The Antiactivator of Type III Secretion, OspD1, Is Transcriptionally Regulated by VirB and H-NS from Remote Sequences in Shigella flexneri

Transcriptional silencing and antisilencing mechanisms regulate virulence gene expression in many important bacterial pathogens. In Shigella species, plasmid-borne virulence genes, such as those encoding the type III secretion system (T3SS), are silenced by the histone-like nucleoid structuring protein H-NS and antisilenced by VirB. Previous work at the plasmid-borne icsP locus revealed that VirB binds to a remotely located cis-acting regulatory site to relieve transcriptional silencing mediated by H-NS. Here, we characterize a second example of remote VirB antisilencing at ospD1, which encodes a T3SS antiactivator and effector. Our study highlights that remote transcriptional silencing and antisilencing occur more frequently in Shigella than previously thought, and it raises the possibility that long-range transcriptional regulation in bacteria is commonplace.

ABSTRACT

Shigella species, the causal agents of bacillary dysentery, use a type III secretion system (T3SS) to inject two waves of virulence proteins, known as effectors, into the colonic epithelium to subvert host cell machinery. Prior to host cell contact and secretion of the first wave of T3SS effectors, OspD1, an effector and antiactivator protein, prevents premature production of the second wave of effectors. Despite this important role, regulation of the ospD1 gene is not well understood. While ospD1 belongs to the large regulon of VirB, a transcriptional antisilencing protein that counters silencing mediated by the histone-like nucleoid structuring protein H-NS, it remains unclear if VirB directly or indirectly regulates ospD1. Additionally, it is not known if ospD1 is regulated by H-NS. Here, we identify the primary ospD1 transcription start site (+1) and show that the ospD1 promoter is remotely regulated by both VirB and H-NS. Our findings demonstrate that VirB regulation of ospD1 requires at least one of the two newly identified VirB regulatory sites, centered at −978 and −1270 relative to the ospD1 +1. Intriguingly, one of these sites lies on a 193-bp sequence found in three conserved locations on the large virulence plasmids of Shigella. The region required for H-NS-dependent silencing of ospD1 lies between −1120 and −820 relative to the ospD1 +1. Thus, our study provides further evidence that cis-acting regulatory sequences for transcriptional antisilencers and silencers, such as VirB and H-NS, can lie far upstream of the canonical bacterial promoter region (i.e., −250 to +1).

IMPORTANCE Transcriptional silencing and antisilencing mechanisms regulate virulence gene expression in many important bacterial pathogens. In Shigella species, plasmid-borne virulence genes, such as those encoding the type III secretion system (T3SS), are silenced by the histone-like nucleoid structuring protein H-NS and antisilenced by VirB. Previous work at the plasmid-borne icsP locus revealed that VirB binds to a remotely located cis-acting regulatory site to relieve transcriptional silencing mediated by H-NS. Here, we characterize a second example of remote VirB antisilencing at ospD1, which encodes a T3SS antiactivator and effector. Our study highlights that remote transcriptional silencing and antisilencing occur more frequently in Shigella than previously thought, and it raises the possibility that long-range transcriptional regulation in bacteria is commonplace.

INTRODUCTION

Shigella flexneri is a facultative, intracellular bacterial pathogen that is the etiological agent of shigellosis or bacillary dysentery. To invade the colonic epithelium and adopt intracellular residency, a type III secretion system (T3SS) is used to directly inject two distinct waves of virulence proteins, known as effectors, into the targeted host cell cytosol (reviewed in reference 1). Genes encoding the T3SS (e.g., apparatus, effectors, and regulators) are primarily clustered on the large (∼220 kb) virulence plasmid pINV (2–4) in a 31-kb region containing the ipa mxi spa operons, collectively known as the “entry region” (5, 6). While genes in the entry region are sufficient for the production of a functional T3SS apparatus (7), additional T3SS regulatory and effector genes outside the entry region (i.e., virF [8, 9], icsA [10], virK [11], etc.) are required for Shigella maintenance and survival in host cells (5, 12, 13). Together, many of the pINV-associated virulence genes are transcriptionally controlled by a three-tiered regulatory cascade (Fig. 1) (reviewed in reference 14) initiated at human body temperature (15).

FIG 1.

Model of type III secretion system regulatory cascade in Shigella. (A) At 37°C, VirF transcriptionally activates virB. The antisilencing protein VirB then counters H-NS-mediated transcriptional silencing of pINV genes, including those required for the T3SS apparatus and first wave of effectors. (B) The transcription of the second wave of T3SS effectors is prevented when the T3SS apparatus, which spans the inner and outer bacterial membrane, is not in contact with the host cell. Under these conditions, MxiE is sequestered by the VirB-regulated antiactivator OspD1 and coantiactivator/chaperone Spa15. The coactivator IpgC is independently sequestered by either the VirB-regulated anticoactivator IpaB or IpaC. (C) Upon host cell contact, the T3SS translocon pore is formed, and the first wave of effectors is secreted, including the antiactivator OspD1 and anticoactivators IpaB and IpaC. Thus, MxiE and IpgC are free to associate and transcriptionally activate the second wave of effector genes.

In response to 37°C, the first-tier regulator VirF is produced (16–19), which transcriptionally activates the virB gene that encodes the second-tier regulator VirB (Fig. 1A) (9, 20, 21). The transcriptional antisilencing protein VirB then counters silencing mediated by the chromosomally encoded histone-like nucleoid structuring protein H-NS (22–28), which engages A/T-rich DNA sequences (29–31) at pINV-associated genes. While the transition to 37°C is sufficient to relieve H-NS silencing at some genetic loci, such as virF (16–19, 32), at other loci, VirB is required to counter H-NS silencing (22, 24, 25, 27). The large VirB regulon includes genes encoding the T3SS (i.e., secretion apparatus and first wave of effectors), other virulence-associated factors (i.e., IcsP), and the third-tier activator MxiE and its coactivator IpgC (33).

Prior to T3SS-dependent contact with the host cell, the third-tier activator MxiE is sequestered by the VirB-dependent antiactivator OspD1 and coantiactivator/chaperone Spa15 (Fig. 1B) (34). The third-tier coactivator IpgC is sequestered by either of the two anticoactivators, IpaB or IpaC. Upon contact, the first wave of VirB-dependent effectors is secreted, which includes the antiactivator OspD1 and anticoactivators IpaB and IpaC (Fig. 1C) (35–39). In doing so, MxiE is liberated to associate with its coactivator IpgC and transcriptionally activate genes required for the second wave of T3SS effectors (Fig. 1C) (40). The coupling of T3SS secretion to partner switching, exemplified by MxiE and IpgC, is an important host contact-dependent response that allows for the temporal control of effector secretion and is found in other bacterial pathogens such as Pseudomonas aeruginosa, Salmonella enterica serovar Typhimurium, and Yersinia enterocolitica (reviewed in reference 41).

Despite the role of OspD1 as an important temporal regulator of the second wave of T3SS effectors, the transcriptional regulation of the monocistronic ospD1 gene, which lies outside the entry region, is not well understood. Indeed, results from previous studies on ospD1 regulation are conflicting. Initially, the ospD1 promoter was observed to not be responsive to temperature (i.e., 30°C and 37°C) or active T3SS secretion conditions (40). However, subsequent macroarray analysis showed that ospD1 mRNA levels increased at 37°C but only in the presence of virB (33). Since our previous work on icsP showed that VirB is capable of remotely regulating from a cis-acting VirB-binding site located over 1 kb upstream of the icsP promoter (25, 27), we reasoned that the VirB regulatory elements that control ospD1 may not have been captured in the initially analyzed 500-bp ospD1 promoter region (40). To address this, we examined an extended region upstream of ospD1 to identify key regulatory elements leading to the transcriptional control of ospD1.

Here, we identify and characterize the DNA sequences required for the transcriptional regulation of ospD1. We characterize ospD1 promoter elements and confirm that the primary ospD1 transcription start site is VirB regulated. We identify two putative VirB-binding sites located over 1 kb upstream of ospD1 and show that either one is sufficient for VirB-dependent regulation of ospD1. We also show that ospD1 is transcriptionally silenced by H-NS and that VirB primarily functions to overcome this silencing rather than to act as a transcriptional activator. Since the region required for H-NS-mediated silencing of the ospD1 promoter is also remotely located, this study is the second to report remote regulation by both VirB and H-NS on pINV, the first being at the icsP promoter (25). Taken together, our findings suggest that regulatory sequences for transcriptional antisilencers and silencers, such as VirB and H-NS, can be located far upstream of the canonical bacterial promoter region (42, 43). The implication of these findings will be discussed.

RESULTS

Identification of the VirB-dependent transcription start site for ospD1.

First, to identify the ospD1 promoter(s) and to characterize its regulation by VirB, primer extension analysis was used. Total cellular RNA was extracted from cells of wild-type S. flexneri 2a strain 2457T and an isogenic virB mutant strain AWY3 that either did or did not (indicated by + or −, respectively) (Fig. 2A) carry the low-copy-number lacZ reporter plasmid for the ospD1 promoter, pPospD1-lacZ (described in Table 1). The primer extension products generated in wild-type S. flexneri were similar regardless of whether pPospD1-lacZ was present or not (Fig. 2A). This suggests that the transcripts produced from the native ospD1 locus are the same as those produced from pPospD1-lacZ. The most abundant ospD1 extension product mapped to a guanine (indicated by the +1) (Fig. 2A) 30 nucleotides upstream of the ospD1 translation start site (ATG) and aligned well with the +1, −10, and −35 promoter elements (Fig. 2B) predicted by the Softberry software BPROM (44). While additional longer ospD1 extension products (Fig. 2B, in bold) were detected (Fig. 2A), they were less abundant. No extension products were detected in the virB mutant background, thus allowing us to conclude that ospD1 transcripts are produced in a VirB-dependent manner. In sum, these primer extension analyses identify the primary transcription start site of ospD1 and reveal that the transcription of ospD1 is VirB dependent.

FIG 2.

Mapping of the VirB-dependent ospD1 transcription start site. (A) Primer extension analysis of ospD1 transcripts generated in the wild type (2457T) and an isogenic virB mutant (AWY3) either with or without (indicated by + or −, respectively) pPospD1-lacZ. Extension products were sized using the Sanger sequencing ladder (GATC). The primary ospD1 transcription start site (+1) is indicated by an arrow. (B) The DNA sequence of the ospD1 promoter region. The primary ospD1 +1 site along with the associated putative regulatory elements (−10, −35, and Shine-Dalgarno [SD]) are underlined. Nucleotides in bold represent transcription start sites identified by primer extension analysis (A). The ospD1 open reading frame is highlighted in gray. Primers W600 (binds ospD1 on pINV) and W601 (not depicted; binds pPospD1-lacZ) have identical lengths and anneal to the same position relative to the ospD1 translation start site.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Descriptiona	Source or reference
Strains
2457T	S. flexneri serotype 2a	81
AWY3	2457T virB::Tn5 Knr	28
MC4100	E. coli strain K-12 with araD and lacZ deletion	82
MC4100 hns::Knr	MC4100 where the first 37 amino acids of H-NS are expressed, resulting in a dominant-negative effect on the other H-NS-like protein (i.e., StpA [26, 83, 84]); Knr	85
Plasmids
pBAD18	l-Arabinose-inducible pBAD vector, pBR322 ori; Ampr	86
pATM324	pBAD-virB; Ampr	87
pBluescript KS(+) II	Cloning vector; Ampr	Stratagene
pAFW04a	icsP promoter region transcriptionally fused to lacZ in pACYC184; Cmr. Digest with SalI and XbaI removes the icsP promoter region entirely but retains lacZ and the lambda oop terminator used to prevent transcriptional readthrough.	24
pMAP07	pAFW04 lacking promoter sequences upstream of lacZ; Cmr	24
pPospD1-lacZ	pAFW04 carrying −1970 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−1220)-lacZ	pAFW04 carrying −1220 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−1120)-lacZ	pAFW04 carrying −1120 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−1070)-lacZ	pAFW04 carrying −1070 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−970)-lacZ	pAFW04 carrying −970 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−920)-lacZ	pAFW04 carrying −920 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−820)-lacZ	pAFW04 carrying −820 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(−520)-lacZ	pAFW04 carrying −520 to +79 relative to the +1 of ospD1 on a SalI and XbaI fragment; Cmr	This work
pPospD1(IR-1mut)-lacZ	pPospD1-lacZ with IR-1 (centered at −1270 relative to the ospD1 +1) mutated from 5′-ATTTCAGTATGAAAT-3′ to 5′-GCCCTGATGCAGGGC-3′; Cmr	This work
pPospD1(IR-2mut)-lacZ	pPospD1-lacZ with IR-2 (centered at −978 relative to the ospD1 +1) mutated from 5′-ATTTCAACGTGAAAA-3′ to 5′-GCCCTGGCACAGGGG-3′; Cmr	This work
pPospD1(IR-1/2mut)-lacZ	pPospD1-lacZ with both IR-1 and IR-2 mutated; Cmr	This work

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

Two remote VirB-binding sites contribute to VirB-dependent regulation of the ospD1 promoter.

Next, we searched for sequences required for VirB-dependent regulation of the ospD1 promoter. In previous work, we found that the sequences required for VirB-DNA binding in vivo and VirB-dependent regulation require a VirB-binding site consisting of a near-perfect inverted repeat with the sequence 5′-ATTT(C)C(A/T)(C/T)N(A/G)(A/T)G(G)AAAT-3′ (25, 27, 45). Since VirB-binding sites have been identified at a variety of positions upstream of +1 (22, 25, 46, 47), we scanned the 2-kb region upstream of the ospD1 ATG for putative sites. From this analysis, two putative VirB-binding sites were found centered at −1270 and −978 relative to the primary ospD1 +1 site (Fig. 3A) (each with either a 13/14 or 12/14 respective match to our previously proposed site [45]). We named the putative VirB-binding sites inverted repeat 1 (IR-1; −1270 site) and inverted repeat 2 (IR-2; −978 site).

FIG 3.

Two remotely located sites, IR-1 and IR-2, contribute to VirB-dependent regulation of the ospD1 promoter. (A) Schematic of 5′ ospD1 promoter truncations used to determine regions required for VirB-dependent activity. The putative VirB-binding sites, organized as inverted repeats, are depicted as divergent arrows. (B) Activity of ospD1 promoter truncations in wild-type S. flexneri (2457T) and an isogenic virB mutant (AWY3). (C) VirB-dependent regulation of the ospD1 promoter is lowered by the mutation of IR-1 and IR-2 (Table 1). Promoter activities were measured in wild-type S. flexneri (2457T) and the virB mutant (AWY3). For panels B and C, β-galactosidase activities were measured in triplicate in three independent trials. Representative data are shown; *, P < 0.05. The control was pMAP07, which carries a promoterless lacZ.

To assess the likelihood that IR-1 and IR-2 were involved in VirB-dependent regulation of the ospD1 promoter, 5′ truncations of the promoter region were generated in the context of pPospD1-lacZ (Fig. 3A). The activities of the resulting promoter constructs were then measured in wild-type S. flexneri (2457T) and an isogenic virB mutant (AWY3) using β-galactosidase assays (Fig. 3B). The resulting data show that VirB-dependent ospD1 promoter activity was significantly higher (∼40-fold change; P < 0.05) when the region containing both IR-1 and IR-2 was present. Removal of the region containing IR-1 (−1220 construct) (Fig. 3B) resulted in a modest decrease (P < 0.05) in VirB-dependent ospD1 promoter activity, from a 40- to 35-fold change, thus raising the possibility that only the downstream region containing IR-2 was required. Once the region with both IR-1 and IR-2 was truncated (−920 construct) (Fig. 3B), VirB-dependent ospD1 promoter activity was completely lost; however, baseline promoter activity increased, suggesting that the region required for H-NS-dependent silencing of the ospD1 promoter had been partially removed. These data suggest that the region between −1220 and −920 relative to the ospD1 +1 is required for VirB-dependent regulation of the ospD1 promoter.

To address this and any relative contributions of IR-1 and/or IR-2 for VirB-dependent regulation of the ospD1 promoter, site-directed mutagenesis was used to mutate IR-1, IR-2, or both in pPospD1-lacZ. Activities of the ospD1 promoter with and without the mutated IRs were then measured in wild-type S. flexneri (2457T) and an isogenic virB mutant (AWY3) by using β-galactosidase assays (Fig. 3C). Strikingly, the data show that the mutation of either IR-1 or IR-2 did not markedly decrease ospD1 promoter activity in the presence of virB, despite being statistically different (P < 0.05). However, when both IR-1 and IR-2 were mutated, a complete loss of VirB-dependent ospD1 promoter activity was observed (Fig. 3C). Thus, at least one of the remotely located sites, IR-1 or IR-2, is required for VirB-dependent regulation of the ospD1 promoter, explaining our observations as shown in Fig. 3B. To the best of our knowledge, this is the first example of two putative VirB-binding sites independently contributing to the VirB-dependent regulation of a promoter. Moreover, this provides two additional examples of remotely located VirB regulatory sites controlling a VirB-dependent promoter.

The ospD1 promoter is silenced by H-NS and countersilenced by VirB.

While VirB has been characterized only as relieving transcriptional silencing mediated by H-NS (23–25, 27) and its homologs (i.e., StpA and Sfh [26]), other transcriptional antisilencing proteins such as ToxT (48, 49) and, under specific conditions, PhoP (50) can function to upregulate transcription independent of H-NS (reviewed in references 26, 51 and 52). Thus, we next chose to examine if the VirB-regulated ospD1 promoter was silenced by H-NS and, if so, which regions were required for this activity. To do this, the activities of 5′ ospD1 promoter truncations were measured in wild-type Escherichia coli K-12 (MC4100) and an isogenic hns mutant carrying a dominant negative allele (MC4100 hns::kn; described fully in reference 26) by using β-galactosidase assays (Fig. 4A), as we have performed previously (24, 26–28). These E. coli strains are suitable proxies to examine H-NS-dependent regulation of an S. flexneri promoter, as the E. coli and S. flexneri chromosomes are 99.9% identical and complications arising from pINV instability or pleiotropic effects in Shigella hns mutants (26, 53, 54) are avoided. The resulting data show that at 37°C, ospD1 promoter activity was lower in the presence of hns than in its absence, demonstrating that the ospD1 promoter is indeed H-NS dependent. By truncating the upstream boundaries of the ospD1 promoter region, a gradual increase in promoter activity was observed between −1120 and −820 relative to the primary ospD1 +1 in the presence of hns, whereas in the absence of hns, similar ospD1 promoter activity was observed throughout the truncation series (Fig. 4A). These results show that H-NS-mediated silencing of the ospD1 promoter requires a long region (−1120 to −820 relative to the ospD1 +1) (Fig. 4A) rather than a short discrete site.

FIG 4.

H-NS-mediated silencing of the ospD1 promoter requires the region between −1120 and −820 relative to the ospD1 +1. (A) Activities of the ospD1 promoter truncation series were measured in wild-type E. coli (MC4100) and an isogenic hns mutant carrying a dominant negative allele (MC4100 hns::kn) by using β-galactosidase assays. (B) Relative contributions of IR-1 and IR-2 in VirB antisilencing of H-NS-mediated silencing at the ospD1 promoter. Promoter activity from ospD1 was measured under inducing conditions in wild-type E. coli (MC4100) and an isogenic hns mutant (MC4100 hns::kn) in the presence (pBAD-virB) or absence (pBAD-empty) of VirB with or without IR-1 and/or IR-2 mutated. For panels A and B, β-galactosidase activities were measured in triplicate in three independent trials. Representative data are shown; *, P < 0.05. The control was pMAP07, which carries a promoterless lacZ.

Next, we determined if VirB functions solely to antisilence H-NS-mediated silencing of the ospD1 promoter and, if so, the relative contributions of IR-1 and IR-2. To test this, pPospD1-lacZ and its derivatives with IR-1, IR-2, or both mutated were introduced into wild-type E. coli (MC4100) and an isogenic hns mutant (MC4100 hns::kn) carrying either the l-arabinose-inducible virB expression plasmid pBAD-virB or an empty plasmid control, pBAD-empty. Promoter activity was measured under inducing conditions (0.2% l-arabinose) by using β-galactosidase assays. In the wild-type background, an 11- to 13-fold change in ospD1 promoter activity was observed in the presence of virB (pBAD-virB) compared to that in its absence (pBAD-empty) but only if IR-1 and/or IR-2 was present (Fig. 4B). In contrast, in the hns mutant, the fold changes in VirB-dependent ospD1 promoter activity were similar regardless if IR-1 and/or IR-2 was present (1.6- to 1.9-fold change) (Fig. 4B) or absent (1.4-fold change) (Fig. 4B). In both backgrounds, no noticeable differences in the relative contributions of IR-1 and IR-2 to ospD1 promoter activity were observed (Fig. 4B). In sum, the major effect of VirB at the ospD1 promoter is to relieve transcriptional silencing mediated by H-NS. Remarkably, either IR-1 or IR-2 in the presence of VirB can equivalently achieve this effect regardless of the sites being 292 bp apart and thus positioned differently with respect to the +1 and the region required for H-NS-mediated silencing of the ospD1 promoter.

Comparison of sites required for VirB-dependent regulation at ospD1 and icsP reveals three near-identical 193-bp DNA sequences on pINV.

Comparison of IR-1 and IR-2 to the well-characterized VirB-binding sites found at the icsB (22, 47) and icsP (25, 27) promoters revealed that specific nucleotides in these sites are conserved (Table 2). Intriguingly, when comparing extended regions from the ospD1 IR-1 site and the icsP VirB-binding site, specifically, 200 bp from each flank, a nearly identical (341/348-bp match) (see Fig. S1 in the supplemental material) DNA sequence was found. Comparison of this sequence to pINV revealed yet another instance of this sequence nearly 2 kb downstream of virA (Fig. 5A); however, this time, the conserved sequence was much shorter. Thus, a 193-bp near-identical core sequence containing a VirB regulatory site (Fig. 5B, blue box) is found three times in pINV. Strikingly, the positions of these sequences are conserved relative to nearby genetic loci (i.e., ospD1, virA, and icsP) (Fig. 5A) in all analyzed Shigella species, including a related E. coli strain (Table 3). We therefore chose to name these sequences the Shigella VirB-binding site repeat region 1 (SVRR1) through SVRR3 (Fig. 5A).

TABLE 2.

Comparison of IR-1 and IR-2 to well-characterized VirB-binding sites

VirB-dependent gene	VirB-binding site (5′→3′)a	Location relative to +1
ospD1	GGGGATTTCAGTATGAAATGAAG	−1270
	TGGGATTTCAACGTGAAAACTTA	−978
icsP	GGGGATTTCAGTATGAAATGAAG	−1137
icsB	TGGGATTTCATGATGAAACGAGC	−76

^aConserved residues are in boldface font (ATTTCANNNTGAAA).

FIG 5.

Locations and sequences of the Shigella VirB-binding site repeat regions (SVRRs). (A) Locations of SVRR1 to -3 (in blue) on pINV (not to scale; NCBI accession no. AF386526.1). In order, the SVRR1 to -3 coordinates and orientations are 17516 to 17708 (upstream and in the same direction as ospD1), 145636 to 145828 (downstream and in the opposite direction of virA), and 219054 to 219246 (upstream of the divergent icsP and ospZ genes but only running in the same direction as icsP). The relative location of the ipa mxi-spa operons is also indicated in gray. (B) Nucleotide alignment of SVRR1 to -3 with conserved nucleotides indicated by an asterisk (data generated using Clustal Omega [80]). The 193-bp SVRR is underlined. The VirB regulatory sites from ospD1 (IR-1 only), icsP, and their identical counterparts are highlighted in blue.

TABLE 3.

Representative list of bacterial species with SVRRs

Strain	Location	No. of SVRRs	Query cover (%)	E value	Identity (%)	Accession no.a
S. flexneri 2a strain 301	pCP301	3	100	1e−94	100	AF386526.1
S. flexneri 5a strain 501	pWR501	3	100	1e−94	100	AF348706.1
S. boydii strain CDC3038-94	pBS512_211	3	100	1e−94	100	CP001062.1
S. sonnei strain Ss046	pSS_046	3	100	1e−94	100	CP000039.1
E. coli strain CFSAN029787	pCFSAN029787_01	3	100	1e−94	100	CP011417.1
S. dysenteriae strain 1617	pSLG231	3	100	2e−91	100	CP006737.1
S. flexneri 2a strain 2457T	Chromosome	1	87	5e−78	98.82	AE014073.1

^aGenBank accession number (National Center for Biotechnology Information [NCBI]).

Surprisingly, an additional SVRR, named SVRR4, was also identified on the S. flexneri 2a chromosome (NCBI accession no. AE014073.1; position 1460799 to 1460967 bp; 167/193-bp coverage; 98.8% identity) downstream of a tehAB operon, which is predicted to confer tellurite resistance (reviewed in reference 55). In contrast to the SVRRs on pINV, SVRR4 was not found in this location on other Shigella chromosomes (NCBI nucleotide database). Thus, it seems likely that SVRR4 is a remnant of replicative transposition between pINV and the chromosome, because transposon-mediated integration and excision of large pINV fragments on the Shigella chromosome are well established (3, 56–58) and all SVRRs are situated between insertion (IS) elements. The conservation of SVRR1 to -3 and their locations on pINV have implications for our understanding of the evolution of VirB regulatory elements and hence transcriptional antisilencing of Shigella virulence plasmid genes.

DISCUSSION

In this study, we characterized the transcriptional regulation of ospD1, which encodes a T3SS effector that also functions as a key temporal regulator of the second T3SS effector wave (Fig. 1) (34–38). We identified putative ospD1 regulatory elements (+1, −10, −35, SD) (Fig. 2) and showed that this promoter is transcriptionally regulated by VirB and H-NS (Fig. 3 and 4). Intriguingly, the sequences required for both VirB- and H-NS-dependent regulation were identified far upstream of the canonical bacterial promoter region (<250 bp [42, 43]). The VirB regulatory sites IR-1 and IR-2 are centered at −1270 and −978, respectively, relative to the primary ospD1 +1 (Fig. 6), making them much more remote than other distal cis-acting elements documented to control bacterial transcription (59–65). Since IR-1 and IR-2 are the second and third examples of remote and functional VirB regulatory sites, our work demonstrates that remote transcriptional antisilencing of virulence genes found on pINV is more common than previously thought (Table 2). Moreover, the striking similarity of IR-1 and IR-2 to sequences required for VirB binding both in vivo and in vitro makes it likely that these regulatory sequences are indeed bona fide VirB-binding sites (22, 27, 45) (Table 2).

FIG 6.

Schematic of elements identified at the ospD1 locus. The region required for H-NS-mediated silencing (gray box; −1120 to −820) and the VirB regulatory sites IR-1 and IR-2 (blue arrows centered at −1270 and −978, respectively) are shown relative to the primary ospD1 transcription start site (+1). SVRR1 is located between −1352 and −1160 relative to the ospD1 +1 (bracket) and encompasses IR-1. The positions of 5′ ospD1 promoter truncations created in this study are indicated (short lines). The scale represents 500 bp.

The remote region required for H-NS-mediated silencing of ospD1 is located between −1120 and −820 relative to the ospD1 +1 (Fig. 6) and is relatively A/T rich compared to its flanking regions (70% versus 56%, respectively). The latter is not surprising given that H-NS preferentially binds A/T-rich DNA (29–31). Similarly, it was not too surprising to learn that a contiguous 300-bp region, rather than a short discrete site, was necessary for full H-NS-mediated silencing of ospD1, because a similarly large upstream region is required for H-NS-mediated silencing of the well-characterized icsP promoter (27). Even the upstream position of IR-1 relative to the region required for H-NS-dependent silencing of ospD1 was consistent with the architecture of the icsP promoter (27). However, a key difference did exist, namely, a second VirB regulatory site, IR-2, was found within the region required for H-NS-mediated silencing of the ospD1 promoter (Fig. 6). This is striking for two reasons. First, at all other well-characterized VirB-regulated loci, single sites are required for VirB-dependent regulatory control (22, 24, 25, 27, 47, 66, 67). Second, it is surprising that IR-2 is functional because it is located within the region required for H-NS-mediated silencing of ospD1. This suggests that IR-2 is VirB accessible even though H-NS-mediated silencing is traditionally thought to involve the formation of H-NS-DNA filaments that coat and sequester DNA (68).

Initially, the presence of two VirB regulatory sites at the ospD1 locus raised the possibility that these sites function cooperatively (69). However, our analyses revealed that IR-1 and IR-2 do not additively or synergistically (70) contribute to ospD1 promoter activity but instead are functionally redundant. This finding is consistent with our observation that VirB functions solely to counter H-NS-mediated silencing at this locus, since the presence of either IR-1 or IR-2 only increased VirB-dependent ospD1 promoter activity by 1.2-fold (Fig. 4B). Moreover, these observations raise questions about the role and evolution of the functionally active but redundant VirB regulatory sites, IR-1 and IR-2, at ospD1. In their current state, it appears that these sites function as a necessary “backup” for one another in the event that one site is mutated or lost. The need for this type of backup might be attributed to the importance of OspD1 in regulating the T3SS, a key component of Shigella virulence. It is also plausible that the maintenance of IR-1 and/or IR-2 is needed for the transcriptional regulation of nearby VirB-dependent loci. Coincidentally, the genes closest to ospD1 are also VirB dependent (33) and include ospF (∼8.7 kb upstream), orf13 (∼7 kb upstream), orf22, which encodes a hypothetical YnfC family lipoprotein (∼0.5 kb downstream), and ipgB2 (∼1.1 kb downstream). The involvement of IR-1 and IR-2 in the regulation of these genes is currently being pursued within our group.

Regarding the evolution of two VirB regulatory sites at ospD1, it appears that IR-1 is a relic of transposon-mediated duplication events early in Shigella history. This prediction is based on our finding that IR-1 is found on a 193-bp DNA sequence that is repeated three times on pINV. Strikingly, the location of these repeats is conserved in the pINV of all Shigella spp. examined (Fig. 5 and Table 3). Hence, we have named these repeats the Shigella VirB-binding site repeat regions (SVRRs). The SVRRs share characteristics with bacterial miniature inverted repeat transposable elements (MITEs). These nonautonomous transposable elements are commonly A/T rich, <200 bp, and located in intergenic regions and can significantly alter the regulatory network of nearby genetic loci (reviewed in references 71, to ,73). We speculate that the acquisition of these SVRRs may have provided a key step in the evolution of a molecular mechanism to counter H-NS-mediated silencing of A/T-rich and horizontally acquired virulence genes. Indeed, SVRR1 and SVRR2 carry VirB regulatory sites that function in the antisilencing of H-NS-mediated silencing of ospD1 and the icsP-ospZ region (24, 25, 27, 28). Thus, we speculate that the acquisition(s) of the SVRRs sets the stage for VirB, a member of the ParB superfamily of plasmid and chromosomal partitioning factors (74), to co-opt a new role in Shigella as a transcriptional antisilencer, where it functions to offset silencing of horizontally acquired genes mediated by H-NS. Thus, the identification of these SVRR elements highlights the mosaic nature of pINV and that these genetic elements may have contributed to the evolution of a molecular mechanism to counter H-NS-mediated silencing of horizontally acquired virulence genes in Shigella species.

In summary, this study has characterized remote transcriptional regulation by VirB and H-NS of ospD1, which encodes a key regulatory component of the T3SS (34–38). Since remote regulation by VirB and H-NS also controls transcription of icsP, our findings raise the possibility that long-range regulatory effects of transcriptional silencing and antisilencing are commonplace on pINV. Moreover, since transcriptional silencing and antisilencing mechanisms control gene expression in a wide variety of bacteria (reviewed in references 26, 51, and 52), it is possible that transcriptional regulation from remote cis-acting sites in bacterial genomes is more common than previously thought (42, 43). Our other major finding that one of the VirB regulatory sites found at the ospD1 locus (IR-1) is located on a 193-bp sequence, which resembles an ancient mobile element, suggests that its acquisition may have had an important role in the eventual potentiation of transcriptional silencing imparted by H-NS. Thus, these investigations have provided further insight into the transcriptional regulatory properties of two key regulators of Shigella virulence, VirB and H-NS, and provide insight into the genetic events that may have allowed H-NS-mediated silencing of horizontally acquired virulence genes to be overcome, a key event in the evolution of Shigella species and their virulence.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are listed in Table 1. E. coli strains were routinely grown at 37°C in Luria-Bertani (LB) broth (75) with aeration or on LB agar (LB broth containing 1.5% [wt/vol] agar). S. flexneri strains were routinely grown at 25°C or 37°C in LB broth with aeration or on Trypticase soy agar (TSA; Trypticase soy broth containing 1.5% [wt/vol] agar). Where appropriate, antibiotics were added to achieve the following final concentrations: ampicillin, 100 μg ml⁻¹; chloramphenicol, 25 μg ml⁻¹; or kanamycin, 50 μg ml⁻¹. To ensure that S. flexneri strains had maintained pINV during manipulation, Congo red binding was tested on TSA plates containing 0.01% (wt/vol) Congo red (Sigma Chemical Co., St. Louis, MO).

Construction of the ospD1 promoter reporter plasmids.

The 5′ ospD1 promoter truncations were created by PCR amplification from the pINV of S. flexneri 2a strain 2457T using the reverse primer W580 (primers are listed in Table S1 in the supplemental material) in combination with the forward primers W579, W593, W639, W640, W642, W599, W614, or W617. PCR amplicons and holding vector, pBlueScript KS(+) II (Stratagene), were digested with SalI and XbaI prior to ligation and verification by DNA sequencing. Then, the 5′ ospD1 promoter regions were digested out using SalI and XbaI and ligated to a similarly digested lacZ reporter pAFW04 (24). The resulting plasmids carry different lengths of the ospD1 promoter (−1970, −1220, −1120, −1070, −970, −920, −820, and −520 relative to the primary ospD1 +1), the first 79 bp of the ospD1 coding region cloned upstream of a translation stop site in each reading frame, and a promoterless lacZ; thus, the expression of lacZ is directly regulated by the ospD1 promoter.

To create constructs bearing either mutated IR-1 or IR-2 [pPospD1(IR-1mut)-lacZ and pPospD1(IR-2mut)-lacZ], gBlock4 (10 μg ml⁻¹) and gBlock5 (10 μg ml⁻¹) (see Table S2) were digested with SbfI and BsrGI and ligated to a similarly digested pPospD1-lacZ. To create the construct bearing both mutated IR-1 and IR-2, each construct bearing a single mutated site was digested with BsrGI and PsiI. The 9,644-bp BsrGI PsiI fragment from pPospD1(IR-1mut)-lacZ was then ligated to the 252-bp BsrGI PsiI fragment from pPospD1(IR-2mut)-lacZ to create pPospD1(IR-1/2mut)-lacZ. All of the resulting plasmids were verified by EcoNI digest and DNA sequencing.

Quantification of ospD1 promoter activity.

Activities of the ospD1 promoter were determined by measuring β-galactosidase activity using the protocol in reference 28, which is based on the Miller protocol (76). First, pPospD1-lacZ, its derivatives, and, where appropriate, the l-arabinose-inducible plasmid pBAD-virB (pATM324) or pBAD-empty control (pBAD18) were freshly introduced into S. flexneri and E. coli strains by electroporation. Cultures were grown overnight (16 h) at either 25°C (S. flexneri) or 37°C (E. coli) in LB broth with aeration. When using the l-arabinose-inducible plasmids, 0.2% d-glucose was also added. Overnight cultures were then diluted 1:100 and grown for 5 h at 37°C with constant shaking (325 rpm in a LabLine/Barnstead 4000 MaxQ shaker) in LB broth prior to cell lysis. Routinely, β-galactosidase levels were measured in early-stationary-phase cultures, since VirB-dependent ospD1 promoter activity was most pronounced at this stage. All β-galactosidase assays were run in triplicate and were repeated three times.

Statistical analysis.

All statistical analyses were performed in R (77) using RStudio (78). Data were analyzed by two-way mixed-model analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) post hoc test. A P value of <0.05 was considered statistically significant.

Primer extension analysis.

The ospD1 +1 was identified through RNA extraction and primer extension analysis as described by Hensley et al. (46) using a protocol adapted from that described by Aiba et al. (79). Total cellular RNA was extracted using the hot phenol method from 10⁹ cells harvested from early-stationary-phase cultures. Samples were rigorously digested with TURBO DNase (10 U) for 1 h at 37°C per the Invitrogen guidelines (Invitrogen catalog number AM1907), and complete DNA digestion was verified by PCR using primers that bind to the intergenic region upstream of the icsP gene, W69/W70. Total RNA integrity was verified by ethidium bromide gel electrophoresis and optical density. To radiolabel primers, W600 or W601 was 5′ end labeled with [γ-³²P]ATP (specific activity, 3,000 Ci mmol⁻¹) using T4 polynucleotide kinase (10 U μl⁻¹, catalog number EK0031; Thermo Fisher). Unincorporated radionucleotides were removed using illustra ProbeQuant G-50 Micro Columns (supplier number 28903408; GE Healthcare). To produce and label cDNA, approximately 30 μg of total RNA and 1 picomole of the [γ-³²P]ATP-labeled primer were ethanol precipitated in 25 μl of ethanol and 1 μl of 3 M sodium acetate (pH 7.0). Precipitates were then dissolved in 30 μl of hybridization buffer (20 mM HEPES, 0.4 M NaCl, 80% [vol/vol] formamide) and vigorously vortexed for 5 min. The annealing reaction mixture was heated at 50°C for 5 min, incubated at 75°C for 15 min, and maintained at 45°C for 3 h. Samples were ethanol precipitated, and cDNA was generated using SuperScript IV reverse transcriptase (200 U μl⁻¹; Invitrogen) with recombinant RNAsin RNase inhibitor (40 U μl⁻¹, N2515; Promega) at 50°C for 10 min per the manufacturer’s guidelines (catalog number 18090010; Invitrogen). Reactions were aborted by heating samples to 80°C for 10 min, and RNA was removed with 10 mg ml⁻¹ of RNase A (catalog number EN0531; Thermo Fisher) for 30 min at 37°C. Samples were ethanol precipitated in 5 μl of 4 M ammonium acetate (pH 4.8) and 125 μl of ethanol prior to being dissolved in 5 μl of loading dye (95% [vol/vol] formamide, 20 mM EDTA, 0.05% [wt/vol] bromophenol blue, and 0.05% [wt/vol] xylene cyanol; USB Sequenase version 2.0 DNA sequencing kit, Affymetrix 70775Y/Z). Primer extension products were separated on a 6% [vol/vol] denaturing polyacrylamide gel (SequaGel UreaGel system EC-833; National Diagnostics). Sequencing gels were transferred to Whatman paper and vacuum dried. Dried gels were exposed to a phosphor screen overnight and visualized using a Typhoon 9410 variable-mode imager (Amersham). A sequencing ladder generated from m13mp18 single-stranded DNA and a −40 M13 reverse primer from the USB Sequenase version 2.0 DNA sequencing kit (Affymetrix 70775Y/Z) was routinely used to determine the sizes of primer extension products.