Quantifying promoter activity during the developmental cycle of Chlamydia trachomatis

Chlamydia trachomatis is an important human pathogen that undergoes a characteristic development cycle correlating with stage-specific gene expression profiles. Taking advantage of recent developments in the genetic transformation in C. trachomatis, we constructed a versatile green fluorescent protein (GFP) reporter system to study the development-dependent function of C. trachomatis promoters in an attempt to elucidate the mechanism that controls C. trachomatis adaptability. We validated the use of the GFP reporter system by visualizing the activity of an early euo gene promoter. Additionally, we uncovered a new ompA promoter, which we named P3, utilizing the GFP reporter system combined with 5′ rapid amplification of cDNA ends (RACE), in vitro transcription assays, real-time quantitative RT-PCR (RT-qPCR), and flow cytometry. Mutagenesis of the P3 region verifies that P3 is a new class of C. trachomatis σ66-dependent promoter, which requires an extended −10 TGn motif for transcription. These results corroborate complex developmentally controlled ompA expression in C. trachomatis. The exploitation of genetically labeled C. trachomatis organisms with P3-driven GFP allows for the observation of changes in ompA expression in response to developmental signals. The results of this study could be used to complement previous findings and to advance understanding of C. trachomatis genetic expression.

promoter activity in a desirable in situ setting, we constructed a transcriptional reporter vector, pPvGFP::SW2 (Fig. 1a) using the backbone of the Escherichia coli and C. trachomatis shuttle plasmid, pGFP::SW2 18 . Plasmid pPvGFP::SW2 contained (i) a core promoter region (− 38 to + 6 relative to the transcription start site (TSS) + 1) from the C. trachomatis euo gene 26 (P euo ) that is flanked with a multiple cloning site (MCS) to facilitate cloning and to validate the reporter vector, (ii) a 34 bps region containing a ribosome binding site (RBS) from the C. trachomatis tuf gene 11 , which codes for translation elongation factor EF-Tu at the 5′ end of the gfp gene. Previous studies showed that tuf RBS functioned in both C. trachomatis and in E. coli 10,11 , ensuring translation of reporter genes. The plasmid also contained the bla gene encoding β -lactamases for positive selection in C. trachomatis. The plasmid pPvGFP::SW2 was transformed into C. trachomatis L2/25667R, a naturally occurring plasmid-free strain. After four rounds of selection with ampicillin, GFP-expressing inclusions were observed by fluorescent microscopy. A clone of the transformed strain, named L2/pPvGFP::SW2, was used to infect HeLa cells. GFP-expressing inclusions appeared at 16 hours post-infection (h pi) (Fig. 1b). The levels of GFP increased proportionately to the expansion of C. trachomatis inclusions and accumulation of organisms. The control, transformed with the promoter-less plasmid, pPLGFP::SW2, did not express GFP. This data indicates that GFP expression is specifically driven by P euo .
To facilitate GFP quantitation in C. trachomatis using an internal protein control, a DNA fragment containing MCS-P euo -tuf RBS from pPvGFP::SW2 was subcloned into pBOMB4-tet-mCherry 20 (hereafter called pBOMBm), which contained the fluorescent protein mCherry gene driven by a tetracycline controlled promoter (P tet ). The resultant plasmid, pBOMB-P euo , was transformed into L2/25667R cells to generate L2/pBOMB-P euo . Infection of L2/pBOMB-P euo in HeLa cells formed GFP-expressing inclusions at 16 h pi and beyond, a phenotype similar to that of L2/pPvGFP::SW2. P tet -driven mCherry expression was induced in the presence of anhydrotetracycline hydrochloride (α TC) (Fig. 1c). Thus, the level of mCherry expressed from the same plasmid provided a control to normalize the levels of GFP expression. Previous studies indicated that the euo gene mRNA was moderately expressed as early as 2 h pi from Chlamydia spp. 6,26 . The failure to detect P euo -driven GFP expression prior to 16 h pi by fluorescence microscopy was perhaps due to the small inclusion size and low bacterial accumulation. It is also possible that the core P euo region used in the construct is intrinsically weak and (an) additional DNA element(s) upstream or downstream may be required for its optimal expression in C. trachomatis. Nevertheless, our results indicate that GFP can be produced under the control of P euo and the tuf translation machinery ribosome complex.
Prediction of ompA P3. C. trachomatis surface-exposed MOMP, encoded by the ompA gene, is a critical component that directly mediates pathogen-host interactions 25 . Developmentally regulated ompA transcription has been a topic of extensive study 12,14,27,28 . Although ompA is expressed as multiple mRNAs, only the P1-derived short transcript and the P2-derived long transcript have been studied in detail 12,14,29 . To quantify the effects of DNA regions on ompA promoter activity, we created a set of lacZ transcriptional fusions in E. coli based on Whipple's system 30 . The regions carrying the promoters (either combined P2 and P1, or P1 or P2 alone) were cloned into the lacZ expression vector, pFW11, and the promoter-lacZ fusion was relocated to an F′ episome in the recipient cells as detailed in the Materials and Methods section. The resultant strains, designated FW/P21, FW/P1, or FW/P2 (Fig. 2a), were subjected to β -galactosidase (β -gal) assays to measure the activity of ompA promoters. We observed a 6-fold increase in β -gal activity in FW/P21 cells compared to the promoter-less control (Fig. 2b). However, no β -gal levels exceeding that of the promoter-less control were produced in FW/P2 or FW/ P1. These results are comparable to previous studies using multi-copy plasmid-encoded chloramphenicol acetyltransferase (CAT) as a reporter in E. coli 12 . Although the activity of P2 has been authenticated in C. trachomatis, the role of P1 is still under debate 12,29 . It is unsurprising that P2 is not active in E. coli because P2 has a GC-rich − 10 hexamer of TATCGC, which is not recognized by E. coli σ 70 in vitro 29 . The high levels of β -gal observed in strain FW/P21 could therefore not be explained by the lack of β -gal activity in strains FW/P1 and FW/P2. We suspected that an additional promoter recognized by E. coli transcription machinery must be present between the P2 and P1 regions.
We next searched for the putative promoter (s) from the 607 bp intergenic sequences (IGSs) upstream of the ompA of C. trachomatis strain L2/434/Bu in silico using a dynamic program-based MotifSearch tool 10 . A two-block-motif, TTAACA (−35)-n 16-19 -TATAAT (−10), was used for C. trachomatis σ 66 recognition sequences 31 and motif TAAAGTTT (−35)-n 10-14 -GTTGACAA (−10) was used for σ 28 recognition sequences 10 . Up to 3 base-pair mismatches (a total 6 bases) were allowed in the − 35 hexamer or −10 hexamer. These predictions resulted in the identification of the known ompA P2 region and a region containing the sequence TTACGA-n 17 -TATGGT, which we named P3 (Fig. 2c). The same P3 region was predicted using a position weight matrix-based program called Footy 32 . P1 was excluded, as it contained mismatches in five of six positions in the putative − 35 hexamer and in three of six positions in the −10 hexamer. No σ 28 recognition sequences were found. An alignment of Chlamydia spp. IGSs upstream of ompA available in the NCBI database shows a minor variation in P3 and P2 regions (Table S1). This suggests that these regions may have a conserved role in ompA expression in Chlamydia spp. There are more variations in the ompA P1 region in terms of the length of the "spacer" separating the putative − 35/− 10 hexamers, ranging from 18 to 26 bps. Chlamydia spp. σ 66 -dependent promoters typically have a 16-18 bp spacer 11,31 , resembling those of E. coli σ 70 , which also senses the promoter spacing and preferentially recognizes a spacer with an average of 17 bps 33 .
Next, 5′ RACE assays were performed to determine the transcription start site (TSS) of P3-derived mRNA. Initially, total RNA isolated from the strain L2/434/Bu infected HeLa cells harvested at 18 h pi were used. Our attempt to detect the 5′ end of P3 mRNA was unsuccessful, although a site corresponding to the reported P1 The PCR products next to the DNA ladder on the agarose gel were subjected to sequencing, and the TSS of P3 appears to be "A" (in blue font), marked with an arrow. The putative − 35 and − 10 hexamers are indicated in red font. mRNA 14,27 was noted. We hypothesized that if the undetectable 5′ end of the P3 transcript is the result of low steady-state levels of P3 mRNA, the use of the C. trachomatis tuf RBS region should improve the stability of P3 mRNA in the reporter system. To test this hypothesis, P3 transcript was determined using a 5′ RACE assay with total RNA isolated from L2/pBomB-P3 infected HeLa cells harvested at 18 h pi. The DNA sequencing data showed that P3 transcript originated with an adenine located at 39 bp upstream from the gfp start codon (Fig. 3c). Based on the TSS at position + 1, the sequences of TTACGA and TATGGT are putative − 35 and − 10 hexamers of P3. Yuan et al. 28 . have previously reported a similar site in C. psiticci Mn Cal 10 and C. psiticci GPIC (now C. caviae according to current chlamydiae taxonomy) 34 . We conclude that P3, lying between the previous reported P2 and P1, indeed functions actively in C. trachomatis.
Defining the core P3 promoter. To precisely define the relevant sequence features needed for the recognition of P3, we investigated how specific mutations of P3 could affect transcription in vitro. Plasmids containing wild-type (WT) or mutated P3 regions (Fig. 4a) and a control promoter of the E. coli fliC (P fliC ) 35 were used as templates. A functional RNAP holoenzyme was reconstituted with core enzyme from E. coli and a hybrid C. trachomatis σ factor, σ 66R24 , in which σ 66 regions 2 to 4 (amino acid residues 315-571) were translationally fused to the region 1 of E. coli σ 70 (amino acid residues 1-372). The use of this holoenzyme allows for the study of σ 66 -dependent promoter activity that relies on the function of σ 66 regions 2 to 4. E. coli σ 70 or C. trachomatis σ 28 were used as controls. In the presence of σ 66R24 -or σ 70 -RNAP holoenzyme, transcripts from WT P3 were evident (Fig. 4b,c). In contrast, no P3-derived transcript was observed in the presence of σ 28 . These data indicate that P3 is specifically recognized by C. trachomatis σ 66 and its homolog E coli σ 70 , consistent with our β -gal reporter assay data in E. coli (Fig. 2). Substitution of the TATGGT (− 10 hexamer) with GCATGC abolished P3 transcription activity when either σ 66R24 RNAP or σ 70 RNAP was used, indicating the importance of the TATGGT hexamer for P3 activity. Interestingly, substitutions of the TTACGA (− 35 hexamer) to GGATCC reduced 80% of P3 transcript in comparison to the WT P3 using σ 66R24 RNAP, while the use of σ 70 RNAP reduced P3 transcript only by 50%, suggesting that the TTACGA hexamer is more recognizable by C. trachomatis σ 66 . These results indicate that the TTACGA sequence of P3 functions as the − 35 promoter element and confirm that this sequence is required for the full activity of P3. We noted that the −16 TGt −14 sequences in P3 matched the consensus extended -10 TG motif recognized by σ 70 from E. coli 36,37 . Substitution of TG with CA resulted in a significant decrease in P3 transcript levels, suggesting that the TG motif is important for P3 activity (Fig. 4). C. trachomatis σ 66 -dependent promoters have typically been characterized as − 10/− 35 promoters. Our data, for the first time, shows that P3 activity requires the − 35 hexamer and an extended − 10 TGn motif, presenting a new promoter class in C. trachomatis.
Promoters harboring the TGn motif have been identified from several microbial systems, including E. coli 37 , Streptococcus pneumoniae 38 , Bacillus subtilis 36 , and Mycobacterium tuberculosis 39 . It has been shown that the recognition of the TGn motif by the principle σ factor plays a role in the formation of an open complex during transcription initiation. Moreover, the TGn motif may compensate for non-canonical − 10 hexamers or suboptimal spacer length between −10 and −35 hexamers. The TGn-promoters appear to more often have significant deviations in the -35 hexamer than typical -10/-35 promoters in E. coli. It seems that ompA P3 deviates substantially from both the consensus − 10 hexamer and − 35 hexamer, because of its GC-rich sequences in these regions. Like Quantifying the contribution of P3 to ompA transcription. To define the role of P3 in ompA transcription and to understand the potential relationship between P3 expression and the action of the well-studied P2 14,27,29 , we quantified the levels of ompA P2 and/or P3 mRNA in the context of chromosomal loci using real-time RT-qPCR. Total RNA from L2/434/Bu infected cells, which were sampled at 0, 2, 4, 12, 24, and 36 h pi, were analyzed (Fig. 5a). P3 transcript was detected early at a low level and continued to accumulate in abundance, exceeding that of P2 at both 2 and 4 h pi (Fig. 5b). P2 transcript rapidly increased with bacteria multiplication and peaked at 12 h pi. However, at 12 h pi, the use of P3 decreased and that of P2 drastically increased. Expression of both ompA P2 and P3 was significantly eliminated at 36 h pi. These data support the notion that P2 is the primary transcript during RB replication and RB-to-EB differentiation, as previously reported 27 , whereas P3 plays a role in the early stage of the development. Therefore, temporally expressed P3 and P2 coordinate ompA transcription during the C. trachomatis developmental cycle.
Evaluating the use of P3-GFP as a reporter in C. trachomatis infected epithelial cells. C. trachomatis growth is accompanied by quantitative and qualitative changes in ompA expression (Fig. 5). To investigate the relationship between P3-GFP level and C. trachomatis growth, GFP in L2/pBOMB-P3 infected HeLa cells was monitored throughout the C. trachomatis developmental cycle using fluorescence microscopy and flow cytometry. In parallel, a one-step growth curve was constructed to enumerate infectious EB progeny yields. We visualized GFP-expressing chlamydial organisms associated with the host cells instantly post-infection, and the levels of GFP expression increased through 36 h pi as C. trachomatis inclusions expanded (Fig. 6a). When quantitatively measured using flow cytometry, the GFP signal was detected instantly after infection but the signal-to-noise ratio was low, preventing meaningful quantitative measurements of cells. The distinct cell populations with bona fide P3-GFP expression were separate from the GFP negative cells until 8 h pi (Fig. 6b). The population of GFP-expressing cells increased from ~3.1% at 8 h pi to 30% at 24 h pi; concurrently, the mean florescence intensity (MFI) significantly increased (Fig. 6b). These increases were correlated with the rapid growth of C. trachomatis (Fig. 6c). Interestingly, while GFP-expressing cells increased to 58.5% at 32 h pi, there were more heterogeneous cell populations which displayed a lower average MFI with a larger deviation compared to that at 24 h pi. This could be explained by the asynchronous growth of C. trachomatis, which occurs when decreasing numbers of RBs continue replicating and increasing progeny EBs accumulate and exit from the host cells to initiate a new cycle of infection. The overall P3-driven GFP levels seemed to be quite stable. Thus, it is likely that the GFP signal detected at the late stage is skewed by GFP accumulation along with bacteria accretion and inclusion expansion. We observed that L2/ Relative amounts of transcript were obtained by normalizing the levels of P2 or P3 transcript to C. trachomatis genomic DNA levels, which were assessed using primers specific to the 16S rRNA gene (Table S3). The values are presented as mean ± SD from an experiment with duplicates. * P < 0.05, * * P < 0.005, and * * * * P < 0.0001. Three independent experiments were performed.
Scientific RepoRts | 6:27244 | DOI: 10.1038/srep27244 pBOMB-P3 exhibited delayed growth pattern relative to L2/434/Bu (Fig. 6c), possibly due to the stress induced by GFP overexpression. Isolated L2/pBOMB-P3 organisms from infected cells harvested at 40 h pi appeared green-tinged, a change detectable by eye (Fig. 6d), indicating high levels of GFP associated with bacteria. Long-lived P3-GFP used in this study may be applicable for long-term bacteria cell-labeling, which allows for the imagining to track the C. trachomatis infection process in intact cells, e.g. invasion, multiplication, and dissemination.
Determining the strength of P3 in C. trachomatis. To evaluate the strength of P3, we assessed the P3-driven GFP levels in L2/pBOMB-P3 compared to those driven by the Neisseria meningitidis promoter, P nm , in L2/pBOMBm 18,20 . Four different quantitative methods (microscopy, flow cytometry, microspectrometry, and immunoblotting) were used. First, to facilitate a direct comparison, an equivalent mixture of L2/pBOMB-P3 and L2/ pBOMBm was used to infect HeLa cells, followed by fluorescence microscopy analysis. In the same field of vision, L2/pBOMB-P3 exhibited a brighter green fluorescence than L2/pBOMBm (Fig. 7a). In the presence of equivalent mCherry levels, which were induced by the addition of α TC, L2/pBOMB-P3 displayed higher levels of GFP signal than L2/pBOMBm. Next, cells infected with either L2/pBOMB-P3 or L2/pBOMBm with similar infectivity (~30%) were harvested at 24 h pi and subjected to flow cytometry. As determined by assessment of the MFI, the average level of P3-driven GFP expression was ~32 times stronger than P nm -driven GFP at 24 h pi, (Fig. 7b). Further, isolated C. trachomatis organisms from cells harvested at 24 h pi were measured for their relative fluorescence intensities (RFI) using microspectrometry. Figure 7c shows that the RFI of L2/pBOMB-P3 was ~10-fold higher than that of L2/pBOMBm. Lastly, the level of GFP protein in isolated C. trachomatis organisms was determined by immunoblotting. The levels of GFP in lysates of L2/pBOMB-P3 were ~24-fold higher than those of L2/pBOMBm (Fig. 7d). With the use of similar plasmid constructs, these results demonstrate that, relative to P nm , P3 is highly active in C. trachomatis. Furthermore, bacterial organisms expressing P3-GFP are more noticeable and easier to measure using fluorescence microscopy, flow cytometry and immunoblotting, rather than microspectrometry. OmpA P3 displays weak transcription activity in E. coli. Our next step was to determine whether P3 activity varies with the E. coli σ 70 paradigm. This information may provide new insight into the diversity of bacterial gene transcription and also may provide an explanation for the previous detection of P3 in E. coli systems. To this end, E. coli strains carrying pBOMB-P3, pBOMBm, and promoter-less pBOMB-PL were harvested at the exponential phase and their GFP expression was measured using fluorescence microscopy, microspectrometry, and immunoblotting. A significant increase in GFP expression in E. coli cells was observed as the bacterial culture grew (Fig. 8a), confirming that P3 was recognized by E. coli RNAP. This data is consistent with the earlier observations as shown in Figs 2 and 4b. Unlike in C. trachomatis, where P3 is highly expressed, P3-driven GFP expression was much lower than P nm -driven GFP in E. coli (Fig. 8b,c). The weak activity of P3 in E. coli is not surprising, as perfect P3 function may be achieved only in the context of a regulatory network unique to C. trachomatis; different regulatory mechanisms exist in E. coli.
Given the high similarity between C. trachomatis σ 66 and E. coli σ 70 in their regions 1 to 4, the use of heterogeneous E. coli systems has provided useful information to understand some questions regarding chlamydial promoter recognition as described both previously and in this study. However, these heterogeneous systems do not allow for the expression of all C. trachomatis genes, such as ompA P2 29 . P3 functions weakly in E. coli and appears to rely on the availability of endogenous conditions for optimal expression in C. trachomatis. Such variety might be attributed to the non-conserved regions between C. trachomatis σ 66 and E. coli σ 70 , including the area between regions 1 and 2, as well as the regions at the N-and C-termini of σ factors 42,43 . Differences between C. trachomatis σ 28 and E. coli σ 28 have been studied previously 10,44 . Additionally, physiological signals within the host cells (such as immunity and metabolic capacity) that C. trachomatis encounters affect bacterial gene expression. Therefore, it is necessary to validate the promoters that were predicted or identified by the heterologous systems in C. trachomatis. A major advantage of the transcription reporter shuttle plasmid with an MCS and tuf RBS upstream of GFP is to allow a preferred promoter to be inserted conveniently so that promoter functioning can be studied by assessing the levels of GFP expression. Promoters detected in C. trachomatis could be directly compared to those in E. coli, thus providing an opportunity to examine the extent of the differences in their activity.

Conclusion
This study resulted in three significant conclusions. First, the ompA regulatory region is complex and exhibits novel features. In particular, ompA transcription is exactly derived from both mid-cycle expressed P2 and the newly identified, early-expressed P3. Second, P3 represents a new class of C. trachomatis promoter, which requires a − 35 DNA element and an extended − 10 TGn motif for its transcription. Third, expression of ompA P3-GFP, Tandemly arranged promoters have been reported to involve developmentally regulated C. trachomatis gene expression, including ompA 27,28 and tuf 11 . To complement the earlier finding that P2 plays a primary role in ompA synthesis, we have shown that P3 also contributes to ompA transcription (Figs 2-5). Previous studies have indicated the detection of steady-state levels of ompA transcript mid-cycle 6,7 . However, the levels of ompA transcripts often do not fully reflect the relative activities of the promoters because the fate of each transcript may be varied and is affected by various processes. The switch in utilization of P3 and/or P2 is evidently a regulatory mechanism of ompA transcription. The manner in which P3 and P2 activity is coordinated or mutually influenced can be affected by multiple factors. A 9-bp inverted repeat adjacent to the − 35 hexamer of the P3 region may play a role in this process. Mathews and Stephens 12 have proposed that this repeat, resembling an operator in DNA structure, might be a target of a yet not-identified protein repressor. Alternatively, a stable stem-loop structure that resembles a rho-independent terminator capable of stopping P2 transcription could be formed by this region. Additionally, DNA supercoiling has been reported to modify C. trachomatis transcription, including ompA P2 45 . Only a subset of early genes can be affected by DNA supercoiling 46 . In general, plasmids are supercoiled to a higher degree than chromosomal DNA. We found similar strong activity of P3 regardless of its plasmid and chromosomal locations, suggesting that supercoiling is less likely to be a main factor involved in P3 expression. Moreover, posttranscriptional regulation of ompA mRNA may occur during development. Transcript derived from P3 seems less stable than that of P2 by mRNA structural prediction. Despite its role in early transcription, P3 mRNA might limit its activity through quick turnover when RBs rapidly replicate and convert to EBs. We cannot exclude the possibility that P1 signal detected is the processed product of longer transcripts 14 , perhaps from P3 and/or P2. It is worth noting that the utility of tuf RBS in mediating the translation of gfp may be mechanistically different from, and possibly more efficient than, the native ompA RBS. Because of the apparent conservation of the P3 and P2 regions, their importance in Chlamydia spp. adaptation to changing environments can be predicted. The proposed P1 remains unproven, but phylogenetic analysis suggest that P1 is less likely to be the conserved regulatory mechanism of ompA in Chlamydia spp.
Our findings underscore the role of new promoter element, the extended − 10 TGn, in C. trachomatis transcription. By sequence inspection, early promoters in C. trachomatis, P euo and the promoter from C. trachomatis groES 13 , also contain potential extended − 10 TGn motifs. It is possible that such a motif represents a signature element for early promoter-RNAP recognition and is affected differently from the typical − 35/− 10 promoters by a transcription factor. For example, we have found that anti-σ factor, CT663 (or Scc4) 40 , strongly inhibited transcription from the − 35/− 10 promoters but was less effective in suppressing transcription from an extended Bacterial lysates were used for immunoblotting with antibodies to GFP or RpoB which was used as a loading control for protein amounts. The relative amount of GFP was obtained by normalizing the GFP intensity to the corresponding RpoB intensity and value is reported as a percentage relative to L2/pBOMB-P3, as indicated at the bottom.
Scientific RepoRts | 6:27244 | DOI: 10.1038/srep27244 − 10 promoter lacking a − 35 element. Despite the limited number of transcription factors in C. trachomatis, the arrangements of promoter modules may be largely diverse and greatly contribute to developmentally regulated gene expression profiles.
The consistency of the different quantitative methods employed show that robust P3-GFP signal is present throughout the developmental cycle. These properties make P3-GFP valuable for long-term bacterial cell-labeling to facilitate the visualization and tracking of infectious process in live cells. Coupled with flow cytometry and microscopy, this GFP-reporter assay may also be applicable in probing ompA changes induced by various insults at both population and single cell level. Adverse growth conditions, such as exposure to IFNγ or antibiotics, cause C. trachomatis ompA downregulation and the formation of an "altered persistent form" 17,47 . With regard to this, low levels of GFP expression driven by a weak promoter, such as P euo , could be challenging to interpret. At the same time, the use of long-lived GFP may be limited, as it is incapable of fully reflecting the highly dynamic process of gene expression. An improved short-lived GFP reporter may be more suitable for studying the transient mRNA dynamics in C. trachomatis. All together, the results obtained and approaches used in this study will allow for the design of future studies in defining promoter signatures, relationships between promoter structure, RNAP recognition, and transcriptional activity in C. trachomatis.

Materials and Methods
Cell cultures and bacteria growth. HeLa 229 cells (human cervical epithelial carcinoma cells; ATCC CCL-2) were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 1 mM glutamic acid, 10% fetal bovine serum and 20 μ g/ml gentamycin (DMEM-10) at 37 °C in an incubator supplied with 5% CO 2 . HeLa cells were infected with C. trachomatis as previously described 17 . To construct a one-step growth curve, cells infected at an MOI of 1 were lysed and cultured on fresh HeLa cells at 0, 4, 12, 24, 32, and 48 h pi. Infectious EB progeny was then evaluated by enumeration of inclusion forming units (IFUs) in 1 ml. E. coli DH5α was used as the host for cloning. A methylation deficient E. coli strain (ER2925, New England Biolabs) was used to prepare the plasmids for C. trachomatis transformation. The E. coli cells were grown in Luria-Bertani (LB) broth or agar plates containing the appropriate antibiotics. Molecular cloning. Plasmids, primers, and oligonucleotides used in this study are listed in Table S2 and   Table S3, respectively. Primers and oligonucleotides were synthesized by Integrated DNA Technologies. To construct pPvGFP::SW2 (Fig. 1a), two separated PCR fragments were obtained with primers, Euo_tuf priF/ORF2 priF or euoPriR/ORF1priF, using pGFP::SW2 18 (a gift form Dr. Ina Clarke, University of Southampton, UK) as template. These PCR fragments were mixed, annealed, and used as templates for subsequent PCR using primers ORF2 priF/ORF1priF. The Bam HI-cut PCR fragment was then inserted into the sites of Bam HI in pGF-P::SW2. To create pP3GFP::SW2, the two annealed complementary oligonucleotides consisting of ompA P3 were cloned into pPvGFP::SW2 at the Spe I/Nae I sites. The Puv I/Nco I fragment of the pP3GFP::SW2 was cloned into the Pvu I/Nco I sites in pBOMB4-tet-mCherry (a gift form Ted Hackstadt, Rocky Mountain Lab, NIH) to yield pBOMB-P3 (Fig. 3a). To create a promoter-free plasmid, pPvGFP::SW2 was digested by Spe I/Nae I to remove P euo . After Kenow fragment treatment, the DNA was self-ligated to create pPLGFP::SW2. The Puv I/Nco I fragment was subcloned into pBOMB to create pBOMB-PL. The plasmids used for the in vitro transcription assay were derived from pP flic 35 . Two annealed complementary oligonucleotides consisting of WT or mutated P3 were inserted into pP flic at the Xba I/EcoR V sites to yield pP3WT-P fliC , pP3m10-P fliC , pP3mTG-P fliC or pP3m35-P fliC . The expression plasmid, pLN-σ 66R24 , containing genes encoding C. trachomatis σ 66 regions 2-4 and E. coli σ 70 region 1, was derived from pLNH-12. The PCR fragment consisting of region 1 of E. coli σ 70 was amplified with primers s70priF/Sig7066Rg24BLo. The PCR fragment containing σ 66 regions 2-4 was amplified with primers sig7066Rg24CUp/S66prRBam. These DNA fragments were annealed and used as a template for subsequent PCR using primers EC70priF/S66prRBam. The resultant PCR fragments were digested with Nco I/Bam HI and were then inserted into the Nco I/BamH I sites of pLN-7066 to generate pLN-σ 66R24 , which encodes a fusion σ 66R24 protein. The identities of all constructs were confirmed by PCR and DNA sequencing.
Generating ompA promoter-lac Z fusion in E. coli. A DNA fragment containing the ompA P1 and P2 regions (+ 375 to + 18 relative to the translation start codon ATG of ompA) was amplified by PCR with primers ompApriF/P1priR using the genome of C. trachomatis serovar D as a template. The EcoR I/Sal I digested PCR fragment was then inserted into the Eco RI/Sal I sites of pFW11 to yield pFW/P21. Two annealed oligonucleotides containing P2 (+ 317 to + 243 relative to the ompA start codon) or P1 (+ 87 to + 18 relative to the ompA start codon) were inserted into pFW11 at Eco RI/Sal I sites, resulting in pFW/P2 or pFW/P1 (Fig. 2a). These constructs were introduced into E. coli CSH100 for homologous recombination of the ompA promoter::lacZ onto an F' episome as previously described 30 . Further mating with E. coli strain FW102 was performed to finalize the construction of the reporter strains, designated as FW/P21, FW/P2, and FW/P1, respectively. The identities of all constructs and strains were confirmed by PCR and DNA sequencing.
Transformation of C. trachomatis. C. trachomatis was transformed with the shuttle plasmids according to the method described 18 with minor modifications. Briefly, plasmid-free C. trachomatis L2/25667R EBs (1 × 10 7 ) were mixed with 7 μ g of plasmid DNA in 100 μ l of CaCl 2 buffer (5 mM Tris HCl, pH7.4, 100 mM CaCl 2 ) and incubated at room temperature for 30 minutes. Freshly trypsinized HeLa cells (6 × 10 6 ) resuspended in 200 μ l CaCl 2 buffer were added to the plasmid/EB mixture and incubated at 37 °C for an additional 20 min. Aliquots of this mixture were then added to a 6-well plate with 1.0 ml of pre-warmed medium in each well. After culturing in DMEM-10 without antibiotics at 37 °C for 24 h, cells were incubated in the presence of ampicillin (5 μ g/ml) and cycloheximide (20 μ g/ml) for an additional 24 hours. The infected cells were harvested and lysed by vortexing with glass beads. The cell debris was removed by spinning at 233 g for 10 min. The Chlamydia-containing supernatant was collected and added onto a HeLa monolayer in a T175 flask. After incubation at 37 °C for 1 hour with gentle shaking, the infected cells were cultured in DMEM-10 containing ampicillin and cycloheximide for 48 hours. The C. trachomatis organisms were isolated by 30% renografin density gradient purification to enrich infectious EBs. For the first passage, a HeLa monolayer in a 6-well plate was inoculated with EBs, followed by centrifugation to enhance infection. Passages were continued 2-3 times, until inclusions positive for green fluorescence were observed. To isolate a single clone of C. trachomatis transformant, GFP-expressing cells were lysed in sterile dH 2 O, serially diluted, and inoculated onto a HeLa cell monolayer in a 96-well plate. At 24 h pi, the plate was examined by fluorescence microscopy. The culture wells containing only a single inclusion were marked and continued to be cultured until subsequent harvest for C. trachomatis large scale amplification.
Phase contrast and fluorescence microscopy. HeLa cell monolayers grown in a glass culture chamber (NUNC) or 24-well culture plate were infected with C. trachomatis to achieve ~30% infection. At various times post infection, the culture wells were observed and photographed with an inverted fluorescence microscope (Zeiss Axio Observer D1). Images were processed using AxioVision software version 4.8.

Microspectrometry. Isolated chlamydial organisms from infected cells harvested at 36 h pi were washed
with PBS, diluted to OD 600 = 0.4, and subjected to fluorescence intensity analysis using a Synergymx microplate reader (BioTek) with excitation wavelengths of 48 nm and 528 nm. A volume of 100 μ l of bacterial cell dilutions was added into each well of a 96-well plate. The following formula was used to determine relative fluorescence intensity (RFI): RFI = (bacterial fluorescence intensity-background)/bacterial OD 600 . The overnight cultures of E. coli DH5α carrying GFP expression plasmids were diluted in fresh LB (1:100) and grown at 37 °C. E. coli cells grown at log phase (4 hours after inoculation) were collected (0.5 ml) by centrifugation and resuspended in the same volume of PBS. The fluorescence intensity of E.coli was obtained in the same manner as described above for C. trachomatis.
Flow cytometry. C. trachomatis infected HeLa 229 cells were detached by trypsinization at different times as indicated in the results and fixed with 2% paraformaldehyde at room temperature for 20 minutes. Cells were then detected in a FACSVantage flow cytometer (BD Biosciences) using the FL-1 (green) channel. Flow cytometry data was recorded for at least 3 × 10 4 cells per sample. Mock infection cells were used as a blank. Data were analyzed using FlowJo software Version 7.6 (TreeStar Inc) for both the percentage of GFP-expressing cell populations and mean fluorescence intensity.
Immunoblotting analysis. Chlamydial organisms were lysed with 2 × SDS loading buffer, separated on 4-20% SDS-PAGE, and transferred onto a PVDF membrane for immunoblotting. The membrane was incubated with a polyclonal anti-GFP antibody (Pierce) or a monoclonal antibody to bacterial RNAP β subunit (RpoB) (Neoclone), followed by incubation with horseradish peroxidase-conjugated goat anti-mouse IgG. RpoB was used as a protein loading control. The protein bands were visualized by an enhanced chemiluminescence kit (Pierce). 5′ Rapid amplification of cDNA ends (RACE). C. trachomatis-infected cells with ~90% infectivity were harvested at 18 hours. Total RNA was isolated using TRIzol reagent (Invitrogen). The transcription start site for P3 in C. trachomatis was probed with the rapid amplification of the cDNA ends (5′ -RACE) using the FirstChoice ® RLM-RACE Kit version 2.0 (Life Technologies) following the manufacturer's protocol. Briefly, 10 μ g of total RNA were treated with calf intestine alkaline phosphatase to remove free 5′ -phosphates. The RNA was then treated with tobacco acid pyrophosphatase to remove the cap structure from full-length mRNA, leaving a 5′ -monophosphate. An RNA adapter oligonucleotide was then ligated to the RNA with 5′ -phosphate using T4 RNA ligase. The 5′ end of the gfp transcript, which is initiated by P3, was then amplified with random-primer reverse transcription and nested PCR using gfp specific primers with the provided primers in the kit. PCR products were then inserted into pUC19 (Promega, Madison, WI) for DNA sequencing.

Real-time reverse transcription quantitative PCR (RT-qPCR). C. trachomatis infected HeLa cells
were harvested at indicated times post infection. Genomic DNA was isolated from cells using the DNeasy ® Blood & Tissue Kit (Qiagen). RNA was isolated from an equivalent number of cells using the Direct-zol ™ RNA Kit (Zymo). DNase treatment was performed to remove residual DNA. A total of 1 μ g of RNA was reverse transcribed into cDNA and the genes of interests were amplified using the VeriQuest Fast SYBR Green qPCR (USB) with appropriate primers (Table S2). The PCR cycle conditions were as follows: 50 °C for 2 min., 95 °C for 5 min., 95 °C for 3 sec., and 60 °C for 30 sec. The latter two steps were repeated for 35 to 45 cycles, and fluorescence was detected at the end of each cycle. Transcripts were normalized to C. trachomatis genomic DNA levels, which were quantified by qPCR with primers specific to the 16S rRNA gene 48 (Table S3). The 2-ddCt method was used to obtain relative transcript levels.

Recombinant protein purification, RNAP and the in vitro transcription assays. Recombinant
C. trachomatis σ 66R24 was expressed in E. coli Rosetta TM (DE3) pLysS cells harboring pLN-σ 66R24 . Cells were grown in LB broth at 37 °C to A 600 = 0.8, and then protein expression was induced by the addition of isopropyl-1-thio-β -d-galactopyranoside (IPTG) to a final concentration of 0.5 mM. Protein purification was performed using anion exchange chromatography (Source Q, GE Healthcare), followed by gel filtration purification as previously described 40 . Recombinant C. trachomatis σ 28 was purified as previously described 35 . To make a functional RNAP holoenzyme, the purified σ 66R24 (2.0 μ g) was mixed with 1 unit of E. coli RNAP core (Epicentre Technologies) and incubated on ice for 30 min. The σ 28 -RNAP holoenzyme was reconstituted using E. coli core and purified σ 28 . E. coli σ 70 RNAP holoenzyme was purchased from USB. An in vitro transcription assay was performed as described previously 10,40 , in a 10 μ l reaction containing 1 μ l 10 × RNA Pol reaction buffer (New England Biolab), 1 μ l RNAP holoenzyme, 1 μ g of plasmid template, 400 μ M ATP, 400 μ M UTP, 1.2 μ M CTP, 0.2 μ M [α -32 P] CTP, 100 μ M 3′ -O-methylguanosine 5′ -triphosphate (GE HealthCare), 2 mM DTT, and 20 U RNase inhibitor (USB). The reaction was incubated for 15 min at 37 °C and terminated by adding loading buffer. The transcripts were separated by electrophoresis with 6% polyacrylamide/8 M urea gel and visualized by autoradiography.
Statistical analyses. Data analyses were performed using GraphPad PRISM software. Statistical significance was determined by two-way analysis of variance (ANOVA). Values of * P < 0.05 were considered statistically significant.