Chlamydia trachomatis {sigma}28 recognizes the fliC promoter of Escherichia coli and responds to heat shock in chlamydiae

Chlamydia trachomatis is an obligate intracellular parasitic bacterium of humans that causes a wide spectrum of diseases (Schachter, 1999). As successful bacterial pathogens, chlamydiae have a unique developmental cycle, involving infection by a metabolically inert, small, spore-like elementary body (EB), followed by transition of the EB into a vegetative, larger and non-infectious reticulate body (RB), differentiation of RB back to EB and release of EB following host-cell lysis (Moulder, 1991; Hatch, 1999). Another characteristic of chlamydiae is their ability to cause persistent infection with chronic inflammation (Moulder, 1991; Beatty et al., 1994; Schachter, 1999). Varying growth conditions, such as exposure to antibiotics (Kramer & Gordon, 1971; Matsumoto & Manire, 1970), nutrition deprivation (Harper et al., 2000; Allen et al., 1985; Raulston, 1997) and presence of cytokines (Beatty et al., 1993; Shemer-Avni et al., 1989) may influence the outcome of Chlamydiahost cell interaction and lead to disruption of the transition from RB to EB, resulting in the accumulation of atypical morphological forms in vitro (Moulder, 1991; Beatty et al., 1994). It has been suggested that these forms may be capable of persisting unrecognized, contributing to the pathological changes that occur during chlamydial infections. When the stress conditions are removed, the development cycle may continue unabated. In parallel, the pattern of gene expression changes is coupled with the interconversion of morphological forms during the developmental cycle (Hatch, 1999; Beatty et al., 1994; Nicholson et al., 2003; Belland et al., 2003).

One of the important mechanisms in the switch of gene expression in bacteria is the use of alternative σ factors to alter promoter selectivity of RNA polymerase (RNAP) (Ishihama, 2000; Burgess & Anthony, 2001; Gross et al., 1998). The basic composition of chlamydial RNAP resembles that of other eubacterial RNAPs in containing subunits α₂ββ'σ (Stephens et al., 1998). Three σ genes, rpoD encoding σ⁶⁶, rpoN encoding σ⁵⁴ and rpsD (fliA or whiG) encoding σ²⁸, and several genes encoding the homologues to σ regulators RsbW, RsbU and RsbV in Bacillus subtilis, exist in the chlamydial genome (Stephens et al., 1998; Kalman et al., 1999). Chlamydial σ²⁸ is homologous to the σ²⁸ family of bacterial σ factors, including σ²⁸ in Escherichia coli and σ^B and σ^D in B. subtilis. Chlamydial RNAP containing σ⁶⁶ is capable of initiating transcription from both E. coli σ⁷⁰ consensus and non-consensus promoter sequences (Douglas & Hatch, 1995; Shen et al., 2000; Tan et al., 1996; Hatch, 1999). The major σ factor rpoD transcripts were detected at all times post-infection (p.i.), consistent with their expected function in the expression of housekeeping genes (Douglas & Hatch, 2000; Mathews et al., 1999). In contrast, transcripts of the alternative σ factors rpoN and rpsD were temporally expressed. Recently, Yu & Tan (2003) demonstrated that recombinant σ²⁸, combined with chlamydial core RNAP, transcribes the late-stage-specific histone-like protein gene hctB from a promoter that resembles the E. coli σ²⁸ consensus recognition sequence. However, information on the conditions under which the chlamydial σ²⁸ is activated and the spectrum of promoter sequences recognized by σ²⁸ remains unclear. Homologues to σ²⁸ in other organisms regulate the expression of flagellar, sporulation, stress response, type III secretion and virulence components (Stephens, 1999). Therefore, it is of great importance to identify the regulatory mechanism of gene expression in this medically important pathogen.

In the present work, we set out to characterize the biological role of the chlamydial alternative σ factor, σ²⁸. Our data show that the product of chlamydial rpsD is an E. coli σ²⁸ homologue. We also report for the first time that expression of rpsD is heat-responsive in C. trachomatis serovar F. These findings suggest that σ²⁸ may be involved in the transcription of a group of genes that are required for an adaptation response enabling the completion of the development cycle.

Organisms and growth conditions.
C. trachomatis strain F/IC-Cal-13 was propagated in mouse fibroblast L929 (ATCC CCL-1) suspensions. Infected cells were harvested at 16 or 30 h p.i. by centrifuging at 4000 r.p.m. (Sorvall GSA rotor) for 10 min at 4 °C. Purification of RB and EB were then carried out following the protocols described by Mathews et al. (1993) and Caldwell et al. (1981), respectively. E. coli strain TOP10 (Table 1) was used as a host for transformation. E. coli K-12 strain YK410 and its isogenic fliA^- mutant YK4104 (Table 1) (generously provided by Dr Robert M. Macnab, Yale University) were used as hosts for expression of the putative chlamydial σ²⁸ and E. coli σ²⁸.

Table 1. Strains and plasmids used in this study

Molecular cloning and plasmid constructions.
Plasmid DNA isolation, ligation, transformation and other DNA manipulations were carried out according to well-established procedures (Sambrook & Russell, 2001).

The plasmids and oligonucleotide primers used are listed in Table 1 and Table 2, respectively. An expression vector pBAD24H was derived from pBAD24 (Guzman et al., 1995) by insertion of a 5' terminal 6x His-tag and more cloning sites, and contained an arabinose-inducible/glucose-repressible ara gene promoter, P_BAD. Primers for the coding region of chlamydial rpsD were designed according to the published sequence of C. trachomatis serovar D (Stephens et al., 1998). The forward primer, S28pri-F, contained an NcoI site and the sequence downstream of the translation initiation codon. The reverse primer, S28pri-R, contained an SphI site, the stop codon TAA and its upstream sequence. Genomic DNA from purified EB of C. trachomatis F/IC-cal-13 was used as template for PCR. The PCR product of the chlamydial rpsD coding region was digested with NcoI and SphI, and cloned into pBADH. Using a similar strategy, the coding region of E. coli fliA was also cloned into pBADH. Oligonucleotide primers E28pri-F and E28pri-R were designed according to the published sequence of E. coli K-12 (Blattner et al., 1997) and genomic DNA from E. coli DH5α was used as template for PCR. These constructs were transformed into E. coli TOP10. The transformants were analysed by restriction mapping and nucleotide sequencing. Resultant plasmids were designated as pS28H (containing chlamydial rpsD) and pES28H (containing E. coli fliA), respectively.

Table 2. Primers used in this study

A low-copy-number expression vector, pLC3, was generated by ligating a blunted NsiI/BsrBI fragment of pBAD24 (Guzman et al., 1995) and an XmnI/AvaI fragment from pACYC184 (Chang & Cohen, 1978). This plasmid contained an araC gene, P_BAD, a ribosome-binding site, a multiple cloning site, the transcriptional terminator rrnB T1, the tetracycline resistance gene and the p15A origin of replication. NcoI/SalI fragments containing chlamydial rpsD coding sequences from pS28H and E. coli fliA coding sequence from pES28H were cloned into pLC3. The resultant plasmids were designated pLF28 and pLE28, respectively.

A number of G-less cassette-based plasmids were constructed for in vitro transcription analysis. Plasmid pMT504 (gift of Dr Ming Tan, University of California, Irvine, CA, USA) (Tan & Engel, 1996), which contained a longer promoterless G-less cassette and a shorter G-less cassette under the control of the promoter of chlamydial rrn P1, was used for cloning the P1 and P3 promoters of tuf (encoding translation elongation factor EF-Tu) (Shen et al., 2000). The P1_tuf and P3_tuf PCR products were generated using primer pairs Ptufpri-F/P1tufpri-R and PtufpriF/P3tufpriR (Table 2), digested with EcoRI and EcoRV and inserted into EcoRI and EcoRV sites of pMT504. The resultant plasmids were designated as pGP1tuf (containing chlamydial P1_tuf and P1_rrn) and pGP3tuf (containing P3_tuf and P1_rrn), respectively. Plasmid pGL was derived from pMT504 by removing the chlamydial rRNA P1 promoter and introducing a PacI site by inverse PCR using primers Gpri-F and Gpri-R. Oligonucleotide pairs, fliCpri-U and fliCpri-L, containing the promoter region of E. coli fliC (Blattner et al., 1997) were annealed and cloned into PacI and SacI sites of pGL, resulting in pGLC (Fig. 1). The promoter region of groE (Tan et al., 1996) was amplified by PCR using GroEpri-F and GroEpri-R (Table 2) and inserted into EcoRI and EcoRV sites of pGLC, resulting in pGPgroE (containing P_groE and P_fliC). Oligonucleotide pairs Psig28-U and Psig28-L (Table 2), containing the putative promoter P1 of chlamydial rpsD, were annealed and cloned into EcoRI and EcoRV sites of pGLC, resulting in pGPS1. The DNA sequences of all cloned genes were confirmed by restriction mapping and nucleotide sequencing.

Table 1). The shorter G-less cassette is under the control of PfliC. In addition, the multiple cloning sites shown on the left side of the figure were available to clone the test promoter, enabling transcription of the longer G-less cassette. The presence of two different lengths of G-less sequences (130 and 152 nt) in one plasmid make it possible to directly compare two transcripts driven by different promoters in one in vitro transcription reaction.

RNA preparation, primer extension analysis and RT-PCR assay.
The suspension culture of L929 cells infected with C. trachomatis serovar F were collected periodically. Whole infected cells were lysed immediately with TRIzol Reagent (Gibco-BRL) and RNA was prepared following the manufacturer's instructions.

Primer extension was carried out as described previously (Shen et al., 2000). Synthetic oligonucleotide primers were complemented to chlamydial rpsD, covering 3858 bp downstream sequence of the GTG translation initiation codon (fliA-pe1, Table 2) and of the G-less region of pMT504 (Gless-pe1, Table 2) by PCR. The Gel Doc system with molecular analysis software Quantity One (Bio-Rad) was used for calculation of integrated volumes of signals, which were proportional to the levels of the transcripts.

Total RNA was reverse-transcribed using AMV reverse transcriptase (Promega) and the resulting cDNA was amplified by PCR using Taq DNA polymerase (USB). Primers P1s28-F and S28pri-R2 were expected to generate a 447 bp fragment from C. trachomatis rpsD mRNA. A 302 bp fragment of 16S rRNA was amplified using primers 16sRT-F and 16sRT-R (Table 2) by PCR and used as the internal standard for relative comparison of gene expression at each time point p.i.

Western blot analysis.
Levels of σ²⁸ expression at different time points during the developmental cycle were measured by Western blot analysis (Sambrook & Russell, 2001). Briefly, infected cells were collected by centrifugation of aliquots of suspension culture at 2800 r.p.m. (Centra GP8R; IEC) for 10 min at 35 °C. Cells were gently resuspended in 0·1 M PBS (pH 7·4) and sonicated briefly to break cell membranes. Chlamydial EB and RB were collected by microcentrifuge at 15 000 r.p.m. for 20 min, quickly resuspended in distilled water, followed by the immediate addition of 2x loading buffer and boiled for 10 min. After centrifugation for 10 min at 15 000 r.p.m. at 4 °C, the supernatant fluids were carefully collected. Uninfected cells served as negative controls. Following separation by SDS-PAGE, proteins were transferred onto Immobilon-P (Millipore). Anti-RpsD, a polyclonal antibody directed against the putative chlamydial σ²⁸ of C. trachomatis was generously provided by Dr Thomas P. Hatch (University of Tennessee, USA) and was used to detect chlamydial σ²⁸. GP-45, an mAb against chlamydial EF-Tu (Zhang et al., 1994) was used to probe EF-Tu in companion samples or its dilutions. Blots were developed by using mouse anti-rabbit immunoglobulin G-horseradish peroxidase (Sigma) and a SuperSignal Chemiluminescent Detection kit (Pierce).

Overexpression of rpsD genes in E. coli and purification of the recombinant proteins.
E. coli YK4104 (Table 1) carrying plasmids pS28H or pES28H was grown at 30 °C in LB containing 100 µg ampicillin ml^-1, induced with 0·002 % L-arabinose in the exponential growth phase (OD₅₉₅=0·50·7) for 2 h and then harvested by centrifugation. Three grams of cell pellet were resuspended in 20 ml buffer A (50 mM Tris/HCl, pH 8·0, 1 mM EDTA, 5 mM DTT, 1 mM PMSF). The 6x His-tagged recombinant protein was purified by metal-chelate affinity chromatography as described previously (Zhang et al., 1997) and dialysed against buffer B (20 mM Tris/HCl, pH 8·0, 0·2 M NaCl, 0·5 mM DTT, 0·1 mM EDTA, pH 8·0, 5 % glycerol), followed by a HiTrap HP Q Column (Amersham Pharmacia Biotech). After washing the column with buffer B, the proteins were eluted with a linear gradient of NaCl (0·21 M) in buffer B. Purified protein fractions were monitored by SDS-PAGE, pooled, dialysed against a storage buffer (10 mM Tris/HCl, pH 8·0 at 4 °C, 10 mM MgCl₂ 0·1 mM EDTA, 0·2 M NaCl, 50 % glycerol, 1 mM DTT, 0·1 % Triton X-100) and stored at -20 °C until use.

In vitro transcription analysis.
An RNAP holoenzyme containing chlamydial σ²⁸ was reconstituted by adding 1 unit E. coli core RNAP (Epicentre Technologies) to 4·0 pmol recombinant chlamydial σ²⁸ in protein dilution buffer (10 mM Tris/HCl, pH 8·0, 10 mM NaCl, 1 mM DTT, 0·1 mM EDTA, pH 8·0, 0·4 mg bovine serum albumin ml^-1, 0·1 % Triton X-100) and incubating on ice for 15 min. Similarly, a holoenzyme containing E. coli σ²⁸ was reconstituted using E. coli core and recombinant σ²⁸. E. coli RNAP holoenzyme saturated with σ⁷⁰ (Eσ⁷⁰) was obtained from Epicentre Technologies. In vitro transcription was performed in a 10 µl reaction volume containing 1 µg G-less supercoiled template plasmid, 2 µl holoenzyme, 400 µM ATP, 400 µM UTP, 1·2 µM CTP, 0·20 µM [α-³²P]CTP, 100 µM 3'-O-methylguanosine 5'-triphosphate (Amersham Pharmacia Biotech) and 20 U RNase inhibitor, following the conditions described by Tan & Engel (1996). One volume of gel loading buffer containing 95 % (v/v) formamide, 0·025 % (w/v) xylene cyanol, 0·025 % (w/v) bromophenol blue and 0·5 mM EDTA (pH 8·0), was then added to the mixture to stop the reaction. The products were run on a 6 % polyacrylamide/8 M urea gel as described by Sambrook & Russell (2001) and gels were subjected to autoradiography. Transcripts were quantified by determining the integrated volume of signal using the Gel Doc system with molecular analysis software Quantity One (Bio-Rad).

Genetic complementation.
E. coli K-12 strain YK410 is wild-type for motility and chemotaxis, while its isogenic fliA^- mutant YK4104 (Table 1) lacks this phenotype (Komeda et al., 1980). YK4104 was used as host for expression of the putative chlamydial σ²⁸ and E. coli σ²⁸. Plasmids pLF28 (carrying chlamydial rpsD), pLE28 (carrying E. coli fliA) and pLC3 (the vector control) were transformed into YK4104, respectively. Motility of the transformants was tested on 0·35 % agar (10 g Bacto tryptone, 3 g Bacto agar, 5 g NaCl in 1 l water) plates using swarm assays. Swarm assays were performed by stabbing fresh bacteria onto a semisolid agar plate and incubating the plates at 30 °C for 8 h. Motility was assessed by examining the circular swarm formed by the growing motile bacterial cells.

Identification of chlamydial rpsD transcriptional initiation sites
As the first step to understanding the regulatory mechanisms involved in the expression of rpsD, we identified the 5' ends of transcripts at 16 h p.i., when inclusions contained mostly RB, and again at 30 h p.i., when there was a relatively high population of EB. Primer extension analysis revealed two 5' ends, indicating two possible transcription start sites at 18 nt (transcript I) and 54 nt (transcript II) upstream of the translation initiation codon GTG of rpsD (Fig. 2a and b). The putative promoters for these transcripts were designated P1 and P2, respectively. A sequence located around the -35 region of P1 (TAAA) is identical to the -35 region of the consensus sequence of the E. coli σ²⁸-dependent promoter. However, P1 has only 3 bp identical to the -10 consensus sequence of the E. coli σ²⁸-dependent promoter (GCCTTATT vs GCCGATAA) (Fig. 2a). The sequence organization for the distal promoter, P2, was not evident based on homology to known σ-recognition sequences.

Table 2). DNA sequence ladders (GATC) of pXE2 (Table 1), generated with the same primer, are shown next to the primer extension reaction. (c) RT-PCR analysis of gene transcript levels in chlamydiae treated with heat shock. Gels were stained with ethidium bromide to ascertain levels of transcripts from a gene encoding 16S rRNA (equivalent at each time point before or after heat shock) and rpsD (upregulated by heat shock at 16 and 30 h p.i.)

Chlamydial rpsD is responsive to heat shock
From the intensity of primer extension signals, both transcripts appeared to be poorly expressed at 30 h p.i. when chlamydiae were grown at 35 °C and not subjected to heat shock; transcript II, but not transcript I, was easily detected at 16 h p.i. under these conditions. To explore the potential contribution of rpsD to environmental cues, we tested the expression of rpsD in C. trachomatis serovar F after shifting the temperature from 35 to 42 °C. After 10 min at 42 °C, the levels of transcript I increased significantly at both 16 and 30 h p.i. In contrast, levels of transcript II increased only slightly after heat shock. The ratio of transcript I to transcript II changed from less than one at 35 °C to greater than one after heat shock at 42 °C (Fig. 2b). To explore whether transcription of rpsD varied in aberrant RB, we isolated RNA from infected cells incubated in the presence of benzylpenicillin (100 µg ml^-1). Penicillin treatment has been shown to interrupt the transition from RB to EB, resulting in the accumulation of large aberrant RB in the inclusions (Kramer & Gordon, 1971; Matsumoto & Manire, 1970). However, we found it had no visible effect on transcription of rpsD in these experiments (data not shown).

Expression of rpsD transcript I was further examined by an RT-PCR assay of total RNA extracted from normal and heat-shocked cells. 16S rRNA was used as the internal standard for relative comparison of gene expression. As shown in Fig. 2(c), the levels of control transcript were almost equal at each time point p.i., enabling direct comparison of rpsD transcript I levels between normal and heat-shocked cells. Consistent with the result of primer extension analysis, rpsD transcript I apparently increased on temperature upshifting from 35 to 42 °C at both 16 and 30 h p.i., but the most significant change occurred at 16 h p.i. The finding that σ²⁸ was upregulated as a result of heat shock suggested that it might play a role in the chlamydial stress response.

Expression of chlamydial σ²⁸ levels during the developmental cycle
The levels of σ²⁸ protein in infected cells harvested at different time points p.i. were measured by Western blot analysis using anti-RpsD. We examined the amounts of sequentially appearing protein in an equal number of infected cells (Fig. 3a). A single immunoreactive band that migrated at approximately 31 kDa (the predicated relative molecular mass of the chlamydial putative σ²⁸ is 28·9 kDa) was observed in lysates of infected cells harvested after 24 h p.i.; the band persisted at subsequent time points, but was not present in lysates harvested at 16 h p.i. and earlier (Fig. 3a). The slow migration of σ²⁸ seen on SDS-PAGE was similar to E. coli σ²⁸, which has an anomalous electrophoretic motility because of its acidic nature (Kundu et al., 1997; Liu & Matsumura, 1995). In a separate experiment shown in Fig. 3(b), we normalized σ²⁸ levels against the level of EF-Tu, which is essential for protein biosynthesis and constitutively expressed (Nicholson et al., 2003), to compensate for the increase in numbers of chlamydiae as the infection progressed. Whereas the ratio of σ²⁸ to EF-Tu appeared nearly constant at 1628 h p.i., it declined after 28 h p.i. A common observation made in these experiments was that, relative to EF-Tu, the level of σ²⁸ decreased after 28 h p.i. and remained detectable at later times p.i.

(59K): Fig. 3. Evaluation of expression levels of chlamydial σ28 by Western blot analysis. Expressed chlamydial σ28 and EF-TU levels in whole infected cells were disclosed with polyclonal anti-RpsD and mAb GP-45 using duplicate samples. (a) Sequentially appearing chlamydial σ28 and EF-Tu at different time points. A constant number of infected cells was loaded to each lane. (b) Normalizing the levels of chlamydial σ28 to EF-Tu at mid to later stages of the developmental cycle. Single normalization was calculated as the ratio of the mean value of integrated volume of signal for chlamydial σ28 to those for EF-Tu in the companion sample. Note that the 16 h lane was loaded with a larger number of cells to bring the EF-Tu level up a level similar to that in other lanes.

Promoter-specific transcription activities of RNAP containing the recombinant chlamydial σ²⁸
Chlamydial rpsD was cloned and expressed in an E. coli fliA^- mutant, which does not make E. coli σ²⁸. His-tagged recombinant chlamydial σ²⁸ and recombinant E. coli σ²⁸ (made in a strain expressing fliA) were purified as described in Methods; expression and purification of recombinant σ²⁸ are shown in Fig. 4. Consistent with the slow electrophoretic motility of native chlamydial σ²⁸, His-tagged recombinant chlamydial putative σ²⁸ ran at 32 kDa on SDS-PAGE, though the predicated mass is 30 kDa.

(52K): Fig. 4. Overproduction and purification of the chlamydial rpsD gene product in E. coli. (a) Coomassie brilliant blue-stained SDS-PAGE profile. (b) Western blots of recombinant 6x His-tagged chlamydial σ28. Lanes: 1 and 4, whole-cell lysates of uninduced E. coli YK4104 (isogenic fliA- mutant of E. coli K-12 strain YK410) transformed with pS28H; 2 and 5, whole-cell lysates of induced E. coli YK4104 transformed with pS28H; 3 and 6, the combined pool of purified protein. The induced expression bands are marked with *.

The activity of the chlamydial σ²⁸ was examined by in vitro transcription with E. coli RNAP core enzyme (E) reconstituted with the His-tagged recombinant chlamydial σ²⁸ (Eσ²⁸) and the G-less plasmid template containing a well-characterized E. coli σ²⁸-dependent promoter, P_fliC (Kundu et al., 1997; Liu & Matsumura, 1995; Macnab, 1996). Several reasons encouraged us to study the functions of the chlamydial σ²⁸ using this system: (1) the inability to undertake direct genetic investigations in C. trachomatis; (2) the high degree of sequence identity between the RNAP core subunits of C. trachomatis and other eubacterial core RNAP subunits; (3) sequence similarity of the main functional domains between putative chlamydial σ²⁸ and E. coli σ²⁸; and (4) the ability of the chlamydial ompA P1 promoter to be recognized by chlamydial σ⁶⁶ using an E. coli gene expression system (Mathews & Stephens, 1999).

A fixed amount (1 µg) of supercoiled DNA template pGLC, which contains the fliC promoter from E. coli, and the 2 µl of reconstituted RNAP holoenzyme (1 unit of E. coli core enzyme reconstituted with increasing amounts of recombinant chlamydial σ²⁸) were used in each reaction of the in vitro transcription assay. As shown in Fig. 5(a), Eσ²⁸ was able to initiate transcription from P_fliC and generated a single transcript of 130 bp. This represents the σ²⁸-driven transcription for the following reasons. First, recombinant σ²⁸ was expressed in E. coli YK4104, which lacks the functional σ²⁸. Thus, the potential artefact caused by contamination of E. coli σ²⁸ could be excluded. Second, E. coli core RNAP provided no signal in the absence of σ²⁸ (lane 1). Third, the level of transcription increased with increasing amounts of σ²⁸ (Fig. 5b). Fourth, there was no 130 nt product when vector control pGL was used (data not shown). Finally, there was no transcription when Eσ⁷⁰ was used (Fig. 6).

(24K): Fig. 5. In vitro transcription activity of RNAP holoenzyme-reconstituted core enzyme of E. coli with recombinant σ28. (a) Autoradiogram obtained from the transcription assay using a constant amount of reconstituted E. coli core enzyme with increasing amounts of His-tagged recombinant σ28. Lanes: 1, E. coli core enzyme alone; 2, recombinant σ28 alone; 37, 0·5, 1·0, 2·0, 4·0 and 6·0 pmol chlamydial σ28 and E. coli core RNAP. (b) Curve generated from densitometry tracings of autoradiograph shown in (a).

(70K): Fig. 6. Promoter recognition specificity of holoenzyme-containing recombinant chlamydial σ28 in vitro. (a) Sequences of promoters. The consensus of -35, separation and -10 regions of E. coli σ28 and σ70 promoters are listed on the top of each group in bold type. PfliC, promoter of E. coli fliC; PgroE, promoter of chlamydial groE; P1rrn, promoter 1 of chlamydial rrn; P1tuf and P3tuf, promoters of chlamydial tuf. PfliC is a well-characterized E. coli σ28-dependent promoter; PgroE, P1rrn, P1tuf and P3tuf are recognized by chlamydial σ66 and E. coli σ70 (Tan et al., 1996; Shen et al., 2000). Sequences of P3tuf are identical to the consensus E. coli σ70 promoter sequence; PgroE, P1rrn and P1tuf present slight variation. Plasmids containing these promoters are listed in the right column. (b) Runoff transcription profile. The 152 nt transcripts were derived from PgroE in pGPgroE, P1tuf in pGP1tuf and P3tuf in pGP3tuf, and the 130 nt transcripts were derived from E. coli PfliC (in pGLC and pGPgroE) or chlamydial P1rrn (in pMT504, pGP1tuf and pGP3tuf). The template plasmids used are indicated at the top of the lanes. The holoenzymes used in the reactions are shown on the left. (c) Comparison of Eσ70 and Eσ28 transcription using pGPgroE as template, which contains the σ70-consensus-like PgroE and the E. coli σ28-dependent fliC promoter. Lanes 1, Eσ28 alone; 2, both Eσ70 and Eσ28; 3, Eσ70 alone.

When RNAP holoenzyme containing the recombinant E. coli σ²⁸ was used as a positive control, the levels of transcripts initiating from P_fliC were higher than those from RNAP combined with chlamydial σ²⁸ (data not shown). The 5' ends of the fliC mRNA that were transcribed by either chlamydial Eσ²⁸ or E. coli Eσ²⁸ mapped to the same adenosine by primer extension analysis (data not shown). These data indicate that the recombinant chlamydial σ²⁸ recognizes the E. coli fliC promoter; therefore, it could be a homologue of E. coli σ²⁸.

To examine the specificity of recombinant chlamydial σ²⁸ on promoter recognition, templates containing four chlamydial promoters with high similarity to σ⁷⁰ consensus sequences were tested using the in vitro transcription assay consisting of either Eσ⁷⁰ or Eσ²⁸, or both, and compared with templates containing P_fliC (Fig. 6a). Templates pGP1tuf, pGP3tuf, pMT504 and pGPgroE are predicted to produce the following σ⁷⁰-dependent transcripts: pMT504, 130 nt (from P1_rrn); pGP1tuf, 130 and 152 nt (from P1rnn and P1_tuf); pGP3tuf, 130 and 152 nt (from P1rnn and P3_tuf); and pGPgroE, 130 nt (from P_groE). In addition, templates pGLC and pGPgroE are predicted to produce 130 nt transcripts from their fliC promoters in the presence of Eσ²⁸. When the reconstituted RNAPs were tested individually, transcripts of the predicted size were produced, including the 130 nt σ²⁸-dependent transcripts when σ²⁸ was present (Fig. 6b). Most significantly, the pGPgroE template, which contains the σ⁷⁰ consensus-like groE promoter and the E. coli σ²⁸-dependent fliC promoter, yielded the 152 nt σ⁷⁰ transcript and the 130 nt σ²⁸ transcript when σ⁷⁰ and σ²⁸ were simultaneously present in the in vitro transcription reaction (Fig. 6c). This analysis further confirmed that Eσ²⁸ and Eσ⁷⁰ only recognize their own promoter sequences in the in vitro transcription assay, providing evidence that there is no cross-activity on promoter recognition between chlamydial housekeeping σ factor σ⁶⁶ and the alternative σ factor σ²⁸.

The putative promoter P1 of chlamydial rpsD was also tested using the in vitro transcription assay. pGPS1 was used as a template in which chlamydial P1_rpsD served as the test promoter and P_fliC served as the control promoter. Consistent with our observations above, both chlamydial Eσ²⁸ and E. coli Eσ²⁸ initiated transcription from P_fliC; however, there were no detectable transcripts generated with P1_rpsD (data not known). This suggested that the putative P1 of the rpsD, although it bears some similarity to the E. coli σ²⁸ consensus (Fig. 2a), may not be recognized by chlamydial σ²⁸ or that an additional molecule(s) is required for the efficient transcription.

Expression of chlamydial σ²⁸ was unable to restore motility to an E. coli fliA^- mutant
The similarity of promoter specificity and selectivity between chlamydial σ²⁸ and E. coli σ²⁸ prompted us to perform complementation experiments in an E. coli fliA^- mutant, which is non-motile because it does not express several σ²⁸-dependent genes required for motility, with the corresponding cloned rpsD gene from C. trachomatis. Plasmids pLF28, containing the coding region of chlamydial rpsD, pLE28, containing the coding region of E. coli fliA, and pLC3, the vector control, were transformed into E. coli fliA strain YK4104, respectively. Expression of recombinant chlamydial σ²⁸ and E. coli σ²⁸ in YK4104 under arabinose induction were confirmed by Western blot analysis (data not shown), indicating that the system was functional. On semisolid agar, wild-type strain YK410 formed a migrating ring of bacteria indicative of flagellar activity, whereas strain YK4104 lacking functional σ²⁸ failed to form a migrating ring. E. coli YK4104 harbouring pLF28 was unable to form a migrating ring on both arabinose-containing and arabinose-free semisolid agar. In contrast, E. coli YK4104 harbouring pLE28 was able to complement the fliA defect and to restore motility on arabinose-containing semisolid agar (Fig. 7a and b), but not on arabinose-free semisolid agar. In addition, there was no observable effect of plasmid pLC3 on the motility of the tested bacterium (data not shown). These results indicated that, unlike E. coli σ²⁸, the expression of chlamydial σ²⁸ could not restore the defect in motility of E. coli mutant.

(64K): Fig. 7. Detection of motility phenotype of E. coli fliA mutants using a swarm assay. Fresh bacteria were stabbed into semisolid LB agar and incubated at 30 °C for 8 h. (a) Semisolid agar. (b) Semisolid agar containing 0·002 % (w/v) L-arabinose. YK410, wild-type E. coli; YK4104, an isogenic fliA- mutant of YK410; YK4104(pLE28), YK4104 containing E. coli fliA in plasmid pLE28; YK4104(pLF28), YK4104 containing chlamydial rpsD in plasmid pLF28. pLE28 contains E. coli fliA. pLF28 contains chlamydial rpsD.

Previously published data (Mathews et al., 1999; Douglas & Hatch, 2000) showed that by RT-PCR, unlike the chlamydial major σ factor gene rpoD, the alternative σ factor genes, rpoN and rpsD, were expressed at low levels when EB converted to RB, but were easily detected throughout the exponential growth phase of RBs and later time points. In the present study we examined expression of rpsD transcripts by primer extension analysis and identified the putative transcription initiation sites. In addition, Western blot analysis showed that, relative to EF-Tu, a molecular switch of protein synthesis (Miller & Weissbach, 1977) and chaperone-like protein (Caldas et al., 1998), chlamydial σ²⁸ was present a higher levels during the exponential growth phase than at very late times in the developmental cycle. On the basis of low levels of σ²⁸ at early time points, it is unlikely that σ²⁸ plays a key role in the transition from EB to RB. More significantly, σ²⁸ is present late in the middle stage of the developmental cycle, a time that coincides with the programmed transition from metabolically active RB to inert EB. We reasoned that σ²⁸ might play an important role in regulating genes that are required for completion of the developmental cycle. This supposition is consistent with the recent observations of Yu & Tan (2003) who demonstrated that the promoter of the chlamydial late-stage hctB gene is recognized by σ²⁸ in vitro. It is likely that numerous signals, such as developmental signals, the density of RB in inclusions and nutritional conditions trigger the alternation of gene expression. However, it should be noted that σ²⁸ detected by Western blotting may not represent functional σ²⁸.

At this point, it is not clear how chlamydial σ²⁸ binds to RNAP core enzyme and how it interacts with the regulatory proteins affecting σ²⁸ activation. In E. coli and Salmonella, while fliA is itself transcriptionally regulated (Liu & Matsumura, 1996; Ikebse et al., 1999), it is also post-translationally controlled by anti-σ factor FlgM, which binds σ²⁸ and blocks transcription initiation (Hughes & Mathee, 1998), and by protease digestion (Tomoyasu et al., 2002). Chlamydial σ²⁸ is also homologous to B. subtilis σ^B. The activity of σ^B is inhibited when it interacts with anti-σ factor RsbW, the activity of which is controlled by RsbV and RsbU. Homologues of all of these σ regulators are encoded by the chlamydial genome. Therefore, σ²⁸ may be present but not functional under a given environmental condition, if it is bound to the anti-σ factor RsbW.

Our results also showed that expression of chlamydial rpsD was induced by heat shock. The chlamydial genome lacks genes with homology to the heat-shock regulator σ^H (σ³²) and σ^E (σ²⁴) of E. coli. The sequences located upstream of known heat-shock-regulated genes, dnaK and groE, contain chlamydial σ⁶⁶ promoters, which are homologous with the consensus of σ⁷⁰ in E. coli (Tan et al., 1996). An alternative regulatory mechanism used by chlamydiae involves repression: the HrcA repressor acts at a cis-acting regulatory element (controlling inverted repeat of chaperone expression, CIRCE) to repress the transcription of dnaK and groE (Wilson & Tan, 2002). However, no obvious CIRCE is present upstream of other heat-induced genes based on sequence homology searches. At present, regulatory mechanisms that respond to universal stresses in addition to heat shock, such as nutrition deprivation, phase of growth, high osmolarity and acid stress, that chlamydiae may encounter during infection are unidentified. They are mediated by σ^B in B. subtilis (Vicente et al., 1999; Dufour et al., 1996) and σ^S in E. coli (Hengge-Aronis, 1999). Under stress conditions or upon entering into stationary phase, they are synthesized and activated and can then express numerous genes required to protect cells from stress (Vicente et al., 1999; Helmann et al., 2001; Yura & Nakahigashi, 1999). Our findings, along with the homology between chlamydial σ²⁸ and σ^B of B. subtilis, suggest that the function of chlamydial σ²⁸ may be associated with adaptation to stress, which may provide enhanced resistance to changing environments. On the basis of in vitro transcription analysis, σ²⁸ may activate a set of genes in a mechanism different from that of σ⁶⁶ because σ²⁸ was unable to transcribe from the well-defined σ⁶⁶-recognized promoters, including heat-responsive P_groE. Recently, chlamydial σ²⁸ was shown to bind chlamydial putative RbsW, a homologue of the stress-responsive σ^B regulator in B. subtilis (A. L. Douglas & T. P. Hatch, personal communication), suggesting that interaction of chlamydial σ²⁸ and its putative regulators may be involved in chlamydial σ²⁸-specific transcription. The possibility that the interplay of σ²⁸ with other σ factors or additional regulatory factors may mediate rpsD activation in response to environmental stresses remains to be examined.

We further demonstrated that the holoenzyme containing chlamydial σ²⁸ was able to direct E. coli RNAP to initiate transcription from a σ²⁸-dependent promoter, P_fliC. The overall amino acid identity between chlamydial σ²⁸ and E. coli σ²⁸ is 34·7 %. The ability of chlamydial σ²⁸ to function with core E. coli RNAP in vitro is consistent with the high homology within regions 2.1, 2.2 and 3.2, which are implicated in core-binding (Burgess & Anthony, 2001; Lonetto et al., 1992). The conserved amino acid sequence at subregions 2.4 and 4.2 that are known to be involved in the recognition of promoter -10 and -35 sequences (Lonetto et al., 1992), respectively, may account for the similarity of promoter recognition between chlamydial Eσ²⁸ and E. coli Eσ²⁸. Although transcripts of P_fliC were initiated from the same start sites using chlamydial and E. coli Eσ²⁸, the transcription signal initiated by chlamydial holoenzyme was weaker. Sequence analysis of chlamydial σ²⁸ also revealed some important differences from the E. coli homologue. For instance, they lack homology in the N-terminal region 1.2 sequence, which can affect promoter binding, open complex and initiation complex formation, and the transition from abortive transcription to elongation (Nicole & Dombroski, 2001). The differences between the gene product of C. trachomatis rpsD and E. coli fliA may allow chlamydial σ²⁸ to recognize promoters with more divergent sequences. Complementation studies showed that the expression of chlamydial σ²⁸ was unable to restore motility in E. coli lacking its own functional σ²⁸. Possible explanations for this include the inability of chlamydial σ²⁸ to transcribe all the σ²⁸-dependent genes that are required for motility in E. coli, or expression of chlamydial σ²⁸ might indirectly affect multicellular processes that are required for swarming motility (Fraser & Hughes, 1999).

Our observation that chlamydial rpsD expresses two steady-state transcripts, which were differentially regulated upon heat shock, suggested that each of them might be derived from different promoters or regulated by different mechanisms. In E. coli, the fliA operon belongs to class 2 in the transcriptional hierarchy of flagellar genes and can be transcribed by Eσ⁷⁰ in the presence of activator FlhD/C, and by Eσ²⁸ (Macnab, 1996; Liu & Matsumura, 1996). It would be interesting to determine if the transcription of rpsD itself is dependent upon σ²⁸ and regulated by other transcription factors. The putative promoter P1 of chlamydial rpsD was not recognized by holoenzymes containing either chlamydial σ²⁸ or E. coli σ²⁸ and σ⁷⁰ in vitro using the G-less plasmid, which is designed to test the initiation of transcription at a specific site. Corroboration experiments are needed to clarify the activities of the putative promoters using alternative approaches.

The findings of this study suggest that chlamydial σ²⁸ may participate in transcription of a group of genes that are involved in the adaptive response of chlamydiae to environmental conditions. Having a relatively small genome, Chlamydia species may employ a complex strategy for expression of developmental genes through the participation of σ factors and their regulators (Mathews et al., 1999; Douglas & Hatch, 2000; Nicholson et al., 2003). Such a regulatory strategy may reflect the ability of Chlamydia to respond to stressful conditions with economy and efficiency.

We thank Drs Thomas P. Hatch and Annemarie L. Douglas for providing polyclonal antibody against chlamydial putative σ²⁸, anti-RpsD, and for helpful discussions; Dr Ming Tan for providing the G-less-based plasmid and as a consultant on the in vitro transcription assay; Dr Robert M. Macnab for providing E. coli strains. Also, we would like to thank Drs Peter A. Rice, Monty Montano, Robin R. Ingalls and Sanjay Ram for reading the manuscript and for their suggestions. This work was supported by National Institutes of Health grants AI38515.