatures, leading to a mucoid phenotype as a consequence of the expression of cps genes [15]. In addition, other regulatory molecules are implicated, either directly or indirectly, in control of CA synthesis. Thus, the anti-termination factor RfaH is required for transcription of genes downstream of wzc [5,18], while DsrA is a small molecule of RNA which increases cps transcription through its negative regulation of H-NS synthesis [19].
Although most of the targets present in the Rcs regulon are positively regulated, it has been suggested that the Rcs regulon also may repress the synthesis of type II capsules in E. coli [20]. Through deletion of rcsB and rcsA genes, in this report we show that RcsA and RcsB act as negative regulators of PA synthesis at 37 • C in E. coli K92, although the mechanism remains unclear. Moreover, both proteins enhance the expression of the cps operon, which is required for maximal CA production at low temperatures. However, they barely play a role in CA synthesis at high temperatures (37 • C). More importantly, neither rcsA nor rcsB deletion completely abolished CA synthesis under any of the conditions tested. In sum, these data suggest that the Rcs system is a critical regulator of the adaptation response of E. coli K92 to different environmental conditions through the expression of PA and CA capsules.

Strains, culture media and growth conditions
The strains and plasmids used in the present study are shown in Table 1. Bacterial cultures were inoculated and grown at 37 or 19 • C as previously described in Navasa et al. [21]. Bacterial cultures were grown in LB (Luria-Bertani) complex medium, LA [LB supplemented with 2% (w/v) agar] and Xyl-Asn (xyloseasparagine) or Glc-Pro (glucose-proline) MM (minimal media) for E. coli [22]. Where indicated, Glc-Pro MM was supplemented with agar 2 % (w/v). We chose Xyl-Asn or Glc-Pro MM because they induce maximal PA and CA production in E. coli K92, respectively [2,14]. During the gene allelic exchange experiments, LA medium supplemented with 5 % (w/v) sucrose and without NaCl was used to select plasmid excision from the chromosome [23]. When required, the following supplements were added to the culture media: rifampicin (25 and 10 μg/ml for liquid and solid media, respectively), kanamycin (25 and 12.5 μg/ml for liquid and solid media, respectively), ampicillin (100 μg/ml) and chloramphenicol (60 μg/ml). .

qRT-PCR (quantitative real-time PCR)
The DNase-treated RNA was reverse transcribed with the ThermoScript TM RT-PCR (reverse transcriptase-PCR) System (Invitrogen S. A.). For qRT-PCR, each cDNA product was used as a template for DNA amplification, using primer pairs as previously described [21], 10 μl of SYBR ® Green PCR Master Mix (Applied Biosystems) and up to 20 μl of water. Primers were  previously described [28] using the equation: WT genotype) and then transformed into relative changes (n-fold) using 2 − C T . C T (threshold cycle value) is the cycle number at which the real-time amplification curve crosses the user-defined threshold, x • C is the temperature at which the RNA was isolated (37 • C or 19 • C) and Y is the mutant strain (E. coli K92ΔrcsA or E. coli K92ΔrcsB). The data represent the average change (n-fold) determined from at least three independent experiments. As a control we used the housekeeping gene gapdh, which was carefully validated before its use in the quantitative mRNA assays, with 16S rRNA expression as internal control obtained under the same conditions and determined from at least three independent experiments.

sacB-assisted allelic exchange mutagenesis
pDS132-based allelic exchange plasmids [23] were electroporated into E. coli S17-1λpir cells for conjugation into E. coli K92 by filter mating [29]. Transconjugants containing single crossovers of the allelic exchange plasmid integrated into the E. coli K92 genome were selected in LA supplemented with chloramphenicol and rifampicin. To force the second recombination, single-crossover strains were plated onto LA containing rifampicin and 5 % (w/v) sucrose and then incubated at 37 • C for 24-48 h. Sucrose-resistant colonies were placed on LA-sucrose plates and screened for a loss of chloramphenicol resistance encoded by the vector.

Deletion of rcsA and rcsB genes from E. coli K92
The 0.3-kb upstream and downstream sequences included in both rcsA or rcsB loci were PCR amplified using the primers indicated in Table 2. These 0.3-kb amplicons were individually cloned into pGEM-T Easy, yielding plasmids pGEM-W, pGEM-X, pGEM-Y and pGEM-Z, respectively. The cloned sequences were excised using EcoRI. Downstream and upstream sequences for each gene were ligated and the products were amplified using the primer pairs rcsAup5 and rcsAdown3 and rcsBup5 and rcsBdown3 , and cloned again into pGEM-T Easy, yielding pGEM-WX and pGEM-YZ. The rcsA and rcsB upstreamdownstream DNA sequences (ΔrcsA and ΔrcsB, respectively) were excised from pGEM-WX and pGEM-YZ using SacI and SphI restriction enzymes and cloned into pDS132 previously digested with the same enzymes, yielding plasmids pDS132-WX and pDS132-YZ, respectively. The suicide vectors carrying ΔrcsA or ΔrcsB were electroporated into E. coli S17-1λpir for biparental conjugation into E. coli K92. The deletions were recombined into the chromosome of E. coli K92 by using the standard two-step sucrose-resistance-assisted allelic exchange method described above. The correct allelic exchange of the WT allele for each mutant allele was confirmed by PCR using the primers rcsAup5 and rcsAdown3 , and rcsBup5 and rcsBdown3 . The ΔrcsA and ΔrcsB E. coli K92 mutants were named E. coli K92ΔrcsA and E. coli K92ΔrcsB, respectively. For each mutant, loss of the DNA fragment was confirmed by PCR and the absence of expression of the deleted gene was confirmed by qRT-PCR.

Quantification of exopolysaccharides
Quantitative determination of CA and PA production by E. coli K92 cultures was performed as previously described [11]. Briefly, E. coli K92 cells were removed by centrifugation and the cell-free supernatant obtained was dialysed against 1000 vol of distilled water for 24 h at 4 • C. Dialysed supernatant samples were used for quantitative determination of CA by the orcinol method [30], according to the amount of uronic acids. Dialysed supernatant samples were also used for quantitative determination of PA following the resorcinol protocol described by Svennerholm [31].

Capsule staining
The presence or absence of CA surrounding the bacteria was evaluated by using a combination of negative and fuchsine staining procedures. Briefly, a very small drop of bacterial culture was placed near one end of a well-cleaned slide. Once dry, one drop of fuchsine (primary colorant) was added to the bacterial culture without spreading for 2 min. After that, the sample was washed with water and dried. Next, a drop of nigrosin (India ink) no greater than the drop of bacterial culture was added and the mixture was spread over the slide using another clean slide. Finally, a cover glass was placed on the sample, avoiding the formation of bubbles and examined under a Nikon Eclipse E600 optical microscope.

Statistical analysis
The results are presented as means + − S.E. Significant differences between means were calculated with Student's t test. P values of 0.05 or less were considered statistically significant.

rcsA and rcsB gene products control CA synthesis in E. coli K92 at low temperatures
We previously showed that E. coli K92 (WT) synthesizes CA as capsular polymers in a temperature-dependent manner [11,21]. It is well known that Rcs phosphorelay and the auxiliary protein RcsA act as positive regulators of CA synthesis [15]. To investigate the role of RcsA and RcsB in CA synthesis by E. coli K92, we performed gene deletion experiments to obtain E. coli K92ΔrcsA and E. coli K92ΔrcsB mutant strains lacking rcsA and rcsB genes, respectively. WT and both mutant strains were grown in Glc-Pro and Xyl-Asn MM at 19 • C, the optimal growth temperature for synthesis of CA by E. coli K92 [11,12], and production of this polymer was determined after 120 h. Neither mutant showed any change in growth under the conditions tested ( Figure 2) and both deletions resulted in a dramatically decreased CA production at  19 • C (measured as glucuronic acid content) in the media tested ( Figures 3A and 3B), but it was not completely abrogated. We ensured that all glucuronic acid detected belonged to a high molecular mass structure, such as CA polymer, through prior dialysis of supernatants using a 10 kDa membrane pore size. To assess the absence of CA capsules surrounding the bacteria, WT and E. coli K92ΔrcsA were grown on Glc-Pro MM agar plates at 19 • C and the Burri method using China ink was applied. This staining technique revealed large amounts of capsular polymer produced by WT in contrast to the mutant strain ( Figure 4). Next, we examined whether RcsA and RcsB control CA production at 37 • C. At this temperature, the levels of CA produced by both mutants were similar to those produced by WT (Figures 3C and 3D). Overall, these results indicate that both RcsA and RcsB positively regulate CA synthesis by E. coli K92 at low temperatures and suggest an Rcs phosphorelay-independent CA synthesis at high temperatures.

Deletion of rcsA or rcsB genes down-regulates the expression of CA synthesis genes
To determine whether RcsA and RcsB regulate the expression of the cps operon responsible for CA synthesis, WT and both mutant strains were grown at 19 • C to enhance CA production, and RNA samples harvested from mid-exponential phase were used to analyse the expression of several genes belonging to the cps operon ( Figure 1B) by RT-PCR. Consistent with the reduced CA production, deletion of rcsA resulted in decreased cps gene expression of 1.4-and 8.5-fold (Table 3). This effect was greater in the absence of RcsB, resulting in a reduction of up to 35-fold. We also analysed expression of the ugd gene, which is located outside the cps operon but is also involved in CA synthesis [32]. Deletion of rcsB reduced ugd expression 2.3-fold, whereas deletion of rcsA barely increased it (1.4-fold).
Since deletion of rcsB resulted in a higher reduction in expression of ugd and cps genes at 19 • C, we also investigated whether RcsB controls their transcription at 37 • C, even though its absence barely decreased CA production at this temperature. Deletion of rcsB resulted in a lower decrease in cps gene expression (between 2.1 and 4.9-fold) ( Table 3) than that observed at 19 • C, and only the expression of gmd and fcl genes was reduced in a similar manner. We compared relative gene expressions for both temperatures given that temperature barely changed cps expression in E. coli K92 [21]. In addition, deletion of rcsB decreased ugd expression 3.3-fold.
rcsA and rcsB gene products repress PA synthesis in E. coli K92 E. coli K92 predominantly synthesizes PA type II capsules at 37 • C, and no production was detected below 20 • C [2,10,21]. It has been suggested that the Rcs phosphorelay system may negatively regulate bacterial group II capsule synthesis [20].To investigate this, WT and both mutant strains were grown in Glc-Pro and Xyl-Asn MM at 37 • C, the optimal growth temperature .    for synthesis of PA by E. coli K92 [11,21], and the production of this polymer was measured after 120 h. Deletion of either rcsA or rcsB resulted in increased PA production under all conditions tested ( Figure 5). This increment was not significant when E. coli K92ΔrcsA was grown in Xyl-Asn MM. Dialysis of the supernatants ensured that all sialic acid detected belonged to a high molecular mass structure. These results show that RcsA and RcsB act, directly or indirectly, to repress PA synthesis in E. coli K92. However, this effect was insufficient to overcome inhibition of PA capsule synthesis at low temperatures, since neither E. coli K92ΔrcsA nor E. coli K92ΔrcsB were able to generate PA synthesis at 19 • C (results not shown).

Dual role of RcsB in transcriptional regulation of PA metabolism genes
We next investigated whether RcsB controls the expression of the kps operon responsible for PA synthesis. WT and E. coli K92ΔrcsB were grown at 19 • C and 37 • C, and RNA samples harvested from mid-exponential phase were used to analyse the expression of several genes belonging to the kps operon [1,10]. At 19 • C, the deletion of rcsB resulted in an increase (between 1.4-and 7.7-fold) in the expression of all tested genes (Table 4), although no PA production was observed. In contrast, E. coli K92ΔrcsB showed a reduction (up to 2.3-fold) in kps gene expression at 37 • C with respect to WT (Table 4), which does not correlate with the increased PA synthesis (see Figure 4). To determine whether increased PA production at 37 • C could be due to a diminished PA catabolism, we also analysed the expression of several genes belonging to the nan (PA catabolism) operon ( Figure 1C) and the regulator NanR. Interestingly, the expression of nanAET genes was greatly reduced at 37 • C (up to 10-fold), suggesting a diminished PA catabolism, whereas nanR expression remained unchanged. Finally, deletion of rcsB resulted in a slight increase in the expression of nan genes at 19 • C (Table 4) Putative colanic acid piruviltransferase − 6.8 †Description of the products encoded by genes. ‡The relative levels of gene expression were calculated as described in the Materials and methods, and then transformed to relative change using the formula 2 − CT . As a control gene we used the housekeeping gene gapdh.
Sialic acid transport and polymerization †Description of the products encoded by genes. ‡The relative levels of gene expression were calculated as described in the Materials and methods, and then transformed to relative change using the formula 2 − CT . As a control gene we used the housekeeping gene gapdh Transcriptional dual regulator Transcriptional activator *Genes involved in metabolism and regulation of CA and/or PA. †Description of the products encoded by genes. ‡The relative levels of gene expression were calculated as described in the Materials and methods, and then transformed to relative change using the formula 2 − CT . As a control gene we used the housekeeping gene gapdh.

RcsB as part of a complex network regulating both CA and PA synthesis
In order to gain a better understanding of the regulatory mechanism involved in the thermoregulated synthesis of CA and PA in E. coli K92, we further investigated whether RcsB controls the expression of several genes implicated in the regulation of CA and/or PA synthesis. Firstly, we analysed the expression of rcsC, rcsF and rcsA genes belonging to the Rcs pathway. Deletion of rcsB resulted in a greater reduction in rcsA expression at both temperatures compared with WT (29.5-and 128-fold at 19 and 37 • C, respectively) ( Table 5), consistent with the key role of RcsB in high-level expression of rcsA [33]. rcsC gene expression was up-regulated at 19 • C (2.8-fold) but down-regulated at 37 • C (3.5-fold) in the mutant strain, whereas rcsF expression remained unchanged.
We also analysed the expression of dsrA, rfaH, h-ns and slyA genes, which are temperature-dependent in E. coli K92 [21]. DsrA RNA positively regulates rcsA expression [15] and the anti-terminator RfaH is required for the expression of large polysaccharide clusters, including kps [5,34] and presumably cps, whereas H-NS and SlyA are required not only for maximum transcription of the kps operon at 37 • C but also to repress it at 19 • C [7]. Deletion of rcsB resulted in an increased expression of all four genes at 19 • C, especially dsrA (7.6-fold), and a reduced expression at 37 • C (between 1.4-and 2.4-fold).

DISCUSSION
The mechanism by which the Rcs system promotes CA synthesis is well known. As a homodimer, or forming heterodimers with RcsA, RcsB binds to the RcsAB box to enhance cps transcription [34]. Consistent with this, we found a decreased cps expression in rcsA and rcsB mutants at both temperatures (Table 3). This effect was greater in the rcsB mutant at 19 • C, probably because deletion of rcsB has a double effect, abrogating both RcsB-and RcsAB-mediated cps transcription. Thus, it has been described that RcsB is required as the main regulator for cps transcription and no expression is detected in its absence, whereas RcsA merely enhances it [15]. However, we found that cps expression was readily detectable at both temperatures, even in rcsB mutants. Since RcsA alone is unable to promote cps expression [21], these results suggest an RcsB-independent cps transcription. The cps cluster not only codifies enzymes involved in CA synthesis but also in other pathways. For example, Wzc not only participates in the polymerization and exportation of CA capsules [1] but also enhances Ugd activity through phosphorylation [35]. The ugd product is involved in many cellular processes such as capsular polysaccharide [36,37] and LPS (lipopolysaccharide) synthesis, polymyxin B resistance [38] and regulation of the levels of the sigma factor RpoS [39]. Thus, it would not be surprising that the regulation of Wzc was under the control of regulators other than RcsB, reflecting the variety of processes which are involved.
Several findings prompt us to believe that transcriptional regulation of the Rcs system on cps expression does not determine CA synthesis at high temperatures. Thus, reduced cps transcription in the rcsB mutant did not result in a reduced CA production at 37 • C ( Figures 3B and 3D). Although CA production at 37 • C was lower than at 19 • C, it cannot be considered as residual since values of CA production observed at 37 • C were close to, for example, those observed in Xyl-Asn MM at 19 • C, and in this case, CA production substantially decreased to a statistically similar degree in both rcsA and rcsB mutants (Figures 3A and 3C). In addition, we might expect that a higher expression of rcsB would result in an increased cps transcription, and consequently, in enhanced CA production; however, previous results showed that the rcsB gene expresses 6-fold higher at 37 • C than at 19 • C in WT [21]. Furthermore, the thermoregulation of CA production is not probably explained through changes in cps expression, whether or not the Rcs pathway is involved. This is supported by the fact that E. coli K92 produced large amounts of CA at 19 • C in contrast to 37 • C, but minimal changes in cps transcription were detected [21], suggesting that post-transcriptional modifications take place. In this sense, phosphorylation of Ugd by Wzc was found to participate in the regulation of CA production [35].
Furthermore, the phosphorylation/dephosphorylation cycle in tyrosine kinase Wzc was shown to control the production and size distribution of CA polymer in E. coli K-12 and, more importantly, an external desiccation signal was directly linked to the phosphorylation state [40]. Whether the Rcs pathway plays a role at the post-transcriptional level remains unknown.
Unlike CA synthesis, the thermoregulation of group II capsule synthesis in E. coli occurs at a transcriptional level [4,7,41]. Consistent with this, we previously observed that kps operon expression was down-regulated by up to 500-fold in E. coli K92 at low temperatures with respect to high temperatures [21]. Russo and Singh [20] have suggested that group II K54 capsule expression is negatively regulated by RcsA, and this effect appears to be mediated through RcsB. We found that deletion of either rcsA or rcsB enhanced kps transcription by up to approximately 8-fold at 19 • C ( Table 4). This effect is probably physiologically insignificant since kps expression at low temperatures is highly reduced [21]; however, it might reveal a role of the Rcs pathway negatively regulating kps transcription. We also found that deletion of either rcsA or rcsB resulted in an enhanced PA production at 37 • C ( Figure 5), but we failed to explain this effect through transcriptional regulation of the kps operon, whose expression was slightly reduced (Table 4). One possibility might be that increased PA production in both mutants is due to a reduction in PA degradation rather than an increment in its synthesis, since deletion of rcsB resulted in a greater reduction of nan gene expression (Table 4); however, this hypothesis awaits confirmation. Nevertheless, although the Rcs pathway diminished PA production at high temperatures, it does not seem to be responsible for thermoregulated inhibition of group II PA capsules in E. coli at low temperatures, since no detectable PA was observed in either rcsA or rcsB mutants.
Lastly, we further investigated the possible role of RcsB controlling the expression of other molecules implicated in the transcriptional control of cps and/or kps operons. It has been shown that RcsB auto-regulates the Rcs system by modifying the expression of rcsA [33] and rcsD [42] genes. Thus, we found that RcsB greatly enhanced rcsA expression at both temperatures (Table 5). RcsB also affected rcsC expression in a temperature-dependent manner (Table 5). Since RcsC can sense external signals to trigger the action of the Rcs pathway, this result would provide a mechanism by which RcsB modifies the capacity of the bacteria to perceive changes in growth temperature. In contrast, although RcsB does not seem to regulate rcsF expression, activation of RcsF does not require increased transcription [15].
The regulatory mechanism of kps operon transcription remains unclear, although recent studies have identified a set of molecules, including H-NS and SlyA, as important factors in this regulation [4,7]. Our results show that RcsB up-regulates and downregulates the expression of all tested genes (dsrA, rfaH, h-ns and slyA) at high and low temperatures, respectively. Although more studies are required to clarify this complex regulatory mechanism, we want to emphasize some aspects. Deletion of rcsB repressed the expression of rfaH, slyA and h-ns genes at 37 • C (Table 5), which may explain the decreased kps expression observed in the rcsB mutant, since all three proteins are necessary for maximal transcription of the group 2 capsule gene cluster at high temperatures [4,7]. On the other hand, kps transcription was repressed from PR3 at 20 • C by H-NS [7]. SlyA was able to promote kps transcription from PR3 at low temperatures in an h-ns mutant; however, there was insufficient SlyA to activate transcription in WT [7]. Thus, a critical point in this regulation seems to be the relative concentration of these two proteins. We and others previously found that slyA expression is temperaturedependent, with a reduced expression at low temperatures [4,21]. Since slyA expression was slightly enhanced at low temperatures in an rcsB mutant, we speculate that this increment may overcome H-NS repression and explain the increment in kps transcription at 19 • C in the rcsB mutant (Table 5). Furthermore, the rcsB mutant showed a great increase in dsrA expression at 19 • C. dsrA is an RNA molecule in which commencement and stability of transcription are higher at low temperatures [15]. dsrA RNA binds to h-ns RNA and blocks its translation [19], which would allow SlyA to promote kps transcription. Recently, it was found that the heterodimer RcsB-BglJ strongly activates leuO transcription [43], a known antagonist of H-NS [44,45]. Thus, we propose that RcsB may repress kps transcription at low temperatures by tipping the balance in favour of SlyA.
Overall, these results suggest that the Rcs phosphorelay, and particularly the RcsB regulator, are global keys in the capacity of E. coli K92 to adapt to different environments either inside or outside host cells, and they provide a better understanding of the complex thermoregulatory network governing capsule synthesis in E. coli.

AUTHOR CONTRIBUTION
All authors contributed actively and similarly in the development of this work. Nicolás Navasa has been mainly responsible for the experimental development.