Epigenetic regulation of the nitrosative stress response and intracellular macrophage survival by extraintestinal pathogenic Escherichia coli

Authors

  • Stacey L. Bateman,

    1. Department of Molecular Genetics and Microbiology
    2. Center for Microbial Pathogenesis
    Search for more papers by this author
  • Patrick C. Seed

    Corresponding author
    1. Department of Molecular Genetics and Microbiology
    2. Center for Microbial Pathogenesis
    3. Department of Pediatrics, Duke University School of Medicine, Durham, NC 27710, USA
    Search for more papers by this author

E-mail patrick.seed@duke.edu; Tel. (+1) 919 684 9590; Fax (+1) 919 681 2089.

Summary

Extraintestinal pathogenic Escherichia coli (ExPEC) reside in the enteric tract as a commensal reservoir, but can transition to a pathogenic state by invading normally sterile niches, establishing infection and disseminating to invasive sites like the bloodstream. Macrophages are required for ExPEC dissemination, suggesting the pathogen has developed mechanisms to persist within professional phagocytes. Here, we report that FimX, an ExPEC-associated DNA invertase that regulates the major virulence factor type 1 pili (T1P), is also an epigenetic regulator of a LuxR-like response regulator HyxR. FimX regulated hyxR expression through bidirectional phase inversion of its promoter region at sites different from the type 1 pili promoter and independent of integration host factor (IHF). In vitro, transition from high to low HyxR expression produced enhanced tolerance of reactive nitrogen intermediates (RNIs), primarily through de-repression of hmpA, encoding a nitric oxide-detoxifying flavohaemoglobin. However, in the macrophage, HyxR produced large effects on intracellular survival in the presence and absence of RNI and independent of Hmp. Collectively, we have shown that the ability of ExPEC to survive in macrophages is contingent upon the proper transition from high to low HyxR expression through epigenetic regulatory control by FimX.

Introduction

Extraintestinal pathogenic Escherichia coli strains (ExPEC) are a commensal–pathogen subset of E. coli that are of agricultural, veterinary and medical importance in avian and mammalian hosts. In humans, ExPEC may transition from benign occupation of the enteric and vaginal tracts to sterile sites such as the urinary tract, bloodstream and central nervous system (Orskov and Orskov, 1992). To mount successful infections after breeching host barriers, ExPEC must circumvent, suppress or tolerate the host innate immune system, including chemical stresses such as reactive nitrogen and oxygen species as well as cellular immune effectors. As ExPEC transitions from the enteric and urinary tracts, for instance, it may encounter professional phagocytes such as macrophages. Recent research suggests that the K1 serotype of ExPEC, a leading cause of neonatal meningitis (Harvey et al., 1999; Gaschignard et al., 2011), requires macrophages for dissemination to the central nervous system, spleen and lungs (Mittal et al., 2010). The regulatory cues that provoke the commensal to pathogen transition and the factors required for survival in macrophages as a potential vehicle for disseminated infections are incompletely understood.

One major, but perhaps still underexplored, form of virulence regulation is at the epigenetic level. A well-known example of epigenetic regulation in E. coli is that of type 1 pili (T1P), a virulence determinant required for adherence to and invasion of the bladder epithelium (Mulvey et al., 1998). Bacteria lacking type 1 pili are attenuated during experimental cystitis secondary to reduced adherence and invasion (Connell et al., 1996; Wright et al., 2007), and corroborating vaccination studies with pili subcomponents have been shown to reduce the severity and duration of infection (Abraham et al., 1985; Palaszynski et al., 1998; Langermann et al., 2000). Among virtually all E. coli, two DNA recombinases called FimB and FimE act as site-specific invertases, controlling the orientation of the type 1 pili promoter (denoted as fimS) such that it is positioned in an OFF or ON orientation (Abraham et al., 1985). FimB and FimE recombination acts in concert with integration host factor (IHF), an essential DNA-bending accessory factor, without which the promoter remains phase-locked (Dorman and Higgins, 1987; Eisenstein et al., 1987; Blomfield et al., 1997). This binary switch produces the hierarchically dominant control over the expression of T1P such that in the OFF orientation, transcription of the T1P operon is completely arrested. In the ON orientation, like most promoters, there are subsequent rheostat controls that alter the level of transcription, including CRP-cAMP and H-NS, among others (Schembri et al., 1998; O'Gara and Dorman, 2000; Muller et al., 2009).

Recently, additional FimB homologues were identified among ExPEC. Prior studies have demonstrated that ExPEC may encode up to five such homologues with FimX and IpuA sharing invertase activity with FimB and FimE at the T1P promoter (Bryan et al., 2006; Xie et al., 2006; Hannan et al., 2008). Through molecular epidemiology studies, we have recently found that fimX is the only Fim-like family member highly associated with ExPEC strains, being present in > 80% of such isolates, while only present in < 25% of commensal strains (S.L. Bateman and P.C. Seed, unpubl. data). Prior work has shown that FimX is sufficient to mediate the phase OFF to ON transition and thus drive T1P expression in the urinary tract, ultimately resulting in the initiation and perpetuation of cystitis (Hannan et al., 2008). However, the apparent functional redundancy among the Fim-like recombinases, particularly FimB and FimX, combined with the molecular epidemiology data suggested that FimX may have activities beyond regulating type 1 pili that are important for maintaining the ExPEC lifestyle.

In our current work, we demonstrate that fimX expression produces unidirectional phase inversion of the T1P promoter and bidirectional inversion of the promoter for hyxR, a gene immediately adjacent to fimX and encoding a LuxR-like response regulator. We demonstrate that FimX produces inversion of the hyxR promoter at inversion sites dissimilar compared with those flanking the type 1 pili promoter and, unlike the type 1 pili promoter, in a manner independent of IHF and H-NS. The consequence of HyxR expression is suppression of ExPEC tolerance and survival during reactive nitrogen intermediate (RNI) stress. We further show that HyxR acts to suppress RNI-dependent and -independent intracellular survival during experimental infection of macrophage-like cells. Together, this study demonstrates the functional versatility of the FimX recombinase and identifies novel epigenetic and transcriptional regulatory controls for ExPEC survival under RNI and macrophage challenges.

Results

FimX regulates hyxR through promoter phase inversion

Many recombinases regulate genes adjacent to their coding loci, similar to FimB- and FimE-mediated recombination of T1P. Therefore, we hypothesized that FimX, whose gene is immediately adjacent to hyxR, hyxA and hyxB on the pathogenicity islet called PAI-X (S.L. Bateman and P.C. Seed, unpubl. data), has a role in phase variation of their respective promoters, thus providing epigenetic control over their expression. To monitor promoter inversion of the hyx genes on the PAI-X islet, we amplified the 5′ intergenic region plus flanking sequences of each gene (hyxR, hyxA, hyxB) and then digested the amplicon with multiple restriction enzymes, predicting that phase inversion would result in re-orientation of the asymmetric restriction site evident by gel electrophoresis. We amplified the putative promoter regions in human ExPEC strain UTI89 derivatives expressing arabinose-inducible constructs under control of the PBAD promoter (Guzman et al., 1995) to ensure that both FimB and FimX were present at high levels, given previous data suggesting that FimX is poorly expressed under laboratory conditions (Hannan et al., 2008). We predicted that phase variation in any of the intergenic regions due to FimX or FimB activity would yield a unique four band pattern, compared with a double band pattern in the UTI89/pBAD33 control background (Fig. 1A).

Figure 1.

FimX phase variation of the hyxR promoter region. A. Comparison of FimX and FimB phase variation of T1P (fimS) and hyxR promoter regions. hyxR phase variation is shown by restriction length dimorphism, corresponding to a unique asymmetric NspI cut site in the promoter after PCR amplification. Lanes 1, 2 and 3 indicate UTI89/pBAD33, UTI89/pBAD:fimB and UTI89/pBAD:fimX respectively. B. FimB and FimX inversion of the type 1 and hyxR promoters in wt and recombinase-null backgrounds using phase-specific primers and PCR analysis. Promoter orientation is labelled relative to transcriptional state as ‘OFF’ or ‘ON’ as previously determined for T1P and as shown in Fig. 2B for hyxR. Type 1 pili expression was measured by immunoblot against the major subunit, FimA. Lanes represent UTI89 or ΔfimBEX with the following plasmids: 1 = pBAD33, 2 = pBAD:fimB and 3 = pBAD:fimX. C. Overview of the genomic organization of the fimX locus PAI-XUTI89 and location of the PhyxR inverted repeats. Directionality of the open reading frames, representing forward or reverse strand orientation, is shown. The 16 bp inverted repeats (▸) for the hyxR 5′ UTR (PhyxR-IRproximal and PhyxR-IRdistal) are shown with the repeat sequence listed below. The NspI (closest to hyxR) and SspI (closest to fimX) restriction sites used to map the approximate region of inversion are indicated by vertical lines. D. Comparison of inverted repeats and flanking sequence upstream of hyxR and T1P. The region shown is known to be important for FimB and FimE binding to fimS (Dove and Dorman, 1996; Gally et al., 1996; Kulasekara and Blomfield, 1999; Holden et al., 2007). The conserved 5′ CA dinucleotides of the fimS repeats are shown in white text on a black background.

We determined that FimX was able to phase vary the promoter of hyxR, but not hyxA or hyxB, based on the restriction sites tested. Digestion of the hyxR 5′ UTR amplicon with the restriction enzyme NspI showed the expected bands at 466 bp and 117 bp when only vector control was present (Fig. 1A). Expression of FimX produced changes in the hyxR promoter region consistent with phase variation, creating two additional fragments of approximately 350 and 225 bp (Fig. 1A). In contrast, we found no evidence of FimX phase variation upstream of hyxA or hyxB (data not shown). Expression of FimB did not produce any change in restriction patterns for any of the regions tested (Fig. 1A and data not shown).

To confirm that the expression of FimX produced phase inversion of the hyxR 5′ UTR, we designed phase-specific primers that discriminate between both orientations of the promoter based on size. We compared the inversion patterns of the type 1 promoter and hyxR promoter produced by FimB and FimX expression in wt and recombinase-null (ΔfimBEX) strains. ΔfimBEX is unable to undergo T1P or hyxR phase variation in the absence of expression of FimB, FimE or FimX, creating a clean background to assay individual recombinase activities. The promoter orientation is labelled relative to the transcriptional state as ‘OFF’ or ‘ON’, as previously published for type 1 pili (Klemm, 1986; Olsen and Klemm, 1994; Gally et al., 1996) and as determined later in this study for hyxR through transcript analysis. As previously described, we have shown that both FimB and FimX can mediate phase inversion of the fimS promoter to the ON orientation in a wt background and also in a Fim recombinase-null background (Fig. 1B) (Klemm, 1986; McClain et al., 1991; Gally et al., 1996; Bryan et al., 2006; Xie et al., 2006; Hannan et al., 2008). Concordant with the T1P phase PCR results, both FimB and FimX expression in wt and recombinase-null backgrounds showed increased levels of the major T1P pilus protein, FimA, as demonstrated by Western blot (Fig. 1B). However, only FimX expression showed phase inversion of the hyxR promoter (Fig. 1B).

The hyxR promoter region is inverted at 16 bp repeats

We next sought to localize the sites of FimX-associated inversion at PhyxR. Canonical sites of inversion resembling those flanking the T1P promoter were not apparent. Based on the NspI and SspI restriction pattern with and without FimX expression (Fig. 1A and data not shown), we determined that the hyxR promoter invertible region was between 270 and 290 bp in length. The location of the inversion sites was mapped to within 40 bp: the proximal inversion site (PhyxR IRproximal) was located approximately 30–70 bp upstream of the putative translational start site of hyxR, while the distal inversion site (PhyxR IRdistal) was located approximately 290–330 bp upstream (Fig. 1C). To determine the exact sites of inversion upstream of hyxR, we amplified the proximal inversion region plus flanking sequence using phase-specific primer pairs from UTI89 expressing pBAD:fimX in trans and then sequenced through the region of inversion in both orientations (for detailed methods see Experimental procedures). We determined that FimX expression resulted in inversion of the hyxR promoter starting at 16 bp inverted repeats (Fig. 1C). These sequencing results concur with our previous results mapping the inversion sites by restriction digest size analysis.

The 16 bp inverted repeats are located 278 bp apart with the proximal repeat located 39 bp upstream of the putative translational start site for hyxR (Fig. 1C). The 16 bp inverted repeats in the 5′ UTR of hyxR have minimal similarity to the 9 bp inverted repeats centred 314 bp apart that are the sites of phase inversion for type 1 pili (Fig. 1D) (Abraham et al., 1985; 1986). Previous in vivo footprinting analysis has defined 13 bp half-sites adjacent to the type 1 pili inverted repeats that serve as FimB and FimE binding sites (Gally et al., 1996; Holden et al., 2007; McCusker et al., 2008). Comparison of the inverted repeats and half-site flanking sequence at the two loci shows that the sites of inversion, and analogous putative DNA binding sites upstream of hyxR, are dissimilar in length and sequence composition to the known binding sites upstream of T1P (Fig. 1D).

The dissimilarity between the hyxR sites of phase inversion and the sites in the type 1 pili promoter region raised questions about the actual sites of inversion for FimX-mediated phase variation of type 1 pili. We hypothesized that, like FimX homologues FimB and FimE, FimX-mediated inversion upstream of type 1 pili would occur at the same 9 bp inverted repeats, although this had not yet been experimentally demonstrated (Gally et al., 1996; Kulasekara and Blomfield, 1999; Holden et al., 2007). To determine inversion sites of FimX upstream of the type 1 pili operon, we used a similar approach as used to determine the sites of hyxR promoter inversion, performing PCR and sequencing the regions of inversion flanking the T1P promoter in strains segregated phase OFF or ON. We were able to confirm that FimX-mediated inversion of the fimS promoter occurs at the previously defined 9 bp inverted repeats (data not shown).

Expression of FimX produces bidirectional inversion of the hyxR promoter region and promotes HyxR expression

In UTI89 ΔfimBEX starting with the hyxR promoter in the OFF orientation, we noted that the expression of FimX in trans produced only an ∼ 50% ON population. This suggested that either a large proportion of the population was unresponsive to FimX-associated recombination or FimX had bidirectional activity. This led us to determine if FimX and FimB expression results in bidirectional phase variation of the hyxR promoter region. Accordingly, we derived strains with all four possible combinations of phase-locked states at T1P and hyxR (Fig. 2A, compare lane 1 for each strain). Using the recombinase null, phase-locked derivatives, we monitored phase inversion at both loci following FimX or FimB expression. FimX and FimB were expressed in trans under induction of the PBAD promoter to ensure tightly controlled expression of each. As previously observed, expression of FimX, but not FimB, produced inversion at the hyxR promoter from OFF to ON (Fig. 2A). Interestingly, FimX was also able to invert the hyxR promoter from ON to OFF in contrast to its unidirectional activity turning T1P ON (Fig. 2A). FimB, which produced bidirectional inversion at the T1P promoter, was unable to mediate inversion in either direction at hyxR (Fig. 2A). FimX-associated inversion of T1P was independent of HyxR, as FimX expressed in trans was capable of producing T1P inversion in an hyxR deletion background (data not shown). Our results suggest that FimX is a unique site-specific, tyrosine recombinase family member in that it appears to recognize two divergent targets and has differing biochemical functions at each locus.

Figure 2.

fimX expression results in bidirectional inversion of the hyxR promoter region independent of IHF and H-NS. A. fimX expression results in bidirectional inversion of hyxR as assayed by phase PCR. Lanes represent ΔfimBEX phase-locked derivatives with the following plasmids: 1 = pBAD33, 2 = pBAD:fimB and 3 = pBAD:fimX. Results denote representative experiment. B. hyxR transcript measured by qRT-PCR. Transcript levels are shown as relative fold change as determined using the ΔΔ C(t) method and were normalized to the expression of a control gene transcript (16S rRNA). P-values that reached significance, defined as reaching a P-value < 0.05, are indicated as follows: *P-value < 0.05; ***P-value ≤ 0.001. qRT-PCR results represent at least three independent experiments. C. Investigation of FimX and FimB phase inversion of T1P and hyxR promoter region was undertaken in four additional urinary tract isolates: J96, CFT073, ASB1298 and NU14. Results denote representative experiment. D. FimX and FimB phase inversion of the type 1 promoter and hyxR in either UTI89, ΔH-NS or ΔhimA (IHF−) deletion strains. Results denote representative experiment.

We have shown that FimX is specifically able to regulate hyxR through phase variation of its promoter region (Fig. 1A and B); therefore, our next goal was to ascertain whether FimX-associated inversion of the hyxR promoter affected the expression of hyxR. To determine hyxR transcript levels, qRT-PCR was performed with FimX, FimB or vector control expressed in trans in the triple recombinase-null strain UTI89 ΔfimBEX. Expression of FimX in trans showed a twofold increase in hyxR transcript levels compared with vector control (P-value = 0.0345, one-sample t-test) (Fig. 2B). Transcript levels of hyxR were also compared between wt and isogenic fimX or fimB deletion strains. Compared with the wt strain, UTI89 ΔfimX had a 3.5-fold decrease in hyxR transcript (P-value = 0.0006, one-sample t-test). A single fimB deletion showed a non-significant 1.6-fold reduction in hyxR transcript levels compared with wt (P-value = 0.0806, one-sample t-test) (Fig. 2B). To determine hyxR transcript levels in the hyxR phase-locked OFF or ON strains, we again performed qRT-PCR. The hyxR phase-lock ON strain showed a 14.6-fold increase in hyxR transcript levels compared with the phase-lock OFF baseline (P-value < 0.0001, one-sample t-test) (Fig. 2B). Taken together, FimX, but not FimB, is able to mediate the phase inversion of the hyxR promoter region, thereby promoting hyxR transcription. Accordingly, the respective hyxR promoter orientations were designated as OFF or ON.

To determine if FimX induced phase inversion of the hyxR 5′ UTR was conserved in unrelated ExPEC carrying the PAI-X locus, we expressed the UTI89 FimX in trans under arabinose induction in four epidemiologically unrelated clinical isolates (see Experimental procedures for a description of clinical isolates) and assayed hyxR phase variation by PCR. FimX, expressed in trans, was able to mediate phase inversion of the hyxR promoter in all four clinical isolates tested (Fig. 2C). FimB was unable to mediate phase inversion of the hyxR promoter in three of the four isolates. However, in pyelonephritis isolate J96, the expression of FimB produced a minor ON population at the hyxR intergenic region, but not to nearly the same extent as FimX expression (Fig. 2C). Together, these data suggest that FimX is the principal epigenetic regulator of hyxR among a diverse group of UTI strains.

Inversion of the hyxR promoter independent of the accessory DNA-bending factor IHF and the global regulator H-NS

The Fim-like recombinases are part of the larger tyrosine recombinase family in prokaryotes for which IHF has been regarded as an essential cofactor for the inversion reaction. Previous work has established the essential role of IHF, a DNA-bending accessory factor that binds to the invertible T1P promoter, in providing scaffolding for the correct conformation of the FimB and FimE recombinase-mediated inversion reaction (Blomfield et al., 1997). FimB- and FimE-mediated recombination of the T1P promoter does not occur in the absence of IHF and the conformational change induced by IHF binding to the fimS promoter region (Dorman and Higgins, 1987; Eisenstein et al., 1987). We sought to investigate whether the action of IHF was required for FimX-mediated inversion of the T1P and hyxR promoter regions. As anticipated, IHF was required for FimX inversion of the T1P promoter (Fig. 2D). In accordance with previous work, we also observed that FimB-mediated inversion of T1P required IHF. In both cases, inversion of the T1P promoter was locked in strains carrying an isogenic deletion in himA, encoding the α subunit of IHF (Fig. 2D). Overexpression of FimB or FimX in trans was unable to overcome the absence of IHF, and no inversion at T1P was observed in these strains (Fig. 2D). In contrast, FimX inversion of the hyxR promoter region was independent of IHF. Baseline hyxR inversion was maintained in the himA deletion background, and overexpression of FimX was still able to robustly invert the hyxR promoter from OFF to ON, even in the absence of IHF (Fig. 2D).

Previous research has shown that the nucleoid-associated protein H-NS can influence the inversion of the T1P invertible promoter (Kawula and Orndorff, 1991; O'Gara and Dorman, 2000), primarily through repression of fimB and fimE (Olsen and Klemm, 1994; Olsen et al., 1998). However, follow-up studies revealed that although IHF plays the dominant role in biasing the T1P invertible switch towards the ON state, H-NS may also play a role in the orientation bias of the T1P promoter in an IHF-null background (Corcoran and Dorman, 2009). Recent work has demonstrated the role of H-NS in binding AT-rich regions of the chromosome, particularly within pathogenicity islands, and silencing their expression via transcriptional repression (Lucchini et al., 2006; Navarre et al., 2006; Fang and Rimsky, 2008). Given that both FimX and HyxR are encoded on a small pathogenicity islet (PAI-X) with low GC content and that IHF appears dispensable for FimX inversion at hyxR, we hypothesized that H-NS may play a role in regulating the promoter orientation of hyxR. However, FimX, expressed in trans, was capable of inverting both the T1P and hyxR promoters in the absence of H-NS (Fig. 2D). Interestingly, although H-NS and IHF were dispensable for FimX-associated recombination of the hyxR promoter, both IHF and H-NS do appear to promote an orientation bias towards OFF, as noted by the increased basal level of hyxR phase ON in either an hns or himA deletion background (Fig. 2D). This same effect was not noticed at the T1P promoter region. Even though both H-NS and IHF are not required for FimX-mediated inversion of hyxR, either one or both factors may play some role in the directional orientation of the promoter.

HyxR suppresses RNI tolerance in vitro

Our next goal was to determine the functional consequences of HyxR expression on how ExPEC responds to environmental stresses encountered during infection. For instance, UTI infection results in recruitment of nitric oxide (NO)-producing professional phagocytes to the bladder, which constitutes a key host defence mechanism (Hang et al., 1999; Haraoka et al., 1999; Horvath et al., 2011) and leads to a 3- to 50-fold increase in NO levels (Lundberg et al., 1996; Poljakovic et al., 2001). ExPEC may also encounter RNIs outside the urinary tract, possibly during passage through the gastrointestinal tract, where intestinal epithelial cells can secrete high levels of nitrate (NO3−) and nitrite (NO2−) (Kolios et al., 1995), or during invasive infections such as meningitis, where ExPEC interacts with macrophages (Mittal et al., 2010), key cellular RNI producers. NO is a precursor to a variety of RNIs, such as peroxynitrite, nitrous oxide and nitrosothiols, which can lead to extensive damage of nucleic acids, lipids and proteins. Compared with K-12 reference strains, most ExPEC isolates have increased resistance to acidified sodium nitrite (ASN; Bower et al., 2009) and other NO generators (Svensson et al., 2006).

To model RNI generation, we used ASN, which has been used widely to generate RNI and test resistance to its effects in vitro (Firmani and Riley, 2002; Bower et al., 2009). Sodium nitrite, when added to acidic media, is converted to nitrous acid, which spontaneously forms NO and other RNI. We investigated the tolerance of UTI89, an isogenic hyxR deletion and an HyxR constitutive expression strain to nitrosative stress in vitro. All strains tested grew equally well in pH-matched control media lacking RNI (Fig. 3A); however, there were differences in growth rate and tolerance when the strains were exposed to 3 mM ASN (Fig. 3B). The hyxR deletion strain had a growth advantage under high-RNI conditions that was reversed when HyxR was overexpressed (Fig. 3B). The consequence of constitutive HyxR expression was suppressed tolerance of RNI, whereas, conversely, mutation of hyxR resulted in a significant growth advantage to ExPEC under RNI stress modelled in vitro.

Figure 3.

HyxR negatively regulates RNI tolerance in vitro and its expression is responsive to nitrosative stress. A. Growth kinetics of UTI89, ΔhyxR and a constitutive HyxR-expressing strain in MES-LB (100 mM, pH 5.0). All strains grew equally well in pH-matched MES-LB. B. Growth kinetics of UTI89/pTrc99a, ΔhyxR/pTrc99a and a constitutive HyxR-expressing strain in MES-LB (100 mM, pH 5.0) + 3 mM ASN. Results represent at least three independent experiments. C. T1P and hyxR promoter orientation in UTI89 (top two panels) post ASN challenge. Bottom panel shows UTI89/pBAD:fimX after arabinose induction of FimX expression, increasing the starting hyxR phase ON population prior to challenge with ASN. Results denote a representative experiment. D. Transcript analysis of hyxR by qRT-PCR in each of the strains ± ASN challenge. Transcript levels are shown as relative fold change using the ΔΔ C(t) method and were normalized the expression of a control gene transcript (16S rRNA). P-values that reached significance are indicated as follows: *P-value < 0.05; ***P-value ≤ 0.001. Error bars represent standard deviation. Results are a composite of at least three independent experiments.

We hypothesized that if HyxR was acting, either directly or indirectly, as a repressor of RNI resistance, then hyxR expression might be regulated by nitrosative stress. To address this hypothesis, UTI89 was grown in the presence or absence of 3 mM ASN. Samples were taken at initial subculture, mid-log (OD600 = 0.5) and stationary (OD600 = 1.2) phase to assess the hyxR promoter orientation by phase PCR. If HyxR represses RNI resistance, then we predicted that, upon challenge with RNI, hyxR would be repressed; however, we did not know whether this would occur at the level of phase inversion or transcription. After growth in ASN, the invertible hyxR promoter region was predominantly in the OFF orientation (Fig 3C, middle panel). The invertible T1P promoter did not undergo a significant shift in orientation following RNI exposure (Fig. 3C, top panel). To increase the basal hyxR phase ON state from the start of the experiment, we grew UTI89/pBAD:fimX with arabinose (Fig. 3C, bottom panel). As we expected, the hyxR promoter showed a more dramatic shift from phase ON to phase OFF after exposure to ASN (Fig. 3C). Sampling at earlier time points (1, 4 and 6 h post ASN challenge; data not shown) indicated that the shift OFF of the hyxR promoter was not rapid, however, suggesting that the phase OFF state may have a fitness benefit, but was not likely being actively inverted. Therefore, to investigate whether hyxR repression was occurring at the level of transcription, we quantified hyxR transcript levels by qRT-PCR with or without a 1 h exposure to ASN. At 1 h we did not observe phase inversion of the hyxR promoter following ASN exposure, suggesting that any change in hyxR transcript levels by qRT-PCR were due to transcriptional or post-transcriptional regulation rather than phase inversion. Challenge of UTI89 with ASN resulted in a 2.1-fold reduction in hyxR transcript relative to the no ASN control (P-value = 0.0081, Student's t-test) (Fig. 3D), which was statistically indistinguishable from RNA collected from the negative ΔhyxR genetic control (P-value = 0.4636). As expected, expression of hyxR from Ptrc in trans resulted in an ∼ 40-fold upregulation of hyxR transcript levels compared with UTI89 wt under control conditions (P-value < 0.0004, Student's t-test). However, hyxR transcript levels did not vary with ASN challenge when driven by Ptrc, suggesting that regulation of hyxR by RNI occurs at the transcriptional level (Fig. 3D). Together, these data indicate that HyxR represses RNI tolerance in vitro and that nitrosative challenge results in suppression of hyxR transcript levels.

HyxR represses expression of Hmp, a key bacterial NO detoxification enzyme

We sought to determine the mechanism through which HyxR controls tolerance to RNI. We hypothesized that HyxR may regulate the expression of the NO detoxification flavohaemoglobin called Hmp. Hmp has been shown to play a key role in NO consumption in pathogenic E. coli strains, as well as other pathogenic species, including Salmonella enterica serovar Typhimurium (S. Typhimurium) (Stevanin et al., 2002; 2007; Svensson et al., 2010). Hmp expression has been shown to be responsive to nitrosative stress, and challenge of EXPEC isolate J96 with the RNI generator DETA/NO has been shown to result in increased hmp expression (Svensson et al., 2010).

As shown in Fig. 4A, hmp transcription under non-RNI conditions was similar between the wt and hyxR deletion strains. However, hmp transcript levels were repressed when measured under the same conditions with UTI89 constitutively expressing HyxR (2.0-fold decrease; P-value = 0.0154, one-sample t-test). After 1 h growth in ASN, the UTI89 wt and hyxR deletion strains had significantly increased levels of hmp transcript (14.0- and 15.6-fold respectively; P-value < 0.0001 for both strains compared with UTI89/pTrc99a without ASN, Student's t-test) (Fig. 4A). Constitutive expression of HyxR in ASN growth conditions resulted in significantly lower hmp transcript compared with either the wt or hyxR deletion strains (fivefold upregulation under ASN; P-value < 0.02 for both comparisons, Student's t-test) (Fig. 4A). These data indicate that HyxR is a repressor of hmp in vitro with and without RNI stress.

Figure 4.

HyxR represses transcript of a key bacterial NO detoxification enzyme, HmpA, correlating with decreased nitrite detoxification in vitro. A. hmpA transcript levels were measured by qRT-PCR in the indicated strains ± exposure to 3 mM ASN for 1 h. Transcript levels are shown as relative fold change using the ΔΔ C(t) method and were normalized the expression of a control gene transcript (16S rRNA). P-values that reached significance are indicated as follows: *P-value < 0.05; **P-value ≤ 0.01. Error bars represent standard deviation. Results are a composite of at least three independent experiments. B. Growth kinetics of UTI89/pTrc99a, Δhmp/pTrc99a, ΔhmpΔhyxR/pTrc99a and ΔhmpΔhyxR/pTrc:hyxR in MES-LB (100 mM, pH 5.0) + 3 mM ASN. C. Growth kinetics of UTI89/pBAD33, Δhmp/pBAD33 and Δhmp/pBAD:hmp in MES-LB (100 mM, pH 5.0) + 500 µM ASN. D. Growth kinetics of UTI89/pTrc99a/pBAD33, ΔhyxR/pTrc99a/pBAD33, ΔhyxR/pTrc:hyxR/pBAD33 and ΔhyxR/pTrc:hyxR/pBAD:hmp in MES-LB (100 mM, pH 5.0) + 3 mM ASN. E. Experimental design and growth curve for the RNI detoxification assay. Growth was monitored for the duration of the experiment at indicated time points. F. At T0 (addition of ASN) and subsequent indicated times, aliquots of the cultures were taken, centrifuged to remove bacteria and assayed for nitrite levels using Greiss Reagent. A sample of 500 µM ASN without bacteria (control) was included for reference. Results in (B), (C) and (D) represent at least three independent experiments.

We predicted that HyxR control over hmp transcription would have a major impact on the tolerance of EXPEC to RNI stress. The growth dynamics of UTI89 wt and isogenic derivatives with combinations of hmp and hyxR deletions and HyxR expression in trans were measured in media with and without RNI stress. All strains grew equally well in pH-matched MES-LB without ASN (Fig. S1A). In media containing ASN, strains carrying a deletion in hmp had a significant lag in growth (Fig. 4B). Constitutive expression of HyxR did not produce an Hmp-independent effect on RNI tolerance by UTI89, suggesting that it is through regulation of hmp that HyxR exerts its major effect on RNI tolerance (Fig. 4B). Expression of Hmp in trans under the inducible PBAD promoter produced partial complementation of the hmp single deletion (Fig. 4C). Expression of Hmp in trans in the constitutive HyxR expression strain was able to partially rescue the growth impairment of the HyxR overexpression strain in ASN (Fig. 4D). These data further demonstrate that Hmp has a major role in tolerance of RNI in vitro and the hmp is repressed by HyxR.

We next assessed if HyxR regulation of hmp altered the overall detoxification of RNI by ExPEC strain UTI89. Strains in early logarithmic growth were challenged with 500 µM ASN (Fig. 4E). At varying times following ASN challenge, nitrite concentrations were determined. In the absence of bacteria to consume NO, nitrite levels remained essentially constant at ∼ 500 µM (Fig. 4F). All of the strains carrying deletions in hmp, including those with or without HyxR expression, had significant defects in detoxification of RNI as measured as higher nitrite level relative to UTI89/pTrc99a (Fig. 4F). In the wt and hyxR deletion backgrounds where hmp was intact, detoxification was reciprocal to the levels of hyxR expression: The strain with constitutive HyxR expression had the highest nitrite levels while the non-complemented hyxR deletion strain had the lowest levels of nitrites (Fig. 4F).

HyxR suppresses intracellular survival in macrophages through RNI-dependent and -independent mechanisms

The apparent role of HyxR in regulating RNI resistance in vitro led us to investigate whether this system has a role during intracellular survival in macrophage-like cells. This question is particularly relevant given evidence that E. coli potentially traffics within macrophages during disseminated infection (Mittal et al., 2010). RAW 264.7 murine macrophage-like cells were pre-incubated with either l-arginine, an NO precursor, or l-NAME, an inducible NO synthase (iNOS)-specific inhibitor, to yield high and low NO physiological states respectively. Macrophage cells were then infected with UTI89 and isogenic derivatives to determine the contribution of HyxR in responding to RNI and intracellular stress in vivo as shown schematically in Fig. 5A.

Figure 5.

HyxR negatively regulates intracellular survival in macrophages through RNI-dependent and -independent mechanisms. A. Schematic diagram of the experimental design and treatment regimen. B. Bacterial titres at 24 h post infection. RAW 264.7 cells were pre-treated with 1 mM l-arginine, an NO precursor or l-NAME, an iNOS-specific inhibitor 1 h prior to infection and for the duration of the experiment. Unless indicated otherwise, statistical comparisons are between UTI89/pTrc99a and all other strains within each treatment arm. Statistical significance indicated as follows: ns = not significant; *P-value < 0.05; **P-value ≤ 0.01; ***P-value ≤ 0.001. C. Relative fitness of different strains under high NO conditions by assessing the ratio of the bacterial burden at 24 h for each strain between a high NO condition (l-arginine) and a low NO condition (l-NAME). D. Cell-free supernatant nitrite levels were measured from the supernatant of infected RAW 264.7 cells using Greiss Reagent (see Experimental procedures for detailed methods). Unless indicated otherwise, statistical comparisons are between UTI89/pTrc99a and all other strains within each treatment arm. Asterisk (***) represents a P-value ≤ 0.001. E. Nitrite levels were measured from the supernatant of RAW 264.7 cells infected with heat-killed (80°C for 15 min) or arrested (inclusion of 20 µg ml−1 chloramphenicol to inhibit protein synthesis) bacteria using Greiss Reagent. No comparisons reached statistical significance. Abbreviations: ND, not detected.

Paralleling our in vitro results, we found that HyxR played a significant role in repressing intracellular survival. An hyxR deletion strain showed an increase in intracellular bacterial burden compared with the wt strain (P-value = 0.0043, Student's t-test) while the HyxR overexpression strain showed a reduced capacity for intracellular survival (P-value ≤ 0.001 compared with UTI89/pTrc99a or ΔhyxR/pTrc99a; Student's t-test) (Fig. 5B). The intracellular survival defect was partially dependent on RNI generation as constitutive expression of HyxR still produced a significant survival defect even under low-RNI conditions (P-value = 0.0079 for UTI89/pTrc99a versus ΔhyxR/pTrc:hyxR under l-NAME treatment, Student's t-test) (Fig. 5C). Unexpectedly, all three UTI89 derivatives had a higher intracellular burden in macrophages in the presence of elevated RNI compared with the low-RNI state (Fig. 5B, compare left and right panels). Under conditions to stimulate NO production, the intracellular survival advantage of the non-complemented hyxR deletion was most evident (Fig. 5C).

Our previous results indicating that HyxR suppression of RNI tolerance in vitro is dependent on Hmp led us to investigate the contribution of Hmp to growth during macrophage infections in the context of HyxR deregulation. A single hmp deletion showed a small but significant reduction in intracellular survival under high-RNI conditions (P-value = 0.0187 compared with UTI89/pTrc99a, Student's t-test) that was completely restored under low-RNI conditions (P-value = 0.6498 compared with UTI89/pTrc99a, Student's t-test). Under high-RNI conditions, the ΔhyxRΔhmp double-deletion strain showed a significant increase in intracellular bacterial burden compared with the single hmp deletion strain (P-value = 0.0012, Student's t-test), while the HyxR overexpression strain still demonstrated a reduced capacity for intracellular survival in the hmp deletion background (P-value ≤ 0.0001 compared with Δhmp/pTrc99a, Student's t-test) (Fig. 5B). The intracellular survival defect due to constitutive HyxR expression was still evident even in the hmp deletion background even under low-RNI conditions (P-value = 0.0002 compared with Δhmp/pTrc99a, Student's t-test) (Fig. 5B, right panel). Despite lacking the major NO-detoxifying enzyme Hmp, Δhmp/pTrc99a still showed a relative fitness advantage under high-RNI conditions compared with l-NAME treatment conditions (4.53-fold) (Fig. 5C). Overexpression of HyxR in the context of the hmp deletion ablated the relative fitness benefit of high RNI (1.08-fold) (Fig. 5C).

We next quantified NO in the supernatant of infected macrophages. Supernatants of cells infected with ΔhyxR/pTrc99a contained a reduced amount of NO (measured as nitrite) compared with infection with UTI89/pTrc99a (Fig. 5D). Infection with ΔhyxR/pTrc:hyxR produced NO significantly greater than the non-complemented deletion or UTI89 wt and a level comparable to all of the strains carrying hmp deletions (Fig. 5D). These data suggest that the levels of NO in the macrophage supernatant reflect detoxification of NO by Hmp and the degree of repression of hmp by HyxR. Suppression of NO production by the low hyxR expression strains required metabolically active bacteria, as heat inactivation and inhibition of protein synthesis abrogated the differences between the HyxR high and low expression strains (Fig. 5E). Together, we have shown that HyxR is able to repress hmp in vitro both at baseline and upon exposure to nitrosative stress; however, HyxR regulation of hmp is unable to fully explain the observed HyxR-dependent intracellular survival defect under high-RNI conditions.

As another physiologically relevant condition with high NO induction, we looked at the ability of hyxR derivatives for intracellular survival in IFNγ-treated macrophages. Figure 6A shows a schematic of the treatment and sampling protocol. INFγ produced an ∼ 1 log decrease in the bacterial counts for each of the strains (Fig. 6B and C), to a level similar to that observed previously under l-NAME-treated conditions (Fig. 5B). Of all the strains, constitutive expression of HyxR produced the lowest intracellular burden in INFγ-treated and untreated cells (Fig. 6C). The hyxR deletion strain showed the largest difference between untreated and treated conditions (Fig. 6D). Uninfected, IFNγ-treated macrophages showed a high level of basal NO induction (∼ 78 µM; Fig. 6E). Upon infection, NO levels were suppressed to some extent by all of the strains relative to the IFNγ-treated, uninfected control (Fig. 6E). However, constitutive expression of HyxR continued to result in higher relative NO levels as seen in previous experiments (Fig. 6E). IFNγ was able to mediate some modest control of the bacterial burden, as relative fitness was higher in untreated cells (Fig. 6C). We still observed an HyxR-dependent defect in intracellular bacterial counts in IFNγ-treated cells, again suggesting HyxR may control other pathways contributing to intracellular survival and/or proliferation (Fig. 6C).

Figure 6.

ExPEC actively suppresses NO production by IFN-activated macrophages, and HyxR negatively regulates intracellular survival in IFNγ-activated macrophages through RNI-dependent and -independent mechanisms. A. Schematic diagram of the experimental design and treatment regimen. B and C. Bacterial titres from 24 h RAW 264.7 cell infections. Cells were not treated (B) or treated with 1 ng ml−1 IFNγ (C) 18 h prior to infection. Unless indicated otherwise, statistical comparisons are between UTI89/pTrc99a and all other strains. Statistical significance indicated as follows: ns = not significant; *P-value < 0.05; **P-value ≤ 0.01; ***P-value ≤ 0.001. Error bars indicate 1 standard deviation. D. Relative fitness of different strains in activated macrophages by assessing the ratio of the bacterial burden at 24 h for each strain between IFNγ-activated (C) and unactivated (B) macrophages. E. Nitrite levels were measured in the supernatant of infected RAW 264.7 cells using Greiss Reagent. Unless indicated otherwise, statistical comparisons are between UTI89/pTrc99a and all other strains within each treatment arm. Statistical significance indicated as follows: *P-value < 0.05; **P-value ≤ 0.01; ***P-value ≤ 0.001. Error bars indicate standard deviation. Abbreviations: ND, not detected.

Discussion

Pathogens like ExPEC have complex regulatory networks to co-ordinate the transition between commensal and pathogenic states and to maximize survival during their engagement with the host. The transition to the invasive phenotype must be tightly controlled since premature induction of virulence may have significant metabolic and immune costs. To keep tight regulatory control over these elements, a hierarchy of regulatory controls may be employed, including epigenetic, transcriptional and post-transcriptional regulation. Many pathogenic organisms have acquired unique genomic segments associated with novel regulatory inputs. Alternatively, many global regulators are part of the ‘core genome’, meaning they can be found in the genomes of related commensal organisms, but may have altered functionality in pathogenic isolates. In this study, we demonstrate that ExPEC has acquired the PAI-X locus that encodes FimX, a tyrosine recombinase-like invertase, and HyxR, a major repressor of RNI tolerance that is regulated through FimX-associated epigenetic control and by RNI stress itself. We have shown that HyxR represses transcription of hmp, which encodes a major RNI detoxification enzyme. We demonstrated that without appropriate relief of HyxR repression, ExPEC is severely attenuated under RNI stress in vitro and in vivo during intracellular persistence in macrophage-like cells.

Among prokaryotic epigenetic mechanisms, phase variation acts to couple environmental signals to rapid expression or repression of a specific factor, with one of the most well-studied examples being T1P. In the case of T1P, phase inversion provides a mechanism to rapidly transition between piliated and non-piliated states during the progression of EXPEC through different host niches. While T1P are critical for adherence to and invasion into the bladder epithelium during UTI (Mulvey et al., 1998; Anderson et al., 2003), the expression of T1P in the bloodstream can result in opsonization by phagocytes (Silverblatt and Ofek, 1983; Weinstein and Silverblatt, 1983) and rapid clearance (Xie et al., 2006). Phase variation through the Fim recombinases provides a rapid and complete switch needed for the transition.

The recombinases FimB and FimE, members of the Lambda Integrase superfamily, are preeminent examples of prokaryotic epigenetic regulation through phase variation. The Fim recombinase family members are the smallest members of the superfamily; yet, despite their small size, they still retain complex functions as site-specific invertases (Esposito and Scocca, 1997; Nunes-Duby et al., 1998). Like other Lambda Integrase family members, FimB and FimE require the DNA bending factor IHF for phase inversion of the T1P promoter (Dorman and Higgins, 1987; Eisenstein et al., 1987; Blomfield et al., 1997). However, phase variation of the hyxR promoter by FimX regulation does not require IHF, which deviates significantly from the canonical phase inversion precedence of other integrase superfamily members. Future studies will investigate whether other accessory factors, like LRP, have compensated for the roles of IHF and H-NS in phase inversion or whether there is something intrinsic about the hyxR promoter that alleviates the necessity for DNA-bending accessory factors. Unlike the T1P promoter region, PhyxR phase variation is also independent of H-NS.

The expression of FimX produces phase inversion at the T1P and hyxR promoters, suggesting that FimX is responsible for mediating the observed inversion at both loci and that FimX may co-ordinately regulate cross-talk between the loci through divergent DNA binding specificities. However, the sites of inversion between the two invertible regions are significantly different in length and sequence composition. Our data to this point are sufficient to draw a strong association that FimX mediates direct interaction with each promoter region. Alternatively, FimX may regulate, either transcriptionally or through phase inversion, another factor that may be responsible for inversion at the hyxR promoter. This scenario seems unlikely given that fimX expression is capable of producing inversion at T1P and hyxR in a ΔfimBEX background in which no other FimB-like homologues are found; however, at this time, we cannot rule out the possibility that FimX regulates another recombinase that is responsible for hyxR inversion. Additional ongoing studies should provide direct biochemical evidence of FimX binding and define direct interactions between FimX and the DNA to model specificity. FimX interaction with these very different invertible elements would constitute a significant functional divergence between FimX and other Lambda Int homologues and run contrary to the site-specific precedence of this family of proteins.

Through regulation of HyxR, FimX controls tolerance to RNI in ExPEC. A major mechanism through which the FimX-HyxR axis regulates RNI tolerance is by HyxR-mediated repression hmp transcription. Although Hmp is a key NO consumption enzyme under negative regulation by HyxR in vitro and in vivo, Hmp does not appear to play a significant role in intracellular survival in macrophages alone or during HyxR suppression of intracellular persistence. As evident in Fig. 5B, Hmp has only a small role in intracellular survival in macrophages under high-RNI conditions without strict correlation between detoxification of RNI and intracellular survival. Independent of Hmp, HyxR has substantial control over ExPEC intracellular survival. This may be of particular importance as prior studies have shown that macrophages are rapidly recruited into the urinary tract during acute infection and may be encountered by E. coli during the course of infection (Horvath et al., 2011). Future studies will be aimed at investigating the level of HyxR control of hmp expression, either direct or indirect, and investigating the effectors of HyxR-mediated suppression of RNI-independent intracellular survival.

HyxR is not a virulence determinant in the classical sense, as deletion does not cause attenuation. From the perspective of intracellular survival in macrophages, HyxR appears to suppress virulence. Other anti-virulence or virulence-modulating factors are known in other pathogenic organisms, such as the transcription factor Lrp in S. Typhimurium (Baek et al., 2009), the ACE2 transcription factor in Candida glabrata (Kamran et al., 2004) and the RcsC/YojN/RcsB phosphorelay system in S. Typhimurium (Cano et al., 2001; Mouslim et al., 2004). Deletion of the genes for these virulence moderators leads to increased proliferation, invasion and, sometimes, immune response compared with the wt isogenic parent. Lrp, for instance, is known to transcriptionally repress key virulence genes, indicating that proper regulation of Lrp expression in the host environment is central to Salmonella pathogenesis (Baek et al., 2009). Similarly, our data suggest that proper regulation of the presumptive HyxR regulon is important for intracellular survival in macrophages, as a constitutive HyxR-expressing strain shows RNI-dependent and -independent effects on intracellular survival.

The macrophage is increasingly becoming known as an important cellular niche for K1 E. coli establishment of distal invasive sites of infection, including the brain, spleen, liver and kidney in animal models infected intranasally (Mittal et al., 2010). It may then be beneficial to ExPEC to evolve multiple regulatory systems to tightly control the expression of RNI tolerance systems in response to certain environmental stimuli or host niches. It is also interesting that FimX appears to play a pivotal role in regulating both T1P and HyxR, suggesting co-ordinated regulation between a major adherence/invasion factor and the nitrosative stress response. Proper phase-variable expression of ExPEC T1P in the bloodstream is important for binding and invasion of human brain microvascular endothelial cells (Teng et al., 2005) and establishment of bacteraemia (Xie et al., 2006; Smith et al., 2010). Epigenetic control by a factor like FimX may co-ordinate pilus expression with the physiological alterations necessary to prepare a bacterium for host cellular entry and persistence. In host sites where ExPEC may encounter professional macrophages or high RNI concentrations, constitutive expression of hyxR has a survival cost to ExPEC.

In summary, we have shown that FimX, a regulator of the major virulence factor type 1 pili, is able to co-ordinately regulate an additional target through epigenetic phase variation. This previously unrecognized ExPEC-associated target, hyxR, can suppress the nitrosative stress response and contributes to intracellular survival in macrophage-like cells. Combined, these data indicate that (i) FimX is an important ExPEC-associated regulator that epigenetically regulates hyxR and (ii) the ability of ExPEC to survive RNI-mediated stresses within macrophages is contingent upon the proper regulation of HyxR, a negative regulator of RNI response pathways.

Experimental procedures

Bacterial strains and cultivation

All of the strains used in this study are listed in Table S1. The prototypic cystitis strain, UTI89, was obtained from an adult patient with cystitis and has been previously described and fully sequenced (Mulvey et al., 2001; Chen et al., 2006). Cystitis isolate NU14 (Hultgren et al., 1986), pyelonephritis isolate J96 (Hall, 1999), asymptomatic bacteriuria isolates ASB1298 (Garofalo et al., 2007) and pyelonephritis/bacteraemia isolate CFT073 (Mobley et al., 1990) have all been previously described and were all determined by multiplex PCR analysis to contain the pathogenicity islet PAI-X (S.L. Bateman and P.C. Seed, unpubl. data).

Bacteria were routinely cultured in Luria–Bertani (LB) broth (Genese Scientific) containing, where appropriate, kanamycin 50 µg ml−1, ampicillin 100 µg ml−1 and chloramphenicol 20 µg ml−1. For all in vitro studies with UTI89, other EXPEC strains and their derivatives, cultures were grown overnight (typically 16–18 h) shaking at 37°C then back-diluted 1:100 into fresh LB broth and grown shaking at 37°C, with the addition of 0.2% arabinose as necessary to induce expression of pBAD constructs in trans. IPTG was not included for induction of pTrc constructs, which were found to have sufficient expression in the absence of chemical induction. For in vitro nitrosative challenge, cultures were grown as just described, normalized to OD600 = 0.8, then subcultured 1:100 into MES acid-buffered LB (100 mM, pH 5.0) containing the indicated concentration of freshly added sodium nitrite. For all RAW 264.7 murine macrophage infections, UTI89 and derivatives were grown overnight shaking at 37°C, back-diluted 1:100 into fresh media and grown statically at 37°C for 18–24 h to induce expression of type 1 pili.

Construction of deletion mutants

Complete deletions for the coding regions of hyxR, H-NS and himA were made using the red recombinase method, as previously described, using pKD4 as a template and the primers as listed in Table S2 (Datsenko and Wanner, 2000; Murphy and Campellone, 2003). All primers were from Integrated DNA Technologies. The deletions were confirmed by PCR with flanking primers (Table S2). The kanamycin antibiotic insertion was removed through FLP-mediated excision by transforming the mutant strain with the temperature-sensitive plasmid pCP20 expressing the FLP recombinase (Cherepanov and Wackernagel, 1995). The resultant strains had no antibiotic resistance compared with the parental strain. The hmp deletion derivatives were created through P1 phage transduction of a disruption of hmp carrying the kanamycin cassette from the Keio collection as previously described (Baba et al., 2006). The hmp::kanR lesion was transduced into both UTI89 and UTI89ΔhyxR (kanS) backgrounds and screened with flanking primers.

Construction of fimB, hyxR and hmp expression plasmids

The fimB expression plasmid was created by digesting pLR92 (Robinson et al., 2006), a pBAD33 derivative, with SacI/BglII (NEB) to create a linearized vector. The pLR92 vector backbone was gel-purified from a 1% TBE agarose gel (APEX) using the Qiagen gel purification kit. The fimB insert was amplified from pTrc:HAT-fimB using Accuprime pfx high-fidelity polymerase [APEX] and primers HAT-B 5′ SacI and fimB-HAT-5 (Table S2). The fimB amplicon was digested with SacI/BamHI, gel-purified as described above and ligated into the similarly cut linearized pLR92 vector. Putative clones were confirmed by PCR and sequencing. For comparison, pBAD33 (vector control) was created by religating SacI/HindIII-digested pLR92 as previously described (Hannan et al., 2008).

To create the hyxR expression plasmid, pTrc99a was digested with SacI/HindIII to create a linearized vector as described above for pBAD:fimB. The hyxR insert was amplified from UTI89 with primers X HypoI 5′ and X HypoI 3′ (Table S2), digested with SacI/HindIII and ligated with the pTrc99a linearized vector. Putative clones were confirmed by PCR and sequencing.

To create the hmp expression plasmid pBAD:hmp, pLR92 was digested with SacI/HindIII to create a linearized vector for subsequent ligation with hmp amplified from UTI89 with primers listed in Table S2. The hmp amplicon was digested with SacI/HindIII, including a primer-encoded artificial ribosome binding site upstream of hmp, and ligated into the pLR92 linearized vector.

Type 1 pili phase orientation and expression

Type 1 pili expression was routinely checked by phase PCR to determine the orientation of fimS (the type 1 pili invertible promoter) as well as by immunoblot to detect protein expression. Phase PCR was assayed using one of two methods. The first method takes advantage of an asymmetric BstUI restriction site in the promoter of fimA and was performed using a modified method (Horcajada et al., 2005) as previously described (Hannan et al., 2008). Briefly, the fimA promoter was amplified with primers PHASE 1, PHASE 2 and Fim 14 (Table S2), which allowed amplification in both wt and ΔfimBE strain backgrounds, digested with BstUI and visualized by gel electrophoresis. Phase PCR assays were also performed by using a 1:1:1 mixture of three phase primers that included (i) an anchor primer (PHASE 2), (ii) a phase-specific primer for the OFF orientation (POFFfimS.1), and (iii) a phase-specific primer for the ON orientation (PONfimS.1) to increase sensitivity in detecting small phase changes and to more closely resemble the phase PCR assay for hyxR (Table S2). Mixed phase populations showed two bands by PCR, corresponding to the OFF and ON promoter orientations. Products were run on a 2% TBE agarose (APEX) gel. Western blot analysis was performed as previously described (Wright et al., 2005) for expression of type 1 pili using an antibody raised against the major subunit, FimA (Hannan et al., 2008). PCR and Western blot results represent at least three independent experiments.

Determination of location of DNA inversion upstream of hyxR

The orientation of the putative promoter regions for hyxR, hyxA or hyxB was determined using a PCR-based assay that exploited a restriction fragment length dimorphism due to possible orientation-dependent restriction sites. The 5′ UTR plus flanking sequence for each gene was amplified and digested with NspI, SspI or TspRI for hyxR; SspI, SphI or PstI for hyxA; and BsgI, SphI or BstUI for hyxB. Size determination of bands resulting from inversion of the promoter allowed approximate localization of the sites of inversion.

The precise sites of FimX-mediated inversion at the type 1 pili promoter as well as upstream of hyxR were determined by sequencing (Duke University DNA Analysis Facility) both strands of PCR products in both the phase ON and phase OFF orientations. Individual phase orientations were amplified using phase-specific primer pairs across the proximal inversion site from UTI89 expressing pBAD:fimX in trans to increase the proportion of the phase ON population (Table S2). The resulting sequences in both orientations were aligned using the clustalw multiple alignment algorithm (Thompson et al., 1994) in the BioEdit Sequence Alignment Editor (Hall, 1999). Sequence alignments were compared between each orientation to determine the maximal region of inversion, indicated by the region of DNA that did not show sequence alignment. Sequence alignment conservation was observed starting at the hyxR proximal 16 bp inverted repeat and extending outward into the flanking region, indicating that the 16 bp repeat was the site of DNA inversion. Similarly, FimX-mediated inversion upstream of the type 1 pili operon was conserved starting at the proximal 9 bp inverted repeat that had been previously defined for FimB and FimE inversion.

Phase-specific hyxR phase PCR

The promoter orientation of hyxR was determined using a phase-specific PCR approach similar to fimS phase PCR. Briefly, a 1:1:1 mixture of three phase primers (Table S2) was used to assay the promoter orientation. Mixed phase populations show two bands by PCR, corresponding to the OFF and ON promoter orientations. PCR reactions were visualized on a 2% TBE agarose gel.

qRT-PCR

RNA was collected from 5 ml of cultures grown in LB shaking for 3 h using the Masterpure system (Epicentre). For nitrosative stress conditions, duplicate cultures were challenged with 3 mM ASN for the last hour of growth, compared with pH-matched MES-LB as described above. One microgram of RNA was reverse transcribed using 1× iScript buffer (Bio-Rad) containing random hexamers and 1 µl of iScript (Bio-Rad). cDNA was diluted fourfold in DEPC-treated water, and 2 µl of each dilution was used per 25 µl of qPCR reaction. Amplicons were detected using the inclusion of 20× EvaGreen (Biotium) diluted to 1× (final) in a mixture of 1× APEX Taq polymerase buffer (Genesse Scientific), 1 U of APEX Taq polymerase, 2.5 mM MgCl2 (final) and 0.2 µM primers (IDT). Relative transcript abundance was assayed for hyxR and hmpA, which were normalized to 16S rRNA transcript levels. cDNA for 16S rRNA analysis (calibrator) was performed by one cycle of 95°C (10 min) then 40 cycles of 95°C (10 s), 55°C (30 s), 72°C (15 s), 80°C (5 s) and plate reading. hyxR and hmpA cDNA analysis was performed by one cycle of 95°C (10 min) then 40 cycles of 95°C (10 s), 57°C (30 s), 72°C (15 s), 80°C (5 s) and plate reading, with the exception that the annealing temperature for hyxR cDNA analysis was 55°C. Melting curve analysis and gel electrophoresis were used to ensure amplicon homogeneity. Relative fold change was derived by the ΔΔ C(t) method in comparison with UTI89 ΔfimBEX/pBAD33, UTI89, UTI89 ΔfimBEXT1P-OFF/hyxR-OFF or UTI89/pTrc99a, depending on the experiment (ABI User Bulletin #2). Reactions lacking reverse transcriptase (−RT) were performed to ensure the adequacy of DNase treatment. All qRT-PCR reactions were performed on the MJ Mini MiniOpticon Personal Thermal Cycler (Bio-Rad).

Macrophage infections

RAW 264.7 murine macrophages were routinely grown in DMEM (HG) + 10% FBS (Sigma) and subcultured 1:4 approximately every 3 days using a cell scraper to detach adherent cells for maintenance. For infections with E. coli, RAW 264.7 cells were seeded into 24-well trays at a density of 1 × 105 cells per well and allowed to adhere and grow to confluence for 36–42 h prior to infection. Where indicated, cells were pre-treated with 1 mM l-Arginine or the iNOS-specific inhibitor l-NAME (Chemicon) for 1 h prior to infection. l-Arginine and l-NAME were subsequently included in the media for the duration of the experiments. In some cases, cells were pre-treated with 1 ng ml−1 IFNγ for 16–18 h prior to infection to stimulate NO expression and the production of cytokines. IFNγ was not included in the media after the initiation of infections. Infections were done by adding 10 µl of bacterial suspension [1 × 107 cfu or multiplicity of infection (moi) of 10] in PBS to confluent cell monolayers (∼ 1 × 106 cells per well). Plates were centrifuged at low speed to bring bacteria in contact with the cell monolayer (5 min at 1000 r.p.m.) and then incubated for 1 h at 37°C with 5% CO2. Initial adherence/invasion (1 h) of UTI89 derivatives was used to normalize cfu per well at 24 h post infection; however, there were no significant differences in the 1 h adherence/invasion between the strains tested. Infected monolayers were washed three times with PBS and incubated for the remainder of the 24 h experiment using a step-down gentamicin protocol (100 µg ml−1 for 2 h then 50 µg ml−1 for 21 h). Finally, cells were washed three times with PBS, lysed in 1 ml of PBS + 0.1% Triton-X with vigorous pipetting, and bacterial cfu was enumerated by dilution plating on LB with ampicillin.

Determination of RNI concentration

To measure RNI, we used Griess Reagent (Biotium) and followed the manufacturer's protocol. Briefly, supernatants from infected macrophage monolayers were centrifuged at 13 000 r.p.m. for 10 min to pellet any contaminating eukaryotic or prokaryotic cells then filter sterilized using a 0.2 µm low-protein binding filter. Supernatant (50–150 µl) was mixed with 20 µl of Griess reagent and 130–230 µl of dIH2O. Samples were measured at 548 nm, and the values were converted to µM nitrite by using a standard curve generated with sodium nitrite.

Statistical analysis

All statistical determinations were made using GraphPad Prism (GraphPad Software). Student's t-test (unpaired) or one-sample t-tests were used to determine statistical differences in mean values as indicated in the text. qRT-PCR results were analysed using a one-sample t-test to compare whether the means were different from either 1 or −1. Two-tail tests were used in the determination of statistical significance, which was defined by attaining P-values ≤ 0.05. Statistical significance is indicated in figures with *P-value < 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001.

Acknowledgements

We would like to thank J. St Geme III and K. Kreuzer for their critical readings of this manuscript. We thank C. Goller and T. Prest for helpful insights and suggestions along the way. This work was supported fully or in part by the National Institutes of Health P50 DK64540 and DK07444301.

Ancillary