Global Regulator of Virulence A (GrvA) Coordinates Expression of Discrete Pathogenic Mechanisms in Enterohemorrhagic Escherichia coli through Interactions with GadW-GadE

ABSTRACT

Global regulator of virulence A (GrvA) is a ToxR-family transcriptional regulator that activates locus of enterocyte effacement (LEE)-dependent adherence in enterohemorrhagic Escherichia coli (EHEC). LEE activation by GrvA requires the Rcs phosphorelay response regulator RcsB and is sensitive to physiologically relevant concentrations of bicarbonate, a known stimulant of virulence systems in intestinal pathogens. This study determines the genomic scale of GrvA-dependent regulation and uncovers details of the molecular mechanism underlying GrvA-dependent regulation of pathogenic mechanisms in EHEC. In a grvA-null background of EHEC strain TW14359, RNA sequencing analysis revealed the altered expression of over 700 genes, including the downregulation of LEE- and non-LEE-encoded effectors and the upregulation of genes for glutamate-dependent acid resistance (GDAR). Upregulation of GDAR genes corresponded with a marked increase in acid resistance. GrvA-dependent regulation of GDAR and the LEE required gadE, the central activator of GDAR genes and a direct repressor of the LEE. Control of gadE by GrvA was further determined to occur through downregulation of the gadE activator GadW. This interaction of GrvA with GadW-GadE represses the acid resistance phenotype, while it concomitantly activates the LEE-dependent adherence and secretion of immune subversion effectors. The results of this study significantly broaden the scope of GrvA-dependent regulation and its role in EHEC pathogenesis.

IMPORTANCE Enterohemorrhagic Escherichia coli (EHEC) is an intestinal human pathogen causing acute hemorrhagic colitis and life-threatening hemolytic-uremic syndrome. For successful transmission and gut colonization, EHEC relies on the glutamate-dependent acid resistance (GDAR) system and a type III secretion apparatus, encoded on the LEE pathogenicity island. This study investigates the mechanism whereby the DNA-binding regulator GrvA coordinates activation of the LEE with repression of GDAR. Investigating how these systems are regulated leads to an understanding of pathogenic behavior and novel strategies aimed at disease prevention and control.

INTRODUCTION

Enterohemorrhagic Escherichia coli (EHEC) O157:H7 is a virulent intestinal pathogen causing foodborne outbreaks of bloody diarrhea (hemorrhagic colitis) and the life-threatening kidney disease hemolytic-uremic syndrome (1–3). Competitive colonization of the intestine by EHEC requires a type III secretion (T3S) mechanism and is characterized by the formation of attaching-and-effacing (A/E) lesions (1, 2). The T3S apparatus is encoded on a pathogenicity island, referred to as the locus of enterocyte effacement (LEE), containing 41 genes organized into five operons (LEE1 to LEE5) (4). The master LEE regulator, Ler, encoded as the first gene of LEE1, positively stimulates LEE transcription by relieving H-NS-mediated repression. Ler further activates the LEE-encoded regulators GrlA and GrlR, which in turn activate and repress ler transcription, respectively (5–10). Numerous non-LEE-encoded regulators also converge on ler and grlA promoters to coordinate LEE-dependent colonization with environmental/physiological cues (11–19).

One such regulator, RcsB, has been shown to both activate and repress LEE transcription and more recently has been shown to be required for induction of the LEE in response to the bicarbonate ion (20, 21). RcsB is the response regulator of the Rcs phosphorelay system, a multicomponent signaling pathway including RcsC (sensor kinase), RcsD (histidine phosphotransferase), and RcsF (outer membrane lipoprotein) (22–26). Upon phosphorylation, RcsB can bind to target promoters as a homodimer and as a heterodimer in conjunction with other regulatory auxiliary proteins, such as RcsA, BglJ, and GadE (27–29). Binding occurs at specific RcsB consensus sites (30, 31), and the location of binding relative to the −35 consensus sequence can dictate whether RcsB positively or negatively affects transcription (32). Tobe et al. (21) demonstrated that upregulation of the LEE effector EspB in response to rcsB overexpression also required the transcriptional regulator GrvA (for global regulator of virulence A). Furthermore, the overexpression of grvA was shown to activate transcription from the ler promoter (LEE1 operon) and to increase adherence to Caco-2 cells in vitro. In a subsequent study by Morgan et al. (20), both rcsB and grvA were shown to be required for bicarbonate induction of the LEE. Collectively, these two studies reveal GrvA to be an activator of Ler that can interact with RcsB to communicate the bicarbonate signal to the LEE colonizing mechanism.

How RcsB controls grvA and how GrvA, in turn, controls LEE expression are not known. GrvA shares homology with ToxR-family protein regulators, such as MarT in Salmonella enterica (30% amino acid identity) and CadC in E. coli (13% amino acid identity). Both MarT and CadC are transcriptional activators, with the latter activating the cadBA operon by displacing H-NS-mediated repression (33). The similarity of GrvA to MarT and CadC largely resides in the N terminus, containing a conserved DNA-binding helix-turn-helix domain (COG3710) and a CheY-like response regulator receiver domain (COG0745). In this study, details of the molecular basis for grvA regulation and for GrvA-dependent control of the LEE and LEE-dependent adherence are determined. Furthermore, global RNA sequencing analysis significantly broadens our knowledge of the role for GrvA in regulation to include genes central to acid resistance and metabolic fitness.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The strains and plasmids used in this study are listed in Table 1. Strains were stocked at −80°C in glycerol diluted (final concentration, 15%, vol/vol) in Luria broth (LB) and were maintained in LB or on LB with 1.5% agar (LBA). Unless otherwise noted, overnight (18- to 20-h) cultures grown in LB were used to inoculate fresh LB or LB buffered with sodium bicarbonate (44 mM NaHCO₃) or fresh Dulbecco's modified Eagle's medium (DMEM; 4 g/liter glucose, 4 mM glutamine, 44 mM NaHCO₃, pH 7.4) to a final optical density at 600 nm (OD₆₀₀) of 0.05. Cultures were grown at 37°C in a rotary shaker (200 rpm) using a 1:10 medium-to-flask volume. Antibiotics (Sigma-Aldrich) were added to the cultures when required.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristics	Reference or source
Strains
DH5α	Vector propagation, recA1 endA1
BL21(DE3)pLysS	BL21 with IPTG-inducible T7 polymerase	90
TW14359	Wild-type strain from the 2006 outbreak, western USA	91
EcRJM-1	TW14359 ΔescN	20
EcRJM-6	TW14359 ΔrcsB	20
EcRJM-10	TW14359 rcsB-FLAG	20
EcRJM-11	TW14359 ΔgrvA	20
EcRJM-12	TW14359 ΔrcsB ΔgrvA	20
EcRJM-13	TW14359 tir-FLAG	20
EcRJM-35	TW14359 ΔgrlR::kan	20
EcRJM-72	TW14359 ΔrcsB/pRJM-20	20
EcRJM-73	TW14359 ΔlacZ	This study
EcRJM-74	TW14359/pRJM-23	This study
EcRJM-75	TW14359/pRJM-24	This study
EcRJM-76	TW14359/pRJM-25	This study
EcRJM-77	TW14359/pRJM-26	This study
EcRJM-78	TW14359 ΔgrvA/pRJM-23	This study
EcRJM-79	TW14359 ΔgrvA/pRJM-24	This study
EcRJM-80	TW14359 ΔgrvA/pRJM-25	This study
EcRJM-81	TW14359 ΔgrvA/pRJM-26	This study
EcRJM-82	TW14359 ΔgrvA ΔgadE::kan/pRJM-24	This study
EcRJM-110	TW14359 ΔgrvA ΔgadE::kan/pRJM-26	This study
EcRJM-83	TW14359 ΔgrvA/pRJM-22	This study
EcRJM-84	TW14359 ΔrcsB ΔgrvA/pRJM-22	This study
EcRJM-85	TW14359 ΔrcsB ΔgrvA/pRJM-20	This study
EcRJM-86	TW14359 rcsDB-luxE	This study
EcRJM-87	TW14359 grvAB-luxE	This study
EcRJM-88	TW14359 rcsDB-luxE/pluxCDAB3	This study
EcRJM-89	TW14359 grvAB-luxE/pluxCDAB3	This study
EcRJM-90	TW14359 ΔgadE::kan	This study
EcRJM-91	TW14359 ΔgrvA ΔgadE::kan	This study
EcRJM-92	TW14359 ΔgadE::kan/pRJM-30	This study
EcRJM-93	TW14359 ΔgrvA ΔgadE::kan/pRJM-30	This study
EcRJM-94	TW14359 zinT-FLAG::kan	This study
EcRJM-95	TW14359 ΔgrvA zinT-FLAG::kan	This study
EcRJM-96	TW14359/pRJM-31	This study
EcRJM-97	EPECa E2348/69/pRJM-31	This study
EcRJM-98	MG1655/pRJM-31	This study
EcRJM-99	DH5α/pRJM-31	This study
EcRJM-100	TW14359/pRJM-32	This study
EcRJM-101	EPEC E2348/69/pRJM-32	This study
EcRJM-102	MG1655/pRJM-32	This study
EcRJM-103	DH5α/pRJM-32	This study
EcRJM-104	TW14359 rcsDB-luxE/pluxCDAB3/pRJM-21	This study
EcRJM-105	TW14359 grvAB-luxE/pluxCDAB3/pRJM-21	This study
EcRJM-106	TW14359 ΔgrvA/pRJM-23	This study
EcRJM-107	TW14359 ΔgrvA/pRJM-24	This study
EcRJM-108	TW14359 ΔgrvA/pRJM-25	This study
EcRJM-109	TW14359 ΔgrvA/pRJM-26	This study
Plasmids
pACYC177	Low-copy-no. cloning vector, Ampr Kanr P15A
pBAD22	Ara-inducible expression vector, Ampr M13	92
pMPM-A2	Low-copy-no. cloning vector, Ampr pMB1/f1	36
pMPM-K3	Low-copy-no. cloning vector, Kanr p15A/f1	36
pMPM-T3	Low-copy-no. cloning vector, Tetr p15A/f1	36
pUC19	High-copy-no. cloning vector, Ampr pMB1	93
pRS551	lac fusion vector, Ampr Kanr lacZ+ ColE1
pCP20	FLP recombinase expression vector	34
pKD4	Template plasmid for Kan cassette	34
pKM208	Bacteriophage λ Red recombinase expression vector	34
pET-24d	T7 promoter, His tag vector, Kanr f1, pBR322	Novagen
pSU312	FLAG epitope template, Ampr Kanr, R6K	94
pRJM-20	pACYC177-rcsB	20
pRJM-33	pACYC177-grvAB	This study
pRJM-21	pMPM-K3-rcsDB	This study
pRJM-22	pMPM-A2-grvA	This study
pRJM-23	pRS551-gadEP(−773, −1)b	This study
pRJM-24	pRS551-gadEP(−320, −1)	This study
pRJM-25	pRS551-gadEP(−516, −276)	This study
pRJM-26	pRS551-gadEP(−773, −497)	This study
pRJM-27	pET-24d-6×His-rcsB	This study
pRJM-28	pMPM-T3-kan	This study
pRJM-29	pMPM-T3-luxE-kan	This study
pRJM-30	pMPM-A2-gadE	This study
pRJM-31	pRS551-grvAP(−268, +38)	This study
pRJM-32	pUC-grvA-6×His	This study
placlux8	placlux8	40
pLuxCDAB3	pLuxCDAB3	40

^aEPEC, enteropathogenic E. coli.

^bThe numbers indicate the positions in gadE_P.

Genetic manipulations and complementation.

Primers used for genetic manipulation and complementation are available upon request. For construction of deletion mutants, the bacteriophage λ Red-assisted one-step deletion method adapted for EHEC was used as described previously (20, 34, 35). Complementation of rcsB was performed using vector pRJM-20, as previously described (20). Additionally, a fragment containing the entire rcsDB operon, including the rcsD promoter region (nucleotide positions 3041665 to 3046191; GenBank accession number NC_013008.1), was cloned into BamHI/XbaI-digested low-copy-number expression vector pMPM-K3 (Kan^r) (36) using primers RcsB−3732/BamHI and RcsB+794/XbaI to produce pRJM-21. To complement grvA, a PCR product corresponding to the grvA open reading frame (ORF) and including a 267-bp upstream region of grvA containing the predicted promoter (nucleotide positions 1281984 to 1283134; GenBank accession number NC_013008.1) was produced using primers GrvA−268/XhoI and GrvA+884/BamHI and cloned into XhoI/BamHI-digested low-copy-number expression vector pMPM-A2 (Amp^r), creating pRJM-22. To complement gadE, a PCR product corresponding to the gadE ORF and 762 bp of the sequence upstream of gadE (containing the native promoters, nucleotide positions 4458857 to 4460195) was produced using primers GadE−763/XbaI and GadE+577/BamHI and cloned into the XbaI/BamHI-digested vector pMPM-A2 to create pRJM-30. Confirmation of genetic constructs was done using a combination of restriction mapping and DNA sequencing (MWG Operon).

RNA purification and qRT-PCR.

The primers used for quantitative real-time PCR (qRT-PCR) are available upon request. RNA purification, cDNA synthesis, qRT-PCR cycling conditions, and data analysis followed previously described protocols (37) using a Realplex2 Mastercycler instrument (Eppendorf). Cycle threshold (C_T) data were normalized to the level of rrsH (16S rRNA gene) expression, and normalized cycle threshold (ΔC_T) values were transformed to arbitrary transcript expression levels using 2^−ΔCT/10⁻⁶, as described previously (38, 39). Expression levels were compared statistically by the appropriate t test or by Tukey's honestly significant difference (HSD) test following a significant F test (n ≥ 3, α = 0.05; R software, version 3.1.0).

Construction of a single-copy grvAB-luxE operon fusion.

A strain containing a single-copy chromosome-plasmid luxE reporter fusion was constructed using a protocol adapted from that of Shimizu et al. (40). To make a grvAB-luxE fusion, the kan cassette and flanking FLP recombination target sites were amplified from pKD4 (34) using primers pKD4forward/SacI and pKD4reverse/BamHI. This product was SacI/BamHI digested and cloned into the BamHI/SacI site of pMPM-T3 to produce pMPM-T3-kan. A XhoI/BamHI-digested PCR product containing the luxE gene and native ribosome binding site was amplified from placlux8 (40) using primers LuxE-18/XhoI and LuxE+1,450/BamHI and cloned into the XhoI/BamHI site of similarly digested pMPM-T3-kan to create pMPM-T3-luxE-kan. The luxE-kan PCR product amplified from pMPM-T3-luxE-kan using primers GrvA+1,283/LuxE and GrvA+1,431/P2 and Phusion high-fidelity DNA polymerase (NEB) was fused to a region 13 bp downstream of the grvAB operon using bacteriophage λ Red recombination (34). Kanamycin resistance cassettes were removed using FLP recombinase as described previously (34), and the grvAB-luxE construct was validated using a combination of restriction mapping and DNA sequencing (MWG Operon). The lux operon genes luxCDAB were constitutively expressed in trans in TW14359 grvAB-luxE from pluxCDAB3 (40, 41).

Luciferase activity.

Overnight cultures grown in LB and DMEM were used to inoculate fresh LB or DMEM, respectively, to an initial OD₆₀₀ of 0.05. To measure the effect of sodium bicarbonate (NaHCO₃) on grvA expression, LB test cultures were grown with and without addition of 44 mM NaHCO₃ (all DMEM cultures contained 44 mM NaHCO₃). Cultures (0.2 ml) were inoculated into 96-well, clear-bottom white-walled plates (catalog number 655098; Greiner Bio-One) and incubated at 37°C in a rotary shaker (200 rpm). Luciferase and OD₆₀₀ measurements were taken every hour for 10 h using a BioTek Synergy 2 plate reader (1 s integration) prewarmed to 37°C. Mean luciferase activity was compared between treatments (medium and NaHCO₃ effects) using the appropriate t test (n = 4, α = 0.05; R software, version 3.1.0).

EMSAs.

Electrophoretic mobility shift assays (EMSAs) were performed as previously described (42). Briefly, the forward primers RcsB+4/NcoI and RcsB+682/XhoI, used to PCR amplify rcsB from TW14359 DNA, contained a new start codon and an N-terminal 6×His epitope tag, and the PCR product was cloned into the NcoI/XhoI sites of similarly digested vector pET-24d to create pRJM-27. N-terminal 6×His-RcsB fusion proteins have wild-type RcsB activity (43). 6×His-RcsB was purified from strain BL21(DE3)pLysS using Ni-nitrilotriacetic acid spin columns (Qiagen) according to the manufacturer's protocol. Briefly, overnight LB cultures containing 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) and kanamycin (50 μg/ml) grown at 18°C (200 rpm) were pelleted and resuspended in 1.5 ml of lysis buffer (10 mM imidazole, 300 mM NaCl, 50 mM NaH₂PO₄, pH 8.0) and lysed by sonication for 3 min (50% amplitude, 10-s intervals, 30-s pause) while they were chilled on ice. The lysate was centrifuged at 21,000 × g to remove insoluble cellular debris, and the supernatant was used for column purification and subsequently eluted with elution buffer (500 mM imidazole, 300 mM NaCl, 50 mM NaH₂PO₄, pH 8.0). Purified 6×His-RcsB protein was quantified using a Bradford protein assay. Five prime biotin-labeled primers (Integrated DNA Technologies oligonucleotides) were used to generate a probe containing the grvA promoter (grvA_P) region, including putative RcsB consensus binding sites (Frag-2F and Frag-2R) (Fig. 1B). An additional upstream grvA promoter probe with no predicted RcsB consensus sequence was amplified for use as a negative control (Frag-1F and Frag-1R) (Fig. 1B). Biotin-labeled probes were gel purified using QIAquick gel extraction kits (Qiagen) and diluted to a concentration of 0.5 ng/μl. All EMSAs were performed using a LightShift chemiluminescent EMSA kit (Thermo Pierce) according to the manufacturer's specifications. Prior to electrophoresis, purified 6×His-RcsB and biotin-labeled fragments were coincubated for 40 min in binding buffer (1× binding buffer, 50 ng/μl sheared salmon sperm DNA, 2.5% glycerol, 0.05% NP-40, 5 mM MgCl₂, 50 mM KCl, 1 mM EDTA) at room temperature. Samples were loaded into a prerun and cooled 8% native polyacrylamide gel and run at 20 V/cm for 1 h. Following electrophoresis, biotin-labeled DNA was transferred (25 V for 20 min) to a Biodyne B precut modified nylon membrane (Thermo Pierce) and UV cross-linked for 60 s. Biotin-labeled DNA fragments were detected using streptavidin-horseradish peroxidase-conjugated antibodies (1:300) in blocking buffer, and the membranes were washed, equilibrated, and subsequently detected using luminol/enhancer solutions in a ChemiDoc XRS+ imaging system and Image Lab (version 3.0) software (Bio-Rad).

FIG 1.

Rcs response regulator RcsB directly activates transcription from the grvAB promoter. (A) (Left) Average luciferase activity (in relative light units [RLU]) plotted as a function of time for a single-copy grvAB-luxE fusion grown in DMEM, LB, or LB supplemented with 44 mM NaHCO₃. Asterisks denote significance by Student's t test (*, P < 0.05; **, P < 0.01; n ≥ 3). Plots at each time point did not vary from the mean by more than 5%. (Right) Growth and grvAB promoter expression (luciferase activity) (bottom) on LB agar plates for strain TW14359 grvAB-luxE/pluxCDAB3 transformed with empty vector pMPM-K3 (left) or pMPM::rcsDB (right). (B) (Left) Map of the grvAB promoter and flanking regions in strain TW14359. Fragments for EMSA (Frag-1 and Frag-2) and the respective primers for probe generation are indicated. Frag-1 is the negative control; Frag-2 contains two putative RcsB binding sites (Box 1 and Box 2). (Right) Alignment of box 1 from the grvAB promoter with experimentally proven RcsB homodimer boxes for rprA and osmC promoters. Conserved bases are in bold and underlined. (C) EMSA for RcsB binding to Frag-2 of the grvAB promoter. The wedge denotes the increasing amount of RcsB added (1 to 12 μg). For Frag-1 (control), RcsB was added at 0 μg (lane −) and 12 μg (lane +).

Construction of lacZ transcriptional fusions and β-galactosidase assays.

Construction of a grvA_P-lacZ reporter transcriptional fusion followed a previously described protocol using pRS551 (Table 1) (20). Briefly, a 306-bp PCR product containing the predicted grvA promoter was amplified from TW14359 genomic DNA using primers GrvA−268/EcoRI and GrvA+38/BamHI, BamHI/EcoRI digested, and then cloned into similarly digested pRS551, resulting in pRJM-31. Four different gadE_P-lacZ promoter fusions were created by cloning BamHI/EcoRI-digested PCR products into similarly digested pRS551 using primers GadE−320/EcoRI and GadE−1/BamHI, GadE−276/EcoRI and GadE−516/BamHI, GadE−773/EcoRI and GadE−497/BamHI, and GadE−773/EcoRI and GadE−1/BamHI. β-Galactosidase activity (Miller units) was measured as previously described (38, 44), and the numbers of Miller units were compared using the appropriate t test or by Tukey's HSD test following a significant F test (n ≥ 3, α = 0.05; R software, version 3.1.0).

RNA sequencing and data analysis.

RNA extractions were performed as described above for qRT-PCR, except that residual genomic DNA was digested using Turbo DNase (catalog number AM2238; Ambion) following the manufacturer's protocol. Three independent RNA samples for each strain (TW14359 and TW14359 ΔgrvA) were pooled into a single sample (n = 1) in a 1:1:1 ratio to account for intersample variability, and rRNA was removed using a MICROBExpress bacterial mRNA enrichment kit (catalog number AM1905; Ambion) according to the manufacturer's instructions. The integrity and concentration of purified and enriched mRNA were determined using an Agilent 2100 bioanalyzer (Agilent Technologies). RNA sequencing (RNA-seq) was performed as previously described (45). Briefly, mRNA was converted to strand-specific cDNA using an Ion Total RNA-Seq kit (version 2; catalog number 4475936; Ion Torrent), and samples were sequenced using an Ion Personal Genome Machine (PGM) system (Ion Torrent). Sequence mapping and data analysis were performed using CLC Genomics Workbench (version 6) software (CLC Bio). Subsequent reads per kilobase per million mapped reads (RPKM) values were converted to the fold change for TW14359 ΔgrvA relative to the value for wild-type strain TW14359. Circular plots were created using Circos (version 0.65) software (http://www.circos.ca) (46).

Intestinal cell adherence assays.

The maintenance and culture of HT-29 colonic intestinal cells were performed as previously described (20). Adherence competition indexes were determined using the method of Gabbianelli et al. (47). Briefly, overnight DMEM cultures were used to inoculate fresh DMEM to an OD₆₀₀ of 0.05, and these were then incubated for 3 h at 37°C with shaking (200 rpm). Cultures were then diluted to an OD₆₀₀ of 0.5 and mixed in a 1:1 ratio (test strain/control strain), and 0.2 ml of each was used to inoculate HT-29 cells in 6-well culture plates. A control strain incapable of utilizing lactose (a Lac⁻ strain) was constructed by deleting the entire lacZ ORF in TW14359 (strain EcRJM-73) using the bacteriophage λ Red approach and deletion primers LacZ−1/P1 and LacZ+3115/P2 (primer sequences are available upon request). Each plate was gently mixed before centrifugation at 500 × g for 5 min and then incubated as described above. Following 3 h of incubation, each well was washed four times using sterile phosphate-buffered saline (PBS) to remove nonadherent cells, and adherent cells were removed using 500 μl of 0.1% Triton X-100. Cells were enumerated (number of CFU per milliliter) by serial dilution in PBS and plating onto MacConkey agar, which is differential for lactose utilization. After overnight growth on MacConkey agar, pink (Lac-positive [Lac⁺]) colonies were scored as the numbers of CFU per milliliter for the test strain, whereas white (Lac-negative [Lac⁻]) colonies were scored as the numbers of CFU per milliliter for strain TW14359 ΔlacZ. The competitive index was derived by dividing the numbers of CFU per milliliter of the test strain by the numbers of CFU per milliliter of TW14359 ΔlacZ, and the indexes were compared statistically using the appropriate t test (n ≥ 3, α = 0.05; R software, version 3.1.0).

Test for acid resistance.

Acid resistance by the glutamate-dependent system was measured as described previously (37) with slight adaptations. Mid-exponential-phase (OD₆₀₀ = 0.5) DMEM cultures were inoculated to a 10⁶-CFU/ml final cell density in E minimal glucose (EG) medium with or without 5.7 mM l-glutamate at pH 7 (control) or acidified with HCl (pH 2). For cell count (number of CFU per milliliter) and percent survival determinations, samples were serially diluted in PBS (pH 7), plated onto LBA, and incubated overnight at 37°C.

RESULTS

Transcription of grvA is growth phase dependent and directly activated by RcsB.

Previous work has shown the response regulator of the rcs phosphorelay system RcsB to be an activator of global regulator of virulence A (grvA) and both factors to be required for full bicarbonate induction of the LEE-encoded regulator Ler and subsequent stimulation of LEE expression in EHEC during exponential-phase growth (20, 21). RcsB is predicted to act upstream of grvA in the regulation of ler, yet the mechanism underlying the control of grvA by RcsB is unknown.

The level of transcription of grvA in reporter strain TW14359 grvAB-luxE∕pluxCDAB3 (EcRJM-89) (Table 1) grown in DMEM with sodium bicarbonate (44 mM) was the highest during early exponential-phase growth, rapidly declining as cultures transitioned into postexponential phase (Fig. 1A). When grown in LB, the level of grvA transcription was similar to that observed when grown in DMEM; however, the levels were consistently lower over the first 7 h. Addition of sodium bicarbonate (44 mM) to LB, previously shown to upregulate RcsB levels (20), had no significant impact on grvA transcription, while overexpression of rcsDB in trans in TW14359 grvAB-luxE∕pluxCDAB3 substantially increased the level of grvA transcription when grown on LB agar in the absence of sodium bicarbonate (Fig. 1A). Thus, the addition of bicarbonate alone is not sufficient for upregulation of grvA. That overexpression of rcsB increased the level of transcription from the grvA promoter suggests that RcsB may directly activate grvA transcription. In support of this, two putative tandem RcsB binding sites proximal to a predicted −35 site (nucleotides 1282968 to 1282982 in TW14359) of the grvA promoter (grvA_P) have been described previously (21, 31) (Fig. 1B, Box 1 and Box 2). To test for interactions between RcsB and the grvA promoter, purified 6×His-tagged RcsB was coincubated with a fragment of the core grvAB_P element, including putative RcsB binding sites (Fig. 1B, Frag-2) and analyzed using electrophoretic mobility shift assays. As anticipated, the 235-bp grvAB_P fragment was visibly shifted with the addition of increasing amounts (1 to 12 μg) of 6×His-RcsB, while no shift was observed for the control fragment (Fig. 1B, Frag-1, and C). These findings suggest that RcsB is a direct transcriptional activator of grvA and that activation occurs at the grvA core promoter region containing at least one putative RcsB binding site.

GrvA is a regulator of pathogenic mechanisms and metabolic fitness in EHEC.

grvA is required for LEE activation by the Rcs phosphorelay system in EHEC (21). How, in turn, GrvA upregulates LEE expression and the scope of genes under the control of GrvA are unknown. To explore this, the transcriptomes of EHEC O157:H7 strain TW14359 and its grvA isogenic derivative (TW14359 ΔgrvA) were measured by RNA-seq analysis during exponential-phase growth (OD₆₀₀ = 0.5) in LEE-inducing medium (DMEM).

In TW14359 ΔgrvA, 765 genes were altered in expression >2-fold above the RPKM cutoff compared to their expression in TW14359; of these, 264 were upregulated and 501 were downregulated (Fig. 2; unpublished data). Known virulence-associated genes made up 6% of the genes whose expression was altered (8 upregulated, 38 downregulated). The majority of these were LEE-encoded structural proteins and secreted effectors, consistent with the role of GrvA as a positive regulator of LEE expression (Fig. 2A and B). Twenty-one LEE genes were downregulated in TW14359 ΔgrvA, including the LEE-encoded regulator ler, type III secretion translocon genes espA, espB, espD, and eaeA (encoding intimin), and the secreted effector tir (translocated intimin receptor) (Table 2). Five non-LEE-encoded effectors (nleC, nleA, nleH2, nleF, and nleE) were also downregulated in TW14359 ΔgrvA. Genes encoded within the 15-kb genomic acid fitness island (AFI), the products of which are required for survival at low pH (48), were upregulated in TW14359 ΔgrvA compared to their level of regulation in TW14359, in which the expression of AFI genes was barely detectable (Table 2). Most notably, genes of the glutamate-dependent acid resistance (GDAR) system were upregulated, including gadA, gadE, gadW, gadX, and the non-AFI-encoded gadCB operon. Thus, GrvA activates the LEE colonization mechanism while repressing acid resistance systems.

FIG 2.

The GrvA regulon in EHEC strain TW14359. (A) Circular plot of RNA-seq data for EHEC strains TW14359 and TW14359 ΔgrvA. Nucleotide positions are based on the TW14359 annotation (GenBank accession number NC_013008.1), indicated by the black outer ring. The heat map denotes the fold change in RPKM for TW14359 ΔgrvA relative to the level of expression in TW14359 for each gene. The two inner histograms indicate the RPKM values of TW14359 (black) and TW14359 ΔgrvA (red). Plots were generated using Circos (version 0.65; http://www.circos.ca). (B) Number of genes in grvA whose expression is altered relative to their expression in TW14359 plotted for each ontological group. The levels of expression of a total of 765 genes were altered by ≥2-fold; upregulation and downregulation are indicated.

TABLE 2.

RNA-seq GrvA transcriptome

Gene identifiera	Gene name	Fold change in ΔgrvA mutantb	RPKMc		Gene product description
			Wild type	ΔgrvA mutant
LEE pathogenicity island genes and genes for non-LEE-encoded effectors
ECSP_4665	espF	−3.1	16.5	5.4	LEE-encoded effector
ECSP_4666	orf29	−3.4	11.9	3.5	LEE-encoded protein
ECSP_4667	escF	−5.8	15.5	2.7	LEE-encoded protein
ECSP_4668	cesD2	−8.3	46.3	5.6	Predicted chaperone
ECSP_4669	espB	−6.6	658.8	100.5	Secreted protein EspB
ECSP_4670	espD	−4.5	208.3	46.4	Secreted protein EspD
ECSP_4671	espA	−4.4	165.3	37.7	Secreted protein EspA
ECSP_4672	sepL	−4.5	46.8	10.3	LEE-encoded T3S component
ECSP_4673	escD	−4.0	4.8	1.2	LEE-encoded T3S component
ECSP_4674	eaeA	−4.8	75.0	15.8	Intimin adherence protein
ECSP_4675	cesT	−5.9	105.0	17.7	Chaperone
ECSP_4676	tir	−4.0	186.7	46.6	Translocated intimin receptor
ECSP_4677	map	−4.4	27.7	6.3	LEE-encoded effector
ECSP_4678	cesF	ND (−)d	3.4	<10−4	Chaperone
ECSP_4679	espH	−4.8	22.9	4.8	LEE-encoded effector
ECSP_4680	escQ	−1.9	9.7	5.0	LEE-encoded protein
ECSP_4681	orf16	−8.1	14.7	1.8	LEE-encoded protein
ECSP_4682	orf15	−1.2	5.9	4.9	LEE-encoded protein
ECSP_4683	escN	−2.7	16.8	6.2	LEE-encoded ATPase
ECSP_4684	escV	−2.3	11.6	5.1	LEE-encoded protein
ECSP_4685	orf12	−1.3	2.6	2.0	LEE-encoded protein
ECSP_4686	espZ	−3.5	78.4	22.6	LEE-encoded effector
ECSP_4687	escI	−1.3	12.7	9.9	LEE-encoded protein
ECSP_4688	escJ	−2.8	18.4	6.6	LEE-encoded protein
ECSP_4689	sepD	−1.3	2.2	1.7	LEE-encoded protein
ECSP_4690	escC	−1.7	7.8	4.5	T3S needle complex subunit
ECSP_4691	cesD	−1.1	6.5	6.0	LEE-encoded protein
ECSP_4692	grlA	−1.1	15.3	13.8	Global regulator of LEE, activator
ECSP_4693	grlR	−1.9	11.4	5.9	Global regulator of LEE, repressor
ECSP_4694	etgA	−1.5	1.3	0.9	Lytic transglycosylase
ECSP_4695	escU	ND (−)	1.6	<10−4	LEE-encoded protein
ECSP_4696	escT	ND (−)	2.0	<10−4	LEE-encoded protein
ECSP_4697	escS	4.0	0.6	2.4	LEE-encoded protein
ECSP_4698	escR	−1.6	4.3	2.6	LEE-encoded protein
ECSP_4699	escL	−2.9	12.2	4.2	LEE-encoded protein
ECSP_4700	orf4	−1.8	10.3	5.6	LEE-encoded protein
ECSP_4701	cesA	−1.0	2.2	2.2	Chaperone
ECSP_4702	escE	−1.1	18.3	16.2	LEE-encoded protein
ECSP_4703	ler	−2.5	55.7	22.7	LEE encoded regulator
ECSP_4704	espG	−3.7	4.5	1.2	LEE encoded effector
ECSP_4705	rorf1	−1.0	0.6	0.6	LEE-encoded protein
ECSP_0061	espY1	2.0	0.3	0.6	Non-LEE-encoded effector
ECSP_0866	nleC	−2.3	4.3	1.9	Non-LEE-encoded effector
ECSP_0868	nleD	−1.7	7.3	4.4	Non-LEE-encoded effector
ECSP_1702	nleA	−4.0	1.5	0.4	Non-LEE-encoded effector
ECSP_1704	nleH2	2.2	3.8	8.6	Non-LEE-encoded effector
ECSP_1705	nleF	−3.7	16.2	4.4	Non-LEE-encoded effector
ECSP_3954	nleE	−4.0	2.6	0.6	Non-LEE-encoded effector
Metabolism and acid resistance genes
ECSP_0704	gltL	−3.9	149.2	38.4	Glutamate and aspartate transporter
ECSP_0705	gltK	−3.0	31.2	10.3	Glutamate and aspartate transporter
ECSP_0706	gltJ	−2.5	29.9	12.0	Glutamate and aspartate transporter
ECSP_0707	gltI	−3.0	156.1	52.4	Glutamate and aspartate transporter
ECSP_0906	glnQ	−2.3	277.1	118.0	Glutamine transporter subunit
ECSP_0907	glnP	−2.3	32.1	13.8	Glutamine transporter subunit
ECSP_0908	glnH	−1.5	141.1	93.5	Glutamine transporter subunit
ECSP_1977	gadC	15.6	1.3	20.5	Glutamate:gamma-aminobutyric acid antiporter
ECSP_1978	gadB	14.6	4.1	59.5	Glutamate decarboxylase B
ECSP_2312	astE	−4.0	9.7	2.4	Succinylglutamate desuccinylase
ECSP_2313	astB	−5.9	10.9	1.8	Succinylarginine dihydrolase
ECSP_2314	astD	−9.1	18.3	2.0	Succinylglutamic semialdehyde dehydrogenase
ECSP_2315	astA	ND (−)	25.5	<10−4	Arginine succinyltransferase
ECSP_2316	astC	−66.0	52.4	0.8	Succinylornithine transaminase
ECSP_2329	gdhA	−1.8	53.4	23.0	Glutamate dehydrogenase
ECSP_3185	argT	−3.5	89.1	25.3	Lysine/arginine/ornithine transporter subunit
ECSP_4065	uxaA	1.7	12.6	21.4	Altronate hydrolase
ECSP_4066	uxaC	1.8	7.8	13.6	Uronate isomerase
ECSP_4486	slp	2.0	37.4	74.3	Outer membrane lipoprotein
ECSP_4498	hdeB	ND (+)	<10−4	13.6	Acid resistance protein
ECSP_4499	hdeA	ND (+)	<10−4	134.4	Acid resistance protein
ECSP_4500	hdeD	ND (+)	<10−4	11.7	Acid resistance protein
ECSP_4501	gadE	ND (+)	<10−4	10.8	GDAR activator
ECSP_4502	mdtE	ND (+)	<10−4	11.6	Multidrug resistance efflux transporter
ECSP_4503	mdtF	ND (+)	<10−4	12.9	Multidrug resistance efflux transporter
ECSP_4504	gadW	ND (+)	<10−4	9.1	DNA-binding transcriptional activator
ECSP_4505		1.0	<10−4	0.0	Hypothetical protein
ECSP_4506	gadX	ND (+)	<10−4	5.0	DNA-binding transcriptional dual regulator
ECSP_4507	gadA	ND (+)	<10−4	18.3	Glutamate decarboxylase A
ECSP_4508	yhjA	1.0	<10−4	<10−4	Cytochrome c peroxidase
ECSP_4509	treF	ND (+)	<10−4	3.4	Cytoplasmic trehalase
ECSP_4510	yhjB	1.0	<10−4	<10−4	DNA-binding response regulator
ECSP_4511	yhjC	ND (+)	<10−4	2.6	DNA-binding transcriptional regulator
ECSP_4512	yhjD	ND (+)	<10−4	5.1	Conserved inner membrane protein
ECSP_4513	yhjE	87.3	0.2	16.3	Predicted transporter
ECSP_4921		ND	2.7	<10−4	Hypothetical protein
ECSP_4922		−1.1	78.6	74.4	Hypothetical protein
ECSP_4923	glnG	−3.7	83.9	22.4	DNA-binding response regulator
ECSP_4924	glnL	−3.7	117.1	31.6	Sensory histidine kinase
ECSP_4925	glnA	−4.3	745.3	172.4	Glutamine synthetase
ECSP_0518	glnK	ND (+)	143.4	<10−4	Nitrogen assimilation regulatory protein
ECSP_0519	amtB	−74.8	185.6	2.7	Ammonium transporter
ECSP_4793	asnC	−2.0	5.2	2.6	DNA-binding transcriptional regulator
ECSP_4794	asnA	−5.0	2,647.6	532.5	Asparagine synthetase A
ECSP_0721	asnB	−7.6	688.2	90.7	Asparagine synthetase B
ECSP_2651	cbl	−4.8	14.1	2.9	DNA-binding transcriptional activator
ECSP_2652	nac	−63.5	32.0	0.5	DNA-binding transcriptional dual regulator
ECSP_3790	galR	−2.1	13.2	6.4	DNA-binding transcriptional repressor
ECSP_3791	lysA	−4.3	23.1	5.4	Diaminopimelate decarboxylase
ECSP_3792	lysR	−2.8	2.8	1.0	DNA-binding transcriptional dual regulator
Metal stress response and import/export genes
ECSP_0334	ykgL	ND (+)	<10−4	1.4	Predicted protein
ECSP_0335	rpmJ	170.7	1.7	284.6	50S ribosomal protein L36
ECSP_0336	rpmE2	565.7	0.6	346.5	Zn(II)-responsive ribosomal protein
ECSP_0622	cusS	−1.3	7.6	5.8	Sensory histidine kinase
ECSP_0623	cusR	1.1	36.3	38.9	DNA-binding response regulator
ECSP_0624	cusC	−1.0	15.6	15.0	Copper/silver efflux system
ECSP_0625	cusF	−1.8	47.5	26.5	Periplasmic copper-binding protein
ECSP_0626	cusB	−1.3	111.6	82.8	Copper/silver efflux system
ECSP_0627	cusA	−1.4	67.5	48.0	Copper/silver efflux system
ECSP_0731	fur	1.7	38.2	64.5	DNA-binding regulator
ECSP_2430	yebA	4.5	17.0	75.9	Predicted peptidase
ECSP_2431	znuA	17.7	13.8	244.0	High-affinity zinc uptake system
ECSP_2432	znuC	2.5	4.5	18.9	High-affinity zinc uptake system
ECSP_2433	znuB	4.2	4.6	11.4	High-affinity zinc uptake system
ECSP_2579	zinT	163.7	2.0	328.1	Conserved metal-binding protein
ECSP_4263	zntR	−1.1	11.4	9.9	DNA-binding transcriptional activator
ECSP_4425	zntA	−1.4	19.9	14.3	Zinc, cobalt, and lead efflux system
Genes for other functions
ECSP_1968	ddpF	1.1	2.3	2.5	d-Ala–d-Ala transporter subunit
ECSP_1969	ddpD	1.5	1.0	1.4	d-Ala–d-Ala transporter subunit
ECSP_1970	ddpC	1.6	1.3	2.1	d-Ala–d-Ala transporter subunit
ECSP_1971	ddpB	−6.5	3.1	0.5	d-Ala–d-Ala transporter subunit
ECSP_1972	ddpA	−28.7	9.3	0.3	d-Ala–d-Ala transporter subunit
ECSP_1973	ddpX	ND (−)	7.3	<10−4	d-Ala–d-Ala dipeptidase
ECSP_2574	yedV	−1.8	2.7	1.4	Sensory kinase
ECSP_2575	yedW	1.0	48.2	26.5	DNA-binding regulator
ECSP_5434	yjiA	2.8	12.3	34.1	Predicted GTPase
ECSP_5435	yjiX	27.8	0.7	19.5	Conserved protein
ECSP_5436	yjiY	3.8	64.6	246.2	Predicted protein
ECSP_2492		−2.4	1,462.1	620.0	Hypothetical protein
ECSP_4094		−1.9	1,576.0	841.5	Hypothetical protein

^aBased on EHEC O157:H7 strain TW14359 (GenBank accession number NC_013008.1).

^bFold change was calculated as the wild-type RPKM value divided by the mutant strain RPKM value.

^cRPKM, reads per kilobase per million mapped reads.

^dND, not determined; (+), upregulated; (−), downregulated.

Twenty-six percent of the genes whose expression was altered (57 upregulated, 140 downregulated) had defined metabolic functions, suggesting a broader role for GrvA in regulation and fitness (Fig. 2B). This included genes for asparagine synthesis (asnA, asnB) and arginine degradation (ast operon) (Table 2). Nine percent (23 upregulated, 45 downregulated) were known or predicted regulators, and of these, global nitrogen metabolism regulators ntrC (also glnG) and nac, as well as cbl for sulfonate utilization, were downregulated. Eight percent (18 upregulated, 42 downregulated) were transport systems, and of these, glutamate (glt operon), glutamine (gln operon, and ammonium (amtB) transporters were downregulated. Thirty-seven percent (112 upregulated, 171 downregulated) had unknown or hypothetical functions (Table 2). Genes for zinc binding and transport, as well as resistance to zinc and other metals, were upregulated in TW14359 ΔgrvA, including znuA, znuCB, zinT (formerly yodA), rpmE2, rpmJ, ykgL, and yebA (49–51).

Alterations in the expression of astC (arginine metabolism), ler (LEE regulator), gadE (acid resistance), and znuC (zinc transport) were validated by qRT-PCR. Complementation of grvA in trans (strain TW14359 ΔgrvA/pgrvA) largely restored wild-type expression for ler and znuA but only partially restored expression for gadE and astC (Table 3). Deletion of rcsB in TW14359 ΔgrvA had no further significant impact on the expression of ler or any other targets tested (Table 3).

TABLE 3.

qRT-PCR validation of RNA-seq data

Strain	Expression unitsa
	ler	gadE	astC	znuA
TW14359	1,054.77 (51.81)	0.18 (0.06)	111.56 (1.55)	8.02 (2.34)
ΔgrvA mutant	78.28 (21.68)	61.11 (8.22)	1.21 (0.67)	551.15 (150.90)
ΔrcsB ΔgrvA mutant	127.97 (33.65)	87.87 (4.10)	2.5 (1.49)	622.03 (46.05)
ΔgrvA pgrvA mutant	556.98 (180.36)	12.94 (0.81)	6.87 (2.24)	20.46 (8.28)

^aFor each gene, expression units relative to the level of rrsH expression were calculated, and standard deviations are in parentheses. Gene name identifiers are based on EHEC strain TW14359 (GenBank accession number NC_013008.1).

GrvA requires acid resistance regulator gadE for control of ler and adherence to intestinal cells.

RNA-seq analysis revealed that deletion of grvA leads to the upregulation of gadE, a known transcriptional repressor of LEE-encoded regulator ler (52, 53). As such, the contribution of gadE to GrvA-dependent control of the LEE and adherence to cultured intestinal epithelial cells was examined.

During exponential-phase growth, gadE transcription is strongly repressed (54, 55). As such, the level of expression of ler in TW14359 ΔgadE only marginally increased compared to that in TW14359 (Fig. 3A). However, when gadE was constitutively expressed in both TW14359 ΔgrvA and TW14359 ΔgrvA ΔgadE, ler expression was reduced to the level of that observed in TW14359 ΔgrvA (Fig. 3A) (P < 0.05).

FIG 3.

GrvA requires gadE for control of LEE-dependent adherence. (A) ler transcript levels determined by qRT-PCR plotted for TW14359, TW14359 ΔgrvA, TW14359 ΔgadE, TW14359 ΔgrvA ΔgadE, and complemented strains TW14359 ΔgadE/pgadE and TW14359 ΔgrvA ΔgadE/pgadE grown in DMEM (OD₆₀₀ = 0.5). Plots which differ by lowercase letter differ significantly by Tukey's HSD test following a significant F test (n ≥ 3, P < 0.05). Error bars denote standard deviations. (B) Adherence ratio box plot for test strains (TW14359 and genetic derivatives) relative to control strain TW14359 ΔlacZ when cultured with HT-29 colonic epithelial cells (see the text for details). Box plot boundaries represent the 25th and 75th percentiles, whiskers represent the maximum and minimum values, and the mean adherence ratio is given by the horizontal line. Asterisks denote a significant difference compared with the results for TW14359, as determined by t test (n ≥ 3; *, P < 0.05; **, P < 0.01; ***, P < 0.001).

The effect of the grvA and gadE interaction on in vitro adherence was determined using a competition assay. Strains were coincubated in a 1:1 ratio with HT-29 intestinal cells, followed by plating and enumeration on MacConkey differential medium. For a control strain, lacZ was deleted from wild-type strain TW14359, and thus, the strain was Lac⁻ and colonies on MacConkey agar were white, whereas the test strains (TW14359 and genetic derivatives) were Lac⁺, producing pink colonies. For each test strain, an adherence index was determined from plate counts as the ratio of the number of CFU per milliliter for the test strain relative to the number of CFU per milliliter for the Lac⁻ wild-type control strain.

In agreement with the ler expression data (Fig. 3A), the adherence index of TW14359 ΔgrvA was significantly decreased compared to the adherence indexes of the wild type, TW14359 ΔgadE, TW14359 ΔgrvA ΔgadE, and grvA-complemented strain TW14359 ΔgrvA/pgrvA (P < 0.05) but not that of type III secretion-defective strain TW14359 ΔescN (Fig. 3B). In strains where gadE was expressed constitutively, the adherence index was uniformly reduced to levels lower than those in all other strains, including TW14359 ΔescN (P < 0.001), suggesting that GadE may impact adherence in both a LEE-dependent and a LEE-independent manner (Fig. 3B). Only in strain TW14359 grlR::kan, in which the LEE is overexpressed (20), was the adherence index increased. The results of these experiments are consistent with the hypothesis that control of the LEE and adherence by GrvA are directed through the repression of gadE.

Negative regulation of gadE by GrvA requires GadW.

GadE is a negative regulator of LEE expression, acting directly on the LEE1 promoter to repress ler transcription (52, 53). In the preceding experiments, the upregulation of ler by GrvA was determined to correspond with gadE repression. Three discrete promoters have been shown to control transcription of the gadE gene in E. coli K-12 MG1655 (56). To better understand the molecular basis underlying the regulation of gadE promoters P1 to P3 by GrvA, the levels of transcription from four different gadE lacZ reporter constructs in TW14359 and TW14359 ΔgrvA were compared during exponential-phase and stationary-phase growth in DMEM (Fig. 4A).

FIG 4.

GrvA represses gadE transcription through GadW, determined from β-galactosidase activity (in Miller units) for gadE-lacZ promoter fusions in TW14359 and genetic derivatives. (A) (Top) Activity (in Miller units) for gadE-lacZ promoter fragments Frag-1 through Frag-4 and the vector control (pRS551) plotted for TW14359 and TW14359 ΔgrvA during exponential-phase (OD₆₀₀ = 0.5) and stationary-phase (OD₆₀₀ = 2) growth in DMEM. (Bottom) Cartoon of the gadE promoter and flanking regions. Promoter (P1 to P3) positions and the positions of each cloned promoter fragment (Frag-1 through Frag-4) relative to position +1 of the initiation codon for the gadE ORF are indicated. (B) Activity (in Miller units) plotted for gadE-lacZ promoter fragments Frag-2 (left) and Frag-4 (right) in TW14359 and mutant derivatives. For panels A and B, asterisks denote significant differences by t test (n ≥ 3; *, P < 0.05; **, P < 0.01). Bars denote standard deviations. (C) Activity (in Miller units) plotted for the gadE-lacZ P3 promoter in TW14359 and mutant derivatives. Plots which differ by lowercase letter differ significantly by Tukey's HSD test following a significant F test (n ≥ 3, P < 0.05).

During exponential-phase growth, the level of transcription from gadE-lacZ containing all three gadE promoters (P1 through P3; Frag-1) was significantly increased in TW14359 ΔgrvA compared to that in TW14359 (P < 0.01) (Fig. 4A) (55). Promoter activity from fragments containing the P1 or P3 promoter alone (Frag-2 and Frag-4, respectively) was also significantly higher in TW14359 ΔgrvA (P < 0.05), while activity from the fragment containing only the P2 promoter (Frag-3) did not differ between TW14359 ΔgrvA and TW14359. As anticipated, promoter activity from all fragments increased significantly during stationary-phase growth for TW14359 (P < 0.01) yet increased only slightly from P2 (P < 0.05). For TW14359 ΔgrvA, promoter activity further increased only from P1 during stationary phase (P < 0.01) and was higher than the activity from P1 in TW14359 (Fig. 4A). This was predicted to be due to autoactivation of the P1 promoter by GadE (54), as deletion of gadE in TW14359 ΔgrvA eliminated P1 activation (Frag-2) during exponential-phase growth but had no effect on P3 activity (Frag-4) (Fig. 4B). On the basis of these findings, it is suspected that the GrvA-dependent regulation of gadE transcription is directed solely through the P3 promoter.

Transcription from the gadE P3 promoter is directly controlled by the AFI-encoded regulators GadX and GadW (56), and RNA-seq analysis of TW14359 ΔgrvA revealed the levels of expression of both genes to be elevated compared to their levels of expression in TW14359 (Table 2). It was thus predicted that repression of P3 by GrvA was mediated through one or both of these regulators. To test this, the effect of gadX and gadW deletion in TW14359 ΔgrvA on P3 activity during exponential-phase growth (OD₆₀₀ = 0.5) in DMEM was measured. Only the deletion of gadW was observed to reduce the level of transcription from P3 (Fig. 4C). Activity from P3 in TW14359 ΔgadW and TW14359 ΔgrvA ΔgadW was reduced significantly compared to that in TW14359 ΔgrvA (P = 0.01 and 0.03, respectively) but not compared to that in TW14359. Conversely, activity from P3 was slightly but significantly increased in TW14359 ΔgrvA ΔgadX compared to that in TW14359 ΔgrvA (P = 0.01) (Fig. 4C). Taken together, these data indicate that GrvA indirectly represses transcription from the gadE P3 promoter during exponential-phase growth in a manner that is dependent on gadW.

GrvA is a novel repressor of glutamate-dependent acid resistance.

The transcription of GDAR genes is growth phase dependent; expression is tightly controlled and low during exponential-phase growth but increases markedly as cells transition into stationary phase (57). Correspondingly, exponential-phase cultures are generally acid susceptible, while stationary-phase cultures are acid resistant. Since deletion of grvA was shown to upregulate GDAR genes (gadA, gadBC, gadE, gadW, gadX) during exponential-phase growth, the contribution of GrvA to GDAR was determined for exponential-phase cultures and was compared to that of the GDAR phenotype of stationary-phase cultures. In addition, since RcsB activates grvA transcription and is a dual regulator of GDAR system genes in E. coli K-12 (28, 58, 59), the interactions of rcsB and grvA in expression of the GDAR phenotype in EHEC O157:H7 were examined.

As expected, no colonies of wild-type TW14359 grown to exponential phase could be recovered on LBA following a 1-h challenge in acidified (pH 2) EG medium (Fig. 5, top). However, for strain TW14359 ΔgrvA, in which GDAR genes are upregulated during exponential-phase growth, 400 CFU/ml, corresponding to 9% of the original inoculum, was recovered. Complementation of TW14359 ΔgrvA with grvA restored acid sensitivity to wild-type levels. GDAR was abrogated following deletion of rcsB in TW14359 ΔgrvA, and complementation with rcsB restored GDAR in TW14359 ΔgrvA ΔrcsB but not in TW14359 ΔrcsB.

FIG 5.

GrvA is a growth phase-dependent repressor of acid resistance. Representative colony-forming units on LB agar for TW14359 and derivative strains following a 1-h challenge in EG medium (pH 7 versus pH 2) are shown. Cultures were tested for acid resistance during exponential-phase (top) and stationary-phase (bottom) growth in DMEM. See the text for details.

In stationary phase, during which GDAR is actively expressed (57), deletion and complementation of grvA had no effect on survival in acid, with 100% of the initial inoculum being recovered following a 1-h challenge in acidified EG medium (Fig. 5, bottom). Thus, control of GDAR by GrvA is relegated to exponential-phase growth. In keeping with the requirement for rcsB in stationary-phase GDAR, no growth was observed for TW14359 ΔrcsB and TW14359 ΔgrvA ΔrcsB unless they were complemented with rcsB (Fig. 5, bottom).

DISCUSSION

This study has examined aspects of grvA transcriptional regulation, has defined the GrvA regulon, and has identified genetic determinants underlying GrvA-dependent control of discrete pathogenic mechanisms in EHEC. RcsB, the response regulator of Rcs phosphorelay in E. coli, is predicted to directly activate transcription of grvA, leading to the upregulation of LEE-dependent adherence and repression of glutamate-dependent acid resistance (Fig. 6). While it is not yet clear what cis-acting element(s) is required for RcsB-dependent activation of grvA, this study demonstrated the binding of RcsB to a grvA promoter fragment containing two putative tandem RcsB binding sites located proximal to a predicted σ⁷⁰ −35 site. These sites share some homology with the RcsAB box consensus sequence described for rcsA, wza, and flhDC promoters (30, 60), yet RcsA, which is highly unstable at elevated temperatures (27), is not required for RcsB-dependent regulation of the LEE (21), and RcsAB generally bind distal to and upstream of the promoter core (31, 61, 62). The binding of RcsB to grvA promoter fragments is more consistent with regulation as a homodimer. RcsB homodimers bind a consensus sequence similar to the RcsAB box and proximal to the −35 site. Of the two putative RcsB binding sites in the grvA promoter, the upstream site (Fig. 1B, Box 1) shares similarities with homodimer sites described for osmC and rprA promoters (63). RcsB has also been shown to negatively regulate grvA expression. In strains of E. coli O157:H7 harboring a 49-bp deletion in rcsB, trans-complementation with wild-type rcsB decreased grvA expression (64). This was observed, however, only during culture at 28°C and correlates with reduced grvA expression during growth at 14°C and 25°C (65). Together, this suggests that regulation of grvA by RcsB is temperature sensitive. There is precedent for this, as repression of flhDC transcription by RcsB occurs at both low and high temperatures. At low temperatures, RcsB partners with RcsA (30), whereas at high temperatures, the LEE-encoded regulator GrlA is required (20).

FIG 6.

Model predicting GrvA-dependent regulation of acid resistance and LEE-dependent adherence. During exponential-phase growth, GrvA represses acid resistance (GDAR) and activates LEE-dependent adherence. This requires the negative regulation of gadW by GrvA. As GadW is a direct activator of gadE, the central regulator of GDAR, its repression by GrvA downregulates GDAR gene expression (gadE, gadA, gadBC) and acid resistance. GadE also directly represses the LEE-encoded regulator ler, the product of which activates the LEE (5 operons encoding the T3S, other regulators, and toxic effectors). The repression of GadW/GadE by GrvA leads to activation of the LEE and LEE-dependent adherence. Bicarbonate in the intestine is hypothesized to stimulate GrvA-dependent control of GDAR and the LEE in a manner requiring the Rcs phosphorelay system. RcsB activates GDAR as a heterodimer with GadE. The role for GrvB in this regulatory pathway is unknown. GrvA regulates the expression of >700 genes in EHEC O157:H7 strain TW14359. The model is inferred from the results of experiments performed in this and previous studies (20, 21, 52, 53, 55). See the text for further details. GDAR, glutamate-dependent acid resistance; LEE, locus of enterocyte effacement.

Overexpression of grvA increases transcription from the LEE1 promoter and correspondingly leads to increased adherence (21). Consistent with this, RNA-seq analysis of TW14359 ΔgrvA in this study revealed the reduced expression of genes belonging to both the LEE1 and the LEE2-LEE3 operons, including LEE-encoded regulator ler, T3S structural genes (esp genes), and the translocated intimin receptor (tir) gene. This study is the first to show that grvA deletion downregulates non-LEE-encoded T3S effector genes (nle genes). These proteins broadly impact host signaling cascades through inhibition of NF-κB (NleC, NleE, NleD, and NleH2) (66–68) and caspase activation (NleF) (69), as well as interfere with host vesicle trafficking (NleA) (70). How these nle genes are regulated by GrvA is unknown. One hypothesis is through its regulatory effects on LEE expression. Both LEE-encoded activators Ler and GrlA regulate nleA expression (71–73). Moreover, LEE-encoded GrlR has been shown to repress nleB and nleH (71). However, non-LEE-encoded regulators involved in quorum sensing (QseA) and nucleoid structuring (H-NS) have also been implicated in the control of nle gene expression (71, 74). Whatever the mechanism, these findings suggest a more comprehensive role for GrvA in EHEC pathogenesis that includes the coordination of T3S-dependent colonization with immune subversion.

This study identifies GrvA to be a new regulator of glutamate-dependent acid resistance (GDAR) genes (gad genes) (Fig. 6). Deletion of grvA led to the upregulation of gad regulatory (gadE, gadW and gadX) and enzymatic/structural (gadA and gadBC) genes and corresponded with a 100-fold increase in acid survival by the GDAR mechanism. These findings greatly expand the role for GrvA in the pathogenic behavior of EHEC to include the control genes essential for the transmission of this pathogen in acidic food matrices, gastric passage, and a low oral infectious dose. Since grvA is needed for full bicarbonate induction of the LEE (20), it is plausible that it helps to coordinate expression of the LEE and GDAR systems upon entry into the intestine (Fig. 6).

Genes with no direct role in virulence were also shown to be regulated by GrvA. Most notably, glnG (also ntrC) and nac expression was positively regulated by GrvA. Nitrogen regulatory protein C (NtrC) is a transcriptional activator of nitrogen assimilation control (Nac), and together these regulators are responsible for coordinating the expression of genes for nitrogen metabolism under limiting conditions (75). That ntrC and nac are positively controlled by GrvA is consistent with the reduced expression of operons for the utilization of glutamate (glt), glutamine (gln), arginine (ast), and asparagine (asn), as well as for the transport of ammonium (amtB) observed in the TW14359 ΔgrvA background. How nitrogen availability influences GrvA-dependent control of these mechanisms and the importance of grvA to nitrogen assimilation are unknown.

GrvA was shown to activate LEE-encoded master regulator ler through downregulation of gadE, the product of which directly represses ler transcription in EHEC (52, 53, 76). Negative control of gadE by GrvA is predicted to occur indirectly through the P3 promoter, requiring an intact gadW (Fig. 6). This is supported by both gadE-lacZ promoter fusions and RNA-seq analysis in this study. GadW is an AraC-family regulator that acts as a homodimer or, when partnered with GadX, to control the acid fitness genes gadBC, hdeA, hdeB, and gadE (54, 55, 77, 78). The details of how GrvA regulates transcription are not yet known. For the regulation of gadW, there are two predicted promoters encoded in tandem and just upstream of the gadW (gadW_P1 and gadW_P2) (77). It is suspected that gadW_P1 is the target of GrvA repression, as transcripts from this promoter, but not those from gadW_P2, were increased in TW14359 ΔgrvA by RNA-seq (data not shown). Multiple trans-acting factors are known to directly influence gadW transcription, including GadE, PhoP, and SdiA (activators) (79–81), as well as GadX, GadW, RutR, Fnr, and H-NS (repressors) (77, 82–84). It is unlikely that gadX contributes to GrvA-dependent repression of gadW, as gadX mutation in TW14359 ΔgrvA was not observed to further alter gadE expression. Save for gadE, no other known regulators of gadW were modified in expression by RNA-seq analysis, and a binding consensus sequence for GrvA has yet to be defined. GrvA shares some homology (29% identity, 79/270 amino acids) with MarT of Salmonella enterica serotype Typhimurium (85–87), most notably, the first 150 residues of the N terminus containing a winged-helix DNA-binding domain. marT is encoded on Salmonella pathogenicity island 3 (SPI3), and its product is a direct transcriptional activator of the misL autotransporter, which binds fibronectin and increases intestinal colonization and invasiveness (88, 89). On the basis of its similarity to CadC of E. coli, MarT is predicted to activate misL by relieving H-NS-mediated repression (33). If the regulation of gadW by GrvA is by a direct mechanism, however, it is unlikely that H-NS is involved, as deletion of grvA increases gadW expression. Future studies will be aimed at identifying cis elements for GrvA-directed transcriptional regulation.