Identification of a GrgA-Euo-HrcA Transcriptional Regulatory Network in Chlamydia

ABSTRACT Chlamydia trachomatis is an obligate intracellular bacterium whose unique developmental cycle consists of an infectious elementary body and a replicative reticulate body. Progression of this developmental cycle requires temporal control of the transcriptome. In addition to the three chlamydial sigma factors (σ66, σ28, and σ54) that recognize promoter sequences of genes, chlamydial transcription factors are expected to play crucial roles in transcriptional regulation. Here, we investigate the function of GrgA, a Chlamydia-specific transcription factor, in C. trachomatis transcriptomic expression. We show that 10 to 30 min of GrgA overexpression induces 13 genes, which likely comprise the direct regulon of GrgA. Significantly, σ66-dependent genes that code for two important transcription repressors are components of the direct regulon. One of these repressors is Euo, which prevents the expression of late genes during early phases. The other is HrcA, which regulates molecular chaperone expression and controls stress response. The direct regulon also includes a σ28-dependent gene that codes for the putative virulence factor PmpI. Furthermore, overexpression of GrgA leads to decreased expression of almost all tRNAs. Transcriptomic studies suggest that GrgA, Euo, and HrcA have distinct but overlapping indirect regulons. These findings, together with temporal expression patterns of grgA, euo, and hrcA, indicate that a transcriptional regulatory network of these three transcription factors plays critical roles in C. trachomatis growth and development. IMPORTANCEChlamydia trachomatis is the most prevalent sexually transmitted bacterial pathogen worldwide and is a leading cause of preventable blindness in underdeveloped areas as well as some developed countries. Chlamydia carries genes that encode a limited number of known transcription factors. While Euo is thought to be critical for early chlamydial development, the functions of GrgA and HrcA in the developmental cycle are unclear. Activation of euo and hrcA immediately following GrgA overexpression indicates that GrgA functions as a master transcriptional regulator. In addition, by broadly inhibiting tRNA expression, GrgA serves as a key regulator of chlamydial protein synthesis. Furthermore, by upregulating pmpI, GrgA may act as an upstream virulence determinant. Finally, genes coregulated by GrgA, Euo, and HrcA likely play critical roles in chlamydial growth and developmental control.

IMPORTANCE Chlamydia trachomatis is the most prevalent sexually transmitted bacterial pathogen worldwide and is a leading cause of preventable blindness in underdeveloped areas as well as some developed countries. Chlamydia carries genes that encode a limited number of known transcription factors. While Euo is thought to be critical for early chlamydial development, the functions of GrgA and HrcA in the developmental cycle are unclear. Activation of euo and hrcA immediately following GrgA overexpression indicates that GrgA functions as a master transcriptional regulator. In addition, by broadly inhibiting tRNA expression, GrgA serves as a key regulator of chlamydial protein synthesis. Furthermore, by upregulating pmpI, GrgA may act as an upstream virulence determinant. Finally, genes coregulated by GrgA, Euo, and HrcA likely play critical roles in chlamydial growth and developmental control.
C hlamydiae are obligate intracellular Gram-negative bacteria with a unique developmental cycle characterized by two cellular forms (1). The small form, known as the elementary body (EB), is capable of short-term extracellular survival but incapable of proliferation. After entering the host cell through endocytosis, the EB differentiates into a larger form known as the reticulate body (RB) inside a vacuole termed the inclusion. The RB replicates with a doubling time of about 2 h (2)(3)(4). Around 24 h, some RBs start to redifferentiate back into EBs while others continue to proliferate (3,5). At the end of the cycle, both infectious EBs and noninfectious RBs are released from host cells through either cell lysis or extrusion of entire inclusions (6,7).
Expression of the chlamydial transcriptome is developmentally regulated. Previous cDNA microarray studies (8,9) enumerate four successive stages of the Chlamydia trachomatis developmental cycle. The immediate early stage is defined as the first hour when EBs are inside nascent inclusions near the plasma membrane. Only a small number of presumably crucial genes are transcribed in this stage to establish an intracellular niche that enables EB survival, RB development, and eventual delivery of the inclusion to a perinuclear region. During the subsequent early stage, an additional number of genes are transcribed to complete the conversion of EBs into RBs. Midcycle commences upon the completion of EB-to-RB conversion and ends when RBs start to differentiate back into EBs. Almost all genes are transcribed during this stage. Last, transcription of a smaller set of genes is initiated and/or upregulated before and during the late stage.
C. trachomatis carries genes that encode three sigma factors (10,11), which guide the RNA polymerase to different promoters (12,13). The vast majority of C. trachomatis promoters are s 66 dependent, while some late genes may possess s 28 or s 54 promoters (14)(15)(16)(17)(18)(19). A few late genes have tandem promoters for s 66 and an alternative s (14). Consistent with their roles in the developmental cycle, expression of the three sigma factors is also temporally regulated (8,9,20). s 66 RNA is detected by microarray as early as 3 h postinoculation (hpi), whereas s 28 and s 54 mRNAs are not detected until 8 hpi (8).
GrgA is the most recently discovered chlamydial TF (39). Identified via promoter DNA pulldown, GrgA physically interacts with s 66 and s 28 and activates transcription from both s 66 -and s 28 -dependent promoters in vitro (39)(40)(41). Transcriptomic studies presented in this report define the regulon of GrgA and identify a GrgA-directed transcriptional regulatory network (TRN). Similar to euo, expression analysis reveals that hrcA is also an immediate early gene and suggests the possibility that the GrgA protein prepackaged within EBs plays a critical role in the activation of euo, hrcA, and other genes after the EB enters the host cell.

RESULTS
GrgA overexpression-mediated global transcriptomic changes include upregulated euo, hrcA, and pmpI expression and downregulated tRNA expression. We performed transcriptome sequencing (RNA-seq) analysis for C. trachomatis L2 (CtL2) transformed with an anhydrotetracycline (ATC)-inducible GrgA expression plasmid (4) to determine the role of GrgA in C. trachomatis transcriptomic expression. In our pilot RNA-seq experiments, GrgA transformants (i.e., CtL2/GrgA) grown in mouse fibroblast L929 cells were treated with or without ATC within two time periods: 12 to 16 hpi and 17 to 21 hpi (n = 1 per experimental condition). The rates of reads mapped to the CtL2 genome of samples prepared at 16 hpi for noninduced and ATC-induced cultures were 13.0% and 16.8%, respectively (see Data Set S1A in the supplemental material). The numbers of mapped reads resulted in 135-fold and 164-fold CtL2 genome coverage, respectively. As expected, the rates of reads mapped to the CtL2 genome of samples prepared at 21 hpi were 3-to 4-fold higher (i.e., 44.1% and 52.6%) (Data Set S1A). Conversely, the rates of reads mapped to the mouse genome of samples prepared at 16 hpi were higher than the rates of those at 21 hpi (data not shown). The total mapped rate (i.e., the chlamydial and murine mapping rates combined) ranged from 95.1% to 96.6% (data not shown).
With the exception of 5S and 16S rRNAs, which were depleted prior to library construction, we detected RNAs of all but seven chlamydial genes at 16 hpi despite the notably lower reads at this time point. In both sets of experiments, RNAs of two transcription repressor-encoding genes euo and hrcA were noticeably increased in ATCinduced cultures. For the 12 to 16 hpi induction, euo and hrcA increased by 3.1-and 2.8-fold, respectively. For the 17 to 21 hpi induction, they increased by 3.4-and 1.9fold, respectively (Data Set S1B).
Subsequent RNA-seq studies were conducted with samples harvested from infected L929 cells at 16 hpi. We chose this time point first because it corresponds to the midlog phase of RB replication, and we are most interested in uncovering how GrgA mediates RB growth regulation, and additionally because the genome coverages in the above pilot study with samples prepared at 16 hpi exceeded the recommended coverage for bacteria with small genomes (42). We repeated RNA-seq analysis with ATCtreated biological replicates for the 12 to 16 hpi time period to generate data that could be analyzed statistically. Consistent with our previous two RNA-seq studies with a single sample per condition (Data Set S1), transcripts of both euo and hrcA also increased significantly in response to ATC treatment in the duplicated experiments. euo increased by 3.3-fold, the highest increase not including ATC-induced increase in the grgA RNA (Data Set S2B). hrcA increased by 2.1-fold, the fifth largest increase. Transcripts of 81 other genes also increased by $1.33-fold with an adjusted P value of ,0.05 (Data Set S2B). We chose this relatively low cutoff to increase the chance of identifying biological targets of GrgA, a strategy used by previous RNA-seq studies (43)(44)(45). Among the 81 other upregulated genes is pmpI, which encodes a polymorphic protein in the outer membrane complex that is a putative virulence factor (46,47). pmpI was upregulated by 2.5-fold, second only to euo (Data Set S2B). Retrospective analysis showed that pmpI was also upregulated by 2.7-fold and 2.5-fold in the aforementioned pilot study with ATC treatment from 12 to 16 hpi and from 17 to 21 hpi, respectively, without biological replicates (Data Set S1B). Collectively, euo, hrcA, pmpI, and the remaining 80 genes upregulated by GrgA fell into 14 different functional groups, as will be described in TRN analysis.
Transcripts of 78 genes decreased by 1.33-fold with an adjusted P , 0.05 in the biologically duplicated RNA-seq study (Data Set S2B). Of these 78 genes, 28 were tRNA genes. Only 9 of the 37 tRNAs were not significantly downregulated (Table 1). Retrospective analysis showed that 26 tRNAs were also decreased by more than 1.33fold in the pilot experiment with ATC treatment from 12 to 16 hpi without biological replicates (Data Set S1B).
Taken together, these RNA-seq studies performed for cultures with 4-h ATC treatment consistently demonstrated upregulated euo, hrcA, and pmpI expression in response to GrgA overexpression. These findings suggest the possibility that GrgA functions as an activator of euo, hrcA, and pmpI. In addition, the studies indicate that GrgA regulates expression of numerous other genes either directly or indirectly.
Upregulation of euo and hrcA but not pmpI depends on r 66 binding of GrgA. Previously published in vitro studies showed that GrgA activates both s 66 -dependent transcription and s 28 -dependent transcription by physically interacting with s 66 and s 28 , respectively (39,40). To determine the roles of GrgA's interactions with s 66 and s 28 in GrgA overexpression-induced transcriptomic changes, we performed RNA-seq for CtL2/GrgADs 66 BD and CtL2/GrgADs 28 BD, which are CtL2 transformed with inducible expression plasmids for GrgA that lacked either the s 66 -binding domain (GrgADs 66 BD) or the s 28 -binding domain (GrgADs 28 BD), respectively. In ATC-treated CtL2/GrgADs 66 BD cultured in L929 cells, upregulation of only a single gene was statistically significant, while downregulation of only four genes was statistically significant ( Fig. 1 and Data Set S2C), even though all forms of transformed GrgA are expressed upon ATC induction (Fig. S1). In ATC-treated CtL2/GrgADs 28 BD, the numbers of upregulated and downregulated genes were both higher than those of ATC-treated CtL2/ GrgA ( Fig. 1 and Data Set S2D).
The sole gene upregulated by GrgADs 66 BD overexpression was pmpI, which was also upregulated by overexpression of both full-length GrgA and GrgADs 28 BD ( Fig. 1A and Data Set S2E). Noticeably, pmpI expression increased after full-length GrgA overexpression and GrgADs 66 BD overexpression by a similar magnitude (2.5-fold versus 1.7fold). These increases were substantially higher than the 40% increase observed after GrgADs 28 BD overexpression (Data Set S2E). Fifty-one genes, including euo and hrcA, were upregulated following both full-length GrgA overexpression and GrgADs 28 BD overexpression ( Fig. 1A and Data Set S2E). Two of the four genes downregulated by GrgADs 66 BD overexpression were ndk (nucleotide diphosphate kinase) and ctl0407 (lipoprotein releasing system ABC transporter ATP-binding protein), which were also downregulated by overexpression of both full-length GrgA and GrgADs 28 BD ( Fig. 1B and Data Set S3E). dppD (oligopeptide ABC transporter ATP-binding protein) is the third GrgADs 66 BD-downregulated gene, which was also downregulated by full-length GrgA. In total, 66 genes were downregulated by both full-length GrgA overexpression and GrgADs 28 BD overexpression (Fig. 1B). Remarkably, 28 of these 66 genes encode tRNAs (Data Set S2E).
Taken together, comparative transcriptomic analysis ( Fig. 1 and Data Set S2) suggests that nearly all transcriptomic changes (including upregulation of euo and hrcA) induced by GrgA overexpression depend on GrgA binding of s 66 . In contrast, fewer changes (e.g., upregulation of pmpI) depend on GrgA binding of s 28 . However, overexpression of GrgADs 28 BD can induce additional transcriptomic changes not seen with full-length GrgA overexpression. Therefore, the s 28 -binding domain may restrict upregulation of certain s 66 -dependent genes in vivo. euo, hrcA, and 11 other genes are upregulated immediately following GrgA overexpression. All three RNA-seq studies presented above were performed in cultures with and without 4-h ATC treatment. To identify genes directly targeted by GrgA, we determined transcriptomic kinetics by extracting RNA at 16 hpi from GrgA transformant-infected L929 cultures with 10 nM ATC treatment initiated at 15.5, 15, or 14 hpi or without ATC treatment. This treatment schedule allowed for 0.5, 1, or 2 h of GrgA overexpression, respectively. Using Gaussian mixture models (48), the entire transcriptome was divided into six groups based on changes in the expression kinetics of individual genes ( Fig. 2 and Data Set S3E). Group A contains six genes whose functions are presented in Fig. 3A. These six genes were upregulated as early as 0.5 h after induction ( Fig. 2A), although the upregulation at 0.5 h was statistically significant for only four of the six genes. The four statistically significantly upregulated genes were euo, pmpI, aroC, and lplA. These genes could be direct targets of GrgA. A total of 174 genes, including hrcA, were upregulated by 1 h (Fig. 2B). Most genes of this large group may be secondary or indirect targets of GrgA. Expression of 444 genes remained relatively constant (Fig. 2C) and are therefore unlikely targets of GrgA considering GrgA acts as a transcription activator (39,40). Expression of the remaining genes decreased by various degrees (Fig. 2D to F), likely in response to expression changes of the primary and/or secondary targets.
A series of quantitative reverse transcription-PCR (qRT-PCR) experiments were performed to validate the RNA-seq data for all genes upregulated by 0.5-h ATC induction and selected genes upregulated by 1-h ATC induction because they might be directly  Fig. 2A and Data Set S3), euo, pmpI, ctl0758, and ctl0418 are nonoperon genes (Fig. 3B). Results showed that expression of euo and pmpI increased about 2-and 3-fold, respectively, at 10 min after induction, and by more than 3-and 4-fold, respectively, at 30 min (Fig. 3B). Smaller but significant increases were detected for the mRNA of ctl0758 from 10 to 30 min (Fig. 3B). However, a 57% increase in ctl0418 expression was not detected until 30 min, and this increase was not statistically significant (Fig. 3B). Based on the genome organization (10,11) and results of a study on C. trachomatis transcriptional start sites (49), lplA and ctl0536 lie within one operon (Fig. 3C), while aroC and three other genes lie within another (Fig. 3D). Noticeably, RNA-seq did not detect concordant upregulation of genes cotranscribed with lplA and aroC (Data Set S3). Therefore, we also included a cotranscribed gene from the same operon in our qRT-PCR analysis. This analysis shows that the expression trends of both lplA and ctl0536 (Fig. 3C) were similar to those of euo and pmpI (Fig. 3B) with a significant increase for lplA and a trending significant increase for ctl0536 starting at 10 min. Statistically significant increases were also found for aroC and aroB by 30 min (Fig. 3D). These results validate lplA and aroC as genes immediately induced by GrgA and further

Gene name
Description Transcripts of hrcA and its cotranscribed genes grpE and dnaK were increased by 10 min, although the increases in grpE were less pronounced with only trending significant adjusted P values. In panels B to F, data are averages plus standard deviations (error bars). All P values were adjusted for multiple comparison. Upregulation detected by RNA-seq at 0.5 h after ATC treatment (Data Set S3) is presented as a black bar for comparison.

Transcription repressor Polymorphic outer membrane protein
suggest that aroL and aroDE, which lie within the same operon as aroB and aroC, are also induced by GrgA though we did not perform qRT-PCR for either aroL or aroDE. The apparent higher detection sensitivity of qRT-PCR analysis over RNA-seq ( Fig. 3C and D) prompted its use to determine whether certain genes whose expression did not increase until 1 h might actually be upregulated at 30 min or even earlier. Our inclusion criteria to select genes for qRT-PCR analysis were (i) a read increase with adjusted P , 0.05, and (ii) at least one FPKM (fragment per kilobase of gene length per million reads of the library) value . 900 for induced samples. Twelve genes met these criteria (Data Set S3). An exception was made for hrcA (FPKM = 369 6 66 [mean 6 standard deviation {SD}]), because HrcA is a TF whose RNA was consistently increased by GrgA overexpression in all previous RNA-seq studies conducted with 4-h ATC induction ( Fig. 1 and Data Sets S1 and 2). qRT-PCR analysis detected apparently increased levels for all mRNAs analyzed; however, only the increases in murE and hrcA were statistically significant (adjusted P , 0.05). Further analysis confirmed that the murE RNA increased at 30 min but not 10 or 20 min after ATC induction (Fig. 3E), while hrcA RNA readily increased at even 10 min (Fig. 3F). hrcA is in the same operon as grpE (which encodes heat shock protein-70 cofactor) and dnaK (a protein chaperone gene), although dnaK also has an additional promoter for itself (50). Similar to the hrcA gene, grpE and dnaK showed increased expression starting at 10 min, although the increase in grpE was borderline significant (P = 0.073) (Fig. 3F).
In addition to analyzing expression at different times after 10 nM ATC induction, we investigated euo and hrcA expression in response to multiple low concentrations of ATC using qRT-PCR. This analysis demonstrated dose-dependent increases in both euo and hrcA expression as early as 30 min after ATC induction (Fig. S2). Taken together, the time-and dose-kinetic results presented in Fig. 3 and Fig. S2 demonstrate that four nonoperon genes (euo, pmpI, murE, and ctl0758) and nine additional genes (ctl0536, lplA, aroL, aroC, aroB, aroDE, hrcA, grpE, and dnaK) in three operons are upregulated by 10-to 30-min induction of GrgA overexpression. Most likely, these 13 genes comprise GrgA's direct regulon.
GrgA stimulates transcription from euo, hrcA, and pmpI promoters in vitro. The earlier obtained RNA-seq data from CtL2/GrgA, CtL2/GrgADs 66 BD, and CtL2/GrgADs 28 BD ( Fig. 1 and Data Set S2) support the notion that GrgA activates genes with s 66 promoters as well as genes with s 28 promoters. To provide further evidence for this proposition, we searched for s 66 and s 28 promoter elements in the promoter regions of the four nonoperon genes and three operons that were upregulated by 30 min of GrgA overexpression (Fig. 3). Among the four nonoperon genes (i.e., euo, pmpI, murE, and ctl0758) and three operons (i.e., the ctl0536-lplA operon, aroLCBDE operon, and hrcA-grpE-dnaK operon) upregulated by GrgA overexpression at 10 to 30 min (Fig. 3), hexameric nucleotide sequences resembling the 210 elements of C. trachomatis s 66 (51,52) were readily recognized 6 to 9 bp upstream of the perspective transcription start sites previously identified for euo, hrcA, murE, and aroL (49,51,52), whereas hexameric nucleotide sequences resembling putative 235 elements of C. trachomatis s 66 promoters (53) were readily found 17 or 18 nucleotides upstream of the putative 210 elements for euo, hrcA, and aroL. The core promoter sequences, including the putative 210 and 235 elements, the spacing sequences between the elements, and the transcription start sites of euo and hrcA are shown in Fig. 4A. In addition, an octameric nucleotide sequence resembling the 210 element of C. trachomatis s 28 promoter (16, 51, 52) was readily found 8 nucleotides upstream of the transcription start site previously identified for pmpI, while another octameric nucleotide sequence resembling the 235 elements of C. trachomatis s 28 promoters (51, 52, 54) was detected 11 nucleotides upstream of the 210 element (Fig. 4A). Next, we constructed three in vitro transcriptional reporter plasmids, of which two carried a putative s 66 promoter of euo or hrcA. The third plasmid carried the putative s 28 promoter of pmpI. All three plasmids were able to direct RNA synthesis, with increased transcripts detected in the presence of GrgA (Fig. 4B). These results confirm that euo, hrcA, and its cotranscribed genes grpE, dnaK, and pmpI are direct targets of GrgA. They provide further support for the notion that GrgA activates transcription from not only the s 66 promoters but also the s 28 promoter in C. trachomatis.
Identification of genes commonly regulated by GrgA, Euo, and HrcA. The consistent upregulation of euo and hrcA by GrgA in both the transcriptomic kinetic studies ( Fig. 3 and Data Set S3) and 4 h induction RNA-seq studies (Data Sets S1 and S2), as well as GrgA-mediated activation of transcription from the euo and hrcA promoters in vitro (Fig. 4), imply the existence of a GrgA-Euo-HrcA transcriptional regulatory network (TRN). As a first step to understand this TRN, we derived CtL2 transformants that carried an inducible plasmid to express either Euo (CtL2/Euo) or HrcA (CtL2/HrcA). RNAseq studies were then conducted to identify transcriptomic changes in these transformants following ATC induction between 12 and 16 hpi. The processed RNA-seq data of these RNA-seq studies are presented in Data Set S4. qRT-PCR analysis was performed for five genes with significantly increased RNA-seq reads and five genes with significantly reduced RNA-seq reads from each transformant following ATC treatment. The qRT-PCR data (Data Set S4D) were consistent with the RNA-seq data (Data Set S4B and S4C).
We identified 27 genes that are commonly regulated in all three transformants (i.e., CtL2/GrgA, CtL2/Euo, and CtL2/HrcA) following ATC treatment ( Fig. 5 and Data Set S4E). Eight of these genes were commonly upregulated ( Fig. 5A and B), while 19 genes were commonly downregulated ( Fig. 5C and D). Noticeably, four of the eight commonly upregulated genes encode proteins involved in DNA replication, while the remaining four encode proteins with various functions (Fig. 5B). Among the 19 commonly downregulated genes, 13 encode functionally known proteins, 5 encode hypothetical proteins, and 1 encodes a glutamic acid tRNA (Fig. 5D). Three (PPA, IspH, and FabI) of the 13 functionally known proteins are enzymes that catalyze metabolic reactions; another three (RpsA, RplM, and RpmB) are ribosomal proteins. OmpA and CTL0887 are the major and a minor component, respectively, of the outer membrane protein complex. At least three proteins are related to type III secretion (T3S): YajC is a T3S protein translocase, CTL0299 is a T3S chaperone, and CTL0874 is virulence factor secreted by the T3S system. Among the remaining two proteins, CTL0476 is an inclusion membrane protein secreted through a yet-undefined mechanism; CTL0742a is a polysaccharide deacetylase. These commonly upregulated and downregulated genes may serve as active regulators of RB growth. GrgA-, Euo-, and HrcA-controlled transcriptional regulatory network. Next, we extracted differentially regulated genes from the RNA-seq data sets generated from experiments with biological duplicates or triplicates (Data Sets S2 to S4). The extracted RNA-seq data (Data Set S5) were combined with qRT-PCR data ( Fig. 3 and Fig. S3) to elucidate GrgA TRNs. We used STRING to generate the TRNs because STRING can integrate the GrgA TRNs with previously identified functional association networks (55). A GrgA TRN within 10 to 30 min of ATC treatment was developed using qRT-PCR data obtained from CtL2/GrgA cultures treated with ATC for 10 to 30 min (Fig. 3). In this network, GrgA upregulated 13 genes and downregulated 8 genes (Fig. 6A). Among the 13 GrgA-upregulated genes, euo, hrcA, and pmpI were demonstrated as direct target genes of GrgA by in vitro transcription from their perspective promoters (Fig. 4). In addition, grpE and dnaK, both with functions in protein folding and stress response, are also considered direct targets of GrgA because they are cotranscribed with hcrA (49, 50, 56-59) and upregulated following GrgA overexpression (Fig. 3). As for the remaining eight upregulated genes, which are likely GrgA targets, four genes in the aroLCBDE operon encode proteins in the chorismate synthetic pathway, which is    (Fig. 3), were used to generate the TRN using STRING v11. Light blue and red nodes are genes upregulated and downregulated by GrgA, respectively. Black lines connect functional or physical associations identified by the STRING v11 database. Solid and dashed lines connect GrgA to upregulated and downregulated genes, respectively. (B) Interregulatory relationship among grgA, euo, and hrcA deduced following 15-min and 30-min ATC treatments of transformants. qRT-PCR data obtained from biological triplicates of CtL2/GrgA (Fig. 3B and F), CtL2/Euo (Fig. S3A), and CtL2/HrcA (Fig. S3B) were used to manually generate the TRNs. Arrows signify (Continued on next page) upstream of the tryptophan synthetic pathway catalyzed by trpB and trpA. Paradoxically, expression of the tribal operon is downregulated in RNA-seq ( Fig. 6A and Data Set S5C).
To determine whether Euo and/or HrcA directly regulate grgA expression and whether euo and hrcA directly regulate each other's expression, we performed qRT-PCR analysis for CtL2/Euo and CtL2/HrcA treated without ATC or with ATC for 15 and 30 min. In ATC-treated CtL2/Euo, grgA was progressively downregulated from 15 to 30 min, whereas hrcA was transiently downregulated at 15 min before it returned to the basal level (Fig. S3A). In ATC-treated CtL2/HrcA, grgA and euo were consistently downregulated at both 15 and 30 min (Fig. S3B). Combined with findings from previous RNA-seq ( Fig. 2A and Data Set S3) and qRT-PCR analyses (Fig. 3B and F), the new qRT-PCR data in Fig. S3 revealed an interregulatory network among grgA, euo, and hrcA (Fig. 6B). While GrgA overexpression consistently upregulated both euo and hrcA, Euo overexpression demonstrated dynamic effects on hrcA. Accordingly, Euo overexpression led to an initial hrcA downregulation but then a complete reversal (Fig. 6B and Fig. S3A). HrcA overexpression consistently resulted in downregulation of both grgA and euo ( Fig. 6B and Fig. S3B).
A GrgA TRN at 1 h of ATC treatment was developed using the RNA-seq data obtained from cultures treated with ATC for 1 h (Data Set S3). In contrast to the TRN developed at 30 min of ATC treatment, a larger number of additional genes were differentially regulated at 1 h. These genes can be classified into 17 functional groups ( Fig. S4 and Data Set S5B). Further, a combined TRN in Fig. 6C was generated from differentially expressed genes detected by RNA-seq of CtL2/GrgA, CtL2/Euo, and CtL2/ HrcA induced with 10 nM ATC for 4 h (Fig. 6C and Data Set S5C). The most striking takeaway from this combined TRN is the downregulation of 28 tRNAs. Of these 28 tRNAs in CtL2/GrgA treated with ATC, 7 were also downregulated in CtL2/Euo treated with ATC. However, four tRNAs downregulated by GrgA overexpression were upregulated by Euo overexpression (Fig. 6C and Data Set S5C).
In summary, network analysis suggests that the direct regulon of GrgA is comprised of 13 genes, which include euo and hrcA (Fig. 6A), and the indirect regulon of GrgA includes approximately 150 genes from a variety of functional groups ( Fig. 6A and C and Fig. S4). Furthermore, our analysis indicates the existence of a dynamic interregulatory network among grgA, euo, and hrcA ( Fig. 6B and C). Interactive versions of Fig. 6A and C and Fig. S4 can be accessed at the Figshare public data repository (https://doi .org/10.6084/m9.figshare.14195447).
Differential developmental expression patterns for GrgA, Euo, and HrcA. The interregulation among GrgA, Euo, and HrcA ( Fig. 6) discussed earlier prompted us to determine the temporal expression pattern of each TF during the developmental cycle in wild-type CtL2 grown in L929 cells. Relative genome copy numbers were used to normalize the expression levels ( Fig. 7 and Fig. S5A). qRT-PCR analysis revealed a low level of grgA mRNA from 0 to 5 hpi, which started to increase at 8 hpi before reaching a peak at 18 hpi ( Fig. 7 and Fig. S5B). Sandwich enzyme-linked immunosorbent assay (ELISA) detected a similar temporal GrgA expression pattern ( Fig. 7 and Fig. S5C). Compared to the GrgA expression pattern, a marked increase in the level of euo mRNA was detected at 1 hpi. The level peaked at 8 hpi, gradually decreasing through 18 hpi before reaching a nadir at 24 hpi that was maintained through the end of the culture period (36 hpi) ( Fig. 7 and Fig. S5D). A similar increase in the hrcA mRNA was also detected at 1 hpi, although the magnitude of change was less than that observed during euo mRNA detection. Similar to euo mRNA, levels of hrcA mRNA reached a peak around 8 hpi. This high level of HrcA expression persisted through 18 hpi before decreasing gradually thereafter ( Fig. 7 and Fig. S5E). The temporal expression patterns of the three TFs in wild-type CtL2 supports the notion that the GrgA-Euo-HrcA TRN is crucial for chlamydial developmental and growth, as will be discussed below.
Effects of GrgA, GrgADr 66 BD, GrgADr 28 BD, Euo, and HrcA overexpression on C. trachomatis growth. The apparent roles that GrgA plays in C. trachomatis transcriptomic expression predict that GrgA overexpression would lead to a chlamydial growth defect. Indeed, treatment of CtL2/GrgA grown in L929 cells with 10 nM ATC between 0 and 35 hpi resulted in apparently smaller inclusions with lower intensities of a red fluorescence protein (RFP) encoded by the mKate gene of the transformed plasmid (Fig. 8A). However, identical treatment of CtL2/GrgADs 66 BD did not cause any observable effects on these growth markers (Fig. 8A). ATC treatment of CtL2/GrgADs 28 BD reduced RFP intensities, albeit the reduction was much less severe than that in CtL2/ GrgA (Fig. 8A). We also demonstrated that treatment of CtL2/GrgA with 10 nM ATC starting at 12 hpi resulted in a 35% reduction in the genome copy number at 16 hpi (Fig. 8B). These findings suggest that delicate regulation of physiological GrgA concentrations is required for adequate CtL2 growth. Further, s 66 -dependent transcriptomic changes are absolutely required for GrgA overexpression-induced inhibition, whereas s 28 -dependent changes also play a significant role.
To probe the contributions of upregulated Euo and HrcA expression to GrgA overexpression-induced growth inhibition, we determined the effects of lower ATC concentrations on genome replication in CtL2/Euo and CtL2/HrcA, respectively. Treatment with subnanomolar ATC from 12 to 16 hpi increased euo and hrcA RNAs nearly 4-fold (Fig. 8C). These degrees of induction were comparable to the euo and hrcA RNA increases induced following 10 nM ATC treatment of CtL2/GrgA cultured in L929 cells ( Fig. 3 and Data Set S2). In parallel, we observed a significant 40% decrease in the genome copy in the CtL2/Euo, and a statistically insignificant 10% decrease in the genome copy in the CtL2/HrcA (Fig. 8D). These data support the notion that dysregulated euo expression contributes to grgA overexpression-induced chlamydial growth inhibition, although an influence of upregulated hrcA expression cannot be ruled out due to the relatively short (4 h) ATC treatment duration. hpi. Genomic DNA (gDNA) was quantified using qPCR (biological triplicates). RNAs of grgA, euo, and hrcA were quantified using qRT-PCR (biological triplicates). GrgA protein was quantified using ELISA (biological duplicates). All expression data were normalized with the amount of gDNA. See Fig. S5 for genome copy number data and expression results of individual genes with error bars.
GrgA overexpression-mediated CtL2 growth inhibition and upregulation of euo and hrcA in infected human cells. All transcriptomic and growth data presented above were obtained using mouse fibroblast L929 cells, a commonly used cellular system for studying C. trachomatis (e.g., references 3, 9, 14, 50, and 60). We next determined the effects of GrgA overexpression on euo expression, hrcA expression, and chlamydial growth using human vaginal carcinoma HeLa cells as the host. As in L929 cells ( Fig. 3B and F), we observed statistically significantly upregulated euo and hrcA transcript levels in HeLa cells following 10 nM ATC treatment for 30 min (Fig. 9A). As was observed in L929 cells (Fig. 8A), we also detected chlamydial growth inhibition in HeLa cells following 10 nM treatment for 35 h (Fig. 9B). These findings indicate that GrgAmediated transcriptional regulation is conserved with either human or mouse cells as the CtL2 host.

DISCUSSION
In this report, we documented the effects of full-length GrgA and domain deletion mutant overexpression on C. trachomatis transcriptomic expression. We identified ATC was added at 0 hpi. Images were acquired at 35 hpi. (B) Genome copy reduction in CtL2/GrgA after treatment with 10 nM ATC for 4 h (from 12 to 16 hpi). Genome copy was determined using qPCR. (C) Moderate increases in euo and hrcA expression in CtL2/Euo and CtL2/HrcA, respectively, after treatment with a low concentration of ATC for 4 h (from 12 to 16 hpi). RNAs of euo and hrcA were quantified using qRT-PCR. Note that mRNA expression values have been normalized with gDNA due to (possible) overexpression-induced growth inhibition (D) Genome copy reduction in CtL2/ Euo but not CtL2/HrcA after treatment with a low concentration of ATC for 4 h. Genome copy was determined using qPCR. In panels B to D, all quantitative data were obtained with biological triplicates. putative direct and indirect regulons of GrgA and uncovered a TRN that encompasses GrgA, Euo, and HrcA. We also revealed temporal expression profiles for grgA, euo, and hrcA. Our findings have important implications for the regulation of chlamydial gene expression and progression of the chlamydial developmental cycle.
Direct and indirect regulons of GrgA. Our RNA-seq and qRT-PCR analyses predict putative direct and indirect regulons of GrgA by detecting time-dependent transcriptomic changes following ATC-induced GrgA overexpression. The direct regulon likely includes 13 genes that are upregulated within 10 to 30 min of ATC treatment (Fig. 3  and 6A). Chromatin immunoprecipitation followed with high-throughput DNA sequencing would be most useful for elucidating the GrgA regulon. In the absence of supporting chromatin immunoprecipitation (ChIP) data, whether or not all 13 genes are directly upregulated by GrgA requires further validation.
Evidence presented in this report reaffirms that GrgA activates both s 66 -and s 28dependent promoters through direct interactions with the two sigma factors, a notion drawn from our previous in vitro studies (39)(40)(41). Full-length GrgA overexpression caused numerous transcriptomic changes, while GrgADs 66 BD overexpression resulted in upregulation of only pmpI expression and downregulation of only four genes ( Fig. 1 and Data Set S2). Additionally, among seven promoter regions upstream of the four nonoperon genes and three operons upregulated within 10 to 30 min of ATC treatment, three have both recognizable 210 and 235 elements of the s 66 -dependent promoter, while one has noticeable 210 and 235 elements of the s 28 -dependent promoter. Finally, the transcription activities of two of the s 66 -dependent promoters (euo and hrcA) and the s 28 -dependent pmpI promoter are stimulated by GrgA in vitro (Fig. 4).
Eight genes are downregulated 30 min after ATC treatment ( Fig. 6A; see also Data Sets S3 and S5A in the supplemental material). These genes are likely subject to indirect rather than direct repression by GrgA. A bacterial TF can activate one of two neighboring genes and repress the other if the genes share an intergenic promoter region (61). Because none of the eight GrgA-downregulated genes is located next to a upregulated gene, the opportunity for direct repression is unlikely. Indeed, RNA-seq data from CtL2/HrcA and CtL2/Euo support possibilities that GrgA downregulates trpB, trpA, and pmpA expression indirectly through HrcA (Data Set S4C) and downregulates ctl0887 and ctl0874 expression through Euo and/or HrcA (Data Set S4B and S4C). Nonetheless, we cannot completely rule out the possibility that GrgA acts as a repressor since some bacterial TFs have dual functions (62,63).
Excluding the 13 putative direct target genes, we consider almost all genes whose RNA levels statistically significantly increased or decreased between 1 h and 4 h as constituents of the indirect regulon of GrgA. It is important to point out that our approach of classifying direct and indirect target genes based on the induction time needed for an increase to be observed could be an overly simplistic approach because it precludes the recognition of direct target genes whose promoters have low affinities for GrgA and thus require longer time to be activated.
Because GrgA is an activator of euo and hrcA, GrgA's indirect regulon is expected to be much larger than its direct regulon. Nearly 150 target genes were identified by 4 h when an arbitrary threshold of 1.33-fold change was applied. This approximation likely underestimates the true size of the regulon for a number of considerations. Technically, we find that RNA-seq (Data Set S3) is not as sensitive as qRT-PCR (Fig. 3) for the detection of mRNA expression changes, a phenomenon also noted recently by Soules et al. in their CtcC transcriptomic study (14). Also, as discussed above, physiological targets that show less than 1.33-fold change are excluded.
Chlamydial growth and development controlled by the GrgA-Euo-HrcA TRN. RNA-seq and qRT-PCR analyses performed for CtL2/GrgA, CtL2/Euo, and CtL2/HrcA suggest that the GrgA, Euo, and HrcA regulate each other's expression. Whereas GrgA serves as an activator of both euo and hrcA, Euo likely represses grgA and hrcA directly. HrcA also represses grgA and euo ( Fig. 6A and B). These observations are consistent with GrgA's role as a transcription activator and Euo and HrcA's roles as transcription repressors. However, a transient hrcA downregulation following Euo overexpression ( Fig. 6B and Fig. S3A) highlights the highly complex nature of transcriptional regulation in the RB, which likely involves other transcriptional regulators.
The rapid and substantial increases in the RNAs of euo and hrcA in chlamydiae within 1 hpi ( Fig. 7 and Fig. S5D and E) suggest that Euo and HrcA as well as GrpE and DnaK (other products of the hrcA-grpE-dnaK operon) play critical roles in immediate early chlamydial development. Even though grgA is not transcribed until 8 hpi, we hypothesize that the GrgA protein prepackaged within EBs is critical for the activation of euo and hrcA immediately after EBs enter host cells, based on the interconnected relationship among grgA, euo, and hrcA (Fig. 6). Although our GrgA protein expression curve indicates that on average, an RB at 18 hpi (when the GrgA level peaks) contains about 5 times more GrgA molecules than an EB ( Fig. 7 and Fig. S5C), the concentration of GrgA in the invading EB would be about 8 times higher than that in the RB since the RB is nearly 40 times larger in volume (3). Previously, multiple groups also detected significant amounts of GrgA protein in EBs (39,64,65). Therefore, the GrgA protein expression profile supports the possibility that GrgA activates euo and hrcA after DNA decondenses during the EB-to-RB transition. The inverse correlation between Euo and GrgA expression between 8 and 18 hpi likely reflects the suppression of grgA by Euo, which is indicated by the interregulatory network (Fig. 6B).
The large increase in GrgA expression during the midcycle (Fig. 7) suggests that GrgA plays important roles in RB replication. Given the large number of genes in its indirect regulon (Fig. 6), GrgA likely fulfills its function as a growth regulator through balanced actions of its direct and indirect target genes with roles in biosynthesis (e.g., polA [DNA polymerase] and glmM [phosphoglucosamine mutase in peptidoglycan biosynthesis]), metabolism (e.g., atpI [ATP synthase]), and other processes (e.g., pknD [serine/threonine protein kinase]). It is conceivable that GrgA downregulates tRNA genes and other genes to coordinate the transition of RBs to EBs in later developmental stages.
We identified 8 commonly upregulated and 19 commonly downregulated genes in CtL2/GrgA, CtL2/Euo, and CtL2/HrcA undergoing growth arrest due to ATC-induced overexpression of respective TFs (Fig. 5 and Data Sets S2B and 4). Their dysregulated expression may contribute to or result from chlamydial growth inhibition. Paradoxically, four of the eight commonly upregulated genes encode proteins involved in DNA replication and repair: topoisomerase I, DNA polymerase III, DNA helicase, and a site-specific tyrosine recombinase (XerD). Interestingly, all these four genes are upregulated during gamma interferon (interferon-g)-induced chlamydial persistence when growth is also reduced to a certain degree. Equally interesting is the observation of downregulated trpBA expression following GrgA overexpression (Fig. 6A and C and Data Sets S2, S3, and S5) because trpBA is activated in response to interferon-g that causes tryptophan degradation in host human cells (23,66,67). In addition, RNAs of ct505 (cotranscribed with GrgA [ct504 in C. trachomatis D] from the same operon) and euo are increased in response to interferon-g treatment (68). We postulate that GrgA could contribute to the increased euo expression and the growth inhibition observed during interferon-g-induced persistence, which would require further study to confirm (68)(69)(70).
Numerous tRNAs are downregulated following GrgA midcycle overexpression (Table 1; Fig. 6C; see also Fig. S4 and Data Sets S1, S2, S3, and S5 in the supplemental material). Among the tRNAs downregulated by GrgA overexpression, several are also downregulated by Euo overexpression, but others are upregulated by Euo overexpression ( Fig. 6C and Data Sets S4 and 5). While the underlying mechanism is unclear, tRNA downregulation almost certainly contributes to growth inhibition. It is probable that expression of chlamydial tRNAs is temporarily delayed in the immediate early developmental stage and also downregulated later as RBs convert to EBs.
In summary, we have identified 13 genes that are likely direct targets of GrgA, the most recently discovered TF in Chlamydia. By activating expression of two major transcription factors (Euo and HrcA) and by regulating expression of numerous additional genes with functions in almost all cellular processes, GrgA acts as a master transcriptional regulator that controls chlamydial growth and development.

MATERIALS AND METHODS
Plasmids. Plasmids used for this study are listed in Table 2. An NEBuilder HiFi DNA assembly cloning kit was used to generate all new plasmids. Expression vectors for Euo and HrcA were constructed by replacing the green fluorescence protein (GFP) gene in pASK-GFP-L2-mKate2 with the appropriate DNA fragments. In vitro transcriptional reporter plasmids pMT1125-Peuo and pMT1125-PhrcA were constructed by cloning an euo promoter fragment (nucleotides 835937 to 835773, GenBank accession number NC_010287.1) and an hrcA promoter fragment (nucleotides 768239 to 768421), respectively, into pMT1125 (58). pMT1125-PpmpI(s 28 ) was constructed in two steps. First, the putative s 28 -dependent (nucleotides 321618 to 321396) and fragments were cloned into the XbaI and EcoRV sites of pMT1125 using T4 DNA ligase. Second, the guanine nucleotide inside the EcoRV site was deleted using Q5 sitedirected mutagenesis kit.
Plasmids constructed were Sanger sequenced at Genscript or Psomagen to confirm sequence authenticity. For chlamydial expression vectors, sequencing analysis covered the CtL2 plasmid-encoded genes, ATC-inducible promoter, and tet repressor-coding sequence, in addition to the coding sequences of TFs or their deletion mutants. Promoter fragments and the reporter cassette in pMT1125-derived vectors were sequenced.
Microscopic analysis of inclusions. L929 or HeLa cells grown on six-well plates were inoculated with EBs of transformants. Following centrifugation at 900 Â g and room temperature for 20 min, cells were switched into above-described medium without or with 10 nM ATC and incubated at 37°C. Phase contrast and RFP images were acquired at 35 hpi (73). The Java-based ImageJ software was then used to quantify RFP intensities (4). C. trachomatis pmpI s 28 -dependent promoter reporter plasmid This study Cellular gDNA and RNA isolation. Induction of GrgA, GrgA domain deletion mutants, Euo, and HrcA expression in C. trachomatis transformant-infected L929 or HeLa cells was performed with 10 nM ATC, unless indicated otherwise. Total host and chlamydial genomic DNA (gDNA) and RNA were isolated using TRI reagent (catalog no. 93289; Sigma), which separates DNA and RNA into different phases. DNA and RNA were purified in accordance with the manufacturer's instructions (74). gDNA was dissolved in a buffer containing 0.1 M HEPES and 8 mM NaOH. These samples were stored at 220°C. RNA was dissolved in diethyl pyrocarbonate (DEPC)-treated H 2 O and further treated with RNase-free DNase I to eliminate residual DNA contamination. Concentrations of purified gDNA and RNA samples were determined using Qubit DNA HS (high-sensitivity) (catalog no. Q32854) and RNA HS (catalog no. Q32855) assay kits, respectively (Thermo Fisher). The resultant DNA-free RNA samples were stored at 280°C.
Quantitative PCR (qPCR) and reverse transcription-qPCR (qRT-PCR). Thermo Fisher QS5 qPCR machine was used for qPCR and qRT-PCR analyses to quantify relative CtL2 genome copy numbers and mRNA levels, respectively. Genomic qPCR was performed using Applied Biosystems PowerUp SYBR green master mix (catalog no. A25742; Thermo Fisher Scientific) following the manufacturer's instructions. For each reaction, 5 ng of purified total host and bacterial gDNA was used as the template. The primers for genomic qPCR were qPCR-ctl0631-F (59CGCGCACGGTTTATTGGTTT39) and qPCR-ctl0631-R (59AAATAGGCCCGTGTGATCCT39). qRT-PCR was performed using Luna Universal one-step qRT-PCR kit (catalog no. E3005E; New England BioLabs [NEB]) following the manufacturer's instructions. For each reaction, 10 ng of purified total host and bacterial RNA was used as the initial template for cDNA synthesis. All genomic and qRT-PCRs were performed in technical duplicates or triplicates. qRT-PCR data of the developmental expression study were normalized with gDNA due to increases in the number of chlamydial cells from midcycle toward late cycle points. qRT-PCR data obtained from experiments with 4-h ATC treatments were also normalized with gDNA due to overexpression-induced growth inhibition in CtL2/GrgA and CtL2/Euo. No normalization was performed for qRT-PCR data obtained from short-term (10 min to 1 h) ATC treatments because qPCR demonstrated no change in the genome copy number during the treatments.
RNA sequencing. Total RNA integrity was determined using Fragment Analyzer (Agilent) prior to RNA-seq library preparation. Illumina MRZE706 Ribo-Zero Gold Epidemiology rRNA removal kit was used to remove mouse and chlamydial rRNAs. Oligo(dT) beads were used to remove mouse mRNA. RNA-seq libraries were prepared using Illumina TruSeq stranded mRNA-seq sample preparation protocol, subjected to the quantification process, pooled for cBot amplification, and sequenced with the Illumina HiSeq 3000 platform with 50-bp single-read sequencing module. On average, 20 to 25 million reads were obtained for each RNA-seq sample. Short read sequences were first aligned to the CtL2 chromosome (accession number NC_010287.1) and the transformed plasmids using TopHat2 aligner, quantified for gene expression by HTSeq to obtain raw read counts per gene, and then converted to FPKM (fragment per kilobase of gene length per million reads of the library) (75)(76)(77). DESeq, a statistical method appropriate for analyzing a small number of RNA-seq replicates (76), was used to normalize data and find group-pairwise differential gene expression based on three criteria: P , 0.05, average FPKM . 1, and fold change $ 1. Genes were clustered into groups based on temporal patterns of transcriptomics using Gaussian mixture models (48).
In vitro transcription assay. Chlamydial RNA polymerase holoenzyme was partially purified from RBs of pTRL2Dgfp-transformed CtL2 using heparin agarose (Sigma) as previously described (39). In vitro transcription assays for s 66 -dependent promoters and s 28 -dependent promoters were performed as previously described (39,40).
Quantification of the GrgA protein. Expression of the GrgA protein during the C. trachomatis developmental cycle was quantified using ELISA. The plates (24-well plates) were seeded with L929 cells. On each plate, four wells were inoculated with wild-type CtL2 434/BU at a multiplicity of infection (MOI) of one inclusion-forming unit (IFU) per two cells. The plates were centrifuged for 20 min at 900 Â g and washed five times thereafter with 100 mg/ml heparin in HBSS to remove free and cell surface-bound EBs. They were cultured at 37°C with medium containing 1 mg/ml cycloheximide. At the intended time (0, 1, 3, 5, 8, 12, 18, 24, 30, or 36 hpi), a plate was removed from the incubator. Cells from two infected wells and two mock-infected wells were harvested in 1% sodium dodecyl sulfate (SDS) (200ml/well) and immediately heated for 10 min at 100°C. After centrifugation (20,000 Â g, 10 min), supernatants were collected and stored in aliquots at 220°C. gDNA was extracted from the remaining two infected wells and further purified as described above.
The capture antibody for the GrgA ELISAs was an antigen-affinity purified rabbit anti-GrgA antibody (81). It was diluted with 0.1 M NaHCO 3 to a final concentration of 2 mg/ml and added (50 ml/well) to ELISA plates (catalog no. 505021; Nest Scientific). After an overnight incubation at 4°C, the plates were washed four times with phosphate-buffered saline (PBS) to remove residual capture antibodies and blocked thereafter with 10% fetal bovine serum (FBS) at room temperature for 1 h. Samples of stored cell extracts were thawed and brought to homogeneity again by gentle pipetting. Cell extracts harvested between 0 to 8 hpi were diluted with equal volume of H 2 O to yield 0.5% SDS after dilution. Those harvested at 12 hpi were diluted fivefold using 0.3% SDS. Those harvested at 18, 24, 30, and 36 hpi were diluted 150-, 500-, 1,500-and 1,500-fold, respectively, with 0.5% SDS. Diluted cell extracts in triplicate were added to ELISA plates (100 ml/well) after removal of FBS and two washes. The plates were incubated on a shaker for 1 h at room temperature (RmT), washed five times with PBS, and sequentially reacted with 1:1,000 diluted polyclonal mouse anti-GrgA (39), and 1:1,000 diluted peroxidase-conjugated goat anti-moue IgG (catalog no. A4416; Sigma) (1 h per reaction with five washes after each). Developing solution was prepared fresh by adding a 0.1 M TMB (3,39,5,59-tetramethylbenzidine) solution (catalog no. 806336; Sigma) and 3% H 2 O 2 with a phosphate citrate buffer (0.2 M Na 2 HPO 4 , 0.1 M citric acid [pH 5.0]) to yield final concentrations at 2.5 mM and 0.02% H 2 O 2 , respectively. The developing solution was added onto the ELISA plates (100 ml/well). After 30-min incubation at RmT, 100 ml of 2 N H 2 SO 4 / well was added to stop the reaction. Absorbance at 450 nm was obtained on a Spectramax iD5 plate reader (Molecular Devices). The concentrations of GrgA in the infected-cell extracts were calculated using standard curves obtained with 0 to 1,000 pg/ml purified recombinant GrgA (39,40) after removing signals generated with mock-infected cell extracts. Genome copy numbers obtained by qPCR for corresponding time points were used to normalize the GrgA expression level.
Statistical analysis. qRT-PCR, in vitro transcription data, and RFP intensity were analyzed using t tests in Excel of Microsoft Office. When applicable, P values were adjusted for multiple comparisons by Benjamini-Hochberg procedure to control the false discovery rate.
Data availability. RNA-seq data have been deposited into the NCBI Gene Expression Omnibus under accession number GSE153747.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. DATA SET S1, XLSX file, 0.