Histone Deacetylase Inhibition Enhances Self Renewal and Cardioprotection by Human Cord Blood-Derived CD34+ Cells

Background Use of peripheral blood- or bone marrow-derived progenitors for ischemic heart repair is a feasible option to induce neo-vascularization in ischemic tissues. These cells, named Endothelial Progenitors Cells (EPCs), have been extensively characterized phenotypically and functionally. The clinical efficacy of cardiac repair by EPCs cells remains, however, limited, due to cell autonomous defects as a consequence of risk factors. The devise of “enhancement” strategies has been therefore sought to improve repair ability of these cells and increase the clinical benefit. Principal Findings Pharmacologic inhibition of histone deacetylases (HDACs) is known to enhance hematopoietic stem cells engraftment by improvement of self renewal and inhibition of differentiation in the presence of mitogenic stimuli in vitro. In the present study cord blood-derived CD34+ were pre-conditioned with the HDAC inhibitor Valproic Acid. This treatment affected stem cell growth and gene expression, and improved ischemic myocardium protection in an immunodeficient mouse model of myocardial infarction. Conclusions Our results show that HDAC blockade leads to phenotype changes in CD34+ cells with enhanced self renewal and cardioprotection.


Introduction
Progenitor cells-based ischemic tissues repair is one of the major endpoints in cardiovascular regenerative medicine. To provide substantial clinical benefits, progenitor cells should be able to engraft in sufficient numbers and differentiate into appropriate cardiovascular cell types, primarily endothelial, vascular smooth muscle cells or cardiac myocytes, or to promote ischemic tissue salvage by paracrine interaction with host tissues cells. Since 1997, the cells primarily deputed to fulfill this role are endothelial progenitor cells (EPCs) [1]. Repair efficiency of patient-derived EPCs may be limited by cell autonomous defects caused by cardiovascular risk factors that greatly reduce EPCs tolerance to stress conditions [2], their ability to produce differentiated progenies or to survive into recipient tissues [3]. Various strategies to restore innate EPC biological activity have been therefore sought [4], based on pre-treatment with drugs that restore pro-survival pathways, on culture in the presence of chemotactic cytokines promoting EPCs migratory activity, or on use of drugs that limit glucotoxicity or oxidative stress [5,6].
The relevance of the so called ''histone code'' [7] for epigenetic control of stem cells differentiation vs. self renewal has been highlighted by studies showing involvement of specific histone tails modifications (acetylation, methylation) in establishment of gene expression signatures specific for pluripotent cells [8], lineagecommitted cells [9] or adult-derived stem cells [10]. Given the emerging role of epigenetic phenomena as fundamental triggers of (stem) cell differentiation and plasticity, in the present study we studied whether modification of the epigenetic landscape by pharmacologic inhibition of histone deacetylases (HDAC) enzymes affects CD34 + cells growth, stemness, phenotype and gene expression; we also assessed whether HDACi-treated CD34 + cells have a modified or enhanced regeneration capacity in a mouse model of myocardial infarction.

Effects of HDAC inhibition on CD34 + cells growth and proliferation
In the present study, a serum free culture method was used to expand CD34 + cells [11] in the presence of increasing amounts of Trichostatin A (TSA) and Valproic Acid (VPA), two wide range HDAC inhibitors. As shown in Figures S1 and S2, morphology of HDACi-treated cells was dose-dependently affected by treatment with both drugs. To identify whether specific subpopulations were affected by treatment with increasing VPA and TSA concentrations, a flow cytometry analysis was performed by grouping cells into three discrete regions: R1, corresponding to CD34 neg cells, R2, corresponding to CD34 dim and R3, corresponding to CD34 bright cells; in parallel, cellular growth was assessed by counting at two consecutive time points. Figure 1A and B show results of a 5 day time course analysis; increasing doses of TSA and VPA decreased cellular growth with a dose-dependent shift toward a homogeneous CD34 bright cell population. To assess the time extension of HDAC inhibition on CD34 and CD133 expression, a flow cytometry analysis was performed in cells cultured in the presence of 2.5 mM VPA at 5, 14 and 21 days ( Figure 1C, D; Figure S3).
TSA was not used in subsequent experiments for the reason that this molecule is not routinely clinically used, such as VPA, for treatment of diseases [12,13]. On the other hand, dose-response treatments indicated that 2.5 mM VPA had a marked effect on CD34 expression enhancement without producing maximal growth retardation, but it did not have a significant effect on cell death (trypan blue exclusion test in cells cultured for 7 days: CTR vs. VPA cells: 13.5%62.7% vs. 15.7%61.77%; P = 0.675 paired ttest) or apoptosis (percent of cells in sub-G1 phase as detected by flow cytometry in PI-stained cells cultured for 7 days: 4.963.2 vs. 1.260.4; CTR vs. VPA; mean 6 SE, n = 3, P = 0.38, paired t-test). VPA 2.5 mM dose was then used in all the following experiments. To clarify the relationship between stem cells generation time and retardation in cell cycle progression, Carboxyfluorescein Succinimidyl ester (CFSE) dye was used to assess CTR and VPA growth progression at 5 and 7 days and to calculate proliferation index and cellular generations number in the presence and absence of VPA [14]. Figure 2A and B show that, at 7 days, proliferation index was significantly reduced in VPA vs. CTR cells and that CD34 bright /CFSE bright cells after VPA treatment were increased. This corresponded to an average 1-2 generations delay in the presence of VPA ( Figure S4A) and to a specific growth retardation of CD34 bright cells ( Figure S4B). Cell cycle analysis and double staining with antibodies recognizing CD34 and Ki67 confirmed that 7 days VPA treatment caused a significant elongation of G1 phase and a slight, but significant, increase in G0 cells (CD34 + / Ki67 2 ; Figure 2 C and D). Finally, as shown in Figure 2E, expression of CDK inhibitors (p14 ARF , p16 INK4 , p21 Cip1/Waf-1 ) was increased at the same time point, although at a different extent, in VPA vs. CTR cells.
Histone hyperacetylation is directly linked to enhanced CD34 + expression and increased self renewal To determine whether shifting toward a homogeneous CD34 bright phenotype is consequent to histone hyperacetylation resulting from HDAC inhibition, Valpromide (VPM), a nonteratogenic VPA derivative with lower HDAC inhibition activity [15], was used in parallel to VPA in a 5 day time course experiment. VPA substitution with VPM inhibited at intermediate levels cellular growth, while it produced a CD34 expression profile in R1, R2 and R3 regions similar to CTR cells ( Figure 3A). Western blotting using pan-acetylated histone H4 (H4Ac) and lysine 9 acetylated histone H3 (H3K9Ac) showed that 7 days VPA treatment specifically enhanced histone tails acetylation ( Figure 3B). An immunoprecipitation (ChIP) experiment was then performed with chromatin extracted by cells grown for 7 days in the presence or the absence of VPA, using anti-H4Ac and anti-H3K9Ac antibodies, followed by real time PCR to detect enrichment of specific CD34 promoter regions. As shown in Figure 3C, H4Ac and H3K9Ac association to various regions upstream and downstream of the CD34 promoter transcriptional start site (TSS) was increased by VPA treatment, suggesting epigenetic regulation of CD34 expression.
Phenotype analysis, clonogenic activity and gene profiling of HDACi treated CD34 + cells Figure S5 shows the experimental design adopted for phenotypic and molecular characterization of HDACi-treated CD34 + cells. Seven days VPA-treated cells were first analyzed to measure the ability to extrude Rhodamine123 (Rho123) drug, a typical primitive stem cell feature relative to the MDR-1 gene product expression [16], and verify expression of Aldehyde Dehydrogenase (ALDH) [17] (Figures 4A, B and S6). These tests showed that VPA maintained a higher percentage of CD34 bright / Rho123 lo and CD34 bright /ALDH + cells compared with controls. Antibodies were then used to detect expression of stem cell (CD34, CD133, KDR) hematopoietic (CD3, CD4, CD8, CD14, CD38, CD45, CD48), endothelial (CD31, CD105, CD144, CD146, LDL uptake), mesenchymal (CD73, CD90, CD130, CD200) and integrin (a 2 b 1 , a 4 b 1 , a 5 b1) markers in multiparametric flow cytometry in CTR and VPA cells at 7 days. Results ( Figure 4C) indicated that VPA significantly enhanced CD34, CD38, CD48, CD133 and KDR expression, but also caused de novo expression of CD90, CD130 and CD146. Data obtained by prolonging the culture time up to 14 days ( Figure  S7) confirmed increased CD34, CD133 and KDR expression, induction of CD146 and increased CD31, and maintenance of a higher LDL uptake ability. Interestingly, expression of CD14 was significantly inhibited at both time points.
Recently, long term proliferating EPCs from cord blood cells characterized by CD34, CD38, CD45 and CD133 markers have been obtained using clonal culture conditions [18]. These cells resemble those named Endothelial Colony Forming Cells (ECFCs) in previous definitions of human EPCs identity [19]. To assess Figure 2. Analysis of CD34 + cells proliferation in the presence and the absence of VPA by flow cytometry. (A) CFSE staining profiles of control and VPA-treated cells at 5 and 7 days of culture. Note that the fluorescence intensity reduction as a consequence in cell proliferation was less pronounced in VPA vs. CTR cells at both time points. Proliferation index at 7 days was significantly reduced. (B) Seven days VPA treated cells had a higher frequency of slow dividing immature (CD34 bright ) stem cells (blue area in contour plots), as detected by co-staining with CFSE and CD34 antibody; plots on the left indicate the fluorescence profile of cells stained with CFSE and CD34 isotype antibody (iso). (C) Cell cycle analysis in 7 days CTR and VPA-treated cells revealed a higher frequency of cells in the G0-G1 and a lower percentage in either S and G2-M phases. (D) The percentage of cells specifically arrested in G0 was also increased, as detected by co-staining with anti Ki67 and CD34 antibodies at the same time point. * indicate P,0.05 by paired t-test (n$3). (E) Quantification of the relative expression level (2 2DDCt method) of small cyclin/CDK inhibitors (p14 ARF , p16 INK4 , p21 Cip1/waf1 and p27) by qRT-PCR analysis. * indicates P,0.05 by unpaired t-test (n$3). doi:10.1371/journal.pone.0022158.g002 whether VPA-treated CD34 + cells have a higher ability to form ECFC clusters, these cells were plated under ECFC inducing conditions [20]. Results showed that derivation of ECFC clusters was higher from VPA cells compared with CTR and freshly isolated CD34 + cells. Despite this, morphologic appearance, expression of ECFC markers and the ability to form capillarylike structures onto Matrigel was not different in CTR vs. VPA cells-derived ECFCs ( Figure 4D).
To investigate possible gene expression changes in 7-days cultured VPA vs. CTR cells, a survey of stem cells-specific transcripts expression was performed using qRT-PCR low density arrays ( Figure 5A). Unsupervised clustering analysis recognized coherently upregulated or downregulated stem cell-associated genes, allowing discriminate between VPA vs. CTR cells signatures. Statistical analysis revealed 34 transcripts significantly modified by VPA. Of these, 6 were down-modulated while 18 were up-regulated. Finally, 10 genes that were not detectable or were at detection limit by qRT-PCR, were expressed de novo by HDAC inhibition (Table 1). A microRNA (miRNA) profiling was finally performed on high-throughput sorted CTR and VPA CD34 bright cellular populations ( Figure S8). Remarkably, unsupervised clustering clearly distinguished miRNA profiles of CD34 bright cells in VPA vs. CTR conditions ( Figure 5B). Results from statistical analysis identified 44 miRNAs whose expression was significantly modified. Of these, 6 were downregulated, 31 were upregulated, while 7 were expressed de novo (Table 2).
Heart repair potency of control and HDACi-treated CD34 + cells Seven days cultured CTR and VPA cells were injected in the ischemic left ventricle of SCID beige mice 15 min after coronary artery ligation (CAL). Six weeks after CAL and cell injection, mice that survived and reached the end of the follow up period were analyzed for heart function and morphometry. Mortality data during the follow up period ( Figure 6A) showed a dramatic difference between animals receiving VPA cells compared to those injected with CTR cells or saline. Echocardiography revealed that in animals injected with VPA cells Left Ventricular Ejection Fraction (LVEF) was significantly increased (33% improvement) compared with saline-injected mice; in addition VPA cells reduced Left Ventricular End Diastolic/Systolic volumes (LVEDV, LVESV) compared with CTR cells and saline ( Figure 6B, Tables S1, Figure S9). To further assess efficacy of CTR and VPA cells, diastole-arrested hearts from sham operated, saline-, CTR-and VPA cells-injected mice were analyzed by histology to obtain LV morphometric parameters and by fluorescent lectin staining to measure capillary density ( Figure 6C, D; Table S2). Surprisingly, VPA-treated cells neither reduced infarct size nor increased capillary density, compared with CTR cells.
Human cells survival at 6 wks after transplantation was evaluated using a qPCR method tailored to detect a human polymorphism with an efficiency of 0.1-1 equivalent human cell genome (0.6-6 pg) in an unrelated DNA mixture [21] ( Figure  S10). By this, it was found that survival of human cells in vivo was low (about 10 2 cells out of the 1.5610 5

injected cells) and comparable in VPA and CTR cells-injected mice.
It has been proposed that cytokines released from cells injected into the ischemic myocardium may interfere, at least in part, with progression of cell death due to hypoxia or may metabolically sustain ischemic myocardium [22]. To better characterize effects of VPA vs. CTR cells, release of inflammatory/pro-angiogenic cytokines was measured in conditioned medium. This tests showed a significantly higher expression of 11 out of 15 tested cytokines in VPA vs. CTR cells ( Figure 7A, B). To assess whether modified cytokine expression in VPA vs. CTR-treated CD34 + cells protects against hypoxia-induced apoptosis, HL-1 cardiomyocyte-like cells [23] were exposed to medium conditioned from VPA and CTR cells in an in vitro hypoxia model. As shown in Figure 7C, VPA cells conditioned medium exerted higher protection of these cells from cell death consequent to hypoxia.
Increased paracrine activity of injected cells might interfere with myocardial adverse remodeling following infarction. To verify this, collagen deposition and myofibroblasts number were determined by picrosirius red staining [24,25] and a-smooth muscle actin (a-SMA) immunofluorescence [26,27] in hearts injected with saline, CTR and VPA-treated cells. Results (Figure 8) showed that collagen deposition and presence of a-SMA + myofibroblasts were (although not significantly) reduced in VPA cells-injected compared with CTR cells and saline-injected mice, suggesting lower adverse remodelling in hearts injected with VPA-treated cells.

Discussion
Direct and pleiotropic roles of HDACi blockade on CD34 + cells markers expression, cell cycle and phenotype For years recognized as an useful treatment for inducing differentiation of transformed cells [28], histone deacetylase . Bar graph on the right indicates quantification of CD34 bright /Rho123 lo cells in C and VPA conditions; dotted line indicate the percentage of CD34 bright /Rho123 lo cells immediately after isolating CD34 + cells from cord blood. (B) Quantification of ALDH expressing cells. Contour plots on the left show the fluorescence profile of cells treated with the ALDH inhibitor DEAB (used as a negative control) and CD34 isotype antibody (i+D), while those on the right show the results of specific staining with CD34 antibody and fluorescent detection of ALDH activity. Note that in the presence of VPA a higher percentage of CD34 bright /ALDH + cells was present (blue area), while in control cells ALDH staining was lower in the CD34 bright gating and present in CD34 dim/neg cells (areas in magenta color). Bar graph on the right indicates quantification of CD34 bright /ALDH + cells in C and VPA conditions; dotted line indicate the percentage of these cells immediately after isolating CD34 + cells from cord blood. * indicate P,0.05 by paired t-test (n = 4). (C) phenotype analysis of control and VPA cells at 7 days of culture by multiparametric flow cytometry experiments. Upregulation of stem cells markers CD34, CD133, CD38 and KDR were found along with enhanced expression of mesenchymal markers CD90, CD146 and CD130. Consistent with an effect of VPA on myeloid differentiation inhibition, CD14 was inhibited. * indicate P,0.05 by paired t-test (n$3). (D) Derivation of ECFCs from fresh, CTR and VPA CD34 + cells. Formation of clusters was observed three weeks after plating these cells onto FN coated dishes. Histogram represents number of ECFC clusters observed in three independent experiments; * indicates P,0.05 by one way ANOVA with Neuman Keuls post-hoc. Pictures on the upper right show the morphology of ECFCs derived from CTR and VPA CD34 + cells and their ability to form capillary-like structures, when plated onto matrigel. The amount of these latter structures formed by either CTR or VPA cells was not different compared with HUVEC cells that were used as a positive control (not shown). Plots in the lower part of the panel show expression of typical ECFC markers [19]. doi:10.1371/journal.pone.0022158.g004 (HDAC) pharmacologic inhibition has been more recently indicated to promote self renewal for expanding the number of immature stem cells in bone marrow repopulation assays [14,29,30,31,32]. In line with these investigations, our experiments show that CD34 + cells were functionally modified by HDACi pretreatment and induced to maintain a ''slow dividing'' phenotype [33] compared with CTR cells. Increased immaturity and retarded growth were not the unique consequences of VPA treatment. In fact we observed: 1) negative and positive modulation of various stem cells and differentiated cells markers such as CD14, VEGFR2/KDR, CD31, CD38, CD48 and uptake ability of Ac-LDL (Figures 4, S7), 2) de novo expression of mesenchymal (CD90, CD130) and ECFCs/mature endothelial cells (CD146) [19,34,35] markers (Figures 4, S7) and, 3) enhanced ability of VPA CD34 + cells to produce ECFCs clones (Figure 4). Taken together, these results suggest that HDAC blockade causes a phenotype change of CD34 + cells due to activation of different, but likely interdependent, molecular pathways targeting at the same time cell proliferation and clonogenicity. It is striking that upregulation of stem cells markers (CD34, CD38, CD133 and KDR) occurred in concert with de novo expression of markers typical of differentiated (CD48, CD146) or mesenchymal (CD90, CD130) cells. This is in line with the evidence that exposure to drugs with genome-wide effects such as generalized histone acetylation has global consequences for cell (re)programming [36,37], but also for fate determination of primitive cells [38,39,40]. mRNA and miRNA profiling of HDACi-treated CD34 + cells reveals a streamlined epigenetic supervision of immature CD34 + cells phenotype The clusterization of mRNAs and miRNAs expressed in VPAtreated cells ( Figure 5) allowed clearly recognize specific gene expression signatures distinguishing them from CTR cells. For mRNAs it was possible to derive functional annotation charts describing occurrence of HDACi-regulated mRNAs into various BIOCARTA/KEGG categories (Table 3). This identified the canonical Wnt-(FZD1, WNT1), Notch-(NOTCH1, DLL3, DLL4) and Hedgehog-activated signaling (DHH, BMP2) as crucial nodes in the generation of the VPA-treated cells phenotype. This is important, as convergence of canonical-Wnt and Notch signaling is recognized to maintain primitive stem cells self renewal in the bone marrow stem cell niche [41], while Notch signaling has specific roles in EPCs neo-vascularization activity [42].
Our observations also reveal a possible interplay between positive and negative stimuli controlling the immature and slow dividing phenotype of VPA-treated cells. In fact, VPA caused downregulation of CDH1 (E-cadherin) and upregulation of FDZ1, WNT1, CCND1, DLL1/3 and mir-9 in VPA-treated CD34 + cells, suggesting enhancement of b-catenin-mediated transactivation and positive effects on CD34 + cells proliferation [43]. On the other hand, HDAC inhibition also determined coherent up-or downmodulation of several miRNAs directly involved in positive (e.g. mir-129-3p, mir-193b, mir-370) or negative (e.g. mir-196b, mir-335, mir-370) control of cell cycle [44,45,46,47,48,49], thus suggesting the existence of an epigenetically regulated negative loop protecting CD34 + cells from unrepressed cellular growth, and reinforcing the anti proliferative effect exerted by small cyclin/ CDK inhibitors such as p14 ARF , p16 INK4 and p21 Cip1/Waf1 gene products ( Figure 2E). Negative effects on cell proliferation may also depend on an HDACi-related modification of the DNA methylation status. This is suggested by the finding that VPA induced mir148, a miRNA targeting DNMT3b methyltransferase, and by the evidence that several of the miRNAs over-or underexpressed in VPA-treated cells are transcriptionally regulated by CpG islands methylation [50,51,52].
Cardiac protection by HDACi-treated CD34 + cells is independent of CD34 + cells regeneration enhancement HDAC inhibitors potently reduce in vitro and in vivo angiogenesis by repressing the ability of mature endothelial cells to form vascular structures [53] or by inhibiting EPCs maturation into endothelium [54]. In addition, HDAC genes targeting (e.g. SIRT-1 or HDAC4,7) impairs vascular development [55,56,57]. Therefore, preconditioning with HDACi should reduce and not increase CD34 + cells pro-angiogenic function into ischemic tissues. Injection of CTR and VPA-treated cells in the ischemic heart ( Figure 6) showed a remarkable effect of HDACi cellular preconditioning on the survival of treated mice. This was associated to a significant enhancement of cardiac function but, surprisingly, neither corresponded to a more efficient reduction of the infarct size, nor to a significant improvement of myocardial tissue regeneration compared with CTR cells (Figure 6, Tables S1, S2). Furthermore, as shown by survival of similar, but low, amount of living human cells in the host myocardium ( Figure S10), this effect was not due to a higher engraftment ability of VPA-treated cells.
How to reconcile these data? Our results call for a generalized increase of CD34 + cells cardioprotection ability or an improved ''paracrine effect'' that may sustain cardiac contractility or interfere with myocardial cells apoptosis as short times after infarction. In support of this hypothesis is the finding that VPAtreated cells showed an enhanced expression of several proangiogenic/pro-inflammatory cytokines at mRNA (FGF2, CXCL12/SDF-1) or protein (bFGF, IL-8, VEGF, Ang-2, IFN-c, TNFa) levels, of cardioprotective factors such as Follistatin [58] and of resident progenitors activating factors such as HGF [59]. Therefore, generalized upregulation of all these gene products in the secretome of CD34 + cells may lead to functional preservation of the left ventricle by sustaining cardiac metabolism and contractility, even in the absence of a net reduction of infarct  A and B, respectively). The analyses were performed initially using the whole datasets of genes (panel A, left heat map) or miRNAs (panel B, left heat map) that passed the quality assurance and filtering criteria (see Supplementary Methods), to assess whether expression profiles discriminates treatment groups. A second round of unsupervised hierarchical clustering was done on differentially expressed genes (panel B, right heat map) or miRNAs (panel C, right heat map), as selected by significance analysis (see Supplementary Methods), to identify biologically relevant co-expressed gene clusters. The mean centered level of expression of each gene/miRNA in each sample is represented with green, black, and red colour scales (green indicates below mean; black, equal to mean; and red, above mean). The dendrograms on top of each heat map display the unsupervised clustering of control and VPA-treated CD34 + cells using the whole or the differentially expressed gene/miRNA lists. The dendrograms on the left side of each heat map show the unsupervised clustering of the genes. Correlation coefficients are reported for both. See Supplementary Methods for data transformation and adjustments, distance metrics and linkage methods. doi:10.1371/journal.pone.0022158.g005 size compared with control CD34 + cells (discussed in [22]). The in vitro data showing significantly higher rescue of HL-1 cardiac myocytes from hypoxia in the presence of VPA cells conditioned medium ( Figure 7C) supports this conclusion.
Presence of bone marrow-derived or resident mesenchymal cells-derived myofibroblasts during early stages after infarction is  an important component of the innate immunity response to ischemia [60,61,62]. By contrast, chronic permanence of these cells after MI [63], or their increased number due to aging [64], is associated to enhanced fibrosis and ventricular dysfunction. In the present study, specific experiments to assess the dynamics of myofibroblasts accumulation at early time points after MI were not performed. However, the lower amount of collagen and the reduced myo-fibroblasts number observed at six weeks in the infarct zone (Figure 8), suggest that injection of VPA-treated cells partially prevented chronic scarring of the myocardial tissue, thus justifying the observed reduction of ventricular dysfunction. Again, this may depend on enhanced cytokine secretion observed in VPA-treated vs. CTR cells. In summary, the data shown in the present study identify a possible novel strategy to enhance the protective ability of human CD34 + cells against the consequences of myocardial infarction. Future studies will be needed to clarify the molecular mechanisms underlying the interplay between VPA-treated CD34 + cells and the ischemic micro-environment, thus allowing to draw a clear scenario for their enhanced repair function.

Ethics statement
Collection of cord blood samples was performed upon written consent and on a voluntary basis. An Institutional Review Board formal approval for cord blood collection at Melzo Hospital was obtained to this aim (December 12, 2008; authorization no 843). All animal studies conformed to national and international ''Guide for the Care and Use of Laboratory Animals'' and the Helsinki declaration. The protocols were reviewed and approved by the Animal Care and Use Committee at Mario Negri Institute, and by the Italian Health Ministry (protocol 0804 base1); they were further compliant with European directives and guidelines (Legislative Decree September 19, 1994

Cord blood collection, expansion and phenotype analysis of CD34 + cells
Isolation and culture of CD34 + cells were performed using a magnetic beads-based method (MINI-MACS) and a serum-free expansion medium [11] as detailed in Materials and Methods S1. At the indicated time points, cells were incubated with suitable combinations of monoclonal antibodies recognizing stem cellspecific, endothelial, hematopoietic or mesenchymal specific markers (1-10 mg/ml final concentration). Cell growth was assessed by incubating cells with CFSE or Propidium Iodide staining and by Ki-67 specific antibodies. To assess stem cell phenotype of VPA vs. control treated CD34 + cells, Rhodamine123 extrusion by MDR-1 gene product and ALDH enzyme activity were measured by flow cytometry. Methodology and reagents used for these tests are described in the Materials and Methods S1.

ECFC Clonogenic expansion
ECFC clonogenic assay was performed as already described [20]. Briefly, fresh, CTR and VPA-treated CD34 + cells were plated at low density into collagen-coated dishes. Clusters of rapidly expanding endothelial-like cells were counted and further expanded for phenotype and functional analyses, as described in Materials and Methods S1.

Chromatin immunoprecipitation
Chromatin immunoprectipitation was performed as described in Nanni et al., 2009 [65], using ChIP-IT Express Enzymatic kit, according manufacture's instruction (Active Motif). Enrichment of DNA sequences associated to segments of the CD34 gene promoter was tested using a quantitative method based on real-time PCR. Further details are provided in Materials and Methods S1.

Transcript and miRNA profiling
For detection of stem cells-associated transcripts, total RNA was extracted from control vs. VPA-treated cells after which the RT 2 Profiler PCR Array system (human stem cell-specific card, catalogue PAHS-405E,Version 4.26; SABiosciences) was used. For microRNAs profiling, total RNA was extracted from high throughput-sorted CD34 bright cells by flow cytometry. Profiling was performed by TaqMan Human MicroRNA A and B Arrays, version 2.0 (Applied Biosystems, USA). Both procedures were performed in a 7900HT (Applied Biosystems, USA) fast real time cycler. A complete description of cell sorting and RNA extraction procedures as well as of raw data (Tables S3, S4 and S5) normalization and statistical handling is provided in Materials and Methods S1.

Animal studies
A SCID beige mouse model of myocardial infarction by permanent left coronary artery ligation was used. Control and VPA-treated cells (1.5610 5 cells/animal) were injected in the left ventricle (LV) at the infarct border zone 15 min after CAL. As controls, sham operated and saline-injected animals were used. After a 6 weeks follow up period, heart functional analyses were performed by transthoracic ultra-imaging echocardiography (VisualSonics, Vevo 770), followed by mice sacrifice for histological analysis and immunofluorescence. Further details are provided in the Materials and Methods S1.
Cytokine, chemokine and growth factor detection; apoptosis in HL-1 cell line Bio-Plex assay (Bio-Rad Laboratories, Italy), a bead-based multiplex immunoassay, was used to quantify cytokines, chemokines and growth factors secreted in culture supernatant by Control and VPA-treated cells. Results are expressed as pg/ml/  (C) Effect of CTR and VPA CD34 + cells conditioned medium on rescue from apoptosis of HL-1 cardiomyocytes cell line exposed to hypoxia conditions. Data in the graph represent the percent variation in apoptotic death of HL-1 cells exposed to hypoxia in the presence of VPA treated CD34 + cells conditioned medium in comparison with medium conditioned by CTR cells; * indicate P,0.05 by paired t-test (n = 5). doi:10.1371/journal.pone.0022158.g007 10 5 cells. Conditioned medium from these cells was also used to assess protection from hypoxia-induced apoptosis in HL-1 cardiomyocyte cell line [23]. Further details are provided in Materials and Methods S1.

Cardiac engraftment analysis
Analysis of CD34 + cells engraftment was performed using a qPCR method allowing detection of a human-specific polymor-phism in an excess mouse DNA [21]. Further details are provided in Materials and Methods S1.

Data analysis
All results are expressed as mean 6 standard error. Statistical significance was determined with paired or unpaired Student's ttest, one-way Anova with Newman-Keuls post-hoc analysis or Two-ways Anova with Bonferroni post-hoc analysis. Collagen data are shown as percentage of the areas containing collagen normalized to total area sections, while myofibroblasts were determined by counting the number of a-SMA + cells in the infarct zone (n$6). Statistical analysis of these data by one-way ANOVA with Newman-Keuls post-hoc test did not reveal differences between treatment groups, although lower amount of Collagen and smaller myofibroblasts number were found in VPA cells-injected mice. doi:10.1371/journal.pone.0022158.g008 Figure S10 Determination of the human cells survival in the mouse heart. qPCR was performed by a Taqman amplification protocols to detect the human SNP C/T (rs6625561 Reference: NCBI SNP). Amplification plots in the upper right show the threshold cycle of increasing (106 higher at each dilution) amounts of human DNA into a fixed amount of mouse DNA, while the other amplification plots are derived from amplification of DNA extracted by saline injected or cells injected mice. Graph on the bottom shows the approximate linearity between threshold cycle and 106increasing amount human of human DNA into the fixed mouse DNA excess. Red and green lines indicate the CT values for Table 3. Gene-enrichment analysis and functional annotation clustering of cell stem related genes in 7-days VPA-treated CD34 + cells. amplification of heart DNA from one mouse receiving CTR cells and another receiving VPA-treated cells.

Supporting Information
(TIF)    Materials and Methods S1 Detailed description of experimental procedures. (DOCX)