The Chromatin-Remodeling Enzymes BRG1 and CHD4 Antagonistically Regulate Vascular Wnt Signaling

Abstract

Canonical Wnt signaling plays an important role in embryonic and postnatal blood vessel development. We previously reported that the chromatin-remodeling enzyme BRG1 promotes vascular Wnt signaling. Vascular deletion of Brg1 results in aberrant yolk sac blood vessel morphology, which is rescued by pharmacological stimulation of Wnt signaling with lithium chloride (LiCl). We have now generated embryos lacking the chromatin-remodeling enzyme Chd4 in vascular endothelial cells. Unlike Brg1 mutants, Chd4 mutant embryos had normal yolk sac vascular morphology. However, concomitant deletion of Chd4 and Brg1 rescued vascular abnormalities seen in Brg1 mutant yolk sacs to the same extent as LiCl treatment. We hypothesized that Wnt signaling was upregulated in Chd4 mutant yolk sac vasculature. Indeed, we found that Chd4 deletion resulted in upregulation of the Wnt-responsive transcription factor Tcf7 and an increase in Wnt target gene expression in endothelial cells. Furthermore, we identified one Wnt target gene, Pitx2, that was downregulated in Brg1 mutant endothelial cells but was rescued following LiCl treatment and in Brg1 Chd4 double mutant vasculature, suggesting that PITX2 helps to mediate the restoration of yolk sac vascular remodeling under both conditions. We conclude that BRG1 and CHD4 antagonistically modulate Wnt signaling in developing yolk sac vessels to mediate normal vascular remodeling.

INTRODUCTION

Chromatin, which consists of DNA tightly wrapped around a scaffold of histone proteins, provides a means for the eukaryotic cell to fit a large amount of genetic information into the nucleus. It also controls gene regulation, since large transcriptional machinery cannot readily access gene promoters that are embedded in tightly compacted chromatin. Enzymes that modify chromatin structure therefore play important roles in modulating gene transcription. For example, ATP-dependent chromatin-remodeling complexes transiently disrupt bonds between histones and DNA to allow transcription factors, coregulatory proteins, and large transcriptional machinery to access gene promoters (14). Importantly, these complexes exercise striking specificity in selecting their genomic targets and can positively or negatively regulate transcription to modulate a number of developmental processes (15, 20).

Chromatin-remodeling complexes contain various numbers of proteins that contribute to their specific temporal and spatial effects on gene transcription (37). Importantly, these complexes contain catalytic ATPase subunits that are required for their function. Mammalian switch/sucrose-nonfermentable (SWI/SNF)-like complexes utilize the ATPases brahma (BRM, also known as SMARCA2) and brahma-related gene 1 (BRG1, also known as SMARCA4). While mice with global Brm deletion are viable and fertile (31), Brg1^−/− embryos die at implantation (1). Nevertheless, tissue-specific mutations have revealed a role for BRG1 in a variety of later developmental processes (2, 9, 11, 13, 16, 32, 33, 38, 40). For example, deletion of Brg1 from developing endothelial cells with a transgenic Tie2-Cre line (Brg1^fl/fl:Tie2-Cre⁺) results in yolk sac vascular abnormalities due in part to misregulated Wnt signaling (12).

Canonical Wnt signaling occurs when soluble extracellular Wnt ligands interact with a cell surface receptor complex containing a seven-transmembrane-domain frizzled (Fzd) family protein (5, 25). Interactions between Wnt ligands and their receptors stabilize the intracellular signaling molecule β-catenin by inactivating a cytoplasmic destruction complex that would otherwise target β-catenin for degradation. Stabilized β-catenin translocates to the nucleus, where it coregulates transcription of Wnt target genes by interacting with transcription factors in the T-cell factor (TCF)/lymphoid enhancer factor (LEF) family. BRG1 impacts the Wnt signaling pathway at two levels: it directly coregulates transcription of multiple Fzd receptors and a subset of Wnt target genes in yolk sac endothelial cells (YSECs) (12). As a result, overall Wnt signaling is downregulated in Brg1^fl/fl:Tie2-Cre⁺ yolk sac vasculature. In vivo pharmacological stimulation of Wnt signaling significantly restores Wnt target gene transcription in yolk sac blood vessels and rescues vascular morphology in Brg1^fl/fl:Tie2-Cre⁺ yolk sacs.

BRG1-containing SWI/SNF complexes can coordinate with other ATP-dependent chromatin-remodeling complexes, such as the nucleosome remodeling and deacetylase (NuRD) complexes, to modulate target gene transcription (8, 30). NuRD complexes combine multiple aspects of epigenetic regulation to modulate transcription of target genes. NuRD complexes contain, in addition to ATPase chromatin-remodeling enzymes, subunits that mediate histone deacetylation and demethylation and proteins that bind methylated DNA (29). Together, these subunits typically mediate transcriptional repression of NuRD target genes, although NuRD can promote transcription of certain target genes in vivo (36). The ATPase chromatin-remodeling enzymes associated with mammalian NuRD complexes are the chromodomain helicase DNA-binding (CHD) proteins CHD3 (Mi-2α) and CHD4 (Mi-2β). CHD3 and CHD4 have been studied less extensively than the SWI/SNF ATPases BRG1 and BRM in vivo. No targeted mutation of Chd3 has yet been described, but a conditional allele has been used to delete Chd4 from thymocytes and keratinocytes (18, 36). CHD4 is an autoantigen in a subset of patients with dermatomyositis, which is characterized by inflammation of muscles and skin (27). Despite the fact that multiple vascular pathologies, such as capillary fragmentation, inflammation, and misregulation of vascular markers, are associated with dermatomyositis (28), no role has yet been assigned to Chd4 or NuRD remodeling complexes in vascular development or homeostasis.

We have now deleted Chd4 from developing endothelial cells with a Tie2-Cre transgene and report that a number of Wnt target genes were upregulated in Chd4-floxed (Chd4^fl/fl):Tie2-Cre⁺ yolk sac vasculature. This study demonstrates that CHD4 modulates Wnt signaling in endothelial cells during vascular development by directly regulating expression of both the Wnt-responsive transcription factor Tcf7 and a subset of Wnt-responsive target genes. Moreover, concurrent deletion of Chd4 and Brg1 from vascular endothelium strikingly rescued vascular abnormalities seen in Brg1^fl/fl:Tie2-Cre⁺ yolk sacs. We conclude that Brg1 and Chd4 antagonistically modulate vascular Wnt signaling to mediate yolk sac angiogenesis.

MATERIALS AND METHODS

Mice.

A Tie2-Cre⁺ transgene was used to delete floxed Brg1 and Chd4 alleles from embryonic endothelial cells in vivo. Brg1-floxed mice (Brg1^fl/fl) (9), Chd4-floxed mice (Chd4^fl/fl) (36), and Tie2-Cre⁺ transgenic mice (19) were maintained on a mixed genetic background at the Oklahoma Medical Research Foundation animal facility. All animal use protocols were approved by the Institutional Animal Care and Use Committee.

Genotyping.

PCR genotyping of Brg1-floxed and Tie2-Cre⁺ transgenic embryos and mice was performed as described previously (12). Chd4-floxed mice and embryos were genotyped by PCR using the following primers: 5′-TCCAGAAGAAGACGGCAGAT-3′ (forward) and 5′-CTGGTCATAGGGCAGGTCTC-3′ (reverse). These primers flank the 5′ LoxP site and yield a 400-bp floxed allele and a 278-bp wild-type allele. The PCR was performed at an annealing temperature of 56°C. All histological sections were scraped from paraffin or OCT (Tissue-Tek) into DEXPAT reagent (TaKaRa) before genotyping.

LiCl injections.

Lithium chloride (LiCl) (400 mg/kg of body weight, dissolved in water) was injected intraperitoneally into pregnant female mice at embryonic day 8.5 (E8.5) and E9.5. Embryos were harvested for endothelial cell isolation at E10.5.

Yolk sac staining.

Whole-mount yolk sac immunostaining was performed as described previously (12). Hematoxylin and eosin staining, terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining, and benzidine staining were performed on histological sections of yolk sacs as described previously (11).

Immunofluorescence.

Brg1^fl/fl:Tie2-Cre⁺, Brg1^fl/fl; Chd4^fl/fl:Tie2-Cre⁺, and littermate control embryos were cryoembedded and sectioned (8 μm). Double immunostaining for BRG1 and PECAM-1 was performed as follows: cryosections were thawed and blocked in 3% normal donkey serum (Jackson ImmunoResearch)–3% bovine serum albumin (BSA; Rockland Immunochemicals)–0.3% Triton X-100–phosphate-buffered saline (PBS) for 1 h at room temperature. Anti-PECAM-1 (1:500; BD Biosciences catalog number 553370) was diluted in 1% normal donkey serum–1% BSA–0.1% Triton X-100–PBS and applied to sections for 1 h at 37°C. Sections were washed three times (2 min each) in 0.1% Triton X-100–PBS. Cy3–donkey anti-rat IgG (1:500; Jackson ImmunoResearch) and Hoechst stain (20 μg/ml) were diluted as described above and applied to sections for 1 h at room temperature. Sections were washed as described above and then blocked and stained with anti-BRG1 (1:100; Santa Cruz Biotechnology catalog number sc-17796) using the M.O.M. kit (Vector Laboratories). Alexa 488-streptavidin (1:500; Invitrogen) was used to detect the biotinylated mouse IgG supplied in the M.O.M. kit. Sections were washed three times (3 min each) in PBS, and a coverslip was applied with 2.5% 1,4-diazabicyclo[2.2.2]octane (DABCO)–90% glycerol–PBS, pH 8.6. Double immunostaining for CHD4 and PECAM-1 was performed similarly with the following exception: cryosections were blocked in 3% normal donkey serum–3% normal goat serum (Jackson ImmunoResearch)–3% BSA–0.3% Triton X-100–PBS for 1 h at room temperature. Anti-PECAM-1 (1:500) and anti-CHD4 (1:1,000; Active Motif catalog number 39289) were diluted in 1% normal donkey serum–1% normal goat serum–1% BSA–0.1% Triton X-100–PBS and applied to sections for 1 h at room temperature. Cy3–donkey anti-rat IgG, Alexa 488–goat anti-rabbit IgG (1:500, Invitrogen), and Hoechst dye were diluted and applied to sections for 1 h at room temperature.

Microscopy.

Fluorescent and light microscopic images were acquired as previously described (12).

Primary endothelial cell isolation.

Yolk sac endothelial cells (YSECs) were isolated with anti-PECAM-1-conjugated magnetic beads from E10.5 embryos as described previously (12).

Custom array.

Total RNA from primary YSECs was isolated from 4 to 6 independent experiments as described previously (12), and cDNA was prepared using the RT² first-strand kit (SABiosciences). Custom-designed Wnt signaling target gene RT² Profiler PCR arrays (SA Biosciences) were used as described previously (12). Data analysis and statistical determinations were performed using the Web-based PCR array data analysis tool available through the SABiosciences website.

qPCR.

To analyze transcript levels, total RNA from primary YSECs was isolated using TRIzol (Invitrogen) according to the manufacturer's instructions. The DNA-free kit (Ambion) was used to digest any contaminating DNA. cDNA was prepared using the iScript cDNA synthesis kit (Bio-Rad), and real-time quantitative PCR (qPCR) was performed using RT² Fast SYBR green qPCR master mix (SABiosciences) and the CFX96 detection system (Bio-Rad) with gene-specific primers.

qPCR primers.

The following qPCR primers were used: for β-actin, 5′-TGTTACCAACTGGGACGACA-3′ and 5′-GGGGTGTTGAAGGTCTCAAA-3′; for glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-TCAACGGCACAGTCAAGG-3′ and 5′-ACTCCACGACATACTCAGC-3′; for Pitx2, 5′-CGTTGAATGTCTCTTCTCCA-3′ and 5′-CTGGCCCTTATCTTTCTCAT-3′; for Chd4, 5′-GCTATGCCCGGTGGCAGGAC-3′ and 5′-CGGGTGAGAGGGGTCCTCGG-3′ for β-catenin, 5′-TGGCAGCAGCAGTCTTAC-3′ and 5′-GAGGTGTCAACATCTTCTTCC-3′; for Tcf3, 5′-CAGATGGTGGCCTGGATACT-3′ and 5′-CATCCCTGCTGTAGCTGTCA-3′; for Tcf4, 5′-GTCCTCGCTGGTCAATGAAT-3′ and 5′-CCCTTAAAGAGCCCTCCATC-3′; for Tcf7, 5′-GCCAGAAGCAAGGAGTTCAC-3′ and 5′-ACAGGGGGTAGAGAGGAGGA-3′ and for Lef1, 5′-TATGAACAGCGACCCGTACA-3′ and 5′-TCGTCGCTGTAGGTGATGAG-3′.

qPCR analysis.

The relative fold change in transcription was determined using the comparative threshold cycle (C_T) method and the β-actin and GAPDH housekeeping genes as internal controls. Data from six independent experiments were combined and are presented as means ± standard errors of the means (SEM). Statistical differences were detected using a two-tailed Student t test.

Cell culture and transfections.

C166 yolk sac endothelial cells (ATCC accession number CRL-2581) were maintained and transfected with 100 nM CHD4 siGENOME SMARTpool or nontargeting control small interfering RNA (siRNA) oligonucleotides (Dharmacon catalog number M-052142-01 or D-001210-01, respectively) as described previously (12).

Western blotting.

Total protein harvested from siRNA-transfected C166 endothelial cells was fractionated on a 9% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane for Western blotting with antibodies to CHD4 (Abcam; catalog number 72418), β-catenin (BD Biosciences; catalog number 610153), and GAPDH (Sigma; catalog number G9545). Relative band intensity was determined using ImageJ software (National Institutes of Health).

ChIP.

Chromatin immunoprecipitation (ChIP) assays were performed as described previously (12) with modifications. Chromatin was immunoprecipitated using a CHD4-specific antibody (Abcam; catalog number 70469). Mouse IgG (Invitrogen; catalog number 100005291) or the polyhistidine epitope tag (His) antibody (Rockland; catalog number 600-401-382) was used as a negative control. For total histone H3 ChIP, chromatin was harvested as described previously (12) from cells transfected with nonspecific or CHD4-specific siRNAs and immunoprecipitated using the histone H3-specific and negative-control antibodies supplied in the ChIPAb+ histone H3 kit (Millipore; catalog number 17-10046). Real-time quantitative PCR was performed using RT² Fast SYBR green qPCR master mix (SABiosciences) and the CFX96 detection system (Bio-Rad) with primer sets within the promoter region of Tcf7 (5′-TGTTCACACCAAGGTTCCAA-3′ and 5′-GGCCCACTGGGAATAATCTT-3′), Lef1 (5′-ATTTTCGCTAGGGTGGTGTG-3′ and 5′-TTCAGCAAAGGCAAACAGAA-3′), Pitx2 (5′-CGGTTTTCCTGGAGACTGAA-3′ and 5′-CAGGCAAACAAACTTGCTCA-3′), Myc (5′-AGGGATCCTGAGTCGCAGT-3′ and 5′-CGCTCACTCCCTCTGTCTCT-3′), Ccnd1 gene (5′-CACACGGACTACAGGGGAGT-3′ and 5′-CGCGGAGTCTGTAGCTCTCT-3′), plaur (5′-ACTGAGCCGCTCTGAGTGAT-3′ and 5′-CCAGGGGAAAAACAAGTTGA-3′), or a negative-control region ∼5 kb upstream of the Fzd5 promoter designated Fzd5UP (5′-GGTGACTTAGGGCAAAACCA-3′ and 5′-AGGCCACCATACCAGGTTCT-3′). Data from three independent experiment were combined and are presented as n-fold levels of enrichment over the level of expression with the negative-control antibody ±SEM. Statistical differences were detected using a two-tailed Student t test.

RESULTS

Vascular deletion of Chd4 rescues vessel morphology in Brg1^fl/fl:Tie2-Cre⁺ yolk sacs.

Brg1^fl/fl:Tie2-Cre⁺ (here referred to as Brg1^fl/fl:Cre⁺) yolk sac vasculature is thin and disconnected by E10.5 (Fig. 1B and F) (12). These abnormalities result in part from misregulated vascular Wnt signaling, since pharmacological stimulation of the Wnt signaling pathway with LiCl treatment, which stabilizes the intracellular signaling molecule β-catenin, significantly rescues vascular morphology in vivo (12).

Fig 1.

Chd4 deletion rescues Brg1^fl/fl:Cre⁺ yolk sac vascular morphology. (A to H) Anti-PECAM1 staining of E10.5 yolk sacs for blood vessel visualization. Vessels in Brg1^fl/fl:Cre⁺ yolk sacs (B and F) are thin and disconnected (arrowheads) compared to those in control (A and E) and Chd4^fl/fl:Cre⁺ (C and G) yolk sacs. (D and H) Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sacs have a substantial restoration of vessel size and interconnectedness. (I to L) Hematoxylin and eosin (H&E)-stained paraffin sections of E10.5 yolk sac vessels. Brg1^fl/fl:Cre⁺ yolk sac vessel lumens (J) are flat compared to those in control (I) and Chd4^fl/fl:Cre⁺ (K) yolk sacs. (L) Luminal space is rescued in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sac vessels. (I to L) Arrows indicate embryonic blood cells within yolk sac vascular luminal spaces. Scale bars, 100 μm (A to D) and 50 μm (E to L).

In order to assess the role of the NuRD complex in vascular development, we deleted the NuRD chromatin-remodeling enzyme Chd4 from embryonic endothelial cells using the Tie2-Cre transgenic line. In contrast to Brg1^fl/fl:Cre⁺ yolk sacs, Chd4^fl/fl:Cre⁺ yolk sacs displayed normal vascular development and remodeling at E10.5 (Fig. 1C and G). Furthermore, vascular luminal spaces, which were flattened in Brg1^fl/fl:Cre⁺ yolk sacs (Fig. 1J) (12), presumably due to plasma leakage through discontinuities in the endothelial cell lining (12), were normally inflated in Chd4^fl/fl:Cre⁺ yolk sacs (Fig. 1K).

We predicted that deletion of both Brg1 and Chd4 from yolk sac vasculature would result in more severe morphological defects than those seen in Brg1^fl/fl:Cre⁺ yolk sacs if SWI/SNF and NuRD played redundant roles in yolk sac vascular development. To our surprise, the vascular thinning and misconnections observed in Brg1^fl/fl:Cre⁺ yolk sac microvasculature were significantly rescued in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sacs (Fig. 1D and H). Likewise, Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sac vascular luminal spaces were inflated comparably to those of control and Chd4^fl/fl:Cre⁺ vessels (Fig. 1L, I, and K). These data indicate that vascular deletion of Chd4 rescues Brg1^fl/fl:Cre⁺ yolk sac vascular thinning, disconnectedness, and luminal flattening.

Brg1 and Chd4 are efficiently excised in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ endothelial cells.

In order to generate a Brg1^fl/fl; Chd4^fl/fl:Cre⁺ mutant, the Tie2-Cre transgene must effectively delete 4 floxed alleles from embryonic endothelial cells in vivo. Thus, we immunostained E10.5 embryos for BRG1 to confirm that the morphological rescue observed was attributable to genetic interaction between Brg1 and Chd4 and not an artifact due to diminished Cre efficiency and mosaic Brg1 excision. BRG1 was normally expressed in the nuclei of endothelial cells and surrounding tissues of the developing embryo at E10.5 (Fig. 2A and D). In contrast, BRG1 was undetectable in the endothelium of Brg1^fl/fl:Cre⁺ (Fig. 2B and E) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ embryos (Fig. 2C and F). Likewise, CHD4 was detectable in endothelial cells from control embryos (Fig. 2G and J) and was efficiently excised in both Chd4^fl/fl:Cre⁺ (Fig. 2H and K) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (Fig. 2I and L) endothelial cells. These data indicate that the phenotypic rescue documented in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sac vasculature (Fig. 1) results from a genetic interaction between Brg1 and Chd4.

Fig 2.

Tie2-Cre efficiently excises BRG1 and CHD4 in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ endothelial cells. E10.5 embryos were cryosectioned and immunostained with an anti-BRG1 antibody (green) (A to F) or an anti-CHD4 antibody (green) (G to L) and an anti-PECAM antibody (red) to mark endothelial cells. Nuclei were stained with Hoechst dye (blue). Arrows designate individual endothelial cells. (A to F) Brg1^fl/fl:Cre⁺ (B and E) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (C and F) endothelial cells display significantly reduced expression of BRG1 compared to control cells (A and D). Likewise, Chd4^fl/fl:Cre⁺ (H and K) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (I and L) endothelial cells have considerably reduced expression of CHD4 compared to controls (G and J). (A to L) Scale bars, 50 μm.

Deletion of Chd4 from primitive erythrocytes does not rescue Brg1^fl/fl:Cre⁺ anemia.

The Tie2-Cre transgene that we used to excise Brg1 and Chd4 from vascular endothelium is also expressed in a subset of hematopoietic cells during embryonic development (19). Brg1^fl/fl:Cre⁺ embryos die from anemia at midgestation due to excision of Brg1 from primitive erythrocytes (11). BRG1 is required for expression of embryonic globin genes, and Brg1-deficient primitive erythrocytes undergo apoptosis due to insufficient hemoglobin production (11).

Despite the role that CHD4 plays in erythrocyte and megakaryocyte maturation and in T cell development in vivo (10, 36), we saw no evidence of hematopoietic abnormalities in Chd4^fl/fl:Cre⁺ embryos by E10.5. Unlike Brg1^fl/fl:Cre⁺ embryos, Chd4^fl/fl:Cre⁺ embryos were not grossly anemic (Fig. 3B and C). Likewise, primitive Chd4^fl/fl:Cre⁺ erythrocytes did not undergo apoptosis (Fig. 3G) or display deficient hemoglobin accumulation (Fig. 3K).

Fig 3.

Chd4 deletion does not rescue Brg1^fl/fl:Cre⁺ anemia. (A to D) Gross photos of E10.5 control and mutant embryos with attached yolk sacs and placentae. Vessels in control (A) and Chd4^fl/fl:Cre⁺ (C) yolk sacs contain visible red blood. Vessels in Brg1^fl/fl:Cre⁺ (B) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (D) yolk sacs are pale. Arrowheads designate blood vessels. (E to H) Cryosections from E10.5 control and mutant yolk sacs were stained by TUNEL (green) to identify apoptotic cells. Primitive erythrocytes in control (E) and Chd4^fl/fl:Cre⁺ (G) yolk sac vessels are TUNEL negative. A subset of primitive erythrocytes in Brg1^fl/fl:Cre⁺ (F) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (H) yolk sac vessels are TUNEL positive (arrows). Nuclei were stained with Hoechst dye (blue). Yolk sac vessels are outlined (yellow). (I to L) Cryosections from E10.5 control and mutant embryos were stained with benzidine (yellow) for detection of hemoglobin in developing blood cells. The majority of primitive erythrocytes in control (I) and Chd4^fl/fl:Cre⁺ (K) vessels stained with benzidine. In contrast, the majority of primitive erythrocytes in Brg1^fl/fl:Cre⁺ (J) and Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (L) vessels failed to stain with benzidine (arrows). Scale bars, 1 mm (A to D), 50 μm (E to H), and 100 μm (I to L).

Since blood flow biomechanics can contribute to yolk sac vascular remodeling (24), we sought to determine whether the rescued vascular morphology observed in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sacs could result from rescued primitive erythrocyte survival. Gross assessment of Brg1^fl/fl; Chd4^fl/fl:Cre⁺ embryos revealed that they were anemic, like Brg1^fl/fl:Cre⁺ embryos at E10.5 (Fig. 3D and B). Likewise, both Brg1^fl/fl; Chd4^fl/fl:Cre⁺ embryos and Brg1^fl/fl:Cre⁺ embryos had apoptotic primitive erythrocytes (Fig. 3H and F) that failed to accumulate normal levels of hemoglobin (Fig. 3L and J). Therefore, Chd4 deletion did not rescue erythroblast hemoglobin production and apoptosis or pallor in Brg1^fl/fl:Cre⁺ embryos, although it did rescue yolk sac vascular patterning (Fig. 1D). These data indicate that BRG1 and CHD4 have separable tissue-specific roles during development. They also provide strong evidence that the vascular anomalies seen in Brg1^fl/fl:Cre⁺ yolk sacs are caused primarily by genetic rather than biomechanical factors.

Wnt signaling is upregulated in Chd4^fl/fl:Cre⁺ yolk sac endothelial cells.

The extent to which vascular morphology was rescued in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sacs (Fig. 1D) is highly reminiscent of that seen when we treat Brg1^fl/fl:Cre⁺ embryos with LiCl in vivo (12). Because LiCl stimulates Wnt signaling by stabilizing the intracellular signaling molecule β-catenin (25), we questioned whether Chd4 deletion similarly rescued Brg1^fl/fl:Cre⁺ yolk sac vascular morphology by stimulating Wnt signaling. Using a custom-designed qPCR array, we assessed the relative transcript levels of 28 direct Wnt target genes in primary YSECs isolated from E10.5 yolk sacs (see Table S1 in the supplemental material). We found that 64% of these target genes were downregulated in Brg1^fl/fl:Cre⁺ YSECs (Fig. 4A), which is consistent with our previous finding that BRG1 promotes vascular Wnt signaling (12). In contrast, 79% of the 28 Wnt target genes were upregulated in Chd4^fl/fl:Cre⁺ YSECs (Fig. 4B). In addition, several of the genes that were downregulated in Brg1^fl/fl:Cre⁺ YSECs were upregulated in Chd4^fl/fl:Cre⁺ YSECs (Fig. 4C). These data indicate that BRG1 and CHD4 modulate Wnt signaling in opposite directions during yolk sac vascular development.

Fig 4.

CHD4 acts downstream of β-catenin to modulate vascular Wnt signaling. (A to C) Endothelial cells from littermate control and mutant yolk sacs were isolated, RNA was purified, and cDNA was synthesized. qPCR was performed using a custom-designed array containing primer sets for 28 Wnt target genes. Data from at least four independent experiments were combined and analyzed using SABiosciences Excel-based software. Pie charts display the percentage of genes in each group that were downregulated, upregulated, or unchanged in Brg1^fl/fl:Cre⁺ (A) or Chd4^fl/fl:Cre⁺ (B) YSECs compared to littermate control YSECs. (C) A Venn diagram shows the transcripts that were downregulated in Brg1^fl/fl:Cre⁺ YSECs compared to control YSECs and upregulated in Chd4^fl/fl:Cre⁺ YSECs compared to control YSECs. (D to E) C166 yolk sac endothelial cells were transfected with nonspecific (NS) or CHD4-specific siRNA oligonucleotides for 48 h. (D) Western blot analysis was performed using antibodies that recognize CHD4, β-catenin, or GAPDH. β-Catenin band intensity was determined and normalized to the intensity of GAPDH. The results from five independent experiments were combined, and data are presented as the means ± SEM. (E) RNA was isolated, cDNA was synthesized, and qPCR was performed using gene-specific primers (Chd4, β-catenin, Tcf3, Tcf4, Tcf7, and Lef1). Relative fold change was calculated for transcripts in CHD4 knockdown cells and normalized to nonspecific siRNA-transfected samples (dotted line). Error bars represent ±SEM of results from four independent experiments. Significant differences were calculated using a two-tailed Student t test (*, P < 0.05).

In order to determine how CHD4 inhibits Wnt signaling, we analyzed β-catenin protein levels following siRNA-mediated knockdown of CHD4 in the C166 yolk sac endothelial cell line (35). Western blots revealed that β-catenin levels were normal following CHD4 depletion (Fig. 4D). This differs from the significant β-catenin degradation seen in BRG1-depleted endothelial cells, which results from reduced Fzd receptor transcription (12). Since we detected normal β-catenin expression in CHD4-depleted endothelial cells, we suspected that CHD4 impacted the Wnt signaling pathway downstream of β-catenin.

β-Catenin mediates Wnt target gene transcription by interacting with and activating TCF/LEF transcription factors (25). To address the possibility that numerous Wnt target genes were upregulated in Chd4^fl/fl:Cre⁺ YSECs due to misregulated TCF/LEF factors, we assessed the expression of all known mammalian TCF/LEF genes. We detected significant upregulation of Tcf7 and Lef1 upon siRNA-mediated depletion of CHD4 in C166 cells (Fig. 4E). We also performed qPCR on primary YSECs isolated from Chd4^fl/fl:Cre⁺ embryos and confirmed that the transcript level of Tcf7 was significantly upregulated in mutant cells (not shown).

Because transcript levels of Tcf7 and Lef1 were increased with CHD4 depletion, we were interested in determining whether these genes are direct targets of CHD4. Chromatin immunoprecipitation (ChIP) assays indicated that CHD4 associates with the promoter region of the Tcf7 gene, but not the Lef1 promoter, in C166 endothelial cells (Fig. 5A). To determine the functional consequence of CHD4 association with the Tcf7 promoter, we performed ChIP following siRNA-mediated knockdown of CHD4 in the C166 yolk sac endothelial cell line with an antibody that recognizes total, unmarked histone H3. This assay allowed us to measure nucleosome density at the Tcf7 promoter in the presence or absence of CHD4. We found greater H3 enrichment in CHD4 knockdown cells than in cells that had been transfected with nonspecific siRNA (Fig. 5B), indicating that CHD4 mediates chromatin decondensation at the Tcf7 promoter along with transcriptional repression. This is consistent with the hypothesis that NuRD complexes mediate transcriptional repression by first using CHD4 to remodel chromatin in order to make it accessible to NuRD-associated histone deacetylases (34, 39). Our data indicate that CHD4 epigenetically modulates vascular Wnt signaling downstream of β-catenin through chromatin remodeling and subsequent transcriptional repression at the Tcf7 promoter.

Fig 5.

CHD4 modulates vascular Wnt signaling at two levels. (A) Chromatin immunoprecipitation (ChIP) assays were performed using antibodies against CHD4 or a polyhistidine epitope tag as a negative control. DNA was isolated and amplified by qPCR to determine whether CHD4 bound the promoter region of Tcf7 or Lef1. (B) Chromatin harvested from nonspecific (NS) or CHD4 siRNA-transfected C166 yolk sac endothelial cells was immunoprecipitated with an antibody against histone H3 or a negative-control antibody. DNA was isolated and amplified by qPCR to examine the relative nucleosome density at the Tcf7 promoter. (C) Chromatin immunoprecipitation assays were carried out using antibodies against CHD4 or IgG (negative control) to determine whether CHD4 was associated with the promoter region of the Wnt target genes Pitx2, Myc, Ccnd1, and Plaur. (A to C) A region greater than 5 kb upstream of the Fzd5 transcription start site (Fzd5UP) was used as a negative control. Data from three independent experiments were combined and are presented as fold enrichment over the level with the negative-control antibody. Significant differences were calculated using a two-tailed Student t test (*, P < 0.05).

CHD4 associates with the promoter region of a subset of Wnt-responsive target genes.

Since BRG1 impacts the Wnt signaling pathway through transcriptional regulation of multiple Fzd receptors as well as a subset of Wnt target genes in yolk sac endothelial cells (12), we examined the possibility that CHD4 might also be directly involved in Wnt-responsive target gene transcription. Although several Wnt-responsive genes were upregulated in Chd4^fl/fl:Tie2-Cre⁺ yolk sac vasculature (see Table S1 in the supplemental material), we randomly selected four genes, Pitx2, Myc, Ccnd1, and Plaur, to test by ChIP assay in C166 endothelial cells. We found that CHD4 associates with the promoter region of the Myc, cyclin D1, and uPAR genes but not the Pitx2 promoter (Fig. 5C), indicating that CHD4 impacts the Wnt signaling pathway through regulation of the Wnt-responsive transcription factor Tcf7 but also directly coregulates transcription of a subset of Wnt target genes.

Vascular deletion of Chd4 rescues Wnt signaling in Brg1^fl/fl:Cre⁺ yolk sacs.

Since BRG1 and CHD4 influence Wnt target gene transcription in opposing manners in primary YSECs (Fig. 4A and B), we sought to determine how Wnt target genes were affected upon concomitant deletion of both Brg1 and Chd4 from the vasculature. Thus, we analyzed Wnt target gene qPCR arrays with YSECs isolated from Brg1^fl/fl; Chd4^fl/fl:Cre⁺ mutants. We found that 50% of Wnt target gene transcripts were upregulated, 18% were downregulated, and 32% were unchanged in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ YSECs compared to YSECs from littermate controls (Fig. 6A). Furthermore, a subset of genes that were downregulated in Brg1^fl/fl:Cre⁺ YSECs and/or upregulated in Chd4^fl/fl:Cre⁺ YSECs were normalized upon deletion of both Brg1 and Chd4 from the vasculature (Fig. 6C). We propose that this subset of genes may contribute to the vascular morphological rescue observed in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ yolk sacs.

Fig 6.

A subset of Wnt target genes are rescued in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ YSECs and LiCl-treated Brg1^fl/fl:Cre⁺ YSECs. Endothelial cells from littermate control and mutant yolk sacs were isolated, RNA was purified, and cDNA was synthesized. (A to D) qPCR was performed using a custom-designed array containing primer sets for 28 Wnt target genes. Data from at least four independent experiments were combined and analyzed using SABiosciences Excel-based software. (A to B) Pie charts display the percentage of genes in each group that were downregulated, upregulated, or unchanged in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ (A) or LiCl-treated Brg1^fl/fl:Cre⁺ (B) YSECs compared to littermate control YSECs. (C to D) Venn diagrams summarize the transcripts that were downregulated in Brg1^fl/fl:Cre⁺ YSECs compared to control YSECs, upregulated in Chd4^fl/fl:Cre⁺ YSECs compared to control YSECs, and unchanged (i.e., rescued) in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ YSECs compared to control YSECs (C) or downregulated in Brg1^fl/fl:Cre⁺ YSECs compared to control YSECs and unchanged (i.e., rescued) in LiCl-treated Brg1^fl/fl:Cre⁺ YSECs compared to LiCl-treated control YSECs (D). (E) qPCR was performed using gene-specific primers for Pitx2. Relative fold change was calculated and normalized to littermate control samples (dotted line). Error bars represent ±SEM of results from six independent experiments. Significant differences between littermate control and mutant samples were calculated using a two-tailed Student t test (*, P < 0.05).

Pharmacological stabilization of β-catenin rescues transcription of select Wnt target genes.

We previously showed that pharmacological stimulation of Wnt signaling with LiCl treatment rescues select Wnt target genes in Brg1^fl/fl:Cre⁺ YSECs (12). In order to expand upon these results, we examined 28 Wnt target genes in YSECs isolated from LiCl-treated Brg1^fl/fl:Cre⁺ yolk sacs and nontreated Brg1^fl/fl:Cre⁺ yolk sacs using our customized qPCR array (compare Fig. 6B and 4A). We found fewer genes downregulated in Brg1^fl/fl:Cre⁺ YSECs following LiCl treatment (39%) than following no treatment (64%), implying that LiCl treatment rescued the transcription of specific Wnt target genes. Since a subset of Wnt target genes require BRG1 for coregulation of their transcription during yolk sac vascular development (12), we hypothesize the genes that were not rescued in LiCl-treated Brg1^fl/fl:Cre⁺ YSECs are direct targets of BRG1. In support of this hypothesis, BRG1 directly associates with the promoter of Wisp1 (12), a Wnt target gene that was downregulated in Brg1^fl/fl:Cre⁺ YSECs and was not rescued by LiCl treatment (Fig. 6D). Furthermore, we propose that genes that were rescued by LiCl treatment represent a subset of genes that may be involved in the morphological rescue of the Brg1^fl/fl:Cre⁺ yolk sac vasculature observed upon treatment of Brg1^fl/fl:Cre⁺ embryos with LiCl (12).

Identification of genes coordinately regulated by BRG1, CHD4, and Wnt signaling.

Because of the comparable rescues of vascular morphology in Brg1^fl/fl; Chd4^fl/fl:Cre⁺ and LiCl-treated Brg1^fl/fl:Cre⁺ yolk sacs, we hypothesized that similar gene targets were rescued under both conditions. Using our custom Wnt target gene arrays, we found a small subset of genes that were downregulated in Brg1^fl/fl:Cre⁺ YSECs but were normalized in either Brg1^fl/fl; Chd4^fl/fl:Cre⁺ or LiCl-treated Brg1^fl/fl:Cre⁺ YSECs (Fig. 6C and D). Only one of these genes, Pitx2, was downregulated in Brg1^fl/fl:Cre⁺ YSECs and unchanged in both Brg1^fl/fl; Chd4^fl/fl:Cre⁺ and LiCl-treated Brg1^fl/fl:Cre⁺ YSECs. We verified these array data by direct qPCR in YSECs and confirmed that Pitx2 was downregulated in Brg1^fl/fl:Cre⁺ YSECs, upregulated in Chd4^fl/fl:Cre⁺ YSECs, and normalized in both Brg1^fl/fl; Chd4^fl/fl:Cre⁺ and LiCl-treated Brg1^fl/fl:Cre⁺ YSECs (Fig. 6E). Our results indicate that BRG1 and CHD4 act in opposing manners to regulate Pitx2 in the yolk sac vasculature during development and furthermore suggest that Pitx2 may contribute to the rescue phenotype observed in both Brg1^fl/fl; Chd4^fl/fl:Cre⁺ and LiCl-treated Brg1^fl/fl:Cre⁺ yolk sac vasculatures.

DISCUSSION

In order to understand how different ATP-dependent chromatin-remodeling complexes influence vascular development, we generated embryos lacking the SWI/SNF catalytic subunit BRG1 or the NuRD catalytic subunit CHD4 in vascular endothelium. We demonstrated that these two chromatin-remodeling enzymes work in opposition to regulate vascular Wnt signaling. While BRG1 primarily promoted Wnt target gene transcription in yolk sac endothelial cells, CHD4 repressed expression of most Wnt target genes that we tested. Furthermore, vascular deletion of Chd4 rescued Brg1 mutant yolk sac phenotypes and restored a subset of Wnt target genes to basal transcription levels. These findings indicate that BRG1 and CHD4 antagonistically modulate Wnt signaling in developing yolk sac vessels.

BRG1 impacts Wnt signaling at two different levels in vascular endothelium: (i) through transcriptional activation of multiple Wnt receptor genes within the Fzd family and (ii) through coregulation of a subset of Wnt/β-catenin target genes (Fig. 7) (12). Therefore, BRG1 interfaces with the Wnt signaling pathway both upstream and downstream of β-catenin to promote Wnt target gene transcription. Our new data support a model in which CHD4 modulates Wnt signaling downstream of β-catenin (Fig. 7). Genetic evidence that CHD4 does not impact Wnt signaling upstream of β-catenin comes from our observation that Chd4^fl/fl:Cre⁺ yolk sacs do not phenocopy β-catenin gain-of-function mutants (4), which have dramatic yolk sac vascular patterning defects presumably due to upregulation of all vascular Wnt target genes. Like BRG1, CHD4 impacts Wnt signaling at two different levels in vascular endothelium: through transcriptional repression of the Wnt-responsive transcription factor Tcf7 and also through direct coregulation and transcriptional repression of a subset of Wnt/β-catenin target genes (Fig. 7). Thus, upregulated Wnt target genes in Chd4^fl/fl; Cre⁺ endothelial cells may be a primary or secondary consequence of CHD4 depletion. More extensive ChIP experiments will be required to distinguish the genes that are direct targets of CHD4 and those that are secondarily affected by misregulation of Tcf7.

Fig 7.

Model for how BRG1 and CHD4 antagonistically influence Wnt signaling during yolk sac vascular development. BRG1 activates the Wnt signaling pathway in two ways: through transcriptional regulation of multiple Fzd receptor genes upstream of β-catenin and through coregulation of Wnt target genes downstream of β-catenin (12). CHD4 represses the Wnt signaling pathway downstream of β-catenin through transcriptional regulation of the Wnt-responsive transcription factor Tcf7 and select Wnt target genes.

We are particularly encouraged that our Wnt target gene arrays provided a novel candidate for mediating yolk sac vascular development. The Wnt target gene Pitx2 was downregulated in Brg1^fl/fl:Cre⁺ yolk sac endothelium but normalized in both Brg1^fl/fl; Chd4^fl/fl:Cre⁺ and LiCl-treated Brg1^fl/fl:Cre⁺ endothelium. Thus, we predict that Pitx2 may contribute to the rescue of vascular thinning and disconnectedness observed in each of these mutants. Although Pitx2 is involved in several developmental processes, including heart development, it has not yet been directly implicated in vascular development (6, 7, 17, 21). Nevertheless, PITX2 has been shown to repress Bmp4 and enhance Fgf8 in the regulation of cell motility during craniofacial development (22, 23). Since roles for BMP and FGF signaling in vascular development have been previously described (3, 26), we speculate that misregulation of these pathways may contribute to the aberrant vascular development observed in Brg1^fl/fl:Cre⁺ mutant yolk sacs. Therefore, further studies are warranted to elucidate the unanticipated role that PITX2 may play in yolk sac vascular remodeling.

Our finding that BRG1 and CHD4 have opposing effects on vascular Wnt signaling in vivo complements previous in vitro evidence that these enzymes antagonistically regulate target genes involved in B cell specification and inflammatory responses (8, 30). Notably, previous studies demonstrate that BRG1 and CHD4 can bind and act antagonistically on the same gene promoters in plasmacytoma and macrophage cell lines. We now report that BRG1 and CHD4 can act on separate targets within the same pathway to regulate Wnt signaling in endothelial cells. While BRG1 modulates transcription of Fzd receptors upstream of β-catenin, CHD4 inhibits transcription of Tcf7, a transcription factor that acts downstream of β-catenin to regulate Wnt target gene transcription. Furthermore, our current data demonstrate that both BRG1 and CHD4 directly modulate Wnt target genes. However, further ChIP experiments will be necessary to determine whether BRG1 and CHD4 bind simultaneously to a subset of Wnt target genes.

We predict that there is more coordination between chromatin-remodeling complexes than has been previously appreciated. Tandem deletion of Brg1 and Chd4 in additional cell types will help to elucidate more developmental processes that utilize SWI/SNF and NuRD to regulate and titrate transcription. In addition, deletion of Brg1 and Chd4 from blood vessels at later embryonic and postnatal time points will reveal whether these chromatin-remodeling enzymes antagonistically modulate Wnt signaling under a variety of angiogenic conditions.