Selective Promoter Recognition by Chlamydial σ²⁸ Holoenzyme

Abstract

The σ transcription factor confers the promoter recognition specificity of RNA polymerase (RNAP) in eubacteria. Chlamydia trachomatis has three known sigma factors, σ⁶⁶, σ⁵⁴, and σ²⁸. We developed two methods to facilitate the characterization of promoter sequences recognized by C. trachomatis σ²⁸ (σ²⁸_Ct). One involved the arabinose-induced expression of plasmid-encoded σ²⁸_Ct in a strain of Escherichia coli defective in the σ²⁸ structural gene, fliA. The second was an analysis of transcription in vitro with a hybrid holoenzyme reconstituted with E. coli RNAP core and recombinant σ²⁸_Ct. These approaches were used to investigate the interactions of σ²⁸_Ct with the σ²⁸_Ct-dependent hctB promoter and selected E. coli σ²⁸ (σ²⁸_Ec)-dependent promoters, in parallel, compared with the promoter recognition properties of σ²⁸_EC. Our results indicate that RNAP containing σ²⁸_Ct has at least three characteristics: (i) it is capable of recognizing some but not all σ²⁸_EC-dependent promoters; (ii) it can distinguish different promoter structures, preferentially activating promoters with upstream AT-rich sequences; and (iii) it possesses a greater flexibility than σ²⁸_EC in recognizing variants with different spacing lengths separating the −35 and −10 elements of the core promoter.

Chlamydia trachomatis is an obligate intracellular bacterial pathogen of humans. It is the major cause of bacterial sexually transmitted diseases in the United States and the leading cause of preventable blindness worldwide (47). A unique developmental cycle, characterized by the conversion of a metabolically inactive but infectious elementary body (EB) to an intracellularly replicating but noninfectious reticulate body (RB) and the return to the EB form at the end of the cycle, makes Chlamydia distinct from other intracellular bacteria (37). Little is known about how gene expression is regulated in Chlamydia, but some control is effected at the level of transcription by σ factors and other transcriptional regulators (1, 16, 39). By binding to the RNA polymerase (RNAP) core enzyme (composed of subunits α₂ββ′ω), the σ factors confer on the enzyme the ability to recognize specific DNA elements (promoters) located upstream of the transcription start site and interact with another transcriptional regulator(s) (10, 15, 20, 57). Based on the homology of amino acids, σ factors can be classified into two groups: those similar to E. coli primary σ factor σ⁷⁰ and those similar to σ⁵⁴ (σ^N) (32). Members of the σ⁷⁰ family contain σ⁷⁰, σ²⁸, and most of the other alternative σ factors. Four conserved functional domains (regions 1 to 4) with subdomains have been proposed for members of the σ⁷⁰ family. Of these domains, regions 2.4 and 4.2 of σ⁷⁰ are involved in the recognition of the −10 and −35 elements of the promoters, respectively (20, 32).

Unlike Escherichia coli, which has seven σ factors, Chlamydia encodes only three known σ factors (22, 42, 52): σ⁶⁶, encoded by rpoD; σ⁵⁴, encoded by rpoN; and C. trachomatis σ²⁸ (σ²⁸_Ct), encoded by rpsD (also called fliA or whiG). The role of each σ factor during the developmental cycle of Chlamydia is not clear.

Studies of molecular mechanisms in Chlamydia are challenging due to the difficulty of propagation and the inability to genetically manipulate this intracellular bacterium. As a result, the promoter specificities of the chlamydial σ factors have been investigated only in vitro, and only a few chlamydial promoters have been characterized (9, 16, 33, 50, 53, 58). It is presumed that σ⁶⁶, a homologue of E. coli σ⁷⁰, is the major σ factor and is responsible for the transcription of housekeeping genes (16, 26). Indeed, several essential genes in Chlamydia have been transcribed in vitro by RNAP containing σ⁶⁶ (9, 34, 50, 53, 55). Much less is known about the target genes and the promoter sequences recognized by chlamydial σ⁵⁴ and σ²⁸. Two σ⁵⁴-promoter-like sequences have been identified upstream of the putative open reading frames CT652.1 and CT683 in Chlamydia, which are constitutively transcribed during the developmental cycle (35, 39). While several late-stage genes are dependent on σ⁶⁶ for their expression (12), a σ²⁸_Ct-dependent promoter has been identified upstream of the late-stage-expression hctB gene (58). This gene encodes a histone-like protein, which is believed to be involved in the conversion of the RB to EB forms of Chlamydia (5). Transcription of the rpsD gene, which encodes σ²⁸_Ct, is heat responsive (49). Thus, σ²⁸_Ct may play a role in controlling the differentiation of Chlamydia and its adaptation response during adverse environmental conditions.

Specifically, our interests are in the mechanism of σ²⁸_Ct activation of transcription. Previously, we demonstrated thatσ²⁸_Ct, when associated with E. coli core RNAP, was able to recognize the E. coli σ²⁸ (σ²⁸_EC)-dependent promoter of fliC (P_fliC), the flagellar filament gene, in vitro (49). However, σ²⁸_Ct was unable to restore the motility to an E. coli mutant defective in the σ²⁸ structural gene fliA. These results suggest that σ²⁸_Ct recognizes a subset of σ²⁸_EC-dependent promoters but not the promoters of all genes that are required for motility in E. coli. Because it is not possible to study chlamydial σ factor specificity in Chlamydia in vivo, we have taken advantage of the different ways that σ²⁸_Ct is able to recognize different σ²⁸_EC promoter sequences in order to characterize the range of promoter elements that may be recognized by σ²⁸_Ct and to identify potential upstream enhancing elements. By microarray analysis and sequence analysis, we report that RNAP containing σ²⁸_Ct (RNAP-σ²⁸_Ct) shares a basic promoter recognition capability with that of RNAP containing σ²⁸_EC (RNAP-σ²⁸_EC). However, we also found, as we predicted, that not all σ²⁸_EC promoters are recognized by σ²⁸_Ct. Specifically, σ²⁸_EC promoters recognized by σ²⁸_Ct tend to have an AT-rich sequence upstream of the −35 promoter element. We confirmed many of our in vivo findings with E. coli by in vitro transcription studies with the chlamydial σ²⁸_Ct-dependent hctB promoter and selected σ²⁸_EC-dependent promoters. We also demonstrated by in vitro transcription analysis that σ²⁸_Ct tolerates greater variation in spacer lengths between the −10 and −35 regions of the hctB promoter than does σ²⁸_EC.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

C. trachomatis strain F/IC-Cal-3 (ATCC VR-346) was propagated in mouse fibroblast L929 (ATCC CCL-1) suspension cultures. EBs were purified by centrifugation through discontinuous Renografin (E. R. Squibb & Sons, Princeton, NJ) as described previously (49). E. coli TOP10 (Invitrogen) was used in cloning experiments. The E. coli strain YK4104 (F⁻ araD139 lacU139 rpsL thi pyrC46 gyrA thyA his fliA) (gift from Robert Macnb, Yale University) (27) was used for complementation studies of the σ²⁸ structural gene fliA. E. coli strains were routinely grown in Luria-Bertani (LB) medium supplemented with appropriate antibiotics.

Plasmids construction.

Plasmids used in this study are listed in Table 1. Oligonucleotides containing target sequences and cloning sites were synthesized by Invitrogen (Carlsbad, CA). Promoter activity in E. coli was assessed using a promoter-probe vector, pCG (Fig. 1), which carries a promoter-less cat-gfp operon. Two cloning strategies were used. First, the promoter regions of chlamydial hctB (P_hctB) and groE (P_groE) were amplified by PCR using PfuUltra high-fidelity DNA polymerase (Stratagene, La Jolla, CA). Genomic DNA from purified EBs of C. trachomatis F/IC-Cal-13 was used as a template for PCR. The PCR product was digested with BamHI and EcoRV and then cloned into pCG through BamHI and EcoRV sites. In the second strategy, the promoter regions of the E. coli flgK, ycgR, tar, or groE gene (P_flgK, P_ycgR, P_tar, or P_groEEc, respectively) were separately generated by annealing two complementary oligonucleotides, which were designed to incorporate XbaI restriction sites near the 5′ end and a blunt end near the 3′ end. The annealed double-stranded DNA was inserted into pCG at the XbaI and EcoRV restriction sites. The resulting constructs contained the different promoters upstream of the cat-gfp reporter genes.

TABLE 1.

Names and descriptions of plasmids

Plasmid	Description	Source
pLC3	A low-copy-number expression vector; a recombinant of pBAD24 and pACYC184; the origin of replication of p15A, Tcr	49
pLF28	pLC3 carrying chlamydial rpsD	49
pLE28	pLC3 carrying E. coli fliA	49
pS28H	A derivative of pBAD24; expression vector with an N-terminal His6 tag; carries 5′ chlamydial rpsD; Apr	49
pES28H	A derivative of pBAD24; expression vector with an N-terminal His6 tag; carries E. coli fliA; Apr	49
pRVB	A recombinant of pRV1 and pBR322; ColE1; Apr	This study
pCG	A promoter probe vector containing promoterless cat::gfp; the origin of replication of ColE1; Apr	This study
pPgroE::cat-gfp	pCG with the promoter region of chlamydial groE (PgroE)	This study
pPgroEEc::cat-gfp	pCG with the promoter region of E. coli groE (PgroEEc)	This study
pPflgK::cat-gfp	pCG with the promoter region of E. coli flgK (PflgK)	This study
pPhctB::cat-gfp	pCG with the promoter region of chlamydial hctB (PhctB)	This study
pPtar::cat-gfp	pCG with the promoter region of E. coli tar (Ptar)	This study
pPycgR::cat-gfp	pCG with the promoter region of E. coli ycgR (PycgR)	This study
pMT504	A cloning vector containing a promoterless guanine (G)-less cassette and a chalmydial rRNA P1-controlled G-less cassette; Apr	53
pGLC	A cloning vector containing a promoterless guanine (G)-less cassette and an E. coli fliC promoter-controlled G-less cassette; Apr	49
pGPgroEEc	pGLC with the promoter region of E. coli groE (PfliC and PgroEEc)	This study
pGPgroE	pGLC with the promoter region of chlamydial groE from serovar F	49
pGPflgK	pGLC with the promoter region of E. coli flgK (PfliC and PflgK)	This study
pGPtar	pGLC with the promoter region of E. coli tar (PfliC and Ptar)	This study
pGPycgR	pGLC with the promoter region of E. coli ycgR (PfliC and PycgR)	This study
pGPhctB	pGLC with the promoter region of chlamydial hctB from serovar F	This study
pGPhctBΔUP	A derivate of pGPhctB deleting the upstream sequence of the PhctB−35 region	This study
pGPhctB-2	A derivate of pGPhctB fusing a DNA sequence upstream of Ptar to the core promoter of PhctB	This study
pGPtar+	A derivate of pGPtar fusing a DNA sequence upstream of PhctB to the core promoter of Ptar	This study
pGPlacUV5	pGLC with the promoter region of E. coli PlacUV5	This study
pGPlacUV5+	A derivate of pGPlacUV5 containing the upstream DNA −65 to −38 sequence fused to core the promoter of PlacUV5	This study
pGS9	A derivate of pGPhctB containing 9 bp between the −35 and −10 regions	This study
pGS10	A derivate of pGPhctB containing 10 bp between the −35 and −10 regions	This study
pGS11	A derivate of pGPhctB containing 11 bp between the −35 and −10 regions	This study
pGS13	A derivate of pGPhctB containing 13 bp between the −35 and −10 regions	This study
pGS14	A derivate of pGPhctB containing 14 bp between the −35 and −10 regions	This study

FIG. 1.

pCG vector for assaying in vivo promoter activity. pCG is a derivative of pRV1 (50). The PvuI/AvaI fragment, which contains rrnB T₁ multiple cloning sites (MCS), chlamydial tuf ribosomal binding site 1 (RBS1), and the reporter cat gene from pRV1, was ligated with the PuvI/AvaI fragment of pBR322, which contains the ColE1 replication origin and rop, resulting in pRVB. pRVB was then modified by (i) replacement of the MCS; (ii) insertion of an enhanced E. coli RBS (RBS2), an improved gfp gene, encoding GFP (36), and the second reporter gene directly downstream of the cat gene; and (iii) addition of a synthetic spf terminator (spfT) downstream of the gfp gene, creating pCG. Thus, pCG contains the MCS convenient for cloning test promoters, promoterless cat-gfp, the ColE1 replicon, a bla gene allowing ampicillin selection, the transcriptional terminator rrnB T₁, preventing readthrough into the cat-gfp operon, and spfT, preventing interference from opposing transcription. The sites used for cloning test promoters in this study are underlined.

Plasmids used for in vitro transcription analysis were constructed by inserting the annealed double-strand DNA fragments, which contained wild-type or mutant promoter regions, into the XbaI and EcoRV sites of pGLC (49) or pMT504 (53). The resulting constructs each contained a test promoter and an internal control promoter (E. coli P_fliC in pGLC derivatives and the chlamydial rRNA P1 in pMT504 derivatives) (Table 1). The final cloned inserts were confirmed by restriction mapping and nucleotide sequencing. The expression plasmids pLF28 and pLE28, which encode σ²⁸_Ct and the σ²⁸_EC, respectively, were described previously (49).

RNA extraction and gene array chip hybridization.

Total RNA was isolated from cells grown in LB medium supplemented with 0.02% (wt/vol) arabinose at 30°C using the QIAGEN RNeasy kit (QIAGEN, Valencia, CA). RNA concentrations were determined spectrophotometrically at 260 nm. The E. coli GeneChip antisense genomic array (Affymetrix, Clara, CA) with oligonucleotides representing 4,200 E. coli open reading frames and 1,350 intergenic sequences was used for analysis of mRNA expression profiles according to the specifications of the manufacturer. Briefly, cDNA was prepared from 10 μg of total RNA through randomly primed reverse transcription using SuperScript II (Invitrogen, Carlsbad, CA). The cDNA was purified using a QIAquick PCR purification column (QIAGEN, Valencia, CA) and then fragmented to a size range of 50 to 200 bases by treatment with DNase I (1 U/μg cDNA) at 37°C for 10 min. Inactivation of the enzyme followed (98°C for 10 min). Fragmented cDNA was biotin labeled using an Enzo BioArray terminal labeling kit with biotin-ddUTP (Enzo Scientific, Farmingdale, NY). Biotinylated cDNA was hybridized to Affymetrix E. coli antisense genome arrays on a rotor (60 rpm) for 16 h at 45°C. The chips were washed, stained, and scanned the following day according to the Affymetrix microarray protocol.

Sequences and computational analyses.

The E. coli K-12 genomic sequence (2) was obtained from GenBank (accessionr no. NC0009134). To define genes that are likely to be in the same transcription unit (TU) (under the control of a common promoter), the genes were grouped as a TU by using the E. coli transcription regulatory databases, called RegulonDB (http://regulondb.ccg.unam.mx/index.html) (46) and EcoCyc (http://EcoCyc.org) (24). C. trachomatis D/UW-3/Cx genomic sequences (52) were obtained from STD Sequence Databases (http://www.stdgen.lanl.gov/). DNA motif searches were performed using BioProspector (http://robotics.stanford.edu/∼xsliu/BioProspector/) (30).

Signal intensities for each feature on the microarrays were subjected to quantile normalization (3). Only features with a minimum signal intensity of 100 pixels in the high-voltage scan were considered for further analysis. Results from three independent experiments were averaged for each strain. The accuracy and statistical significance of the expression profiles were determined by analysis of variance (http://lgsun.grc.nia.nih.gov/ANOVA/). Differentially expressed genes were selected based on false discovery rates (FDRs) of ≤0.05 (significant) (FDR = P × n/rank, where P is the P value, n is the total number of genes, and rank is the rank of a gene ordered in increasing P values) and a change in expression level (n-fold) of ≥3 for the first gene in a TU. Genes with an FDR value of >0.05 were not considered to be regulated and were assigned not significant status, regardless of their differential expression levels.

Determination of promoter activity in E. coli.

In vivo promoter activity was determined by measurement of the chloramphenicol (CAT) and green fluorescent protein (GFP) reporter genes in an E. coli fliA mutant strain, YK4104. This strain does not encode a functional σ²⁸ due to a frameshift mutation in the fliA open reading frame (38). Cells harboring an appropriate plasmid(s) were grown overnight in LB medium supplemented with 0.4% (wt/vol) glucose at 30°C. Cultures were diluted 1:100 in fresh LB medium to an optical density at 600 nm (OD₆₀₀) of 0.3, and then the expression of σ²⁸_Ct or σ²⁸_EC was induced by adding 0.02% (wt/vol) arabinose to the culture. To quantify the fluorescence intensity of GFP expressed in E. coli, 200-μl portions of cultured cells were collected at different times, washed once with 0.1 M phosphate-buffered saline (pH 7.4), resuspended in 100 μl phosphate-buffered saline, and subjected to fluorescence examination in a 96-well plate (Costar, Corning, NY) using a fluorescence microplate reader GENios (TECAN, Grodig/Salzburg, Austria). Cell densities at 600 nm were also determined. The relative fluorescence unit (RFU) was used as a measure of GFP expression [RFU = |(fluorescence intensity of test strains − fluorescence intensity of the strain carrying pCG)|/|(OD₆₀₀ of test strain − OD₆₀₀ of strain carrying pCG)|, where pCG contained cat-gfp lacking a promoter and was used as the intrinsic background fluorescence control]. CAT assays were carried out as described previously (50).

Determination of promoter activity in vitro.

N-terminally His-tagged σ²⁸_Ct and σ²⁸_EC were purified as described after overexpression in E. coli YK4104 from the vectors pS28H and pES28H (49), respectively. RNAP holoenzyme containing σ²⁸ (RNAP-σ²⁸_Ct or RNAP-σ²⁸_EC) was made by addition of a fourfold excess of purified σ²⁸_Ct or σ²⁸_EC to E. coli RNAP core enzyme (Epicenter, Madison, WI). Holoenzyme saturated with E. coli σ⁷⁰ (RNAP-σ⁷⁰) was purchased from Epicenter. To allow for open-complex formation, assays were carried out by preincubating RNAP holoenzyme (20 nM) with pGLC or pMT504-derived supercoiled plasmid DNA (Table 1) (1 μg) in buffer (10 mM Tris-HCl, pH 8.0, 200 mM NaCl, 5 mM dithiothreitol) for 5 min at 37°C. Transcription was initiated by adding 400 μM ATP, 400 μM UTP, 1.2 μM CTP, 0.20 μM [α-³²P]CTP (3,000 μCi/mmol), 100 μM 3′-O-methylguanosine 5′-triphosphate (Amersham Pharmacia Biotech), and 100 μg/ml heparin. Reaction mixtures were incubated for 15 min at 37°C, and reactions were stopped by the addition of a solution of 95% (vol/vol) formamide, 0.025% (wt/vol) xylene cyanol, 0.025% (wt/vol) bromophenol blue, and 0.5 mM EDTA (pH 8.0). Transcripts made by RNAP were detected using 6% (wt/vol) polyacrylamide-7 M urea in polyacrylamide gel electrophoresis. Signal intensities from autoradiographs were determined with Bio-Rad Quantity One software (Bio-Rad, Hercules, CA).

Primer extension analysis.

Primer extension analysis was performed as described previously (50). Primer hctB-pe1 (5′-TCTTGTGCTGCATTTCTTTTGT-3′) was labeled with [α-³²P]ATP (Amersham Biosciences, Piscataway, NJ) using T4 polynucleotide kinase (New England Biolabs, Beverly, MA) and annealed with total RNA isolated from E. coli YK4104 carrying plasmid pP_hctB::catgfp and pLF28 or pLE28. The primer was extended with reverse transcriptase (Promega, Madison, WI) at 42°C for 1 h, and the products were subjected to electrophoresis on a DNA sequencing gel. A comparison DNA ladder that was generated with the same primer using a DNA sequencing kit (USB Corporation, Cleveland, OH) was used as a control.

RESULTS

Comparative transcription profiles of σ²⁸-dependent genes in E. coli.

E. coli strain YK4104, which is defective in the σ²⁸ structural gene (fliA), was transformed with pLF28 (containing rpsD, the gene encoding σ²⁸_Ct), pLE28 (containing fliA, the gene encoding σ²⁸_EC), or pLC3 (the empty vector). Sigma factor gene expression was induced by the addition of arabinose, and total RNA was isolated and used for microarray analysis of the E. coli genome. Specific transcript levels from the σ²⁸-expressing cells were compared with that from pLC3-carrying cells (i.e., transcription ratios, σ²⁸_Ct/pLC3 versus σ²⁸_EC/pLC3) to determine the influence of plasmid-encoded σ²⁸ upon transcription.

We found 26 genes/operons (referred to herein as TUs) to be significantly upregulated by σ²⁸_EC and 15 TUs to be significantly upregulated by σ²⁸_Ct (see Materials and Methods for the criteria for significance). The TUs upregulated following the expression of σ²⁸ were clustered into three groups: cluster A, upregulated by both σ²⁸_Ct and σ²⁸_EC (Table 2); cluster B, upregulated only by σ²⁸_EC (Table 3), and cluster C, upregulated only by σ²⁸_Ct (Table 4). Thirteen genes in nine TUs were upregulated by both σ²⁸_Ct and σ²⁸_EC, most of which showed higher transcription ratios in the σ²⁸_EC-expressing strain than in the σ²⁸_Ct-expressing strain (Table 2). Of the genes recognized by both σ²⁸_EC and σ²⁸_Ct, eight have been identified previously as members of the class 3 flagellar regulon that is controlled by σ²⁸_EC; these genes are involved in the late stage of flagellar assembly in a hierarchical regulatory cascade in E. coli (7, 14, 25). Genes recognized by σ²⁸_EC but not by σ²⁸_Ct included 14 class 3 and 7 class 2 flagellar genes, which are controlled by both FlhD/FlhC proteins (the master regulator) and σ²⁸_EC and encode components required for the early phase of flagellar assembly (7, 19, 31).

TABLE 2.

Cluster A genes upregulated by σ²⁸_Ec and σ²⁸_Ct

Transcription unita	Gene	Transcription ratiob		Function
		σ28Ec/pLC3	σ28Ct/pLC3
Known flagellar genes
fliC	fliC	68.046	14.365	Flagellin, filament structural protein
fliD-fliS-fliT	fliD	30.346	3.181	Flagellin, enables filament assembly
	fliS	22.055	2.934	Repressor of class 3 operons
	fliT	14.194	2.882	Repressor of class 3 operons
flgM-flgN	flgM	23.431	22.065	Anti-FliA, anti-sigma factor
	flgN	11.679	11.008	Protein of flagellar biosynthesis
flgK-flgL	flgK	18.720	19.925	Hook-filament junction protein
	flgL	25.817	30.061	Hook-filament junction protein
ycgR	ycgR	10.032	16.638	Hypothetical protein
Other genes
yjdA-yjcZ	yjdA	38.842	22.019	Hypothetical protein (YjdA)
	yjcZ	5.092	3.694	Hypothetical protein
flxA	flxA	30.960	12.277	Hypothetical protein
ykfB	ykfB	16.263	9.590	Hypothetical protein
yfiD	yfiD	7.328	4.709	Alternate pyruvate formate lyase subunit

^aUpregulated genes are in bold type.

^bTranscription ratios of the genes in a σ²⁸_Ct- or σ²⁸_Ec-expressing E. coli strain versus that in the vector-carrying strain.

TABLE 3.

Cluster B genes upregulated by σ²⁸_Ec only

Transcription unita	Gene	Transcription ratiob (σ28Ec/pLC3)	Function(s)
Known flagellar genes
fliA-fliY-fliZ	fliA	171.87	FliA, sigma 28 factor
	fliY	42.707	Possible cell density-responsive regulator
	fliZ	5.453	Putative periplasm-binding transport protein
tar-tap-cheR-cheB-cheY-cheZ	tar	23.378	Methyl-accepting chemotaxis protein II
	tap	33.558	Methyl-accepting chemotaxis protein IV
	cheR	6.437	Response regulator for chemotaxis
	cheB	11.395	Response regulator for chemotaxis
	cheY	13.858	Chemotaxis regulator
	cheZ	18.823	Chemotaxis protein
motA-motB-cheA-cheW	motA	14.025	Proton conductor component of motor
	motB	32.779	Flagellar motor rotation
	cheA	13.633	Sensory transducer kinase
	cheW	9.598	Chemotaxis protein, regulation
yhjH	yhjH	22.024	Flagellar-function protein
	yhjG	4.105	Hypothetical protein
aer	aer	8.553	Aerotaxis sensor receptor, flavoprotein
trg	trg	6.088	Methyl-accepting chemotaxis protein
fliL-fliM-fliN-fliO-fliP-fliQ	fliL	6.008	Flagellar biosynthesis
	fliM	3.273	Flagellar motor switch protein
	fliN	4.415	Flagellar motor switch protein
trs	tsr	5.642	Methyl-accepting chemotaxis protein
Other genes
ynjH	ynjH	23.436	Hypothetical protein
modA-modB-modC	modA	9.975	Molybdate ABC transporter
	modB	4.227	Molybdate ABC transporter
b1742 (ves)	ves	9.763	Cold-induced member of the CspA family
yfiA	yfiA	6.624	Stationary-phase translation inhibitor
galE-galT-galK-galM	galK	6.012	Galactokinase
b1345-ydaQ	b1345	4.902	Putative transposase
ompF	ompF	4.735	Outer membrane protein
araC	araC	4.265	AraC transcriptional dual regulator
uspA	uspA	4.081	Universal stress protein, regulator

^aUpregulated genes are in bold type.

^bTranscription ratio of the genes in a σ²⁸_Ec-expressing E. coli strain versus that of the vector-only control strain.

TABLE 4.

Cluster C genes upregulated by σ²⁸_Ct only

Transcription unita	Gene	Transcription ratiob (σ28Ct/pLC3)	Function(s)
melR	melR	6.518	Transcriptional dual regulator
ynaF (uspF)	ynaF	5.168	Stress-induced protein, ATP-binding protein
groES	groES	5.064	GroES, 10-kDa chaperone
malK-lamB-malM	malK	4.948	Maltose ABC transporter
	lamb	4.411	Maltose high-affinity receptor
yfcZ	yfcZ	3.591	Conserved hypothetical protein
yeiY	yeiY	3.098	Putative dihydrothymine dehydrogenase

^aUpregulated genes are in bold type.

^bTranscription ratio of the gene in a σ²⁸_Ct-expressing E. coli strain versus that in the vector-carrying strain.

Analysis of the putative DNA recognition sites by σ²⁸_Ct in E. coli.

We investigated potential conserved DNA elements upstream of σ²⁸-upregulated genes in E. coli. To do so, 300 bp of sequence immediately upstream of the ATG start codons of the first gene in each predicted TU was retrieved to construct sequence sets that were likely to contain regulatory elements. Almost all of the known regulatory sites in E. coli K-12 are located within this range (6). Using the motif search program BioProspector (30), we conducted a de novo search for motifs that contained two 8-bp blocks separated by a gap ranging from 10 to 14 bp within the sequence sets, as σ²⁸_EC recognizes promoters with a consensus of TAAAGTTT-N₁₁-GCCGATAA (18). The highest-scoring motif was found to be TAAAGTTT-N_X-GCCGATAA. This motif is preceded by seven of the nine TUs in cluster A (Table 2), including the five flagellum genes, flxA and ykfB (not known to be a σ²⁸-activated gene), and 9 of 17 TUs in cluster B (Table 3), including 8 flagellar TUs and modA-modB-modC, which has a σ²⁸-like promoter that has not been experimentally confirmed (40). The sequence identified upstream of ykfB is TGATGAAT-N₁₂-GCCGATAA, which matches 4 of the 8 bp in the consensus −35 box and perfectly matches the −10 box but has a 12-bp spacer instead of the canonical 11-bp spacer. We were unable to identify a TAAAGTTT-N_X-GCCGATAA motif in the upstream sequence regions of cluster C genes.

From the 26 TUs that are significantly induced by σ²⁸_EC (Tables 2 and 3), we found 14 that contained experimentally verified σ²⁸_EC-dependent promoters (14, 18, 25, 31) (Fig. 2). Two highly conserved regions resembling the reported consensus −35 and −10 elements (18) were observed. We next investigated the adenine (A) and thymine (T) content 30 bp immediately upstream of the −35 element from each of the 14 genes in an attempt to identify potential upstream elements, which enhance transcription initiated by some σ factors (13, 43). Using Spearman's correlation test we found that the correlation between the AT content and the transcription levels directed by σ²⁸_Ct for these genes was 0.6697, with a P value of 0.0063 (significant). In contrast, the correlation between the AT content and the transcription levels directed by σ²⁸_EC is 0.4206, with a P value of 0.119 (insignificant). This suggested that σ²⁸_Ct-directed transcription might be influenced by AT-rich sequence features upstream of the −35 element of the promoter.

FIG. 2.

Sequence alignments and subsets of promoters from upregulated genes in E. coli. (a) Subset 1 promoters that were significantly upregulated by both σ²⁸_EC and σ²⁸_Ct. AT-rich sequences are present upstream of the −35 regions that center mainly at −41, −51, and −56. (b) Subset 2 promoters that were upregulated by σ²⁸_EC only. In this subset, a lower AT content was observed upstream from the −35 region. The results are also shown as logos in which the heights of the letters in bits are proportional to their frequencies. Logos were generated by Weblogo (http://weblogo.berkeley.edu/logo.cgi). For ease of interpretation, the promoter sequences are numbered from the +1 position of the fliC promoter (in bold and italics) and aligned according to their −35 boxes. The regions of high homology (−35 and −10) are highlighted in red, green, blue, and black. The AT contents of the 30-bp sequence immediately upstream of −35 (−65 to −37) are shown at the right.

To further characterize the effect of the upstream AT-rich sequences on selective σ²⁸_Ct-dependent transcriptional activation, we divided the 14 experimentally confirmed σ²⁸_EC-dependent genes into two additional subsets (Fig. 2): subset 1 contains genes that were significantly upregulated by both σ²⁸_EC and σ²⁸_Ct, and subset 2 contains genes that were upregulated by σ²⁸_EC but were relatively unaffected by σ²⁸_Ct. Two statistical tests were used to address the question of whether or not the two subsets have different AT contents. The two-sample t test gave a P value of 9e-5 (significant), and the Wilcoxon rank sum test gave a P value of 8e-4 (significant). These results further suggest that an upstream AT-rich sequence in the promoter region might be favorable for selective transcription activated by σ²⁸_Ct compared to that activated by σ²⁸_EC.

These observations raised three questions that we sought to further address experimentally. (i) Was the observed increase in σ²⁸_Ct-dependent transcription due to recognition of specific promoter sequences by σ²⁸_Ct? (ii) Did RNAP-σ²⁸_Ct initiate transcription from an authentic transcription start site in E. coli? (iii) What promoter configuration, including upstream elements and −35 and −10 spacer lengths, is preferentially recognized by σ²⁸_Ct?

RNAP-σ²⁸_Ct recognizes specific promoters in E. coli.

We attempted to confirm the promoter activities of three σ²⁸_EC TUs (P_flgK, P_ycgR, and P_tar) identified in our microarray assay, as well as the σ²⁸_Ct-dependent promoter from the one known chlamydial σ²⁸_Ct-dependent gene, hctB (P_hctB) (58). We selected P_flgK and P_ycgR because they contain consensus-like core promoter elements and were positive in our microarray assay when both σ²⁸_Ct and σ²⁸_EC were expressed in E. coli. In contrast, we selected P_tar because it contains a consensus core element but was negative in the microarray assay with σ²⁸_Ct. We also tested P_groEEc, because expression of the groE gene, which is recognized by σ³²_EC and lacks a σ²⁸ core element, was increased in the σ²⁸_Ct-expressing strain by microarray analysis but not in the σ²⁸_EC-expressing strain. As a control, we analyzed the σ⁶⁶-dependent promoter from the chlamydial groE gene (P_groE) (55). Promoter activity was measured in vivo using the cat-gfp-based reporter system by cloning each promoter into the cat-gfp plasmid pCG (Fig. 1). pCG contains a multiple cloning site sandwiched between the transcriptional terminator rrnB T₁, preventing readthrough into the cat-gfp operon, and spfT, preventing interference from opposing transcription (Fig. 1). In control experiments (data not shown), no fluorescence or chloramphenicol resistance was noted in strains carrying the empty vector pCG.

Following the induction of σ²⁸_Ct by the addition of arabinose to the media, constant increases in the GFP fluorescence levels were observed over 8 h for all promoters tested, although the increase of P_tar was barely detectable (Fig. 3a). The promoter sequences tested are listed in Fig. 4a. Increases were also noted in strains that harbored both pLE28 (fliA⁺) and each of above promoter fusions (Fig. 3a). To assess the specific effect of σ²⁸_EC and σ²⁸_Ct on transcription, GFP expression was examined at the 4-h time point with and without the addition of arabinose (Fig. 3b). Consistent with the known promoter specificities (σ⁶⁶ for P_groE and σ³² for P_EgroE), the groE promoters did not show significantly increased transcription upon induction of the σ²⁸ genes; presumably, expression was driven by chromosomally encoded σ factors in E. coli strain YK4104. GFP expression from P_hctB, P_ycgR, and P_flgK was significantly increased following the induction of both σ²⁸_Ct and σ²⁸_EC, clearly demonstrating σ²⁸ recognition of these promoters with consensus-like sequences. In contrast, GFP expression from P_tar was insignificant following the induction of σ²⁸_Ct but significant following the induction of σ²⁸_EC (Fig. 3b). Our GFP expression data mirrored our microarray data, except with P_groE, which was positive by microarray analysis but negative by GFP analysis with σ²⁸_Ct. The expression of CAT was also measured, and the levels correlated well with the levels of GFP expression (data not shown).

FIG. 3.

Measurements of GFP levels associated with different promoters. (a) Variation of GFP levels expressed from different transcriptional cat-gfp fusion plasmids in the E. coli strain coharboring plasmid-encoded σ²⁸_Ct (left) or σ²⁸_EC (right). (b) Effect of σ²⁸_Ct or σ²⁸_EC on the expression of promoter-driven GFP levels after a 4-h addition of arabinose (ara+) or without inducer (ara−). Numbers of RFUs were taken as the measurement of GFP levels. RFU = |(fluorescence intensity of test strains − fluorescence intensity of strain carrying pCG)|/|(OD₆₀₀ of test strain − OD₆₀₀ of strain carrying pCG)|, where pCG containing a promoterless cat-gfp operon was used as the intrinsic background fluorescence control. *, t tests give a P value of >0.05 (insignificant).

FIG. 4.

In vitro activities of σ²⁸-dependent promoters. (a) Promoter sequences inserted into the test vector. (b) Gel analysis of the in vitro transcription products using RNAP-σ²⁸_Ct (upper panel) or RNAP-σ²⁸_EC (lower panel). Supercoiled DNA templates used in this experiment are indicated on the top of each lane. The RNAP holoenzymes used are indicated on the left. Note that all transcripts were generated from the same starting amount of DNA template and run in parallel on the same gel. The ratios between the amounts of the test promoter and control promoter transcripts (P_test/P_fliC) are indicated and were used to normalize the data.

We performed in vitro transcription assays to test the activities of the six promoters described above (Fig. 4a) in the presence of RNAP holoenzyme but in the absence of any additional factors that might be present in whole E. coli cells. RNAP holoenzyme was made with E. coli core enzyme and recombinant σ²⁸_Ct or σ²⁸_EC. Transcription experiments were performed three or more times for each promoter; the results of a representative experiment are shown in Fig. 4b. Activity was not detected when either σ²⁸_Ct or σ²⁸_EC was omitted from the reaction mixture (not shown). To estimate the comparative in vitro activities of the test promoters, we normalized the amount of transcript generated by a test promoter to that of the P_fliC transcript. Neither chlamydial P_groE nor E. coli P_groE was stimulated by RNAP-σ²⁸_Ct or RNAP-σ²⁸_EC in the in vitro reaction (Fig. 4b). These results indicate that neither σ²⁸_Ct nor σ²⁸_EC displayed overlapping recognition of σ⁶⁶_Ct- and σ³²_EC-dependent promoters; they were consistent with results of the GFP reporter assay but were inconsistent with our observations from the microarray assay.

RNAP-σ²⁸_Ct was transcribed efficiently from chlamydial P_hctB, despite the presence of a suboptimal 12-bp spacer and two substitutions from the σ²⁸_EC consensus in the −10 region (C→T at position −14 and A→G at position −8) (Fig. 4b). Based on a P_hctB/P_fliC ratio of >1, we conclude that P_hctB is the stronger of the two promoters. RNAP-σ²⁸_EC also initiated transcription more efficiently from P_hctB than from P_fliC (Fig. 4b). E. coli P_tar was strongly activated by RNAP-σ²⁸_EC (P_tar/P_fliC ratio, >1) but weakly stimulated by RNAP-σ²⁸_Ct (P_tar/P_fliC ratio, <1). Transcription from P_ycgR, which has partial homology to the σ²⁸ −35 and −10 consensus sequences, was very weak with σ²⁸_EC and only slightly above the P_tar level with σ²⁸_Ct. Transcript initiation from P_flgK was good in the presence of RNAP-σ²⁸_Ct but was weak in the presence of RNAP-σ²⁸_EC. Our in vitro transcription studies indicate that transcripts initiating from the chlamydial P_hctB and the E. coli P_ycgR and P_flgK promoters are directly controlled by σ²⁸. The failure of σ²⁸_Ct to efficiently initiate transcription from P_tar, which includes the consensus −35 and −10 elements, suggests that that sequence, in addition to the core element, may be required for efficient σ²⁸_Ct transcription.

RNAP-σ²⁸_Ct utilizes an authentic transcription initiation site in E. coli.

In a previous study, we have demonstrated that hybrid RNAP-σ²⁸_Ct and RNAP-σ²⁸_EC initiated transcription in vitro from the E. coli fliC gene at identical transcription start sites (49). Here we address whether hybrid RNAP-σ²⁸_Ct initiates transcription in E. coli using an authentic transcription start site in the chlamydial hctB gene. The 5′ end of the mRNA derived from the hctB gene was determined by primer extension analysis of total RNA extracted from E. coli strain YK4104, which carries pLF28 (rpsD⁺) and pP_hctB::cat-gfp. When expression of σ²⁸_Ct was induced with arabinose, the 5′ end of the hctB fusion transcript was mapped to an adenine residue (103 bp upstream of the AUG codon of the cat gene) (Fig. 5). This start site coincides with the one previously identified in C. trachomatis serovar L2 (5), indicating that σ²⁸_Ct recognizes the chlamydial P_hctB promoter in E. coli, which is similar to that in Chlamydia. A similar primer extension analysis indicated that RNAP-σ²⁸_EC recognizes the same start site when initiating transcription from P_hctB (Fig. 5).

FIG. 5.

Mapping of the 5′ end of hctB mRNA. (a) Sequence of the regulatory region of the chlamydial hctB gene. (b) Primer extension analysis. Total RNA extracted from E. coli YK4104 harboring pLF28 plus pP_hctB::cat-gfp (lanes 1 and 2) and harboring pLF28 plus pP_hctB::cat-gfp (lanes 3 and 4) were used. Transcripts were detected after induction with arabinose (lanes 2 and 4) or repression with glucose (lanes 1 and 3). A DNA sequencing ladder (lanes A, C, G, and T) was prepared using pP_hctB::cat-gfp (Table 1). The transcription start site (+1) is as indicated. The −10 and −35 regions of the promoter sequence are underlined and in boldface type.

The presence of AT-rich sequences located upstream of the −35 region favors σ²⁸_Ct activation.

In silico analysis suggested that the level of σ²⁸_Ct-directed transcription might be positively associated with the AT-rich sequences upstream of the −35 element in σ²⁸_EC-dependent promoters (Fig. 2). Chlamydial P_hctB is a strong promoter (Fig. 3 and 4) and is enriched in A and T residues upstream of its −35 element. These AT-rich sequences are also present in other hctB genes from different species of Chlamydia (not shown). We therefore examined the contribution of these AT-rich sequences to σ²⁸_Ct-specific transcription in vitro using derivates of pGP_hctB that carried the P_hctB core promoter (−37 to +5 sequences) with or without its native AT-rich upstream sequences (−65 to −38) (Fig. 6a). Because this assay excludes factors other than RNAP, the resultant transcript should reflect the direct interaction between RNAP-σ²⁸ and the promoter. We constructed pGP_hctBΔUP, containing only the core promoter of P_hctB, and pGP_hctB-2, containing the sequence from −65 to −38 of the P_tar promoter (CG-rich sequences) fused to the core promoter of P_hctB (Fig. 6a). In the presence of RNAP-σ²⁸_Ct, the relative levels of transcript initiating from mutant P_hctB and P_hctB-2 were sharply reduced (P_hctB/P_fliC ratio, <1; P_hctB-2/P_fliC ratio, <1), compared to that from P_hctB (P_hctB/P_fliC ratio, >1), where native −65 to −38 sequences of hctB were present (Fig. 6b, upper panel). Diminished transcription from mutant P_hctB and P_hctB-2 was also observed when RNAP-σ²⁸_EC was used. The effects were smaller than with RNAP-σ²⁸_Ct (Fig. 6b, lower panel).

FIG. 6.

Effect of AT-rich sequences upstream of P_hctB on transcription in vitro. (a) Sequences of test promoters. Sequences were aligned according to their −35 boxes. Numbering of the upstream sequences of these promoters was done from the +1 base (in bold and italics) that was determined for the native promoter of the hctB gene. The nonnative upstream sequences are shown in lowercase letters. mut, mutant. (b to d) Gel analysis of the in vitro transcription products from promoters that differ in their upstream sequences. (b) Derivatives of P_hctB; (c) derivatives of P_tar; (d) derivatives of P_lacUV5. The RNAP holoenzymes used are indicated on the left. Plasmid templates are shown on the top. Transcription ratios are shown below each lane. CtrRNA, C. trachomatis rRNA.

To investigate whether the −65 to −38 region of P_hctB functioned independently away from its native core element, we placed these AT-rich sequences upstream of the core promoter of σ²⁸_EC-dependent P_tar to construct pGP_tar+. The hybrid promoter, P_tar+, was considerably more responsive to RNAP-σ²⁸_Ct stimulation than the native promoter, P_tar (Fig. 6c, upper panel). This observation suggests that the poor expression from native P_tar in our microarray and GFP reporter assays, in spite of the presence of a consensus core element, may be due to the absence of an upstream AT-rich sequence. Only a slight increase of transcription level stimulated by RNAP-σ²⁸_EC was observed at P_tar+ (Fig. 6c, lower panel). Similarly, we fused the sequence from −65 to −38 of P_hctB to the core promoter of P_lacUV5, which is σ⁷⁰_EC dependent and lacks AT-rich enhancer sequences (UP element) (43), to generate pGP_laUV5+. In the presence of RNAP-σ⁷⁰_EC, we found over a twofold increase in transcription from P_lacUV5+ compared to that from native P_lacUV5 (Fig. 6d), suggesting that the −65 to −38 sequences of P_hctB also had an enhancing effect within the context of core P_lacUV. It is unclear why the AT-rich hctB sequence had a greater effect on P_tar than on P_lacUV.

RNAP-σ²⁸_Ct tolerates an altered spacer length between the −10 and −35 elements.

By microarray and sequence analysis, an σ²⁸_EC promoter-like ykfB promoter with a 12-bp spacer was identified in E. coli, although the known σ²⁸_EC-dependent promoters contain 11-bp spacers (18, 40). RNAP-σ²⁸_Ct was able to initiate transcription from P_ykfB efficiently in vitro, while RNAP-σ²⁸_EC was only slightly transcribed from P_ykfB (data not shown). This suggested that σ²⁸_Ct and σ²⁸_EC may differ in recognizing promoters with an altered length of the spacer.

To test this hypothesis, we systemically analyzed how a change in spacer length affected the RNAP promoter interactions with chlamydial P_hctB that ordinarily has a 12-bp spacer between its −35 and −10 elements (58). We examined variants of pGP_hctB, where the −10 and −35 regions of P_hctB were kept constant while the lengths of spacers varied from 9 to 14 bp (Fig. 7a). RNAP-σ²⁸_Ct was able to transcribe from P_hctB variants with spacers ranging from 9 to 14 bp (Fig. 7b, upper panel). A relative increase in transcription level was observed from the promoter with an 11-bp spacer, suggesting that the optimal spacing for the recognition of P_hctB by RNAP-σ²⁸_Ct may be 11 bp, similar to what occurs with σ²⁸_EC-dependent promoters (18). Although the levels of transcription dropped using templates with 9-, 10-, 13-, and 14-bp spacers, the levels stimulated by RNAP-σ²⁸_Ct were still higher than those from P_fliC. RNAP-σ²⁸_EC transcribed variants of P_hctB with 10-, 11-, 12-, and 13-bp spacers (Fig. 7b). The level of transcription initiating from the promoter with the 11-bp spacer was similar to that using the native 12-bp spacer. An obvious decrease in transcript levels was observed using templates with 10- or 13-bp spacers. Compared to RNAP-σ²⁸_Ct, RNAP-σ²⁸_EC failed to stimulate a promoter with a larger (14-bp) or smaller (9-bp) spacer (Fig. 7b). This suggests that RNAP-σ²⁸_Ct may have greater flexibility in recognizing and stimulating a modulated promoter DNA structure.

FIG. 7.

Influence of altered spacer size on the transcription of chlamydial P_hctB in vitro. (a) Derivatives of pGP_hctB differing in spacer lengths are depicted. The wild-type −10 and the −35 regions are underlined. The short lines indicate deletion of residues. Inserted extra residues are indicated in lowercase letters. The transcription start sites at the +1 positions are italicized and boldface. (b) Gel analysis of the in vitro transcription products using RNAP-σ²⁸_Ct (upper panel) or RNAP-σ²⁸_EC (lower panel).

DISCUSSION

The σ²⁸ protein plays a pivotal role in flagellin biosynthesis and chemotaxis in E. coli and Salmonella (7, 28, 31); however, chlamydiae are not motile. In other bacteria, σ²⁸ recognizes the promoters of genes involved in many functions, including sporulation and the response to certain forms of stress, but orthologous genes are not present in Chlamydia, and the role of σ²⁸ in Chlamydia is not obvious. One chlamydial gene, hctB, has been demonstrated by in vitro transcription analysis to possess a promoter recognized by σ²⁸_Ct (58). The hctB gene is expressed late in the cycle, but many other late genes are dependent on the major σ factor σ⁶⁶ for initiation of transcription (12). The hctB gene promoter strongly resembles the E. coli σ²⁸ consensus; however, it is not clear what range of core promoter sequences and what other regulatory elements might be recognized by RNAP-σ²⁸_Ct.

In the absence of a tractable genetic system in Chlamydia, we took four experimental approaches to identify potential promoter elements that might be recognized by σ²⁸_Ct. The first approach, in which we used microarray-based transcriptional analysis, allowed us to examine the entire E. coli genome (four times the size of the chlamydial genome) for promoters recognized by σ²⁸_Ct in this heterologous system. We found 15 TUs that appeared to be activated by σ²⁸_Ct (Tables 2 and 4), nine of which were also upregulated upon the induction of σ²⁸_EC (Table 2) and six of which were not upregulated by σ²⁸_EC (Table 4). We also found 17 TUs that were upregulated by σ²⁸_EC but were not upregulated by σ²⁸_Ct (Table 3, cluster B). We next used a promoter probe vector, which isolated promoters from upstream and downstream transcripts, to confirm the promoter activities identified by microarray analysis (Fig. 3). Then, we analyzed selected promoters by in vitro transcription analysis, an approach that allowed us to determine the ability of RNAP-σ²⁸_Ct to recognize promoter sequences in the absence of other factors that might influence gene expression in E. coli (Fig. 4). Last, we carried out mutational analysis of selected promoters to determine the effects of upstream sequence on core promoter recognition (Fig. 6) and −35 to −10 spacing (Fig. 7).

We were surprised to find by microarray analysis E. coli groE and five other genes that were not upregulated by σ²⁸_EC to be upregulated by σ²⁸_Ct (Table 4). Previously, it has been reported that B. subtilis σ²⁸ shares overlapping promoter specificity with E. coli σ³² and that B. subtilis σ²⁸ can use the E. coli σ³²-dependent promoter from the rpoD gene (4). In our studies, none of these genes possess −35 and −10 elements resembling the σ²⁸_EC consensus, and our GFP and in vitro transcription analyses failed to confirm the ability of σ²⁸_Ct to initiate transcription from one of these promoters, P_groEEc. It is possible that the observed upregulation of groE was not direct but was rather a response to stress induced by our experimental system and manipulations. It is not obvious why the other genes were upregulated; most likely the upregulation of these genes (and groE) is affected by the indirect influence of multiple regulatory networks in vivo. For this reason, we did not further analyze cluster C genes.

In a recent study, we noted that σ²⁸_Ct could complement an σ²⁸_EC-deficient mutant (fliA mutant) but did not restore motility in E. coli (49). From these observations, we speculated that σ²⁸_Ct recognizes only a subset of σ²⁸_EC promoters. Our microarray analysis confirmed this suspicion. Interestingly, two of the TUs not recognized by σ²⁸_Ct are members of the class 2 flagellar regulon that requires additional transcriptional regulators for activation (7, 19, 31), factors that may not interact with RNAP-σ²⁸_Ct. In silico analysis revealed only minor differences in the core elements of the promoters recognized by σ²⁸_Ct and σ²⁸_EC (cluster A) and those recognized by σ²⁸_EC but not by σ²⁸_Ct (cluster B). Indeed, one of the promoters not recognized by σ²⁸_Ct, P_tar, possesses perfect σ²⁸ −35 and −10 elements. This suggested to us that the context of the core elements is important for the interaction of RNAP-σ²⁸_Ct with promoters. Subsequently, we found that all of the upstream sequence of the known σ²⁸_EC promoters in cluster A (Table 2 and Fig. 2) are AT rich (0.67 or higher) but that only two of the eight known σ²⁸_EC promoters in cluster B (Table 3 and Fig. 2) had an AT content above 0.60; the AT content of poorly recognized P_tar was only 0.47. This observation raised the possibility that the AT-rich upstream sequence may enhance the function of RNAP-σ²⁸_Ct.

DNA sequences upstream of −35 elements can bind to the C-terminal domain of the RNAP α subunit and are important for transcription initiation in bacteria (44, 45). These are named UP elements and may consist of a proximal and a distal subsite (11). The proximal subsite is more important in the stimulation of E. coli σ⁷⁰; the distal subsite favors selectivity of a general stress-responsive σ factor, σ³⁸ (11, 56). In Chlamydia, evidence supporting the existence of a UP element has been reported for the σ⁶⁶-specific chlamydial promoters rRNA P1 (54) and major outer membrane protein gene, omp1 (9). However, 14 of 19 known chlamydial σ⁶⁶ promoters have AT-rich sequences around the −41 regions, and two of them have further upstream AT-rich sequences (16). Thus, frequencies of UP-like sequences are likely common in Chlamydia. Whether or not the AT-rich sequence in the cluster A genes functions as an UP element per se is not known and must be experimentally confirmed. In general, these AT-rich sequences do not resemble the E. coli consensus UP element (NNNAAAWWTWTTTTNNAAAANNN) (11), nor do they resemble the UP-like element sequence (TAGGGAAGTTGAGAAAAAATAG) located immediately upstream from the chlamydial rRNA operon −35 element (53). However, at least three lines of evidence indicate a functional interaction between RNAP-σ²⁸_Ct and the AT-rich sequences upstream of the −35 element in the hctB promoter. First, deleting or exchanging the AT-rich elements with the UP-disparate sequences compromised hctB activity; second, fusing UP-like sequences to the “weak” promoters increased significantly their strength; and third, transcription analysis performed in vitro using RNAP in the absence of additional accessory factors reflected the direct interaction of RNAP holoenzyme and promoters.

The presence and influence of UP elements have been found to be more significant in some bacteria than others (17). For example, a strong influence of an UP element was observed for σ²⁸-specific promoters in B. subtilis (13), but little effect was observed on the σ²⁸-specific promoters in E. coli and Salmonella (18, 48). Chlamydial genomes contain about 59% AT (22, 42, 52), which is similar to that in B. subtilis (57%), whereas the E. coli K-12 genome has a lower (50%) AT content. It is not clear whether the difference in use of the UP element is related to the general AT contents in genomes from divergent bacterial species.

We report here that RNAP-σ²⁸_Ct can tolerate variations in the spacer length of P_hctB (Fig. 7), in contrast to what occurs with known RNAP-Eσ²⁸ promoters, which have an 11-bp spacer. It has been proposed that an optimal spacer length facilitates the spatial alignment of the −10 and −35 elements to accomplish binding to the 2.4 and 4.2 regions of the σ factor (8, 29). This may depend on the structure of the holoenzyme and vary across different organisms possessing different amino acid structures of RNAP subunits. Interestingly, promoters with different spacer lengths, such as the promoters for motA in Vibrio cholerae (9 bp) (41), the promoter for the flaA gene in Helicobacter pylori (10 bp) (21), and, here, Chlamydia P_hctB (12 bp), have been identified in other bacteria. Our observation that RNAP-σ²⁸_Ct can potentially recognize a range of “suboptimal” spacer lengths in P_hctB and yet maintain function implies a structural flexibility of the enzyme and an ability to undergo conformational transition. This could be a point of regulation, if subsets of σ²⁸_Ct promoters with different spacers are somehow better recognized under different environmental conditions.

The differences in levels of transcription activation by σ²⁸_Ct and σ²⁸_EC, as well as the failure of σ²⁸_Ct to activate cluster B TUs, may result from different σ²⁸ protein levels, different σ²⁸ core enzyme binding affinities, or other steps in the transcription cycle. Conformational changes in the σ or core enzyme or both occurring during holoenzyme formation are a consideration (29). It was shown that extension sequences at both the amino and carboxy termini of chlamydial σ⁶⁶ were responsible for modulating transcription from the chlamydial major outer membrane protein promoter (34). Amino acid alignments of σ²⁸_Ct and σ²⁸_EC indicate a lack of homology at the N terminus of domain 2.1 as well as at the linker region between domains 3 and 4, which serves as a central pillar of the σ²⁸ conformation (23, 51). Possibly, differences between σ²⁸_Ct and σ²⁸_EC might also contribute to the disparity of promoter recognition.

In summary, we have established that σ²⁸_Ct can recognize many but not all σ²⁸_EC consensus-like promoters and have found a correlation between high AT content and the ability of σ²⁸_Ct to recognize σ²⁸_EC promoters and the chlamydial hctB. We have also demonstrated that that σ²⁸_Ct tolerates a greater range of spacer distances between −35 and −10 elements. We have used these findings to search the chlamydial genome for a putative σ²⁸_Ct-controlled promoter using MotifSearch (X. Feng et al., unpublished data). Mismatches to the motif sequence were allowed in the range of 0 to 4 bases for each block and for a spacer of 10 to 14 bp. These searches generated 78 candidates, including the known hctB promoter. These putative promoter sequences are listed in Table S1 in the supplemental material. It is possible and perhaps likely that the promoter recognition properties of the σ²⁸_Ct-containing holoenzyme might differ from the promoter recognition properties of the hybrid holoenzyme used in this study. Thus, further validation with chlamydial σ²⁸_Ct-containing holoenzyme will be needed. We are now in the process of cloning these potential core elements and upstream sequences into our GFP promoter probe vector and the in vitro transcription vector to validate these sequences as σ²⁸_Ct promoters. Positive candidates will then be further analyzed by mutational analysis to establish the significance of potential upstream AT-rich elements.