Transcriptional corepressors HIPK1 and HIPK2 control angiogenesis via TGF-β-TAK1-dependent mechanism

Abstract

Several critical events dictate the successful establishment of nascent vasculature in yolk sac and in the developing embryos. These include aggregation of angioblasts to form the primitive vascular plexus, followed by the proliferation, differentiation, migration, and coalescence of endothelial cells. Although transforming growth factor-β (TGF-β) is known to regulate various aspects of vascular development, the signaling mechanism of TGF-β remains unclear. Here we show that homeodomain interacting protein kinases, HIPK1 and HIPK2, are transcriptional corepressors that regulate TGF-β-dependent angiogenesis during embryonic development. Loss of HIPK1 and HIPK2 leads to marked up-regulations of several potent angiogenic genes, including Mmp10 and Vegf, which result in excessive endothelial proliferation and poor adherens junction formation. This robust phenotype can be recapitulated by siRNA knockdown of Hipk1 and Hipk2 in human umbilical vein endothelial cells, as well as in endothelial cell-specific TGF-β type II receptor (TβRII) conditional mutants. The effects of HIPK proteins are mediated through its interaction with MEF2C, and this interaction can be further enhanced by TGF-β in a TAK1-dependent manner. Remarkably, TGF-β-TAK1 signaling activates HIPK2 by phosphorylating a highly conserved tyrosine residue Y-361 within the kinase domain. Point mutation in this tyrosine completely eliminates the effect of HIPK2 as a transcriptional corepressor in luciferase assays. Our results reveal a previously unrecognized role of HIPK proteins in connecting TGF-β signaling pathway with the transcriptional programs critical for angiogenesis in early embryonic development.

Figure 1. Increased proliferation and poor adherens junction formation in the endothelial cells of Hipk1 ⁻

^/− ;Hipk2 ^{− /−} embryos. (A–D) Whole-mount and confocal immunofluorescent images of the developing vasculature in the yolk sacs of wild-type and Hipk1 ^−/− ;Hipk2 ^−/− embryos. (E) Quantification of the branch points, vessel lengths, and avascular areas in E9.5 yolk sacs. (F) Whole-mount CD31 staining of E9.5 control and Hipk1 ^−/− ;Hipk2 ^−/− embryos. (G) Transverse sections of E9.5 embryos at the trunk level. Arrowheads in panels (G) and (G') indicate the blood vessels in control and Hipk1 ^−/− ;Hipk2 ^−/− mutant embryos. (H) Quantification of CD31+;BrdU+ endothelial cells in E9.5 yolk sacs, trunk blood vessels, or endocardium. (I–J') EM analyses show reduced size and density of adherens junctions in Hipk1 ^−/− ;Hipk2 ^−/− endothelial cells. Panels (I) and (I') are low-magnification EM images, whereas (J) and (J') are high-magnification images. Arrowheads in (I) and (I') indicate intact endothelial cells that show no evidence of leakiness. Arrows in (J) and (J') highlight the presence of adherens junctions in both control and Hipk1 ^−/− ;Hipk2 ^−/− mutants, though the size and density of adherens junction are reduced in Hipk1 ^−/− ;Hipk2 ^−/− mutants. (K) Quantification shows the reduced length and density of adherens junction in E9.5 Hipk1 ^−/− ;Hipk2 ^−/− mutants.

Figure 2. HIPK1 and HIPK2 suppress MEF2C-mediated Mmp10 and Vegf expression.

(A) Expression profiles of Hipk1 ^−/−;Hipk2 ^−/− embryos reveal marked up-regulation of several angiogenic genes. The mRNA levels in Hipk1 ^−/− ;Hipk2 ^−/− mutants are normalized to those in control. (B) Immunohistochemical analyses confirmed the increased expression of VEGFA, MMP10, and PAI1 in the endocardium and endothelial cells of Hipk1 ^−/−;Hipk2 ^−/− embryos. Arrowheads indicate endothelial cells, and VL stands for vascular lumen. (C) Schematic diagrams of the 886-bp upstream regulatory sequence of the Mmp10 locus. The potential binding sites for SBE and MEF2 and the mutations of SBE or MEF2 sites are shown. (D) MEF2 binding element, but not the SBE, is required for HIPK2 to suppress Mmp10-luciferase activity. (E) Mmp10-luciferase reporter assays in HEK293T cells show that the kinase activity and the protein–protein interacting domain of HIPK2 are required to suppress MEF2C-mediated activation of Mmp10 expression. (F) HIPK1 and HIPK2 cooperatively suppress MEF2C-mediated activation of Mmp10 reporter. (G) Acute knockdown of Hipk2 in HEK293T cells using siRNA promotes the activation of Mmp10-luciferase reporter in the absence or presence of MEF2C. (H) Vegf-luciferase reporter can be activated by MEF2C and suppressed by HIPK2 in HEK293T cells. Student's t test, n = 3 (*p<0.05, **p<0.01, when compared to Vegf-Luc alone; ^# p<0.05, ^## p<0.01, when compared to the same condition without exogenous HIPK2). (I) HIPK1 and HIPK2 show cooperative and additive effects in suppressing MEF2C-mediated Vegf-Luc activity. Data are shown as mean ± s.e.m. Student's t test, n = 3 (*p<0.05, **p<0.01).

Figure 3. HIPK2 suppresses Mmp10 expression through interaction with MEF2C and HDAC7.

(A) Co-IP assays using protein lysates from HEK293T cells expressing HIPK2 and MEF2C show that HIPK2 can be detected in a protein complex with MEF2C (upper panels). Similar protein complex formation between endogenous HIPK2 and MEF2C can also be detected in wild-type MEF cells (lower panel). The interaction appears to depend on HIPK2 kinase activity as the kinase inactive HIPK2-K221A shows much reduced interaction with MEF2C. (B) ChIP assays using native chromatin from HUVECs show that HIPK1, HIPK2, and MEF2C can be detected in the promoter sequence of Mmp10. (C) Co-IP assays reveal that HIPK2 can be detected in a protein complex with HDAC7 and MEF2C. (D) HIPK2 and HDAC7 cooperatively suppress MEF2C-dependent activation of Mmp10-luciferase reporter activity. In the presence of the HDAC7 that lacks MEF2C interaction domain, HIPK2 can still suppress MEF2C in the activation of Mmp10 reporter. (E) HDAC7 continues to suppress MEF2C-dependent activation of Mmp10 in HEK293T cells where the endogenous Hipk2 mRNA is reduced by siRNA. (F) Co-IP assays showing that TGF-β and TAK1 enhance the interaction between MEF2C and HIPK2. (G) TGF-β and TAK1 enhance the corepressor effects of HIPK2 on MEF2C-mediated Mmp10 expression. (H) Co-IP assays using HUVEC cell lysates show that TGF-β promotes the interaction of endogenous HIPK2, TAK1, and MEF2C. Data are shown as mean ± s.e.m. Student's t test, n = 3 (*p<0.05, **p<0.01).

Figure 4. TGF-β–TAK1 promotes HIPK2 activity through protein–protein interaction and protects HIPK2 from proteasome-mediated degradation.

(A) TGF-β promotes HIPK2 kinase activity in HEK293T cells, whereas kinase inactive HIPK2-K221A shows no incorporation of γ-³²P-ATP upon TGF-β treatment. (B) The ability of TGF-β to activate HIPK2 kinase activity can be blocked by TGF-β type I receptor inhibitor SB431542. (C and D) TGF-β and wild-type TAK1 activate HIPK2 kinase and maintain the stability of HIPK2 protein. In contrast, dominant negative TAK1 (DN-TAK1) promotes HIPK2 degradation via the proteasome pathway.

Figure 5. TGF-β activates HIPK2 by phosphorylating a highly conserved tyrosine residue on position 361.

(A) Amino acid sequence alignment of the HIPK protein family from human and mouse reveals a stretch of highly conserved residues from position 346 to 371 in the activation segment of the subdomain VII in HIPK2. (B) Alignment of the similar regions of HIPK2 (346 to 371) from different species confirms that these amino acid residues are highly conserved from nematodes to the vertebrates. Conserved amino acids that can potentially be phosphorylated in MAPK signaling pathway are shown in bold. (C) The combined immunoprecipitation and in vitro kinase (IP-IVK) assays show that TGF-β treatment promotes the ability of wild-type HIPK2 to incorporate γ-³²P-ATP. In contrast, kinase inactive HIPK2-K221A fails to incorporate γ-³²P-ATP. While HIPK2-S359A and HIPK2-T360A mutant proteins can still incorporate γ-³²P-ATP in response to TGF-β treatment, the Y361F mutation in HIPK2 completely eliminates its ability to incorporate γ-³²P-ATP. (D) TGF-β and TAK1-induced phosphorylation of HIPK2 occurs primarily on Y361 residue in HIPK2. HIPK2-Y361F mutant completely loses its ability to incorporate γ-³²P-ATP upon activation by TGF-β or TAK1. Data are shown as mean + s.e.m., n = 3. Statistics in (C) and (D) use Student's t test. *p<0.05, **p<0.01, ns = not significant.

Figure 6. Mutation in Y361 of HIPK2 abolishes its ability to suppress Mmp10-Luc activity.

(A) Phospho-specific antibody for HIPK2-P-Y361 confirms that TGF-β–TAK1 can indeed promote the phosphorylation of wild-type HIPK2 protein, but not HIPK2-Y361F mutant, in HEK293T cells. (B) TGF-β type I receptor (TβRI) inhibitor SB431542 blocks TGF-β–induced phosphorylation, but not basal phosphorylation, on Y361 residue in HIPK2 in HUVEC cells. (C and D) Phosphorylation on the Y361 residue of HIPK2 is required for the transcriptional suppressor effect of HIPK2 on Mmp10 in HEK293T cells (C) and HUVEC cells (D). Data in (C) and (D) are shown as mean ± s.e.m., n = 3. Statistics in (C) and (D) use two-way ANOVA. *p<0.05, ns = not significant.

Figure 7. Perturbations of TGF-β signaling in endothelial cells recapitulate angiogenesis defects in Hipk1 ⁻

^/− ; Hipk2 ^{− /−} mutants. (A and B) Confocal images for the analyses of BrdU-incorporation in endothelial cells using anti-CD31 (green) and anti-BrdU (red) antibodies. Images are obtained from the blood vessels at the trunk region (panels A, A') and heart (panels B, B') of E9.5 control (TβRII^fl/fl) and Tie2-Cre;TβRII^fl/fl embryos. Arrows indicate CD31 and BrdU double positive cells. (C) Quantification of CD31 and BrdU double positive cells from blood vessels at the trunk level and the endocardium. Data are shown as mean ± s.e.m. Student's t test, n = 3. (D) qRT-PCR analyses show similar abnormalities in TGF-β target genes and angiogenic genes in Tie2-Cre;TβRII^fl/fl embryos at E9.5. Data are shown as mean ± s.e.m, n = 3. (E–G) Matrigel assays show capillary-like structure formation in HUVECs treated with control siRNA, Hipk1/2 siRNA, or TβRI siRNA. (H) Quantification of the number of branch points in Matrigel assays. (I–M) siRNA knockdown of Hipk1/2 and TβRI promotes BrdU incorporation (I–L) and up-regulation of Mmp10, Vegf, and Pai1 mRNA levels (M) in HUVEC cells. Arrowheads in panels (I–K) indicate BrdU+ cells. (N and O) The ability of TGF-β to suppress the expression of Mmp10 and Vegf mRNA (panel N) and BrdU incorporation in HUVEC cells can be blocked by siRNA knockdown of Hipk1/2. Numbers in panels (H), (L), (M), and (N) are represented as means ± s.e.m. Student's t test, n = 3. *p<0.05, **p<0.01.

Figure 8. A working model for HIPK proteins in the transcriptional control of angiogenic gene expression in the downstream of TGF-β-TAK1 signaling pathway.