Expression of Chlamydia trachomatis genes encoding products required for DNA synthesis and cell division during active versus persistent infection

Authors


  • The first two authors contributed equally to this work.

*For correspondence. E-mail ahudson@med.wayne.edu; Tel. (+1) 313 993 6641; Fax (+1) 313 577 1155.

Abstract

During persistent infection, the intracellular bacterial pathogen Chlamydia trachomatis is viable but severely attenuates the production of new, infectious elementary bodies (EBs). To investigate the reasons for this lack of new EB output, we analysed the expression of chlamydial genes encoding products required for DNA replication and cell division, using in vitro models of active versus persistent infection and synovial tissue samples from patients with chronic Chlamydia-associated arthritis. Hep-2 cells were infected with K serovar C. trachomatis and harvested at t = 0–48 h post-infection (p.i.; active). Human monocytes were infected similarly and harvested at t = 1–7 days p.i. (persistent). RNA preparations from infected/uninfected cells and patient samples were subjected to reverse transcription–polymerase chain reaction (RT–PCR) targeting polA, dnaA, mutS and parB mRNA, related to chlamydial DNA replication/segregation; these were expressed in infected Hep-2 cells from 11 to 48 h p.i.; ftsK and ftsW, related to cell division, were expressed similarly. Real-time PCR analyses demonstrated that significant accumulation of chlamydial chromosome began at about 12 h p.i. in infected Hep-2 cells. In infected human monocytes, polA, dnaA, mutS and parB mRNA were produced from days 1–7 p.i. and were weakly expressed in patient samples. Real-time PCR indicated the continuing accumulation of chlamydial chromosome during the 7 day monocyte infection, although the rate of such accumulation was lower than that occurring during active growth. However, transcripts from ftsK and ftsW were detected only at 1 day p.i. in infected monocytes but not thereafter, and they were absent in all patient samples. Thus, genes whose products are required for chlamydial DNA replication are expressed during persistence, but transcription of genes whose products are required for cytokinesis is severely downregulated. These data explain, at least in part, the observed attenuation of new EB production during chlamydial persistence.

Introduction

The intracellular bacterial pathogen Chlamydia trachomatis undergoes a biphasic developmental cycle. In the first phase, the infectious extracellular form of the organism, the elementary body (EB), locates and binds to an appropriate eukaryotic host cell. After binding, the organism is brought into a membrane-bound cytoplasmic vesicle, within which it reorganizes into the metabolically active form, the reticulate body (RB). During this intracellular phase, RBs undergo several rounds of cell division, at the termination of which 70–80% of dividing RBs reorganize back to the EB form. These new EBs are released from the host cell via either lysis or exocytosis and then go on to find other host cells to propagate infection (reviewed by Wyrick, 1998; Hackstadt, 1999). Research from many groups has modified this standard view of the chlamydial developmental cycle. That is, it is now clear that, under some conditions, C. trachomatis undergoes persistent infection, during which the normal cycle is altered, and the organism does so in vivo in disease states and in in vitro model systems (e.g. Moulder, 1991; Beatty et al., 1994a,b; for a review, see Köhler et al., 1998). Influential studies based on observations using infected host cells in culture treated with either penicillin or, in more recent publications, interferon gamma (IFNγ), have proved to be important in this regard (Beatty et al., 1994a). In those studies, it was shown that C. trachomatis in cytoplasmic inclusions within IFNγ-treated host cells assume an aberrant morphological form. Persistent organisms also show unusual transcriptional characteristics and attenuate the production of new EBs (for a review, see Inman et al., 2000).

Urogenital infection with C. trachomatis is associated with the development of reactive arthritis (ReA; Köhler et al., 1998; Inman et al., 2000). Initial attempts to establish the aetiological relationship between chlamydial infection and arthritis used culture of the organism from synovial fluid or synovial tissue. Methods for this have been established for many years, and some investigators did successfully isolate the bacterium from the joints of arthritis patients (e.g. Schachter et al., 1966); however, other attempts were unsuccessful (e.g. Gordon et al., 1973; Keat et al., 1986), engendering the consensus that Chlamydia-associated ReA results from a sterile immune-mediated inflammatory process involving only chlamydial antigens in the joint. However, studies from several laboratories showed that, in addition to protein antigens from the organism, C. trachomatis DNA is present in the joints of patients with Chlamydia-associated arthritis (e.g. Bas et al., 1995; Branigan et al., 1996). Moreover, morphologically aberrant chlamydial forms, similar to those known from in vitro systems, are present in synovial tissue of many ReA patients who are polymerase chain reaction (PCR) positive for the organism (Beutler et al., 1994; Nanagara et al., 1995). This suggested that persistent chlamydiae are the dominant form of the organism in the joints of such patients, and that the culture negativity of synovium results from the attenuated EB production characteristic of persistent C. trachomatis (Beatty et al., 1994a,b; Köhler et al., 1998; Inman et al., 2000). One group provided data indicating that persistent, morphologically aberrant synovial chlamydiae are metabolically active (Gérard et al., 1998a). Persistent, metabolically active C. trachomatis have been found in fallopian tubes from women with fertility deficits as well, suggesting that this unusual biological state is a general aspect of chlamydial pathogenesis (e.g. Gérard et al., 1998b).

Some biological attributes of persistent C. trachomatis, as well as aspects of the pathogenic process engendered by the organism, have been elucidated. In situ hybridization of synovial tissue from patients with Chlamydia-associated ReA indicated that extracellular organisms are virtually absent in those tissues; rather, the majority of organisms are intracellular, and all display the aberrant morphology mentioned above (e.g. Beatty et al., 1994a; Nanagara et al., 1995). This is also the case for C. trachomatis undergoing persistent infection of normal human monocytes in culture (e.g. Köhler et al., 1997) and in Chlamydia-infected HeLa and other cell types treated with IFNγ (Beatty et al., 1994a). In both the in vivo disease context and the in vitro models of persistence, new EBs are not formed or are generated at only an extremely low level. Persistent C. trachomatis cells show an unusual transcriptional profile for several important genes. In the in vitro system of Chlamydia-infected human monocytes and in synovial tissue from patients with Chlamydia-associated ReA, expression of omp1, encoding the C. trachomatis major outer membrane protein (MOMP), is downregulated (e.g. Beatty et al., 1994a; Köhler et al., 1997; Gérard et al., 1998a,b). This is surprising as, during normal growth, mRNA from this gene accounts for as much as 30% of chlamydial protein synthesis (Stephens, 1994); the absence of MOMP in persistent organisms probably explains the aberrant morphology both in vivo and in vitro (Beatty et al., 1994a,b; Köhler et al., 1997). Transcription of hsp60, encoding a highly immunogenic protein, is upregulated in persistent C. trachomatis, explaining in part synovial inflammation in patients with Chlamydia-associated ReA (Beatty et al., 1994a; Köhler et al., 1997; Gérard et al., 1998a,c); upregulation of hsp60 and downregulation of omp1 also occur in persistent Chlamydia in fallopian tubes (Gérard et al., 1998b).

Although some understanding exists concerning the molecular biology of persistent C. trachomatis, it is not known currently why so few or no new infectious EBs are produced by organisms in that state. We and many others have argued that persistent organisms are arrested at some point late in the life cycle, presumably before or at the point at which RB to EB reorganization is initiated (Moulder, 1991; Beatty et al., 1994a; Gérard et al., 1998c). Importantly, a recent analysis of C. trachomatis-infected, IFNγ-treated cells in vitro showed that the aberrantly shaped, persistent bacteria accumulate properly segregated copies of the bacterial chromosome in the absence of subsequent cell division (Beatty et al., 1995). Here, we provide a molecular explanation for that accumulation of chlamydial chromosome copies and for the attenuated production of EBs during persistence, in both an in vitro model system and in vivo in synovial tissues from patients with chronic Chlamydia-associated arthritis.

Results

Accumulation of chlamydial chromosome copies during active infection in vitro

Hep-2 cells can be used as an in vitro host for the growth of C. trachomatis. That is, in this host cell type, C. trachomatis undergoes normal EB to RB reorganization within the host inclusion, active growth of RBs and normal RB to EB dedifferentiation at the end of the developmental cycle; in this system, that cycle requires about 50 h for completion. As a control for the reverse transcription (RT)–PCR studies described below, we used a real-time PCR assay system to monitor the accumulation of chlamydial chromosomal DNA over time during 48 h post-infection (p.i.) in Hep-2 cells infected with K serovar C. trachomatis. The representative assays given in Fig. 1 indicated a slight increase in chlamydial chromosomal DNA as early as 6–10 h p.i., suggesting that some low-level replication of bacterial DNA had been initiated by this time; however, levels of such DNA began to increase significantly at 12 h p.i., somewhat later than reported for C. trachomatis LGV-434 (serovar L2) growing in HeLa cells (Shaw et al., 2000; see Discussion).

Figure 1.

Representative results from quantitative PCR assays to assess the relative level of accumulation of chlamydial DNA over time during active infection. Hep-2 cells were infected with K serovar C. trachomatis as described in Experimental procedures, and flasks were harvested at the times indicated. Total DNA was prepared, and real-time PCR assays were used to determine the relative level of Chlamydia chromosomal DNA at each time point. Input into each quantitative assay was normalized to the host 18S rRNA genes. Data are expressed as fold increase in Chlamydia DNA over time relative to the value obtained at 2 h p.i.; ≈ 105 chlamydial genome equivalents are represented by that initial value.

Expression of DNA replication- and cell division-related genes during active infection in vitro

In order to determine the time p.i. at which chlamydial DNA replication- and cytokinesis-related genes are expressed under conditions of normal growth, we assessed transcripts from a panel of representative genes whose products are required for each process, using as starting material total RNA from Hep-2 cells infected with C. trachomatis as above. DNA sequence information from the chlamydial genome project identified homologues of the Escherichia coli dnaA, polA and mutS genes, all of whose products are involved in chromosomal DNA replication or repair (Stephens et al., 1998). Specifically, the dnaA product is involved in initiation of replication (e.g. Vinella and D'Ari, 1995); the C. trachomatis chromosome encodes two dnaA proteins (ct250 and ct275), both of which appear to be functional but share limited homology (see below). polA encodes the chlamydial DNA polymerase I, and mutS specifies a DNA mismatch repair protein (reviewed by Bremer and Churchward, 1991; Donachie, 1993). The sequencing project also identified chlamydial homologues of the ftsK, ftsW and parB genes; the product of the latter gene is involved in plasmid partitioning/segregation after replication, whereas the first two genes specify proteins required for cell division (Bremer and Churchward, 1991; Donachie, 1993). Using DNA sequence information from the C. trachomatis genome project, primers were designed for RT–PCR analyses targeting mRNA from each of these genes (Table 1).

Table 1.  Primers used for RT–PCR analyses of C. trachomatis mRNA.
adt1 5′-CGACAACAACGGCTAGCATTAGTGGTG-3′(Outer)
 5′-AACCTAAGATGAGCATGGGCGAAAGC-3′(Outer)
 5′-TGGGGAACTTCTCTTTCTGGACAGGAGT-3′(Inner)
 5′-CCCAAAGCAGCAACTAAACCTGTAGCAT-3′(Inner)
dnaA 5′-CCGTTTAAGGAGACTCAGATCCAGCAAG-3′(Outer)
 5′-TATTAAAGGACTAACGGGAATCGAAATC-3′(Outer)
 5′-TCCCCCACTCAAAGCGACTAATTAATC-3′(Inner)
 5′-TCACTATGAACTATGGGGACATTCATC-3′(Inner)
polA 5′-TATTCCTGTGCGTTCTGAAAGAGGGAG-3′(Outer)
 5′-TGTGTAATAAATTCTGCTGCTAGAGGATAAC-3′(Outer)
 5′-CCGCTGATTATTCCCAGATTGAGTTACG-3′(Inner)
 5′-CCATACGCCTGCTGTCCATACACTAAG-3′(Inner)
mutS 5′-ACGTTTCTTGGGCTTTCGAGTATCAGTC-3′(Outer)
 5′-CTGAGCATAACAGACTTTGAGTTTCTTAATG-3′(Outer)
 5′-TGAGAGAGCATAGAGGGGGAGCAATCA-3′(Inner)
 5′-CACACTATCCGAATACGTTTCTTGGGC-3′(Inner)
parB 5′-GAGTTAATAGCTGGTGAGCGTCGTTGGC-3′(Outer)
 5′-TCCATCTCTGCTTCGCGTACTGCTAACT-3′(Outer)
 5′-GGCAGCAGAGGCGACTTTAATAGA-3′(Inner)
 5′-CCTTTCACTAGACGTTTAGCCTCCATCTC-3′(Inner)
ftsK 5′-TCCTGTGATCACATCTCGAGAAGGGCG-3′(Outer)
 5′-GCTTCTAGCATCCGCATCATCGCCCCT-3′(Outer)
 5′-GAGCTAAGCGTACAATAGGCGTTTCGA-3′(Inner)
 5′-CTCCCCCTACAGACATAAAACTCGTAATC-3′(Inner)
ftsW 5′-CTTGGATTGGGACTTGGTATCGCTTC-3′(Outer)
 5′-CGATTCTGAACATAAGGGAGCCGATAGG-3′(Outer)
 5′-GCTAAGCGTTGGCTAGGAGTGGGGCAG-3′(Inner)
 5′-GCAAAGGAAGCAGCCAATAGCGTAATC-3′(Inner)

Previous RT–PCR-based studies have shown that transcription of the chlamydial rRNA operons is initiated early in the EB to RB reorganization process in active infection of HeLa cells (Gérard et al., 1997; Shaw et al., 2000). As shown in Fig. 2, RT–PCR analyses targeting primary rDNA transcripts in infected Hep-2 cells indicated that the production of these molecules had begun by 5 h p.i. (Fig. 2A), as expected. Transcription of adt1, which specifies an ATP/ADP exchange protein (Stephens, 1998; see also Hatch et al., 1982), and omp1, encoding the chlamydial MOMP, also began at an early point after infection (Figs 2B and C). Messengers from chlamydial genes, whose products are required for DNA replication/repair (dnaA, polA and mutS;Fig. 2D–F respectively), and post-replication DNA partitioning (parB;Fig. 2G) were identifiable in actively growing chlamydia by 11 h p.i., as was mRNA from genes whose products are required for cytokinesis (ftsK and ftsW;Figs 2H and I respectively). Expression of all these genes continued up to 48 h p.i. and was not completely terminated at that time because not all RBs dedifferentiate back to EBs. In Fig. 2D, transcripts from the dnaA gene designated ct250 are shown, but the same results were obtained in assays targeting the dnaA designated ct275 (data not shown); transcripts from the chlamydial minD gene, whose product is required for chromosome segregation, appeared in our assays at the same time as did parB transcripts (data not shown). Thus, initiation of measurable expression for chlamydial genes whose products are required for DNA replication, partition and cell division immediately preceded the time at which chromosome copies began to accumulate significantly during active infection of Hep-2 cells.

Figure 2.

Representative RT–PCR analyses targeting C. trachomatis mRNA encoding products required for DNA replication/segregation and cell division over time during active infection in vitro. Hep-2 cells were infected with K serovar, harvested at the times indicated p.i., and RNA was prepared from each such harvested culture, all as given in Experimental procedures. RT–PCR analyses were performed using primers given in Table 1 and Gérard et al. (1998a). RT–PCR analyses are shown targeting: (A) chlamydial primary rDNA transcripts; (B) the adt1 transcript; (C) the omp1 transcript; (D) the dnaA transcript; (E) the polA transcript; (F) the mutS transcript; (G) the parB transcript; (H) the ftsK transcript; and (I) the ftsW transcript. In each panel, the C+ lane is a positive PCR control for each primer set, using C. trachomatis DNA as amplification template. The C– lane is a negative RT–PCR control, using cDNA from uninfected Hep-2 cells as amplification template. The RT– lanes are negative controls showing the results of PCR with the omp1 primer set in the absence of reverse transcription for all RNA preparations used; a different RNA preparation was assessed in each panel, with the exception of that from 3 h-infected cells, which showed no signal in any assay. The ACT lane shows the amplification product from an RT–PCR assay targeting host actin mRNA, used to ensure the quality of the RNA preparation analysed; as with the RT– lanes, a different RNA preparation was assessed in each panel. Input into each assay was normalized to the host actin transcript. Amplification product sizes are: actin, 510 bp; primary chlamydial rRNA transcripts, 429 bp; adt1, 193 bp; omp1, 745 bp; dnaA, 501 bp; polA, 208 bp; mutS, 630 bp; parB, 324 bp; ftsK, 351 bp; ftsW, 274 bp.

Accumulation of C. trachomatis chromosome during persistent infection in vitro

To investigate the accumulation of chromosomal DNA in persistent Chlamydia, the real-time PCR assay system was used to assess bacterial chromosome levels over several days p.i. in persistently infected human monocytes in culture; this system reflects all the known molecular and morphological aspects of persistence known from in vivo studies of patients with Chlamydia-associated ReA. As indicated by the representative data from such assays shown in Fig. 3,C. trachomatis underwent DNA replication during persistent infection in the monocyte system, although the accumulation of chromosome copies was slower than that shown by the organism during active infection of Hep-2 cells (compare Fig. 3 with Fig. 1). Importantly, accumulation of bacterial chromosome continued well beyond 3 days p.i., although previous studies using the monocyte model have shown that C. trachomatis assumes the aberrant morphological form, attenuates EB production and displays the unusual transcriptional characteristics that characterize persistence by this time (e.g. Köhler et al., 1997; see also below).

Figure 3.

Representative results from quantitative PCR assays to assess the relative accumulation of chlamydial DNA over time during persistent infection in vitro. Normal human monocytes were prepared and infected with K serovar C. trachomatis as described in Experimental procedures, and flasks were harvested at the times indicated. Total DNA was prepared, and real-time PCR assays were used to determine the relative level of Chlamydia DNA at each time. Input into each quantitative assay was normalized to the host 18S rRNA genes. Data are expressed as fold increase in Chlamydia DNA over time relative to the value obtained at 1 day p.i.; ≈ 104 chlamydial genome equivalents are represented by that initial value.

Expression of chlamydial DNA replication- and cell division-related genes during persistent infection in vitro

As with persistent C. trachomatis in vivo in the human synovium, the organism does not produce titratable levels of new EBs in the monocyte model of persistence (Köhler et al., 1997; Gérard et al., 1998c). Because the data shown in Fig. 3 indicated that replication of bacterial DNA continued even after the point when EB production has been shown to be highly attenuated in infected monocytes, synthesis of chlamydial transcripts for the panel of genes representative of those involved in DNA replication and cell division was assessed. Representative results from RT–PCR assays are shown in Fig. 4. These data indicated that, as in previous studies, intracellular persistent Chlamydiae were viable and metabolically active during long-term monocyte infection; that is, primary transcripts from the organisms' rRNA operons were identifiable at all times assessed from 1 to 7 days p.i. (Fig. 4A). Interestingly, transcripts from adt1 were also found at all times assayed throughout the extended course of this experiment (Fig. 4B), but omp1 expression became highly attenuated at day 3 p.i. and thereafter as in previous studies (Fig. 4C; Gérard et al., 1998c). mRNA from the chlamydial polA, dnaA, mutS and parB genes was clearly identifiable at day 1 p.i. (Fig. 4D–G), consistent with the observation that chromosome replication and partition do occur in persistent organisms in this in vitro model system; transcripts from all these genes remained plentiful even at 7 days p.i., the most extended time assayed. Importantly, transcripts from both ftsK and ftsW were present in total RNA prepared from monocytes harvested 1 day after infection (Figs 4H and I), but we could identify no transcripts from either of these cytokinesis-related genes in this system at 2 days p.i. or later. This suggests that, although DNA replication/segregation continued at some level during persistence, chlamydial cell division was abrogated largely or completely soon after infection in monocytes. In turn, these data suggest that attenuated production of new EBs during persistent C. trachomatis infection results from low-level or non-expression of bacterial genes required for the cell division process, once persistence is established.

Figure 4.

Representative RT–PCR analyses targeting C. trachomatis genes whose products are required for chromosomal DNA replication/segregation and cell division, as a function of time during persistent infection in vitro. Normal human monocytes were infected with the K serovar, harvested at the times indicated p.i., and RNA was prepared from each such harvested culture, all as given in Experimental procedures. RT–PCR analyses were performed using primers given in Table 1 and Gérard et al. (1998a). RT–PCR analyses are shown targeting: (A) chlamydial primary rDNA transcripts; (B) the adt1 transcript; (C) the omp1 transcript; (D) the dnaA transcript; (E) the polA transcript; (F) the mutS transcript; (G) the parB transcript; (H) the ftsK transcript; and (I,) the ftsW transcript. In each panel, the C+ lane is a positive PCR control for each primer set, using C. trachomatis DNA as amplification template. The C– lane is a negative RT–PCR control, using cDNA from uninfected Hep-2 cells as amplification template. The RT– lanes are negative controls showing the results of PCR with the omp1 primer set in the absence of reverse transcription for each RNA preparation; (F–I) repeat the RT– controls for the day 1–5 samples. The ACT lane shows the amplification product from an RT–PCR assay targeting host actin mRNA, used to ensure the quality of the RNA preparation analysed; a different cDNA preparation is shown in each blot. Input into each assay was normalized to the host actin transcript. Amplification product sizes are as given in the legend to Fig. 2.

Expression of chlamydial DNA replication- and cell division-related genes in synovial tissue

As mentioned, in previous studies, the monocyte model of C. trachomatis persistence has exhibited all the characteristics shown by the organism in vivo in patients with Chlamydia-associated ReA. To assess whether chlamydiae in that tissue display the transcriptional repression of cell division-related genes identified in the in vitro model, mRNA from these genes was assayed in RNA from synovial biopsies from eight patients PCR positive in that tissue for C. trachomatis; as a control, we assessed RNA from synovial tissues of two patients with psoriatic arthritis (PsA), both of whom were PCR negative for the organism. As shown in Fig. 5, primary rDNA transcripts were identifiable, as were transcripts from adt1, in each relevant patient RNA preparation (Figs 5A and B); as expected, mRNA from the chlamydial omp1 gene was not demonstrable in any patient sample (Fig. 5C). RT–PCR targeting chlamydial genes whose products are required for DNA replication/partitioning also indicated that, as in the in vitro model system, those transcripts were present in most patient sample (Fig. 5D–G). Importantly, however, we could identify no transcripts from ftsK or ftsW in any patient samples tested (Figs 5H and I), suggesting that, as in the monocyte model, production of these mRNA was severely attenuated or absent in persistent C. trachomatis in vivo. As expected, samples from the PsA patients, both of whom were PCR negative for the organism, were negative for all chlamydial transcripts assayed.

Figure 5.

Representative RT–PCR analyses targeting C. trachomatis genes whose products are required for chromosomal DNA replication/segregation and cell division in RNA prepared from synovial tissues of patients PCR positive in that tissue for the organism, plus two control patients. Samples were procured and RNA prepared from each as given in Experimental procedures. RT–PCR analyses were performed using primers given in Table 1 and Gérard et al. (1998a). Patients 1–8 were PCR positive for C. trachomatis; diagnoses given are: ReA, reactive arthritis; UM, undifferentiated monoarthritis; UO, undifferentiated oligoarthritis; PR, palindromic rheumatism. Patients 9–10 were PCR negative for the organism; both had psoriatic arthritis (PsA). RT–PCR analyses are shown targeting: (A) chlamydial primary rDNA transcripts; (B) the adt1 transcript; (C) the omp1 transcript; (D) the dnaA transcript; (E) the polA transcript; (F) the mutS transcript; (G) the parB transcript; (H) the ftsK transcript; (I) the ftsW transcript; and (J) the host actin gene. In each panel, the C+ lane is a positive PCR control for each primer set, using C. trachomatis DNA as amplification template. The C– lane is a negative RT–PCR control, using cDNA from uninfected Hep-2 cells as amplification template. The RT– lanes in the various panels are negative controls showing the results of PCR with the omp1 primer set in the absence of reverse transcription for each patient RNA preparation studied. Input into each assay was normalized to the host actin transcript. Amplification product sizes are as given in the legend to Fig. 2.

Discussion

The basic biphasic developmental cycle of C. trachomatis has been understood for many years, but numerous studies have established that, when this organism disseminates from its site of primary infection, it can depart from that normal cycle to engage in persistent infection. Moreover, increasingly cogent evidence indicates that persistence is a relatively common aftermath of active genital infection with the organism, and that persistent C. trachomatis infections engender significant clinical consequences at sites to which the organism disseminates. This bacterium displays unusual characteristics when in the persistent state, including attenuated production of new EBs (Beatty et al., 1994a; Inman et al., 2000). In the work presented here, we investigated the molecular basis of this aspect of persistent chlamydia.

Earlier electron microscopic studies indicated that, during persistent infection of human monocytes in culture, a limited increase in C. trachomatis cell numbers occurs within cytoplasmic inclusions (Köhler et al., 1997). That is, at early times p.i., chlamydial inclusions in the monocyte cytoplasm usually contain a single bacterial cell, whereas at later times, more organisms are often present in those inclusions; this number does not increase significantly after about 48 h p.i., and it is after this time that EB production becomes highly attenuated (Köhler et al., 1997). The observation that DNA replication-/segregation-related genes, and those required for cell division, are expressed during early phases of infection in this in vitro model of persistence provides an explanation for the initial accumulation of chlamydiae, and the lack of transcripts from ftsK and ftsW after about 24 h p.i. explains the termination of cell division at those later times. Moreover, data from another group has indicated that C. trachomatis undergoing IFNγ-induced persistent infection in vitro accumulate fully replicated and partitioned chromosome copies within the aberrant intracellular form of the organism (Beatty et al., 1995). The results given here, showing differential expression of gene products required for chlamydial DNA replication/partition and cell division, probably explain that observation, even though the in vitro model systems studied are somewhat different. We are currently assessing mRNA from our panel of representative genes in the IFNγ-based model of persistence. As developed below, the results given here may also explain some aspects of chlamydial pathogenesis that have until now been poorly understood.

Chlamydia-associated ReA is an inflammatory disease, and many laboratories have shown that DNA from the organism, as well as some protein antigens, are present in the joints of patients with the disease (e.g. Keat et al., 1986; Schumacher et al., 1988; Beutler et al., 1994; Bas et al., 1995; Branigan et al., 1996). A puzzling aspect concerning Chlamydia-associated arthritis has been the culture negativity of synovial tissue, the primary site for long-term persistence of the organism in patients with this disease (Branigan et al., 1996). For ReA induced by enteric bacteria, including that associated with species of the genera Salmonella, Yersinia, Campylobacter and others, it is clear that the organisms rarely reach the joint in an intact, viable state (Granfors et al., 1989; 1990; Hammer et al., 1990; Merilahti-Palo et al., 1991; but see Gaston et al., 1999). This is certainly not the case for C. trachomatis. The organism is viable during persistent synovial infection, as well as during similar infection of the fallopian tubules in women with ectopic pregnancy, despite the frequent culture negativity of those tissues (Gérard et al., 1998a,b). Moreover, in the case of at least some patients with Chlamydia-associated ReA, the organism remains viable in the synovium after extended and aggressive antibiotic treatment (Beutler et al., 1997). Indeed, some data indicate that treatment of chlamydial infection with specific antibiotics induces persistence, rather than clearing the organism (Dreses-Werringloer et al., 2000). The results given here explain, at least in part, the earlier observation from in situ hybridization studies that extracellular, infectious EBs cannot be found in synovial tissue (Beutler et al., 1994) and the consequent culture negativity of chronically C. trachomatis-infected joint materials. Lack of cytokinesis after chromosome replication would preclude organisms attaining the end-stage of the chlamydial developmental cycle. Thus no, or extremely few, new infectious EBs would be formed and released from synovial host cells.

It is of interest that an earlier, similarly structured study of C. trachomatis L2-434-infected HeLa cells indicated that chromosome accumulation was apparent by about 3 h p.i. in that system, and that expression of dnaE, encoding the α-subunit of the bacterial DNA polymerase, was initiated just before that time (Shaw et al., 2000). The data presented here indicate that accumulation of chromosome in serovar K-infected Hep-2 cells begins in earnest somewhat later, ≈ 12 h p.i., slightly after the identifiable initiation of expression of genes whose products are required for DNA replication and partition. Although the method used in that earlier study for the analysis of bacterial chromosome accumulation (quantitative competitive PCR) was different from that used here, we suspect that the difference in observed time of bacterial DNA replication initiation, and in relevant transcript production, in L2-434 versus serovar K is not simply a function of the analytical techniques used. Rather, the early start to the DNA replication process seems likely to be an underlying factor in the far more severe and aggressive pathogenesis elicited by the LGV strain compared with that of the standard genital serovar. More study will be required to resolve this issue.

We have not attempted to quantify transcript levels for the chlamydial genes assayed here. Nonetheless, it is unlikely that the ftsK and ftsW transcripts were missed in these analyses in either the monocyte system or the patient samples. Control studies (not shown) indicated that the relative sensitivities of the RT–PCR assay systems used are approximately equivalent and, in the monocyte model, we easily identified amplification products from the parB, dnaA, mutS and polA cDNA after the first round of nested PCR, whereas we found no such products for ftsK and ftsW even after the second, nested round. Similarly, in the patient samples analysed, the signal was unequivocal for DNA replication-/repair- and chromosome partition-related cDNA amplification products after nested assays using RNA prepared from most samples, although the majority of such samples were not universally RT–PCR positive for each of the dnaA, polA, mutS and parB transcripts; we presume that those biopsies negative for one, but not all, these DNA replication- and segregation-related mRNAs simply possessed extremely low levels of that absent transcript. The important point is that we could identify no products derived from the cytokinesis-related mRNA in any patient sample examined, supporting the contention that the expression of these two cell division-related genes is severely attenuated, even if it is not completely abrogated, during persistent C. trachomatis infection.

Data given here confirm that transcription of omp1, encoding the chlamydial MOMP, is selectively downregulated during persistent infection, both in vitro in infected monocytes and in vivo in synovial tissues of patients with Chlamydia-associated ReA (Gérard et al., 1998a,c). The results also indicate that expression of ftsK and ftsW is similarly attenuated or abrogated during persistence. The control by which transcript initiation is governed in C. trachomatis during normal EB to RB reorganization, standard growth or RB to EB dedifferentiation is currently only poorly understood. More immediately germane to the topic at issue, the mechanism(s) controlling selective transcriptional downregulation of omp1, ftsK and ftsW during persistence is not understood. A computer-assisted analysis of the 5′ portion of the coding sequence, plus the 5′ flanking regions, for omp1, ftsK and ftsW did not identify any commonly held DNA sequence that might function as a binding site for a repressor molecule. Similarly, no sequence 5′ to dnaA, polA or other genes expressed during persistence was identified that might function selectively to trans-activate transcriptional initiation in the manner of the E. coli crp gene product at the lac operon. The C. trachomatis genome encodes three σ factors for its RNA polymerase, specified by genes designated rpoD, rpsD and rpoN (Stephens et al., 1998). A recent study indicated that these genes are expressed at different times during the chlamydial developmental cycle, with the first two expressed early and rpoN expressed relatively late (Mathews et al., 1999). We developed RT–PCR assays targeting transcripts from each of these genes, and preliminary analyses of monocytes persistently infected with C. trachomatis indicated that all transcripts are present, even at extended times p.i., when the persistent state is well established. Thus, the simple absence of specific σ factors cannot explain the differential transcriptional governance of the chlamydial genes studied here. We are now assessing quantitatively those various transcript levels over time p.i. in monocytes persistently infected with the K serovar organism using a real-time PCR system.

It is tempting to speculate that some factor, such as IFNγ or tumour necrosis factor (TNF)α, produced by the host in response to the organisms' presence is responsible for the selective transcriptional downregulation observed in the studies presented but, to our knowledge, no direct evidence exists to support such a contention. Indeed, published results indicate that removal of IFNγ and TNFα from the culture medium, among other manipulations, of infected cells does not engender the reactivation of persistent chlamydial infection, or relevant transcriptional upregulation, in the monocyte system (Köhler et al., 1997; Gérard et al., 1998c). Regardless, our working hypothesis is that persistence, with all its various transcriptional consequences, is induced in C. trachomatis via some interaction with or effector from its long-term host cell. We are now using biochemical and molecular genetic approaches to elucidate the details of that interaction.

Experimental procedures

Patient samples

Synovial tissue samples were procured from patients attending the Arthritis Clinics at the Philadelphia D.V.A. Medical Center and the University of Pennsylvania Hospital, Philadelphia, PA, USA. Synovial biopsies were procured by the Parker–Pearson method (Schumacher and Kulka, 1972), and samples were immediately snap-frozen at −80°C until used for the preparation of nucleic acids (see below). Eight patients studied here were included simply because they were known to be PCR positive for chromosomal DNA from C. trachomatis but no other organisms, because most had disease of relatively long duration, and because enough nucleic acids of high quality were present in each sample. Patients in the PCR-positive group had diagnoses of ReA, undifferentiated mono- or oligoarthritis and palindromic rheumatism; diagnoses were made according to the criteria of the American College of Rheumatology (Schumacher et al., 1993). Two patients with psoriatic arthritis were included as controls because they were known to be PCR negative for the organism. Half the patients studied were female, and overall disease durations ranged from 0.2 to 16 years.

Tissue culture

As a correlate to the in vivo persistent synovial chlamydial infection of ReA patients, we adapted an in vitro model system of persistently infected human monocytes (Rothermel et al., 1989), which shows all the biochemical and metabolic characteristics of the organism known from long-term synovial infection in vivo. Briefly, normal human peripheral monocytes were isolated from healthy volunteer donors, infected at a multiplicity of infection (MOI) of 1:1 with C. trachomatis (serovar K) and allowed to grow for several days as described (Köhler et al., 1997; Gérard et al., 1998c); in most such experiments, 10–15% of monocytes become infected with C. trachomatis. At several times p.i., cells were harvested by the standard method, and cell pellets were snap-frozen at −80°C until processed for study as described below. Overall cell number usually declines somewhat over the first 3–4 days p.i. in this monocyte system, and cells were counted at every harvest time to monitor that decline. In our hands, monocytes differentiate to macrophages slowly in this culture system, with a minority of cells having undergone that differentiation by 7 days p.i.; by 11–12 days p.i., a majority of monocytes in the culture have made the transition to macrophages (Köhler et al., 1997). As a control for normal chlamydial growth and gene expression, Hep-2 cells were infected at an MOI of 1:1 with the same chlamydial serovar, and infected cells were harvested at multiple times p.i., as described previously (Gérard et al., 1997); host cell populations were also determined at most harvest times for infected Hep-2 cultures, but no significant change was seen in these determinations. All harvested cell pellets were immediately snap-frozen at −80°C until processed for the preparation of nucleic acids. In Hep-2-based experiments, 30–35% of cells were infected.

Nucleic acid preparation

Total nucleic acids were prepared from synovial tissue samples, infected and uninfected (control) human monocyte cell pellets and cell pellets from infected and uninfected (control) Hep-2 cells via homogenization in hot phenol, then extensive extraction in phenol–chloroform (24:1), as described (Köhler et al., 1997; Gérard et al., 1998a). Pure RNA for RT–PCR analyses was isolated from total nucleic acid preparations by extensive treatment with RNase-free DNase I (RQ1; Promega Biotech), followed by further extractions and precipitation in ethanol (e.g. Harwood, 1996). All RNA preparations were confirmed to be DNA negative by PCR targeting the host actin gene in the absence of reverse transcription (e.g. Gérard et al., 1997). The quality of each RNA preparation (i.e. the level of RNA degradation) was assessed by analysis on ethidium bromide-stained formaldehyde–agarose gels and by RT–PCR targeting the host actin gene (Gérard et al., 1997; 1998c; Harwood, 1996).

RT–PCR analyses

Reverse transcription (RT) was performed using 5 µg of total RNA from each preparation with random hexamers as primers; RT was accomplished by the MuLV enzyme (Life Technologies), as described (Gérard et al., 1997). cDNA from each reaction was treated with RNase A, RNase T1 and RNase H; then, the mixture was extracted with phenol–chloroform (24:1) several times. cDNA was recovered by precipitation in ethanol. RT–PCR for C. trachomatis transcripts used ≈ 10% of cDNA preparations and targeted representative coding sequences specifying proteins required for bacterial DNA replication and cell division. The genes targeted and the primer sequences used are given in Table 1. Primer sequences were designed using the GeneRunner system (Hastings Software), based on information from the chlamydial genome sequencing project; full sequence information for each gene studied here is available at website http://www.stdgen.lanl.gov. Extensive testing of primer sets showed that each nested assay system amplified only the target C. trachomatis sequences intended but nothing within the host genome. Control studies demonstrated that the relative sensitivities of the various assays were comparable, and each was able to identify 10–50 chlamydial cells. Amplification conditions for the first round of the nested reactions were: 4 min at 95°C; then 35 cycles of 40 s at 95°C, 40 s at annealing temperature, 40 s at 72°C; then 10 min at 72°C; annealing temperatures varied somewhat among the several primer sets. For the second, nested amplification round, cycling parameters were the same as above with appropriate annealing temperatures, and 10% of the first-round amplification reaction mixture was used. To assess the viability of infecting chlamydia in all systems, RT–PCR assays targeting primary transcripts from the bacterial rRNA operons were used; this system, including the primers used, has been described previously (Gérard et al., 1997; 1998a). Amplifications were performed using Ampli-Taq DNA polymerase (Perkin-Elmer) in an MJ thermocycler (model PTC-100; MJ Research), and products were visualized on agarose electrophoretic gels via ethidium bromide staining (Harwood, 1996). Representative RT–PCR products were hybridized with appropriate internal probes for chlamydial DNA sequences to ensure the authenticity of those products.

Real-time PCR analyses

To assess the accumulation of chlamydial chromosome over time p.i. during both active and persistent infection in vitro, a highly quantitative real-time PCR assay system designed, tested and characterized by others was used (Mathews et al., 1999). Briefly, the chlamydia-directed primers target the two copies of the 16S rRNA gene on the bacterial chromosome, whereas assay input is normalized to host 18S rRNA sequences. The chlamydia-directed primers were: 5′-GGAGAAAAGGGAATTCACG-3′ and 5′-TCCACATCAAGTATGCATCG-3′ (Mathews et al., 1999). The human 18S-directed primer system used for input normalization was purchased from PE Biosystems and was designed for precisely this purpose. All assays were performed several times, each in triplicate, using a PE Biosystems model 7700 sequence detector with the SYBR green method (Hiratsuka et al., 1999). Data from real-time PCR assays were calculated using the version 1.7 sequence detection software from PE Biosystems.

Acknowledgements

This work was supported by NIH grants AR-42541 (A.P.H.), AI-44493 (J.A.W.-H.) and AR-47186 (H.C.G.), grants from the D.V.A. Medical Research Service (H.R.S. and A.P.H.), a grant from the Arthritis Foundation (J.A.W.-H.) and a grant from the German Ministry of Education and Science (01 GI 9950; L.K.).

Ancillary