The AraC/XylS Protein MxiE and Its Coregulator IpgC Control a Negative Feedback Loop in the Transcriptional Cascade That Regulates Type III Secretion in Shigella flexneri

ABSTRACT

Members of the AraC family of transcriptional regulators (AFTRs) control the expression of many genes important to cellular processes, including virulence. In Shigella species, the type III secretion system (T3SS), a key determinant for host cell invasion, is regulated by the three-tiered VirF/VirB/MxiE transcriptional cascade. Both VirF and MxiE belong to the AFTRs and are characterized as positive transcriptional regulators. Here, we identify a novel regulatory activity for MxiE and its coregulator IpgC, which manifests as a negative feedback loop in the VirF/VirB/MxiE transcriptional cascade. Our findings show that MxiE and IpgC downregulate the virB promoter and, hence, VirB protein production, thus decreasing VirB-dependent promoter activity at ospD1, one of the nearly 50 VirB-dependent genes. At the virB promoter, regions required for negative MxiE- and IpgC-dependent regulation were mapped and found to be coincident with regions required for positive VirF-dependent regulation. In tandem, negative MxiE- and IpgC-dependent regulation of the virB promoter only occurred in the presence of VirF, suggesting that MxiE and IpgC can function to counter VirF activation of the virB promoter. Lastly, MxiE and IpgC do not downregulate another VirF-activated promoter, icsA, demonstrating that this negative feedback loop targets the virB promoter. Our study provides insight into a mechanism that may reprogram Shigella virulence gene expression following type III secretion and provides the impetus to examine if MxiE and IpgC homologs in other important bacterial pathogens, such as Burkholderia pseudomallei and Salmonella enterica serovars Typhimurium and Typhi, coordinate similar negative feedback loops.

IMPORTANCE The large AraC family of transcriptional regulators (AFTRs) control virulence gene expression in many bacterial pathogens. In Shigella species, the AraC/XylS protein MxiE and its coregulator IpgC positively regulate the expression of type III secretion system genes within the three-tiered VirF/VirB/MxiE transcriptional cascade. Our findings suggest a negative feedback loop in the VirF/VirB/MxiE cascade, in which MxiE and IpgC counter VirF-dependent activation of the virB promoter, thus making this the first characterization of negative MxiE- and IpgC-dependent regulation. Our study provides insight into a mechanism that likely reprograms Shigella virulence gene expression following type III secretion, which has implications for other important bacterial pathogens with functional homologs of MxiE and IpgC.

INTRODUCTION

Many Gram-negative bacterial pathogens, including Shigella species, use type III secretion systems (T3SS) to directly inject virulence proteins, known as effectors, into host cells to invade and/or subvert host cell machinery (1). Without a functional T3SS, these pathogens are often avirulent in animal infection models (2). Frequently, genes encoding the T3SS are transcriptionally activated by members of the AraC family of transcriptional regulators (AFTRs) (3, 4). Examples of AFTRs that regulate T3SS-encoding genes are VirF (5, 6) and MxiE (7, 8) in Shigella species, BsaN in Burkholderia pseudomallei (9), and InvF in Salmonella enterica serovars Typhi and Typhimurium (10). Of the well-characterized proteins in the large AraC family of transcriptional regulators (≥ 830 proteins [11]), most are described as activators that regulate the transcription of genes involved in carbon metabolism, stress response, or pathogenesis (12). A distinguishing feature of AFTRs is a conserved C-terminal DNA-binding domain comprised of two helix-turn-helix motifs (12, 13). Due to the high homology of this domain, the DNA-binding and/or regulatory activities of some AFTRs have been shown to be interchangeable when substituted for each other (14–16). In contrast, the N-terminal domains of AFTRs are functionally variable and involved in dimerization and/or ligand binding (4, 12, 17, 18). Due to the notorious insolubility of purified AFTRs in vitro, functional characterizations for the mechanism(s) of AFTR transcriptional regulation have proven difficult (19).

In Shigella flexneri, a causal agent of bacillary dysentery, a T3SS is used to inject two distinct waves of effectors to invade human colonic epithelial cells and adopt intracellular residency (20–24). Most genes encoding the T3SS (e.g., apparatus, effectors, and regulators) are clustered in the 31-kb ipa mxi spa operons (25, 26) on the large (~220 kb) virulence plasmid pINV (27–29). The pINV-associated virulence genes are regulated by the three-tiered VirF/VirB/MxiE transcriptional cascade, of which the first- and third-tier regulators VirF and MxiE belong to the AFTRs. Upon a switch to 37°C, virF expression is upregulated (30–33), whereby VirF transcriptionally activates virB (34–36). VirB then counters transcriptional silencing mediated by the chromosomally encoded histone-like nucleoid structuring protein H-NS (37–45), which engages AT-rich DNA sequences (46–48) at pINV-associated genes. While a temperature change to 37°C can be sufficient to relieve H-NS-mediated transcriptional silencing at some promoters, like PvirF (30–33, 49), the expression of approximately 50 T3SS-encoding genes requires VirB to counter silencing by H-NS before transcription can proceed (37, 39, 40, 42, 44). The large VirB regulon (50) includes genes encoding the T3SS (e.g., secretion apparatus and first wave of effectors), other virulence-associated factors (e.g., OspD1 [44], IcsP [40, 42, 43]), and the third-tier activator MxiE and its coactivator IpgC (51).

Prior to T3SS-dependent contact with the host cell, MxiE is sequestered by the antiactivator OspD1 and co-antiactivator/chaperone Spa15 (44, 52), whereas IpgC is independently sequestered by either the anti-coactivator IpaB or IpaC. Upon contact, the first wave of VirB-dependent effectors is secreted, which includes the antiactivator OspD1 and anti-coactivators IpaB and IpaC (7, 51, 53–55). In doing so, MxiE becomes free to associate with the chaperone IpgC, which lacks a DNA-binding domain (56), and together MxiE and IpgC transcriptionally activate genes encoding the second wave of T3SS effectors (7, 8, 50, 56–59; a figure summarizing these events can be found in reference 44). Due to the dormancy of MxiE- and IpgC-dependent transcriptional regulation prior to type III secretion, characterization of this complex regulatory system has proven challenging. In laboratory settings, active type III secretion can be induced to allow MxiE to associate with IpgC using chemicals (e.g., Congo red [8], the bile salt sodium deoxycholate [60]) or an S. flexneri mutant that lacks the T3SS apparatus tip proteins IpaB or IpaD (53). More recently, MxiE- and IpgC-dependent regulation has been more directly tested by tightly controlling the expression of mxiE and/or ipgC on plasmids in an S. flexneri strain cured of pINV (52). Through a combination of these strategies, the expression of over a dozen pINV-associated (i.e., ipaH7.8, ipaH9.8, ospB, ospC1, ospE1, ospF, and virA [8, 50, 58, 61]) and chromosomal (i.e., ipaHa, ipaHc, ipaHd, gem1, and gem3 [58, 61]) genes have been demonstrated to be MxiE and IpgC dependent. By aligning these promoter regions, a 17-bp putative cis-acting MxiE regulatory site, known as the MxiE box was identified, which has subsequently been shown to be required for MxiE- and IpgC-dependent transcriptional activation (8, 58, 59). However, as with many AFTRs, direct evidence for MxiE binding to a DNA recognition site remains elusive.

In this study, we identify and characterize a novel negative MxiE- and IpgC-dependent feedback loop in the three-tiered VirF/VirB/MxiE transcriptional cascade that regulates the expression of T3SS genes. We show that MxiE and IpgC negatively regulate the virB promoter, thus decreasing VirB protein production and VirB-dependent promoter activity at ospD1. This is the first description of negative MxiE- and IpgC-dependent regulation. We propose and test a model for negative MxiE- and IpgC-dependent regulation, whereby MxiE and IpgC interfere with VirF-dependent activation to negatively and specifically impact virB promoter activity. Our work has implications for other important bacterial pathogens with MxiE/IpgC homologs that control type III secretion systems, including BsaN/BicA in Burkholderia pseudomallei and InvF/SicA in Salmonella enterica serovars Typhi and Typhimurium (9, 10, 62).

RESULTS

MxiE- and IpgC-dependent regulation of the VirB-dependent ospD1 promoter is negative and likely indirect.

A prior investigation of the VirB-dependent ospD1 promoter (44) revealed a sequence similar to the MxiE box in both composition, 13/17-nucleotide (nt) match (Fig. 1A, bolded), and position, 29 nt between this site and the upstream flank of the −10 promoter element of ospD1 (Fig. 1A) (8, 58, 59). This putative site also contained 8 out of the 9 strictly conserved nucleotides (Fig. 1A, underlined) demonstrated to be required for MxiE- and IpgC-dependent transcriptional regulation (58). Since MxiE- and IpgC-dependent regulation of ipaHa only requires a site with a 14/17-nt match to the MxiE box consensus, we reasoned that MxiE and IpgC may also positively regulate ospD1. To test this, the putative MxiE box was mutated by site-directed mutagenesis in the context of our low-copy lacZ reporter for the ospD1 promoter, pPospD1-lacZ (44). Activity of the ospD1 promoter was measured in wild-type S. flexneri (2457T) and an isogenic virB mutant (AWY3) using β-galactosidase assays. To circumvent the low levels of free MxiE and IpgC (52, 63) associated with cells grown in broth where type III secretion is only weakly active (<5%) (53), l-arabinose inducible plasmid vectors carrying mxiE and ipgC (pBAD18-mxiE-ipgC) or no additional genes (pBAD18) were introduced into each cell background. These cells were then grown in the presence of 0.2% l-arabinose prior to promoter activity assays.

FIG 1.

MxiE- and IpgC-dependent regulation of the VirB-dependent ospD1 promoter is negative and indirect. (A) Comparison of the sequence (in bold) and location of the MxiE box consensus to the putative MxiE box identified at the ospD1 promoter (PospD1). Nucleotides strictly conserved within the MxiE box consensus (58) are underlined. Site directed mutagenesis was used to mutate the putative MxiE box at PospD1. (B) Activities of the ospD1 promoter were measured under inducing conditions (0.2% l-arabinose) in wild-type S. flexneri (2457T) and an isogenic virB mutant (AWY3) in the presence of exogenous MxiE and IpgC (pBAD18-mxiE-ipgC) or the empty control (pBAD18) using β-galactosidase assays. Representative data of three independent trials are shown. Data are represented as mean ± standard deviation. Significance calculated using two-way analysis of variance (ANOVA) with Šidák’s correction. ****, P < 0.0001.

Consistent with previous reports, ospD1 promoter activity was VirB dependent (18- to 19-fold change in ±virB conditions) (Fig. 1B) (44, 50). While MxiE and IpgC were expected to increase ospD1 promoter activity, this was not observed. Instead, ospD1 promoter activity decreased by 14- to 15-fold when mxiE and ipgC expression was induced compared to the pBAD18 empty control in wild-type S. flexneri (Fig. 1B). Furthermore, mutation of the putative MxiE box did not affect ospD1 promoter activity (Fig. 1B, left versus right), suggesting either indirect MxiE- and IpgC-dependent regulation of the ospD1 promoter or the involvement of an unidentified MxiE box. Since other putative MxiE boxes with strong matches to the consensus were not identified in the ospD1 promoter region contained within pPospD1-lacZ, we concluded that the VirB-dependent ospD1 promoter is negatively regulated by MxiE and IpgC and that this regulatory effect is likely to be indirect.

The virB promoter is negatively regulated in a MxiE- and IpgC-dependent manner.

From what is known regarding the VirF/VirB/MxiE transcriptional cascade, MxiE and IpgC most likely indirectly regulate the ospD1 promoter by negatively impacting the transcription of virB (Fig. 2, #1) or virF (Fig. 2, #2). Prior transcriptomic analyses (50, 61) have shown that virB mRNA levels decrease during constitutively active T3SS secretion (i.e., a condition favorable for MxiE- and IpgC-dependent regulation [52]) and in cells with mxiE when compared to an mxiE mutant. However, virF mRNA levels remained consistent in those cell backgrounds. Therefore, we hypothesized that virB is negatively regulated in a MxiE- and IpgC-dependent manner, thus decreasing VirB protein production and VirB-dependent promoter activity like that of ospD1.

FIG 2.

The virB promoter is negatively regulated in a MxiE- and IpgC-dependent manner. (A) Potential transcriptional inputs for negative MxiE- and IpgC-dependent regulation at either virB (#1) or virF (#2) that may explain MxiE and IpgC downregulation of the ospD1 promoter. (B) Differential MxiE- and IpgC-dependent regulation of the virB and ospF promoters. Promoter activities were measured using lacZ reporter plasmids, pPvirB(-1946)-lacZ and pPospF-lacZ, under inducing conditions (0.2% l-arabinose) in the S. flexneri mxiE mutant strain JAI04 (2457T mxiE2::aphA-3) and pINV-cured strain BS103 in the presence of exogenous MxiE (pBAD18-mxiE), MxiE and IpgC (pBAD18-mxiE-ipgC), or the empty control (pBAD18) using β-galactosidase assays. Representative data of three independent trials are shown. Data are represented as mean ± standard deviation. Significance is calculated using two-way ANOVA with Tukey’s correction. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

To address this hypothesis, we first determined if virB promoter activity is MxiE and IpgC dependent. To do this, the activity of the virB promoter was measured using β-galactosidase assays in an S. flexneri mxiE mutant strain JAI04 (2457T mxiE2::aphA-3) and the pINV-cured strain BS103 carrying pBAD18-mxiE, pBAD18-mxiE-ipgC, or the pBAD18 empty control under inducing conditions. As a control for canonical MxiE- and IpgC-dependent transcriptional regulation, the ospF promoter was used since it has been well-established to be positively regulated by MxiE and IpgC (Fig. 2B) (8, 50, 58, 59). As expected, ospF promoter activity significantly increased in the presence of exogenous MxiE or both MxiE and IpgC in the mxiE mutant cell background, exemplifying the previously described role of MxiE and IpgC as positive transcriptional regulators. In contrast, under identical assay conditions and in the same strain background, virB promoter activity significantly decreased when either mxiE or both mxiE and ipgC were induced compared to that of the empty pBAD18 control in the mxiE mutant cell background (Fig. 2B). Furthermore, in the BS103 cell background that lacks pINV, and hence mxiE and ipgC, we observed that ospF promoter activity only significantly increased when mxiE and ipgC were induced from pBAD18-mxiE-ipgC. Notably, basal virB promoter activity did not significantly change in the presence of mxiE or mxiE and ipgC in this strain background (Fig. 2B). This suggests that negative MxiE- and IpgC-dependent regulation of the virB promoter requires an additional pINV-associated factor. These data represent the first recorded observation of MxiE and IpgC either directly or indirectly negatively regulating transcription.

To address if MxiE and IpgC also decrease VirB protein levels, we measured VirB protein in wild-type S. flexneri (2457T), an isogenic virB mutant (AWY3), and an isogenic mxiE mutant strain JAI04 (2457T mxiE2::aphA-3) carrying either pBAD18 or pBAD-mxiE-ipgC using the same inducing conditions as our reporter assays. As expected, VirB protein was only detectable in the wild-type but not the virB mutant background (15-fold difference in ±VirB production based on average arbitrary densitometry units) (Fig. 3A and B). In the presence of exogenous MxiE and IpgC, VirB protein was also undetectable (15-fold decrease in ±MxiE and IpgC conditions in the mxiE mutant cell background) (Fig. 3A and B). These data demonstrate that VirB protein levels drop precipitously in the presence of MxiE and IpgC, which is consistent with our observation that MxiE and IpgC cause a decrease in virB promoter activity. Furthermore, since mxiE and ipgC mRNA expression is VirB dependent (50), our findings suggest a negative feedback loop in the VirF/VirB/MxiE transcriptional cascade that regulates T3SS-encoding genes.

FIG 3.

MxiE and IpgC decrease VirB protein levels. (A) Western analysis of VirB in the presence (pBAD18-mxiE-ipgC) and absence (pBAD18) of induced MxiE and IpgC in the S. flexneri mxiE mutant strain JAI04 (2457T mxiE2::aphA-3). The S. flexneri wild-type (2457T) and isogenic virB mutant (AWY3) are positive and negative controls, respectively. Blotting was probed using anti-VirB antibody (1:1,000). A representative blot is shown. (B) Densitometry analysis of Western analyses depicting the mean ± standard deviation of three independent trials (n = 3). Significance calculated using an unpaired two-tailed Student's t test that assumed either equal or unequal variance. **, P < 0.01.

The regions required for negative MxiE- and IpgC-dependent regulation and positive VirF-dependent regulation of the virB promoter are coincident.

To identify a potential mechanism for negative MxiE- and IpgC-dependent regulation at the virB promoter, the region upstream of virB was scanned for a putative MxiE box (58, 59). Four putative MxiE boxes were identified (site 1, 5′-AAATAGTAATTTTTaAG-3′; site 2, 5′-GATAAGCATTTTTTcAT-3′; site 3, 5′-CTGCCGATTCTCTTtCT-3′; and site 4, 5′-AGACTGATTTTTTAtCA-3′ centered at −52, −150, −299, and −873 relative to the +1 of virB, respectively). However, these putative MxiE boxes were either located far upstream of the −10, exhibited low matches to the consensus sequence (7- to 10-nt match), or both. The lack of a traditional MxiE box suggested that a different sequence and/or mechanism may be used for negative MxiE- and IpgC-dependent transcriptional regulation. As such, we mapped the region required for negative MxiE- and IpgC-dependent regulation of the virB promoter using 5′ promoter truncation analysis. The activities of the 5′ virB promoter truncations were measured using β-galactosidase assays in an S. flexneri mxiE mutant strain JAI04 (2457T mxiE2::aphA-3) carrying pBAD18-mxiE-ipgC or the pBAD18 empty control under inducing conditions. The resulting data (see Fig. S1 in the supplemental material) are represented as fold repression by MxiE and IpgC (Fig. 4B) and reveal that the downregulation of the virB promoter by MxiE and IpgC is greatest (4- to 6-fold repression) when sequences upstream of −402 relative to the virB +1 are present. MxiE- and IpgC-dependent repression gradually drops from 3.5- to 2-fold with the removal of sequences between −402 and −116 relative to the virB +1 despite the overall increase in virB promoter activity (see Fig. S1). Once the established region required for positive VirF-dependent regulation of the virB promoter is removed (−110 to −80 relative to the virB +1) (Fig. 4A) (36), the negative regulatory effect of MxiE and IpgC falls below 2-fold (Fig. 4B).

FIG 4.

The regions required for negative MxiE- and IpgC-dependent regulation and positive VirF-dependent regulation of the virB promoter are coincident. (A) Genetic locus of the virB promoter (−1946 to the virB +1). Drawn so panels align with corresponding 5′ virB promoter truncation in graphs below. (B) Regions required for negative MxiE- and IpgC-dependent regulation of the virB promoter were mapped using 5′ promoter truncations. Activities of the 5′ virB promoter truncations were measured under inducing conditions (0.2% l-arabinose) in the S. flexneri mxiE mutant strain JAI04 (2457T mxiE2::aphA-3) in the presence of exogenous MxiE and IpgC (pBAD18-mxiE-ipgC) or the empty control (pBAD18) using β-galactosidase assays. Data are represented as the average in fold repression ± standard deviation (pBAD18/pBAD-mxiE-ipgC) in virB promoter activity from three independent trials. (C) Regions required for positive VirF-dependent activation of the virB promoter were mapped using 5′ promoter truncations. Activities of the 5′ virB promoter truncations were measured under inducing conditions (0.2% l-arabinose) in the S. flexneri pINV-cured strain BS103 in the presence of exogenous VirF (pBAD18-virF) or the empty control (pBAD18) using β-galactosidase assays. Data are represented as the average in fold activation ± standard deviation (pBAD18-virF/pBAD18) in virB promoter activity from three independent trials. (D) Schematic of the coincident regions required for negative MxiE- and IpgC-dependent regulation and positive VirF-dependent regulation of the virB promoter.

While the region required for VirF-dependent regulation of the virB promoter has been mapped (36) (Fig. 4A), those studies focused on a relatively short promoter region of ~200 bp immediately upstream of the virB transcription start site (+1). Since our research has revealed that long-range regulatory effects modulate genes encoded by the Shigella virulence plasmid, pINV (40, 42, 44), we reexamined VirF-dependent regulation of the virB promoter but with longer upstream promoter fragments. To do this, we measured VirF-dependent regulation of the virB promoter using the 5′ promoter truncations in the S. flexneri pINV-cured strain (BS103) carrying pBAD18-virF or the pBAD18 empty control under inducing conditions. The resulting data (see Fig. S2 in the supplemental material) are represented as fold activation by VirF (Fig. 4C). As previously observed, VirF-dependent activation of the virB promoter is lost when DNA sequences upstream of −58 are removed, consistent with sequences between −110 and −80 being required for VirF-dependent regulation (36). Our results show that VirF-dependent regulation was significantly higher when sequences −1946 to −976 relative to the virB +1 were present, indicating that these upstream regions also contribute to the VirF-dependent regulation of the virB promoter. Importantly, these findings demonstrate that the regions required for VirF-dependent regulation and MxiE- and IpgC-dependent regulation are coincident despite the opposing regulatory effects (Fig. 4B to D; see also Fig. S1 and S2). These data suggest that the AFTR MxiE and its coregulator IpgC or a chromosomal MxiE- and IpgC-regulated factor interfere with the positive regulatory activity of VirF, another AFTR, from coincident regions upstream of the virB gene.

Negative MxiE- and IpgC-dependent regulation of the virB promoter functions to counter VirF-dependent activation of virB.

Since the regulatory regions for VirF and MxiE/IpgC at the virB promoter were superimposed, we next investigated if MxiE and IpgC negatively regulate virB by interfering with VirF-dependent activation of the virB promoter. To test this, MxiE- and IpgC-dependent regulation of the virB promoter was measured in an S. flexneri strain lacking pINV (BS103) but carrying either a virF (pBAD42-virF) or an empty expression plasmid (pBAD42). This strategy would allow these potentially complex and interconnected regulatory inputs to be assessed in a background without interference from other pINV-associated factors. As expected, virB promoter activity was positively regulated when VirF was present (pBAD18) (Fig. 5A). In contrast, VirF-dependent virB promoter activity significantly decreased by 8- to 9-fold in the presence of MxiE and IpgC (pBAD-mxiE-ipgC) compared to the empty control (pBAD18) (Fig. 5A). In the absence of virF, the activity of the virB promoter was not significantly altered regardless of the presence of mxiE and ipgC (Fig. 5A). These findings are consistent with data gathered in the pINV-cured BS103 cell background (Fig. 2B) since virF is encoded by pINV and consistent with the coincident pattern of VirF- and MxiE/IpgC-dependent regulation observed in the 5′ virB promoter truncation analysis (Fig. 4). Together, these data suggest that negative MxiE- and IpgC-dependent regulation of the virB promoter is mediated by these regulators interfering with VirF-dependent transcriptional activation of virB.

FIG 5.

Negative MxiE- and IpgC-dependent regulation is observed at the VirF-dependent virB and not icsA promoter. (A) Negative MxiE- and IpgC-dependent regulation of the virB promoter is solely observed in the presence of VirF. Activities of the virB promoter were measured from pPvirB(−1946)-lacZ under inducing conditions (0.2% l-arabinose) in an S. flexneri strain cured of pINV (BS103) carrying either the pBAD42 empty vector or pBAD42-virF in the presence of exogenous MxiE and IpgC (pBAD18-mxiE-ipgC) or the empty control (pBAD18) using β-galactosidase assays. (B) Activities of the icsA promoter were measured under inducing conditions (0.2% l-arabinose) in BS103 carrying either pBAD42 or pBAD42-virF in the presence of pBAD18-mxiE-ipgC or pBAD18 using β-galactosidase assays. Representative data of three independent trials are shown. Data are represented as mean ± standard deviation. Significance calculated using two-way ANOVA with Tukey’s correction. *, P < 0.05; **, P < 0.01.

MxiE and IpgC do not negatively impact VirF-dependent activation of the icsA promoter.

Lastly, we examined if MxiE and IpgC can counter VirF-dependent transcriptional activation of a different promoter. Since VirF has only been characterized to directly bind to and transcriptionally activate the pINV-associated virB (35, 36) and icsA (64–67) promoters, we tested if the icsA promoter was also negatively regulated by MxiE and IpgC. Activity of the well-characterized icsA promoter was measured in the presence (pBAD42-virF) and absence of virF (pBAD42) in an S. flexneri strain lacking pINV (BS103) that carries either pBAD18-mxiE-ipgC or the pBAD18 empty control under inducing conditions using β-galactosidase assays. Contrary to our results at the virB promoter, icsA promoter activity did not significantly decrease in the presence of MxiE and IpgC (pBAD-mxiE-ipgC) compared to the empty control (pBAD18) when in a BS103 cell background carrying pBAD42-virF (Fig. 5B). Rather, icsA promoter activity increased in an MxiE- and IpgC-dependent manner. Although the reason for this increase remains unclear, it may be caused by a MxiE box within the icsA-virA intergenic region that is needed for positive MxiE- and IpgC-dependent regulation of virA (58, 59), or it is possible that MxiE and IpgC can substitute for VirF to upregulate the icsA promoter. Regardless, these results demonstrate that MxiE and IpgC do not interfere with VirF-dependent activation of the icsA promoter, making it unlikely that MxiE and IpgC exert a dominant negative effect on the VirF protein or lead to the expression of a negative regulator of VirF (68). Instead, the negative feedback loop characterized by this work and orchestrated by MxiE and IpgC appears to downregulate the activity of the virB promoter and hence the transcriptional cascade controlling Shigella virulence.

DISCUSSION

Our study identifies a novel negative feedback loop in the VirF/VirB/MxiE transcriptional cascade that regulates the expression of T3SS genes in S. flexneri. We demonstrate that MxiE and IpgC, both VirB-dependent products, negatively regulate the virB promoter, leading to decreased VirB protein levels and the downregulation of ospD1, a representative of the large VirB regulon (>50 genes [50]) (Fig. 6). This is the first description of negative regulation by MxiE and IpgC, which have been well established to function as transcriptional activators of genes encoding the second wave of T3SS effectors during Shigella infection (7, 8, 58, 59). Additionally, our findings corroborate prior transcriptomics data that showed that both ospD1 and virB expression increased in the absence of mxiE compared to the wild type (50).

FIG 6.

Model of negative MxiE- and IpgC-dependent feedback loop in the VirF/VirB/MxiE transcriptional cascade that regulates T3SS-encoding genes. The VirB-dependent third-tier regulators MxiE and IpgC negatively feedback at the virB promoter by countering VirF-dependent activation of the virB promoter. As a result, VirB-dependent promoter activity like that of ospD1 is significantly decreased.

The negative feedback loop characterized in this work likely reprograms Shigella virulence gene expression at the stage following secretion of the first wave of T3SS effectors and invasion of host cells. This stage is favored since the secretion of the first wave of T3SS effectors triggers MxiE and IpgC to associate and positively regulate the transcription of second wave effector genes (7, 8, 52, 58, 59). We report that when MxiE and IpgC levels rise in the cell, these proteins negatively regulate virB and in turn downregulate VirB-dependent promoters, such as ospD1 (Fig. 1). Although not exhaustively tested in this study, the regulatory impacts of the MxiE- and IpgC-dependent negative feedback loop may be considerable. The VirB regulon consists of nearly 50 genes (50), including those that encode (i) the T3SS secretion apparatus, (ii) the first wave of effectors, (iii) other virulence-associated factors (e.g., OspD1 [44], IcsP [43]), and (iv) the third-tier regulators MxiE and IpgC. Therefore, it is tantalizing to consider that negative MxiE- and IpgC-dependent regulation of virB may “tap the brakes” on VirB-dependent gene expression, which could modulate or attenuate the expression of virulence genes in Shigella cells during an infection. Such an event could lead to an increase in Shigella persistence within host cells and/or the maintenance of Shigella virulence plasmid genes postinvasion (note that prolonged VirB-dependent gene expression is energetically costly and can lead to pINV instability or loss [69]).

In addition, this work provides evidence for a potentially novel mechanism of negative AFTR-dependent regulation. We demonstrate that MxiE and IpgC negatively regulate the virB promoter in a manner dependent upon the transcriptional activator of the virB promoter VirF, which like MxiE also belongs to the AFTRs. Furthermore, we show that the regions required for negative MxiE- and IpgC-dependent regulation are coincident with the regions required for positive VirF-dependent regulation of the virB promoter. These data suggest that MxiE and IpgC may interfere with VirF-dependent activation but specifically and only at the virB promoter. In support of this, our data show that the negative regulatory impact of MxiE and IpgC is solely observed at virB but not icsA (the only other VirF-activated locus on pINV [64–67]). Thus, MxiE and IpgC are unlikely to interfere with VirF protein levels by exerting a dominant negative effect on VirF or by upregulating the production of an AraC negative regulator that targets VirF (70). Instead, it is possible that MxiE and IpgC occlude VirF from the virB promoter to prevent VirF-dependent activation. A similar mechanism of AFTR regulatory interference has been described at the rhaSR operon in Escherichia coli (68). There, two AFTR members, RhaR and RhaS, recognize and bind to an overlapping region in the rhaSR promoter region. Interference can occur upon an increase in RhaS concentration that results in RhaS outcompeting RhaR for binding sites at the rhaRS promoter, thus preventing the synergistic activation of this locus with another transcriptional activator, CRP (68). While our work is consistent with a mechanism of AFTR interference being responsible for the MxiE- and IpgC-dependent negative feedback loop that controls the Shigella virulence gene cascade, in vitro experiments using all three purified proteins (MxiE, IpgC, and VirF) are required to gain direct evidence for this mechanism. Due to the notorious insolubility exhibited by AFTRs, this is currently beyond the scope of this study (7, 19, 57). Nevertheless, our finding that the regions required for AFTR-mediated positive and negative regulation of the virB promoter are coincident is significant and may guide others when similar regulatory controls are detected.

While the specific sequence(s) responsible for negative MxiE- and IpgC-dependent regulation of the virB promoter was not identified in this work, our analysis suggested that this sequence(s) will not be organized like the MxiE box consensus (58, 59). Upon reevaluation of the putative MxiE boxes identified at the virB promoter region (sites 1 to 4), it was found that sites 1, 2, and 4 (−52, −150, and −873 relative to the +1 of virB, respectively) contained the direct repeat 5′-AnTTTTTnA-3′ which may constitute a half-site. Since AraC binds half-sites to loop DNA and negatively regulates transcription (71–73), it is possible that MxiE and IpgC negatively regulate via a similar mechanism. The significance of these sequences has yet to be determined but will frame future work that provides mechanistic insight into these opposing regulatory controls.

Finally, our finding that MxiE and IpgC create a negative feedback loop in the transcriptional cascade that regulates T3SS in S. flexneri has implications for MxiE and IpgC homologs that regulate T3SS-encoding genes in other bacterial pathogens like Burkholderia pseudomallei (i.e., BsaN and BicA [9]) and Salmonella enterica serovars Typhi and Typhimurium (i.e., InvF and SicA [10]). Of note, the MxiE homolog BsaN in B. pseudomallei has also been suggested to positively and negatively influence transcription in a transcriptomic analysis (74). Therefore, other MxiE-like regulators, like BsaN or InvF, may also differentially regulate the expression of T3SS genes. We anticipate that other AFTR interference mechanisms may coordinate important negative feedback loops, like that described herein, to regulate virulence.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are listed in Table 1. S. flexneri strains were routinely grown at 25 or 37°C in Luria-Bertani (LB) broth (75) with aeration by constant shaking (325 rpm in a LabLine/Barnstead 4000 MaxQ shaker) or on Trypticase soy agar (TSA) (Trypticase soy broth containing 1.5% [wt/vol] agar). Where appropriate, antibiotics were used at the following final concentrations: ampicillin (100 μg mL⁻¹), chloramphenicol (25 μg mL⁻¹), kanamycin (50 μg mL⁻¹), or spectinomycin (50 μg mL⁻¹). To ensure that S. flexneri strains maintained pINV during manipulation, Congo red binding was tested prior to each assay on TSA plates containing 0.01% (wt/vol) Congo red (Sigma Chemical Co., St. Louis, MO) at 37°C.

TABLE 1.

Bacterial strains and plasmids used in this study

Bacterial strain or plasmid	Descriptiona	Source
Strain
2457T	Shigella flexneri serotype 2a	81
AWY3	2457T virB::Tn5; Knr	43
JAI04	BS611 (8); 2457T mxiE2::aphA-3; Knr	This work
BS103	2457T cured of the large virulence plasmid pINV	82
Plasmid
pBluescript KS(+) II	Multicopy cloning vector; Ampr	Stratagene
pBAD18	pBAD18 expression vector, pBR322 ori; Ampr	78
pBAD18-virF	pHJW4; pBAD18 expression vector carrying virF; Ampr	43
pBAD18-mxiE	pBAD18 expression vector carrying mxiE; Ampr	This work
pBAD18-mxiE-ipgC	pBAD18 expression vector carrying mxiE and ipgC; Ampr	This work
pBAD42	pBAD42 expression vector, pSC101 ori; Spcr	78
pBAD42-virF	pBAD42 expression vector carrying virF; Spcr	This work
pAFW04a	icsP promoter transcriptionally fused to lacZ in the low-copy pACYC184; Cmr. Digest with PstI/SalI or XbaI/SalI completely removes PicsP but retains lacZ and the lambda oop terminator used to prevent transcriptional read-through	39
pMCS-lacZ	pAFW04a-derived plasmid carrying the pBluescript KS(+) II multiple cloning site in place of the entire icsP promoter region; Cmr	This work
pSFUM131	IPTG inducible mxiE(1–251) and his6::ipgC in pNEB193; Ampr	This work
pPvirB(−1946)-lacZ	pAFW04a with −1946 to +54 relative to the virB +1; Cmr	This work
pPvirB(−1437)-lacZ	pAFW04a with −1437 to +54 relative to the virB +1; Cmr	This work
pPvirB(−976)-lacZ	pAFW04a with −976 to +54 relative to the virB +1; Cmr	This work
pPvirB(−402)-lacZ	pAFW04a with −402 to +54 relative to the virB +1; Cmr	This work
pPvirB(−350)-lacZ	pAFW04a with −350 to +54 relative to the virB +1; Cmr	This work
pPvirB(−250)-lacZ	pAFW04a with −250 to +54 relative to the virB +1; Cmr	This work
pPvirB(−200)-lacZ	pAFW04a with −200 to +54 relative to the virB +1; Cmr	This work
pPvirB(−116)-lacZ	pAFW04a with −116 to +54 relative to the virB +1; Cmr	This work
pPvirB(−58)-lacZ	pAFW04a with −58 to +54 relative to the virB +1; Cmr	This work
pPospD1-lacZ	pAFW04a with −1970 to +79 relative to the ospD1 +1; Cmr	44
pPospD1(mutMxiE)-lacZ	pPospD1-lacZ with the MxiE box 5′-CAATGGTTTTTTTAGAT-3′ mutated to 5′-CAATGCAAAACCATGGT-3′; Cmr	This work
pPospF-lacZ	pAFW04a with −2000 to +49 relative to the ospF ATG; Cmr	This work
pPicsA-lacZ	pMCS-lacZ with −440 to +442 relative to the icsA +1; Cmr	This work
pMAP07	pAFW04 lacking promoter sequences upstream of lacZ; Cmr	39

^aAmp^r, ampicillin resistance; Cm^r, chloramphenicol resistance; Kn^r, kanamycin resistance; Spc^r, Spectinomycin resistance.

Construction of mxiE mutant strain.

To construct a 2457T mxiE mutant strain, mxiE2::aphA-3 was transduced from BS611 (8) into 2457T by P1 phage transduction. The resulting strain, JAI04, was diagnosed by PCR for the mxiE2::aphA-3 allele using primers W672 and W677. The integrity of pINV and key virulence loci (icsA [W11/W12], virK [W13/W14], icsP [W15/W16], virF [W17/W18], and virB [W19/W20]) were also verified by PCR. While BS611 is a 2457T derivative, JAI04 was created for lab strain consistency.

Construction of the promoter-lacZ reporter plasmids.

To create pPospD1(mutMxiE)-lacZ, the putative MxiE box 5′-CAATGGTTTTTTTAGAT-3′ was mutated to 5′-CAATGCAAAACCATGGT-3′ using site-directed mutagenesis in the context of pPospD1-lacZ (44). Briefly, pPospD1-lacZ and gBlock 12 were digested with AflII and XbaI and then ligated to create pPospD1(mutMxiE)-lacZ. An NcoI restriction site engineered into the mutated MxiE box was used for diagnostic purposes.

To create the 5′ virB promoter truncations carrying −1946 to −58 relative to the virB +1, the upstream boundaries of virB were PCR amplified from the pINV of S. flexneri serotype 2a strain 2457T using the reverse primer W476 in combination with one of the following forward primers: W696, W746, W747, W557, W558, W559, W560, W561, or W493 (primers described in Table 2). Each PCR amplicon and the lacZ reporter plasmid, pAFW04 (39), were then digested with XbaI in combination with either PstI or SalI prior to ligation. All constructs were verified by DNA sequencing.

TABLE 2.

Oligonucleotide primers used in this study

Primera	Sequence (5′ to 3′)b	Purpose
W11	CTCCCCTGATTTTGTTAGGGATTTTC	Verify icsA locus
W12	GCCATCACAGGAAGCAGCCTC
W13	GGATATAGAAGAGCGGTTTG	Verify virK locus
W14	ACTTTATAATTTCAAGGGTACGGGTCCG
W15	GCACTTTGTGTACCTGCGATC	Verify icsP locus
W16	GCACTATTTTTAATGGTGCCAG
W17	CGAATCGCTGCAGGATATTATGATGCTGGAGTTTTGCGAAGC	Verify virF locus
W18	CGAATCGAATTCCCATCTGGCAATAGCGATGGGC
W19	CGAATCGAATTCTGAATTGGGCAGTTTACATCAGTG	Verify virB locus
W20	CGAATCGCTGCAGATTCTCTTTCTCTGATTGAAATGCTGG
W696	CGCAGAgtcgacCACAGTATTCGGAACTAATTATAAAAGATAAATTATCCC	Make pPvirB(−1946)-lacZ
W746	CGCAGAgtcgacGCCAATGAGAAAACATCCCAACC	Make pPvirB(−1437)-lacZ
W747	CGCAGAgtcgacCTCATCAGATACAACTAAAAGTAGCGC	Make pPvirB(−976)-lacZ
W557	CGCAGActgcagAATATGATAAAAGAAAAAATATTATCAATAGTGGC	Make pPvirB(−402)-lacZ
W558	CGCAGActgcagCGTACAGCAAACTATCTGAAGAA	Make pPvirB(−350)-lacZ
W559	CGCAGActgcagGATTCTCTTTCTCTGATTGAAATGC	Make pPvirB(−250)-lacZ
W560	CGCAGActgcagTCTACGTATAGATGAATCTACATTAGAAC	Make pPvirB(−200)-lacZ
W561	CGCAGActgcagTTATTTCTGTAGTCAAAAATAGTACAAAATCA	Make pPvirB(−116)-lacZ
W493	CGCAGActgcagATAGTAATTTTTAAGACTACCGTTGAC	Make pPvirB(−58)-lacZ
W476	TCTGCGtctagaATCACACCCTGTTTATTCATATTGAT	Reverse primer to make PvirB truncations
W690	CGCAGAgtcgacCCGCATCCCCTAAACGGT	Make pPospF-lacZ
W691	CGCTACtctagaCATTCAAAGAATCTAAATTTAGTTTTAGACAGGGCT
W672	CGCAGAgaattcCATGAACCAGTTAACGTTGAGC	Make pBAD18-mxiE and pBAD18-mxiE-ipgC
W677	TCGATAATTCTCTTGCAGAAAAGCC
W723	CGCAGAgaattcACCATCCTTATAGAGAAGGATGGG
W674	CGCAGAgtcgacGCCGGATCCTTAATTAAGTCTAGAAC	Make pBAD18-mxiE
W675	CGCAGAgtcgacCCCTAATAATTACTCCTTGATATCCTGAATTG	Make pBAD18-mxiE-ipgC

^aPrimers supplied by Integrated DNA Technologies.

^bEnzyme restriction sites PstI (in bold), XbaI (in italics), EcoRI (underlined), and SalI are in lowercase.

To create pPospF-lacZ, the region spanning −1974 to +70 relative to the predicted +1 of ospF (59) was PCR amplified from the pINV of S. flexneri serotype 2a strain 2457T using W690 and W691. The PCR amplicon was then inserted into pBluescript KS(+) II by digesting both the amplicon and holding vector with XbaI and SalI prior to ligation. The amplified ospF regulatory region was verified by DNA sequencing. Subsequently, the ospF regulatory region was removed from pBluescript KS(+) II using XbaI/SalI and inserted into a similarly digested pAFW04 (39) to create pPospF-lacZ, which was verified by diagnostic digest.

To create pMCS-lacZ, gBlock 14 (described in Table 3) carrying an engineered multiple cloning site was digested with PstI/XbaI. This insert was then ligated to pAFW04a (39), which had been digested with PstI and XbaI to entirely remove the icsP promoter region. The resulting construct, pMCS-lacZ, was verified by diagnostic digest.

TABLE 3.

gBlock gene fragments used in this study^a

Name	Sequence (5′ to 3′)
gBlock 12b	ACGCCCcttaagGACTGGCGTATGGCAATGAGCTGCTTTATTATCAAGTTTGGTGATCGCCTTGACGGTCACTTCTGAGTGAAGGCGCTTGCACAGAATGGTAAACGGACTCGATTATTGAATCTTAGGGATAAGATATATCTACGATATGTCAATATACGAAATCGAAAGAATTTATTGGTGAATATAGCCATTCAATGTGGCAATGCAAAACCATGGTTAGCAATAGACTAATATCTATTACATTACAAAAATAGGTTATTGTTTTTTTTTAGCAAAAAGGATGAACAAAAATGTCAATAAATAACTATGGATTACATCCAGCAAACAACAAAAATATGCTCTAGATAATTT
gBlock 14c	ATATActgcagGCGAATTGGAGCTCCACCGCGGTGGCGGCCGCACTAGTGGATCCCCCGGGGAATTCGATATCAAGCTTATCGATACCGTCGACCTCGAGGGGGGGCCCGGTACCCAGCTTTTGTTCCCTTTAGTGATCTAGAATATA

^agBlock gene fragments are trademarked by Integrated DNA Technologies.

^bThe mutated putative MxiE box at the ospD1 promoter is in gray shading. The restriction sites AflII (in lowercase) and XbaI (in bold) are indicated.

^cThe multicloning site from pBluescript KS(+) II is in gray shading. The restriction sites PstI (in lowercase) and XbaI (in bold) are indicated.

To create pPicsA-lacZ, the region spanning −440 to +1610 relative to the +1 of icsA (65, 67) was PCR amplified from the pINV of S. flexneri 2a strain 2457T using W692 and W693. The PCR amplicon was then inserted into pBluescript KS(+) II by digesting both the amplicon and holding vector with XbaI and SalI prior to ligation. The entire inserted icsA regulatory region was verified by DNA sequencing. Then, the −440 to +442 region (includes all VirF and H-NS binding regions identified in reference 67) relative to the icsA +1 was removed from pBluescript KS(+) II using XbaI and inserted into pMCS-lacZ linearized with XbaI to create pPicsA-lacZ, which was verified by diagnostic digest.

Construction of protein expression plasmids.

To create pBAD18-mxiE and pBAD18-mxiE-ipgC, mxiE and ipgC were PCR amplified from pSFUM131, which expresses the in frame and contiguous form of mxiEab(1–251) without transcriptional slippage and his₆::ipgC from the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible lac promoter in pNEB193 (gift from Maria Carolina Pilonieta and George P. Munson [76, 77]). Construction and verification of the pSFUM131 counterpart pSFUM139 (malE::mxiE[1–251] his₆::ipgC) is described in reference 57. To create pBAD18-mxiE, mxiEab(1–251) was amplified from pSFUM131 using W672 and W674. The PCR amplicon and pBAD18 (78) were then digested with EcoRI/SalI and ligated. The resulting construct was digested with EcoRI/BglII and ligated to a similarly digested PCR fragment amplified from pSFUM131 using W723 and W677. To create pBAD18-mxiE-ipgC, W672 and W675 were used to amplify mxiEab and his₆::ipgC from pSFUM131. Then, the PCR amplicon and pBAD18 were digested with EcoRI/SalI and ligated. The resulting construct was digested with EcoRI/BglII and ligated to a similarly digested PCR fragment amplified from pSFUM131 using W723 and W677. These cloning strategies capture the alternative GTG translation start site required for mxiEab(1–251) (79).

To create pBAD42-virF, the virF gene was digested from pHJW4 (43) using EcoRI/SalI and ligated to a similarly digested pBAD42 (78). The resulting construct was verified by digest with EcoRV.

All expression plasmids were verified by DNA sequencing and checked for regulatory activity and/or protein production by Western blotting.

Quantification of promoter activity.

Promoter activity was determined in a variety of strain backgrounds by measuring β-galactosidase activity (protocol adapted from reference 80 and described in 43) from lacZ plasmid reporters (Table 1). Where indicated, lacZ reporter plasmids and pBAD18/42 derivatives were introduced into the different strain backgrounds (Table 1) by electroporation. All cultures were grown overnight (16 h) at 25°C in LB broth (75) containing the respective antibiotics (Table 1) with aeration by constant shaking at 325 rpm. When using pBAD18, 0.2% d-glucose was added to the overnight LB broth. Prior to the β-galactosidase assay, overnight cultures were diluted 1:100 and grown for 5 h at 37°C with constant shaking in LB broth with respective antibiotics prior to cell lysis. Where necessary, to induce expression from pBAD18 or pBAD42, 0.2% l-arabinose was added after 3 h of the 5-h growth period. Data represent β-galactosidase activities generated in triplicate.

Protein analysis by Western Blotting.

Overnight cultures were diluted in 1:100 in LB containing the respective antibiotics and grown for 5 h at 37°C with aeration by constant shaking at 325 rpm. To induce expression from pBAD18, 0.2% l-arabinose was added after 3 h of growth. Cells were normalized to cell density (optical density at 600 nm [OD₆₀₀]), washed with 0.2 M Tris buffer (pH 8.0), and resuspended in 10 mM Tris (pH 7.4) containing 4× SDS-PAGE buffer and β-mercaptoethanol. VirB was detected using an affinity purified anti-VirB polyclonal antibody (1:1,000; Pacific Immunology) and an anti-rabbit IgG-horseradish peroxidase (HRP) (1:2,000; GE; NA9340) secondary antibody. All blots were imaged on the Azure 500 (Azure Biosystems), and densitometry was completed using the AzureSpot analysis software.

Statistical analysis.

Statistical calculations were done using GraphPad Prism version 9.1.1 software. Statistical tests are indicated in the respective figure legends. Statistical significance is represented as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.