Correspondence:Öjar Melefors, Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet, SE-17177 Stockholm, Sweden. Tel.: +46 8 45 72 414; fax: +46 8 302 566; e-mail: firstname.lastname@example.org
The carbon storage regulatory (Csr) system is a complex network controlling various phenotypes in many eubacteria. So far, the external conditions by which the system is regulated are poorly understood. Here we show that the expression of the two noncoding small RNAs CsrB and CsrC in Escherichia coli is strongly increased in cultures grown in minimal medium. Addition of tryptone, casamino acids or a mixture of amino acids to a culture grown in minimal medium led to a rapid reduction in the levels of CsrB. Based on this we propose that the expression of the Csr sRNAs is controlled by the amino acid availability in the growth medium.
The ability of bacteria to rapidly adapt to changing environmental conditions largely depends on complex regulatory networks that orchestrate cellular changes in response to external stimuli. One such complex network is the carbon storage regulatory (Csr) system, which is widely distributed among eubacteria and controls various global responses in bacterial metabolism and physiology (Babitzke & Romeo, 2007; Lapouge et al., 2008; Lucchetti-Miganeh et al., 2008). The central player of the Csr system is the RNA-binding protein, CsrA, which post-transcriptionally down- or upregulates the expression of target genes. In Escherichia coli CsrA was found to control a broad range of phenotypes, such as biofilm formation, motility as well as many functions in metabolism, including glycolysis, gluconeogenesis and acetate and glycogen metabolism (Romeo, 1998). Nevertheless, only a few direct target genes of CsrA have to this date been identified. CsrA has been shown to directly downregulate the expression of the glgCAP operon (Baker et al., 2002), encoding the glycogen synthesis apparatus, the cstA gene (Dubey et al., 2003), involved in carbon starvation, the pga operon, encoding the biofilm polysaccharide poly-β-1,6-N-acetyl-d-glucosamine (Wang et al., 2005), the RNA chaperone gene hfq (Baker et al., 2007), as well as ycdT and ydeH, both of which encode GGDEF domain proteins with functions in the turnover of the second messenger c-di-GMP (Jonas et al., 2008). The only target, which has been demonstrated to be directly upregulated by CsrA is flhDC, encoding the master regulator of flagella synthesis (Wei et al., 2001).
The global role of CsrA in metabolism and physiology requires a tight regulation of its activity in response to environmental conditions. In E. coli, regulation of CsrA is mediated by the two noncoding small RNAs (sRNAs), CsrB and CsrC, which were demonstrated to antagonize CsrA activity by sequestering the protein (Liu et al., 1997; Romeo, 1998; Weilbacher et al., 2003). The levels of the sRNAs are positively controlled through direct transcriptional activation of csrB and csrC expression by the response regulator UvrY (Suzuki et al., 2002; Weilbacher et al., 2003), which together with the tripartite histidine kinase BarA comprises a two-component system (TCS) (Pernestig et al., 2001). Notably, activation of csrB and csrC transcription only occurs in the presence of CsrA. Thus, CsrA controls its own activity by an autoregulatory mechanism (Gudapaty et al., 2001; Suzuki et al., 2002). Besides BarA, UvrY and CsrA, another factor has recently been identified to modulate the levels of the CsrB and CsrC sRNAs, called CsrD (YhdA) (Jonas et al., 2006; Suzuki et al., 2006). The membrane anchored CsrD protein, which contains unconventional GGDEF and EAL domains, was demonstrated to target CsrB and CsrC for RNAseE-mediated degradation, thereby acting as a negative regulator of the sRNAs (Suzuki et al., 2006).
Although the key players of the Csr system have been identified, the external conditions by which the system is regulated are poorly understood. Signal integration into the Csr system is predicted to occur in part through the BarA sensor. Despite its extended periplasmic loop, which might possess sensing functions, the molecular stimuli to which BarA responds are still unknown. However, it has been shown that BarA sensing is pH dependent (Mondragon et al., 2006) and that expression of csrB and csrC is activated in monocultures at the entry into stationary phase (Gudapaty et al., 2001; Weilbacher et al., 2003). Based on our previous finding that BarA-UvrY is needed for efficient switching between glycolytic and gluconeogenic carbon sources, we earlier proposed that a stimulus might reflect the energy/growth status of the cell (Pernestig et al., 2003). These findings encouraged us to explore the effect of different growth media on csrB and csrC expression.
Materials and methods
Strains, plasmids and growth conditions
All strains and plasmids used in this study are listed in Table 1. Bacteria were grown in Luria–Bertani broth (LB) or in M9 minimal medium (MM), which was prepared according to the manufacturer's protocol (Sigma-Aldrich) and supplemented with 0.5% glucose, 50 mM sodium pyruvate, 50 mM succinate or 50 mM sodium acetate. All media used in this study were buffered with 100 mM MOPS and the pH was adjusted to pH 7.0. If required, antibiotics were added in the following concentrations: ampicillin 100 μg mL−1, chloramphenicol 30 μg mL−1 and kanamycin 50 μg mL−1. Tryptone and casamino acids were used in the final concentration of c. 10 g L−1. Amino acids (Sigma-Aldrich) were mixed in a composition, which corresponds approximately to the composition of a casamino hydrolysate: 2.8% Ala, 3.6% Arg, 6.3% Asp, 0.3% Cys, 2.1% Glu, 2.7% His, 5.6% Ile, 8.4% Leu, 7.5% Lys, 2.7% Met, 4.6% Phe, 9.9% Pro, 5.6% Ser, 4.2% Thr, 1.1% Trp, 6.1% Tyr, 5.0% Val, 2.2% Gly. Tryptone, casamino acids and the mixture of amino acids were prepared as concentrated stock solutions and were added in small volumes (<200 μL) to maintain the culture density.
Bacterial cultures were mixed with two volumes of RNAprotect Bacterial Reagent (Qiagen) and incubated for 5 min at room temperature. Total cellular RNA was subsequently prepared using the RNeasy Mini Kit with on-column DNA digestion (Qiagen). RNA concentrations were determined using the NanoDrop® ND-1000 UV-Vis Spectrophotometer (NanoDrop Technologies, Wilmington, DE).
Five hundred nanograms of total RNA were used to synthesize cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Template, 0.1 ng, was used for the real-time PCR reaction using the Power SYBR Green PCR Master Mix (Applied Biosystems) and primers that were designed using the primer express software v3.0 (Applied Biosystems). Analysis was performed with an ABI 7500 Real Time PCR System (Applied Biosystems) using the standard run mode of the instrument. For detection of primer dimerization or other artefacts of amplification, a dissociation curve was run immediately after completion of the real-time RT-PCR. Individual gene expression profiles were normalized against the rrnD gene (16S rRNA gene), serving as an endogenous control. All results were analysed using the 7500 sds software v1.3.1 (Applied Biosystems) and further prepared using excel (Microsoft). The data values presented in all figures represent the mean expression level of quadruplicates from one real-time RT-PCR assay, relative to a calibrator value (LB exponential phase, wild type or time point, 0 min). The error bars represent the SEM expression level calculated by the sds software using the confidence value 95%.
β-Galactosidase activity was measured in 10-min reactions using the Miller protocol (1972). One hundred microliters of bacterial culture was used for each reaction. Each measurement was carried out in duplicates, from which mean and SD were calculated. The entire experiment was repeated twice with similar results.
Western blot analysis
Bacteria were grown to an OD600 nm of 0.5 and subsequently lysed by freeze thawing and treatment with Triton X-100 (20%). After determining the amount of total protein using the Bradford standard method, the normalized protein samples were subjected to electrophoresis in a sodium dodecyl sulphate-15% polyacrylamide gel and electrotransferred onto a 0.2-μm-pore size nitrocellulose membrane. Immunodetection using 5000-fold-diluted rabbit antiserum raised against purified CsrA–CsrB complex (Gudapaty et al., 2001) was performed according to a standard protocol.
CsrB and CsrC levels are strongly increased in minimal medium
Growth rates of E. coli MG1655 were monitored in buffered LB and MM, supplemented with glucose, sodium pyruvate, succinate or sodium acetate at 37 °C (Fig. 1a). To measure the levels of the CsrB and CsrC sRNAs in the different media by quantitative real-time RT-PCR over growth, total RNA was sampled during the exponential phase (OD600 nm 0.4–0.6), at the entry into the stationary phase (OD600 nm 1.3–1.7, for acetate OD600 nm 1.0) and during the late stationary phase (after the OD600 nm has reached the maximum value). In accordance with previous data (Gudapaty et al., 2001; Weilbacher et al., 2003), CsrB and CsrC RNA levels increased at the entry of the stationary phase, before slightly decreasing in the late stationary phase in cultures grown in LB medium. Strikingly, when the bacteria were cultured in minimal medium with the different carbon sources, the levels of both CsrB and CsrC were drastically elevated at all three time points (Fig. 1b and c). The observed effect was more pronounced for csrB than for csrC, supporting the earlier hypothesis that csrB and csrC are differentially regulated (Weilbacher et al., 2003; Jonas et al., 2006). Although the CsrB and CsrC RNA levels were lower in MM with glucose than in MM with pyruvate, succinate or acetate, we could not detect a direct correlation between the growth rate and the extent of csrB/C expression.
We also monitored the expression of a chromosomal Φ(csrB-lacZ) transcriptional fusion, containing the region from −242 to +4 bp of the csrB gene. In agreement with the data obtained by quantitative real-time RT-PCR, csrB-lacZ expression was strongly elevated in minimal medium compared with LB (Fig. 1d). However, the data from both approaches did not entirely correlate. In MM with succinate CsrB RNA levels were comparably high (Fig. 1b), but the expression of csrB-lacZ was in this medium lower than in MM with the other carbon sources (Fig. 1d). This observation might be explained with the earlier finding that CsrB RNA levels are regulated not only transcriptionally, but also at the level of RNA stability by CsrD (Suzuki et al., 2006).
Other Csr components are only moderately affected by the medium
In contrast to csrB and csrC, the mRNA levels of csrA, encoding the molecular target for CsrB and CsrC, were only affected on a minor scale by the growth medium. In MM with glucose, pyruvate or succinate, csrA mRNA was slightly increased, whereas growth MM with acetate caused a faint decrease in csrA mRNA levels (Fig. 2a). In agreement with these data, also the levels of CsrA protein were somewhat elevated in minimal medium with glucose, pyruvate and succinate, and reduced in minimal medium with acetate (Fig. 2b). Earlier data have demonstrated that in complex medium, the levels of CsrA exceed the binding capacity of the CsrB and CsrC sRNAs (Gudapaty et al., 2001). It is therefore possible that the elevated levels of CsrB and CsrC in nutrient-poor media lead to a more efficient sequestration of CsrA. We also measured the expression of barA, uvrY and csrD in the different media. The mRNA levels of barA and csrD showed an increase in MM with glucose and pyruvate. Otherwise the expression of these genes was hardly affected by the growth medium (Fig. 2a).
Dependence on the BarA-UvrY TCS
In agreement with earlier results (Gudapaty et al., 2001; Suzuki et al., 2002, 2006; Jonas et al., 2006), csrB expression was strongly reduced in strains mutated in uvrY, barA or csrA, when grown in rich medium to the early stationary phase. Disruption of csrD led as expected to slightly increased CsrB levels (Fig. 3a) (Suzuki et al., 2006). We observed similar results in cultures grown in MM with glucose (Fig. 3b), although the effect of a mutation in csrD was more pronounced than in LB medium. When the bacteria were grown in MM with pyruvate, succinate or acetate, we measured higher residual CsrB levels in the uvrY and barA mutants (Fig. 3c–e). These data show that in poor medium the levels of the CsrB sRNA depend, as in LB, on BarA, UvrY and CsrA. However, in some media this dependence is less strong. It might be possible that under certain conditions one or several other factors might influence csrB expression. In several other studies it has already been suggested that additional so far unidentified factors are involved in the regulation of the Csr regulatory system (Suzuki et al., 2002; Weilbacher et al., 2003; Jonas et al., 2006).
Notably, under each of the tested conditions mutants in barA and uvrY did not behave completely identical with respect to csrB expression. A mutation in uvrY led to a somewhat more pronounced reduction in csrB expression than a barA mutation, suggesting that under certain conditions the response regulator UvrY can respond even in the absence of the BarA sensor. This agrees with the earlier finding that in the absence of BarA, UvrY can be phosphorylated by acetyl-phosphate and that overexpression of UvrY can override the effect of a barA mutation but not vice versa (Suzuki et al., 2002; Tomenius et al., 2005; Jonas et al., 2006). It might also be possible that UvrY is activated by an additional sensor, as observed for GacA, the UvrY ortholog in Pseudomonas. GacA responds to the BarA ortholog GacS as well as to two other sensors, RetS and LadS (Ventre et al., 2006). Thereby, GacA is suggested to integrate multiple signals.
In a previous study variants of the BarA sensor have been isolated with an impaired kinase activity and a net dephosphorylating activity due to point mutations in the HAMP linker region (Tomenius et al., 2005). Some of these plasmid-borne BarA mutants could counteract the activity of the wild-type sensor in a dominant-negative fashion. We used one such mutant expressed from pHTbar7 to test whether modulation of the activity of BarA as kinase or phosphatase would result in changes in CsrB levels in MM with glucose, a medium in which csrB expression greatly depended on functional uvrY and barA alleles (Fig. 3b). Our results demonstrate that expression of the BarA phosphatase variant in the wild-type background resulted in as low CsrB sRNA levels as in the barA deletion mutant (Fig. 3f). From this we conclude that csrB expression in MM with glucose depends on the kinase activity of BarA.
Addition of rich medium or amino acids causes CsrB levels to decrease
Our data demonstrate that csrB and csrC expression is high when the bacteria grow in a poor medium. To examine whether this effect is reversible upon addition of complex medium we grew cultures of MG1655 in MM with glucose to OD600 nm 0.5, added subsequently concentrated tryptone broth and followed changes in CsrB sRNA levels over time by quantitative real-time RT-PCR. Within 30 min after addition of tryptone the levels of CsrB decreased significantly (to c. 20%) (Fig. 4). The fact that tryptone is a trypsin digest of casein, led us to examine whether addition of casamino acids, which is a hydrolysate of casein, would also result in a reduction in csrB expression. Indeed, addition of casamino acids led to similar kinetics in CsrB sRNA levels over time as addition of tryptone. Escherichia coli bacteria grown in complex media, such as LB, tryptone or casamino acids, consume amino acids as nutrient source (Wolfe, 2005). To investigate the effect of pure amino acids on csrB expression, a mixture of the purified l-amino acids was made and added to the growing culture. Similar as for tryptone and casamino acids, addition of amino acids led to a clear reduction in CsrB RNA levels (Fig. 4). These data suggest that the availability of amino acids has a repressing effect on the RNA levels of csrB and possibly also csrC.
Despite the global role of the Csr system in bacterial metabolism and physiology, only little is known about the environmental conditions controlling the regulatory system. Here we show that CsrB and CsrC sRNA levels are controlled by the nutrient availability of the growth medium. In nutrient-poor minimal medium CsrB and CsrC sRNA levels were high in contrast to complex LB medium, in which csrB and csrC expression was low. This finding agrees well with the earlier documented observation that csrB levels increase when the bacteria enter stationary phase (Gudapaty et al., 2001; Suzuki et al., 2002), when the growth medium gets depleted.
A previous work suggested that BarA-UvrY is needed for efficient switching between glycolytic and gluconeogenic carbon sources (Pernestig et al., 2003). This hypothesis was based on findings from competition experiments, in which mutants in uvrY or barA showed a clear growth advantage over wild-type bacteria in media with gluconeogenic carbon sources. In contrast, media with glycolytic carbon sources led to a distinct growth advantage of the wild type. On the basis of these earlier findings we have investigated the effect of minimal media with different carbon sources on csrB and csrC expression. We found that CsrB and CsrC levels were, in general, considerably higher in minimal medium compared with LB (Fig. 1). The identity of the carbon source, whether glycolytic or gluconeogenic, seemed to be less important.
Upon addition of nutrient-rich tryptone broth to a culture grown in minimal medium with glucose we found that CsrB levels rapidly decreased (Fig. 4). Likewise, also the addition of casamino acids or a mixture of pure amino acids led to a clear reduction in CsrB levels. Based on this we suggest that csr expression is regulated in response to the availability of amino acids in the growth medium. The availability of amino acids is known to control various metabolic and signalling pathways. One of the best described examples is the stringent response, involving the small signalling molecule ppGpp that accumulates in response to amino acid starvation and other stress conditions in many bacteria (recent reviews by Magnusson et al., 2005; Potrykus & Cashel, 2008; Srivatsan & Wang, 2008). Also several other signalling molecules or metabolic intermediates, including inorganic polyphosphate or thiamine triphosphate, accumulate in response to changes in amino acid availability (Ault-Riche et al., 1998; Lakaye et al., 2004; Wolfe, 2005). It might be possible that the observed effect of amino acids on csrB/C expression is mediated by such small signalling molecules or metabolic products. In fact, in Legionella pneumophila, ppGpp-mediated stringent response coupled with the response to an excess in short chain fatty acids was shown to control differentiation and virulence in a pathway dependent on the LetS–LetA TCS, an orthologous system to BarA-UvrY (Hammer et al., 2002; Edwards et al., 2009). In E. coli mutations in the genes relA and spoT, both of which are important for ppGpp production, did not affect csrB-lacZ expression in LB (Jonas et al., 2006). Nevertheless, we do not exclude that similar pathways might exist in E. coli as in L. pneumophila. The expression of csrB depended, especially in LB and MM with glucose, on functional uvrY and barA alleles (Fig. 3). Moreover, in MM with glucose csrB expression required the kinase activity of BarA. We therefore consider it likely that the observed effect of the growth medium on csrB and csrC expression is at least in part mediated by the BarA-UvrY TCS. In poor medium the kinase activity of BarA might be favoured leading to an increased transcription of the sRNA genes. In contrast, complex medium containing amino acids might reduce BarA's kinase activity (or enhance its phosphatase activity) leading to decreased transcription of csrB and csrC. It is possible that BarA senses the availability of nutrients directly. Alternatively, BarA might be controlled by an endogenously produced signal, such as a small molecule or metabolite (as discussed above), which reflects the metabolic status of the cells and is produced in a medium-dependent manner. Such a quorum sensing-like activation of the regulatory system has already been observed for pseudomonads and vibrios, which excrete, when grown to high population densities, unidentified signal molecules that activate the GacS/GacA system (Dubuis & Haas, 2007).
We would like to thank Tony Romeo for providing the Anti-CsrA serum and Karina Hentrich for excellent technical assistance. This work was supported by a grant from the Marie Curie Early Stage Training Fellowship of the European Community's Sixth Framework Program (MEST-CT-2004-8475).