Chlamydial infection of polarized HeLa cells induces PMN chemotaxis but the cytokine profile varies between disseminating and non-disseminating strains

Authors


*For correspondence. E-mail pbwyrick@etsu.edu; Tel. (+1) 423 439 6294; Fax (+1) 423 439 8044.

Abstract

While genital infections caused by Chlamydia trachomatis are generally asymptomatic, the density and pattern of inflammation varies considerably. The purpose of this study was to try to dissect the signalling in chlamydiae-infected epithelial cells that triggers innate responses and regulates polymorphonuclear neutrophil (PMN) chemotaxis. Polarized endocervical epithelial HeLa cells, grown in commercial inserts, were inoculated either with the non-disseminating (luminal) serovar E or the disseminating serovar L2. At 12–48 h after infection, the chambers were used in a quantitative chemotaxis assay, and cytokine production by infected cells was examined using cDNA microarray technology and confirmed by enzyme-linked immunosorbent assay (ELISA). Infection of HeLa cells with C. trachomatis E or L2 induced a strong and similar PMN chemotactic response, but larger amounts of interleukin (IL)-8 and IL-11 were released after infection with serovar L2. IL-6 was also produced in modest amounts after infection with either strain, but no IL-1α or tumour necrosis factor (TNF)-α was detected in any of the culture supernatants tested. IL-11 did not appear to influence the PMN response to chlamydial infection, but secretion of large amounts of this anti-inflammatory cytokine, mainly active on macrophages, in the very early stages of the infection may allow C. trachomatis to escape some innate defences to establish infection.

Introduction

The strains of Chlamydia trachomatis causing the majority of urogenital infections in humans belong to non-disseminating serovars D–K. They typically induce asymptomatic lower genital tract infections, such as cervicitis and urethritis, but as chlamydiae escape from the apical domain of infected epithelial cells and ascend canalicularly, chronic infections can evolve with sequelae such as prostatitis, epididymitis, pelvic inflammatory disease, ectopic pregnancy and infertility, especially in the absence of antibiotic treatment ( Stamm and Holmes, 1990). Histopathological analysis of endocervical and endometrial biopsies reveals that primary infection is characterized by a local acute inflammatory response associated with early infiltration of polymorphonuclear neutrophils (PMNs) in the submucosal and mucosal compartments, followed by a subepithelial accumulation of mononuclear leucocytes during the chronic phase ( Patton and Kuo, 1989; Kiviat et al., 1990 ).

Sexually transmitted infections with a somewhat different presentation are caused by the disseminating C. trachomatis biovar Lymphogranuloma venereum (LGV, serovars L1–L3), but only isolated cases resulting from LGV are reported in industrialized countries. The clinical course of Lymphogranuloma venereum is generally divided into three stages. After initial infection of the urethra or cervix, the first stage of the disease is characterized by a transient and often imperceptible primary local lesion. From one to several weeks later, systemic spread of C. trachomatis to the regional lymph nodes occurs, leading to an inguinal lymphadenopathy; most patients recover from this secondary-stage infection without further sequelae. In a few cases, the infection can progress to a tertiary stage that involves the formation of genital ulcers, fistulas and elephantiasis, leading to destruction of the mucosal epithelium and scarring ( Schachter and Osoba, 1983; Goens et al., 1994 ).

It has long been believed that damage from chlamydial infections is immune mediated. For example, in in vitro organ culture, where immunological interactions are excluded, chlamydial infection of fallopian tube produces little or no destruction, whereas repeated inoculations of fallopian tube in vivo organ explant produces severe salpingitis, adnexal adhesions, scarring and tubal occlusion ( Patton et al., 1987; Cooper et al., 1990 ).

Chlamydia is an obligate intracellular bacterium whose primary target host is a mucosal epithelial cell. Entry into genital epithelial cells is thus essential for survival and growth of chlamydiae, as well as a prerequisite for causing disease. However, chlamydial infection can occur, at least in vitro, with surprisingly little damage to its host epithelial cell, suggesting that it is, perhaps, early signalling from the infected epithelial cell that triggers immune surveillance and provokes an inflammatory response. Rasmussen et al. (1998 ; 1997) first reported that infection of non-polarized cervical (HeLa, SiHa and primary endocervical cultures) and colonic (HT-29, SW620) epithelial cells with C. trachomatis LGV or Chlamydia psittaci induced the production of several proinflammatory cytokines, including interleukin (IL)-8, GRO-α, granulocyte macrophage colony-stimulating factor (GM-CSF), IL-1α and IL-6. Subsequently, our laboratory showed that C. trachomatis serovar E infection of polarized human endometrial epithelial cells (HEC-1B) elicited a PMN chemotactic response and that the likely chemoattractant signals were ENA-78 and GCP-2, two members of the C-X-C chemokine family ( Wyrick et al., 1999 ). Interestingly, little IL-8 was detected in the serovar E-infected endometrial cells. These data implied that production of certain chemokines may be tissue specific and that signals resulting from epithelial cells infected with luminal versus systemic chlamydiae may be different.

The purpose of this study was to compare the PMN chemotactic response in vitro to polarized HeLa cells infected with the non-disseminating serovar E or the disseminating serovar L2 and to determine whether differences in PMN transepithelial migration were reflected in chemokine profiles screened by microarray analysis.

Results

Infection of HeLa cells with C. trachomatis induced a PMN chemotactic response

In order to study the PMN response induced by chlamydial infection of endocervical cells, polarized HeLa cells were infected for various times with C. trachomatis serovar E or L2 and examined in chemotaxis assays ( Fig. 1). The results of these chemotaxis experiments are presented in Fig. 2. The number of PMNs migrating in response to control uninfected HeLa cells averaged ≈ 1000, regardless of the time of culture (12–48 h). In the early stages of chlamydial infection, the PMN response was not significantly different from background, with average numbers of 1200 and 950 PMNs migrating to serovar E-infected HeLa cells, and 570 and 1600 PMNs migrating to serovar L2-infected HeLa cells, when tested at 12 and 24 h after infection respectively. However, after 36 h of infection, an increase of 60- to 120-fold in the number of transepithelial migrating PMNs was observed towards HeLa cells infected with both strains, when compared with control uninfected cells (63 000 ± 33 000 PMNs for serovar E; 32 000 ± 24 000 for serovar L2; and 520 ± 120 PMNs for the uninfected negative control). The same pattern of PMN chemotaxis was found at 48 h after infection for cells infected with both strains. While the average number of PMNs migrating in response to serovar E-infected HeLa cells was a little higher than that to serovar L2-infected cells, a slightly higher infectivity titre was found on immunostaining of monolayers infected with serovar E than with L2 (data not shown); importantly, this difference in PMN chemotaxis was not significant by statistical analysis (SD). Thus, for a similar level of infectivity (60–65%), HeLa cells harbouring the non-disseminating serovar E or the disseminating serovar L2 induced a comparable and strong PMN response in our in vitro system; the migrating PMNs represented from 3% to 10% of the total PMN population.

Figure 1.

Schematic representation of the in vitro co-culture model system for PMN chemotaxis. HeLa cells were grown in a polarized orientation on ECM-coated commercial inserts (0.3 cm2 membrane; Matrigel invasion chambers) and then infected with C. trachomatis. At various times after infection, the chamber was inverted, and calcein-loaded PMNs were added at the basal side. Chemotaxis was analysed by the number of fluorescent PMNs migrating from the basal-to-apical chamber after 3 h of incubation.

Figure 2.

PMN chemotactic response to polarized HeLa cells infected with non-disseminating C. trachomatis serovar E or disseminating C. trachomatis serovar L2. Confluent monolayers of polarized HeLa cells, cultivated in commercial inserts precoated with ECM (0.3 cm2), were inoculated with C. trachomatis (60–65% infectivity titre). At the times indicated, 1 × 106 calcein-loaded PMNs were added to the basal domains, and transepithelial neutrophil migration was allowed to proceed for 3 h. The migrating PMNs were collected and counted by fluorescence microscopy; PMN numbers are represented on a log scale. The results shown for C. trachomatis serovar E- (hatched bars) and serovar L2-infected HeLa cells (white bars) and uninfected cells (black bars) represent the mean ± SD from three independent experiments.

Interestingly, the transepithelial PMN migration was not correlated with an upregulation of ICAM-1 expression in HeLa cells following C. trachomatis infection, as examined by fluorescence immunostaining (data not shown).

Infection with C. trachomatis induced expression of IL-6, IL-8 and IL-11 in HeLa cells

To begin to characterize the chemotactic factors potentially released by HeLa cells infected with chlamydial serovars E and L2, the eukaryotic cellular expression of cytokines over the time course of infection was studied at the mRNA level using cDNA array technology and at the protein level by performing enzyme-linked immunosorbent assay (ELISA) on culture supernatants from uninfected and infected cells. In an attempt to correlate the PMN response with cytokine production, the same timed intervals were used for both studies, i.e. 12, 24, 36 and 48 h after infection.

The purpose of using the array technology approach was to obtain a panoramic view of cytokine expression profiles of uninfected and C. trachomatis-infected HeLa cells. A representative example of the results obtained is presented in Fig. 3. To our surprise, IL-11 (spot no. 4) was the only cytokine for which an obvious upregulation of mRNA expression was detected by this method. This upregulation was detected as soon as 12 h after infection and maintained until 48 h after infection, and was much stronger in HeLa cells infected with serovar L2 ( Fig. 3B) than in those cells infected with serovar E ( Fig. 3C). Densitometric analysis of the blots confirmed these observations ( Table 1). At 12 h after infection, a fivefold upregulation in IL-11 mRNA expression was found in HeLa cells infected with serovar E or L2 when compared with uninfected control cells. This upregulation was sustained at the same level (fivefold increase) in HeLa cells infected with serovar E until 36 h after infection then decreased at 48 h after infection to almost the same level as the control (ratio ≈ 1). In contrast, after infection with serovar L2, a progressive increase in the level of IL-11 mRNA upregulation over the time of infection was detected, reaching a level of expression 12.7 times higher than that in uninfected cells ( Table 1).

Figure 3.

Cytokine gene expression profiles by microarray analysis in HeLa cells infected with C. trachomatis serovars E and L2. HeLa cells were seeded in large inserts (44 cm2) coated with ECM and, when confluent, the monolayers of polarized cells were infected with C. trachomatis serovar L2 or E (80–90% infectivity). At 36 h after infection, the cellular total RNA were isolated and used for gene expression analysis in a cDNA array assay. The blots, consisting of cDNA sequences, immobilized in duplicate for 268 human cytokine and cytokine receptor genes and nine housekeeping genes, were exposed to X-ray films and autoradiographed.

A–C. Blots from uninfected, serovar L2- and serovar E-infected HeLa cells respectively. Housekeeping genes, used as positive controls for hybridization, are located at the left of the blots, and the location of ubiquitin, used as a standard for the normalization of the signal intensities, is indicated. The location of IL-1α (spot no. 1), IL-6 (spot no. 2), IL-8 (spot no. 3), IL-11 (spot no. 4) and TNF-α (spot no. 5) hybridization signals are also shown.

Table 1.  Time course of IL-11 mRNA expression in Hela cells during C. trachomatis infection, as detected by the cDNA array method.
 Upregulation of IL-11 mRNA expressiona
Time after inoculationHeLa + serovar EHeLa + serovar L2
  1. a. The autoradiographic signals obtained from the cDNA arrays corresponding to IL-11 mRNA were analysed by densitometry, quantitated as mean intensity values and normalized using ubiquitin as standard. Then, the mean intensity values for infected HeLa cells were compared with the values for control uninfected cells harvested at the same time, and upregulation of mRNA expression following chlamydial infection was evaluated by calculation of the ratio of mean intensity (infected cells)/mean intensity (control cells).

12 h5.35.1
24 h4.86.4
36 h5.38.7
48 h1.412.7

As previous studies reported the detection by reverse transcriptase–polymerase chain reaction (RT–PCR) of IL-8 and IL-1α mRNA in HeLa cells infected with C. trachomatis serovar L2 ( Rasmussen et al., 1997; 1998), we expected to find an upregulation of these genes with the cDNA array method. However, no IL-8 mRNA was detected in polarized HeLa cells infected with either serovar for all times tested. Messenger RNA of IL-6, tumour necrosis factor (TNF)-α and IL-1α was detected in large, average and very low amounts, respectively, in both infected and uninfected control cells ( Fig. 3), and the densitometric analysis failed to detect any differences in the signals related to infection with C. trachomatis (data not shown).

To confirm our results obtained with the cDNA arrays, the culture supernatants from uninfected and infected HeLa cells were tested by ELISA for IL-1α, IL-6, IL-8, IL-11 and TNF-α products ( Table 2). Large amounts of IL-11 protein were detected in all infected cell supernatants tested, and the IL-11 concentration increased over the time of chlamydial infection. At 12 h after infection, the IL-11 concentrations in the culture supernatants of HeLa cells infected with serovar E and L2 were 190 pg ml−1 and 170 pg ml−1, respectively, while the uninfected cells produced only 19 pg ml−1. In the later stages of serovar E infection, only a slight increase in IL-11 concentration was detected (i.e. 300 pg ml−1 at 48 h after infection), whereas L2 infection induced a dramatic increase in IL-11 production by HeLa cells, with the concentration reaching 990 pg ml−1 after 48 h of infection ( Table 2).

Table 2.  Comparison of cytokine induction in HeLa cells infected with C. trachomatis serovar E or serovar L2.
 Concentration (pg ml−1) in basal supernatants
C. trachomatis serovarIL-1αIL-6IL-8IL-11TNF-α
  1. a.–, None detected. The limit of sensitivity of the ELISA kit used was 3.9 pg ml −1 for IL-1α, 3.12 pg ml−1 for IL-6, 31.2 pg ml−1 for IL-8 and 15.6 pg ml−1 for IL-11 and TNF-α.

12 hpi
 Controla6.17.819.0
 HeLa + E8.0190
 HeLa + L213.616.8170
24 hpi
 Control4.58.116.1
 HeLa + E11.08.1180
 HeLa + L217.7101.5390
36 hpi
 Control5.713.224.0
 HeLa + E23.014.5270
 HeLa + L272.6151750
48 hpi
 Control7.011.219.0
 HeLa + E47.043.1300
 HeLa + L243.3147.5990

Despite the fact that no IL-8 mRNA was detected at the times tested with the cDNA array assay, an increase in the protein concentration was found in the supernatants of C. trachomatis-infected HeLa cells. After infection with serovar L2, a significant amount of IL-8 released by HeLa cells was detected at 24 h after infection and peaked at 36 and 48 h after infection to a concentration of about 150 pg ml−1 ( Table 2). In marked contrast, serovar E infection induced a later and more moderate IL-8 production, the concentration being higher than the control only at 48 h after infection (43.1 pg ml−1 versus 11.2 pg ml−1; Table 2). A modest increase in IL-6 concentration over the time of infection was detected for both strains, with peak concentrations for C. trachomatis serovar E-infected cells reaching 47 pg ml−1 at 48 h after infection, whereas peak concentrations for L2-infected cells reached 72.6 pg ml−1 at 36 h after infection and then declined to 43.3 pg ml−1 by 48 h after infection. Lastly, no IL-1α or TNF-α was detected in any of the culture supernatants tested.

At this point: (i) an upregulated IL-11 mRNA expression in infected HeLa cells was correlated with the production of large amounts of the protein, occurred early after infection, was maintained until 48 h after infection and was stronger in cells infected with serovar L2 than in those infected with serovar E; (ii) no IL-8 mRNA was detected with the method used, but increasing moderate and large amounts of IL-8 protein were found in the culture supernatants of HeLa cells infected with serovar E and L2 respectively; (iii) no upregulation of IL-6 mRNA expression was detected in infected HeLa cells with the cDNA array assay, but a similar and moderate increase in the protein amount was found after infection with C. trachomatis serovars E and L2; and (iv) no change in the levels of TNF-α and IL-1α expression following chlamydial infection could be detected at the messenger and protein levels. Thus, it is interesting to note that HeLa cells infected with the C. trachomatis disseminating serovar L2 produced larger amounts of IL-8, a potent PMN chemoattractant, and IL-11, a cytokine with anti-inflammatory properties, but similar amounts of IL-6, than HeLa cells infected with the non-disseminating serovar E.

Effects of antibodies specific for IL-8 and IL-11 on PMN chemotaxis

To substantiate the influence of IL-8 and IL-11 secretion by infected HeLa cells on the PMN response, specific antibodies directed against these two cytokines were added to the culture medium at 12 and 6 h after infection, respectively, and the chemotaxis assay was performed at 48 h after infection. When the modest amount of IL-8 released by HeLa cells infected with serovar E was neutralized by anti-IL-8 monoclonal antibody, a 20% reduction in PMN chemotaxis was found; the effect of neutralization of the large IL-8 release by anti-IL-8 was more dramatic for serovar L2-infected cells, as the PMN response was reduced by 90% ( Fig. 4). This difference in the neutralization/inhibition pattern may be related to the higher IL-8 amount produced by HeLa cells in response to serovar L2 infection compared with serovar E infection ( Table 2). In contrast, no inhibition of PMN migration was detected when an anti-IL-11 neutralizing antibody was exposed to HeLa cells infected with serovars E or L2 ( Fig. 4), despite the large amount of IL-11 released by infected cells. Thus, IL-8 but not IL-11 appeared to be a major chemoattractant factor of PMNs in response to L2 infection.

Figure 4.

Effect of specific antibodies, cycloheximide or dexamethasone exposure on the PMN response to polarized HeLa cells infected with C. trachomatis serovars E and L2. Confluent polarized HeLa cells were inoculated with serovar E (hatched bars) or L2 (white bars) (60–65% infectivity) and, at 6 h after infection for anti-IL-11 or at 12 h after infection for anti-IL-8, specific antibodies were added to the cultures (see Experimental procedures for details). For the experiments using cycloheximide (1 μg ml−1), the drug was added to the HeLa cells 2 h before inoculation (2 h preT) or after the inoculum was adsorbed for 1 h on the monolayers (noted T0), and dexamethasone (10−6 M) was added at T0. At 48 h after infection, the infected cultures were used in a PMN chemotaxis assay. For each set of chlamydiae-infected cells, the average number of migrating PMNs was calculated from the two control samples incubated with no antibodies or any drug, and this value represented the 100% migration. The percentage of PMN chemotaxis (y-axis) compared with this control value was calculated for each set of test samples incubated in the presence of a specific antibody, cycloheximide or dexamethasone. The graph presents the data from a representative experiment.

Effect of inhibitors on cytokine release by C. trachomatis-infected HeLa cells

To clarify the potential influence of secreted HeLa cell signals on PMN migration, chemotaxis assays were performed with HeLa cells cultivated in the presence of cycloheximide, dexamethasone or α-amanitin, chemical reagents used to block chemokine production. However, before application in the chemotaxis experiments, the efficacy of these agents on inhibition of cytokine release by C. trachomatis L2-infected HeLa cells was examined by measuring IL-8, IL-11 and IL-6 concentrations in basal supernatants of exposed and unexposed cultures ( Table 3). When cycloheximide was added to HeLa cells just after inoculation with serovar L2, the production of IL-8 at 48 h after infection was reduced by 46%, whereas a 2 h pretreatment before inoculation led to a decrease of 95% in IL-8 production. A similar reduction in IL-8 concentration was obtained after a 2 h pretreatment of L2-infected cells with α-amanitin ( Table 3), an RNA polymerase inhibitor, suggesting that induction of IL-8 mRNA expression may occur early after initial contact of HeLa cells by C. trachomatis, confirming the previous findings of Rasmussen et al. (1998) . In contrast, cycloheximide exposure of C. trachomatis-infected HeLa cells markedly potentiated IL-6 and IL-11 expression induced by infection. Indeed, when cycloheximide was added to HeLa cells either after inoculation or 2 h before inoculation, strong and moderate increases, respectively, in IL-11 concentration were detected relative to the concentration in unexposed infected cells. It should also be noted that uninfected HeLa cells exposed to cycloheximide produced moderately more IL-11 than unexposed cells ( Table 3). A superinduction of IL-6 release was also detected when cycloheximide was added to HeLa cells after C. trachomatis inoculation, whereas the 2 h cycloheximide pretreatment had almost no effect on IL-6 production. In contrast, when HeLa cells were exposed 2 h before inoculation to α-amanitin, a dramatic reduction in IL-11 and IL-6 release was detected ( Table 3), suggesting that these cytokines may be induced at the transcriptional level following chlamydial infection. A recent study reported overexpression of IL-6 mRNA in cycloheximide-exposed HeLa cells, mediated by activation of the nuclear factor NF-κB ( Faggioli et al., 1997 ) and, as IL-11 also belongs to the gp130 cytokine family, the mechanism of IL-6 and IL-11 superinduction might be similar. At the relatively low concentration of cycloheximide used (1 µg ml−1) in our experiments, translation of the superinduced transcripts may not be completely abolished, thus explaining such increases in the protein amounts.

Table 3.  Effect of inhibitor exposure on cytokine release by HeLa cells infected with C. trachomatis serovar L2 for 48 h.
Inhibitor used a
Per cent of cytokine released in basal supernatant/control b
IL-8Il-11IL-6
  • a. The different inhibitors were added to the cultures after inoculation with serovar L2 (noted T0) or 2 h before inoculation (noted 2 h pre-T), and used at the following concentrations, 1 μg ml −1 for cycloheximide, 10 μg ml−1 for α-amanitin and 10−6 M for dexamethasone.

  • b

    . The amount of IL-8, IL-11 and IL-6 released in the presence of each inhibitor was evaluated as a percentage of the amount found for unexposed cultures that represent the 100%.

  • –, None detected with the ELISA kit used and ND, not done.

Cycloheximide T0
 Uninfected87ND
 HeLa + L254207172
Cycloheximide 2 h pre-T
 Uninfected132ND
 HeLa + L2515894
α-amanitin 2 h pre-T
 Uninfected83ND
 HeLa + L210323
Dexamethosone T0
 UninfectedNDNDND
 HeLa + L22935

As cycloheximide exposure had little effect in abolishing cytokine release by C. trachomatis-infected HeLa cells, infected cultures were subsequently examined, especially for IL-11 and IL-6 production, following exposure to dexamethasone, a glucocorticoid known to modulate cytokine production ( Brattsand and Linden, 1996; Wang et al., 1999 ). At a concentration of 10−6 M, dexamethasone induced a dramatic decrease in IL-8 (71% reduction), IL-11 (97% reduction) and IL-6 (95% reduction) release by L2-infected cells compared with unexposed infected cells ( Table 3).

Differential effects of dexamethasone and cycloheximide exposure of C. trachomatis-infected HeLa cells on PMN chemotaxis

The PMN chemotactic response to HeLa cells infected with C. trachomatis and exposed to dexamethasone was definitely reduced relative to the response found for control unexposed cultures. Indeed, the number of migrating PMNs was reduced by 57% and 73% for serovars E and L2, respectively ( Fig. 4), again confirming the influence of cytokines produced by infected HeLa cells on PMN response, particularly in the case of serovar L2 infection. It is noteworthy that C. trachomatis growth in unexposed and dexamethasone-exposed cultures was similar (data not shown).

Interestingly, PMN migration in response to serovar E- and L2-infected HeLa cells exposed to cycloheximide was about 2.5-fold stronger than the PMN response to control unexposed cultures, regardless of when the cycloheximide was added to the cultures, i.e. either after or 2 h before inoculation ( Fig. 4). Immunofluorescent staining of infected cells substantiated that chlamydial inclusions were much bigger in cultures treated with cycloheximide than in the unexposed, control infected cultures (data not shown). Considering that (i) cycloheximide treatment of infected HeLa cells inhibited IL-8 production; (ii) IL-11 did not influence the PMN response to chlamydial infection; and (iii) enhanced PMN migration was not correlated with IL-6 superinduction in the case of a 2 h cycloheximide pretreatment, it could be speculated that the enhanced PMN chemotactic response to chlamydial infection in the presence of cycloheximide might be related to an enhanced growth of C. trachomatis.

Discussion

Genital infections caused by C. trachomatis are primarily characterized by a moderate inflammatory response that, in the absence of treatment or after repeated infections, tends to become more severe, leading to tissue damage and sequelae. Shortly after infection is initiated, neutrophils are the most predominant effector cells recruited to the infectious foci ( Patton and Kuo, 1989; Kiviat et al., 1990 ), suggesting that PMNs are involved in controlling the early stages of C. trachomatis infection, and confirmed by a study from Barteneva et al. (1996) . The acute host response to Chlamydia in the genital tract is primarily initiated and sustained by epithelial cells, the first and main targets of chlamydial infection ( Rasmussen et al., 1997; 1998; Wyrick et al., 1999 ).

Using a co-culture chamber model, we showed that infection of endocervical HeLa cells with disseminating serovar L2 and non-disseminating serovar E induced a similar PMN chemotactic response, which seemed surprising, considering the difference in virulence of these strains both in vivo and in vitro. As found previously for serovar E-infected endometrial HEC-1B cells ( Wyrick et al., 1999 ), a strong transepithelial PMN migration to infected HeLa cells was detected after 36 h of infection for both strains. However, in this study, analysis of epithelial cytokine production in response to infection and inhibition experiments of PMN chemotaxis showed differences between the C. trachomatis E and L2 infections; IL-8, IL-11 and IL-6 expression were upregulated at the transcriptional level but, for the first two cytokines, higher amounts of protein were found in serovar L2 infection.

IL-8 is one of the most potent C-X-C chemokines and is produced in response to bacterial infection including Chlamydia ( Beatty and Sansonetti, 1997; Burns et al., 1997 ; Rasmussen et al., 1997 ). Although IL-8 could be detected in culture supernatants of L2-infected HeLa cells as soon as 24 h after infection, the PMN response was detectable only at 36 h after infection. When L2-infected HeLa cells were cultivated in the presence of anti-IL-8 antibodies or with dexamethasone, IL-8 activity was suppressed, and the PMN chemotactic response was dramatically reduced, suggesting that, at least for L2 infection, IL-8 is a major PMN chemoattractant. In contrast, the PMN response to HeLa cells infected with serovar E was much less affected by the anti-IL-8 and dexamethasone exposures, which may be related to the relatively moderate amounts of cytokines induced by serovar E infection. These in vitro data tend to correlate with in vivo observations that serovar E infections are more asymptomatic than LGV infections. The apparent contradiction between mRNA and protein analyses for IL-8 and IL-6 may be explained by the limits of the cDNA array technology. If this latter method allows expression analysis of multiple genes simultaneously, it is certainly much less sensitive than RT–PCR used by Rasmussen et al. (1998 ; 1997), and low upregulation is difficult to detect. Moreover, the lack of detection of IL-1α in HeLa cells infected with C. trachomatis for 12–48 h was expected, at least at the protein level, because this cytokine is mainly released at the end of the chlamydial developmental cycle, i.e. after cell lysis ( Rasmussen et al., 1997 ).

The most surprising result from this study was the detection of very large amounts of IL-11 released by C. trachomatis-infected HeLa cells, particularly with disseminating serovar L2. IL-11 production induced by viral or bacterial infections has been reported recently ( Elias et al., 1994; Kernacki et al., 1998 ). IL-11 is a pleiotropic cytokine that exhibits anti-inflammatory effects in a variety of animal models of acute and chronic inflammation, including inflammatory bowel disease, arthritis, inflammatory skin disease and various infection–endotoxaemia syndromes (reviewed by Barton et al., 1996; Schwertschlag et al., 1999 ). The anti-inflammatory properties of IL-11 involve a direct activity on macrophages, independent of other anti-inflammatory mediators, such as IL-10 and TGF-β1 ( Trepicchio et al., 1996 ) . Recombinant human IL-11 (rhIL-11) inhibited TNF-α, IL-1β, IL-12, IL-6 and nitric oxide production from lipopolysaccharide (LPS)-activated macrophages in culture ( Trepicchio et al., 1996 ) through the inhibition of nuclear translocation of NF-κB as a result of an enhanced production of the NF-κB inhibitors, IκB-α and IκB-β ( Trepicchio et al., 1997 ) . Consistent with in vitro data, in in vivo murine models of endotoxaemia or radiation-induced thoracic injury, rIL-11 also downregulated levels of LPS- or radiation-induced TNF-α in serum or lungs respectively ( Redlich et al., 1996 ; Trepicchio et al., 1996 ) . In contrast, rhIL-11 did not affect phagocytosis, chemotaxis or IL-8 production of LPS- and zymosan A-activated PMNs, although IL-11 receptor α chain RNA can be detected in human peripheral blood neutrophils ( Bozza et al., 1998 ). The lack of effect of IL-11 on PMN chemotaxis was implied in this study, as culture of C. trachomatis-infected HeLa cells in the presence of anti-IL-11 neutralizing antibodies did not affect the PMN chemotactic response to infection. The effect of IL-11 on TNF-α release by activated PMNs has not, to our knowledge, been investigated.

Darville et al. (1995) observed that TNF-α could be detected as early as 3 days after infection in genital secretions of female guinea pigs infected intravaginally with the chlamydial guinea pig inclusion conjunctivitis agent. Similar results were observed when TNF-α levels were measured in genital secretions from mice infected with C. trachomatis biovar mouse pneumonitis, but C57BL/6 mice, which produced significantly higher levels of TNF-α, were able to resolve their infection more rapidly than C3H/HeN or BALB/c mice ( Darville et al., 1997 ). Thus, these data suggest that TNF-α may have a protective role in the early host response to chlamydiae, and resident macrophages and, possibly, neutrophils of the genital tract epithelium may constitute the main sources of the early TNF-α response. In this context, very early secretion of large amounts of IL-11 by endocervical epithelial cells following C. trachomatis infection may reduce TNF-α release by macrophages and facilitate the establishment of the infection. In fact, one might argue that the larger amount of epithelial IL-11 induced by serovar L2 compared with serovar E may allow L2 to escape host innate defences for dissemination better.

Interestingly, cycloheximide exposure of HeLa cells enhanced epithelial IL-11 and IL-6 production. The IL-11 and IL-6 genes, like the genes for ICAM-1, IL-8 or GM-CSF ( Wertheimer et al., 1992; Xing et al., 1993 ), belong to the group of superinducible genes for which an enhancement of transcript accumulation is induced by protein synthesis inhibitors, albeit via different mechanisms ( Maier et al., 1993; Faggioli et al., 1997 ; Roger et al., 1998a ). The IL-8 gene that was superinduced by cycloheximide treatment (10 µg ml−1) in lung epithelial H292 cells ( Roger et al., 1998b ) was not overexpressed in C. trachomatis-infected HeLa cells exposed to 1 µg ml−1 inhibitor, at least at the protein level. Rasmussen et al. (1998) showed that, in HeLa cells infected with C. trachomatis serovar L2, the late maximal induction of cytokine resulted from a NF-κB-independent process. Thus, the mechanisms of cytokine induction by chlamydial infection and cycloheximide may involve several transcription factors and appear to be complex.

In addition to effects on macrophages, rhIL-11 also reduces CD4+ T production of Th1 cytokines, such as interferon (IFN)γ induced by IL-12, while enhancing Th2 cytokine production ( Schwertschlag et al., 1999 ). Further, in a murine model of endotoxaemia, rhIL-11 blocked LPS-induced elevation of IFNγ serum levels ( Trepicchio et al., 1996 ) . These effects might be important in the context of chlamydial infection, as Th1 cells and IFNγ are crucial in the resolution of the infection (reviewed by Ward, 1999). We cannot exclude the possibility that IL-11 may also show an autocrine activity on endocervical cells or, at least, interact with other genital cell types, as IL-11 receptor is expressed in the uterus and ovary ( Davidson et al., 1997 ). In addition, elevated IL-11 concentrations were detected in seminal plasma of infertile patients with urogenital infection, but no information about the nature of the infection was available ( Matalliotakis et al., 1998 ).

The reason for using a simple in vitro co-culture system was to discover and dissect the cellular microbiology effects of early epithelial cell signalling following C. trachomatis infection of HeLa cells, as a potential model of endocervical infection. Of course, in the context of an in vivo infection, the cytokines released and the timing thereof by infected genital epithelial cells may be somewhat different. In conclusion, very early production by C. trachomatis-infected epithelial cells of large amounts of the anti-inflammatory cytokine IL-11, combined with the weak potency of C. trachomatis LPS to induce an inflammatory response and a minimal and delayed induction of proinflammatory cytokines may explain, in part, how chlamydiae sustain infection and may delay the initiation of an immune response.

Experimental procedures

Chlamydiae and cell lines

C. trachomatis serovar E/UW-5/CX and C. trachomatis biovar LGV L2/434/Bu were grown in McCoy cells cultivated on microcarrier beads as described previously ( Wyrick et al., 1996 ). The chlamydial harvests were aliquoted and stored at −80°C.

HeLa 229 cells were grown at 37°C in an atmosphere of 5% CO2 in minimal essential medium (MEM) with Earle's salts (Gibco BRL) supplemented with 10% fetal bovine serum, 2 mM glutamine and 10 µg ml−1 gentamicin.

Infected cultures for the chemotaxis experiments

For the PMN chemotaxis experiments, 3 × 104 HeLa cells were seeded in commercial inserts (BioCoat Matrigel invasion chambers, Becton Dickinson Labware) containing a 0.3 cm2 translucent membrane with 8 µm pores, precoated with extracellular matrix (ECM). The inserts were incubated at 37°C in an atmosphere of 5% CO2 until the HeLa cells became confluent and polarized. The culture medium was then replaced by 50 µl of chlamydial inoculum corresponding to a dilution in culture medium of the serovar E and serovar L2 stocks to give 60–65% infectivity. After 1 h of incubation at 35°C, 5% CO2, to permit attachment/entry of chlamydiae, the inoculum was removed and replaced by 750 µl of culture medium. The infected cultures were reincubated at 35°C for times varying from 12 h to 48 h. In all experiments, uninfected HeLa cells were included as a negative control, and a minimum of two culture chamber inserts were prepared per timed assay.

In chemotaxis inhibition assays using specific antibodies, the inoculation step was performed using the same protocol as described above, but at 6 h or 12 h after infection, the culture medium was replaced by 750 µl of medium containing dilutions of the anti-IL-11 or anti-IL-8 antibodies respectively. The inserts were then reincubated at 35°C for 42 and 36 additional hours respectively.

In some experiments, HeLa cells were grown in the presence of (i) cycloheximide (1 µg ml−1), an inhibitor of eukaryotic protein synthesis, which was added to the culture medium either 1 h after adsorption of chlamydiae or 2 h before inoculation; or (ii) dexamethasone (10−6 M) added after inoculation. In these experiments, PMN chemotaxis was evaluated at 48 h after infection, and chlamydial growth was evaluated by immunofluorescence staining for chlamydial inclusions by exposure of infected cells to a pool of fluorescein isothiocyanate (FITC)-conjugated monoclonal antibodies, generated against the chlamydial major outer membrane protein, and suspended in an Evans blue counterstain (Syva).

Preparation of PMNs for the chemotaxis assay

Peripheral blood was obtained from healthy donors, and the PMNs were separated and isolated by the standard Hypaque–Ficoll–dextran procedure ( Wyrick et al., 1999 ). After hypotonic lysis of the red cells, the PMNs were resuspended in Hanks' balanced salt solution (HBSS) and counted in a haematocytometer; PMN purity was > 95% as assessed by Giemsa staining. The PMNs were centrifuged at 500 g for 5 min and then loaded with the fluorescent dye calcein-AM using a protocol derived from previous studies ( Sunder-Plassman et al., 1996; Frevert et al., 1998 ). Briefly, the PMNs were resuspended in culture medium containing 2.5 µM calcein-AM (Molecular Probes) at a final concentration of 5 × 106 cells ml−1 and incubated for 30 min at 37°C. The cells were centrifuged for 5 min, washed once with 5 ml of culture medium and resuspended again in 5 ml of medium. The number of fluorescent PMNs was determined by microscopy, using a FITC filter (488 nm excitation wavelength). The percentage of fluorescent PMNs was > 99%. The PMNs were centrifuged and resuspended in medium at a concentration of 25 × 106 PMNs ml−1 before being added to the chemotaxis chamber inserts.

Chemotaxis assay and quantification of migrating PMNs

From 12 to 48 h after infection, the culture inserts were prepared for the chemotaxis assay as described in Fig. 1. Briefly, the culture medium was removed and replenished with 1 ml of fresh medium to prevent drying of cells. A round coverslip was sealed with grease to the top of the chamber, the insert was inverted, transferred to a 12-well plate and an O-ring was sealed at the bottom of each insert. Then, 1 × 106 calcein-loaded PMNs were added to the basal side of the chamber insert, and the inserts were incubated for 3 h at 37°C to allow for PMN migration. Afterwards, the non-migrating PMNs were removed by wiping the basal side of the insert with a paper towel, followed by rinsing with HBSS. The coverslip was removed and the medium discarded. The HeLa cell monolayers and migrating PMNs were collected by pipetting up and down in the presence of 100 µl of a dispase solution (Becton Dickinson Labware). The cell suspension was washed with HBSS, centrifuged and resuspended in 100 µl of HBSS. This suspension, containing HeLa cells and fluorescent, chemotactic PMNs, was transferred to a 96-well microtitration plate, and a 10-fold dilution series in HBSS was performed. PMNs present in the wells were counted by inverted fluorescence microscopy (50× magnification), and the total number of migrating PMNs per insert was calculated.

Antibodies

The assays for inhibition of PMN migration were performed using an anti-hIL-8 mouse monoclonal antibody (Genzyme Diagnostics) and an anti-hIL-11 goat polyclonal neutralizing antibody (R and D Systems).

Culture for the cytokine production and analysis

The production of cytokines by HeLa cells after infection with C. trachomatis was analysed at the mRNA level (cDNA microarrays) and at the protein level (ELISA). Because mRNA analysis required large numbers of host cells, HeLa cells were grown in 44 cm2 inserts (Costar). The inserts were coated with 500 µl of ECM (Matrigel; Becton Dickinson Labware), diluted 1:3 in culture medium, and 4 × 106 HeLa cells were seeded onto the solidified matrix. After 3 days of incubation, the confluent monolayers of polarized cells were inoculated with 1 ml of C. trachomatis serovar E or L2 inoculum, chosen to give 80–90% infection in HeLa cells. The culture inserts were incubated at 35°C for 1 h, and the inoculum was distributed every 15 min by gently shaking the dishes. Then, the inoculum was replaced by 10 ml of culture medium, and the inserts were incubated for varying times. At 12, 24, 36 and 48 h after infection, the basal supernatants were collected, aliquoted and stored at −80°C for ELISA analysis; infected and uninfected control cells were collected in cold PBS by scraping the insert membranes with a rubber policeman, centrifuged at 500 g for 5 min at 4°C and used immediately for total RNA isolation.

In some experiments, cytokine synthesis was inhibited by culture of HeLa cells in the presence of cycloheximide, α-amanitin (10 µg ml−1; Sigma) or dexamethasone, as described above, and at 48 h after infection, the amounts of IL-8, IL-11 and IL-6 in the basal supernatants were determined by ELISA.

RNA isolation and cDNA arrays

Total RNA from uninfected or infected HeLa cells was isolated using the Atlas Pure Total Labelling System kit (Clontech Laboratories), according to the recommended procedure and including a DNase I treatment. After a precipitation step in the presence of 2 M NaOAc/96% ethanol and a wash with 80% ethanol, the RNA pellet was resuspended in RNase-free water at a final concentration of 1–2 µg µl−1. Total RNA concentration and purity were evaluated by absorbance reading at 260 and 280 nm, and RNA integrity was confirmed on a denaturing agarose gel. The mRNA was then isolated from 45 µg of total RNA with biotinylated oligo-dT and streptavidin-magnetic beads, and used to synthesize the probes for the cDNA arrays with the Atlas Human cDNA Expression Array Cytokine/Receptor kit (Clontech). Briefly, a mixture containing 1 µl of 10× CDS primers (Clontech) and the 6 µl bead/polyA+ complex was heated for 2 min at 65°C and then cooled for 2 min at 50°C. A volume of 13.5 µl containing 4 µl of 5× reaction buffer, 2 µl of 10× dNTP, 5 µl of [α-32P]-dATP (10 µCi µl−1, 3000 Ci mmol−1; Amersham Pharmacia), 0.5 µl of dithiothreitol (100 mM) and 2 µl of Moloney Murine Leukaemia Virus reverse transcriptase (50 units µl−1) was added, and the mixture was incubated for 25 min at 50°C. The reaction was stopped by the addition of 2 µl of 10× termination mix at room temperature. The reverse-transcribed sample was subsequently purified on a Chroma Spin-200 column (Clontech) in a total elution volume of 300 µl. The activity of the labelled probe, measured in a scintillation counter (Beckman LS 6500), was generally 1–10 × 106 counts per minute (c.p.m.).

For each cDNA array experiment, two RNA samples were run in parallel, one corresponding to the infected HeLa cells (with serovar E or L2), and the other corresponding to the uninfected control cells; both samples were harvested at the same time. Importantly, the microarray membranes used in each experiment belonged to the same lot. The Atlas cDNA Expression Array membranes contain cDNA sequences, immobilized in duplicate, specific for 268 human genes of cytokines and cytokine receptors. The cDNA corresponding to nine housekeeping genes and to plasmid sequences are also included as positive and negative controls respectively. Hybridization was performed with the same number of c.p.m. in both samples, i.e. labelled cDNA derived from uninfected and infected HeLa cells. A mixture containing 5 µl of human Cot-1 DNA (1 µg µl−1; Clontech) and the labelled cDNA was boiled for 2 min and immediately cooled on ice. The denatured, labelled cDNA was then added to the Atlas cDNA Expression Array membrane, which was prehybridized for 30 min at 68°C in 5 ml of ExpressHyb hybridization solution containing 0.5 mg of denatured salmon sperm DNA (Gibco BRL). The hybridization proceeded at 68°C overnight with continuous agitation. Membranes were washed stringently four times for 30 min each at 68°C with 200 ml of prewarmed solution 1 (2× SSC, 1% SDS) followed by one final wash with 200 ml of prewarmed solution 2 (0.1× SSC, 0.5% SDS). Afterwards, membranes were rinsed for 5 min at room temperature in 2× SSC and exposed to X-ray film (BioMax MS; Kodak) for various times (overnight to 3 days) at −80°C. The two membranes run in parallel were exposed in the same autoradiographic cassette to limit the variations in signal intensities resulting from exposure conditions.

Semi-quantitative analysis of gene expression

The X-ray films were scanned with a ScanMaker 5 scanner (Microtek Lab), and the level of gene expression was analysed using the Kodak 1D image analysis software (Kodak Digital Science). The signal intensity corresponding to each gene tested was expressed as a mean intensity value considering the duplicated dots. Then, the data obtained from the two samples tested on a single film (i.e. uninfected and infected cells, harvested at the same timed interval of chlamydial infection) were normalized using the ubiquitin housekeeping gene as a reference for constant level of expression. Finally, for each sample time tested, the difference in the level of gene expression in C. trachomatis-infected and uninfected HeLa cells was expressed as a ratio of the respective mean intensity values found.

ELISA for cytokines

The basal supernatants from polarized uninfected and infected HeLa cells cultivated in the large Costar inserts were tested by ELISA for IL-1α, IL-6, IL-8, IL-11 and TNF-α (Quantikine; R and D Systems).

Acknowledgements

This study was supported by Public Health Service grants AI 31496 to the North Carolina Sexually Transmitted Disease Cooperative Research Center, and AI 13446, both from the National Institute of Health, NIAID.

Footnotes

  1. Present address: Department of Microbiology, Box 70579, VA#1-41, James H. Quillen College of Medicine, East Tennessee State University, Johnson City, TN 37614, USA.

Ancillary