Anaerobic Regulation of Shigella flexneri Virulence: ArcA Regulates fur and Iron Acquisition Genes

Bacterial strains and plasmids are described in Table 1. E. coli strains were grown in Luria-Bertani (LB) broth or on LB agar (16). S. flexneri strains were grown in RPMI (RPMI 1640 [Invitrogen] with l-glutamine, without phenol red [Gibco; Invitrogen], and buffered with 100 mM HEPES) supplemented with 2.5 μM FeSO₄ where indicated or on tryptic soy broth agar plus 0.01% Congo red dye (Congo red agar) at 37°C. Antibiotics were used at the following concentrations (per milliliter): 125 μg of carbenicillin (Car), 200 μg of streptomycin (Str), 20 μg of kanamycin (Kan), and 7.5 μg of chloramphenicol (Cam).

For green fluorescent protein (GFP) reporter and real-time reverse transcription (RT)-PCR assays, S. flexneri strains were plated on Congo red agar with appropriate antibiotics and grown overnight at 37°C. Isolated colonies were used to inoculate 2 ml RPMI with 2.5 μM FeSO₄ and appropriate antibiotics for plasmid maintenance. After overnight growth at 37°C, cultures were diluted to an optical density at 600 nm (OD₆₀₀) of 0.05 into 2 ml RPMI containing antibiotics and grown aerobically at 37°C for 2 h. These were subcultured to an OD₆₀₀ of 0.03 into the same medium and grown to exponential phase either aerobically with vigorous shaking or anaerobically in Oxoid AnaeroJars with AnaeroGen sachets and Anaerobic Indicators (Oxoid Ltd., Hampshire, England), as indicated. All data are the average of at least three independent experiments.

Henle cells (Intestine 407 cells; American Type Culture Collection, Manassas, VA) were grown in minimal essential medium (MEM) or RPMI 1640 (Invitrogen) supplemented with 10% Bacto tryptose phosphate broth (Difco; Becton Dickinson Company), 2 mM glutamine, MEM nonessential amino acid solution (Invitrogen Corporation), and 10% fetal bovine serum (Invitrogen) in a 5% CO₂ atmosphere at 37°C. Plaque assays were performed as described previously (52), except that the medium was supplemented with 100 mM HEPES (pH 7.5) and agarose was omitted from the overlay. For the anaerobic plaque assays, the plates were incubated in the BD BBL GasPack pouch anaerobic system (Becton, Dickinson, and Company) for the duration of the experiment.

The S. flexneri arcA::kan mutant was constructed by bacteriophage P1 transduction (56) from the E. coli arcA::kan mutant ECL5331 (30). Insertional inactivation of the arcA gene in MBF200 was confirmed by PCR using primers flanking the arcA gene, IWDArcAOut1 and IWDArcAOut2.

The S. flexneri fnr strain was made by allelic exchange. The fnr region was amplified from SA100 by PCR using primers MWDFnrUS1 and MWDFnrDS1, digested with PstI and SalI, and ligated with the pDRIVE vector. A cam cassette was excised from pMTLcam using SmaI and was ligated into pDRIVEfnr digested with BclI and made blunt with the Klenow fragment of DNA polymerase I (New England Biolabs, Ipswich, MA). fnr::cam was then excised from pDRIVEfnr::cam using SmaI and ligated into SmaI-digested pCVD442N. The resulting plasmid, pCVD442fnr::cam, was then mated into SM100 by triparental conjugation. Primary integrants were selected by growth in the presence of Car, Str, and Cam and verified with primers MWFFnrDS1 and FNRintFor. The fnr::cam mutant was isolated by growth in the presence of sucrose and Cam and was confirmed by PCR using primer pair FNRintFor and FNRintRev and primer pair MWFnrUS1 and MWFFnrDS1. MBF100 (SA100 fnr::cam) and MBF300 (SA100 arcA::kan fnr::cam) were obtained by P1 transduction of fnr::cam from SM100 fnr::cam to SA100 and MBF200, respectively, and verified by PCR. All primer sequences are listed in Table S1 in the supplemental material.

A single-copy, IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible arcA vector was engineered by first cloning the arcA gene under the inducible T5 promoter of plasmid pQE-2 (QIAGEN). SA100 genomic DNA was used as a template for PCR using the primers MBarcAforQE and MBarcArevQE (see Table S1 in the supplemental material), and the fragment was cloned into pQE-2 digested with BseRI and HindIII to generate pMBarcAQE. The FspI fragment, including the lacI and T5-arcA genes, was excised from pMBarcAQE and ligated into the blunt-cloning-ready pCC1 vector (Epicenter Biotechnologies, Madison, WI), resulting in pMBarcAccQE, which was verified by DNA sequencing.

A plasmid for IPTG-inducible expression of fur was constructed by PCR amplification of the fur gene from SA100 DNA using primers MBfurForQE and MBfurRevQE (see Table S1 in the supplemental material). The PCR product and pQE-2 vector were digested with BseRI and HindIII and ligated, and the resulting pMBfurQE plasmid was confirmed by DNA sequence analysis.

Microarrays were printed and postprocessed as described previously (45). Wild-type S. flexneri batch cultures were grown aerobically in RPMI medium without added iron to mid-logarithmic phase in a BIOFLO 110 Fermentor/Bioreactor (New Brunswick Scientific, Edison, NJ) to maintain constant pH, dissolved oxygen concentration, temperature, and agitation. A portion of the culture was removed to obtain the aerobically grown bacteria. The dissolved oxygen concentration was then depleted in the remaining culture by the addition of nitrogen gas, and 15 min after depletion, the anaerobic bacteria were isolated. RNA was purified with RNeasy Midi kits (QIAGEN). Reverse transcription of RNA to generate amino allyl-dUTP-incorporated cDNA, Cy3 and Cy5 coupling, probe generation, and array hybridization were performed as described previously (45), with RNA derived from aerobically grown bacteria labeled with Cy3 and from anaerobically grown strains labeled with Cy5. Microarrays were scanned by the Genepix array scanner 4000A (Axon Instruments, Union City, CA). Preliminary analysis of microarrays was performed with Genepix 5.0 software, and normalization of microarray data was carried out by the Longhorn Array Database, an open-source, MIAME-compliant implementation of the Stanford Microarray Database (28). Normalized data were filtered so that spots with a regression correlation of lower than 0.6 and those that were in areas of high background were excluded. Additionally, genes that did not exhibit greater than a twofold difference in expression in at least two arrays and those that showed inconsistent patterns of induction or repression were excluded from further analysis.

gfp transcriptional fusions (Table 1) were constructed by PCR amplification of the promoter of each gene (primers are listed in Table S1 in the supplemental material) and ligation into the promoterless gfp vector pLR29. To measure GFP, overnight cultures were diluted to an OD₆₀₀ of 0.05 into RPMI with Car and 100 mM HEPES and grown aerobically at 37°C for 2 h. These were subcultured to an OD₆₀₀ of 0.03 into the same medium and grown for 2 h either aerobically with vigorous shaking or anaerobically in Oxoid AnaeroJars with AnaeroGen and Anaerobic Indicators (Oxoid Ltd., Hampshire, England). The cultures were diluted to an OD₆₀₀ of 0.08, and fluorescence was measured in a VersaFluor fluorometer (Bio-Rad Laboratories, Hercules, CA). The instrument was blanked using the parent strains with the pLR29 control plasmid, and the range was set to 15,000 relative fluorescence units (RFU) with the constitutive pMBminC vector, which expresses gfp constitutively. The data are shown as relative expression levels, with the culture giving maximal expression set at 100%. The average RFU/OD₆₀₀ of each transcriptional fusion was normalized to the RFU/OD₆₀₀ of the pMBminC vector in the same strain. Results are the average of three independent experiments. The plasmids pMBnarG and pMBlldP served as positive controls for Fnr and ArcA regulation, respectively.

S. flexneri strains SA100/pCC1, MBF200/pCC1, and MBF200/pMBarcAccQE were grown as for the GFP reporter assays, except that 1 μM IPTG was added to the final subculture medium to induce ArcA expression; anti-ArcA (generously provided by P. Silverman [59]) immunoblots showed that this amount of IPTG induced wild-type levels of ArcA. RNA was isolated on RNeasy Mini columns (QIAGEN) following the addition of 1/5 volume of 95% ethanol-5% phenol (vol/vol) to logarithmically growing, anaerobic cultures. RNA was DNase treated (DNase I; QIAGEN) on the RNeasy column and again after elution with amplification-grade DNase I (Invitrogen) according to the manufacturers' instructions. cDNA was generated from approximately 5 μg of each RNA sample with the High Capacity cDNA Archive kit (Applied Biosystems). Real-time RT-PCR mixtures in a total volume of 25 μl contained 1× Power SYBR green PCR Master Mix (Applied Biosystems), 800 nM concentrations of the indicated primers, and 1/200 of the cDNA reaction mixture. fur cDNA was detected with primers MBfurRT1 and MBfurRT2. rrsA cDNA was detected with primers RrsA.for and RrsA.rev. Real-time RT-PCR and analyses were carried out with an Applied Biosystems 7300 Real Time PCR System and software. Standard curves for each primer set were generated by using cDNA obtained from 10-fold dilutions of SA100 RNA, and the amount of cDNA in each sample was extrapolated from the standard curve. The relative amounts of fur cDNA were normalized by dividing the values by the relative amounts of rrsA control cDNA in each sample.

The promoter-gfp fusion plasmids served as templates for PCRs to generate probes for the promoter regions, using primers listed in Table S1 in the supplemental material. After gel extraction of the PCR fragments, the DNA was digested with XmaI for probes generated with the pLR29EMSAfor 5′ primer, which cuts just upstream of the BamHI restriction enzyme site in the pLR29 vector, while the iuc probe was digested with XbaI. The probes were gel purified, and the ends were filled in using Klenow fragment (New England Biolabs, Ipswich, MA) and a mixture of nucleotides for cold probes. dCTP was replaced with [α-³²P]dCTP (Perkin Elmer, Boston, MA) for radiolabeled probes. Unincorporated nucleotides were removed with Micro Bio-Spin P-30 Tris chromatography columns (Bio-Rad Laboratories), and all probes were phenol-chloroform extracted and ethanol precipitated (1). Radioactivity was measured by liquid scintillation counting, and probe concentrations were determined by measuring the absorbance.

Cell extracts from MBF200/pCC1 and MBF200/pMBarcAccQE were prepared by the method of Tardat and Touati (61), with the exception of growth conditions. S. flexneri cultures were grown overnight in HEPES-buffered RPMI with 2.5 μM FeSO₄ and antibiotics, subcultured at 1:100 into the same medium without added iron, and grown aerobically at 37°C to mid-logarithmic phase. The cultures were then diluted into the same medium containing 1 μM IPTG to induce arcA expression, and the cultures were grown for 2 h under anaerobic conditions.

The antibody supershift assays were performed essentially as described by Ausubel et al. (1). Binding buffer consisted of 10 mM Tris-HCl (pH 7.4), 10% glycerol, 10 mM CaCl₂, 100 mM KCl, 1 mM EDTA, 5 μg/ml bovine serum albumin, 1 mM dithiothreitol, and 1 μg poly(dI-dC) in a reaction volume of 30 μl, also containing 5 μg of crude protein extract, 1 μl of antiserum diluted 1/1,000, and approximately 1 ng of labeled probe. Fragments were separated by electrophoresis in a 5% polyacrylamide-Tris-borate-EDTA gel, and radioactive bands were visualized with a Bio-Rad Molecular Imager FX after overnight exposure of the dried gel to a phosphor screen (Bio-Rad Laboratories). Fluorophore band intensity was analyzed with Quantity One software (Bio-Rad Laboratories).

The plaque assay, a standard measure of S. flexneri virulence, is performed under aerobic conditions, which may not allow assessment of factors influenced by the anaerobic environment of the colon. Therefore, we developed a modified plaque assay in which infected Henle cell monolayers were incubated anaerobically. As shown in Fig. 1, wild-type S. flexneri formed plaques under both aerobic and anaerobic conditions, whereas the avirulent Crb⁻ strain was unable to form plaques under either condition. S. flexneri mutants defective in one or more iron transport systems (29, 52) were also tested in this assay (Fig. 1), since we had previously shown that the S. flexneri iuc, feo, and sit iron transport systems contribute to iron acquisition in cultured cells. A mutant lacking all three of these systems, which is defective in aerobic plaque formation (52), also failed to form plaques anaerobically (Fig. 1), while all strains expressing at least two of these three iron transport systems formed plaques in either environment (data not shown). The strain with only the Sit transporter was also able to form plaques under both conditions, although under anaerobic conditions the plaque size was reduced compared to that of the wild type (feo iuc mutant [Fig. 1]). The iuc sit double mutant, in which only the Feo system is functional, formed plaques under anaerobic but not aerobic conditions. This is consistent with the ferrous iron ligand for the Feo system being enriched under anaerobic conditions. In contrast, the feo sit double mutant, which expressed only the aerobactin system, formed plaques solely under aerobic conditions, as might be expected for a ferric iron transporter. This effect of oxygen availability on the function of these iron transport systems suggested that the regulation of their genes could also be influenced by oxygen.

Because plaque formation was influenced by the availability of oxygen, microarray analysis was used to screen S. flexneri iron transport genes or other virulence genes for the effect of oxygen on expression. Wild-type cultures grown aerobically were compared with those subjected to oxygen depletion. Genes previously reported to be influenced in E. coli by oxygen availability or by the anaerobic transcription factor ArcA or Fnr (25, 30, 54, 55) showed the expected regulation in S. flexneri (Table 2). Representative genes induced aerobically included the TCA cycle loci acnA, mdh, and lpd-ace, the F₁F_o ATPase atpCD GAHF genes, the ndk/nrd aerobic ribonucleotide reductases, and genes encoding Fe-S cluster biogenesis and oxidative stress proteins. Under anaerobic conditions, the genes encoding nitrate reductase (narGHJI) and fermentation genes (adhE and pflB) were elevated (Table 2). This indicates that the conditions used for analysis of S. flexneri were appropriate for detecting anaerobically induced or repressed genes. The transcription of several iron acquisition genes also responded to oxygen availability. The feo genes were induced anaerobically (Table 2). Conversely, the sit and iuc transcripts were more abundant aerobically (Table 2). dps, which has been shown to bind ferrous iron (4), was also elevated anaerobically (Table 2).

The regulatory effects detected by microarray were further analyzed with GFP reporter fusion assays (Fig. 2). Transcriptional gfp reporter fusions in the vector pLR29 were made using promoters upstream of the feoABC, sitABCD, and iucABCD iutA operons. Strains of wild-type S. flexneri containing these reporter fusions were grown aerobically or anaerobically to compare the relative fluorescence. The reporter data confirmed that the iuc and sit promoters were more active aerobically than anaerobically, while the feo promoter was stronger anaerobically (Fig. 2).

To determine whether the transcription factors ArcA and Fnr have a role in the regulation of iron acquisition genes in response to oxygen availability, arcA and fnr mutants of S. flexneri were constructed and the relative expression from the gfp reporter fusions was determined in these strains. The activity of iuc promoter fusions was the same in wild-type and fnr strains (data not shown) but was slightly elevated in the arcA mutant (Fig. 3A). This suggests that ArcA represses iuc, and a putative ArcA box was found in the promoter sequence (Fig. 3B). In contrast, the activity of the feo promoter was approximately half of the wild-type level in the arcA and fnr single mutants and one-fifth of the wild-type level in the arcA fnr double mutant (Fig. 4A), indicating that the anaerobic induction of feo transcription in the wild-type strain is due to Fnr and ArcA and that their effects are additive.

The Fnr regulation of feo transcription was expected since induction of the feo promoter under anaerobiosis was shown previously to be Fnr dependent in E. coli. The S. flexneri feo promoter sequence is identical to that of E. coli feo, including a site with near identity to the Fnr consensus sequence (TTGAT[n₄]ATCAA) (24). However, ArcA regulation of the feo promoter has not been shown previously. Although consensus sequences for the ArcA recognition site have been reported in the literature, the optimal ArcA binding sequence is not known (13, 14, 30, 32, 39, 40). Using parameters from one of these predictions (14) and adding sequences from newly identified ArcA binding sites, we generated an ArcA consensus sequence (Fig. 5) and used this sequence to identify a putative ArcA binding sequence in the feo promoter (Fig. 4B). To determine whether this site is important for ArcA regulation of feo, several bases were changed to the least-common bases found at those positions in the weight matrix of the ArcA box consensus sequence. The activity of the altered feo promoter (feo _Alt) was similar in wild-type and arcA strains (Fig. 4A), indicating that ArcA did not stimulate expression from the altered feo promoter. This confirms that ArcA stimulates feo transcription and that the sequences changed in the feo _Alt promoter are required for stimulation. The activity of the altered feo promoter in the fnr mutant was reduced to approximately half its level in the wild-type strain (Fig. 4A), which indicates that Fnr binding was not disrupted by the base changes. This is consistent with the sequence prediction that the Arc and Fnr binding sites do not overlap (Fig. 4B).

The sit promoter fusion was also examined for ArcA regulation. The average relative gfp expression of the reporter fusion in the arcA mutant grown under aerobic conditions (0.98 ± 0.01) was the same as under anaerobic conditions (0.98 ± 0.2), indicating that the higher level of expression seen under aerobic conditions in the wild type (Fig. 2) required ArcA. However, no sequence matching the ArcA consensus was identified in the sit promoter region, suggesting that an effect of ArcA on sit expression might not be direct.

Since several iron acquisition genes were regulated in response to oxygen availability (Table 2), the effect of ArcA and Fnr on transcription of fur, which encodes the regulator of iron transport genes, was determined. Real-time RT-PCR was performed with RNA isolated from anaerobically grown wild-type S. flexneri and the arcA, fnr, and arcA fnr mutants. While there was no difference in the level of fur mRNA between the wild type and the fnr mutant (data not shown), there was a significant increase of the fur transcript in strains lacking arcA (Fig. 6A), suggesting that fur transcription was repressed in an ArcA-dependent manner. fur repression was restored in the arcA strain by inducing arcA expression from a plasmid (Fig. 6A).

To confirm that ArcA regulates fur promoter activity, a gfp reporter fusion to the fur promoter was constructed and the relative fluorescence of gfp was determined in the wild-type and arcA mutant strains. Relative to the wild type, the level of GFP was elevated in the arcA mutant (Fig. 6B), indicating that this promoter was negatively regulated by ArcA. Several bases matching the ArcA box consensus sequence in the fur promoter were then changed to the least-common bases occurring at those positions in the weight matrix derived from ArcA boxes of known ArcA-regulated promoters (fur _Alt [Fig. 5 and Fig. 6C]). The activity of the altered fur promoter was higher than that of the native fur promoter in the wild-type ArcA⁺ strain, and the difference between the amount of GFP in the arcA strain relative to the wild type was significantly decreased (Fig. 6B). These results were consistent with ArcA repression of fur transcription and implicated the altered bases in contributing to ArcA binding at this promoter.

The presence of putative ArcA binding sites in the fur, iuc, and feo promoters suggested that the observed regulation of these genes by oxygen availability was mediated directly by ArcA. To detect binding of ArcA to the feo, iuc, and fur promoters, anti-ArcA antibody supershift assays were performed (Fig. 7). The lld (lct) promoter, which contains a known ArcA binding site (32), was used as a positive control. Incubation of the lld, iuc, feo, or fur promoter with an extract containing ArcA and anti-ArcA antibody slowed electrophoretic mobility (Fig. 7). Incubation of the probes with extract from the arcA mutant did not affect mobility of the radiolabeled probe (Fig. 7). The altered feo and fur probes were also tested, and as predicted from the gfp expression results, the altered probes showed reduced anti-ArcA-dependent supershifting (Fig. 7; compare feo with feo _Alt and fur with fur _Alt). Some binding of ArcA to the altered probes was still observed, particularly with fur _Alt. The amount of probe in the supershifted bands was quantified by measuring the radioactivity; the mean counts in the supershifted region were 3,910 for the native feo, compared with 1,511 for feo _Alt, and 1,918 for fur, compared with 1,607 for fur _Alt. The relatively small effect of changing the fur promoter may reflect the fact that there are regions with homology to the consensus ArcA box on both DNA strands of the fur probe. The base changes introduced alter each site differently, and there may be residual ArcA binding to one or both sequences on the altered fur probe. ArcA did not bind the sit probe, as indicated by the lack of supershifting (Fig. 7), which was in agreement with our failure to find an ArcA box sequence in the promoter. These data demonstrate that ArcA directly binds the feo, iuc, and fur promoters and that the altered bases of feo are within the region required for DNA recognition by ArcA.

The expression of iron acquisition genes is regulated by ArcA, Fnr, and Fur, and since iron uptake is required for plaque formation by S. flexneri, we assessed the importance of these regulators in aerobic and anaerobic plaque assays. The arcA and fnr single mutants formed plaques under anaerobic conditions (Fig. 8A). In the absence of oxygen, these mutants also formed plaques but there was a slight reduction in plaque size (Fig. 8A). These data indicate that ArcA and Fnr exhibit a partial redundancy in the ability to regulate one or more of the steps required for plaque formation. In contrast, the arcA fnr double mutant was able to form plaques only in the presence of oxygen (Fig. 8A). Additionally, the double mutant formed plaques when the infected Henle cells were incubated aerobically for 2 days following the usual period of anaerobic incubation (data not shown). This indicates that under anaerobic conditions, the double mutant could invade and remain viable inside epithelial cells but was defective in intracellular growth or cell-to-cell spread. Since these transcription factors influence iron acquisition, the defective anaerobic plaque formation by arcA fnr mutants may reflect reduced feo expression and aberrant expression of fur.

The effect of altered fur expression on the ability of S. flexneri to form plaques was then investigated. An S. flexneri fur mutant and the wild-type strain expressing fur from an IPTG-inducible plasmid were assayed for aerobic and anaerobic plaque formation in the Henle cell model of infection. Overexpression of fur prevented plaque formation, while a mutation in fur reduced the number of plaques formed (Fig. 8B). The effects of altered expression of fur on plaque formation occurred in both aerobic and anaerobic environments, indicating that proper regulation of fur is critical for the virulence of S. flexneri regardless of oxygen availability.