Use of a promiscuous, constitutively-active bacterial enhancer-binding protein to define the σ⁵⁴ (RpoN) regulon of Salmonella Typhimurium LT2

Abstract

Background

Sigma54, or RpoN, is an alternative σ factor found widely in eubacteria. A significant complication in analysis of the global σ⁵⁴ regulon in a bacterium is that the σ⁵⁴ RNA polymerase holoenzyme requires interaction with an active bacterial enhancer-binding protein (bEBP) to initiate transcription at a σ⁵⁴-dependent promoter. Many bacteria possess multiple bEBPs, which are activated by diverse environmental stimuli. In this work, we assess the ability of a promiscuous, constitutively-active bEBP—the AAA+ ATPase domain of DctD from Sinorhizobium meliloti—to activate transcription from all σ⁵⁴-dependent promoters for the characterization of the σ⁵⁴ regulon of Salmonella Typhimurium LT2.

Results

The AAA+ ATPase domain of DctD was able to drive transcription from nearly all previously characterized or predicted σ⁵⁴-dependent promoters in Salmonella under a single condition. These promoters are controlled by a variety of native activators and, under the condition tested, are not transcribed in the absence of the DctD AAA+ ATPase domain. We also identified a novel σ⁵⁴-dependent promoter upstream of STM2939, a homolog of the cas1 component of a CRISPR system. ChIP-chip analysis revealed at least 70 σ⁵⁴ binding sites in the chromosome, of which 58% are located within coding sequences. Promoter-lacZ fusions with selected intragenic σ⁵⁴ binding sites suggest that many of these sites are capable of functioning as σ⁵⁴-dependent promoters.

Conclusion

Since the DctD AAA+ ATPase domain proved effective in activating transcription from the diverse σ⁵⁴-dependent promoters of the S. Typhimurium LT2 σ⁵⁴ regulon under a single growth condition, this approach is likely to be valuable for examining σ⁵⁴ regulons in other bacterial species. The S. Typhimurium σ⁵⁴ regulon included a high number of intragenic σ⁵⁴ binding sites/promoters, suggesting that σ⁵⁴ may have multiple regulatory roles beyond the initiation of transcription at the start of an operon.

Background

Transcription in eubacteria is mediated by the RNA polymerase holoenzyme (Eσ), which has five constant subunits (α₂ββ’ω) and a variable subunit (σ). The constant subunits constitute the RNA polymerase core (RNAP), which has the polymerization activity; the σ subunit determines promoter recognition and functions in the Eσ-promoter transition from closed complex to open complex (isomerization). The primary σ factor in a bacterium, such as σ⁷⁰ in Escherichia coli, controls transcription of most housekeeping genes in the cell; alternative sigma factors have specialized regulons that function in the response to environmental stressors or morphological changes, or in developmental systems (for review see [1]). In many bacteria the alternative σ factor σ⁵⁴ (also called RpoN or NtrA) has unusually diverse regulons, with genes that function in a variety of cellular processes, including flagellar biogenesis, response to nitrogen starvation, transport and metabolism of carbon substrates, and tolerance to heavy metals [2-6].

Multiple features, including protein structure, promoter consensus sequence, and mode of activation, distinguish σ⁵⁴ from all other primary and secondary sigma factors, which constitute the σ⁷⁰ family (reviewed in [1,7]). Although both σ⁵⁴- and σ⁷⁰-type sigma factors associate with the β and β’ subunits of RNAP and mediate the binding of Eσ to specific promoter sequences, σ⁵⁴ differs extensively from σ⁷⁰-type sigma factors in primary amino acid sequence and domain organization (reviewed in [8]). The essential promoter features for Eσ⁵⁴ recognition and binding center around conserved GG and TGC elements at -24 and -12, respectively, relative to the transcription start site (TSS) [9], while holoenzymes with the various σ⁷⁰-type sigma factors generally recognize and bind promoter elements at -35 and -10 with the consensus sequences TTGACA and TATAAT, respectively (reviewed in [1]). Perhaps the most important feature of Eσ⁵⁴ that differs from Eσ⁷⁰ is the isomerization process (Figure 1A). For Eσ⁷⁰ the transition from closed complex to open complex is usually spontaneous and rapid, so regulation of transcription initiation frequently occurs at the level of closed complex formation. Initiation of transcription by Eσ⁵⁴ more closely resembles eukaryotic Pol II systems in that Eσ⁵⁴ forms a stable closed complex that requires a bacterial enhancer-binding protein (bEBP) and ATP hydrolysis for isomerization to open complex (reviewed in [10]). The bEBPs add an additional level of complexity to the σ⁵⁴ regulon.

Figure 1.

Activation of σ⁵⁴-dependent transcription and activator structure. A) σ⁵⁴ (red subunit) directs binding of the RNA polymerase (dark blue subunit) holoenzyme (Eσ⁵⁴) to the -12, -24 promoter elements (light blue box). This closed complex is stable and cannot transition to open complex. In response to an environmental or cellular signal, the activator (bEBP; yellow dimers) oligomerizes. For most bEBPs, the oligomer binds to an enhancer (green box) 80 to 150 bp upstream of the promoter and DNA looping brings the activator in contact with σ⁵⁴ in the Eσ⁵⁴ closed complex. Hydrolysis of ATP by bEBP causes remodeling of Eσ⁵⁴, which leads to open complex formation and transcription. There are a few bacteria with bEBPs that are missing the DNA binding domain; after oligomerization, these activators can bind to Eσ⁵⁴ in closed complex with any promoter to stimulate open complex formation (promiscuous activation). B) The domain structure for the Sinorhizobium meliloti bEBP, DctD, is typical of most bEBPs. The amino-terminal regulatory domain (dark blue box) inhibits assembly of the bEBP oligomer until it interacts with an activation signal; the AAA+ ATPase domain (red box) mediates ATP binding and hydrolysis, as well as the protein-protein interactions between bEBPs (oligomerization) and between bEBP and σ⁵⁴; the carboxyl-terminal DNA binding domain (aqua box) contains a helix-turn-helix motif for binding the enhancer. The truncated DctD variant, DctD250, is missing the regulatory and DNA binding domains, so that it is constitutively active and promiscuous in stimulating transcription from σ⁵⁴-dependent promoters.

bEBPs have a modular structure that is generally conserved: an N-terminal regulatory domain, a central AAA+ ATPase domain, and a C-terminal DNA binding domain (Figure 1B; reviewed in [8]). These proteins activate transcription from σ⁵⁴-dependent promoters in three basic steps (Figure 1A). First, the bEBP receives an environmental stimulus through phosphorylation, ligand binding, or protein-protein interactions with the N-terminal regulatory domain that stimulates the bEBP to multimerize through the AAA+ ATPase domain and bind to an upstream activator sequence (UAS or enhancer) via the C-terminal DNA binding domain. The bEBP-UAS complex is then brought into contact with the Eσ⁵⁴-promoter closed complex via a DNA looping event and interactions between highly conserved regions of the AAA+ ATPase domain of bEBP and σ⁵⁴. Finally, ATP hydrolysis drives isomerization, allowing the initiation of transcription.

The requirement for bEBP-mediated activation of σ⁵⁴-dependent transcription presents two problems for global analysis of a σ⁵⁴ regulon. The first is the need for the proper environmental stimulus to activate bEBPs. Since the Eσ⁵⁴ closed complex requires an activated bEBP, σ⁵⁴-dependent promoters are usually transcriptionally silent in the absence of the specific stimulus for the bEBP [8]. Analysis of transcription from σ⁵⁴-dependent promoters under any single growth condition would miss operons whose bEBPs are not activated under the condition tested. Secondly, the requirement for the UAS or enhancer by most bEBPs presents a challenge for predicting whether a Eσ⁵⁴ binding site is functioning as a promoter or not. There is no common consensus sequence for the enhancer and their position relative to the promoter can be quite variable. For many σ⁵⁴-dependent promoters the UAS sequence lies ~70-150 bp upstream of the promoter, but other configurations have been characterized, such as enhancers located 1.5 kb downstream of the rocG promoter in Bacillus subtilis[11] and up to 3 kb upstream of the promoter in artificial constructs of the glnA operon from E. coli[12]. If a σ⁵⁴ binding site is examined for promoter activity in isolation, such as in a promoter-reporter vector, it is difficult to discern whether a site is inactive because it is not a promoter or because the enhancer was not included in the cloned sequence.

Previous studies to define the σ⁵⁴ regulons of Escherichia coli[13], Vibrio cholerae[14] and Geobacter sulfurreducens[15] have recognized the limitations presented by the requirement for activated bEBPs in the characterization of the full σ⁵⁴ regulon, even when σ⁵⁴ is overexpressed from a heterologous promoter. Our approach to overcoming these problems in the global characterization of σ⁵⁴ regulons in bacteria is the utilization of a constitutively-active, promiscuous bEBP, the AAA+ ATPase domain of Sinorhizobium meliloti DctD [16,17]. We chose to assess the efficacy of this approach in Salmonella enterica subsp. enterica serovar Typhimurium LT2 (hereafter referred to as S. Typhimurium LT2), a widely-used laboratory strain, because it has a moderately-sized σ⁵⁴ regulon with 13 known or predicted bEBPs [18], providing sufficient diversity in bEBPs to test our hypothesis.

We report here that use of this constitutively-active, promiscuous bEBP in DNA microarrays and promoter function assays permitted detection of nearly all known and predicted σ⁵⁴-dependent operons. These studies also revealed a new σ⁵⁴-dependent promoter expressing a putative cas1 gene in S. Typhimurium LT2 (STM2938). In addition, chromatin immunoprecipitation-microarray (ChIP-chip) analysis combined with bioinformatics identified 70 Eσ⁵⁴ or σ⁵⁴ binding sites, of which 41 appear to be within open reading frames (ORFs). This surprising number of intragenic sites suggests regulatory roles for σ⁵⁴ or Eσ⁵⁴ that may involve repression, transcriptional interference, or expression of cis- or trans-acting small RNA (sRNA) [19,20].

Results and discussion

Utility of a promiscuous, constitutive bEBP in characterizing the σ⁵⁴ regulon

Since all known σ⁵⁴-dependent promoters require an activated bEBP for transcription initiation, it is a challenge to find a condition under which all promoters can be detected within the σ⁵⁴ regulon of a bacterium. In the recent mapping of the S. Typhimurium SL1344 transcriptome using early stationary phase cultures in rich media (Lennox broth), only one of the known or predicted σ⁵⁴-dependent gene transcripts was detected, pspA[21]. The currently favored approach is over-expression of σ⁵⁴ to facilitate detection of σ⁵⁴-dependent promoters, which assumes a reasonable basal level of activation of the bEBPs. Using relatively low cutoffs for the fold-change (1.5- to 2-fold) in transcript levels between the σ⁵⁴-overexpression strain and wild type or ΔrpoN strains, a considerable portion of the σ⁵⁴-dependent transcriptome was defined in Escherichia coli[13], Vibrio cholerae[14] and Geobacter sulfurreducens[15]. However, not all previously-identified σ⁵⁴-dependent operons were detected for E. coli and G. sulfurreducens, and evidence from the V. cholera and G. sulfurreducens studies suggests that overexpression of σ⁵⁴ may repress expression from some σ⁵⁴-dependent promoters and alter expression of σ⁵⁴-independent promoters [13-15]. We hypothesize that a promiscuous and constitutive variant of the bEBP DctD from S. meliloti can activate transcription from all σ⁵⁴-dependent promoters in S. Typhimurium LT2 at wild-type levels of σ⁵⁴ under a single growth condition, thereby facilitating global characterization of the σ⁵⁴ regulon without overexpression of σ⁵⁴. This promiscuous and constitutive DctD variant is missing the N-terminal response regulator and C-terminal DNA binding domains, leaving only the central AAA+ ATPase domain, residues 141 to 390 of DctD and referred to hereafter as DctD250 [17]. Previous work showed that DctD250 was able to interact with Eσ⁵⁴ in E. coli to drive transcription from the chromosomal glnA promoter and from the S. meliloti dctA promoter in the absence of native DctD and without an enhancer sequence [16,17].

The σ⁵⁴-dependent promoters of S. Typhimurium LT2 are normally responsive to one or more of thirteen known and predicted bEBPs under various growth conditions [18], so to initially assess DctD250 activation of transcription from σ⁵⁴-dependent promoters that respond to different bEBPs in Salmonella, the σ⁵⁴-dependent promoters for the glnKamtB (STM0462) and rtcBA (STM3521) operons were introduced upstream of a promoter-less lacZ gene and the reporter plasmids were transformed into a derivative of S. Typhimurium LT2 (wild-type; WT) and WT containing the DctD250 expression plasmid (WT + DctD250) to perform β-galactosidase assays. The glnKamtB and rtcBA promoters were chosen because neither has predicted σ⁷⁰-dependent promoters within the cloned promoter region and each is responsive to a different bEBP: NtrC for glnKamtB[22] and RtcR for rtcBA[23]. In the WT strain, the glnKamtB and rtcBA operon promoters expressed lacZ at very low levels; but in the presence of DctD250, lacZ was expressed at 150- and 16-fold higher levels, respectively (Table 1). To compare the level of expression stimulated by DctD250 to the level that is seen under physiological conditions that activate the promoter-associated bEBP, lacZ expression from the glnKamtB promoter was assayed in the WT strain in nitrogen-limiting medium, which activates NtrC. Under nitrogen-starvation conditions NtrC multimerizes, binds the enhancer in the cloned promoter region, and hydrolyzes ATP to stimulate transcription by Eσ⁵⁴ at the glnKamtB promoter (see Figure 1A). In the presence of activated NtrC, the glnKamtB promoter expresses lacZ at a nearly 10-fold higher level than in the presence of DctD250. This reduced level of activation by DctD250 relative to the cognate bEBP under activation conditions is consistent with previous studies comparing the activity of truncated versions of bEBPs, which must interact with Eσ⁵⁴ from solution, to that of the wild type bEBPs, which are directed to the target σ⁵⁴ promoter via binding to the enhancer sequence [17,24]. The control reporter plasmids pDV6, which has the σ⁷⁰-dependent, circle junction promoter from IS492[25], and the promoter-less pDS12 expressed lacZ at approximately the same level in WT as WT + DctD250 (Table 1). Based on these results, DctD250 activates transcription from σ⁵⁴-dependent promoters that are normally responsive to different bEBPs under different growth conditions. Therefore, we performed DNA microarray and promoter-reporter analyses in the presence of the promiscuous, constitutive activator DctD250 to assess the efficacy of this approach in defining the σ⁵⁴ regulon of S. Typhimurium LT2.

Table 1.

DctD250-dependent activity of predicted and potential σ⁵⁴-dependent promoters

		Miller unitsb
Promotera	WT	WT+DctD250	WT-N
Intergenic:
STM0462 (glnKamtB)	1.1 ±0.1	180 ±22	1778 ±28
STM3521 (rtcBA)	14 ±6.8	220 ±55	N.D.
Intragenic:
STM0699	5.3 ±1.6	56 ±6.2	N.D.
STM2430	19 ±3.8	90 ±5.5	N.D.
STM2939	2.2 ±1.4	76 ±17	N.D.
Controls:
IS492-CJ (pDV6)	6400 ±2200	9000 ±4100	N.D.
Empty vector (pDS12)	2.0 ±0.5	8.4 ±3.5	4 ±0

^aPromoters assayed in lacZ-reporter plasmids pDS11 and pDS12: intergenic promoters are predicted promoters for the glnALG, glnKamtB and rtcBA operons; intragenic promoters were identified in the ChIP-chip assay (see Results). Controls are the σ⁷⁰-dependent, circle-junction promoter from IS492 and the empty vector, pDS12.

^bβ-galactosidase assays were performed in MOPS minimal medium (WT and WT+DctD250) or in nitrogen-limiting MOPS (WT-N).

Microarray analysis of σ⁵⁴-dependent transcripts in Salmonella expressing DctD250

To determine the genes whose transcription is controlled by σ⁵⁴ in S. Typhimurium LT2 we performed a microarray analysis comparing WT+DctD250 to an isogenic strain with a deletion of rpoN (ΔrpoN+DctD250). RNA collected during mid-log phase growth in nutrient medium was reverse transcribed and cDNAs from each strain were differentially labeled and applied to a complete ORF array containing all annotated open reading frames for S. Typhimurium LT2 [26]. Open reading frames that were transcribed in WT at a level > 3-fold higher than in the ΔrpoN strain, with a p-value <0.02, were considered up-regulated and, for the purpose of the initial categorization of these results, an operon was considered up-regulated if at least one gene met these criteria. In three biological replicates, the same 33 operons were up-regulated in the presence of σ⁵⁴. The microarray results for S. Typhimurium LT2 genes within operons that meet the criteria for up-regulation, or that are known or predicted to be σ⁵⁴-dependent, are shown in Table 2 and Additional file 1. Only 4 genes, STM2722, STM2724, STM2729, and STM2730, which are part of 2 operons in the Fels-2 prophage, were down-regulated >3-fold with a p-value <0.02 in the WT strain as compared to the ΔrpoN strain.

Table 2.

Microarray results for known, predicted, and novel σ⁵⁴-dependent operons and sRNA genes of S. Typhimurium

Locus taga	Gene symbolb	Function	bEBPc	WT/ΔrpoNd	Ref.e
Known σ54- dependent operons and sRNA genes :
STM0368-71	prpBCDE	Proprionate catabolism (putative)	PrpR	45	[4]
STM0830-28	glnHPQ	Glutamine high-affinity transporter	NtrC	7.1	[27]
STM2355	argT	Lysine/arginine/ornithine transport protein	NtrC	3.5	[2]
STM_R0152	glmY	GlmY sRNA	GlrR	0.9	[28]
STM_R0167	glmZ	GlmZ sRNA	GlrR	1.1	[28]
STM4007-05	glnALG	Glutamine synthetase	NtrC	48	[29]
Predicted σ54- dependent operons :
STM0462-63	glnK amtB	hypothetical protein	NtrC	3.6	[22]
STM0577-72		PTS (putative)	STM0571	67	[18]
STM0649.S-53		Hydrolase (putative)	STM0652	11	[18]
STM0665-62	gltIJKL	Glutamate/aspartate transporter	NtrC	1.8	[18,30]
STM1285-84	yeaGH	Serine protein kinase (putative)	NtrC	2.5	[18,30]
STM1303-07	astCABDE	Arginine/ornithine/glutamine metabolism	NtrC	2.4f	[31,32]
STM1690-86	pspABCDE	Phage shock proteins	PspF	17	[5]
STM2360-56	------ubiX	Amino acid transport (putative)	STM2361	100	[18]
STM2840-41	norV ygbD	Nitric oxide reductase	NorR	16	[18,33]
STM2843-42	hydN hypF	Hydrogenase maturation proteins	FhlA	13	[34]
STM2853-44	hycABCDEFGHI-	Hydrogenase 3	FhlA	26	[35]
STM2854-58	hypABCDE	Formate-hydrogen lyase system	FhlA	5.6	[35]
STM3521-18	-rtcBA	RNA repair system (putative)	RtcR	71	[23]
STM3568	rpoH	Heat shock sigma factor (σ32)		1.7	[36,37]
STM3772-66		PTS (putative)	STM3773	39	[18]
STM4172	zraP	Zinc resistance-associated protein	ZraR	16	[3,18]
STM4173-74	hydHG	Zinc resistance two-component system	ZraR	3.7	[3]
STM4244	pspG	Phage shock protein	PspF	1.4	[38]
STM4285	fdhF	Formate dehydrogenase	FhlA	29	[39]
STM4535-40.s		PTS (putative)	STM4534	16	[18]
Novel σ54- dependent operon :
STM2944-2937		CRISPR-associated genes		1.6f	--

^aLocus tags for genes within operons or sRNA genes are grouped by those previously shown to be σ⁵⁴-dependent in Salmonella, previously predicted to be σ⁵⁴-dependent, or identified in this study as encoded in a novel σ⁵⁴-dependent transcript. Locus tags for operons that are not up-regulated are in bold type.

^bGenes that have not been assigned a gene symbol are represented by a dash (−).

^cKnown or predicted bacterial enhancer-binding protein (bEBP) that activates the σ⁵⁴-dependent operon.

^dSignal ratio for the first gene in the operon in WT and ΔrpoN strains expressing DctD250 from pPHBP92. Operons with at least one gene with a signal ratio >3 and p-value <0.02 are considered up-regulated by RpoN; signal ratios above the 3-fold cut off are in bold type. Data for all genes in these operons can be found in Additional file1.

^eReferences for operons shown to be σ⁵⁴-dependent in Salmonella and for operons either determined to be σ⁵⁴-dependent in other bacterial genera or predicted to be regulated by σ⁵⁴ in Salmonella are listed.

^fThe first gene in the operon was <3-fold up-regulated, but other genes in the operon were >3-fold up-regulated.

Known σ⁵⁴-dependent operons and sRNA

If our hypothesis is correct, then in the presence of DctD250 we should observe up-regulation of operons (one or more structural genes) and sRNA genes that are known to have σ⁵⁴-dependent promoters, even though they are normally activated by different bEBPs. Previously, four Salmonella operons have been experimentally shown to be regulated by σ⁵⁴: prpBCDE[4], glnHPQ[27], argT[2], and glnALG[29]. Additionally, two sRNA genes, glmY and glmZ, have also been shown to have σ⁵⁴-dependent promoters [28]. Table 2 summarizes the genes, functions, bEBPs, and microarray results for the known σ⁵⁴-dependent operons and sRNA genes of Salmonella.

The DNA microarrays showed up-regulation of all four known σ⁵⁴-dependent operons in Salmonella, prpBCDE, glnHPQ, argT, and glnALG (Table 2). The two sRNA genes with known σ⁵⁴-dependent promoters did not appear up-regulated by σ⁵⁴. This result was not surprising since in S. Typhimurium both glmY and glmZ possess σ⁷⁰-dependent promoters that fully overlap the σ⁵⁴-dependent promoters, such that the Eσ⁷⁰ and Eσ⁵⁴ compete for binding to their respective promoters [28]. Gopel et al. [28] demonstrated that the level of glmY transcription was similar in wild type and ΔrpoN cells and that transcription of glmZ actually increased in the rpoN mutant, reflecting that the σ⁷⁰-dependent promoter for glmZ is stronger than the σ⁷⁰-dependent promoter for glmY. The presence of a σ⁷⁰ promoter does not necessarily preclude detection of a σ⁵⁴-dependent promoter controlling expression of a gene or operon in these microarray assays, though; the promoter region of glnA has non-overlapping σ⁷⁰- and σ⁵⁴-dependent promoters [29], yet was up-regulated 48-fold. Taken together, these results for the known σ⁵⁴-dependent promoters are consistent with our hypothesis that DctD250 can promiscuously and constitutively activate σ⁵⁴-holoenzyme at a variety of σ⁵⁴-dependent promoters.

Confirmation of predicted σ⁵⁴-dependent operons

There are 20 operons that we define as ‘predicted’ σ⁵⁴-dependent operons in Salmonella. These predictions are based on in silico analyses indicating either homology to known σ⁵⁴-dependent operons in E. coli and other enteric bacteria or promoter sequence homology along with genetic proximity to predicted bEBP genes [3,5,18,22,23,30-39]. However, σ⁵⁴-dependent transcription of these operons has not previously been experimentally demonstrated in Salmonella. In the DNA microarrays, 16 of the 20 operons that have been predicted to have σ⁵⁴-dependent promoters in Salmonella were up-regulated in WT+DctD250 as compared to ΔrpoN+DctD250 (Table 2), providing experimental evidence that these genes are, in fact, regulated by σ⁵⁴ in S. Typhimurium LT2.

For these 16 up-regulated σ⁵⁴-dependent operons there are 11 different bEBPs that either are known or predicted to activate expression from their σ⁵⁴-dependent promoters (Table 2). Five of the up-regulated operons, STM0577-0572, STM0649.s-0653, STM2360-2356, STM3772-3766, and STM4535-4540.s, were predicted to be σ⁵⁴-dependent based on linkage to a predicted bEBP and an upstream sequence with the essential -12 and -24 elements of a σ⁵⁴-dependent promoter [18]. There are no orthologs in E. coli for the predicted bEBPs associated with these operons; three of these predicted bEBPs, STM0571, STM3773 and STM4534, are similar to the LevR-type EBPs found in Gram-positive bacteria [18]. In addition to the microarray evidence presented here for σ⁵⁴ regulation of these operons, we know that STM3773 is the bEBP controlling expression of STM3772-3776 and that this operon encodes the components of a phosphotransferase system permease for D-glucosaminic acid and enzymes required for catabolism of this acid sugar [40]. These results show that DctD250 can activate expression at σ⁵⁴-dependent promoters that are normally regulated by the LevR-type bEBPs.

Of the four predicted σ⁵⁴-dependent operons that did not fulfill our criteria for upregulation in the microarray, at least two have additional σ⁵⁴-independent promoters, which may have masked the effect of σ⁵⁴ on transcription levels. The heat shock sigma factor gene rpoH has been shown to be under the control of additional promoters and other regulatory proteins in E. coli[36]. The conservation of this promoter region for rpoH in S. Typhimurium LT2 suggests that a similar complex regulatory scheme may be involved [37], thereby reducing the effects of the ΔrpoN mutation. The yeaGH operon, which was just below the 3-fold cutoff for up-regulation in the microarray analysis, has previously been shown to be under control of σ^S in Salmonella[41]; however, our assays utilized S. Typhimurium LT2, which has a defective rpoS gene due to a transversion mutation in the start codon [42]. The promoter-reporter assay with the yeaGH promoter region, described below, suggests there is a σ⁵⁴- and σ^S-independent promoter expressing the yeaGH operon in both the WT+DctD250 and ΔrpoN+DctD250 strains.

The frequency of alternate promoters seen for the σ⁵⁴-dependent operons in Salmonella (at least 15% for the known and predicted promoters in our analyses) is not unique. Zhao et al. [13] estimate that 14% of σ⁵⁴-dependent genes in E. coli are transcribed by σ⁷⁰-associated RNA polymerase and suggest that expression of σ⁵⁴–dependent genes from alternate promoters allows for differential expression under various environmental conditions.

New potential σ⁵⁴-dependent genes

In addition to the σ⁵⁴-dependent expression of known or predicted genes and operons, the DNA microarray analysis revealed up-regulation of a gene, STM2938, which has not previously been reported or predicted to be σ⁵⁴-dependent. STM2938 is the penultimate gene in a nine-gene operon that is annotated as a group of CRISPR-associated (cas) genes. Although none of the other genes in this operon seem to be controlled by σ⁵⁴, further evidence is presented below that supports the presence of a σ⁵⁴-dependent promoter within the gene upstream of STM2938. CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) and cas genes constitute an adaptive immune system in bacteria and archaea that protects against invading mobile DNA, such as phage and plasmids [43]. The response to phage infection, which is referred to as phage shock, is regulated by σ⁵⁴ and the bEBP PspF in E. coli[5]; thus, it would not be surprising for essential components of the bacterial immune response in phage infection to be regulated similarly. The potential σ⁵⁴-dependent gene STM2938 is a homologue of the cas1 gene, which is an endonuclease that is associated with all CRISPR loci and is most likely involved in the adaptation phase of the CRISPR-cas immune system [44]. The regulation of this cas1-like gene by PspF in Salmonella is currently under investigation.

There were 12 additional ORFs that met the 3-fold cutoff for up-regulation by σ⁵⁴ in the microarray assay, including genes for pilin biosynthesis (hofB), histidine ammonia lyase (hutH), bEBPs (ygaA, fhlA), propanediol utilization (pduG), siderophore production (iroD), and cell invasion (invG). The whole genome chromatin immunoprecipitation assays described below did not reveal σ⁵⁴ binding sites associated with these ORFs; thus the expression of these genes may be indirectly affected by the absence of σ⁵⁴ in the ΔrpoN mutant, or constitute false positives (Additional file 1).

ChIP-chip analysis of genome-wide σ⁵⁴ binding sites in Salmonella

In the characterization of the σ⁵⁴ regulon of Salmonella, determination of the genomic binding sites for the Eσ⁵⁴ allows confirmation of primary transcripts indicated by microarray analysis and recognition of potential σ⁵⁴-regulated genes that might not have been detected due to instability of the transcripts. To assess the binding of Eσ⁵⁴ in the S. Typhimurium LT2 genome, we isolated σ⁵⁴-DNA complexes from WT and ΔrpoN strains that did not contain the DctD250 expression plasmid. Since bEBPs do not activate transcription by recruiting Eσ⁵⁴ to promoter sequences [8], inclusion of DctD250 should not be necessary to detect binding of holoenzyme to promoter sequences in the ChIP-chip assay. Protein-DNA complexes containing either Eσ⁵⁴ or σ⁵⁴ are pulled down in the ChIP with α-σ⁵⁴. The σ⁵⁴ subunit is most likely to interact with the genome in the context of the RNA polymerase holoenzyme; however, σ⁵⁴ has been shown to specifically bind in the absence of the core RNA polymerase at σ⁵⁴-dependent promoters that have a T-tract upstream of the GC in the -12 promoter element [45]. DNA fragments from the α-σ⁵⁴ ChIP were labeled and applied to the same complete open reading frame arrays as used in the microarray analysis.

Since the use of the ORF arrays did not allow direct mapping of the binding sites, we combined the ChIP-chip data with in silico analysis to determine the potential σ⁵⁴ DNA binding sites. A Position-Specific Score Matrix (PSSM) was created using 27 known or previously predicted σ⁵⁴-dependent promoters from S. Typhimurium LT2 (Additional file 2); the extent of each promoter sequence used for the PSSM (18 bp) was based on the consensus sequence for σ⁵⁴-dependent promoters defined by Barrios et al. [9] and comparison analysis of the known Salmonella σ⁵⁴-dependent promoters. This PSSM was applied with the Motif Locator program [46] to the enriched ORF sequence and 1000 bp of flanking sequence on both sides of the ORF to identify potential σ⁵⁴ DNA binding sites. The size range of DNA fragments that were pulled down via ChIP and amplified by ligation-mediated PCR was 200–1000 bp long, as determined by agarose gel electrophoresis, suggesting that intergenic binding sites up to 1000 bp from the enriched ORF might be detected in the ChIP-chip assays.

σ⁵⁴ binding to promoters for known, predicted, and novel σ⁵⁴-dependent operons

In the ChIP-chip assays with the WT and ΔrpoN strains, the promoter-proximal gene for all the 24 known and predicted σ⁵⁴-dependent operons and the 2 sRNA genes (Table 2) were enriched, as defined by a stringent cut-off, i.e. signal ratio ≥3 and p-value <0.02 (Table 3). The associated promoter sequences, as determined by the in silico analysis, had PSSM scores ranging from 10.9 to 23.6 and were within 27 to 154 bp of the enriched ORF. In the DNA microarrays, six of the known or predicted σ⁵⁴-dependent operons did not appear up-regulated; but in the ChIP-chip assays the promoter regions for all six operons gave signal ratios ranging from 3.2- to 39-fold greater in WT than in ΔrpoN cells. The detection of the σ⁵⁴-dependent promoters for all the other known and predicted σ⁵⁴-dependent operons supports the efficacy of our approach to mapping potential σ⁵⁴-binding sites.

Table 3.

ChIP-chip signal ratios, PSSM scores, and predicted binding sites for ORFs enriched in the presence of σ⁵⁴

Locus taga	Gene name	Signal ratiob	Orientationc	PSSMd	Start	End	Sequence
Sites located within intergenic regions:
STM0368	prpB	11	+	20.7	417914	417931	TGGCATAGCCTTTGCTTT
STM0448	clpP	4.6	+	14.6	503028	503045	TGTCACGTATTTTGCATG
STM0462	glnK	3.2	+	20.0	520445	520462	TGGCACATCCTTTGCAAT
STM0577		8.3	+	17.9	636883	636866	TGGCACGCCGTTTGCCAT
STM0649.S		6.9	+	18.7	711945	711962	TGGCACGCCTTTTGATTA
STM0665	gltI	3.2	+	22.1	730107	730090	TGGCACGTCTATTGCTTT
STM0830	glnH	16	+	20.8	897079	897062	TGGCATGATTTTTTCATT
STM1285	yeaG	4.1	+	21.5	1363884	1363867	TGGCATGAGAGTTGCTTT
STM1303	astC	4.0	+	21.7	1382105	1382122	TGGCACGAATGCTGCAAT
STM1690	pspA	23	+	20.3	1782486	1782469	TGGCACGCAAATTGTATT
STM2355	argT	3.5	+	16.2	2466359	2466376	TGGCATAAGACCTGCATG
STM2360		4.8	+	23.6	2472731	2472714	TGGCATGCCTTTTGCTTT
STM_R0152	glmY	31	+	20.6	2707874	2707857	TGGCACAATTACTGCATA
STM2809	proV	9.5	-	14.6	2955839	2955822	TGGCATGAATATTGCGAG
STM2840		6.5	+	20.1	2985009	2985026	TGGCACACTAGCTGCAAT
STM2843	hydN	23	+	17.1	2990721	2990704	TGGCACGATTCGTGTATA
STM2853	hycA	31	+	17.9	2999639	2999622	TGGCATGGAAAATGCTTA
STM2854	hypA	71	+	22.4	2999753	2999770	TGGCATAAATATTGCTTT
STM3521	rsr	15	+	21.2	3684734	3684717	TGGCACGCTGGTTGCAAT
STM3568	rpoH	22	+	18.9	3736836	3736819	TGGCACGGTTGTTGCTCG
STM3772	dgaA	3.6	+	20.3	3972484	3972467	TGGCACAACCTTTGCTCT
STM_R0167	glmZ	15	+	19.5	4141620	4141637	TGGCACGTTATGTGCAAT
STM4007	glnA	4.2	+	19.2	4217110	4217093	TGGCACAGATTTCGCTTT
STM4172	zraP	29	+	17.4	4388217	4388234	TGGCACGGAAGATGCAAG
STM4173	hydH	4.8	+	20.1	4388385	4388402	TGGCATGATCTCTGCTTA
STM4244	pspG	39	+	19.4	4465042	4465059	TGGCATGATTTTTGTAAG
STM4285	fdhF	10	+	18.2	4527564	4527547	TGGCATAAAACATGCATA
STM4367	yjeB	3.8	+	14.1	4610407	4610424	TGGCAGATATTTTGCTTG
STM4535		12	+	18.3	4794881	4794898	TGGCACGCCGCTTGCTCT
Sites located within the enriched ORF:
STM0131	ftsQ	7.6	+	6.2	153598	153615	TGGAACGCGTCTTGCAGG
STM0155		4.1	+	9.5	182767	182784	CGGCATGGCATTTGCCAG
STM0322	proA	7.8	-	11.3	368058	368041	CGGCACAGTTTATGCAAG
STM0332		3.0	-	8.1	376286	376269	TGGCCAGAAATATGCTTA
STM0526	ylbA	4.3	+	9.1	588233	588216	TGGCATTAATGCTGCATC
STM0699		14	+	13.7	761691	761674	TGGCATCGATATTGCAAA
STM0879^	potH	5.2	+	12.1	951550	951567	TGGCAGGAGTTTTTCAAT
STM0884	ulaA	5.2	+	10.0	955545	955562	CGGCACGATTTTTTCCAT
STM0901		3.9	+	11.7	971761	971778	TGGCATGAAACTTGTCAC
STM0940		9.5	+	13.7	1018097	1018080	TGGCCTGAATCTTGCTAA
STM0961		7.6	-	17.7	1041686	1041669	TGGCATGAAAGCTGCTCA
STM1361	ydiM	3.6	+	11.6	1443903	1443886	TGGCATTCTTTATGCTCA
STM1390	orf242	8.9	-	12.9	1475563	1475546	TGGCATCATTATTGCCTA
STM1409	ssaJ	5.0	+	6.5	1490273	1490290	TGGCATGAAGGTTCATCG
STM1586		6.6	-	13.5	1672845	1672862	TGGCAAGAATATTGCCAT
STM1594	srfB	4.8	+	13.4	1681565	1681582	TGGCACACGTTTTGCGCT
STM1665		4.2	-	11.7	1759185	1759168	TGGCATCATTTTTTCAAG
STM1904^	yecN	3.6	+	5.9	1998988	1999005	TGGCAAACCTGTGGTATA
STM1928	otsA	5.3	+	8.1	2023398	2023381	TGGCAGGAGCGTTTTATT
STM1990	yedA	5.6	+	14.4	2072998	2073015	TGGCGCGCTTTTTGCCTT
STM2033	cbiC	4.4	-	2.2	2111221	2111238	CGGTATAAATAATGCACG
STM2115	wcaA	4.3	-	9.5	2198775	2198792	TGGCATATAAATTGAGAT
STM2181	yohJ	15	+	4.9	2277993	2278010	AGGCATTTTTCTTGCATC
STM2430	cysK	4.8	-	11.7	2544207	2544190	TGGCATCACTGTTGCAGT
STM2475		9.0	-	1.0	2585621	2585638	TGGCACATCAGGCAAAAG
STM2476	ypfG	3.1	+	12.7	2586874	2586857	TGGCAGGTCACCTGCAAT
STM2517	sinH	4.6	-	11.1	2650462	2650479	TGGTACGGATCTTGCCAT
STM2563	yfhG	4.7	-	6.8	2705786	2705803	CGGCGTAATTTTTGCATC
STM2939	ygcH	10	+	10.9	3080061	3080044	CGGCACAGCTCTTGCATC
STM2957	rumA	5.5	+	14.5	3105809	3105792	TGGAACGCTTTTCGCATT
STM3072		4.1	-	6.9	3234181	3234164	TGGCCCATTGAATGCATC
STM3302	yhbE	5.5	+	12.6	3472042	3472025	TGGCATGATGGTCGCCAG
STM3535	glgA	8.0	+	11.6	3702315	3702298	AGGCATGTTTTATGCAAA
STM3721	rfaP	13.5	+	8.3	3916283	3916300	TGGTACGTAAAATGCACG
STM3863^		7.5	+	11.1	4072959	4072942	TGGCGCGATTATTGCCAG
STM3919	wzzE	4.2	+	11.0	4128295	4128312	TGGCCTGCTATTTGCCCT
STM3924	wecD	22	+	11.8	4133232	4133249	TGGCGCGGAAATTGCACA
STM4013.S		3.6	-	13.6	4222708	4222725	TGGCATAAAACCTGAAAA
STM4226	yjbA	6.4	-	4.2	4446318	4446301	AGGCGCGAATAATGCATC
STM4290	proP	13	+	10.1	4532022	4532039	TGGCCTGATTTTTGCAGG
STM4572	stjB	8.3	-	8.2	4826908	4826925	TGGCGTGGCGATTTCAAT

^aLoci listed in bold are known or predicted σ⁵⁴-dependent promoters.

^bRatio of signals between WT and ΔrpoN cells.

^cOrientation of the predicted binding site with respect to the listed ORF. (+) binding site is in same direction as ORF; (−) binding site is in opposite direction as ORF.

^dPSSM score for the best predicted binding site within 1 kb of enriched ORF using the position-specific scoring matrix derived from the sequences in Additional file 2. As a reference to interpret the PSSM scores, the S. Typhimurium LT2 chromosome contains 3 sites with PSSM scores ≥22.0, 21 sites with PSSM scores ≥18.0, 61 sites with PSSM scores ≥14.0, and 401 sites with PSSM scores ≥10.0.

^The predicted binding site is a potential promoter for a neighboring gene based on its orientation and location within 250 bp of the 5’ end of a neighboring gene or a long intergenic region (>100 bp) that may encode a sRNA.

The ChIP-chip analyses also showed that only one (STM2938) of the 13 newly-identified, potential σ⁵⁴-dependent operons from the DNA microarray assays has a σ⁵⁴ DNA binding site associated with it (Table 3), suggesting that the other 12 operons may be indirectly regulated by σ⁵⁴. The σ⁵⁴ DNA binding site associated with STM2938, the cas1-like gene, is within the upstream gene, STM2939 (539 bp from the start of STM2938). Further characterization of this potential σ⁵⁴ promoter is described below in the promoter-reporter analysis.

σ⁵⁴ binding to newly identified potential promoter and regulatory sites

In total, 70 ORFs were each found to be enriched in 3 replicate samples for the WT cells as compared to the ΔrpoN cells in the ChIP-chip assays (Table 3). The potential σ⁵⁴ binding site with the highest PSSM score within each enriched ORF or up to 1000 bp of flanking intergenic sequence was identified by Motif Locator and is reported in Table 3. For the 70 enriched ORFs, 29 of the associated binding sites mapped to intergenic regions and 41 of the potential σ⁵⁴ binding sites were located within the enriched ORF (Figure 2). In determining the most likely binding site for an enriched ORF, sequence within an adjacent non-enriched ORF was not considered for potential σ⁵⁴ binding sites, since the ORF containing the binding site should be enriched; therefore, even if a site with a higher PSSM score was located in an immediately adjacent non-enriched ORF, the next highest scoring site found in either the enriched ORF or adjacent intergenic sequence was reported as the potential binding site in Table 3. This reflects a limitation of the in silico prediction of σ⁵⁴ binding sites based on a PSSM that was created with known and predicted intergenic promoter sequences; the sequences for intragenic promoters or for σ⁵⁴ binding sites that are regulatory sites, but not promoters, may differ enough to appreciably affect PSSM scores.

Figure 2.

Binding sites predicted by ChIP-chip analysis. A) Location of the predicted binding sites for the 70 ORFs enriched by α-σ⁵⁴ pulldown. Outer bars represent further breakdown by location and orientation of the binding site relative to the enriched ORF, as diagrammed in (B). A (+) indicates that the binding site is in the same orientation as the ORF while (−) indicates that the binding site is in the opposite orientation as the enriched ORF.

Consensus sequences were generated using WebLogo [47] for the intergenic and intragenic potential σ⁵⁴ binding sequences and for the promoter sequences used to generate the PSSM (Figure 3). Noteworthy differences in the consensus sequence for the intragenic σ⁵⁴ binding sites, as compared to the consensus sequences for the intergenic σ⁵⁴ binding sites and PSSM promoters, are at the −23 and −11 positions, which each contribute in different ways to σ⁵⁴-promoter DNA interactions. The −23 A-T base pair is important in promoter recognition by σ⁵⁴; the winged helix-turn-helix DNA binding motif of σ⁵⁴ makes base-specific contacts with the top strand GG at positions −26 and −25 and with the bottom strand T at position −23 [48]. The base pairs immediately adjacent to the conserved GC element in the -12 region of the promoter are involved in Eσ⁵⁴ binding to form the stable closed complex; the bases on the bottom strand of the promoter at the -12 and -11 positions interact with σ⁵⁴ in a short region of ‘early melting’ that stabilizes closed complex until the bEBP binds σ⁵⁴ and activates the holoenzyme to transition to open complex [49]. The reduced conservation of nucleotide sequence at the −23 and −11 positions for the potential intragenic σ⁵⁴ binding sites may reflect varied functionality of these intragenic sites, or a level of inaccuracy inherent to in silico prediction of the binding sites associated with enriched ORFs in the ChIP-chip assays.

Figure 3.

Alignment of σ⁵⁴ binding sites. Weblogos show the consensus sequence for A) 27 known/predicted promoter sequences used to generate the position-specific scoring matrix B) 29 predicted intergenic binding sites for ORFs enriched in ChIP-chip analysis or C) 41 predicted intragenic binding sites from within ORFs enriched in ChIP-chip analysis. Weblogos were generated using the online program available at http://weblogo.berkeley.edu/.

The position and orientation for each potential σ⁵⁴ binding site are indicated in Table 3 and summarized for all the binding sites in Figure 2. This information is useful in considering possible functions for the binding sites. For example, the 16 intragenic σ⁵⁴ binding sites oriented in the opposing direction of the gene might regulate by transcription interference and/or anti-sense RNA [19]. Four intragenic σ⁵⁴ binding sites are within 250 bp of the 5’ end of a downstream gene, or a large intergenic region (>100 bp), and oriented in the direction such that they might act as promoters for the downstream gene or a sRNA [20]. Binding sites located near a functional σ⁵⁴ promoter may serve to accelerate the search for the promoter by Eσ⁵⁴ sliding from the secondary sites [50]; while binding sites adjacent to, or overlapping, a σ⁷⁰- or σ⁵⁴-promoter may bind Eσ⁵⁴ or σ⁵⁴ and repress or activate transcription from the other promoter [51]. The possible functions of the σ⁵⁴ binding sites are quite varied and many are dependent on whether the binding site can function as a promoter.

It is likely that our initial approach to defining the global binding sites of σ⁵⁴ in S. Typhimurium LT2 resulted in an underestimation of the number of binding sites. Multiple sites within ~2,000 bp of an ORF would enrich one or two adjacent ORFs and, in our analysis, would have been counted as one site. In addition, since Eσ⁵⁴-promoter closed complexes are reversible [52], some complexes might have not been detected due to high disassociation rates; detection of these sites may be improved in the presence of DctD250, which stimulates conversion of closed complex to the more stable open complex, and rifampicin, which prevents extension of RNA past the second or third nucleotide [53], thus improving the chances of cross-linking Eσ⁵⁴ at the promoter sequence [54].

Promoter-reporter analysis to determine activity for predicted promoter sequences

To assess the functionality of σ⁵⁴ binding sites defined by the ChIP-chip and PSSM analyses, promoter-lacZ fusion assays were performed using several of these sequences. We had two goals in performing these assays. First, we wanted to further confirm the σ⁵⁴-dependent promoter activity for some of the predicted σ⁵⁴-dependent promoters that were up-regulated in the DNA microarrays and enriched in the ChIP-chip assays. Secondly, we wanted to test the σ⁵⁴ binding site predictions from the ChIP-chip combined with PSSM analyses; i.e. does a potential σ⁵⁴ binding site equate to a σ⁵⁴-dependent promoter? This promoter function assay is the initial exploration of the roles for σ⁵⁴ binding sites located within intragenic regions.

Potential promoters were introduced upstream of a promoter-less lacZ gene in a reporter vector, either pDS11 or pDS12 (which differ only in their MCS sequence). These promoter-reporter plasmids were co-transformed along with the DctD250 expression plasmid into either WT or ΔrpoN cells. After induction of DctD250 expression, standard β-galactosidase assays were performed. The results from WT were compared to those from the ΔrpoN mutant to determine whether activity seen was σ⁵⁴-dependent (Figure 4). For intergenic sequences that were known or predicted to be σ⁵⁴-dependent promoters, the results matched those observed in the DNA microarray assays (Figure 4A, Table 2). The glnA, glnK, and STM3521 (rtcBA operon) promoters showed strong σ⁵⁴-dependent activity. For the glmY, glmZ, rpoH, and yeaG promoters, transcription in the ΔrpoN mutant was either as high as or higher than in wild type cells. This is likely due to the presence of σ⁷⁰-type promoters in the cloned sequence. In addition to the σ⁵⁴ dependent promoters, other promoters have been reported upstream of glmY, glmZ, rpoH, and yeaG[28,36,41].

Figure 4.

Promoter location, orientation, and activity for selected σ⁵⁴ binding sites. The ratio of β-galactosidase activity (Miller Units) in WT+DctD250 vs. ΔrpoN+DctD250 cells is shown for (A) known (light blue bars) and predicted (dark blue bars) σ⁵⁴-dependent promoters, and (B) potential intragenic promoter sequences (red bars) in the promoter reporter vectors, pDS11 or pDS12 (black bars). Double asterisks denote significant increase in β-galactosidase activity in WT+DctD250 versus ΔrpoN+DctD250 (p-value <0.02). Circled numbers below locus tags indicate orientation of the potential promoter sequence, as illustrated in (C). Orientation of potential intragenic promoter sequence is: 1) same as ORF and >300 bp from 3’ end; 2) same as ORF and <300 bp from 3’ end of a convergent downstream gene; 3) opposite of ORF and >300 bp from 5’ end; 4) opposite of ORF and <300 bp from the 3’ end of an upstream gene; and 5) opposite of ORF and <300 bp from 5’ end of gene, but >300 bp from the 3’ end of an upstream gene.

A total of eight intragenic sites identified in the ChIP-chip assay were selected for functional analysis (Figure 4B). All of these predicted sites had PSSM scores >10. As shown in Figure 4C, the sites chosen represent a variety of configurations with regard to their position and orientation within the ORF as well as the position and orientation of downstream ORFs. Given the possible functions for an intragenic promoter sequence (e.g. promoter for a downstream gene or sRNA, generation of antisense RNA, etc.), results of our analysis allow us to determine which, if any, of these roles may be attributable to any of these promoters.

Comparing the levels of lacZ expression in wild type cells to those in ΔrpoN mutants, we found that four of the eight intragenic sites were able to function as σ⁵⁴-dependent promoters. For these sites, the difference observed between WT and ΔrpoN cells varied from 4.6-fold for the sequence located in STM2957 to 8.9-fold for the sequence within STM2939. Overall, the level of transcription from these promoters was relatively low, with Miller units ranging from ~30-100. The low activity levels may indicate that Eσ⁵⁴ has a low affinity for these sequences or that the DctD250 is inefficient in productively engaging closed complexes formed at these sites. A subset of the promoter-reporter plasmids with intragenic sites that exhibited σ⁵⁴-dependent transcription were also assayed in WT cells versus WT-DctD250 to determine the dependence of transcription on the promiscuous, constitutive bEBP (Table 1). All three intragenic promoters assayed, STM0699, STM2430, and STM2939, gave low levels of β-galactosidase activity in the absence of DctD250 and from 4.7- to 34-fold higher levels of β-galactosidase activity in the presence of DctD250. The σ⁵⁴-dependent transcription from intragenic binding sites suggests previously unrecognized regulatory functions for σ⁵⁴ in Salmonella; however, it will be critical to characterize transcription from their chromosomal loci before biological functions can be ascribed.

Some of the potential promoter sequences that were assayed failed to show any transcriptional activity. There are a number of possible reasons for the lack of promoter activity for these sites. Two likely explanations are: 1) the wrong sequence was chosen as the binding site based on the PSSM score and proximity to the enriched ORF in the ChIP-chip assays, i.e. a lower scoring sequence near the enriched ORF was the actual σ⁵⁴ binding site; or 2) the σ⁵⁴ binding site does not function as a promoter but serves another regulatory role, such as an operator site for regulating promoter activity, a site for transient binding in facilitated diffusion, or a site for sequestering Eσ⁵⁴ in order to increase local concentration (since σ⁷⁰ has a higher affinity for RNAP [55]).

Summary of S. Typhimurium LT2 σ⁵⁴ regulon and comparison to σ⁵⁴ regulons of other bacteria

Figure 5 summarizes the results from the DNA microarray and promoter-fusion assays performed in the presence of DctD250 and ChIP-chip in the absence of DctD250 to characterize the σ⁵⁴ regulon of S. Typhimurium LT2. Based on DNA microarray, there are 33 up-regulated operons (76 genes; Additional file 1); global ChIP-chip combined with in silico analysis revealed at least 70 σ⁵⁴ binding sites (Table 3), of which 21 were associated with up-regulated operons from the DNA microarrays. The promoter-lacZ fusions with seven of the 29 intergenic σ⁵⁴ binding sites and eight of the 41 intragenic σ⁵⁴ binding sites showed DctD250- and σ⁵⁴-dependent expression for three intergenic sites (associated with up-regulated operons) and four intragenic σ⁵⁴ binding sites (Table 1, Figure 4). The cellular functions impacted by genes in the σ⁵⁴ regulon of S. Typhimurium LT2 are quite diverse, ranging from carbon-source and amino acid metabolism to response to stressors, such as nitric oxide and toxic levels of zinc (Table 2). Our results suggest that a new cellular process may be added to this extensive list—cell immunity through the CRISPR system; the role of σ⁵⁴ in regulating a cas1-related gene within an operon of CRISPR-associated genes is presently being investigated.

Figure 5.

Comparison of positive results from characterization of the σ⁵⁴ regulon for S. Typhimurium LT2. + Positive results are promoter sequences that were up-regulated >3-fold, or displayed a significant increase in β-galactosidase activity in WT+DctD250 compared to ΔrpoN+DctD250 in DNA microarray, and promoter-lacZ fusion assays, respectively or enriched >3-fold in WT compared to ΔrpoN cells in ChIP-chip assays. Regions of overlap indicate promoters that were positive in multiple experiments.

The σ⁵⁴ global regulon of S. Typhimurium LT2 may differ from that of virulent S. Typhimurium isolates due to accumulated mutations in this extensively-used, laboratory strain, particularly the rpoS mutation that contributes to attenuation of the LT2 strain [42]. Changes in the level of expression of one sigma factor can alter the expression of genes that are expressed by different sigma factors [56]; for example, it has been shown that deletion of rpoN alters expression of σ^S-dependent promoters in E. coli[57]. We are currently characterizing the σ⁵⁴ global regulon of the virulent strain S. Typhimurium 14028s.

The σ⁵⁴ regulons in other δ/γ-proteobacteria have been characterized experimentally to varying extents [13-15,58-60]. Only in Vibrio cholera 037 strain V52 have both global transcripts and binding sites been characterized experimentally [14]. In E. coli MG1655 and Geobacter sulfurreducens, the global σ⁵⁴ transcriptomes were determined and local σ⁵⁴ binding sites associated with up-regulated genes were assessed by computational analysis and selected promoters were assessed experimentally [13,15]. The number and diversity of the operons that are directly controlled by σ⁵⁴-promoters in these δ/γ-proteobacteria are comparable to that of S. Typhimurim LT2. The greatest variability in the σ⁵⁴ regulons of the γ-proteobacteria appears to be the location of σ⁵⁴ binding sites. Zhao et al. [13] estimated 70 σ⁵⁴ promoters in E. coli MG1655, of which 13 (18%) were intragenic or located between convergently transcribed genes. In V. cholera, Dong and Mekalanos [14] identified a total 68 σ⁵⁴ binding sites, of which 35 (51%) were intragenic and, similarly, we found 70 potential σ⁵⁴ binding sites of which 41 (58%) appear to be located in intragenic regions.

Does the success with DctD250 in characterizing the S. Typhimurium σ⁵⁴ regulon predict utility of this constitutive, promiscuous activator in defining σ⁵⁴ global regulons in bacteria from other classes in the Proteobacteria phylum, or from other phyla? The key to activation of Eσ⁵⁴ by DctD250 in diverse bacteria is the ability of the activator to make the appropriate interactions with σ⁵⁴ in the context of the Eσ⁵⁴-promoter closed complex; thus, comparison of interacting regions of σ⁵⁴ and bEBPs between S. Typhimurium and phylogenetically diverse bacteria is a good predictor of success. Extensive characterization of bEBP activation of Eσ⁵⁴ in closed complex has shown that the GAFTGA motif of the AAA+ ATPase domain plays a primary and essential role for productive interactions with Eσ⁵⁴, which lead to transcriptional activation (reviewed in [8]); the GAFTGA motif is very highly conserved among bEBPs in all bacteria that encode σ⁵⁴, which includes bacteria from a majority of the eubacterial phyla [61]. It has not yet been determined which specific residues of σ⁵⁴ are contacted by Loop 1 of the bEBP AAA+ ATPase domain, but it has been clearly demonstrated that multiple residues within the amino-terminal 50 amino acids of σ⁵⁴ (Region I) are key determinants for activator interaction [62] and there is extensive conservation of amino acid sequence in Region I for σ⁵⁴ from phylogenetically diverse bacteria [63]. Thus, the comparison of interacting regions of σ⁵⁴ and the AAA+ ATPase domain among diverse bacteria predicts that DctD250 will be a valuable tool in characterizing the σ⁵⁴ regulons in many bacteria.

Conclusions

The results of DNA microarray and promoter-lacZ fusion analyses of the σ⁵⁴ regulon of S. Typhimurium LT2 in the presence of DctD250 support our initial hypothesis: the AAA+ ATPase activation domain of DctD can stimulate transcription from σ⁵⁴-dependent promoters in a constitutive and promiscuous manner, thereby facilitating the global characterization of σ⁵⁴ regulons. Sixteen previously predicted σ⁵⁴-dependent operons were confirmed, and a new σ⁵⁴-dependent gene, cas1, was identified by the DNA microarray and ChIP-chip analyses. In addition, the ChIP-chip analyses indicate an excess of σ⁵⁴ binding sites compared to the number of σ⁵⁴-dependent transcripts and a high percentage of intragenic binding sites, suggesting that Eσ⁵⁴ and σ⁵⁴ may have more regulatory functions than transcription initiation at the start of an operon or sRNA. The number of functional promoters located inside genes suggests a need to consider such promoters in bioinformatic analyses of transcription factor binding sites.

Methods

Bacterial strains, media, and enzymes

The parental strain, designated wild-type, in these experiments was Salmonella enterica subspecies enterica serovar Typhimurium LT2 derivative MS1868 [leuA414(Am) hsdSB(r^-m⁺)Fels^-] [64]. An isogenic derivative, TRH134, has a deletion in rpoN (ntrA) from codons 8 through 455, rendering it auxotrophic for glutamine [65]. S. Typhimurium strains were cultured in either nutrient broth (NB; Difco Laboratories), MOPS minimal media [66], or nitrogen-limiting MOPS [67]. Media supplement concentrations were 5 mM L-glutamine (Gln), 40 μg/ml L-Leucine (Leu), and 10 mM L-glutamate (Glu). Cloning procedures were performed in E. coli DH5α cultured in Luria-Bertani medium (LB; Fisher Scientific). All strains were grown at 37°C. Antibiotics (Sigma-Aldrich) were used at the following concentrations (μg/ml) for E. coli/S. Typhimurium (NB)/ S. Typhimurium (MOPS), respectively: ampicillin (Amp) 80/120/50; spectinomycin (Spc) 50/125/50; streptomycin (Str) 25/75/0. All enzymes were purchased from New England Biolabs, unless otherwise indicated, and were used according to manufacturer’s recommendations.

Plasmids

Plasmid pPBHP92 is a derivative of the expression vector pTrcHisC (Invitrogen) that expresses the Sinorhizbium meliloti DctD AAA+ ATPase domain (E141-S390, designated DctD250) with an N-terminal 6x-His tag. This plasmid was constructed by digestion of pHX182 [17] with NdeI, filling in the 5’-overhang with T4 DNA polymerase and subsequent digestion with XhoI. The blunt-XhoI fragment containing the truncated dctD was cloned into pTrcHisC, which had been cut with NheI, blunt-ended, and cut with XhoI. The truncated dctD is under control of P_trc and subject to repression by the vector-encoded LacI. The reporter plasmids used in these studies, pDS11 and pDS12, are both derivatives of pDV6 [25] that contain a promoter-less copy of lacZ downstream of a MCS region. The MCS region was generated by annealing two oligonucleotide primers (Additional file 3) which were then ligated into a pDV6 backbone that had been digested with BamHI and HindIII. pDS11 and pDS12 differ only in MCS sequence. Potential promoter sequences were amplified from S. Typhimurium LT2 genomic DNA using Taq polymerase and the primers in (Additional file 3) and cloned into pCR2.1 (Invitrogen). Sequencing analysis to determine accuracy and orientation was performed for all plasmids by Genewiz, Inc. (South Plainfield, NJ). Depending on their orientation in pCR2.1 potential promoter sequences were sub-cloned into pDS11/12 using XbaI and either KpnI or HindIII. Plasmid pTG4, which encodes the DctD AAA+ ATPase domain under control of P_tac/lacI^q, was created by amplifying the corresponding region of pPBHP92 using primers DctD-F/R (Additional file 3), digesting the product with BamHI and HindIII, followed by ligation into the similarly digested pKH66 [68].

Transcriptional profiling by microarrays

S. Typhimurium strains MS1868 and TRH134, each bearing plasmid pPBHP92 (WT+DctD250 and ΔrpoN+DctD250, respectively), were grown overnight at 37°C in NB-Amp. Cultures were sub-cultured in fresh medium and grown to mid-log phase (OD₆₀₀ ≈ 0.8). Since the basal level of DctD250 expression from pPBHP92 was shown to optimally activate transcription from a σ⁵⁴-dependent dctA’-‘lacZ reporter [17], IPTG induction was not used for these cultures. RNA isolated using the RNAeasy kit (Qiagen) was used to generate differentially labeled cDNA using reverse transcriptase as previously described [69]. Labeled cDNA was hybridized to DNA microarrays containing complete open reading frames (ORFs) from S. Typhimurium LT2 printed in triplicate [70]. Microarrays were scanned with a ScanArray Lite laser scanner (Packard BioChip Technologies, Billerica, MA) using ScanArray Express 1.1 software. Signal intensities were quantified using QuantArray 3.0 (Packard). The ratio of WT+DctD250 signal to ΔrpoN+DctD250 signal was determined for each of the triplicate spots and the median value for each ORF was used in the statistical analysis [70]. Data shown is the result of three biological replicates with statistical analysis performed using the WebArrayDB program [71,72]. The intensity values for the three biological replicates of WT+DctD250 and of ΔrpoN+DctD250 were compared for the calculation of the p-values, where the null hypothesis was that the intensities for WT+DctD250 and ΔrpoN+DctD250 would be equivalent. Genes that displayed a WT+DctD250/ΔrpoN+DctD250 signal ratio of >3-fold with a p-value of <0.02 were considered to be up-regulated.

Chromatin immunoprecipitation (ChIP)

ChIP was carried out using the ChIP Assay kit (USB Corporation) essentially as described by the manufacturer’s instructions. Briefly, 100 ml cultures of S. Typhimurium strains MS1868 (WT) and TRH134 (ΔrpoN) were grown overnight in NB at 37°C and sub-cultured in fresh medium the next day. Once cultures reached mid-log phase (OD₆₀₀ ≈ 0.7), cells were treated with formaldehyde (3 ml of a 37% solution per 100 ml of culture) for 10 min. at room temperature to cross-link proteins to DNA. Cross-linking was quenched by the addition of glycine (10 ml of 1.33 M solution per 100 ml of culture) and incubation at 4°C for 30 min. Cells were harvested, washed and lysed in accordance with kit instructions. Cells were lysed in two passages through a French pressure cell at 10,000 psi. Cell extracts were clarified and pre-cleared with the provided protein A-Sepharose bead slurry per the kit instructions. 0.6 ml of the resulting extracts were mixed with 2 μl of rabbit anti-serum against S. Typhimurium σ⁵⁴[73] and incubated with gentle shaking overnight at 4°C. The next day, 50 μl of protein A-Sepharose bead slurry was added to each sample, incubated 1 hr at room temperature and collected by centrifugation. The beads were washed, and protein-DNA complexes were eluted from the beads and disrupted per the supplier’s instructions. DNA was purified from each sample using the Qiagen PCR purification kit.

ChIP-chip assays

Purified ChIP DNA was amplified by ligation-mediated PCR, adapting the procedure found at [http://www.flychip.org.uk/protocols/archive_protocols/lm_pcr.php]. Linkers consisting of complementary oligonucleotides (LM-PCR; Additional file 3) were ligated to the ends of purified DNA repaired with T4 DNA polymerase. Ligated was purified using the Qiagen PCR purification kit and the DNA was amplified with Taq polymerase (Fermentas; Burlington, ON) using LM-PCR-R as the PCR primer and the following cycling conditions: 55°C—2 min (1×); 72°C—5 min (1×); 94°C—5 min (1×); 94°C—1 min, 55°C—1 min, 72°C—1 min (24×); 72°C—5 min (1×); 4°C—hold. The resulting amplicons, most of which were 300–800 bp, were purified using the Qiagen PCR purification kit and to prepare dye-labeled DNA (Cy3 or Cy5) for hybridization to the S. Typhimurium complete ORF microarray. Microarrays were scanned and analyzed as above. ChIP-chip was performed on three biological replicates for the WT and ΔrpoN strains; the statistical analysis of the data was performed as described for the microarray data.

Identifying candidate σ⁵⁴ binding sites in the S. Typhimurium genome

The Motif Locator program [http://www.cmbl.uga.edu/software.html] was used to identify candidate σ⁵⁴ binding sites. The program applies the standard position-specific score matrix (PSSM) described in [46]. We used a PSSM derived from the alignment of 27 high-confidence sites supported by experimental evidence in either Salmonella or E. coli (Additional file 2). Background nucleotide frequencies were assigned in accordance with the genomic G+C content. Pseudo-counts equal to the background frequencies were used in PSSM construction. For ORFs discovered in the ChIP-chip assay, this matrix was used to determine the most likely binding site either within the ORF itself or in the region ±200, 500, or 1000 bp surrounding the gene.

β-Galactosidase assays

The DctD250 expression plasmid pTG4 was introduced into S. Typhimurium MS1868 and TRH134 by electroporation using a GenePulser 2 system (BioRad; Hercules, CA) and the resulting transformants were electroporated with pDS11 or pDS12 reporter constructs containing potential σ⁵⁴-dependent promoter sequences. Overnight cultures grown in MOPS-LeuGln, or nitrogen-limiting MOPS-Glu, with the appropriate antibiotics were sub-cultured into fresh medium, grown to OD₆₀₀ ≈ 0.2, and induced with 50 μM IPTG (empirically determined IPTG concentration for optimal expression of DctD250 from pTG4 to activate known σ⁵⁴-dependent promoters on the reporter plasmids). Cultures were induced for 6 hours and β-galactosidase activity was measured as described previously [74] with the following changes: 1) assays were performed at 37°C, and 2) after stopping reaction, samples were centrifuged and OD₄₂₀ of the supernatant was measured, eliminating the OD₅₅₀ correction for cell debris. Activity was calculated as Miller units: [(OD₄₂₀ × 1000)/(OD₆₀₀ × Time(min) × volume (ml))] [74]. Ratios of activity in wild type/ΔrpoN cells were compared and analyzed using a 2-tailed Student’s T-test. Data shown for each promoter construct represents ≥3 biological replicates.

Accession number for microarray and ChIP-chip data

The DNA microarray and ChIP-chip data were deposited in NCBI GEO under accession number GSE25849.

Abbreviations

Amp: Ampicillin; bEBP: Bacterial enhancer-binding protein; ChIP: Chromatin immunoprecipitation; Eσ: RNA polymerase holoenzyme; LB: Luria-bertani media; LM-PCR: Ligation-mediated PCR; NB: Nutrient broth; RNAP: RNA polymerase; Spc: Spectinomycin; Str: Streptomycin; UAS: Upstream activation sequence.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

DS harvested RNA, carried out promoter-reporter assays, and drafted and revised and prepared the manuscript. JF prepared cDNA and preformed DNA microarrays assays and applied DNA from the ChIP to the microarrays. SP and MM produced the microarrays and protocols and performed the statistical analysis of the microarray therein. JM performed the bioinformatic analyses. TH conceived, designed, and coordinated the study, harvested RNA, and performed ChIP pulldowns. AK conceived, designed, and coordinated the study, harvested RNA, and drafted and revised the manuscript. All authors read and approved the final manuscript.

Supplementary Material

Additional file 1

Complete list of microarray results. The file contains data for all genes within operons that showed up-regulation by σ⁵⁴.

Additional file 2

Sequences used to generate Position-Specific Scoring Matrix.

Additional file 3

Oligonucleotides used.

Acknowledgements

We would like to thank the following people for their assistance on this project: Ashley Bono, Uchenna Ewulonu, and Trevor Wright generated reporter constructs. Tanya Grancharova sub-cloned DctD250. Sonya Chelliah performed ChIP pull-downs. Jennifer Turpin prepared and applied cDNA to the microarrays. This work was supported by National Science Foundation Grant MCB-1051175 (to T.R.H. and A.C.K.). MM and SP were supported in part by NIH Contract No. HHSN272200900040C from PATRIC and grants AI039557 AI052237, AI073971, AI075093, AI077645 AI083646, USDA grants 2009-03579-30127 and 2011-67017-30127, the Binational Agricultural Research and Development Fund. JM was supported in part by the National Science Foundation grant DBI-0950266.