HDAC1 inhibition by MS-275 in mesothelial cells limits cellular invasion and promotes MMT reversal

Peritoneal fibrosis is a pathological alteration of the peritoneal membrane occurring in a variety of conditions including peritoneal dialysis (PD), post-surgery adhesions and peritoneal metastases. The acquisition of invasive and pro-fibrotic abilities by mesothelial cells (MCs) through induction of MMT, a cell-specific form of EMT, plays a main role in this process. Aim of this study was to evaluate possible effects of histone deacetylase (HDAC) inhibitors, key components of the epigenetic machinery, in counteracting MMT observed in MCs isolated from effluent of PD patients. HDAC inhibitors with different class/isoform selectivity have been used for pharmacological inhibition. While the effect of other inhibitors was limited to a partial E-cadherin re-expression, MS-275, a HDAC1-3 inhibitor, promoted: (i) downregulation of mesenchymal markers (MMP2, Col1A1, PAI-1, TGFβ1, TGFβRI) (ii) upregulation of epithelial markers (E-cadherin, Occludin), (iii) reacquisition of an epithelial-like morphology and (iv) marked reduction of cellular invasiveness. Results were confirmed by HDAC1 genetic silencing. Mechanistically, MS-275 causes: (i) increase of nuclear histone H3 acetylation (ii) rescue of the acetylation profile on E-cadherin promoter, (iii) Snail functional impairment. Overall, our study, pinpointing a role for HDAC1, revealed a new player in the regulation of peritoneal fibrosis, providing the rationale for future therapeutic opportunities.


MCs derived from peritoneal effluent of PD patients undergo reversal of MMT-related marker gene expression when treated with MS-275, a selective HDAC1-3 inhibitor. To analyze the role of
HDACs in the maintenance of a mesenchymal-like state in MCs that have undergone MMT in vivo, we used MCs isolated from peritoneal effluent of PD patients. When cultured in vitro these cells have been shown to maintain a stable cobblestone-like or a fibroblastic phenotype, and their stage of trans-differentiation correlates with the progression of peritoneal damage and with the time of PD treatment 19 . Even with a cobblestone-like phenotype, these cells are different from normal healthy MCs since they express lower levels of E-cadherin and cytokeratins 8-18, and express Snail 11,20 .
To analyze whether histone deacetylation activity plays a role in epithelial gene expression changes during MMT, histone H3 acetylation on E-cadherin promoter was analyzed by chromatin immunoprecipitation (ChIP) in epithelial-like (epithelioid) and mesenchymal-like (non-epithelioid) MCs. The observation of higher H3 acetylation levels on E-cadherin promoter in epithelial-like with respect to mesenchymal-like cells suggests a role for HDAC activity in MMT induction during PD (Fig. 1A).
To demonstrate the effect of HDACs in the regulation of genes relevant for the maintenance of a mesenchymal phenotype, ECM production and invasiveness, MCs were treated for 3 days with Trichostatin A (TSA), a pan HDAC inhibitor; with MC1568, an HDAC4-6-8 inhibitor, and with MS-275, a HDAC1-3 inhibitor [21][22][23] (Fig. 1B). While treatment with TSA and MC1568 was ineffective, MS-275 significantly decreased the expression of mesenchymal markers such as type I collagen (Col1A1), MMP2, and PAI-1 (Fig. 1C). Accordingly, the expression of both TGFβ1 and TGFβRI were decreased upon treatment with MS-275 (Fig. 1C). On the other hand, the expression of epithelial markers (E-cadherin and Occludin) was increased upon treatment with MS-275 (Fig. 1C). While TSA and MC1568 induced a partial recovery of E-cadherin expression, they failed to promote a more general reprogramming towards an epithelial-like phenotype.
Interestingly, MCs treated with MS-275 were unresponsive to contemporary TGFβ1 stimulation. The increase of E-cadherin expression was confirmed at protein level (Fig. 1D). Increase of TSA concentrations up to 250 nM led to a partial rescue of epithelial, but not to a decrement of mesenchymal markers (Suppl. Fig. 1A-C). As a verification of MS-275 specificity, treatment with compound MC2500, an inactive meta isomer of MS-275, did  not modify the expression of these epithelial/mesenchymal genes. (Suppl. Fig. 2A-D) 24 . In the same conditions, immunofluorescence (IF) analysis confirmed an increase of histone H3 acetylation in the nuclei of MCs treated with MS-275, but not with compound MC2500 ( Fig. 2A,B). Since MS-275 is up to 13-fold more potent against HDAC1 than against HDAC2 and -3, and at the used dose (250 nM) it is specific towards HDAC1 inhibition, we checked the effect of genetic HDAC1 silencing in our model 21 .
Notably, the specific HDAC1 targeting, by siRNA approach, shown in Fig. 3A,B, was found sufficient to induce E-cadherin and to markedly limit TGFβ1 expression. To directly analyze the effect of HDAC1 in mediating the changes induced by PD fluids, epithelial-like MCs were treated with stay safe balance PD fluid. The induced MMT was reversed by subsequent treatment with MS-275, as demonstrated by RT-PCR and WB analysis (Fig. 3C,D).
These results demonstrate that while the effect of TSA and MC1568 is limited to a partial rescue of E-cadherin expression, MS-275 treatment, targeting HDAC1 function, is causal for MMT reversal. Moreover, MS-275 was able to restore an epithelial signature after MMT induction upon exposure in vitro to PD fluids. Treatment with MS-275 mediates the reacquisition of an epithelial-like phenotype and the inhibition of cellular motility and invasion. The effect of MS-275 has been analyzed also at morphological level by IF analysis.
Treatment with MS-275 induced the recovery of fibroblastic MCs towards a cobblestone-like cell shape, which was associated to a marked relocalization of ZO-1 at cellular junctions ( Fig. 4A and Suppl. Fig. 3A), and to a general decrease of α-SMA expression (Fig. 4B). Interestingly, MCs reacquiring a cobblestone phenotype were insensitive to contemporary TGFβ1 stimulation (Fig. 4A,B and Suppl. Fig. 3B,C). In accordance with molecular data, MS-275 blocked the induction of a fibroblastic-like shape and the loss of ZO-1 membrane localization in epithelial-like MCs treated with high glucose PD fluid ( Fig. 4C and Suppl. Fig. 4).
During the process of peritoneal fibrosis, invasion of the sub-mesothelial stroma by MCs is a key event leading to abnormal production of ECM proteins and pathological angiogenesis 25 . Here, we analyzed the impact of MS-275 in MCs motility performing an in vitro scratch assay on confluent monolayers of MCs from PD patients.  Treatment of MCs with MS-275 led to a partial reduction of scratch closure, whereas treatment with MC2500 was ineffective (Fig. 5A,B). We then analyzed whether exposure to MS-275 may affect three-dimensional (3D) invasion through Matrigel. We found that treatment with MS-275 totally abolished MCs invasion (Fig. 5C,D).
These results demonstrate that treatment with MS-275 promotes reversal towards an epithelial-like morphology of MCs having undergone MMT in vivo, strongly limits MCs directed migration while totally abolishing 3D invasion, thus impacting important features of the MMT/fibrotic program.

Treatment with MS-275, while inducing Snail expression, hampers its activity. When looking
at the molecular mechanisms controlling MET, we first focused on Snail, the EMT master gene and direct repressor of E-cadherin expression. Unexpectedly, Snail expression was found markedly increased upon exposure to MS-275 both at mRNA and at protein level (Fig. 6A,B). These results were confirmed in MeT5A, a mesothelial cell line, where increased levels of both E-cadherin and Snail expression upon treatment with MS-275 were found (Suppl. Fig. 5A,B). IF experiments also showed increased nuclear staining of Snail in MCs having undergone a reversal towards a cobblestone phenotype after treatment with MS-275 (Fig. 6C).
The contradictory evidence of a bona fide MMT reversal in the presence of increased levels of the EMT-inducer Snail, suggests a MS-275-mediated Snail functional impairment.
We wondered whether Snail and HDAC1 binding to E-cadherin promoter was inhibited by the treatment with MS-275. We performed HDAC1 and Snail immunoprecipitation followed by RT-PCR (ChIP) on E-cadherin promoter in MeT5A cells in the presence of TGFβ1 to induce Snail expression.
As shown by Fig. 6D, in the presence of MS-275 neither Snail nor HDAC1 binding to E-cadherin promoter was impaired. Next, we verified histone H3 acetylation status on E-box within E-cadherin promoter upon treatment with MS-275. By performing acetylated histone H3 immunoprecipitation followed by RT-PCR on E-cadherin promoter, we found that histone H3 acetylation on E-cadherin promoter, reduced by TGFβ1, was restored in the presence of MS-275 (Fig. 6E). These results indicate that, despite increased expression of Snail and maintenance   of Snail-HDAC1 binding to E-cadherin promoter, Snail functional repressive activity is HDAC1-dependent and thus is hampered by treatment with MS-275. Last, to evaluate the persistence of the epithelial-like phenotype obtained after treatment with MS-275, expression of Snail and of epithelial/mesenchymal markers were evaluated three days after withdrawal of MS-275. While expression of Snail was markedly reduced, the cobblestone-like morphology and ZO-1 membrane localization was maintained (Fig. 7A and Suppl. Fig. 6). Also, the expression of epithelial/mesenchymal markers was generally maintained after withdrawal of MS-275 (Fig. 7B,C). These results suggest that the treatment with MS-275 promotes the repositioning of MCs towards a new stable phenotype with epithelial-like features.
Overall, we demonstrated that inhibition of HDAC1 by MS-275 is sufficient to promote MMT reversal and inhibition of cellular invasiveness of MCs from effluent of PD patients having undergone MMT in vivo and in a model of PD fluid exposure in vitro. Moreover, treatment with MS-275 hampers Snail activity through altered histone H3 acetylation status on specific promoters (Fig. 7D).

Discussion
Aim of this study was to evaluate the effect of HDACs in the re-acquisition of epithelial-like features in MCs isolated from effluent of PD patients and to elucidate mechanisms involved. The rationale for the use of HDAC pharmacological inhibitors was provided by our finding that epithelial-like MCs showed higher H3 acetylation on E-cadherin promoter with respect to mesenchymal-like cells.
When analyzing different HDAC inhibitors, we found that while the effect of TSA (pan-HDAC inhibitor) and MC1568 (HDAC4-6-8 inhibitor) were limited to a rescue of E-cadherin expression, treatment with MS-275 (HDAC1-3 inhibitor) was sufficient to promote a bona fide MMT reversal characterized by increase of epithelial markers, decrease of mesenchymal markers, conformational change, reduction of directed migration (scratch assay) and total inhibition of three dimensional invasiveness through Matrigel. Due to their mesodermal origin, MCs have in basal conditions an epithelial-like morphology co-expressing epithelial and mesenchymal markers, such as E-cadherin, cytokeratins, vimentin and desmin 20 . Once stimulated by TGFβ1 often in combination with TLR ligands, MCs undergo MMT and may became indistinguishable from myofibroblasts obtained from other sources 26 .
The ability to stably acquire mesenchymal features makes these cells a privileged experimental model for the study of mechanisms controlling EMT dynamics and especially the role of epigenetics. In 'truly epithelial cells' such as hepatocytes, TGFβ1 withdrawal is sufficient to determine recovery of E-cadherin levels and to induce Snail downregulation 27 . In MCs, induction of MMT leads to a stable acquisition of a 'mesenchymal-like' phenotype, characterized by reduced or absent E-cadherin and constitutively elevated Snail expression levels 11,20,28 .
To our knowledge, our study first deals on the effect of HDAC1 inhibition on MMT of MCs. A recent report analyzed the role of a selective HDAC6 inhibitor, tubastatin, in the inhibition of peritoneal MMT and fibrosis 29 . Interestingly, in apparent contradiction to our results, treatment with the inhibitor of histone acetyltransferase C646 was reported to limit the MMT induced by glucose in a MC line 30 .
The specificity of MS-275 was warranted by the use of its meta isomer MC2500, which was ineffective. MS-275 is an inhibitor of HDAC1 and to lesser extent, of HDAC2 and HDAC3 activity 31 . At the concentration used in this study (250 nM), this pharmacologic inhibitor is specific for the HDAC1 isoform. Genetic silencing experiments confirmed that HDAC1 plays a major role in MMT reversal observed in MCs. As a further confirmation of cellular activity, MS-275 inhibited MMT induced by treatment of epithelial-like MCs with PD fluid in vitro.
The role of MS-275 in EMT has been evaluated in other experimental systems, especially in tumors. MS-275 at higher concentrations was demonstrated to reverse EMT in ERa-negative breast cancer cell lines 17 . In a non-tumoral experimental setting, MS-275 was demonstrated to limit renal fibrosis induced by renal fibroblasts activation, although no information was provided about renal epithelial cells 32 .
As previously mentioned, TSA and MC1568 effects are limited to a rescue of E-cadherin expression. No data are available in the literature about the effect of MC1568 in E-cadherin expression and EMT. Since MC1568 at the used dose (10 µM) inhibits HDAC6 and HDAC8, conceivably inhibition of other HDACs is needed to obtain a more general reprogramming towards the epithelial-like phenotype in MCs 22,23 . With regard to TSA, there is an apparent conflict with other studies performed generally in epithelial cell lines or tumors where TSA treatment inhibited EMT 33,34 . However, in these studies TSA was used at toxic concentrations for primary MCs (data not shown).
Besides epithelial/mesenchymal marker expression and cell morphology, we focused on the regulation of cellular functions. Since cells coexpressing epithelial cytokeratin and a mesenchymal marker (α-SMA or fsp1) have been found in the submesothelial stroma of fibrotic peritoneum, it has been hypothesized that MCs invasion and subsequent production of ECM proteins, VEGF and inflammatory/chemotactic factors is a main mechanism of fibrosis 9,35 . Alternatively, it has been proposed that MCs undergoing a partial EMT and remaining at the mesothelial layer would participate to the EMT process secreting proinflammatory/pro-fibrotic factors 36 . Studies of lineage tracing demonstrated that MCs may invade the sub-mesothelial stroma, although the number of human E-cadherin promoter. Values derived from 3 independent experiments are reported as means ± SEM and expressed as ((IP/IgG)/Input). Statistically significant differences are reported. (E) qPCR analysis of ChIP assays with anti-acH3 antibody and, as controls, normal rabbit IgG on chromatin from MeT5A cells treated with MS-275 for 72 h or left untreated (NT) and treated with TGFβ1 for 24 h (TGFβ1) when indicated. Data show the enrichment of H3 acetylation on Snail consensus binding sites of human E-cadherin promoter. Values derived from five independent experiments are reported as means ± SEM. and expressed as ((IP-IgG) %Input). Statistically significant differences are reported. P < 0.05 was considered significant. of MCs generating myofibroblasts and the relevance of this process may change in different models of peritoneal fibrosis 37,38 . Our data demonstrate that, while the maintenance of a reduced horizontal mobility may allow devoid areas of peritoneum to be repopulated even in the presence of HDAC inhibitors allowing re-peritonealization 39,40 , potentially harmful invasion of the sub-mesothelial stroma is totally blocked.
Previous studies focused on mechanisms maintaining a mesenchymal-like phenotype in MCs from PD patients. Due to the relevance of TGFβ signaling, SMAD dependent and independent mechanisms were elucidated 3,41 . Constitutive hyper-activation of MEK-ERK1/2-Snail and TAK1/NF-κB pathways was described in these cells and pharmacological blockage induced a partial MMT reversal 11,28 .
Exogenous administration of BMP7, an antagonist of TGFβ1, was sufficient to induce MMT reversal 12,42 . MCs exposed to pro-inflammatory stimuli secrete large amounts of TGFβ1, whose production may induce signaling loops playing a role in the maintenance of a mesenchymal-like phenotype 43 .
To this extent, our studies complement these previous discoveries, since we demonstrate that sensitivity of MCs to TGFβ1 is blocked by treatment with an HDAC1-3 specific inhibitor, MS-275. The lack of cellular response to TGFβ1 observed upon treatment with MS-275 may arise by: (i) reduction of endogenous TGFβ1 expression; (ii) reduced levels of TGFβRI; (iii) inhibition of downstream mediators, such as Snail.
When focusing on Snail expression, we unexpectedly found that Snail levels were increased upon treatment with MS-275. Since direct targets, such as E-cadherin and Occludin, were increased as well, we hypothesized that Snail activity would be hampered despite increased expression.
Peinado et al. demonstrated that E-cadherin repression is mediated by Snail through the recruitment of the Sin3A/HDAC1/HDAC2 complex 44 . The recruitment to Snail binding site of other pivotal epigenetic factors including Polycomb Repressive Complex 2 (PRC2) have also recently been demonstrated to be crucial for Snail activity 45 .
We observed that treatment with MS-275, while non influencing Snail/HDAC1 recruitment on E-cadherin promoter, did not result in the expected local H3 histone deacetylation, thus causing the impairment of the TGFβ1-induced Snail-mediated repression (Fig. 7D).
Interestingly, this inhibition of Snail activity is inhibitor-specific or dependent on cellular context, since other HDAC inhibitors such as valproic acid, sodium butyrate and SAHA have been demonstrated to promote EMT through upregulation of Snail in epithelial tumors 15,16 .
Other mechanisms proposed for the effect of MS-275 in EMT were reduction of SMAD3, EGFR, STAT3 phosphorylation 32 . Our proposed mechanism, focusing on the effect of increased H3 histone acetylation on E-cadherin promoter, explains the observed functional inactivation of the TGFβ1/Snail axis. However, we do not exclude that other mechanisms such as those above cited, may play a role in HDAC1 inhibition especially in vivo.
Since HDAC inhibitors are already used in cancer therapy and are on trials for other non tumoral/fibrotic diseases, the understanding of the role of HDAC1 in the maintenance of a mesenchymal state in MCs has both a basic and a translational relevance. These discoveries may have also a broader impact in relevant clinical issues unrestricted to PD practice, such as in the prevention of post-operative peritoneal adhesions, and in the control of peritoneal metastases.

Methods
Cells. Effluent-derived MCs were isolated from 17 clinically stable PD patients as described previously 10 .
Baseline clinical data from these patients are reported in Table 1. MCs from PD effluents express the standard mesothelial markers intercellular adhesion molecule (ICAM)-1 and cytokeratins 8-18, although at lower levels than healthy HPMCs. MC cultures are negative for the endothelial marker CD31 and the pan-leukocyte marker CD45 6,11,28 . Effluent-derived MCs were cultured in Earle's M199 supplemented with 10% FBS (GIBCO ® Life Technology, Monza, Italy) and antibiotics.
Chromatin immunoprecipitation assay (ChIP). ChIP analysis was performed as reported previously 45 .
Scratch assay. MCs were allowed to reach 100% confluency. Cells were pretreated with DMSO, 250 nM MS-275 or MC2500 at the same concentration for 48 h in culture medium until reaching 100% confluency. A scratch wound was created using the culture-insert 2 wells in µ-Dish from ibidi (Martinsried, Germany). Micrographs were taken at time 0 and 18 h after the scratch. Three independent experiments were performed. For microscopy time lapse experiment, after insert removing the wound closure was monitored for 15 h acquiring bright-field images every 30 minutes. Images were acquired at the Olympus iX83 FluoView1200 laser scanning confocal microscope using an UPLSAPO10x2, NA 0.40 air objective. Images ware stitched using Olympus FluoView software.

3-dimensional Invasion
Assay. 3-dimensional invasion assays were performed as in 35 , MCs (1.5 × 10 5 ) were treated with 250 nM MS-275, 250 nM MC2500 or DMSO for 12 h and then seeded in triplicate in ibidi 15-well slides (µ-Slide Angiogenesis) and allowed to attach for 3 h. 40% Matrigel (40 μl) in serum-free medium was laid over the cells. After 1 h, 50 μl full medium containing 20% serum (and MS-275or vehicle) was added and cells incubated for 72 h. Cells were fixed with 4% paraformaldehyde (PFA), permeabilized with 0.25% Triton PBS, and stained for 12 h with rhodamine-phalloidin (to stain F-actin) and Hoechst (to stain nuclei) in PBS. After washes, cell were mounted using the ibidi mounting medium. Confocal images were acquired at the Olympus iX83 FluoView1200 laser scanning confocal microscope using an UPLSAPO10x2, NA 0.40 air objective. Each stack consists of individual images with a z-step of 5 μm. The 3D rendering was performed using the Imaris image analysis software v.8.1.2 (Bitplane). The invasion quantification was performed with the software Huygens Professional v.17.04 (Scientific Volume Imaging, The Netherlands). Three independent experiments were performed.
Statistical analysis. Statistical significance was determined with a t-test using GraphPad Prism version 5.0 (La Jolla, CA, USA). Differences were considered significant at P < 0.05.