Up‐regulation of HDACs, a harbinger of uraemic endothelial dysfunction, is prevented by defibrotide

Abstract Endothelial dysfunction is an earlier contributor to the development of atherosclerosis in chronic kidney disease (CKD), in which the role of epigenetic triggers cannot be ruled out. Endothelial protective strategies, such as defibrotide (DF), may be useful in this scenario. We evaluated changes induced by CKD on endothelial cell proteome and explored the effect of DF and the mechanisms involved. Human umbilical cord vein endothelial cells were exposed to sera from healthy donors (n = 20) and patients with end‐stage renal disease on haemodialysis (n = 20). Differential protein expression was investigated by using a proteomic approach, Western blot and immunofluorescence. HDAC1 and HDAC2 overexpression was detected. Increased HDAC1 expression occurred at both cytoplasm and nucleus. These effects were dose‐dependently inhibited by DF. Both the HDACs inhibitor trichostatin A and DF prevented the up‐regulation of the endothelial dysfunction markers induced by the uraemic milieu: intercellular adhesion molecule‐1, surface Toll‐like receptor‐4, von Willebrand Factor and reactive oxygen species. Moreover, DF down‐regulated HDACs expression through the PI3/AKT signalling pathway. HDACs appear as key modulators of the CKD‐induced endothelial dysfunction as specific blockade by trichostatin A or by DF prevents endothelial dysfunction responses to the CKD insult. Moreover, DF exerts its endothelial protective effect by inhibiting HDAC up‐regulation likely through PI3K/AKT.

traditional risk factors in this setting, and classical CV risk scores, such as the Framingham risk score, underestimate the CV risk in this population. 3 Further, the results of pharmacological interventions that have proven a cardio-protective benefit in the general population or other high risk populations (eg in diabetes or in patients with established cardiovascular disease), such as statins or inhibitors of the renin-angiotensin system, have failed to show a benefit in patients with endstage renal disease. [4][5][6][7] Because of that, new or uraemia-related CV risk factors have been postulated to play a role in the CV risk of these patients. 8 Thus, there is a need for the search of new therapies able to reduce the high CV risk and the associated mortality in this population.
There is wide evidence indicating that endothelial dysfunction is an initial stage for the development of atherosclerosis, and in vivo and in vitro studies have demonstrated that endothelial dysfunction is present in CKD. 9 In previous in vitro studies, endothelial dysfunction induced by the CKD milieu has been characterized. Exposure of endothelial cells (ECs) to sera from CKD patients results in morphological alterations, enhanced thrombogenicity of the extracellular matrix and changes towards a pro-inflammatory phenotype, with increased expression of adhesion receptors and activation of signalling pathways, such as nuclear factor-kappa B (NFkB), 10 or the innate immunity Toll-like receptor 4 (TLR4) and the NALP3 inflammasome. 11 Moreover, the endothelial response to the CKD insult is also characterized by an enhanced oxidative stress and changes in the expression of related proteins. [12][13][14][15] Our previous in vitro results suggest that the CKD setting induces activation of specific genes in the endothelium that promotes a prooxidant, pro-inflammatory, prothrombotic and proproliferative state. In this regard, HDACs enzymes are important epigenetic factors that regulate pro-inflammatory gene expression, in ECs and in other tissues, and also deacetylate non-histone proteins that regulate inflammatory signalling. 16 Therefore, they appear to be promising therapeutic targets for the treatment of atherosclerosis and other cardiovascular diseases. 17,18 Furthermore, HDAC activity is linked to a number of cardio-renal pathologies, including heart failure, 19 hypertension, 20 and diabetes or diabetic kidney disease. 21,22 Thus, the dysfunctional endothelium and its epigenetic regulation arise as an attractive target for interventions designed to reduce the risk and burden of CVD in CKD patients. 23,24 Defibrotide (Defitelio®) (DF) is a drug composed by a complex single-stranded oligodeoxyribonucleotides derived from porcine intestinal mucosal DNA that has demonstrated profibrinolytic, antithrombotic-thrombolytic, antiischaemic, antishock, antiatherosclerotic, antirejection and anti-angiogenic effects. 25 The endothelial protective activity of this drug has been demonstrated in different clinical settings. [26][27][28] This drug interacts specifically with the membrane of ECs where it displays its anti-inflammatory and anti-oxidant effects. 29 In the present study, we applied a translational approach 12 to identify proteins differentially expressed in an endothelium exposed to the CKD milieu and the effect of DF. Functional assays were performed to confirm the role of the most physiologically relevant identified proteins and their implication in the development of endothelial dysfunction.
Moreover, we aimed at identifying the mechanisms involved in the protective effect exerted by DF in this setting.

| Endothelial cell cultures
ECs were isolated from human umbilical cord veins. 30

| Differential proteomic analysis
ECs were isolated from human umbilical cord veins (n = 6) and cultured as previously described. After reaching confluence during the first passage, cells were mixed, seeded in 75-cm 2 flasks and exposed to media containing 20% of pooled sera of controls (n = 20 distributed in 6 different pools) or CKD patients (n = 20 distributed in 6 different pools) with or without DF (100 μg/mL, a dose selected from our previous studies) 27

| Confirmation of proteins by Western blot and immunofluorescence studies
From the differentially up-regulated proteins found comparing the analysed settings, both HDAC1 and HDAC2 were selected to be deeply studied. First, protein lysates from ECs exposed to the same conditions studied in the proteomic assay were analysed by Western blot (WB, n = 4) and immunofluorescence (IF, n = 6) with specific primary antibodies (Abcam) to confirm the proteomic results.
For WB analysis, ECs were exposed to media containing 20% of For IF analysis, ECs were exposed to the same conditions pre-

| Roles of HDAC1 and HDAC2 in endothelial damage induced by CKD sera
To evaluate the role of HDAC1 and HDAC2 in uraemic endothelial dysfunction, a pan-HDAC inhibitor trichostatin A (TSA) (Sigma Aldrich) and DF were used as HDAC inhibitors. ECs on 6-well microplates were pretreated (24 hours) with TSA (50 nmol/L) or DF (100 µg/mL) and exposed to media containing 20% of pooled sera from the uraemic patients or healthy donors, as previously described for the differential proteomic analysis. Cells were then fixed and

| Statistics
Results are expressed as mean ± standard error of the mean (SEM) and correspond to the % of labelled area/% of nuclei area for IF as-

| Endothelial dysfunction in CKD is characterized by an increase in HDAC1 and HDAC2, among other proteins
To study the proteomic changes involved in endothelial uraemic dysfunction and the potential protective role of DF in this setting, a proteomic approach was performed on ECs exposed to sera of healthy controls or CKD patients in the absence and presence of DF. After normalization, 11 proteins were differentially expressed comparing the analysed settings (Table 2). Proteins related to inflammation, such as protein arginine N-methyltransferase 5 or histone arginine methyltransferase; proteins involved in epigenetic regulation and proteins related to cell motility; proliferation and survival (HDAC1, HDAC2, ATP-dependent DNA helicase Q1, TRIO and F-acting-binding protein), among others were overexpressed in the uraemic setting.
Results suggested that DF could re-establish the proper expression of proteins that were found overexpressed in the ECs exposed to CKD conditions.

| DF prevents the overexpression of HDAC1 and HDAC2 induced by CKD sera, as well as increased HDAC1 in both nucleic and cytoplasmic locations
WB and IF techniques were applied to confirm HDAC1 and HDAC2 overexpression in ECs exposed to CKD sera and the effect of DF. WB results confirmed that CKD induced an overexpression of HDAC1 (1.9 ± 0.3 fold vs Control, n = 4, P < .05) and HDAC2 (1.5 ± 0.1 fold vs Control, n = 4, P < .05) that was normalized in the presence of DF (1.1 ± 0.2 for HDAC1, and 0.9 ± 0.2 for HDAC2, n = 4 and P < .05 vs CKD for all experiments) (Panel A, Figure 1).
Similarly, an increase in HDAC1 was detected by IF in ECs exposed to CKD from 3 ± 0.3% of labelled area/% nuclei area in controls to 4.7 ± 0.2% of labelled area/% nuclei area in CKD (n = 6, P < .01).
The effect of uraemic sera on endothelial HDAC1 expression was prevented by DF (2.6 ± 0.1% of labelled area/% nuclei area, n = 6, P < .01 vs CKD) (Micrographs and left scatterplot, Panel B, Figure 1). Specifically at the nucleus, CKD sera induced a significant increase in HDAC1 (77.5 ± 2.9% vs 50.3 ± 2.4% of nuclei labelled area in control samples, n = 6, P < .01), which was prevented by DF (41.8 ± 5.2% of nuclei labelled area, n = 6, P < .01 vs CKD). In addition, the effect of DF was also notable in the cytoplasmic location in ECs exposed to CKD, as the overexpression of the protein in ECs exposed to uraemic sera vs controls was reduced to almost 75% of cytoplasmic labelled area (Micrographs and right scatterplot, Panel B, Figure 1).

| The inhibitory effect of DF on CKD-induced HDAC1 overexpression is dose-dependent
Immunofluorescence assays were performed with different DF doses to test the specificity of the reduction of HDAC1 expression previously detected. In ECs exposed to CKD sera, HDAC1 total expression increased to 4.7 ± 0.2% of labelled area/% nuclei area compared to control, and was dose-dependently inhibited in the presence of 50 μg/mL (4.2 ± 0.3% of labelled area/% nuclei area, n = 6,) and 100 μg/mL (3.8 ± 0.1% of labelled area/% nuclei area, n = 6, P < .05 vs CKD) of DF ( Figure 2).

| CKD-induced endothelial dysfunction is mediated through HDAC1 and HDAC2 overexpression
ICAM-1 and TLR4 expression on cell surfaces and vWF content were higher in ECs exposed to the CKD patients' sera when TA B L E 2 Overexpressed proteins in endothelial cells exposed to CKD sera  Figure 3). Furthermore, DF was also capable to inhibit the expression of ICAM-1, TLR4 and vWF content on cells exposed to uraemic sera in a significant manner, decreasing their expression to control levels (0.4 ± 0.1%, 0.4 ± 0.1% and 4.1 ± 0.2%, respectively, n = 6, P < .05 vs CKD for ICAM and P < .05 vs CKD for TLR4 and vWF) ( Figure 3). DF also exhibited a remarkably and significant inhibitory effect on ROS generation in response to the uraemic sera, reducing its production to control levels (18.1 ± 1.3 of mean fluorescence intensity, n = 6, P < .05 vs CKD) (Figure 3).

| Effect of DF on HDAC1 and HDAC2 is potentially mediated through PI3K/AKT pathway inhibition
ECs were exposed to P740-Y-P, a cell-permeable phosphopeptide activator of the PI3K/AKT pathway in the presence or absence of DF (100 μg/mL). Then, HDAC1 expression was assessed by WB and IF, and HDAC2 by WB (Figure 4).
WB results revealed that the expression of HDAC1 and HDAC2 was increased in ECs incubated with P740-Y-P (5 hours) (fold of 1.9 ± 0.1 and 1.4 ± 0.2, respectively vs control, n = 4, P < .05) and that these increases were prevented by DF (0.9 ± 0.1 and 1.1 ± 0.2 fold vs control, F I G U R E 1 Defibrotide prevents HDAC1 increased expression induced by CKD sera at both nucleic and cytoplasmatic locations. A, Immunoblot images show expression of HDAC1 (left) and HDAC2 (right) when endothelial cells were exposed to control or CKD sera (24 h) in the absence or presence of DF (100 µg/mL). B, Micrographs show HDAC1 expression (green) in endothelial cells exposed to control and CKD sera in the absence or presence of defibrotide (CKD + DF). Left column shows total presence of HDAC1 (green) and nucleic (blue) staining. The middle column shows HDAC1 expression only in the nuclei and the right column shows HDAC1 expression only in the cytoplasm (40× magnification). Left scatterplot (with median) represents HDAC1 expression in terms of the total percentage of labelled area/percentage of nuclei area, and right scatterplot represent HDAC1 expression in terms of the total percentage of labelled area in nuclei or cytoplasm, in endothelial cells exposed to healthy sera (Control) and sera from CKD patients with or without the presence of defibrotide (CKD and CKD + DF, respectively) (n = 6, being **P < .01 vs control and ##P < .01 vs CKD, n = number of independent experiments, statistical analysis was performed with raw data using t test) respectively). Moreover, through an IF assay HDAC1 overexpression in the nuclei was confirmed after the incubation of ECs with P740-Y-P (from 32.6 ± 4.4% of covered area to 44.9 ± 6.4%, n = 6, P < .05).

| D ISCUSS I ON
Our present study explored the protein signature of the endothelium exposed to CKD sera in the presence and absence of DF, and pointed out to HDACs as key molecules that mediate the endothelial response to the CKD milieu. Both TSA and DF prevented the endothelium from developing its pro-inflammatory, prooxidant, prothrombotic and activated innate immunity phenotype induced by the CKD sera. Further, PI3K/AKT signalling pathway was identified as a putative pathway through which DF modulates HDACs expression ( Figure 5). Thus, the results of the present study highlight the relevance of the epigenetic changes associated with endothelial dysfunction in CKD and uncover the potential mechanisms of action by which DF exerts its protective effect on ECs in this setting.
The involvement of vascular endothelium in the initiation and the progression of atherosclerosis in CKD patients has been progressively recognized. So far, the endothelial phenotype in CKD has been extensively characterized, but there is a lack of information regarding the mechanisms through which the uraemic milieu exerts an impact on the endothelial cell and its epigenome. To approach this knowledge, we applied an established translational methodology 12 to look for the signature of endothelial dysfunction in CKD and find key factors that may be susceptible to be regulated by DF. We were able to identify two proteins, HDAC1 and HDAC2, involved in epigenetic regulation, among other up-regulated proteins. Epigenetics refer to chromatin-based mechanisms important in the regulation of gene expression that do not involve changes in the DNA sequence per se. 35,36 Our proteomic results were confirmed by other techniques and revealed that the CKD milieu induces overexpression of both HDAC1 and 2 and the accumulation of HDAC1 in both the nucleus and the cytoplasm of ECs. These two enzymes are widely expressed in human tissues, belong to class I HDACs, which are considered the 'classical' HDACs, whose activities could be inhibited by TSA, and are involved in cell proliferation and cell survival. 37 Although HDAC class I (HDAC 1, 2, 3 and 8) is classically described as being expressed in the nucleus, these proteins are also found in the cytosol. In this sense, several studies have shown that HDAC1 can also be expressed in the cytosol in different cell lines, [38][39][40] and although its expression is mainly located in the nucleus, its expression is enhanced in the cytosol in pathological conditions. 41 Further, it has been reported that the up-regulation of HDAC1 in ECs deacetylates nitric oxide synthase 3 (NOS 3 ), directly resulting in a decrease of endothelial nitric oxide production. 42 HDAC overexpression in ECs in response to CKD sera is in concordance with previous evidence by other authors showing enhanced expression of these proteins in other cardio-renal pathologies. [19][20][21][22] The protective effect of DF in this setting concord with its proved beneficial effect in other clinical entities associated with endothelial dysfunction. [27][28][29] HDACs positively act on pro-inflammatory candidates, such as cytokines (IL-6, IL-12, TNF), chemokines (CCL2, CCL7 and CXCL) and other inflammatory mediators (MMP-9 and endothelin-1). 43,44 In this regard, HDACs inhibitors show broad anti-inflammatory effects through different mechanisms. In our model of uraemic endothelium, both TSA and DF reduced ICAM-1 expression and vWF production. Moreover, there are many studies demonstrating the important role of HDACs both in innate and adaptive immunity. In particular, these enzymes seem to act on the expression of TLR target genes. 45 In this regard, we show in the present study that TLR4 expression is regulated by HDACs in ECs exposed to a uraemic milieu. In addition, the inhibition of ROS production by TSA and DF demonstrates that HDACs are also involved in the prooxidant endothelial response to CKD.

F I G U R E 2
Defibrotide prevents HDAC1 increased expression induced by CKD sera in a dose-dependent manner. Micrographs show endothelial cells exposed to CKD sera in absence (up) and presence of defibrotide (50 µg/mL and 100 µg/mL). Nuclei are labelled with DAPI (blue) (40× magnification). Scatterplot (with median) represents quantification of the area positive for HDAC1 staining normalized by percentage of nuclei area, by immunofluorescence technique (n = 6, being *P < .05 vs control and #P < .05 vs CKD, n = number of independent experiments, statistical analysis was performed with raw data using t test) The pharmacological inhibition of HDACs is a new target for several disorders. These drugs have been approved for the treatment of some lymphomas, or epilepsy (valproic acid) and several others are in clinical trials. Further, these agents have shown its efficacy in animal models of inflammatory diseases, attenuating the progression of renal fibrogenesis in obstructive nephropathy or reducing cyst formation in polycystic kidney disease. 46 This property has been successfully applied to ameliorate endothelial activation and vascular F I G U R E 3 ICAM-1, TLR4, vWF and ROS increases in CKD are mediated through HDACs induction. Micrographs show enhanced expression of the adhesion receptor ICAM-1 (red), TLR4 (red), the adhesive protein vWF (red) and production of ROS (green) when endothelial cells were exposed to CKD patients' sera. The co-incubation of cells with CKD patients sera and TSA (CKD + TSA, middle row) or defibrotide (CKD + DF, lower row) induced a decrease in the expression of the markers of endothelial damage analysed. Cell nuclei are stained with DAPI (blue) (40× magnification). Scatterplots (with median) correspond to the quantification of the endothelial damage markers evaluated. All data correspond to relative expression compared to control levels (n = 6, being *P < .05 vs control and #P < .05 and ##P < .01 vs CKD, n = number of independent experiments, statistical analysis was performed with raw data using t test) pathophysiology in sickle cell anaemia in transgenic mice. In this experimental animal model, and also in HUVECs, 47,48 TSA significantly reduced VCAM-1 and tissue factor expression. 49 Although many of the underlying mechanisms and functions of HDACs in human cells are poorly understood and remain unclear, they appear to be promising targets in the modulation of endothelial dysfunction.
Once we demonstrated that DF is able to reduce HDACs expression and their downstream effects in the CKD-induced endothelial damage, we focused on the search of the signalling pathway potentially used by the drug to modulate these epigenetic regulators.
There is evidence demonstrating that HDAC expression is regulated by the PI3K/AKT pathway. 17,50 In addition, our group has previously F I G U R E 4 Defibrotide acts as a PI3/ AKT inhibitor to interact with HDACs. A, Immunoblot images show expression of HDAC1 (left) and HDAC2 (right) when endothelial cells were exposed to 740 Y-P in absence or presence of DF (100 µg/ mL). B, Micrographs show an increase in HDAC1 expression (green) in endothelial cells exposed to P740-Y-P (+P740-Y-P) and a decrease when DF was added (+740 Y-P + DF). Scatterplot (with median) represents the quantification of HDAC1 expression in the three situations (Control, +740 Y-P, +740 Y-P + DF) in terms of the labelled area (n = 6, being *P < .05 vs control and #P < .05 vs CKD, n = number of independent experiments, statistical analysis was performed with raw data using t test) F I G U R E 5 Visual abstract. A single, concise, pictorial and visual summary of the main findings of the present study in which we demonstrate that HDACs appear as key modulators of the CKD-induced endothelial dysfunction and that DF prevents endothelial dysfunction responses to the CKD insult likely through PI3K/AKT demonstrated that DF inhibits the activation of PI3K/AKT. 27,51 Both facts prompted us to hypothesize that DF was regulating HDACs through this pathway. To prove this hypothesis, ECs were exposed to a PI3K/AKT inducer to stimulate HDAC expression, and then, the effect of DF was explored. Our results indicate that DF could act as a PI3K/AKT pathway inhibitor to modulate HDACs expression. One of the limitations of our study is that we measured protein levels, but not HDACs activity or other HDACs not appearing in our proteomic approach. Another limitation of our study is that we used a pan-HDAC inhibitor rather than specific HDAC1 or 2 inhibitor, and its effects on other HDACs cannot be ruled out.
In conclusion, CKD is an example of a complex disease in which the phenotype arises from a combination of environmental and inheritable factors. 52 Evidence suggests that the contribution of the uraemic milieu may be mediated via modifications of the epigenome. The findings of the present study revealed a role for HDACs as key modulators of the endothelial phenotype in response to the CKD insult. These enzymes seem to mediate, at least in part, the endothelial enhanced oxidative stress, the up-regulation of the innate immunity, and the pro-inflammatory and pro-thrombogenic responses. In addition, we provide strong evidence showing that DF may confer a protection to the endothelium from the uraemic insult acting as a potential HDAC modulator. Finally, DF seems to exert its endothelial protective effect by inhibiting HDAC up-regulation likely through PI3K/AKT signalling pathway. The data that support the findings of this study are available from the corresponding author upon reasonable request.