Thrombomodulin Is Silenced in Malignant Mesothelioma by a Poly(ADP-ribose) Polymerase-1-mediated Epigenetic Mechanism*

Malignant mesothelioma (MM) is often complicated by thromboembolic episodes, with thrombomodulin (TM) playing a critical role in the anticoagulant process. Heterogeneous expression of TM has been observed in cancer, and low or no TM expression in cancer cells is associated with poor prognosis. In this study, we analyzed TM expression in biopsies of MM patients and compared them with normal mesothelial tissue. The role of DNA methylation-associated gene silencing in TM expression was investigated. To evaluate poly(ADP-ribose) polymerase-1 (PARP1) as responsible for gene promoter epigenetic modifications, nonmalignant mesothelial cells (Met-5A) and MM cells (H28) were silenced for PARP1 and the DNA methylation/acetylation-associated TM expression evaluated. A correlation between low TM expression and high level of TM promoter methylation was found in MM biopsies. Low expression of TM was restored in MM cells by their treatment with 5-aza-2′-deoxycytidine and, to a lesser extent, with trichostatin, whereas the epigenetic agents did not affect TM expression in Met-5A cells. Silencing of PARP1 resulted in a strong down-regulation of TM expression in Met-5A cells, while restoring TM expression in H28 cells. PARP1 silencing induced TM promoter methylation in Met-5A cells and demethylation in MM cells, and this was paralleled by corresponding changes in the DNA methyltransferase activity. We propose that methylation of the TM promoter is responsible for silencing of TM expression in MM tissue, a process that is regulated by PARP1.

Malignant mesothelioma (MM) 3 is a relatively rare neoplasm, whose formation is associated with occupational hazard largely due to exposure to asbestos, with rather common metastases to the lungs, liver, bone, and the adrenal gland (1). It has been reported that MM patients frequently suffer from throm-boembolic complications (2)(3)(4)(5). An increased susceptibility to thrombosis of solid malignant diseases develops as a result of the activating effect of malignant cells on the hemostatic system (6). The development of this activating effect is a consequence of interactions between malignant cells and the various components of the coagulation system, including coagulation factors, platelets, endothelial cells, and the fibrinolytic system, which promotes progression from a prothrombotic state to clinically manifested disorders of the hemostatic system (7).
Among natural anticoagulants, thrombomodulin (TM) plays a critical role due to its anticoagulant activity, and it is also involved in pathological processes, including thrombosis, inflammation, and cancer (8). TM is a glycosylated transmembrane protein with high affinity for thrombin to form a complex that inhibits thrombin activity and accelerates protein C activation (9). Although it is mainly expressed by endothelial cells, it is found in a wide range of other cells (10 -12). Heterogeneous expression of TM has been observed in the neoplastic disease, and low or no expression of the protein in cancer cells is associated with poor prognosis (13)(14)(15)(16).
TM is transcriptionally up-regulated by thrombin, the vascular endothelial growth factor, histamine, retinoic acid, cAMP, or heat shock (17,18). Its transcription is down-regulated by shear stress, hypoxia, and the transforming growth factor-␤ (19). The ability of the TM gene promoter to bind relevant transcription factors appears to reflect a summation of events that define competence, including nucleosome position, histone modifications, nuclear localization, and the DNA methylation status.
DNA demethylation is typically required for active transcription and is frequently developmentally regulated (20). In addition, expression of various genes is modulated via modification of histones. Silencing of the TM gene promoter has been implicated in the down-regulation of TM synthesis, and it was found that methylation of the promoter CpG islands (CGI) is one of the mechanisms causing its low expression (21,22).
The mechanism by which CGIs are protected from methylation in nonmalignant cells and become susceptible to methylation in tumor cells is unknown. Accumulating evidence indicates a role for poly(ADP-ribose) polymerase-1 (PARP1) in the regulation of methylation patterns. Besides being an enzyme involved in DNA repair, PARP1 also participates in the transcription regulation of various genes. More specifically, the pro-tein promotes decondensation of the chromatin structure through poly(ADP-ribosyl)lation of histones as well as acting as an enhancer-promoter regulatory complex (23). It has been found that the nuclear balance between unmodified and poly-(ADP-ribosyl)ated PARP1, which depends on the dynamics of the PARP and poly(ADP-ribose)glycohydrolase activity, is a key to the maintenance of the genomic methylation pattern (24).
In this study, we analyzed TM expression in biopsies of MM and investigated the involvement of DNA methylation-associated gene silencing in TM expression. To evaluate the role of PARP1 in epigenetic modifications of the TM gene promoter, nonmalignant mesothelial cells (Met-5A) and MM cells (H28) were silenced for PARP1, and DNA methylation and acetylation-associated gene expression were evaluated. We show here, for the first time, that PARP1 regulates TM expression on the epigenetic level and that this regulation is involved in MM.

EXPERIMENTAL PROCEDURES
Biopsy Sampling-Biopsy specimens were obtained from 22 subjects (aged 69.8 Ϯ 10.1 years; 20 males, 2 females) who underwent thoracoscopy or thoracotomy for suspected MM. The collected tissue was divided into two parts; one was suspended immediately in the RNALater solution (Ambion) and stored at Ϫ80°C until analysis. The other part of the tissue was used for histological examination by the Pathological Anatomy Unit of the Hospital University (Ancona, Italy). According to the diagnosis, the individuals were classified as subjects with MM (the MM group) and as healthy subjects with normal mesothelium (the NM group). The MM group (n ϭ 10) included tissue with clear signs of the pathology, whereas the NM group (n ϭ 5) included nonmalignant tissue. The clinical data were obtained retrospectively and include information on the gender, age, histology, neoadjuvant chemoradiation and therapy administration (before surgery), smoking status, and the pathologic staging. The exclusion criteria were the presence or suspicion of any infectious disease and other malignancies. The patients were not treated with any adjuvant chemotherapy or radiation therapy. All subjects filled in a questionnaire that included their informed consent and provided a blood sample. The study was carried out according to the Helsinki Declaration and approved by the Ethical Committee of the University Hospital of Marche, Italy. The demographic and pathological characteristics of the subjects are summarized in Table 1.
Cell Culture-Nonmalignant immortalized mesothelial cells (Met-5A) and MM cells (H28) were obtained from the American Type Culture Collection. The cell lines were grown in RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 10% FBS and cultured at 37°C and 5% CO 2 in humidified atmosphere.
Knockdown of PARP1-To establish sub-lines with stably silenced PARP1, 0.2 ϫ 10 6 exponentially growing Met-5A and H28 cells were transfected with 1 g of the PARP1 shRNA pRS plasmid, with the PARP1 targeting sequence of TAC CAT CCA GGC TGC TTT GTC AAG AAC AG or 1 g of empty pRS plasmid (OriGene) using the TransIT-LT1 transfection reagent (Mirus). Selection was carried out with puromycin added to the cells at 1 g/ml 48 h post-transfection. Puromycin-resistant clones were isolated, expanded, and analyzed for PARP1 pro-tein and mRNA levels as well as PARP activity. Selected clones were maintained in 1 g/ml puromycin.
Assessment of PARP Activity-PARP activity was assessed using the universal colorimetric PARP kit (Trevigen) based on the incorporation of biotinylated ADP-ribose into histone proteins. Cell and tissue lysates containing 50 g of protein were loaded into 96-well plates coated with histones and biotinylated poly(ADP-ribose), allowed to incubate for 1 h, treated with streptavidin-HRP, and read at 450 nm in an ELISA plate reader (Sunrise). The PARP activity was expressed as units per mg protein.
DNA Methyltransferase Activity Analysis-The activity of DNA methyltransferases (DNMTs) was evaluated in nuclear extracts of Met-5A and MM H28 cell lines and their PARP1silenced counterparts (Met-5A 1947 and H28 1949 cells) with or without 5-aza-2Ј-deoxycytidine (5-aza-dC; Sigma) treatment. The cells were seeded in T25 cell culture flasks and treated with 5 M 5-aza-dC for 96 h. The cells were then harvested, and nuclear extraction was performed (EpiQuik nuclear extraction kit, Epigentek). The DNMT activity was assayed in 5 g of nuclear extract proteins using the EpiQuik DNA methyltransferase activity kit (Epigentek). The results were expressed as the increase in the OD per hour per mg of protein.
Western Blotting-Cells and tissue were lysed in a buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 7.2), 1 mM EDTA (pH 8), 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, and a mixture of protease inhibitors (2 g/ml aprotinin, 2 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 2 g/ml pepstatin) and stored at Ϫ80 C until analysis. The protein level was quantified using the Bradford assay (Sigma). 50 g of protein were separated on a 12.5% SDS-PAGE gel and transferred to nitrocellulose membranes (Protran). After blocking for 1 h with 5% nonfat milk in PBS-Tween, the membranes were incubated with monoclonal anti-PARP1 IgG (C2-10; Trevigen), monoclonal anti-TM IgG (PBS-01; Abcam), and polyclonal anti-␤-actin (Bethyl) at 4°C overnight. After washing with PBS-T, the membranes were incubated with HRP-conjugated secondary antibody for 2 h at room temperature, and the protein bands were visualized using the ECL detection system (Pierce). Band intensities were evaluated by ChemiDoc using Quantity One software (Bio-Rad).

5-Aza-2Ј-deoxycytidine and Trichostatin
Treatment-Met-5A and H28 cell lines and their PARP1-silenced counterparts (Met-5A 1947 and H28 1949 cells) were seeded in a six-well plate at a density of 0.1 ϫ 10 6 . After an overnight incubation, cells were treated with freshly prepared 5 M 5-aza-dC for 96 h or 200 ng/ml trichostatin (TSA, Sigma) for 24 h. For the combination treatment, cells were first treated with 5 M 5-aza-dC for 72 h, then 200 ng/ml of TSA was added, and the cells were incubated for further 24 h.
Immunohistochemistry-Formalin-fixed paraffin-embedded tissue of the subjects affected by MM (n ϭ 10) was collected from the Pathological Anatomy Unit of the Hospital University of Ancona, Italy. The normal mesothelium tissue from NM subjects (n ϭ 6) was used as the control. Formalin-fixed paraffin-embedded sections were cut at 5-m thickness, deparaffinized, and heated in 1 mM EDTA buffer (pH 8.0) for 15 min. After blocking, the sections were incubated with anti-TM IgG (PBS-01; Abcam) at 4°C overnight and then with the FITCconjugated secondary antibody (Pierce). The sections were inspected in a fluorescence microscope (Axiocam MRc5, Zeiss). Hematoxylin and eosin staining was used for histological evaluation.
Immunocytochemistry-Met-5A and H28 cells as well as their PARP1-silenced counterparts were placed in 35-mm dishes on glass coverslips. After an overnight incubation, the cells were treated with 5-aza-dC (5 M) and TSA (200 ng/ml), washed with PBS, fixed with fresh 4% formaldehyde in PBS, and permeabilized with saponine (0.05% saponine in PBS, with 2% FCS). Cells were then incubated with mouse anti-human TM IgG at 4°C overnight followed by FITC-conjugated secondary IgG. The cells were inspected in a fluorescence microscope.
Quantitative RT-PCR (qRT-PCR) Analysis-Total RNA was extracted from biopsies (2-3 mg) and cultured cells (10 6 cells) using the SV total RNA Isolation System (Promega Italia) according to the manufacturer's instructions. The first-strand cDNA was synthesized using the MMLV Kit (Promega Italia). The qRT-PCR analyses were performed in the final volume of 25 l using the specific primer sequences: TM, 5Ј-TAA CGA AGA CAC AGA CTG CGA TT-3Ј (forward) and 5Ј-CTA GCC CAC GAG GTC AAG GT-3Ј (reverse); PARP1, 5Ј-AAG CCC TAA AGG CTC AGA AC-3Ј (forward) and 5Ј-AGA AGG CAC TTG CTG CTT GT-3Ј (reverse); and GAPDH, 5Ј-TCC ACT GGC GTC TTC ACC-3Ј (forward) and 5Ј-GGC AGA GAT GAT GAC CCT TTT-3Ј (reverse). The housekeeping GAPDH gene was used as the loading control. The expression of the genes was assessed by the IQ5 qPCR instrument (Bio-Rad) using the RT 2 SYBR Green qPCR Master Mix (Invitrogen). The results were expressed as ⌬C T , and fold changes in relative mRNA expression were calculated using the equation 2 Ϫ⌬(⌬Ct) .
DNA Isolation and Bisulfite Modification-Genomic DNA was isolated from five noncancerous and 10 cancerous biopsy specimens and from cultured cell lines treated with 5-aza-dC and TSA alone or in combination, using the Methylamp isolation and modification coupled DNA kit (Epigentek) according to the manufacturer's protocol. The extraction of genomic DNA from cultured cells was performed using the DNeasy Blood and Tissue Kit (Qiagen). The isolated DNA was stored at Ϫ20°C before use. The genomic DNA (3 g) was digested by EcoRI (Promega Italia) and purified using the Wizard DNA Clean-up system (Promega Italia). 500 ng of the digested and purified DNA were subjected to bisulfite modification using the Imprint DNA Modification Kit (Sigma) according to the manufacturer's instructions. 100 ng of modified DNA isolated from biopsies and cell lines were used in PCR analysis for bisulfite sequencing. To study the methylation of TM genes, a CpG island in the promoter, 5Ј-UTR, and a first exon region containing 44 CpG dinucleotides, respectively, was analyzed and amplified by PCR using AmpliTaq Gold (Applied Biosystems). For the TM gene, the primer sequences were as follows: 5Ј-GAT TAA GAG ATG AAA GAG GGT TGT A-3Ј (forward) and 5Ј-AAA CCA AAA CCC CAA ACA TAT TA-3Ј (reverse), and the annealing temperature was fixed at 58°C. In the PCR reactions, the forward primer was added after the first eight cycles followed by 35 more cycles. The PCR product was purified using the Silice Bead DNA Gel Extraction Kit (Fermentas), cloned into p-GEM-T-easy (Promega Italia), and transfected into JM109 competent cells according to the manufacturer's instructions. Clones containing the plasmid with the insert were selected using isopropyl 1-thio-␤-D-galactopyranoside/X-Gal plates and verified by PCR using the M13 primers. The plasmid DNA was subsequently extracted using the Wizard Plus SV Minipreps DNA purification System (Promega Italia), according to the manufacturer's instructions; 10 clones for each biopsy and cell line were sequenced. The results were expressed as methylation percentage of the 44 CpG dinucleotides.
Co-immunoprecipitation-Imunoprecipitation (IP) was performed using the Cross-link immunoprecipitation kit (Pierce). Briefly, Met-5A and H28 cell lines (3 ϫ 10 6 ) were lysed in the IP buffer. Aliquots normalized for total protein (1 mg) were incubated overnight with anti-DNMT1 (Epigentek) previously cross-linked to protein A/G plus agarose. The resin was washed with the IP buffer, and the bound protein was eluted. Western blots were performed according to standard procedures. Antibodies against PARP1, DNMT1, and normal mouse IgG (Epigentek) were used to detect the individual proteins.
ChIP-The ChIP assay was performed using the EpiQuik chromatin immunoprecipitation kit (Epigentek). Met-5A and H28 cells as well as their PARP1-silenced counterparts (0.5 ϫ 10 6 ) were cross-linked with 1% formaldehyde in the RPMI medium. The reaction was stopped with 1.25 M glycine. The cells were harvested and incubated in the lysis buffer containing protease inhibitors, followed by DNA sonication. After centrifugation (14,000 ϫ g for 10 min), an aliquot of the supernatant was incubated (2 h) with anti-DNMT1 IgG previously crosslinked to the 96-well strips. An aliquot of each supernatant was used as the input control. Positive and negative controls were processed using anti-RNA polymerase II IgG and anti-normal IgG, respectively. The immunocomplexes and input controls were incubated with proteinase K at 65°C, and the samples were transferred to the column and washed with 70 and 90% ethanol, and the purified DNA was eluted. To verify the binding of DNMT1 to the TM promoter, PCR was carried out according to standard procedures. TM primers were designed for the promoter region (Ϫ352 to Ϫ287). The sequences were as fol-lows: 5Ј-AGG GCA GGG TTT ACT CAT CC-3Ј (forward) and 5Ј-TAC TCG GAT TGC TGG GTT CT-3Ј (reverse). The annealing temperature was fixed at 55°C. GAPDH primers were used as a positive control. The band intensities were evaluated by ChemiDoc using Quantity One software.
Statistical Analysis-Results were expressed as mean values Ϯ S.D. Comparisons between groups were performed using the Mann-Whitney U test for unpaired samples and the Kruskal-Wallis analysis for multiple comparisons. A probability value of p Ͻ 0.05 was considered significant. All statistical analyses were performed with the SPSS statistical software (version 15).

TM Expression in Malignant
Mesothelioma-The expression of TM mRNA was ϳ12-fold lower in MM tissue when compared with NM tissue with high statistical significance (p Ͻ 0.001) (Fig. 1A). Although all MM surgical specimens lacked TM immunoreactivity, strong cell membrane TM staining was documented for NM tissue (Fig. 1B). Low TM protein expression in MM tissue was further corroborated by Western blot analysis (Fig. 1C).

Methylation Status of TM Promoter
Region-Bisulfite sequencing was performed in MM and NM biopsies to define the status of DNA methylation. A CpG-rich section of the TM gene spanning the region from the promoter at Ϫ298 to the first exon at ϩ149 was analyzed for its methylation status. Hypermethylated TM was found in the pathological tissue (Ͼ80%), whereas Ͻ20% of methylation was observed in NM tissue (Fig.  2), suggestive of epigenetic regulation of TM expression of normal and malignant mesothelial tissue.
TM Expression and Its Epigenetic Modulation in Cultured Cells-The gene and protein expression levels of TM were analyzed in the nonmalignant Met-5A and the MM H28 cells by qRT-PCR and immunocytochemistry (Fig. 3). Expression of TM was markedly lower in the MM cells when compared with nonmalignant cells (Ͼ200-fold difference, Fig. 3A). To investigate the role of the methylation and/or acetylation, the cells were treated with 5-aza-dC and TSA, respectively. The DNMT inhibitor 5-aza-dC, and partially the deacetylation inhibitor TSA, increased the TM expression in the MM cells. On the other hand, treatment with the individual epigenetic inhibi-tors separately did not affect the TM expression in Met-5A cells, whereas its level decreased upon combined treatment of the cells with the two agents (Fig. 3B). The methylation involvement was confirmed by assessing the DNMT activity. As shown in Fig. 3C, the activity of DNMT was significantly higher in the MM cells relative to Met-5A cells, and the 5-aza-dC treatment markedly inhibited the DNMT activity in both cell lines. The level of the TM protein expression in Met-5A and H28 cells and the effect of the epigenetic modulators 5-aza-dC and TSA was confirmed by immunocytochemical analysis (Fig. 3D).
TM Expression Is Regulated by PARP1-To examine the role of PARP1 in the methylation-and/or acetylation-mediated silencing of TM, the nonmalignant mesothelial cells Met-5A and the MM cells H28 were stably silenced for PARP1 protein expression to generate the Met-5A 1947 and H28 1949 sub-lines. Silencing of PARP1 in Met-5A cells (Fig. 4A) resulted in a lower PARP activity (Fig. 4B). We then tested these cells for the level of expression of TM mRNA. Fig. 4C documents that silencing of the PARP1 protein expression caused a massive down-  60ϫ). The asterisk refers to significant differences between H28 and Met-5A cells, the degree symbol refers to that of 5-aza-dC-treated versus untreated cells, with p Ͻ 0.05. regulation in the TM transcript. The lower level of TM mRNA in the Met-5A 1947 cells was restored to the level in Met-5A cells transfected with the empty vector (Met-5A pRS cells) when they were treated with the epigenetic modulators, in particular 5-aza-dC (Fig. 4D). The PARP1-silenced Met-5A 1947 cells also exhibited enhanced DNMT activity, which was completely inhibited by 5-aza-dC treatment (Fig.  4E). The level of the TM protein expression is shown in Fig.  4F. Finally, we found a much higher level of methylation within the Ϫ298 to ϩ149 region of the TM gene in the Met-5A 1947 cells compared with their counterparts transfected with the empty vector (Fig. 5).
In the next set of experiments, we investigated the effect of PARP1 silencing on the regulation of TM expression in the MM H28 cells. The H28 1949 cells with down-regulated PARP1 gene (Fig. 6A) and with low PARP1 activity (Fig. 6B) exerted considerable up-regulation of the TM mRNA and protein with no additional effect of the epigenetic modulators 5-aza-dC and TSA (Fig. 6, C, D, and F). Silencing of PARP1 in H28 cells significantly reduced the DNMT activity  , 60ϫ). The asterisk represents significant differences between Met-5A 1947 and Met-5A pRS cells, the degree symbol between 5-aza-dC-treated versus untreated cells, with p Ͻ 0.05. (Fig. 6E). The MM cells with silenced PARP1 also showed massive loss of CpG methylation in the TM locus when compared with their counterparts transfected with the empty plasmid (Fig. 7).
PARP1-DNMT1 Complex and DNMT1/TM Promoter Localization-Coimmunoprecipitation was performed to evaluate whether DNMT1 is associate with PARP1. Met-5A and H28 cell extracts were immunoprecipitated with anti-DNMT1 IgG, and the presence of PARP1 was evaluated. PARP1 was found to be associated with DNMT1 in Met-5A cells (Fig. 8A), whereas no association was found in H28 cells. Next, the co-localization of DNMT1 on the TM promoter and the effect of PARP1 silencing on DNMT1/TM promoter interaction were investigated in Met-5A and H28 cells as well as their PARP1silenced counterparts using ChIP assay. Cross-linked chromatin was immunoprecipitated with anti-DNMT1 IgG, and the region of DNA within the TM promoter was evaluated by PCR. Fig. 8B shows that DNMT1 did not localize in the TM promoter in Met-5A cells, whereas PARP1 silencing markedly increased the DNMT1-TM promoter complex formation. Conversely, DNMT1 specifically localized within the region of the TM promoter in H28 cells, and PARP1 silencing reduced the DNMT1/TM promoter interaction.

DISCUSSION
Tumor cells lacking TM feature low anti-coagulant activity, facilitating adhesion to the endothelium of target tissue and, as a consequence, metastatic spread and thrombus formation (6,25). Thrombosis, a well recognized complication of cancer, is an early manifestation of malignancy and is associated with a high rate of morbidity and mortality (26,27). Clinical observations have documented that malignant mesothelioma patients are highly susceptible to thromboembolic events, including arterial thrombosis (3). MM patients may show increased production of procoagulant factors associated with a reduced secretion of molecules with anticoagulant activity.
To this effect, this study documents that TM expression is silenced in surgical MM specimens (cf. Fig. 1). To evaluate possible involvement of epigenetic modification in this process, we first analyzed the methylation status of CGIs in the TM pro- moter region and its proximity in the MM and NM tissues, more specifically within the locus spanning the promoter region (from position Ϫ298) and the first exon (position ϩ149). We show here that CGIs are methylated in the "TM-silent" MM tissue and demethylated in the TM-expressing NM tissue (cf. Fig. 2). The involvement of epigenetic modifications in the TM silencing was then confirmed by treating immortalized mesothelial cells and MM cells with epigenetic agents. Of the epigenetic modulators tested, only the demethylating agent 5-aza-dC sub-stantially restored TM expression in MM cells, while not affecting mesothelial cells (cf. Fig. 3, A and B). Conversely to what we observed in MM cells, the TM promoter of the nonmalignant mesothelial cells was found to be in a demethylated state (cf. Figs. 5-7). Thus, inhibition of DNMT1 by the epigenetic agent 5-aza-dC did not induce any further demethylation with, as anticipated, no additional effect on TM expression.
It is well established that epigenetic events are controlled by a specific sub-group of proteins, which include DNMTs, his-  60ϫ). The asterisk represents significant differences between H28 1949 and H28 pRS cells, the degree symbol between 5-aza-dC-treated versus untreated cells, with p Ͻ 0.05. tone acetylases, histone lysine methyltransferases, or histone deacetylases. These enzymes modulate the methylation or acetylation patterns of the promoter regions of a variety of genes, whereby controlling their expression (29 -33). Recently, components of chromatin, including histones, have been documented as targets for poly(ADP-ribosyl)lation (34), a posttranscriptional modification of specific proteins mediated by PARPs. Accumulating evidence links poly(ADP-ribosyl)lation with DNA methylation (24,(35)(36)(37). This then epigenetically regulates expression of different genes. Consistent with this notion, we found that PARP1 associated with DNMT1 (cf. Fig.  8A) in the nonmalignant mesothelial cells Met-5A. Silencing of PARP1 thus liberates DNMT1 to methylate DNA. In fact, DNMT1 was found to localize in the TM promoter (cf. Fig. 8B), which induces its methylation, resulting in the inhibition of the TM gene and protein expression associated with enhanced DNMT activity (cf. Figs. 4 and 5). This mechanism of the regulation of TM expression was further confirmed by treatment of the cells with the epigenetic modulatory agent 5-aza-dC, an inhibitor of DNMT.
It was found earlier that normal cells with active PARP1 and, consequently, increased level of poly(ADP-ribosyl)lation, are characterized by widespread DNA hypomethylation (37). Conversely, inhibition of the PARP activity introduces an abnormal hypermethylated pattern in the genomic DNA (35) as well as in the CGI regions (36). This suggests that in the absence of polymers of ADP-ribose, certain DNA regions are no longer protected from methylation. The reason is that PARP1 in its poly-(ADP-ribosyl)ated form makes DNMT1 catalytically inactive and thus inefficient in DNA methylation (38). In this model, the poly(ADP-ribose) groups present on PARP1 interact monocovalently with DNMT1, suppressing its enzymatic activity.
A completely different behavior was observed in MM cells. Coimmunoprecipitation analysis revealed that DNMT1 and PARP1 are not associated in H28 cells (cf. Fig. 8A), allowing DNMT1 to methylate DNA; DNMT1 localizes within the TM promoter in these cells. The methylated status of H28 cells was reversed by silencing PARP1, resulting in enhanced TM expression and reduction in the DNMT1-TM promoter complex and DNMT activity (cf. Figs. 6 -8). Involvement of DNA methylation was confirmed using the epigenetic agent 5-aza-dC. This can be reconciled with the notion that cancer cells undergo dramatic changes upon alterations in the DNA methylation pattern. Certain gene promoters become hypermethylated, whereas the genomic DNA is subject to significant hypomethylation. The detailed mechanisms by which CGIs are protected from methylation in normal cells and the mechanisms whereby these DNA regions become susceptible to methylation in tumor cells are still unknown. A mechanism has been proposed, such that CGIs in the promoter region of DNMT are protected from methylation by poly(ADP-ribosyl)ated PARP1 (39). It is possible that under the condition of poly(ADP-ribose) levels too low to inhibit the DNMT1 activity, new methyl groups were added onto the DNMT1 promoter, which resulted in the downregulation of the gene. However, this hypothesis was not confirmed by our results. We did not find a reduction in DNMT1 gene expression in PARP1-silenced H28 cells with respect to the wild-type counterparts (data not shown). Therefore, an indirect role of PARP1 could be involved.
In conclusion, a dual role of PARP1 has been established in TM gene regulation in the context of mesothelioma. Under physiological conditions, PARP1 is involved in the maintenance of a demethylated status leading to TM expression. In tumors, PARP1 induces TM promoter methylation, thus resulting in TM silencing. We believe that this report shows, for the first time, that PARP1 regulates TM expression on the epigenetic level and that this regulation is involved in MM. These findings suggest a possible benefit of using PARP inhibitors to improve the outcome of MM patients that appears rather grim at this stage (40 -42).