Histone Deacetylase 9 Promotes Angiogenesis by Targeting the Antiangiogenic MicroRNA-17–92 Cluster in Endothelial Cells
Jump to

Abstract
Objective—Histone deacetylases (HDACs) modulate gene expression by deacetylation of histone and nonhistone proteins. Several HDACs control angiogenesis, but the role of HDAC9 is unclear.
Methods and Results—Here, we analyzed the function of HDAC9 in angiogenesis and its involvement in regulating microRNAs. In vitro, silencing of HDAC9 reduces endothelial cell tube formation and sprouting. Furthermore, HDAC9 silencing decreases vessel formation in a spheroid-based Matrigel plug assay in mice and disturbs vascular patterning in zebrafish embryos. Genetic deletion of HDAC9 reduces retinal vessel outgrowth and impairs blood flow recovery after hindlimb ischemia. Consistently, overexpression of HDAC9 increases endothelial cell sprouting, whereas mutant constructs lacking the catalytic domain, the nuclear localization sequence, or sumoylation site show no effect. To determine the mechanism underlying the proangiogenic effect of HDAC9, we measured the expression of the microRNA (miR)-17–92 cluster, which is known for its antiangiogenic activity. We demonstrate that silencing of HDAC9 in endothelial cells increases the expression of miR-17–92. Inhibition of miR-17–20a rescues the sprouting defects induced by HDAC9 silencing in vitro and blocking miR-17 expression partially reverses the disturbed vascular patterning of HDAC9 knockdown in zebrafish embryos.
Conclusion—We found that HDAC9 promotes angiogenesis and transcriptionally represses the miR-17–92 cluster.
Introduction
The family of histone deacetylases (HDACs) is composed of 4 classes based on their homology to yeast proteins: class I (HDAC1, 2, 3, and 8), class II (HDAC4, 5, 6, 7, 9, and 10), class III (Sirtuins), and class IV (HDAC11). Class II HDACs can be subdivided into class IIa with HDAC4, 5, 7, and 9 and the class IIb, which includes HDAC6 and 10. Class II HDACs are signal-responsive regulators of gene expression that shuttle between the cytoplasm and the nucleus. In the nucleus, HDACs remove the acetyl groups from histones leading to the formation of heterochromatin and gene silencing.1 Other than histones, HDACs can deacetylate many other proteins including transcription factors but also cytoplasmic proteins.2
See accompanying article on page 445
Broad-spectrum HDAC inhibitors block angiogenesis in vitro and in animal models3–5 and are in clinical development as antitumor agents.6,7 However, more selective inhibitors may have advantages, and several studies recently elucidated the effect of individual HDACs in the vasculature. HDAC6 and HDAC7 are positive regulators of vessel formation and are enhancing angiogenic functions of endothelial cells.8–12 In contrast, HDAC5 inhibits endothelial sprouting by repression of angiogenic guidance factors.13,14 Although HDAC9 is the closest homolog of HDAC5 (78% identic amino acids in the deacetylation domain and 57% overall identic amino acids) and has redundant effects in cardiac hypertrophy,15 a small interfering RNA (siRNA) screen in endothelial cells suggests that HDAC9 has opposing effects and is a positive regulator of angiogenic sprouting in vitro.13
MicroRNAs (miRNAs/miRs) are endogenous, highly conserved, noncoding small RNA molecules that regulate gene expression on the posttranscriptional level either by inhibiting translation or by promoting the degradation of mRNA. Mature miRs are generated by 2 major RNase III endonucleases, namely Dicer and Drosha.16,17 Inhibition of Dicer or Drosha expression impaired endothelial cell sprouting and neovascularization,18,19 and several proangiogenic and antiangiogenic miRs have meanwhile been identified, among them the miR-17-92 cluster.20 The miR-17-92 cluster is a polycistronic miR cluster encoding for miR-17, miR-18a, miR-19a/b, miR-20a, and miR-92a, which are transcribed as one common primary miR precursor. The miR-17-92 cluster controls tumor growth and hematopoiesis and exhibits effects on the cardiopulmonary system.21 In tumor cells, overexpression of the miR-17-92 cluster promoted tumor angiogenesis through blockade of the expression of endogenous angiogenesis inhibitors Thrombospondin 1 and connective tissue growth factor, which affect angiogenesis in a paracrine manner.22 In contrast, analyzing individual members of the cluster in endothelial cells revealed that the entire miR-17-92 cluster exhibits a cell-intrinsic antiangiogenic function in endothelial cells.23 miR-17 and miR-20a show antiangiogenic effects in vitro and in vivo,23 and miR-17 controls wound healing.24 Moreover, miR-92a blocked angiogenesis in vitro and in vivo, whereas inhibition of miR-92a enhanced neovascularization after hindlimb or myocardial ischemia.25 The miR-17-92 cluster is transcriptionally regulated by a variety of transcription factors, such as c-Myc, signal transducer and activator of transcription 3, p53, and E2F,26 and it was shown recently to be epigenetically controlled by an Egr2-depended recruitment of the histone demethylase Jarid1b.27
Here, we further elucidate the angiogenic function of HDAC9 in comparison with HDAC5 in vitro and in vivo. Because broad-spectrum histone deacetylase inhibitors were shown to control miR expression patterns in tumor cells,28 we further explored the regulation of miRs by HDAC9 in endothelial cells. We show that HDAC9 silencing inhibits endothelial cell sprouting in vitro and reduces vessel growth in Matrigel plug assays and in zebrafish in vivo. Furthermore, genetic deletion of HDAC9 reduces retinal vessel outgrowth and impairs blood flow recovery after hindlimb ischemia. The proangiogenic effects of HDAC9 are mediated by the repression of the miR-17-92 cluster, which controls several genes involved in angiogenesis.
Materials and Methods
Cell Culture
Pooled human umbilical vein endothelial cells (HUVECs) were purchased from Lonza and cultured in endothelial basal medium (Lonza) supplemented with hydrocortisone, bovine brain extract, epidermal growth factor, gentamycin sulfate, amphotericin-B, and 10% FCS (Invitrogen) until the fourth passage. After detachment with trypsin, cells were grown in 6-cm culture dishes for ≥24 hours.
Transfection
HUVECs were transfected at 60% confluence using GeneTrans II (Mobitec) according to the manufacturer’s protocol with 67 nmol/L of siRNA. siRNAs were synthesized by Eurogentec or Sigma-Aldrich. The following sequences were used: control siRNA siScr I against firefly luciferase (5´-CGUACGCGGAAUACUUCGA-3´)29 (used as standard control), siScr II (5´-GUGGGCACCGAUAUCUUGA-3´), siHDAC9 I (5´-GAAAGACACUCCAACUAAU-3´), siHDAC9 II (5´-CACAUUACCAGGAGCACAA-3´), and siHDAC9 3 untranslated region (siHDAC9 3UTR) (5´-GGACUUGAAAGGGCAUUAA-3´). Inhibition of miR-17, miR-18a, miR-19a, and miR-20a in vitro was achieved by transfection of miRIDIAN Hairpin Inhibitors (Dharmacon).
Quantitative Real-Time Reverse-Transcription Polymerase Chain Reaction
Total RNA was isolated using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. Afterward, 1 μg of RNA from each sample was reverse transcribed into cDNA and subjected to quantitative SYBR green PCR (StepOnePlus and Fast SYBR Green Mastermix, Applied Biosystems). GAPDH or RPLP0 expression served as loading control. Primer sequences are available on request.
miR Expression Analysis
Total RNA was isolated using miRNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. For miR detection, we used TaqMan MicroRNA Assays (Applied Biosystems). RNU48 served as control for HUVEC RNA. Quantitative real-time polymerase chain reactions were done on a StepOnePlus device (Applied Biosystems). Gene expression data were normalized to RNU48. The relative expression was determined using the formula 2-ΔCt.
Plasmid Transfection
Plasmids containing flag-tagged HDAC9 wild-type and constructs were described previously.30 Mock pcDNA 3.1 (Invitrogen) served as control. For overexpression HUVECs were transfected 24 hours before analysis using Superfect (Qiagen) according to the manufacturer’s protocol.
Luciferase Reporter Assay
miR-17-92 core promoter constructs were described previously.31 For measuring luciferase activity HEK293FT cells were grown in 24-well plates until 60% confluence. A total of 0.1 µg of Luciferase plasmid was cotransfected with 0.1 µg of pGL4 Renilla plasmid (Promega) as control for transfection efficiency together with 0.3 µg of HDAC9 wild-type, Mef2 interacting transcriptional repressor (MITR), or mock-control plasmid (pcDNA 3.1) using Lipofectamine 2000 (Invitrogen). The activity of Luciferase and Renilla was assessed after 48 hours with the Dual-Luciferase Reporter 1000 Assay System (Promega).
Western Blot Analysis
For Western blot analysis, cells were lysed in radioimmunoprecipitation assay lysis buffer (Sigma-Aldrich) for 15 minutes on ice. After centrifugation for 15 minutes at 20 000g (4°C), the protein content of the samples was determined according to the Bradford method. For separation of the nuclear protein fraction from the cytoplasmic protein fraction the NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (Thermo Scientific) was used according to the manufacturer’s protocol. Equal amounts of protein were loaded onto SDS-polyacrylamide gels and blotted onto polyvinylidene difluoride membrane. Western blots were performed by using antibodies directed against HDAC9 (Biovision), TOPO1 (Santa Cruz clone C15), HSP70 (Cell Signaling), Flag (Sigma-Aldrich clone M2), α-Tubulin (NeoMarkers clone DM1A), and Jak1 (Cell Signaling clone 6G4). Secondary antibodies were purchased from Jackson ImmunoResearch. Densitometry was performed where indicated for the quantification of the Western blots using the Scion Image software (version 4.0.2, Scion, Frederick, MD).
Angiogenesis Assay In Vitro
Endothelial cell spheroids of defined cell number were generated as described previously.32 In vitro angiogenesis was quantified by measuring the cumulative length of the sprouts, the number of sprouts, and the number of branch points that had grown out of each spheroid using a digital imaging software (Axiovision 4.6, Carl Zeiss Imaging Solutions GmbH, Munich, Germany) analyzing 10 spheroids per group and experiment. To measure tube formation, 75 000 HUVECs were seeded in 1 mL of endothelial basal medium (10% FCS) on the Matrigel Basement Membrane Matrix, as described previously.33
Hindlimb Ischemia Model
The animal experiments were approved by the Regional Board of the State of Hessen, Germany. HDAC9 knockout mice were described previously34 and were kindly provided by E.N. Olson (Dallas, TX). Three- to 4-month–old HDAC9 knockout mice and control littermates were subjected to hindlimb ischemia by ligation of the femoral artery and vein. Two weeks after induction of hindlimb ischemia, blood flow was assessed by laser-Doppler measurement, as described previously.35 Relative blood flow was determined as a ratio between blood flow of the ischemic compared with the nonischemic limb.
Murine Retina Vascularization Assay
The eyes of HDAC9 knockout mice or wild-type littermates were removed at day 5 postnatally and were fixed for 2 hours with 4% paraformaldehyde, as described previously.36 The retina was dissected and the hyaloid vessels were removed. Vessels were stained with isolectin B4, postfixed, and mounted for microscopic analysis. Overview images were taken with an Axio Observer Z1 (Axio Vision Rel 4.8, Carl Zeiss, Jena) using a ×20 objective. Number of branch points was assessed by manual counting of four 0.1-mm2 areas per retina at the vascular front between an artery and a vein. For the assessment of filopodia, pictures were taken using a laser scanning microscope (LSM510 META) and a ×40 objective.
In Vivo Spheroid Assay in Mice
In vivo Matrigel plug assays were performed as described previously.8 Briefly, a total of 3×105 HUVECs were transduced with short hairpin RNA virus, and transduced cells were selected with 0.4 mg/mL of puromycin for 24 hours. Cells were grown for 4 days, and spheroids were generated and mixed with Matrigel containing vascular endothelial growth factor (VEGF)-A and fibroblast growth factor 2. Spheroids were subcutaneously injected (1000 spheroids per plug) into severe combined immunodeficiency mice for 3 weeks. Mice received 150 mg of TRITC-lectin (Sigma-Aldrich) in physiological NaCl solution intravenously 20 minutes before euthanization to stain perfused vessels. Immunohistological analysis was performed as described previously.37,38 Microvascular density and size of the vessels were analyzed using ImageJ (National Institutes of Health). Perfusion was analyzed as TRITC-lectin–positive vessels in relation to hCD34-positive vessels.
Zebrafish Lines, Antibodies, and Reagents
Embryos of AB wild-type and the tg(fli1:EGFP) line39 were raised and staged as described.40 Embryos were kept in E3 solution at 28.5°C with or without 0.003% 1-phenyl-2-thiourea (Sigma-Aldrich) to suppress pigmentation and staged according to somite number or hours postfertilization.41
Morpholino Injection
Morpholinos were diluted in 0.1 mol/L of KCl to concentrations 2, 4, 5, 8, 12, and 13 µg/µL, respectively. One nanoliter of the morpholino dilution was injected through the chorion of 1-cell or 2-cell stage embryos. To attenuate possible off-target effects, a p53-Mo was coinjected 1.5-fold to the other morpholinos used. The following splicing-blocking (SB) and translation-blocking (TB) antisense morpholinos (Gene Tools) were used: HDAC9-SB-Mo I: 5´-GTACACATGAAGACAACTTACGTGT-3´ (exon 4-intron 4 junction); HDAC9-SB-Mo II: 5´-TCGTGATGATGAGCTCTTACTCTCT-3´ (exon 3-intron 3 junction); p53-TB-Mo: 5´-GCGCCATTGCTTTGCAAGAATTG-3´; miR-17-Mo: 5´-CTACCTGCACTGTAAGCACTTTGAC-3´; and standard control-Mo: 5´-CCTCTTACCTCAGTTACAATTTATA-3´.
Statistical Analysis
Data are expressed as mean±SEM. Two treatment groups were compared by Student t test. Multiple group comparisons were done by ANOVA using a least significant difference post hoc analysis (SPSS Inc). Results were considered statistically significant when P<0.05.
Results
HDAC9 Regulates Angiogenesis
To address whether HDAC9 regulates in vitro angiogenesis, we used 3 different siRNAs. Two of the siRNAs (siHADC9 I and siHDAC9 II) target the full-length HDAC9 (NM_178425.2) and the enzymatic inactive C-terminal HDAC9-splice variant MITR (NM_014707), which lacks the deacetylation domain. The third siRNA specifically targets the 3´ untranslated region of the full-length HDAC9 (siHDAC9 3UTR). Specificity of the siRNAs was tested 24 hours after transfection in HUVECs demonstrating that all 3 of the siRNAs targeted HDAC9 (Figure 1A), whereas MITR was targeted only by siHDAC9 I and siHDAC9 II (Figure IA in the online-only Data Supplement). Furthermore, silencing of HDAC9 did not reduce the expression of other class II HDAC members (Figure IB in the online-only Data Supplement). Western blot analysis of cytoplasmic and nuclear extracts confirmed the downregulation of HDAC9 by siRNAs and revealed that HDAC9 is preferentially localized to the nucleus of HUVECs (Figure 1B). However, protein expression of MITR was not detectable in HUVECs (data not shown).
Histone deacetylase (HDAC) 9 is required for in vitro sprouting angiogenesis. A, Human umbilical vein endothelial cells (HUVECs) were transfected with control small interfering RNAs (siRNAs) or siRNAs targeting HDAC9 for 24 hours. RNA was isolated and HDAC9 mRNA expression was determined (n=3). B, HUVECs were transfected with control siRNA or HDAC9 siRNAs for 48 hours. Nuclear and cytoplasmic protein fractions were separated and subjected to Western blot analysis. Left, Statistical analysis of nuclear HDAC9 protein expression. Right, Representative Western blot. TOPO1 serves as nuclear loading control and HSP70 represents cytoplasmic equal loading (n=3). C and D, Endothelial tube formation was assessed 24 hours after transfection of control or HDAC9 siRNA. C, Quantitative analysis (n=5). D, Representative images. E through H, HUVECs were transfected with control or HDAC9 siRNA for 24 hours and spheroid assays were applied. E, Quantification of the cumulative sprout length per spheroid (n=3–4). F, Representative images. G, Quantification of the number of sprouts per spheroid (n=3–4). H, Quantification of the number of branch points per spheroid (n=3–4). I, 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide viability assay of HUVECs treated for 24 or 48 hours with control siRNAs or HDAC9 siRNAs. Data are presented as % siScr I (n=3). *P<0.05 vs siScr I; #P<0.05 vs siScr II; statistically significant results were marked with * or #, respectively.
Silencing of HDAC9 significantly decreased endothelial network formation in Matrigel assays (Figure 1C and 1D). Moreover, in a spheroid assay, silencing of HDAC9 with 3 siRNAs reduced the cumulative sprout length (Figure 1E and 1F), the number of sprouts (Figure 1G), and the number of branch points (Figure 1H) per spheroid. In contrast, siRNAs directed against the HDAC9 homolog HDAC5 increased sprout formation, as described previously.13 Viability was not affected by any of the siRNA used (Figure 1I). Together, these data demonstrate that HDAC9 is required for angiogenic sprouting, whereas HDAC5 is an endogenous repressor of sprouting of endothelial cells in vitro.
HDAC9 Is Essential for In Vivo Vessel Formation in Mice and Zebrafish
Because silencing of HDAC5 and HDAC9 exhibits different effects on in vitro angiogenesis, we asked whether silencing of these HDACs might translate to different vascular phenotypes in vivo by using a previously described in vivo spheroid model.8,37 Therefore, we transplanted spheroids generated from HUVECs, which were transduced with HDAC5 or HDAC9 short hairpin RNA viruses, into severe combined immunodeficiency mice. Both short hairpin RNAs efficiently suppressed the respective HDAC isoform (Figure 2A) and regulated in vitro spheroid sprouting in a similar manner as shown for siRNA-mediated silencing (Figure 2B in comparison with Figure 1E and Urbich et al.13) Transplantation of shHDAC9-transduced spheroids in vivo results in a significantly decreased vascular density (Figure 2C and 2D), a reduced average size of the formed human CD34+ vessels (Figure 2E), and a profoundly reduced perfusion (Figure 2F). Silencing of HDAC5, however, slightly increased vascular density (Figure 2C through 2E) and significantly augmented the number of perfused human CD34+ vessels (Figure 2F).
Silencing of histone deacetylase (HDAC) 9 reduces in vivo vessel formation and perfusion, whereas silencing of HDAC5 increases vessel perfusion in vivo. A and B, Human umbilical vein endothelial cells (HUVECs) were transduced with control, HDAC9, or HDAC5 short hairpin RNA (shRNA) virus for 48 hours. A, mRNA expression of HDAC9 (n=4, left) and HDAC5 (n=4, right). B, Spheroid assay (n=3). C through F, HUVECs were transduced with control, HDAC9, or HDAC5 shRNA virus. Spheroids of these cells were generated in vitro and injected with Matrigel for 3 weeks into severe combined immunodeficiency (SCID) mice. Plugs were harvested and analyzed histologically. C, Representative images of the human vasculature (shown in green by staining against hCD34) and the perfusion in the plugs (shown in red by infusion of TRITC-lectin). Nuclei are shown in blue. D, Quantification of the microvascular density of the human vessels (n=3–5). E, Quantification of the average surface area of the human vessel (n=3–5). F, Quantification of the perfusion of the human vessels (n=3–5).
To further explore the in vivo function of HDAC9, we silenced HDAC9 in zebrafish embryos by application of 2 different splice blocking morpholinos (Figure IIA in the online-only Data Supplement). The splice blocking morpholinos directed against HDAC9 dose-dependently reduced HDAC9 expression (Figure 3A) and significantly increased the number of intersegmental vessel defects and dorsal longitudinal anastomotic vessel defects (Figure 3B and 3C). Furthermore, the percentage of animals harboring intersegmental vessel or dorsal longitudinal anastomotic vessel defects was significantly increased (Figure 3D). We observed the most pronounced phenotype in the formation of the parachordal lymphangioblasts, which was absent in nearly all HDAC9 morpholino-treated animals (Figure 3E and Figure IIB in the online-only Data Supplement). In contrast to HDAC9 morpholino treatment, HDAC5 morphant zebrafish embryos show a normal vascular patterning without vascular defects or hypersprouting (Figure IIIA through IIID in the online-only Data Supplement).
Silencing of histone deacetylase (HDAC) 9 by morpholino (Mo) treatment induces vascular defects in zebrafish embryos. A, Aberrant splicing of Danio rerio HDAC9 mRNA after HDAC9-splice blocking Mo injection as shown by reverse-transcription polymerase chain reaction (RT-PCR). Injection of the HDAC9-Mo I and HDAC9-Mo II strongly reduces HDAC9 wild-type (wt) signal demonstrating the functionality of the Mos. Whole zebrafish embryo mRNA was isolated 24 hours after Mo injection and subjected to RT-PCR. Actin mRNA expression serves as loading control. Morpholino design is shown in Figure IIA in the online-only Data Supplement. B through E, Phenotyping of HDAC9 morphants 48 hours postfertilization. For phenotyping of zebrafish embryos, 5 ng of control-Mo (Co-Mo), 5 ng of HDAC9-Mo I, or 2 ng of HDAC9-Mo II were injected. B, Quantification of defects in intersegmental vessel (ISV) and dorsal longitudinal anastomotic vessel (DLAV) for HDAC9 and control morphants (n=45–85; *P<0.05 for ISV; #P<0.05 for DLAV). C, Representative images of Co-Mo–treated embryos and HDAC9-Mo I–treated embryos. Vascular defects are indicated with arrows (red for parachordal lymphangioblasts [PL] defects, white for ISV defects, and grey for DLAV defects). D, Penetration of vessel defects for HDAC9 or control-Mo. Data are shown as percentage of animals with ≥1 ISV or DLAV defect (n=45–85; *P<0.05, Fisher exact test). E, Quantification of parachordal lymphangioblasts phenotype with absent = 0, partial absent = 1, and present = 2 (n=45–85). Statistically significant results were marked with * or #, respectively.
In summary, silencing of HDAC9 impairs vessel formation and maturation of human vessels in mice and vascular patterning in tg(fli1:EGFP) zebrafish, whereas downregulation of HDAC5 improves vessel perfusion.
HDAC9 Knockout Mice Exhibit a Reduced Blood Flow After Ischemia and a Reduced Retina Vascularization
Having shown that silencing of HDAC9 affects angiogenic responses in vitro and impairs vessel formation in zebrafish in vivo, we determined whether a genetic deletion of HDAC9 affects new blood vessel formation in mice. In a pathologically relevant setting of hindlimb ischemia, male HDAC9 knockout mice show a significantly reduced blood flow recovery in the ischemic leg compared with wild-type littermates 14 days after induction of ischemia (Figure 4A and 4B). Moreover, genetic deletion of HDAC9 reduces retinal vessel outgrowth (Figure 4C and 4D) and branching of vessels at the vascular front (Figure 4E and 4F), without affecting the number of tip cell filopodia (Figure IVA and IVB in the online-only Data Supplement). In summary, genetic deletion of HDAC9 abrogates vessel formation in 2 different vascular beds in mice.
Genetic deletion of histone deacetylase (HDAC) 9 causes impaired recovery after hindlimb ischemia and reduces retina vascular outgrowth. A, Representative images of relative blood flow by laser-Doppler measurements of male HDAC9 knockout mice and male control littermates 14 days after induction of hindlimb ischemia. White arrows indicate the ischemic limb. B, Quantification of relative blood flow by laser-Doppler measurements (n=3–4). C through F, Eyes of HDAC9 knockout mice and control littermates at day P5 were removed and retinal vessels were stained by isolectin B4. C, Overview images of wild-type and HDAC9 knockout retinas showing the outgrowth of retinal vessels. D, Quantification of retina vessel outgrowth (n=5–7). E, Representative images of vascular branching at the angiogenic front between an artery (A) and a vein (V). F, Quantification of vascular branching at the vascular front (n=6–7).
Effects of HDAC9 on Angiogenic Sprouting Depend on Its Nuclear Localization
Next we investigated the mechanism by which HDAC9 controls angiogenesis. Therefore, we overexpressed HDAC9 wild-type or different natural occurring splice variants of HDAC9.30 The splice variant lacking exon 7 (HDAC9 Δ 7) is devoid of the nuclear localization sequence. MITR misses the deacetylation domain, whereas the splice variant HDAC9 Δ 12 lacks exon 12, which contains a sumoylation site (Figure 5A). All constructs were appropriately expressed (Figure 5B) and localized (Figure V in the online-only Data Supplement and Reference 30). Overexpression of wild-type HDAC9 significantly increases in vitro sprouting compared with mock-transfected HUVECs (Figure 5C). The splice variant HDAC9 Δ 7, which is mainly localized in the cytoplasm (Figure V in the online-only Data Supplement) does not affect sprouting capacity (Figure 5C), indicating that nuclear localization of HDAC9 is required for angiogenesis. Additionally, overexpression of HDAC9 Δ 12 does not increase sprouting either (Figure 5C), indicating that sumoylation of HDAC9 might be required for proper function of HDAC9. Furthermore, overexpression of MITR, which is localized to dot-like structures within the nucleus (Figure V in the online-only Data Supplement) does not augment sprouting (Figure 5C). In summary, the proangiogenic activity of HDAC9 requires the deacetylation domain, nuclear localization, and sumoylation.
Histone deacetylase (HDAC) 9 increases sprouting angiogenesis and inhibits the expression of the miR-17-92 cluster. A, Scheme of HDAC9 wild-type and HDAC9 constructs. Alternative exon 12 of MITR is labeled with exon 12*. aa indicates amino acid; HDAC9 Δ 7, HDAC9 lacking exon 7; HDAC9 Δ 12, HDAC9 lacking exon 12; MITR, Mef2 interacting transcriptional repressor; and NLS, nuclear localization sequence. B and C, Human umbilical vein endothelial cells (HUVECs) were transfected with different HDAC9-flag tagged constructs. B, Western blot analysis of overexpression 16 hours after transfection. α-Tubulin serves as loading control. C, Spheroid assay with HUVECs overexpressing different HDAC9 constructs. Data are presented as percentage of mock control (n=6). D, HUVECs were treated with 1 µmol/L of TSA for 16 hours. mRNA was isolated and the expression of the primary miR-17-92 cluster was determined by quantitative polymerase chain reaction (qPCR; n=5). E, HUVECs were transfected with control small interfering RNAs (siRNAs) and 3 different HDAC9 siRNAs for 48 hours. Expression of the primary miR-17-92 cluster was determined by qPCR (n=6–9). F and G, HEK293FT cells were transfected with plasmids for firefly luciferase driven by a miR-17-92 core promoter construct or promoter deletion constructs. Firefly luciferase was coexpressed with mock, HDAC9 wild-type (wt), or MITR plasmids for 48 hours. As transfection control serves a Renilla luciferase plasmid which was co-expressed in all conditions. F, Scheme of miR-17-92 cluster core promoter and 2 deletion constructs driving the expression of firefly luciferase. A region that is rich for transcription factor binding sites is indicated as an activator region. Transcription start is marked with an arrow. G, Luciferase normalized to Renilla activity was measured. Data are normalized to P395 + mock condition (n=6–8; Student t test).
HDAC9 Represses the miR-17-92 Cluster
Previous studies demonstrate that broad-spectrum HDAC inhibitors transcriptionally regulate the expression of miRs28 and that miRs, particularly the miR-17-92 cluster, play key roles in angiogenesis signaling.23,25 Therefore, we explored the influence of HDAC9 on endothelial miR expression. Incubation of endothelial cells with the broad-spectrum HDAC inhibitor TSA for 16 hours or siRNA-mediated silencing of HDAC9 for 48 hours increases the expression of the miR-17-92 cluster (Figure 5D and 5E). Consistently, mature miRs of the miR-17-92 cluster were significantly increased by siRNAs directed against HDAC9 (Figure VIA in the online-only Data Supplement). The effect appears to be specific for HDAC9, because siRNA directed against other class II HDACs, namely HDAC4, HDAC5, HDAC6, and HDAC7, did not significantly affect miR-17-92 cluster expression (data not shown).
To determine whether HDAC9 transcriptionally controls the miR-17-92 cluster, we used luciferase reporter assays with the previously described core promoter of the miR-17-92 cluster31 spanning from –72 to +323 bp and –189 to +112/323 bp (Figure 5F). Construct P395 lacks a conserved transcription factor binding region (here called activator region), which is required for efficient promoter activity. Therefore, overexpression of neither HDAC9 nor MITR affects the activity of this construct. However, in the presence of the activator region (P512 and P301), overexpression of HDAC9 but not the deacetylase-deficient MITR construct reduced the expression of both reporter constructs (Figure 5G). Together these data demonstrate that HDAC9 transcriptionally represses the miR-17-92 cluster.
Proangiogenic Function of HDAC9 Is Mediated by the miR-17-92 Cluster
Next, we elucidated whether the HDAC9-dependent repression of the miR-17-92 cluster contributes to the proangiogenic effect of HDAC9. Therefore, we reduced the expression of the 4 most profoundly upregulated miRs of the cluster, namely miR-17, miR-18a, miR-19a, and miR-20a (here called miR-17-20a inhibitor), in HDAC9 siRNA-treated endothelial cells. For that purpose we used low doses of miR inhibitors to reduce the HDAC9 siRNA-mediated upregulation without abolishing the expression of the cluster members (Figure VIB in the online-only Data Supplement). Inhibition of miR-17-20a fully rescued the reduced sprouting and network formation observed after silencing of HDAC9 in vitro (Figure 6A). Because among the potent antiangiogenic members of the miR-17-92 cluster miR-17 was most profoundly upregulated after HDAC9 silencing, we additionally assessed whether inhibition of miR-17 alone is sufficient to rescue the impaired angiogenesis. As shown in Figure 6B, silencing of miR-17 partially rescues the sprouting defect of HDAC9 depletion in vitro. Consistently, silencing of HDAC9 decreases the protein expression of the validated miR-17 target Jak1 (Figure 6C), which was already implicated in miR-17–mediated suppression of angiogenesis.23 To evaluate whether inhibition of miR-17 is sufficient to recover vessel formation in vivo, we used a miR-17 morpholino. Application of the miR-17–morpholino induces a dose-dependent decrease in miR-17 expression in zebrafish embryos (Figure 6D). Silencing of miR-17 does not affect intersegmental vessel and dorsal longitudinal anastomotic vessel formation (data not shown). However, miR-17–morpholino partially rescued the intersegmental vessel and dorsal longitudinal anastomotic vessel defects caused by HDAC9 silencing (Figure VIC in the online-only Data Supplement) and normalizes the impaired parachordal lymphangioblast formation of HDAC9 morphants (Figure 6E), demonstrating that HDAC9 regulates vessel formation at least in part by inhibition of miR-17 in vivo (Figure 6F).
Histone deacetylase (HDAC) 9 regulates sprouting angiogenesis by inhibition of the miR-17-92 cluster. A, Human umbilical vein endothelial cells (HUVECs) were transfected with control small interfering RNAs (siRNAs) or HDAC9 siRNA together with a control microRNA (miR) inhibitor (100 nmol/L) or low doses of a combination of miR-17, miR-18a, miR-19a, and miR-20a inhibitors (each 25 nmol/L; here called miR-17-20a inhibitor). Sprouting capacity and network formation capacity were assessed by spheroid assay and tube formation assay (n=3 each). *P<0.05 vs siScr I + Co-Inh; #P<0.05 vs siHDAC9 I + Co-Inh; B, HUVECs were transfected with control siRNA or HDAC9 siRNA together with a control miR inhibitor (50 nmol/L) or miR-17 inhibitor (50 nmol/L), and sprouting capacity was assessed by a spheroid assay (n=5). C, HUVECs were transfected with control siRNA or HDAC9 siRNAs for 48 hours. Protein lysates were generated and Jak1 protein expression was determined (n=3). α-Tubulin served as loading control. Top, Representative Western blot. Quantification is shown below. D, Control or miR-17 morpholinos were injected in zebrafish embryos. After 24 hours expression of miR-17 was determined by quantitative polymerase chain reaction (qPCR) with RNU6 as loading control (n=3). E, Zebrafish embryos were injected with control- morpholino (Mo; Co-Mo; 13 ng without p53-Mo) or HDAC9-Mo I with or without miR-17-mo (5 ng of HDAC9-Mo I without p53-Mo; 5 ng of HDAC9-Mo I + 8 ng of miR-17- morpholino without p53-Mo). Formation of parachordal lymphangioblast was assessed at 72 hours postfertilization (n=112–165). F, Mechanism of HDAC9 in endothelial cells. HDAC9 augments angiogenesis and vessel formation in vitro and in vivo by repression of the antiangiogenic miR-17-92 cluster, which reduces the expression of proangiogenic proteins like Jak1. Additionally, HDAC9 might regulate angiogenesis by targeting other proangiogenic and antiangiogenic factors. Statistically significant results were marked with * or #, respectively.
Discussion
The data of the present article provide novel evidence that the class II HDAC9 is essential for angiogenic sprouting of endothelial cells in vitro and vessel formation in mice and zebrafish. The proangiogenic function of HDAC9 depends on its nuclear localization and the deacetylation domain, suggesting that HDAC9 transcriptionally controls angiogenesis-relevant genes. Indeed, HDAC9 represses the miR-17-92 cluster and thereby controls the expression of angiogenesis-relevant genes, such as Jak1.
The essential role of HDAC9 in angiogenesis was shown by several in vitro angiogenesis models and in vivo models. Silencing of HDAC9 profoundly reduced vascular growth of implanted human endothelial cells in vivo and disturbed vascular patterning in zebrafish. Furthermore, genetic deletion of HDAC9 impaired retinal vessel outgrowth and branching, as well as ischemia-induced neovascularization in mice. The function of HDAC9 is distinct from its close homolog HDAC5, which exhibits an antiangiogenic function in vitro (consistent with previous publications)13 and reduces perfusion of human vessels in vivo. In contrast to HDAC9, silencing of HDAC5 induced no phenotype in zebrafish embryos. The different function of HDAC9 versus HDAC5 is surprising, because both are close homologues and have redundant functions in regulating cardiac hypertrophy.15 Interestingly, in female mice (but not in male mice) genetic deletion of HDAC5 or HDAC9 was shown to increase capillary density and recovery after myocardial ischemia, and knockout of HDAC9 was shown to increase VEGF-A expression in an estrogen receptor-dependent manner in the injured heart.42 This is in line with our results, showing a striking sex difference of HDAC9–/– mice in the response to hindlimb ischemia. Although male HDAC9–/– mice show a significant reduction in blood flow recovery (Figure 4A and 4B), we observed a trend toward improved perfusion in female HDAC9–/– mice (Figure VII in the online-only Data Supplement). In female HDAC9–/– mice, the increased estrogen signaling in muscle cells was shown to induce a massive release of angiogenic growth factors (eg, VEGF-A), which was not seen in male mice42 and may override the endothelial cell-intrinsic impairment of angiogenic responses caused by HDAC9 deficiency.
Interestingly, in endothelial cells, silencing of HDAC9 inhibits VEGF-A expression (Figure VIIIA in the online-only Data Supplement), which is entirely different to the response seen in muscle cells.42 In addition, VEGF-A augments HDAC9 expression in endothelial cells (Figure VIIIB in the online-only Data Supplement), suggesting that in endothelial cells VEGF-A and HDAC9 are interacting in a positive feedback loop.
One potential mechanism underlying the different responses of muscle versus endothelial cells might relate to the finding that MITR is the predominant HDAC9-splice variant in the heart.34 This splice variant lacks the deacetylation domain and was not detectable on protein level in endothelial cells. Overexpression of MITR did not affect angiogenic signaling and had no effect on the expression of the miR-17-92 cluster in endothelial cells in our study, suggesting that the proangiogenic effects in endothelial cells are mediated by the full-length HDAC9, and one may speculate that the opposing effects on VEGF-A production in muscle cells of female HDAC9–/– mice might be mediated by the highly abundant muscle-enriched MITR splice variant.
The present study additionally discloses a novel mechanism by which the expression of the miR-17-92 cluster is transcriptionally controlled. Silencing of HDAC9 increased the expression of the primary miR-17-92 transcript, resulting in a significantly increased expression of the mature members of the miR-17-92 cluster and a downregulation of known miR-17 targets. Several studies document the direct transcriptional regulation of the miR-17-92 cluster by the transcription factors c-Myc, N-Myc, E2F, and p53.26 In hematopoietic cells, the expression of the cluster is additionally epigenetically regulated by the transcription factor Egr2, which recruits the histone demethylase Jarid1b to demethylate histone H3 at lysine 4, leading to a repression of the cluster.27 Here, we show that HDAC9 represses the expression of the miR-17-92 cluster in endothelial cells and that silencing of the miR-17-92 cluster rescues the antiangiogenic phenotype of HDAC9 depletion. Whether HDAC9 has further targets in addition to the miR-17-92 cluster, which contribute to angiogenesis, needs further evaluation. Furthermore, whether the regulation of the cluster by HDAC9 is mediated by histone modification or by interaction of HDAC9 with nonhistone proteins such as transcription factors or coactivators/repressors is currently not known. Certainly, the mechanism underlying the repression depends on the deacetylation domain of HDAC9, because the splice variant MITR did not repress the expression of the cluster. However, it remains to be determined whether the domain exclusively mediates its function via directly deacetylating proteins or whether it might have additional functions, for example, in recruiting other proteins. HDAC9 exhibits moderate deacetylation activity by its own.30 However, it is well known that HDAC9 and its splice variant MITR can recruit class I HDACs,30,43 which, in turn, mediate gene repression. Indeed, the broad-spectrum inhibitor TSA increased the expression of the miR-17-92 cluster in endothelial cells. Whether this is mediated by a direct repression of HDAC9 or by a repression of class I HDACs, which might be recruited by HDAC9, is not clear. Interestingly, the effect of TSA was cell-type specific, because under similar experimental conditions, TSA profoundly decreased the expression of the miR-17-92 cluster in tumor cells (Figure IX in the online-only Data Supplement), a finding that is consistent with a recent study.44 The reason for the cell type–specific regulation of miR-17-92 expression by TSA is currently unknown and requires further investigation.
Together, our study identified HDAC9 as positive regulator of endothelial cell sprouting and vascular growth, which represses the transcription of the antiangiogenic miR-17-92 cluster. The causal contribution of the regulation of the miR-17-92 cluster for the angiogenesis-regulatory effect of HDAC9 is supported by the finding that inhibition of the most profoundly regulated antiangiogenic miR of the cluster, namely, miR-17, partially rescued the impaired angiogenic sprouting induced by HDAC9 silencing. The transcriptional upregulation of HDAC9 by VEGF-A is consistent with the proangiogenic effects of HDAC9 and suggests that endogenous HDAC9 levels may contribute to the regulation of miRs under proangiogenic conditions.
Acknowledgments
We are grateful to Timothy W. McKeithan and Ming Ji at the Department of Internal Medicine, University of Nebraska Medical Center, for kindly providing the miR-17–92 core promoter luciferase constructs. We are thankful to Nicole Konecny, Andrea Knau, Ariane Fischer, and Eva Besemfelder for expert technical assistance and Katrin Bennewitz for excellent zebrafish work.
Sources of Funding
This study was supported by the Deutsche Forschungsgemeinschaft, SFB/TR23 TR-SFB 23 (projects Z5 to J.K., A3 to H.G.A., A2 to SD, and B5 to C.U.).
Disclosures
None.
Footnotes
The online-only Data Supplement is available with this article at http://atvb.ahajournals.org/lookup/suppl/doi:10.1161/ATVBAHA.112.300415/-/DC1.
- Received September 3, 2012.
- Accepted December 11, 2012.
- © 2013 American Heart Association, Inc.
References
- 1.↵
- 2.↵
- Choudhary C,
- Kumar C,
- Gnad F,
- Nielsen ML,
- Rehman M,
- Walther TC,
- Olsen JV,
- Mann M
- 3.↵
- 4.↵
- Rössig L,
- Li H,
- Fisslthaler B,
- Urbich C,
- Fleming I,
- Förstermann U,
- Zeiher AM,
- Dimmeler S
- 5.↵
- Qian DZ,
- Kato Y,
- Shabbeer S,
- Wei Y,
- Verheul HM,
- Salumbides B,
- Sanni T,
- Atadja P,
- Pili R
- 6.
- 7.
- 8.↵
- Kaluza D,
- Kroll J,
- Gesierich S,
- Yao TP,
- Boon RA,
- Hergenreider E,
- Tjwa M,
- Rössig L,
- Seto E,
- Augustin HG,
- Zeiher AM,
- Dimmeler S,
- Urbich C
- 9.↵
- 10.↵
- Mottet D,
- Bellahcène A,
- Pirotte S,
- Waltregny D,
- Deroanne C,
- Lamour V,
- Lidereau R,
- Castronovo V
- 11.↵
- Martin M,
- Potente M,
- Janssens V,
- Vertommen D,
- Twizere JC,
- Rider MH,
- Goris J,
- Dimmeler S,
- Kettmann R,
- Dequiedt F
- 12.↵
- Margariti A,
- Zampetaki A,
- Xiao Q,
- Zhou B,
- Karamariti E,
- Martin D,
- Yin X,
- Mayr M,
- Li H,
- Zhang Z,
- De Falco E,
- Hu Y,
- Cockerill G,
- Xu Q,
- Zeng L
- 13.↵
- Urbich C,
- Rössig L,
- Kaluza D,
- Potente M,
- Boeckel JN,
- Knau A,
- Diehl F,
- Geng JG,
- Hofmann WK,
- Zeiher AM,
- Dimmeler S
- 14.↵
- Ha CH,
- Wang W,
- Jhun BS,
- Wong C,
- Hausser A,
- Pfizenmaier K,
- McKinsey TA,
- Olson EN,
- Jin ZG
- 15.↵
- Chang S,
- McKinsey TA,
- Zhang CL,
- Richardson JA,
- Hill JA,
- Olson EN
- 16.
- 17.
- Thum T
- 18.↵
- Kuehbacher A,
- Urbich C,
- Zeiher AM,
- Dimmeler S
- 19.↵
- Suárez Y,
- Fernández-Hernando C,
- Yu J,
- Gerber SA,
- Harrison KD,
- Pober JS,
- Iruela-Arispe ML,
- Merkenschlager M,
- Sessa WC
- 20.
- 21.↵
- Ventura A,
- Young AG,
- Winslow MM,
- Lintault L,
- Meissner A,
- Erkeland SJ,
- Newman J,
- Bronson RT,
- Crowley D,
- Stone JR,
- Jaenisch R,
- Sharp PA,
- Jacks T
- 22.↵
- 23.↵
- Doebele C,
- Bonauer A,
- Fischer A,
- Scholz A,
- Reiss Y,
- Urbich C,
- Hofmann WK,
- Zeiher AM,
- Dimmeler S
- 24.↵
- 25.↵
- Bonauer A,
- Carmona G,
- Iwasaki M,
- et al
- 26.↵
- 27.↵
- Pospisil V,
- Vargova K,
- Kokavec J,
- Rybarova J,
- Savvulidi F,
- Jonasova A,
- Necas E,
- Zavadil J,
- Laslo P,
- Stopka T
- 28.↵
- Scott GK,
- Mattie MD,
- Berger CE,
- Benz SC,
- Benz CC
- 29.↵
- 30.↵
- Petrie K,
- Guidez F,
- Howell L,
- Healy L,
- Waxman S,
- Greaves M,
- Zelent A
- 31.↵
- 32.↵
- Diehl F,
- Rössig L,
- Zeiher AM,
- Dimmeler S,
- Urbich C
- 33.↵
- Carmona G,
- Göttig S,
- Orlandi A,
- Scheele J,
- Bäuerle T,
- Jugold M,
- Kiessling F,
- Henschler R,
- Zeiher AM,
- Dimmeler S,
- Chavakis E
- 34.↵
- 35.↵
- Rössig L,
- Urbich C,
- Brühl T,
- et al
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- Epting D,
- Wendik B,
- Bennewitz K,
- Dietz CT,
- Driever W,
- Kroll J
- 42.↵
- van Rooij E,
- Fielitz J,
- Sutherland LB,
- Thijssen VL,
- Crijns HJ,
- Dimaio MJ,
- Shelton J,
- De Windt LJ,
- Hill JA,
- Olson EN
- 43.↵
- Zhou X,
- Richon VM,
- Rifkind RA,
- Marks PA
- 44.↵
- Humphreys KJ,
- Cobiac L,
- Le Leu RK,
- Van der Hoek MB,
- Michael MZ
This Issue
Jump to
Article Tools
- Histone Deacetylase 9 Promotes Angiogenesis by Targeting the Antiangiogenic MicroRNA-17–92 Cluster in Endothelial CellsDavid Kaluza, Jens Kroll, Sabine Gesierich, Yosif Manavski, Jes-Niels Boeckel, Carmen Doebele, Arthur Zelent, Lothar Rössig, Andreas M. Zeiher, Hellmut G. Augustin, Carmen Urbich and Stefanie DimmelerArteriosclerosis, Thrombosis, and Vascular Biology. 2013;33:533-543, originally published February 13, 2013https://doi.org/10.1161/ATVBAHA.112.300415
Citation Manager Formats
Share this Article
- Histone Deacetylase 9 Promotes Angiogenesis by Targeting the Antiangiogenic MicroRNA-17–92 Cluster in Endothelial CellsDavid Kaluza, Jens Kroll, Sabine Gesierich, Yosif Manavski, Jes-Niels Boeckel, Carmen Doebele, Arthur Zelent, Lothar Rössig, Andreas M. Zeiher, Hellmut G. Augustin, Carmen Urbich and Stefanie DimmelerArteriosclerosis, Thrombosis, and Vascular Biology. 2013;33:533-543, originally published February 13, 2013https://doi.org/10.1161/ATVBAHA.112.300415