|Home | About | Journals | Submit | Contact Us | Français|
Human cytomegalovirus (HCMV) contributes to pathogenic processes in immunosuppressed individuals, in fetuses, and in neonates. In the present report, by using reporter gene activation assays and confocal microscopy in the presence of a specific antagonist, we show for the first time that HCMV infection induces peroxisome proliferator-activated receptor gamma (PPARγ) transcriptional activity in infected cells. We demonstrate that the PPARγ antagonist dramatically impairs virus production and that the major immediate-early promoter contains PPAR response elements able to bind PPARγ, as assessed by electrophoretic mobility shift and chromatin immunoprecipitation assays. Due to the key role of PPARγ in placentation and its specific trophoblast expression within the human placenta, we then provided evidence that by activating PPARγ human cytomegalovirus dramatically impaired early human trophoblast migration and invasiveness, as assessed by using well-established in vitro models of invasive trophoblast, i.e., primary cultures of extravillous cytotrophoblasts (EVCT) isolated from first-trimester placentas and the EVCT-derived cell line HIPEC. Our data provide new clues to explain how early infection during pregnancy could impair implantation and placentation and therefore embryonic development.
Human cytomegalovirus (HCMV), a member of the betaherpesvirus family, causes miscarriages, morbidity, and mortality in fetuses and newborns and contributes to the development of numerous diseases in immune-compromised hosts, especially transplant recipients and AIDS patients (for a review, see reference 16). A crucial step in regulation of HCMV replication and reactivation from latency lies in activation of the major immediate-early promoter (MIEP), which controls production of the immediate-early gene products IE1 and IE2, the latter being essential for viral replication. Among transcription factors which have been reported to be involved in MIEP regulation (15), NF-κB is the most studied, as four cognate recognition sites have been identified in the MIEP sequence and because it is clearly one of the most important regulators of proinflammatory gene expression, including those encoding cytokines, such as tumor necrosis factor alpha, interleukin-1β (IL-1β), IL-6, and IL-8, and cyclooxygenase 2 (Cox-2). Interestingly, it was demonstrated that in infected fibroblasts, inhibitors of Cox-2 reduced the yield of HCMV by a factor 100 and blocked the accumulation of IE2 mRNA and protein (28). These data suggested regulation of the MIEP downstream of prostaglandin (PG) production, a by-product of arachidonic acid (AA) conversion by Cox-2. Up to now, mechanisms underlying regulation of IE2 mRNA by prostaglandins, either at the transcriptional or posttranscriptional level, remained unknown. Evidence that induction of Cox-2 and synthesis of prostaglandins including PGE2 were essential for efficient HCMV replication prompted us to consider a possible role for the peroxisome proliferator-activated receptor gamma (PPARγ), a nuclear receptor able to induce transcription of genes involved in lipogenesis and to repress genes involved in inflammatory processes (9). Indeed, Cox-2 is required for the production of prostaglandins, some of which are agonists of PPARγ.
PPARγ is an isoform of the PPAR subset of nuclear receptors that also includes PPARα and PPARβ/δ, which all bind to DNA as heterodimers with retinoid X receptors (RXR). Heterodimers activate expression of target genes by binding to peroxisome proliferator-responsive elements (PPREs) composed of direct tandem repeats of a consensus sequence spaced by a single nucleotide. In addition, PPARs suppress gene expression by interfering with the activity of transcription factors like NF-κB, Stats, and AP-1. Besides its role as regulator of lipid metabolism, inflammation, and the immune response (9), PPARγ was demonstrated to play a major role for trophoblast differentiation and early placental development in knockout mice (3, 12). In human placenta, PPARγ was shown to control trophoblast invasion and differentiation (26), a process known to be impaired by infection with HCMV during pregnancy (6). Here we first demonstrate that PPARγ can be used for HCMV replication through its binding to the major immediate-early viral promoter and that by this way infection dramatically impairs trophoblast migration and invasiveness.
U373MG astrocytoma cells, MRC5 fibroblasts, and Hek-293 cells (InvivoGen) were propagated in medium containing 10% fetal calf serum, and extravillous cytotrophoblasts (EVCT; also HIPEC) were propagated as described previously (22, 26). Chorionic villi from first-trimester placentas (8 to 12 weeks of gestation) were obtained from volunteers who legally chose to terminate pregnancy after written approbation (Broussais Hospital, Paris, France). Cultures of trophoblasts were approved by the local ethical committee. After dissection, EVCT were isolated by trypsin digestion and purified through a discontinuous percoll gradient. Cell viability was determined by flow cytometry analysis of annexin V (AV)-positive cells (fluorescein isothiocyanate [FITC]-AV; Dako). About 90% of viable EVCT stained positive for human invasive extravillous cytotrophoblast markers after culture on matrigel (CK07, HLA-G, CD9, hPL, c-erB2, or the α5 subunit of the fibronectin receptor) and expressed RXRα and PPARγ in their nuclei. Immunohistochemistry for PPARγ and cytokeratin 7 was performed as previously described (8) on paraffin-embedded sections of an implantation site (placenta and myometrium at 16 weeks of pregnancy). Immunocytochemistry with anti-RXRα, -PPARγ, and -CK07 antibodies on EVCT and HIPEC was performed as described in reference 22.
Virus strains used were AD169 (ATCC), Towne (a gift from S. Michelson, Institut Pasteur, Paris, France), and VHLE (a gift from C. Sinzger, Tubingen, Germany). Virus stocks were collected when cytopathic effects on MRC5 cells (bioMérieux, France) were >90%. Supernatants were clarified of cell debris by centrifugation at 1,500 × g for 10 min, ultracentrifuged at 100,000 × g for 30 min at 4°C, and stored at −70°C until use. Virus titers were determined by plaque assay on MRC5 cells using standard methods. UV irradiation of HCMV was performed with a Spectroline irradiator (EF-140/F) for 20 min.
NS-398 (10 μM; Sigma) and G3335 (30 μM; Calbiochem) were added to cells 30 min before infection and cultures were refed with G3335 every 24 h. Gw9662 was from Calbiochem. 15-Deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2) and rosiglitazone were from Cayman Chemicals. Dose-effect experiments were done for all reagents, which were each used at the optimal concentration (peak dose) that did not affect cell viability as tested by trypan blue exclusion, cell morphology, and nuclei condensation or fragmentation (4′6-diamino-2-phenylindole [DAPI] staining). The solvent used for each of them (vehicle control) was assayed as a control in all sets of experiments.
A PPRE-luciferase (Luc) plasmid provided by W. Wahli (University of Lausanne, Lausanne, Switzerland) and a home-made construct containing PPRE sequences were used independently. A reporter vector for PPARγ (pB3xPPRE-Luc) was constructed by modifying the pNF-κB-Luc plasmid which was described previously (18). Briefly, the NF-κB binding sites of the pNF-κB-Luc plasmid were removed by digestion with MluI and BglII and replaced by the phosphorylated, double-stranded MluI-BglII oligonucleotide obtained by annealing 3XPPRE-S (5′-CGCGTAGGTCACAGGTCACAGGTCAGAGGTCAGAGGTCATAGGTCA-3′) and 3X PPRE-AS (5′-GATCTGACCTATGACCTCTGACCTCTGACCTGTGACCTGTGACCTA- 3′). As a control, the MluI/BglII-linearized pNF-κB-Luc plasmid was also directly recircularized, providing a vector deleted of any responsive element (p0RE-Luc).
Three reporter vectors containing the putative PPRE sequences of the MIEP (PP1-Luc, PP2-Luc, and PP3-Luc) were similarly constructed by replacing the NF-κB binding sites of the pNF-κB-Luc plasmid with the phosphorylated, double-stranded MluI-BglII oligonucleotides obtained by annealing PP1-S (5′-CGCGTAGTTCATAGCCCATAGTTCATAGCCCA-3′) and PP1-AS (5′-GATCTGGGCTATGAACTATGGGCTATGAACTA-3′) for PP1, PP2-S (5′-CGCGTAACTTACGGTAAATGGCCCGTAACTTACGGTAAATGGCCCA-3′) and PP2-AS (5′-GATCTGGGCCATTTACCGTAAGTTACGGGCCATTTACCGTAAGTTA-3′) for PP2, and PP3-S (5′-CGCGTACGTCAATGACGGTAAATGGCCCGACGTCAATGACGGTAAATGGCCCA-3′) and PP3-AS (5′-GATCTGGGCCATTTACCGTCATTGACGTCGGGCCATTTACCGTCATTGACGTA-3′) for PP3. All constructs were checked by DNA sequencing. For transfection, U373MG cells were seeded in a 24-well plate at 50,000 cells per well. Sixteen hours later, pPPRE-reporter plasmids and FuGene 6 reagent (Roche) were added according to the manufacturer's instructions. After 24 h of treatment or infection, cells were lysed with 1× cell culture lysis reagent (Promega), and luciferase activity was quantified using a Mithras luminometer.
Sequences of probes were selected according to previous data that demonstrated the importance of the 5′-flanking sequence (10, 20) of the DR-1 elements. Except for PP1, which contains a RARE sequence in its 5′-flanking region, nucleotides upstream of the DR-1 motif are conserved for PP2 and PP3. Two hundred nanograms of each double-stranded probe, PP1, PP2, or PP3, as used for reporter plasmids construction, was labeled with [γ-32P]ATP (Perkin-Elmer) using T4 polynucleotide kinase (New England Biolabs) for 30 min at 37°C and purified on MiniQuick spin oligo columns (Roche). Three micrograms of nuclear protein extract (Q proteome nuclear subfractionation kit; Qiagen) from HCMV-infected U373MG cells (24 h postinfection [p.i.]) were incubated at room temperature for 30 min with 15 ng of each probe, poly(I·C), and 2 μg of either control IgG (Abcam) or PPARγ antibody (Santa Cruz Biotechnology) or 200 ng of cold probe in 10 μl of binding buffer (0.2 M KCl, 0.1 M K+ HEPES, pH 7.6, 5 mM MgCl2, 5 mM EGTA, 4% Ficoll). Mixes were submitted to electrophoresis on a 6% Novex retardation gel (Invitrogen) in 0.5× Tris-borate-EDTA buffer at 100 V for 90 min. Revelation was made on Kodak BMR film at −80°C overnight.
U373MG cells (5 × 104 cells/cm2) were infected with HCMV (AD169; multiplicity of infection [MOI], 3) for 24 h, fixed with 1% formaldehyde, lysed (Abcam), and sonicated (Vibracell 72405; Bioblock), in order to obtain DNA fragments of sizes of <1,000 bp. DNA-protein complexes were immunoprecipitated according to the Abcam procedure with chromatin immunoprecipitation (ChIP)-grade antibodies (Abcam), control IgG, anti-histone-3 (H3), or anti-PPARγ, and then incubated with protein A-Sepharose beads (Sigma-Aldrich). Primers to detect HCMV MIEP were complementary to positions −272 and +13 relative to the MIEP start site as described in reference 24, with the sense primer 5′-ATT ACC ATG GTG ATG CGG TT-3′ and antisense primer 5′-GGC GGA GTT GTT ACG ACA T-3′. The PCR amplification parameters were as follows: 95°C for 5 min, and then 50 cycles at 94°C (40 s), 50°C (40 s), and 72°C (90 s).
Full-length mouse PPARγ2 as well as transduction in Hek-293 cells were done as follows: PPARγ2 was initially N-terminally hemagglutinin (HA) tagged and cloned into p.Shuttle-CMV (Stratagene) as described in reference 19. The CMV-HA-PPAR cassette was transferred to the AdEasy-1 vector by homologous recombination in Escherichia coli to generate the AdHA-PPARγ2 construct (Ad-PPARγ). The plasmid was linearized and transfected into Hek-293 cells for amplification. The amplified adenovirus was purified using CsCl gradients, and the viral titer was estimated by a plaque assay-based approach as recommended in the AdEasy protocol (Stratagene). Transduction with adenovirus encoding PPARγ (Ad-PPARγ) or not (Ad-vector) was performed at an MOI of 0.1 in Hek-293 cells at 80% confluence. After 2 h of incubation, the medium was replaced with fresh medium. The next day, cells were infected with HCMV (MOI, 3) as indicated. Virus titers were determined on Hek-293 supernatants by plaque assay on MRC5 cells.
Cells were cultured on coverglasses, washed, fixed, and permeabilized in methanol for 5 min at −20°C. PPARγ was detected with specific monoclonal antibody (MAb E-8; Santa Cruz Biotechnology) followed by phycoerythrin-conjugated anti-mouse IgG antibody (Beckmann Coulter). IE1 and IE2 proteins were detected with FITC-conjugated MAb (Chemicon). Fluorescence was analyzed with an LSM 510 confocal microscope (Zeiss, Jena, Germany). Images were reconstructed by using the LSM image browser software system (Zeiss).
Cells were fixed and permeabilized in methanol for 2 min at −20°C and incubated for 10 min with 0.3% (wt/vol) Oil Red O in isopropanol. After 30 s of incubation in 60% isopropanol, cells were washed in water and nuclei were counterstained with hematoxylin.
RNA was extracted by using Trizol (Invitrogen), reverse transcribed (RT) with Superscript III (Invitrogen), and cDNA amplified with Mastermix (Eppendorf). Fragments of 215 bp and 217 bp, corresponding to IE1 and IE2 cDNA, were amplified with the following primers: sense (5′-AGG TGC CAC GGC CCG AGA CAC C-3′) and antisense (5′-TCT GTT TGA CCG AGT TCT GCC-3′) for IE1 and sense (5′-AGG TGC CAC GGC CCG AGA CAC C-3′) and antisense (5′-GGC GAG GAT GTC CGA GTT CTG CC-3′) for IE2. A 300-bp fragment from actin was amplified with the sense primer 5′-TCC CTG GAG AAG AGC TAC GAG-3′ and antisense primer 5′-CAT CTG CTG GAA GGT GGA CA-3′. PCR was performed as follows: 95°C for 5 min followed by 35 cycles of 95°C (30 s), 58°C (for actin) or 60°C (for IE1) or 62°C (for IE2) for 1 min, 72°C for 1 min, and 72°C for 2 min.
Cells lysates were submitted to SDS-PAGE on 4 to 12% polyacrylamide gels (Invitrogen). Blots on Hybond C-extra membrane (Amersham) were treated with anti-IE antibody (mouse hybridoma E13; a gift from S. Michelson), anti-UL44 (Abcam), anti-UL99 (Abcam), anti-β-actin (Santa Cruz Biotechnology), or anti-PPARγ (Santa Cruz Biotechnology), and then with polyclonal rabbit horseradish peroxidase-conjugated antibody, and protein bands were detected by using an ECL-plus kit (Amersham).
The PPREfinder program was developed at Loria to exhaustively list PPRE present in DNA sequences (http://bioinfo.loria.fr/projects/pprefinder/PpreFinder1.1.zip/view). This three-step program first identifies and localizes all PPRE half-sites (consensus sequence, AGGTCA) in the input sequence, allowing for mismatches (three in this study) at any place in the consensus 6-base sequence. The second step consists of localizing pairs of half-sites separated for this study by only one nucleotide. The third step filters direct repeats (DR-1 motifs) from the preceding pair list and ranks them according to the total mismatch number over the two half-sites. A statistical module based on a stationary Markov chain provides an estimation of the probability that the finding of a given DR-1 motif reflects a particular base composition in the sequence.
Monolayers of HIPEC cells (80% confluent) were infected with HCMV, scratched (three per well), left for 24 h in culture, washed with phosphate-buffered saline (PBS), and photographed under a light microscope. Wound closure reflecting migration of cells was quantified by using ImageJ software (NIH). EVCT were cultured on matrigel-coated transwell membranes (8-μm-diameter pore membranes) for 48 h in the presence or absence of 1 μM rosiglitazone, 35 μM G3335, and 3 MOI of HCMV. After CK07 immunostaining and counterstaining with DAPI, pseudopodia crossing the porous membrane were quantified and normalized to the total number of nuclei. Results of a representative experiment (performed in triplicate) are expressed as the percent variation relative to the control.
U373MG astrocytoma cells, which constitutively express PPARγ, were transfected with different PPRE-Luc plasmids encoding a luciferase gene under the control of the thymidine kinase minimal promoter and a multimerized PPRE recognized by PPARγ and infected with HCMV (VHLE clinical strain). Figure Figure1A1A shows that expression of luciferase increased throughout infection, which can be assumed to reflect transcription of the Luc gene through binding of PPARγ to the PPRE sequence. Luciferase activity in uninfected cells may be due to ligand-independent activity of PPARγ (19) or to low levels of endogenous agonists. Addition of 15d-PGJ2, a natural ligand of PPARγ, to uninfected cells induced Luc transcription with a much lower efficiency than HCMV (Fig. (Fig.1B).1B). This may reflect a better availability of endogenously synthesized 15d-PGJ2 in infected cells and/or that activation of PPARγ by HCMV could be in part mediated by other metabolites. To provide direct proof that PPARγ was involved in the PPRE-driven expression of luciferase, U373MG cells were infected in the presence of G3335, a cell-permeable dipeptide that acts as a specific and reversible antagonist of PPARγ through blockade of ligand binding. Figure Figure1C1C shows that under these conditions, luciferase activity was strongly decreased, demonstrating that luciferase transcription depends on the interaction of PPARγ with its ligand. Treatment with NS398, an inhibitor of Cox-2 activity, substantially reduced luciferase activity in infected cells (Fig. (Fig.1C),1C), demonstrating that Cox-2 activation takes part in PPARγ transcriptional activity, likely through its role in prostaglandin synthesis from arachidonic acid. Specificity of PPARγ activation by HCMV was further assessed by using one shot of Gw9662 (Gw), an irreversible antagonist known to covalently bind and modify the ligand binding site of PPARγ. Figure Figure1D1D shows the inhibitory effect of Gw on cells transfected with the pB3xPPRE-Luc plasmid and treated with rosiglitazone as well as on those infected with HCMV. To address whether viral neosynthesis was required for activation of PPARγ, luciferase activity was quantified from cells infected with UV-irradiated virus. According to the usual considerations, failure of the irradiated virus (Fig. (Fig.1D)1D) to drive luciferase expression suggests a role for newly synthesized viral proteins in PPARγ activation. Transfection with PPRE-deleted control plasmid (Fig. (Fig.1D,1D, right histogram) did not show any luciferase activity.
To further address whether PPARγ activity was restricted to infected cells, either MRC5 fibroblasts or U373MG cells were infected with HCMV. In MRC5 cells, infection strongly redistributed PPARγ staining from the cytoplasm to the nucleus of infected cells as assessed by confocal microscopy with merged expression of IE proteins and PPARγ in the nucleus of infected cells (Fig. (Fig.1E).1E). Identical results were obtained in U373MG cells although with less intensity (data not shown), and we further assessed the function of PPARγ in these cells through its ability to induce activation of genes involved in lipogenesis. Staining with Oil Red O revealed accumulation of lipid droplets in infected cells, contrary to those treated with the PPARγ antagonist G3335 (Fig. (Fig.1F).1F). These experiments provided evidence that PPARγ is activated in HCMV-infected cells, as demonstrated by nuclear translocation and lipid accumulation.
According to previous data demonstrating that Cox-2 inhibitors block the accumulation of IE2 mRNA and protein and consequently reduce replication of HCMV (AD169 strain) (28), we examined the effects of G3335 on IE1 and IE2 expression and on the yield of infectious virus. Infection of U373MG cells in the presence of G3335 showed that inhibition of PPARγ activity resulted in a huge decrease in the amount of IE2 transcripts, compared to IE1, after 24 h of infection (Fig. (Fig.2A).2A). Western blot analysis of cell lysates obtained from day 1 to day 5 after infection further confirmed a role for PPARγ in modulating expression of IE2 protein, as IE2 expression was delayed and the IE2/IE1 ratio was consistently lower in treated cells (Fig. (Fig.2B).2B). As expected, expression of early (UL44) and late (UL99) proteins was modified in the same way by treatment with PPARγ antagonist (data not shown). Furthermore, production of infectious virus, as assessed by plaque assay every 24 h, was reduced by a factor of 100 in the presence of a PPARγ antagonist (Fig. (Fig.2C),2C), demonstrating that in these cells HCMV uses PPARγ activity for its replication. Identical results were obtained with MRC5 fibroblasts (data not shown). Altogether, our data demonstrate that PPARγ can regulate expression of IE2 mRNA and viral replication in a way that depends on association with its ligand.
To further characterize the responsiveness of putative PPREs contained in the MIEP and to provide direct proof for the role of PPARγ in HCMV replication, Hek-293 cells expressing RXR but not PPARγ were coinfected with adenoviral vectors encoding PPARγ and with HCMV. Under ectopic expression of PPARγ, HCMV-infected Hek-293 cells expressed a higher ratio of IE2 to IE1 mRNA compared with those either nontransduced or transduced with control adenovirus, and the relative amount of IE2 mRNA with respect to β-actin was much higher than in control cells (Fig. (Fig.2D),2D), suggesting a role for PPARγ in production of IE2. Quantification of infectious viruses by plaque assay 48 h after addition of HCMV showed that ectopic expression of PPARγ improved virus production by a factor of 100 (Fig. (Fig.2E).2E). With infection with Ad-vector, production of virus could be due to activation of transcription factors by adenovirus itself. Moreover, no plaques were observed when Hek-293 cells were infected with HCMV alone (mock), despite expression of IE mRNA, or with Ad-PPARγ alone (data not shown), the latter ensuring that the presence of PFU was due to HCMV replication. Altogether, our data provide evidence that activation of PPARγ in Hek-293 cells improves their permissivity to HCMV, which could in part result from an increased expression of IE2.
Because transcriptional activity of PPARγ requires binding to DNA response elements in enhancer or promoter regions of target genes, we asked whether PPREs could be found in the MIEP sequence. Originally, PPAR-RXR heterodimers were shown to bind to a sequence composed of a direct repeat (DR) of the consensus sequence AGGTCA interspaced by 1 nucleotide (AGGTCA N AGGTCA). Besides these so-called DR-1 canonical motifs, other types of PPREs have been reported with various numbers of mismatches but with a conserved ability to drive transcription of genes (4, 5). Scanning of the MIEP sequence for the PPRE canonical motif by using the PPRE finder program revealed that among all the putative PPRE sequences, three (PP1, PP2, and PP3) were identified in the distal enhancer as DR-1 elements, PP2 and PP3 having exactly the same sequence (Fig. 3A and B).
In order to determine whether the putative PPREs were able to mediate activation of gene transcription, plasmid clones containing the putative PPREs upstream of a minimal TK promoter in front of the luciferase gene (PP-Luc) were used for transfection of U373MG cells. In cells transfected with any of the PP-Luc constructs, infection with HCMV induced transcription of the reporter gene, whose efficiency decreased upon treatment with G3335 (Fig. (Fig.3C),3C), suggesting that the identified PPRE sequences could be targets for HCMV-mediated PPARγ-induced transcription. Discrepancies between PP sequence abilities to drive transcription of luciferase may reflect the importance of the 5′-flanking sequence, as previously suggested (10). To address whether PPARγ could bind to PP1, PP2, and PP3, gel mobility shift assays were performed with double-stranded oligonucleotides containing the appropriate sequences. Nuclear extracts from HCMV-infected U373MG were able to form a complex with the 32P-end-labeled PP1, PP2, and PP3 probes, and unlabeled oligonucleotide competed for binding to the probes. Moreover, anti-PPARγ antibodies induced a supershift of the complex, demonstrating that PPARγ was able to bind to PPRE sequences identified in the MIEP (Fig. (Fig.4A).4A). To further determine whether PPARγ could bind to the MIEP in infected cells, ChIP assays were performed on U373MG cells infected for 24 h with HCMV. Figure Figure4B4B shows that immunoprecipitation with antibodies directed against PPARγ as well as histone-3 (H3) allowed amplification of a DNA fragment contained in the MIEP, contrary to results with control IgG. When cells were treated with G3335, ChIP assays no longer revealed association of PPARγ with the MIEP. Altogether, these results give evidence that PPARγ can be associated with the MIEP in infected cells and provide an explanation for the impaired expression of IE2 mRNA and protein in infected cells treated with G3335.
During the first trimester of pregnancy the cytotrophoblasts (EVCT) proliferate to form multilayered columns of cells that rapidly migrate out of the chorionic villi and invade the decidua and the upper third of the myometrium, as illustrated in Fig. Fig.5A.5A. At the implantion site, EVCT are characterized by CK07 and nuclear PPARγ expression all along their differentiating pathway (Fig. 5B and C). To study human trophoblast invasion and migration in vitro, we used either primary cultures of EVCT purified from first-trimester placentas (26) or a cell line (HIPEC) established from primary cultures which share the phenotypic and functional characteristics of primary EVCT (22). In the placenta, trophoblasts constitutively express PPARγ in their nucleus as shown in situ (Fig. (Fig.5A)5A) and in vitro in primary (Fig. (Fig.5D)5D) and transformed (Fig. (Fig.5E)5E) EVCT. We first assessed whether PPARγ could be further activated in HIPEC by infection with HCMV, through staining with Oil Red O. Figure Figure6A6A shows red droplets in HCMV-infected cells, contrary to those infected but treated with G3335, demonstrating the ability of infection to induce activation of PPARγ in cytotrophoblast. Analysis of IE1 and IE2 protein expression levels in infected HIPEC revealed a lower amount of IE2 in the presence of G3335 (Fig. (Fig.6B)6B) compared with untreated cells, even though to a lesser extent than in more permissive cells, such as MRC5 and U373MG, as shown above. Nevertheless, in cytotrophoblasts, the ability of the virus to launch PPARγ activation more than its usage for viral replication is crucial in the sense that it could independently disturb migration and invasion processes.
To address this question, measurements of trophoblast cell migration were done with HIPEC either mock-infected cells or cells infected and treated or not with G3335, by using a “wound healing” assay. The role of PPARγ in inhibition of migration was first confirmed by treating HIPEC with rosiglitazone in the presence of G3335 or Gw antagonists (data not shown). Figure Figure6C6C shows that HCMV infection inhibited cell migration by about 50% and that treatment with G3335 overcame this effect.
To further assess whether HCMV could impair invasiveness of cytotrophoblasts, invasion assays were performed with purified primary EVCT, and the invasion index was determined by counting pseudopodia crossing the porous membrane as visualized by immunostaining with CK07 (Fig. (Fig.7A).7A). As previously demonstrated (26), activation of PPARγ with either rosiglitazone or 15d-PGJ2 inhibited the trophoblastic invasion process by about 50% and 40%, respectively (Fig. (Fig.7B).7B). Incubation of cells with HCMV resulted in a 30% inhibition of invasion, which was totally abrogated following preincubation with G3335 (Fig. (Fig.7C).7C). Altogether, these data provide evidence that activation of PPARγ by HCMV is involved in defective trophoblastic migration and invasion processes.
Implication of Cox-2 and PGE2 in the viral cycle of herpesviruses has been noted for a long time, as anti-inflammatory drugs such as aspirin were shown to modulate primary infection and to decrease reactivation of latent genome. Accordingly, studies conducted by Zhu and collaborators (28) indicated that inflammation may promote HCMV replication and reactivation, and blockades of Cox-2 activity and of prostaglandin production were then considered as potential antiviral therapeutics. Interestingly, in the macaque rhesus model, infection with the rhesus cytomegalovirus (RhCMV) did not induce expression and activation of cellular Cox-2, contrary to human CMV, but the virus encoded a protein with high homology to cellular Cox-2 (vCox-2) (25) which appeared determinant for replication in endothelial cells, highlighting the role of the prostaglandin metabolic pathway in replication of cytomegaloviruses. HCMV replication and reactivation require transcription of the IE genes UL123 and UL122, encoding IE1 and IE2 proteins, respectively, two major products resulting from a differential splicing. Production of IE2 but not IE1 was shown to be crucial for productive replication, and transcription of IE2 but not IE1 was significantly affected by Cox-2 inhibitors (28). A role for cyclic AMP (cAMP) response elements contained in the MIEP was suggested (17) with respect to the known ability of prostaglandins to increase cAMP, but for the moment no direct proof has been obtained to support this hypothesis. Our results demonstrate for the first time a role for PPARγ in production of IE2 through binding to PPRE sequences contained in the MIEP, and experiments demonstrating a huge decrease in viral production in the presence of PPARγ antagonist confirmed the key role of PPARγ in viral replication. Upstream of PPARγ activation, molecular and cellular events underlying activation of Cox-2 are known to involve the enzymatic activity of the cellular phospholipase A2 (c-PLA2). It has been known for a long time that c-PLA2 activity can be induced by HCMV infection (1), but more interestingly, a connection between infection and PPARγ activation was supported by a study demonstrating that HCMV particles carry a cell-derived PLA2 that is boarded during the assembly process, which plays a major role in infectivity of MRC5 fibroblasts (2). PLA2 activity is responsible for the production of arachidonic acid, a substrate for both Cox-2 and lipoxygenase activities, and we assume that it is a reason why c-PLA2-specific inhibitors dramatically impair virus production (2).
Nevertheless, given that cPLA2 is involved in many complex processes, including embryogenesis, inflammation, and reproduction, the potential side effects of inhibitors, provided that highly selective ones are available, might discredit their potential therapeutic indications. With respect to these observations and to those reported in the present paper, we can imagine a scenario where virus entry could initiate the release of arachidonic acid through boarded c-PLA2 activation, a starting point of a process involving Cox-2- as well as lipoxygenase-mediated production of PPARγ ligands, the latter being considered a final requisite for PPARγ-mediated transcription in infected cells. Nevertheless, our data demonstrating the inability of UV-irradiated viruses to induce activation of PPARγ raise the question of a role for neosynthesized viral proteins in the activation process, and they also raise the question of whether irradiation could alter the structure and function of virions constituents including boarded cPLA2.
Surprisingly, and in agreement with previous data using Cox-2 inhibitors (28), PPARγ inhibition differentially impaired the expression of IE1 and IE2 mRNA, suggesting a role for PPARγ in favoring IE2 mRNA production. This was supported by our experiments demonstrating a huge increase in expression of IE2 but not IE1 mRNA under ectopic expression of PPARγ in HCMV-infected Hek-293 cells. Since IE1 and IE2 transcripts result from differential splicing events which are temporally regulated after infection, IE1 being produced before IE2, we can assume that binding of PPARγ to the MIEP could be coordinated for instance by temporal chromatin remodeling as reported to explain IE2-mediated autorepression of the MIEP at late times of infection (23). Interestingly, based on several studies showing that transcription and RNA processing could be spatially and temporally coupled and that the transcriptional PPARγ coactivator 1 (PGC-1) could play a major role in pre-mRNA splicing (11), we speculate that binding of PPARγ to the MIEP may recruit PGC-1, which in turn could control exon 4 skipping to give rise to IE2 mRNA, some issues which remain to be extensively explored.
PPARγ was shown to play a major role in human cytotrophoblast differentiation and function. Implantation of the conceptus involves early invasion of the uterine epithelium and of the underlying stroma by extravillous trophoblastic cells, and we demonstrated that PPARγ activation repressed this trophoblastic invasion process (26). Since migration/invasion is essential for the implantation and remodeling of uteroplacental vascularization necessary for fetal growth, we speculate that through activation of PPARγ, infection of the placenta by HCMV, a phenomenon known to precede infection of the fetus, could take part in miscarriages and low-birthweight babies. In addition to 15d-PGJ2, a metabolite of cyclooxygenase, the metabolite of 15-lipoxygenase (Lox), 15-HETE, was also shown to inhibit human trophoblastic invasion by about 50% (7, 21). Whether HCMV infection might lead to production of 15-HETE by activating Lox in cytotrophoblasts is a relevant question, since incubation of EVCT with [3H]arachidonic acid followed by quantitative high-performance liquid chromatography analysis of [3H]AA metabolites indicated that 15-HETE was the main metabolite produced by EVCT (unpublished data). In the course of the invasion process, degradation of the basal membrane and of the extracellular matrix of the uterine stroma is ensured by metalloproteases (MMP) such as MMP-9, whose secretion is elevated during the early steps of gestation (13). Active MMP-9 is required for invasion to proceed and, for instance, we can assume that HCMV-induced activation of PPARγ could lead to a decrease in expression or activity of MMP-9, thus impairing placental differentiation. Remarkably, PPARγ activation has been shown to induce trans-repression of the MMP-9 protein (14), whose transcription depends on NF-κB activation, and HCMV infection is known to decrease MMP activity and to dysregulate the cell-cell and/or cell-matrix interactions of infected cytotrophoblasts (27). Experiments are in progress to address this issue. In utero, viral infections have been associated with an adverse pregnancy outcome and may have a causative role in the unexplained cause of congenital infections in developed countries, and they are considered one of the most common pathogens in fetal loss cases. Cytomegalovirus accounts for miscarriages and intrauterine growth retardation and congenital anomalies, and our study provides the first evidence that activation of PPARγ by infection could take part in these pathological processes through abnormal placentation related to impairment of the trophoblast migration/invasion mechanisms. Overall, an understanding of the mechanisms underlying activation of PPARγ and characterization of its target genes following infection of cytotrophoblasts by HCMV should help in proposing new prognostic and therapeutic approaches to prevent miscarriage and risks of sequelae in newborns.
We thank Claudie Offer and Hélène Brun for sequencing, Michel Baron for helpful advice in performing gel mobility shift assays, John Sinclair (University of Cambridge, United Kingdom) and Stéphane Chavanas in ChIP assays, Vassilis Tsatsaris for immunochemistry of placental sections, and Ananda Mookerjee and Charlotte Casper for critical reading of the manuscript.
This work was supported by institutional grants from INSERM.
Published ahead of print on 30 December 2009.