Regulatory Circuitry of the CsrA/CsrB and BarA/UvrY Systems of Escherichia coli

Abstract

The global regulator CsrA (carbon storage regulator) is an RNA binding protein that coordinates central carbon metabolism, activates flagellum biosynthesis and motility, and represses biofilm formation in Escherichia coli. CsrA activity is antagonized by the untranslated RNA CsrB, to which it binds and forms a globular ribonucleoprotein complex. CsrA indirectly activates csrB transcription, in an apparent autoregulatory mechanism. In the present study, we elucidate the intermediate regulatory circuitry of this system. Mutations affecting the BarA/UvrY two-component signal transduction system decreased csrB transcription but did not affect csrA′-′lacZ expression. The uvrY defect was severalfold more severe than that of barA. Both csrA and uvrY were required for optimal barA expression. The latter observation suggests an autoregulatory loop for UvrY. Ectopic expression of uvrY suppressed the csrB-lacZ expression defects caused by uvrY, csrA, or barA mutations; csrA suppressed csrA or barA defects; and barA complemented only the barA mutation. Purified UvrY protein stimulated csrB-lacZ expression approximately sixfold in S-30 transcription-translation reactions, revealing a direct effect of UvrY on csrB transcription. Disruption of sdiA, which encodes a LuxR homologue, decreased the expression of uvrY′-′lacZ and csrB-lacZ fusions but did not affect csrA′-′lacZ. The BarA/UvrY system activated biofilm formation. Ectopic expression of uvrY stimulated biofilm formation by a csrB-null mutant, indicative of a CsrB-independent role for UvrY in biofilm development. Collectively, these results demonstrate that uvrY resides downstream from csrA in a signaling pathway for csrB and that CsrA stimulates UvrY-dependent activation of csrB expression by BarA-dependent and -independent mechanisms.

To persist in nature, bacteria must be able to compete and survive under various growth conditions. To accomplish this task, they possess regulatory systems that permit them to recognize and adapt to a changing environment. In Escherichia coli and related species, the transition from exponential growth to stationary-phase growth is accompanied by striking physiological changes, which produce cells that are more stress resistant, slower metabolizing, and better at scavenging nutrients (23, 25). These adaptations are brought about largely through changes in gene expression that are coordinated through global regulatory networks (19, 39). The present study investigates interactions among three different types of global regulatory systems that affect stationary-phase gene expression.

The global regulatory system Csr (carbon storage regulator) represses a variety of stationary-phase genes (reviewed in reference 45). The central component of this system, CsrA, is a 61-amino-acid RNA binding protein. This protein inhibits glycogen biosynthesis and catabolism, gluconeogenesis, and biofilm formation, whereas it activates glycolysis, acetate metabolism, motility, and flagellum biosynthesis in E. coli (27, 47, 49, 58, 59, 63). The dramatic effect of CsrA on biofilm formation is mediated primarily through its regulatory role in directing glycogen biosynthesis and catabolism (27). Homologues of csrA exhibit a broad phylogenetic distribution in the eubacteria (45), repress stationary-phase genes of Pseudomonas fluorescens (8) and genes involved in plant pathogenesis in Erwinia carotovora (11), and regulate genes involved in mucosal invasion by Salmonella enterica (4, 5).

CsrA is capable of posttranscriptional repression or activation, depending upon the particular RNA target. The mechanism by which CsrA represses glycogen synthesis in E. coli has been examined in considerable detail. CsrA binds to the untranslated leader of the glgCAP transcript, which encodes enzymes required for glycogen synthesis, at a site that overlaps the glgC Shine-Dalgarno sequence and a second site within a hairpin that is located upstream of the Shine-Dalgarno sequence (6). Thus, CsrA blocks ribosome binding and inhibits the initiation of glgC translation. Inhibition of translation probably contributes to the observed destabilization of glgCAP mRNA by CsrA (29, 30). CsrA positively regulates motility in E. coli by binding to and stabilizing the flhDC transcript, which encodes the subunits of a tetrameric DNA binding protein (FlhD₂C₂) that activates the expression of genes involved in flagellum biosynthesis, motility, and chemotaxis (59).

A second component of Csr is the 366-nucleotide untranslated CsrB RNA, which binds to ∼18 CsrA subunits, forming a large globular ribonucleoprotein complex (31). In vitro transcription-translation studies of glgCAP expression and in vivo csrB disruption and overexpression studies have revealed that CsrB RNA functions as an antagonist of CsrA, apparently by sequestering this protein (20, 30, 31). A highly repeated sequence element that is located in the loops of predicted CsrB hairpins and is related to the sequences involved in glgCAP recognition sites (6) probably mediates the binding of CsrA to CsrB. The function of CsrB RNA as an antagonist of an mRNA binding global regulatory protein offers a novel paradigm for posttranscriptional control by prokaryotic regulatory RNA molecules (reviewed in references 14 and 57). We recently demonstrated that CsrA indirectly activates csrB transcription, indicative of an autoregulatory mechanism that determines the intracellular activity of CsrA without affecting its level (20).

Bacterial adaptation to environmental changes depends heavily upon two-component signal transduction systems. These typically consist of a membrane-bound sensor kinase which detects environmental changes and its cognate response regulator, which is phosphorylated by the kinase and thereby activated for specific DNA binding (24). Data from the E. coli K-12 genome sequence indicate that it possesses approximately 30 typical two-component systems (35). Recent studies have begun to reveal a relationship between homologues of the BarA/UvrY two-component system and the CsrA/CsrB system. HCN biosynthesis (hcnABC) and extracellular protease (aprA) in P. fluorescens CHA0 are regulated indirectly by GacA, a homologue of UvrY, via a posttranscriptional mechanism involving RsmA, a homologue of CsrA, and a potential CsrB homologue, RsmZ (8, 21). PrrB RNA, a 132-nucleotide transcript in P. fluorescens F113 that is similar to CsrB RNA, is itself regulated by GacS/GacA (1). The barA gene of S. enterica positively affects the expression of hilA, which encodes a regulator of genes of Salmonella pathogenicity island 1 (SPI1) (4). Mutations in csrA or csrB also affect hilA expression (4, 5). Furthermore, gacS and gacA (alternatively called expS and expA) of E. carotovora affect levels of RsmB, a homologue of CsrB (12, 26). BarA/UvrY-homologous systems in gram-negative pathogens control a variety of virulence functions (21, 42).

The sensor kinase BarA of E. coli was identified as a multicopy suppressor of an envZ defect in the expression of outer membrane proteins (38) and was recently shown to activate the transcription of rpoS, which encodes the stationary-phase sigma factor RpoS or σ^s (37). BarA is a member of the subclass of tripartite sensor kinases. These proteins consist of an N-terminal cytosolic domain, a canonical pair of transmembrane regions linked by a periplasmic bridge, a transmitter domain containing a conserved histidine residue, a central receiver domain with a conserved aspartate residue, and a C-terminal phosphotransfer domain with a conserved histidine residue (38, 42). Tripartite kinases catalyze the phosphorylation of their cognate response regulators via an ATP-His-Asp-His-Asp phosphorelay (18, 28). While the sensory mechanism of BarA remains unresolved, the barA gene has been implicated in the virulence of uropathogenic E. coli in the urinary tract (64; A.-K. Pernestig et al., submitted for publication). BarA was recently demonstrated to be a cognate kinase of UvrY, a response regulator of the FixJ family (42). The uvrY gene is cotranscribed with uvrC, which encodes a DNA repair enzyme, although UvrY plays no apparent role in DNA repair (36).

Comprehensive transcript profiling recently revealed that increased sdiA gene dosage causes uvrY transcript levels to increase by about 10-fold (60). The sdiA gene encodes a protein of the LuxR family, the members of which contain one domain for binding N-acylated homoserine lactones (AHL) and a second domain for binding DNA (54, 56). These proteins permit the bacterium to sense and respond to the resident microbial population by binding to AHL beyond a threshold concentration and activating or repressing the transcription of target genes, i.e., they mediate quorum sensing (17). E. coli and Salmonella are not known to synthesize AHL and there are no apparent AHL synthase genes in their genomes (33). However, SdiA of S. enterica has been proposed to be an AHL receptor that detects signals from other species (3, 33).

Here, we examine regulatory interactions of BarA/UvrY, CsrA/CsrB, and SdiA of E. coli. The UvrY response regulator of E. coli directly activates transcription of csrB and mediates the indirect effects of CsrA on csrB. The BarA sensor kinase and the DNA binding protein SdiA also regulate csrB transcription, apparently through effects on UvrY phosphorylation and uvrY expression, respectively. Autoregulatory loops characterize these systems. CsrA activates transcription of its RNA antagonist, CsrB (20), and we show that UvrY stimulates expression of barA, which encodes its cognate sensor kinase. A simple model for this regulatory circuitry is presented and discussed.

MATERIALS AND METHODS

Strains, plasmids, and phage.

The bacterial strains, plasmids, and bacteriophage used in this study are listed in Table 1.

TABLE 1.

Bacterial strains, plasmids, and phages used in this study

Strain, plasmid, or phage	Description	Source or reference
Strains
AKP014	MC4100 barA::kanR	42
AKP023	MC4100 uvrY::cam	42
BA MG1655a	MG1655 barA::kanR (from AKP014)	This study
CAG18642	zfh-3131::Tn10; 57.5 min (near csrA)	52
CF7789	MG1655 ΔlacIZ (MluI)	Michael Cashel
DHB6521	SM551 λInCh1 (Kanr)	9
HS703	MC4100 barA::λplacMu53 [φ(barA-lacZ)1010]	37
KSA712	CF7789 Δ(att-lom)::bla φ(csrA′-′lacZ)1(Hyb) Ampr Kans	27
KSB837	CF7789 Δ(att-lom)::bla φ(csrB-lacZ)1(Hyb) Ampr Kans	20
KSY009	CF7789 Δ(att-lom)::bla φ(uvrY′-′lacZ)1(Hyb) Ampr Kans	This study
KSGA18	CF7789 φ(glgA::lacZ) (λplacMu15)	20
MG1655	Prototrophic	Michael Cashel
RG1-B MG1655a	csrB::cam	20
SM551(= DHB6501)	F− λ− λs Δlac(MS265)mel NalArsupF58 (= suIII+)	9
TR1-5 MG1655a	csrA::kanR	47
UY CF7789a	CF7789 uvrY::cam (from AKP023)	This study
WX2	Δlac-pro met pro zzz::Tn10 thy supD rK− mK−sdiA::kanR	56 (L. Rothfield)
Plasmids
pCBZ1	pGE593 φ(csrB-lacZ)	20
pCZ3-3	pMLB1034 φ(glgC′-′lacZ)	46
pUZ9	pMLB1034 φ(uvrY′-′lacZ)	This study
pCSR10	Minimal csrA in pUC19	47
pCRA16	csrA in blunt-ended VspI site of pBR322, Tetr	This study
pCA9505	Carries uvrY gene, Ampr	36
pUY14	uvrY in blunt-ended VspI site of pBR322, Tetr	This study
pBarA	barA in pGEM-T, Ampr	This study
pBA29	barA in blunt-ended VspI site of pBR322, Tetr	This study
pSdiA	sdiA in EcoRI site of pBR322, Ampr Tetr	Robert A. LaRossa
pSPT18-CsrB	Plasmid for the csrB riboprobe synthesis	20
pGE593	Vector for lacZ transcriptional fusions; Ampr	15
pMLB1034	Vector for lacZ translational fusions; Ampr	51
pGEM-T	T-cloning vector, Ampr	Promega
pBR322	Cloning vector; Ampr Tetr	50
Bacteriophages
P1vir	Strictly lytic P1	Carol Gross
λInCh1	For genomic insertions; Kanr	9

^aStrain designations containing the prefix TR or TR1-5, RG or RG1-B, UY, BA, or SA indicate that the mutant allele csrA::kanR, csrB::cam, uvrY::cam, barA::kanR, or sdiA::kanR, respectively, was introduced by P1vir transduction.

Media and growth conditions.

Luria-Bertani medium (34) was used for routine cultures. Kornberg medium (1.1% K₂HPO₄, 0.85% KH₂PO₄, 0.6% yeast extract containing 0.5% glucose for liquid medium) was used to grow cultures for the gene expression assays and Northern blot analysis. M63 medium supplemented with glucose (0.4%); thiamine (5 μg/ml); adenine and thymine (50 μg/ml); calcium pantothenate (1 μg/ml); and serine, glycine, and methionine (100 μg/ml each) was used for the selection of rel⁺ barA mutants (55). Tryptone broth (pH 7.4) contained 1% tryptone and 0.5% NaCl. Colonization factor antigen medium (pH 7.4) contained 1% Casamino Acids, 0.15% yeast extract, 0.005% MgSO₄, and 0.0005% MnCl₂ (16). The following antibiotics were added, as required, at the indicated concentrations: chloramphenicol, 20 μg/ml; kanamycin, 50 μg/ml; ampicillin, 100 μg/ml; and tetracycline, 10 μg/ml, except that ampicillin and kanamycin were used at 50 and 40 μg/ml, respectively, during the construction of the uvrY′-′lacZ fusion, and kanamycin was used at 100 μg/ml for the selection of csrA::kanR strains. All cultures that were used for gene expression assays were grown at 37°C with rapid rotary shaking (48).

Molecular and genetic techniques.

P1vir transduction or cotransduction of resistance markers, subcloning, PCR amplification, and molecular genetic techniques were performed by standard procedures (34, 50).

The plasmid pBarA was constructed by PCR amplification of the barA gene from −270 to the end of the barA coding region from E. coli MC4100 chromosomal DNA with Pfu polymerase and the primers 5′-GAGAATGCATACGCCAAAATGAGGACAG and 5′-GCGGATCCACTCGACAAGACATCCATTA. The resultant product was cloned directly into the pGEM-T vector (Promega) by using the T-overhang, with the barA gene position in a clockwise direction. A 0.5-kb BamHI-EcoRI fragment containing csrA from pCSR10 (48), a 1.3-kb EcoRI-HindIII fragment containing uvrY from pCA9505 (36), and a 3.0-kb NcoI-NotI fragment containing barA from pBarA were treated with the Klenow fragment of DNA polymerase I and were individually subcloned into the blunt-ended VspI site of pBR322 to generate pCRA16, pUY14, and pBA29, respectively. The open reading frames of the above genes are oriented in the same direction as bla in the vector.

Special precautions were required for two of the P1vir transductions of this study, as follows. The barA::kanR mutation was transduced from AKP014 (42) into MG1655 with selection for kanamycin resistance. Because AKP014 is a relA mutant, and relA is separated by only 1.4 kb of DNA from barA, a relA wild-type transductant was selected by virtue of its ability to grow on M63 supplemented medium and was designated BA MG1655. The uvrY::cam mutation was transduced from AKP023 (42) into CF7789. Because AKP023 is a flhD mutant and flhD is located 21 kb away from uvrY, a motile transductant was identified by the plate assay for motility and designated UY CF7789. The other barA or uvrY mutants were generated by P1vir transduction from BA MG1655 or UY CF7789, respectively.

Construction of a chromosomal uvrY′-′lacZ translational fusion.

A 572-bp fragment containing the upstream regulatory region and 12 codons of uvrY was amplified from MG1655 DNA by PCR with the primers 5′-CAGCATCGCTTTCAGGCAGGAGACTTC and 5′-CAGTTCGTGGTCATCAACAAGTAGAACG, treated with T4 DNA polymerase and polynucleotide kinase, and subcloned into the SmaI site of pMLB1034 (51). The resulting plasmid, pUZ9, contained 26 codons of yecF, which is upstream from and divergently oriented with respect to uvrY (7), the complete upstream flanking region of uvrY, 12 codons of the uvrY coding region, and an in-frame uvrY′-′lacZ translational fusion. DNA sequencing was performed to confirm the presence of the correct fusion and the absence of PCR-generated mutations. The uvrY′-′lacZ fusion in pUZ9 was moved into the E. coli CF7789 chromosome by using the λInCh1 system as described elsewhere (9). The resulting strain, KSY009, which was chosen for subsequent studies, was Amp^r Kan^s and was no longer temperature sensitive. The presence of the uvrY′-′lacZ translational fusion was confirmed by PCR analysis, as recommended elsewhere (9). Oligonucleotide primers used in this study were synthesized by Integrated DNA Technologies Inc., Coralville, Iowa.

Motility assay.

The plate assay was initiated by stabbing a colony from an overnight culture into semisolid agar (tryptone broth or colonization factor antigen medium solidified with 0.35% agar). The plates were kept in a humidified incubator at 30°C and examined at ∼16 h of growth (59).

β-Galactosidase and total protein assays.

β-Galactosidase activity was assayed in 10-min reactions, as described previously (46). Total protein was measured by the bicinchoninic acid method with bovine serum albumin as the protein standard (53).

Purification of His₆-tagged UvrY.

His₆-tagged UvrY was purified as described previously, except that purified protein was dialyzed against 10 mM Tris-acetate (pH 8.0) containing 25% glycerol and concentrated in Centricon 10 units (Amicon) (42).

In vitro transcription-translation.

Effects of UvrY protein on csrB-lacZ expression were examined by using S-30 extracts prepared from a uvrY mutant strain (UY CF7789), as previously described (31, 48), except that reaction volumes were scaled down to 28 μl. Radiolabeled proteins were detected by fluorography with sodium salicylate (10), and methionine incorporation into the LacZ polypeptide was quantified by liquid scintillation counting of H₂O₂-solubilized gel sections (48).

Quantitative biofilm assay.

The quantitative analysis of biofilm formation in microtiter plates by crystal violet staining was previously described in detail (27).

Northern hybridization.

RNA isolation, riboprobe synthesis, and Northern blotting were conducted essentially as described elsewhere (20). RNA (5 μg) was separated on formaldehyde agarose (1.2%) gels, blotted by capillary transfer onto positively charged nylon membranes (Boehringer Mannheim), and immobilized by being baked at 120°C for 30 min. A digoxigenin-labeled csrB riboprobe was synthesized from pSPT18-CsrB by using the DIG Luminescent Detection kit for nucleic acids, as described by the manufacturer (Boehringer Mannheim). The blot was prehybridized and hybridized at a probe concentration of 50 ng/ml with Perfecthyb Plus hybridization buffer (Sigma Chemical). Signal detection used the commercial protocol (Boehringer Mannheim), except that incubation in blocking solution was extended for an additional 10 h. Chemiluminescent signals were detected with Kodak X-Omat AR film. In addition, signals were quantified by phosphorimaging with a GS-525 PhosphorImager (Bio-Rad). Phosphorimaging data were analyzed with Molecular Analyst (version 2.1.2) software (Bio-Rad). Prior to blotting, the gels were stained with ethidium bromide and rRNA bands were photographed. The resulting signals were quantified by densitometry, and values were used to correct for minor variations in sample loading.

RESULTS

Effects of uvrY and barA on the in vivo expression of csrB.

Northern hybridization was used to determine if UvrY or BarA affects CsrB RNA levels (Fig. 1). RNA was isolated from MG1655 and its isogenic uvrY and barA mutants at ∼2 h post-exponential phase, which is optimal for CsrB accumulation (20). CsrB RNA levels were decreased ∼60% in the barA mutant and ∼98% in the uvrY mutant relative to the parent strain. These results suggested that UvrY is an important regulator of csrB expression and that BarA influences csrB expression to a lesser extent.

FIG. 1.

Northern analysis of CsrB levels in E. coli K-12 MG1655 and isogenic barA- and uvrY-knockout mutants. RNA from cultures harvested at 2 h post-exponential phase of growth was probed for the CsrB transcript. (A) Blot representative of the observed results. An overexposed film is shown to reveal CsrB RNA from the uvrY mutant. (B) Quantification of the signal by phosphorimage analysis. Signals were normalized with respect to MG1655, which was assigned a value of 100%. Each bar represents the mean value obtained from two independent experiments. Where error bars depicting the standard deviations are not apparent, the standard deviations were below the resolution of the graph.

To further determine if uvrY and barA regulate expression of csrA and csrB, expression from chromosomal csrA′-′lacZ translational or csrB-lacZ transcriptional fusions was examined in wild-type or uvrY or barA mutant strains. The specific β-galactosidase activity from the csrA′-′lacZ fusion was not altered by either mutation (Fig. 2A), while the activity from the csrB-lacZ fusion was dependent upon both uvrY and barA (Fig. 2B). The uvrY mutation reduced csrB-lacZ expression ∼95% or 20-fold, whereas the barA mutant exhibited a decrease of ∼70%. BarA is a member of the subclass of tripartite sensor kinases, and UvrY has been shown previously to be the cognate response regulator for BarA (42). These results suggest that BarA-phosphorylated UvrY activates transcription of the csrB gene and that either unphosphorylated UvrY can activate csrB expression to a lesser extent or UvrY is activated by an alternative phosphoryl donor.

FIG. 2.

Effects of the uvrY or barA null mutations on expression from csrA′-′lacZ translational or csrB-lacZ transcriptional fusions. β-Galactosidase specific activities expressed from csrA′-′lacZ in strains KSA712 (wild type), UY KSA712 (uvrY::cam), and BA KSA712 (barA::kanR) (A) and from csrB-lacZ in strains KSB837 (wild type), UY KSB837 (uvrY::cam), and BA KSB837 (barA::kanR) (B) are shown. β-Galactosidase activities in the wild-type, uvrY::cam, or barA::kanR strain are shown as closed circles, squares, or triangles, respectively. Growth (A₆₀₀) of the respective strains is depicted by open circles, squares, and triangles, respectively. This experiment was repeated in its entirety with essentially identical results.

Effects of csrA and uvrY on barA expression.

To further examine the regulatory interactions of the CsrA/CsrB and BarA/UvrY systems, the effects of CsrA and UvrY on barA expression were examined by monitoring expression of a chromosomal barA-lacZ transcriptional fusion in wild-type or csrA or uvrY mutant strains. The wild-type strains exhibited approximately twofold-greater β-galactosidase activity than did their isogenic csrA (Fig. 3A) or uvrY (Fig. 3B) mutants. We note that because HS703 was kanamycin resistant, this csrA isogenic strain pair was constructed by cotransduction of the csrA::kanR mutation along with a closely linked tetR marker from TR1-5 CAG18642 into HS703 (barA-lacZ). Tet^r transductants were screened for the csrA glycogen phenotype to distinguish csrA wild-type and mutant colonies, and isogenic csrA mutant and wild-type strains, which each contained the tetR marker that was used for cotransduction, were compared in the assays (Fig. 3A). We note that the tetR mutation itself increased barA-lacZ expression approximately twofold (compare the isogenic parent strains, CAG HS703 and HS703, in Fig. 3A and B). The gene disrupted by the tetR marker is srlD (gutD) and encodes the glucitol-6-phosphate dehydrogenase of the glucitol operon (40, 61). The basis of its effect on barA expression is unknown. In conclusion, UvrY and CsrA each exhibit modest stimulation of barA expression.

FIG. 3.

Effects of csrA and uvrY mutations on expression from a barA-lacZ transcriptional fusion. (A) β-Galactosidase specific activities expressed from a barA-lacZ fusion in strains CAG HS703 (wild type) and TR1-5 CAG HS703 (csrA::kanR) are shown as closed and open circles, respectively. (B) Activities in strains HS703 and UY HS703 (uvrY::cam) are shown as closed and open circles, respectively. In each panel, growth (A₆₀₀) of the respective strains is depicted by closed or open squares. This experiment was repeated in its entirety with essentially identical results.

Effects of csrA, barA, and uvrY on uvrY expression.

To determine if the expression of uvrY is influenced by csrA, we examined expression from a uvrY′-′lacZ translational fusion in csrA wild-type and mutant strains. Expression from this gene fusion was not affected by a csrA mutation (data not shown). Overexpression of csrA from a plasmid resulted in only slight activation of this fusion, which is not likely to be biologically relevant. Similarly, the disruption of barA or uvrY had no effect on the expression of this uvrY′-′lacZ fusion (data not shown).

Complementation studies: effects of ectopic expression of csrA, uvrY, or barA on csrB-lacZ expression in csrA, uvrY, and barA mutants.

We previously showed that CsrA is a strong activator of csrB transcription and provided evidence that this effect was indirect (20). Since uvrY, and to a lesser extent barA, activates csrB expression (Fig. 1 and 2) and csrA modestly stimulates barA expression (Fig. 3), complementation studies were conducted to further delineate the regulatory circuitry of this system. We found that multicopy plasmids containing either csrA or uvrY could restore csrB-lacZ expression in a csrA mutant background (Fig. 4). We also found that only uvrY could restore csrB-lacZ expression in a uvrY mutant (Fig. 4). Finally, barA, uvrY, or csrA was able to enhance csrB-lacZ expression in a barA strain background (Fig. 4). These results are consistent with a late or perhaps terminal role for UvrY in a signaling pathway to csrB. Because csrA stimulates barA expression (Fig. 3) and csrB expression in a barA mutant (Fig. 4), these results indicate that csrA influences csrB expression through BarA-dependent and -independent mechanisms. In addition, csrA has no effect on csrB expression in the uvrY mutant background, indicating that its role in csrB expression is completely dependent upon UvrY. The simplest model that accounts for all of these findings is that CsrA is involved in both BarA-dependent and -independent pathways for UvrY activation. We acknowledge that more complex explanations have not been formally eliminated. Taken in context, the failure of barA to complement a csrA defect also allows for the possibility that CsrA may affect BarA function, e.g., CsrA may indirectly influence BarA activation.

FIG. 4.

Complementation studies: effects of ectopic expression of csrA (pCRA16), uvrY (pUY14), or barA (pBA29) on expression of a csrB-lacZ transcriptional fusion in isogenic csrA, uvrY, or barA mutants of KSB837. The vector control was pBR322 in each case. Specific β-galactosidase activities and growth (A₆₀₀) at 24 h are shown as bars and closed circles, respectively. This experiment was repeated with essentially identical results.

Effects of uvrY and barA on CsrB-activated genes.

To assess the effects of the uvrY and barA mutations on genes that are repressed by csrA and activated by csrB, we monitored expression of a chromosomal glgCA′-′lacZ translational fusion and a plasmid-encoded glgC′-′lacZ translational fusion. The effect of uvrY disruption was similar to that of csrB (20) and was a modest decrease in glgCA′-′lacZ expression (Fig. 5A). Likewise, overexpression of uvrY yielded the opposite effect from uvrY and csrB disruption (Fig. 5B). Ectopic expression of uvrY from the multicopy plasmid pUY14 in a csrB-null mutant strain resulted in modest to negligible effects on glgCA′-′lacZ (Fig. 5C), suggesting that UvrY affects expression of glgCAP primarily through its is role as an activator of csrB. The uvrY and barA mutations also decreased glgC′-′lacZ expression from pCZ3-3 about twofold (data not shown).

FIG. 5.

Effects of the csrB or uvrY null mutations and uvrY overexpression on a chromosomal glgCA′-′lacZ translational fusion. β-Galactosidase activities expressed from glgCA′-′lacZ in strains KSGA18 (closed circles) and UY KSGA18 (uvrY::cam) (open circles) (A), KSGA18[pBR322] (closed circles) and KSGA18[pUY14] (open circles) (B), and RG KSGA18[pBR322] (csrB::cam) (closed circles) and RG KSGA18[pUY14] (csrB::cam) (open circles) (C) are shown. In each panel, growth (A₆₀₀) of the respective strains is depicted by closed or open squares. This experiment was repeated in its entirety, with essentially identical results.

UvrY activates csrB expression in vitro.

To determine whether UvrY directly regulates the transcription of the csrB gene, in vitro transcription-translation of pCBZ1-encoded csrB-lacZ transcriptional fusion was examined in S-30 extracts prepared from the uvrY mutant, UY CF7789, in the presence of various concentrations of purified UvrY protein. The expression of the pCBZ1-encoded csrB-lacZ fusion was activated ∼20-fold by uvrY in vivo, as determined in uvrY wild-type versus mutant strains (data not shown). As shown in Fig. 6, in vitro synthesis of the LacZ protein was stimulated approximately sixfold in the presence of 2.3 μM UvrY protein subunits, a concentration that saturated the reaction. Synthesis of the LacZ polypeptide was also detected in reactions with the control vector, pGE593. This was likely due to read-through transcription, as previously shown for similar derivatives of the vector pMLB1034 (29). However, in contrast to the reactions with pCBZ1, LacZ expression from the control vector was not stimulated by UvrY. Since pCBZ1 contained the upstream region of csrB (−242 to + 4), this experiment indicates that UvrY activates csrB transcription, presumably by binding to csrB DNA. In this experiment, recombinant UvrY protein was used as isolated from the cell. Since BarA is required for maximal expression of csrB in vivo (Fig. 1 and 2), it is likely that phosphorylated UvrY activates csrB transcription. It is possible that UvrY protein was phosphorylated prior to or during the S-30 reaction or that unphosphorylated UvrY may bind to DNA and activate transcription, albeit at reduced affinity relative to the phosphorylated form.

FIG. 6.

In vitro transcription-translation of the csrB-lacZ transcriptional fusion carried on pCBZ1. Reaction mixtures contained pCBZ1 (csrB-lacZ) or vector only (1.6 μg), as indicated. Reactions were conducted in the absence or presence of UvrY protein, in an S-30 extract from UY CF7789 (uvrY::cam). (A) Labeled proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and fluorography. The position of an unlabeled standard of β-galactosidase (LacZ) is shown. (B) Incorporation of [³⁵S]methionine into the LacZ polypeptide was determined by liquid scintillation counting. Approximately 5.5 × 10³ cpm is equivalent to 1 pmol of LacZ polypeptide per reaction per h.

Effects of sdiA on expression of csrA, csrB, and uvrY.

SdiA is a LuxR homologue that possesses a putative AHL binding domain and a second domain for binding DNA (54, 56). Genomic array studies indicated that an increase in the copy number of sdiA significantly increases the levels of uvrY mRNA (60). To determine if csrA, csrB, or uvrY expression is altered by sdiA, we examined expression from chromosomal csrA′-′lacZ, csrB-lacZ, and uvrY′-′lacZ fusions in isogenic sdiA mutant and wild-type strains. We also compared expression from these fusions in strains containing a plasmid clone of sdiA, pSdiA, or the cloning vector, pBR322. No significant effect of SdiA was observed for csrA′-′lacZ expression (Fig. 7A and B). However, expressions from csrB-lacZ and uvrY′-′lacZ fusions were partially dependent upon sdiA, as they were decreased by the sdiA mutation (Fig. 7C and E) and increased 1.5- and ∼6-fold, respectively (Fig. 7D and F), by sdiA overexpression. In order to determine whether sdiA regulates csrB expression via its effect on uvrY, a complementation test was conducted. Expression from the csrB-lacZ fusion in the uvrY mutant was no longer affected by sdiA overexpression (data not shown), suggesting that sdiA stimulates csrB expression through its effects on uvrY. Regulation via the highly conserved SdiA homologue of S. enterica was recently found to respond to AHL, the most active of which was N-(β-ketocaproyl)-l-homoserine lactone (33). Nevertheless, we did not observe any effect of this compound, over a broad concentration range, on lacZ fusions for csrA, csrB, or uvrY, in sdiA wild-type or mutant strains (data not shown).

FIG. 7.

Effects of sdiA disruption and overexpression on expression of chromosomal uvrY′-′lacZ and csrA′-′lacZ translational fusions and a csrB-lacZ transcriptional fusion. (A and B) Expression from csrA′-′lacZ in KSA712 and SA KSA712 (sdiA::kanR) (A) and KSA712[pBR322] and KSA712[pSdiA] (B). (C and D) Expression from csrB-lacZ in strains KSB837 and SA KSB837 (sdiA::kanR) (C) and KSB837[pBR322] and KSB837[pSdiA] (D). (E and F) Expression of uvrY′-′lacZ in strains KSY009 and SA KSY009 (sdiA::kanR) (E) and KSY009[pBR322] and KSY009[pSdiA] (F). In each panel, β-galactosidase activities in the wild-type strain are shown as closed circles, whereas activities in sdiA::kanR or sdiA-overexpressing strains are shown as open circles. Growth (A₆₀₀) of the respective strains is depicted by closed or open squares. This experiment was repeated in its entirety, with essentially identical results.

Effects of uvrY, barA, or sdiA on biofilm formation.

We recently showed that CsrA represses biofilm formation, while CsrB activates this process (27). The effects of uvrY, barA, and sdiA on biofilm formation were monitored in static cultures by the microtiter plate assay, which measures the binding of crystal violet to adherent cells of the biofilm (27). The most striking observation was that the ectopic expression of uvrY activated biofilm formation severalfold (Fig. 8B). This effect was almost as great as that of a csrA-knockout mutation. In addition, uvrY overexpression activated biofilm formation in a csrB mutant strain background, indicating that UvrY has effects on biofilm formation that are mediated independently of csrB. More modest effects were observed for knockouts of csrB, uvrY, barA, and sdiA, which were still statistically significant (Fig. 8A). The parent strain (MG1655) formed approximately threefold-more biofilm than each of these mutants. Overexpression of barA caused an unexplained modest decrease in biofilm formation. The increased gene dosage of sdiA caused a modest increase in biofilm formation, which was not determined to be statistically significant. In a repetition of this entire experiment, all of the above trends, including the relatively modest ones, were found to be reproducible.

FIG. 8.

Effects of csrB, uvrY, barA, and sdiA null mutations and ectopic expression of uvrY, barA, and sdiA on 24-h biofilms grown in microtiter wells. (A) Biofilm formation by the parent strain MG1655 and isogenic mutants, as indicated. (B) Effects of increased gene dosage of uvrY (pUY14), barA (pBA29), and sdiA (pSdiA) on biofilm formation by MG1655. (C) Same as panel B, except in a csrB-null mutant (RG1-B MG1655). Bars show the averages and standard errors of three experiments, with three samples per experiment. Double asterisks indicate statistically significant differences between strains of a given set (P < 0.01).

DISCUSSION

The RNA binding protein CsrA and the untranslated RNA CsrB constitute a posttranscriptional regulatory system that has profound effects on central carbon metabolism, motility, and multicellular behavior of E. coli (reviewed in reference 45). This system also has provided a model for the elucidation of several homologous systems in other gram-negative bacteria, referred to as CsrA/CsrB, RsmA/RsmB, or RsmA/RsmZ, which exhibit additional interesting and important regulatory roles (1, 4, 5, 8, 12, 32, 43). We recently reported that CsrA is autoregulatory and indirectly activates transcription of the gene for its RNA antagonist, CsrB (20). The simplest explanation for this role of CsrA was that CsrA either activates a transcriptional activator or inhibits a transcriptional repressor of csrB. The present study supports the former hypothesis and delineates the intervening signaling circuitry by which CsrA activates csrB transcription, namely, the BarA/UvrY two-component signal transduction system.

The organization of the signaling circuitry that connects the CsrA/CsrB and BarA/UvrY regulatory systems was defined by several kinds of evidence. First, the steady-state levels of CsrB RNA were extremely deficient in csrA (20) and uvrY (Fig. 1) mutants. The in vivo expression of a csrB-lacZ transcriptional fusion containing the region from −242 to +4 bp of csrB, relative to the start of transcription, is also highly dependent upon csrA (20) and uvrY (Fig. 2). Since this fusion is capable of expressing only the first 4 nucleotides of the natural csrB transcript, the latter experiment establishes that UvrY activates csrB transcript initiation, as opposed to stabilizing CsrB RNA. A barA mutant is partially defective for CsrB accumulation and csrB-lacZ expression (Fig. 1 and 2), but this effect of barA is considerably less severe than those of csrA (20) or uvrY (Fig. 1 and 2) mutations. Second, barA, uvrY, and csrA itself (20) have no effects on the expression of csrA (Fig. 2). CsrA and UvrY each stimulate barA-lacZ expression ∼1.5 to 2-fold (Fig. 3). Mutations in csrA, barA, or uvrY itself did not significantly affect uvrY′-′lacZ expression (data not shown). Third, complementation studies with multicopy plasmids showed that uvrY suppresses the defects in csrB-lacZ expression that are caused by csrA, barA, or uvrY mutations (Fig. 4). A csrA plasmid clone suppresses the defects of csrA or barA mutants but has no effect in a uvrY mutant (Fig. 4). Finally, a plasmid clone of barA suppresses the barA defect but does not affect csrB-lacZ expression in strains defective for uvrY or csrA (Fig. 4). Because ectopic expression of csrA has no effect on csrB if uvrY is defective, while ectopic expression of uvrY suppresses a csrA defect, these complementation experiments provide strong genetic evidence that the effects of CsrA on csrB transcription are mediated through UvrY. It should be reemphasized that CsrA does not affect csrB by regulating the expression of the uvrY gene, although it is a modest activator of the expression of its cognate kinase, BarA (Fig. 3). Fourth, purified CsrA protein failed to regulate csrB-lacZ expression in S-30 transcription-translation assays, indicating that CsrA indirectly activates csrB expression (20), while recombinant UvrY protein activated the same csrB-lacZ fusion approximately sixfold (Fig. 6). To our knowledge, the latter result represents the first biochemical evidence that UvrY or any of its homologues directly activates gene expression and positions UvrY immediately upstream from csrB in a signaling pathway. Fifth, mutations in uvrY or barA result in a reduction of the expression of glgC′-′lacZ and glgCA′-′lacZ translational fusions (Fig. 5 and data not shown), which are repressed by CsrA and activated by CsrB (20, 47, 63). While these effects of barA and uvrY were modest, they were in agreement with the modest effects noted for CsrB (20). The relatively weaker effects of CsrB than of CsrA have been observed thus far for all Csr-regulated genes. They are consistent with the finding that CsrB levels in the cell are sufficient to bind only ∼30% of the CsrA protein, assuming full occupation of the approximately 18 CsrA binding sites on CsrB (20, 31). The relatively lower level of CsrB in the cell than of CsrA may also account for the modest effects of BarA/UvrY, since UvrY effects on glgCA′-′lacZ were largely dependent on the presence of a functional csrB gene (Fig. 5). Sixth, comparisons of sdiA wild-type, mutant, and overexpressing strains confirmed that SdiA activates expression of a uvrY′-′lacZ translational fusion (Fig. 7). In addition, sdiA was found to activate csrB expression through its effect on uvrY.

The model shown in Fig. 9 incorporates the findings of this study in context with our previous findings on the regulatory circuitry of the Csr system. The RNA binding protein CsrA is the key regulator of the Csr system and indirectly activates the transcription of its RNA antagonist, CsrB, ∼20-fold (20). Although CsrA binds to CsrB RNA, it does not alter its chemical decay rate, which has a half-life of ∼2 min (20). The effects of CsrA on csrB expression are completely dependent upon UvrY (Fig. 4), which is a direct activator of csrB expression (Fig. 6). While barA is involved in the circuitry, mutagenesis (Fig. 1 and 2) and complementation (Fig. 4) studies suggest that CsrA can activate csrB expression independently of BarA. UvrY is a member of the FixJ family of response regulators, and promoter binding by purified FixJ has been shown elsewhere to require phosphorylation (2, 13, 44). In addition, BarA has been shown elsewhere to phosphorylate UvrY (42). Thus, activation of UvrY by BarA likely occurs by phosphorylation. The relative effects of CsrA (20), UvrY, and BarA on csrB expression and complementation studies with these genes collectively suggest that a BarA-independent mechanism also activates UvrY. The model in Fig. 9 also depicts the finding that UvrY stimulates barA expression, indicative of a positive autoregulatory loop within this system. Finally, the LuxR homologue SdiA activates uvrY expression and in this way affects csrB expression. Although the sigma factor RpoS or σ^s is important in stationary-phase regulation and rpoS transcription was reported elsewhere to be activated by barA (37), we previously found that neither CsrA protein nor CsrB RNA levels were significantly affected by rpoS disruption (20).

FIG. 9.

Summary of the regulatory interactions of CsrA/B, BarA/UvrY, and SdiA. CsrA activates csrB transcription indirectly (20). This effect of CsrA requires functional UvrY, which directly activates csrB transcription. The effect of CsrA on csrB is mediated in part by activation of barA expression but apparently also involves a BarA-independent, UvrY-dependent mechanism, shown as [X]. UvrY also activates the expression of barA, in an autoregulatory loop. SdiA activates the expression of uvrY and, to a lesser extent, that of csrB. Finally, CsrB RNA binds to ∼18 subunits of CsrA protein and antagonizes its regulatory effects in the cell (31, 45).

The UvrY/BarA two-component signal transduction system was recently recognized in E. coli, and biochemical and genetic evidence was presented to demonstrate direct phosphotransfer from BarA to UvrY (42). BarA of E. coli was reported elsewhere to be involved in the bacterial adaptive responses against hydrogen peroxide-mediated stress by activating transcription of the rpoS gene, which encodes a sigma factor involved in the expression of stationary-phase and stress response genes (23, 37). The effect of UvrY on rpoS has not yet been examined. Two-component systems homologous to BarA/UvrY have been identified in other gram-negative species and are known as BarA/SirA in S. enterica, ExpS/ExpA or GacS/GacA in E. carotovora, and GacS/GacA in Pseudomonas (references 22 and 42 and references therein). Genetic evidence for relationships between the Csr-homologous (Rsm) systems and the BarA/UvrY-homologous (GacS/GacA) systems has been obtained from Pseudomonas and Erwinia species. Transcription of rsmB was under the positive control of gacA (expA) in Erwinia carotovora subsp. carotovora (12, 26). In this bacterium, mutants with mutations in gacS or gacA possessed very low levels of RsmB RNA compared with the wild-type strain but possessed similar levels of rsmA mRNA and RsmA protein (12). In contrast, Hyytiäinen et al. (26) noted that rsmA transcription is elevated in the expA (gacA) mutant of this species during exponential growth. A recent report by Heeb et al. (21) revealed that a small RNA, which might be a csrB homologue of P. fluorescens CHA0, is activated through gacA/gacS and suggested that this RNA may affect RsmA by a titration effect, similar to the CsrA/CsrB model. Both E. carotovora and P. fluorescens are plant-associated species, and the GacS/GacA and Rsm systems affect the stationary-phase expression of genes encoding a variety of secreted virulence proteins of E. carotovora and protease and hydrogen cyanide production in P. fluorescens. The csrA, csrB, and barA genes of the mammalian pathogen S. enterica serovar Typhimurium regulate expression of the regulatory gene hilA and several related genes that encode components of the SPI1 type III secretion apparatus, which is required for mucosal invasion (4, 5). While the barA and csrB genes of S. enterica were originally reported to be unrelated (4), both genes affect expression of hilA and related genes. In the context of the results from E. coli and E. carotovora, it is plausible that regulatory interactions also occur between these genes in S. enterica. Thus, it appears that BarA/UvrY systems are intimately related to the CsrA/CsrB systems in a variety of γ-proteobacteria. Despite the important roles played by these systems, the environmental stimulus recognized by BarA remains to be established.

The present investigation confirms the result from a genomic array study, which had indicated that sdiA overexpression results in elevated levels of uvrY transcript in E. coli (60). SdiA is a member of the LuxR family of proteins, which regulate quorum sensing. SdiA facilitates the binding of RNA polymerase to the ftsQP2 promoter and requires an intact α-subunit C-terminal domain for this effect (62). A binding site for SdiA was demonstrated within the −35 region of the ftsQ promoter of E. coli, and the sequence 5′-AAAAGNNNNNNNNGAAAA-3′ was reported to comprise an SdiA box. A similar sequence overlaps the −35 region of the uvrY promoter (data not shown), and we predict that this is an SdiA box that is required for activation of uvrY transcription. Unlike other gram-negative species, in which LuxR homologues bind to AHL, E. coli, Salmonella, and Klebsiella spp. are not known to synthesize the AHL and have no apparent AHL synthase genes, which correspond to luxI or luxLM homologues (3). Nevertheless, Michael et al. (33) provided evidence that SdiA is an AHL receptor of S. enterica, which may detect signals emanating from other species. This suggests that SdiA may provide a means of recognizing other species, e.g., within the intestinal tract, and modulating csrB expression through effects on uvrY. This is an attractive hypothesis, in view of the proposed role of CsrB as an activator of hilA expression in S. enterica (4, 5). However, we tested the most active AHL derivative that was investigated by Michael et al. (33) and found that it had no effect on the expression of our csrB-lacZ fusion in E. coli (data not shown). Therefore, the environmental signal that permits SdiA of E. coli to regulate uvrY expression and, as a consequence, csrB transcription remains to be determined.

We recently reported that CsrA is a repressor of biofilm formation, while CsrB activates biofilm formation (27). In addition, the gratuitous induction of csrA in a preformed biofilm caused it to disperse by liberating viable planktonic cells (27). The effect of CsrA on biofilm formation was mediated largely through its role as a regulator of intracellular glycogen synthesis and turnover in the stationary phase of growth. The most striking finding noted upon examination of biofilm formation by strains in the present study was that ectopic expression of uvrY caused a severalfold increase in biofilm formation, which was almost as great as the increase caused by a csrA mutation (Fig. 8). UvrY was able to activate biofilm formation independently of CsrB. Recently, the gacA gene of Pseudomonas aeruginosa, which is homologous to uvrY, was reported to activate biofilm formation (41). In that study, a number of possible factors that are required for biofilm formation were examined, including twitching and swarming motility, alginate biosynthesis, and autoinducer production, but none accounted for the regulatory effect of GacA. While the effect of UvrY on E. coli biofilm formation remains to be elucidated, UvrY effects on glgA expression and glycogen accumulation (data not shown) appear to be too modest to account for this role of UvrY. Thus, the mechanisms by which UvrY and GacA activate biofilm formation by E. coli and P. aeruginosa, respectively, have not yet been determined.