All the animal experiments were reviewed and approved by the intramural Ethics Committee on Humane Treatment of Experimental Animals. Endothelial-specific deletion of BRG1 was achieved by crossing the Smarca4^f/f strain (Li et al., 2018a, b) with the Cdh5-Cre strain (Li et al., 2018e). Male, 8-week old mice were induced to develop cardiac fibrosis by Angiotensin II (1 μg/kg/min) infusion for two consecutive weeks using subcutaneously implanted minipumps (Alzet 2002). Cardiac hypertrophy was monitored by echocardiography (GE Vivid 7 equipped with a 14-MHz phase array linear transducer, S12, allowing a 150 maximal sweep rate).

Immortalized human endothelial cells (EAhy926, ATCC), mouse macrophage-like cells (RAW264.7, ATCC), and HEK293 cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS, Hyclone). Angiotensin II was purchased from Sigma. MRP8 promoter-luciferase constructs (Grebhardt et al., 2012) and BRG1 expression constructs (Li et al., 2019a) have been previously described. Small interfering RNAs were purchased from Dharmacon. PFI-3 was purchased from Selleck. Transient transfections were performed with Lipofectamine 2000. Luciferase activities were assayed 24–48 h after transfection using a luciferase reporter assay system (Promega) as previously described (Li et al., 2019c). For conditioned media (CM) collection, the cells were switched to and incubated with serum-free media overnight. The next day, the media were collected, centrifuged at 4000 × g for 30 min at 4°C using 3-kDa MW cut-off filter units (Millipore) and sterilized through a 0.4-μm filter.

Whole cell lysates were obtained by re-suspending cell pellets in RIPA buffer (50 mM Tris pH7.4, 150 mM NaCl, 1% Triton X-100) with freshly added protease inhibitor (Roche) as previously described (Li et al., 2018c, d; Liu et al., 2018). Western blot analyses were performed with anti-BRG1 (Santa Cruz, sc-10768), anti-MRP8 (Proteintech, 15792-1-AP), anti-HIF-1α (Santa Cruz, sc-10790), anti-p300 (Santa Cruz, sc-585), anti-KDM3A (Proteintech, 15792-1-AP), and anti-β-actin (Sigma, A2228) antibodies. For densitometrical quantification, densities of target proteins were normalized to those of β-actin. Data are expressed as relative protein levels compared to the control group which is arbitrarily set as 1.

RNA was extracted with the RNeasy RNA isolation kit (Qiagen). For cardiac tissue homogenization, the heart was dissected and cut into small pieces (∼20 mg) using a sterilized blade. The appropriately sized cardiac tissue was placed into a microcentrifuge tube along with stainless steel beads (1.6 mm) and 350 μl lysis buffer. The tubes were placed in the Bullet Blender^TM (Scientific Instrument Services) for homogenization. The homogenized tissue lysates were used for RNA extraction per manual instruction. Reverse transcriptase reactions were performed using a SuperScript First-strand Synthesis System (Invitrogen) as previously described (Li et al., 2018a, b; Liu et al., 2018). Real-time PCR reactions were performed on an ABI Prism 7500 system with the following primers: MRP8, 5′-AATTTCCA TGCCGTCTACAG-3′ and 5′-CGCCCATCTTTATCACCAG-3′; BRG1, 5′-TCATGTTGGCGAGCTATTTCC-3′ and 5′-GGTTCC GAAGTCTCAACGATG-3′; HIF1A, 5′-TTCACCTGAGCCTAA TAGTCC-3′ and 5′-CAAGTCTAAATCTGTGTCCTG-3′; p300, 5′-GCGGCCTAAACTCTCATCT-3′ and 5′-TCTGGTAAGTCG TGCTCCAA-3′; KDM3A, 5′-TTCTTTTCCTCCAAGATTC CC-3′ and 5′-GGGACCATTCGAGCTGTTT-3′. Ct values of target genes were normalized to the Ct values of housekeekping control gene (18 s, 5′-CGCGGTTCTATTTTGTTGGT-3′ and 5′-TCGTCTTCGAAACTCCGACT-3′ for both human and mouse genes) using the ΔΔCt method and expressed as relative mRNA expression levels compared to the control group which is arbitrarily set as 1.

Chromatin immunoprecipitation (ChIP) assays were performed essentially as described before (Li et al., 2019b). In brief, chromatin in control and treated cells were cross-linked with 1% formaldehyde. Cells were incubated in lysis buffer (150 mM NaCl, 25 mM Tris pH 7.5, 1% Triton X-100, 0.1% SDS, 0.5% deoxycholate) supplemented with protease inhibitor tablet and PMSF. DNA was fragmented into ∼200 bp pieces using a Branson 250 sonicator. Aliquots of lysates containing 200 μg of protein were used for each immunoprecipitation reaction with anti-BRG1 (Santa Cruz, sc-10768), anti-anti-acetyl H3 (Millipore, 06-599), anti-dimethyl H3K9 (Millipore, 17-648), anti-HIF-1α (Santa Cruz, sc-10790), anti-p300 (Santa Cruz, sc-585), anti-KDM3A (Abcam, ab-91252), or pre-immune IgG. For re-ChIP, immune complexes were eluted with the elution buffer (1% SDS, 100 mM NaCO₃), diluted with the re-ChIP buffer (1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris pH 8.1), and subject to immunoprecipitation with a second antibody of interest.

Macrophage migration was measured using the Boyden chamber inserts (5 μm, Corning) as previously described (Zhang et al., 2018b). Briefly, RAW264.7 cells were added to the upper chamber whereas the conditioned media collected from endothelial cells were added to the lower chamber. The number of migrated macrophages in the lower chamber was counted in five randomly chosen fields using an inverted microscope. In certain experiments, recombinant human MRP8 (20 ng/ml, R&D) was directly added to the conditioned media. Migrated macrophages were counted in at least five different fields for each well. All experiments were performed in triplicates and repeated three times.

For immunofluorescence staining, paraffin sections (5 μm) of aortic arteries were permeabilized with 0.1% Triton X-100 in PBS for 10 min and then blocked with 5% BSA for 20 min at room temperature followed by incubating with anti-CD31 (Abcam, ab28364) and anti-MRP8 (Abcam, ab17050) overnight at 4°C. The nuclei were counterstained with DAPI (Sigma).

Data are presented as mean ± SD. Data normality was examined by the Shapiro–Wilk test. For normally distributed data, comparisons between two groups were performed using Student’s t-test, and comparisons among multiple groups were performed using ANOVA followed by post hoc Bonferroni correction. For non-normal data comparisons were performed by the Mann–Whitney U test or the Kruskal–Wallis test followed by Dunn’s multiple comparison test. Data homoscedasticity was examined by the Bartlett’s and Brown–Forsythe test. Non-homoscedastic data was analyzed by the Brown-Forsythe ANOVA test followed by Dunnett’s T3 multiple comparisons test. A p-value < 0.05 was considered significant.

We first compared the phenotype of the mice with a deficiency of BRG1 in endothelial cells (Smarca4^f/f; Cdh5-Cre, referred to as ecKO) and the control mice (Smarca4^f/f, referred to as WT) in response to chronic Ang II infusion. Ang II infusion induced a robust pro-hypertrophic response in the heart as evidenced by the elevation of heart weight/body weight ratio (Figure 1A), heart weight/tibia bone length ratio (Figure 1B), left ventricular systolic dimension (Figure 1C), and left ventricular posterior wall dimension (Figure 1D). For each one of these measurements, it was observed that endothelial specific BRG1 deletion attenuated the pro-hypertrophic response. Quantitative PCR analysis of message levels of pro-hypertrophic genes, including Nppa, Nppb, and Myh6, confirmed that cardiac hypertrophy was ameliorated in the ecKO mice compared to the WT mice (Figure 1E). In addition, WGA staining also showed that cross-sectional areas of cardiomyocytes were smaller in the ecKO hearts than the WT hearts indicative of a reduced pro-hypertrophic response (Figure 1F).

Because macrophage-mediated cardiac inflammation plays a key role in the pathogenesis of pathological hypertrophy, we evaluated macrophage infiltration by immunofluorescence staining with an anti-F4/80 antibody or an anti-CD45 antibody. As shown in Figure 2A, macrophage infiltration was less prominent in the ecKO hearts than in the WT hearts. Consistent with this observation, several pro-inflammatory cytokines associated with macrophage function including interleukin 1 (Il-1b), interleukin 6 (Il-6), tumor necrosis factor (Tnfa), and macrophage chemoattractant protein 1 (Mcp1) were detected to be present at lower levels in the ecKO hearts than in the WT hearts (Figure 2B). Because inflammation is normally coupled with fibrosis, we examined the expression levels of several pro-fibrogenic marker genes. Indeed, expression levels of Col1a1, Col3a1, and Acta2 were collectively down-regulated in the ecKO hearts compared to the WT hearts (Supplementary Figure S1).

MRP8 belongs to the well-documented S100A family of calcium-binding chemokines functioning as a damage associated molecular pattern (DAMP) released under stress conditions to promote macrophage homing and tissue inflammation (Wang S. et al., 2018). Compared to the other chemokines/cytokines, expression of MRP8 was more robustly induced by Ang II infusion and more sensitive to BRG1 deficiency (Figure 2B). We hypothesized that decreased macrophage infiltration in the ecKO hearts could be ascribed to down-regulation of MRP8 expression in the endothelial cells. Double immunofluorescence staining showed that MRP8 expression in the vessels (aortic arteries) was largely restricted to the endothelial layer and that MRP8 expression in endothelial cells, as indicated by CD31⁺MRP8⁺ cells, was significantly lower in the ecKO mice than in the WT mice (Figure 2C).

We next tested the hypothesis that BRG1 mediates Ang II induced MRP8 expression to promote macrophage migration in cultured vascular endothelial cells. Ang II treatment led to robust induction of MRP8 expression in human endothelial cells (EAhy926) at both mRNA (Figure 3A) and protein (Figure 3B) levels; BRG1 knockdown significantly attenuated MRP8 induction by Ang II. In accordance, conditioned media (CM) collected from Ang II-treated endothelial cells displayed much stronger chemoattractive potency than those collected from the mock treated endothelial cells whereas BRG1 depletion dampened the chemoattractive ability of the Ang II treated CM (Figure 3C). Of note, the addition of recombinant human MRP8 to the CM collected from the BRG1 deficient cells restored macrophage migration suggesting that MRP8 may constitute a major Ang II inducible and BRG1-dependent chemotactic substance in the endothelial cells. Similarly, inhibition of BRG1 with a small-molecule compound PFI-3 (Wu et al., 2016) dose-dependently suppressed the induction of MRP8 expression by Ang II in endothelial cells (Figures 3D,E). BRG1 inhibition also crippled the ability of the CM to promote macrophage migration, which could be normalized by the addition of recombinant MRP8 (Figure 3F).

We next asked whether BRG1 regulated MRP8 expression at the transcriptional level and, if so, by what mechanism. To this end, a series of MRP8 promoter-luciferase constructs with progressively inward deletions were transfected into endothelial cells. Ang II treatment activated the longest MRP8 promoter construct (−1,000 relative to the transcription start site); BRG1 over-expression further enhanced the MRP8 promoter activity (Figure 4A). The responsiveness to Ang II treatment and BRG1 over-expression was retained when the deletion extended to −500 relative to the transcription start site. However, when the deletion went beyond −100, neither Ang II treatment nor BRG1 over-expression was able to activate the MRP promoter activity (Figure 4A). A previous study has identified a DNA binding element for HIF-1α (HRE) between −500 and −100 within the human MRP8 promoter. Mutation of this region abrogated the activation of the MRP8 promoter by Ang II treatment and BRG1 over-expression (Figure 4B), raising the possibility that HIF-1α might be play an essential role in recruiting BRG1. To address this issue, we performed the following experiments. Ang II treatment up-regulated HIF-1α protein, but not mRNA, levels in endothelial cells (Figure 4C). Under normal conditions, HIF-1α protein is modified by prolyl hydroxylases (PHDs) and the ubiquitin E3 ligase VHL before being targeted for degradation. Ang II treatment decreased expression levels of PHD3, but not PHD1 or PHD2 or VHL, which may account for the stabilization of HIF-1α protein (Supplementary Figure S2). More important, Ang II treatment markedly increased the occupancies of the MRP8 promoter by both HIF-1α and BRG1 with similar kinetics (Figure 4D). Ang II-dependent interaction between HIF-1α and BRG1 was further evidenced by Re-ChIP assay (Figure 4E). HIF-1α depletion by small interfering RNA (Figure 4F) or inhibition by two different HIF inhibitors (Figure 4G) equally compromised recruitment of BRG1 to the MRP8 promoter. Of interest, BRG1 deficiency reciprocally influenced the affinity of HIF-1α for the MRP8 promoter as shown by ChIP assay (Figure 4H). Collectively, these data suggest that an interplay between BRG1 and HIF-1α may underscore Ang II induced trans-activation of MRP8 in endothelial cells.

The recruitment of acetyltransferase p300 represents a paradigm of HIF-1α mediated transcription. On the other hand, BRG1 is known to interact with and engage various histone modifying enzymes to regulate transcription. ChIP assays revealed that the MRP8 promoter was characterized by low levels of histone H3 acetylation (Figure 5A) and high levels of dimethyl H3K9 (Figure 5B) indicative of a transcriptionally silenced state. Upon Ang II treatment, H3 acetylation was up-regulated on the MRP8 promoter with a similar kinetics as the recruitment of HIF-1α and BRG1 (Figure 5A). The removal of dimethyl H3K9 from the MRP8 promoter, however, lagged behind the augmentation of H3 acetylation (Figure 5B). Consistent with these observations, it was further discovered that p300 recruitment to the MRP8 promoter synchronized with the binding of HIF-1α/BRG1 whereas KDM3A, a H3K9 demethylase, was brought to the MRP8 promoter with a slightly lagged kinetics (Figures 5C,D).

We finally examined whether endothelial depletion in either p300 or KDM3A would exert equivalent effects on MRP8 expression and macrophage migration. As shown in Figures 6A,B, siRNA-mediated knockdown of p300 and KDM3A comparably down-regulate MRP8 induction by Ang II in endothelial cells. Consequently, conditioned media collected from endothelial cells depleted of either p300 or KDM3A were compromised in the ability to promote macrophage migration (Figure 6C). Combined, these data suggest that sequential recruitment of histone modifying enzymes by BRG1 contributes to MRP8 trans-activation.