Cbx7a/CBX7 is upregulated in different CCM disease models and lesion material of familial CCM patients.

(A) Volcano plot of microarray data of differentially-expressed genes (calculated with the limma R package (Ritchie et al, 2015) with moderated T-statistic, adjusted p value <0.05 and fold change >1.4) in zebrafish ccm2^m201 mutants (Renz et al, 2015). (B) Shown are expression fold-changes of Cbx7a/CBX7 mRNA in different CCM disease models. Elevated mRNA expression levels of zebrafish cbx7a in krit1^ty219c and ccm2^m201 mutants (RNA-sequencing of extracted hearts at 48 hours post fertilisation (hpf)), murine Cbx7 in Ccm1/Krit1^ECKO-derived brain microvascular endothelial cells (BMECs) (Koskimäki et al, 2019), and CBX7 in CCM1 and CCM2-deficient human iPSC-derived ECs (ECs) (RNA-sequencing). (C–E) Whole-mount in situ hybridization of zebrafish embryo at 56 hpf revealed low levels of cbx7a mRNA in wild-type (wt) embryos (C) and elevated levels in ccm2^m201 (D) and krit1^ty219c (E) mutants throughout the entire heart. The constriction of the atrioventricular canal (AVC) was lost in ccm2^m201 (D) and krit1^ty219c (E) mutants. (F) Quantifications of cbx7a mRNA levels by qRT-PCR from isolated tissue preparations of head, heart, or tail regions in ccm2^m201 and krit1^ty219c mutants as compared to wild-type (wt) (n = 3–4 biological replicates per group, each replicate contained 10–15 pooled tissue samples). Statistical testing is based on the student’s T-test (p values and s.e.m. bars indicated). (G, K) Immunohistochemistry stainings for CBX7 expression. CD34 marks ECs. Only low expression of CBX7 is detected in ECs of healthy brain (K, arrows). In lesions of CCM1 (G, H, J) and CCM2 (I) familial patients, CBX7 staining is markedly increased in ECs (arrows). Asterisks indicate vessel lumen. Ven ventricle, Atr atrium, OFT out flow tract. Experiments in zebrafish were done in three biological replicates. Scale bars are (C–E) 100 µm; (G–K) 200 µm. Source data are available online for this figure.

Cbx7a/CBX7 suppresses CCM-associated phenotypes in zebrafish ccm2^m201 mutants and is controlled by Klf2a and blood flow.

(A) Schematic representation of the cbx7a genomic locus in zebrafish. The cbx7a^pbb62 allele comprises a 510 bp deletion 80 bp upstream of the starting ATG (green bar). Areas in dark gray indicate the coding region and light gray areas highlight 5′ and 3′ untranslated regions (UTR). (B) Quantification of cbx7a mRNA levels show that its expression is completely depleted in cbx7a^pbb62 mutants (p = 0.0289) and in ccm2^m201;cbx7a^pbb62 double mutants (p = 0.0248). In ccm2^m201 mutants, cbx7a mRNA levels are upregulated (p = 0.0102). Four to six replicates per group, pooled 8–10 embryos per each replicate. Statistical testing is based on one-way ANOVA with Tukey’s multiple comparisons test (s.e.m. bars indicated). (C–E) Endocardium at 56 hpf, marked by Tg(kdrl:EGFP)^s843 expression in wild-type (wt) (C), ccm2^m201 (D), and ccm2^m201;cbx7a^pbb62 double mutants (E). The endocardium forms the AVC (red arrowhead) between the ventricle and the atrium at 56 hpf (C). Endocardial cells of the AVC are marked by the expression of the cell junctional protein ALCAM (C’). The asterisk indicates the location of cardiac jelly between the myocardium and endocardium (C’). In ccm2^m201 mutants, there is no AVC constriction (D) and endocardial cells lack ALCAM expression, and the cardiac jelly is markedly reduced (D’). The endocardial defects in zebrafish ccm2^m201 mutants were suppressed when Cbx7a is genetically depleted (E). ccm2^m201;cbx7a^pbb62 double mutants form an AVC constriction (E), and ALCAM expression (E’) and cardiac jelly (E’, asterisk) are restored. (F) Quantifications demonstrate that ccm2^m201 mutants have higher endocardial cell numbers (p < 0.0001) and there is a normalization in ccm2^m201;cbx7a^pbb62 double mutants (p = 0.4099 compared to wild-type; p = 0.0018 compared to ccm2^m201). Each data point represents endocardial cell numbers counted from a single heart. Statistical testing is based on the student’s T-test (s.e.m. bars indicated). (G) Expression levels of klf2a and klf2b mRNA are elevated in ccm2^m201 and ccm2^m201;cbx7a^pbb62 double mutants. Shown are quantifications of whole embryos by qRT-PCR. mRNA levels of klf2a and klf2b are increased in ccm2^m201 mutants as compared to wild-type (p = 0.0072 and p = 0.027, respectively) and do not normalize in ccm2^m201;cbx7a^pbb62 double mutants (p = 0.1018 and p = 0.99 for klf2a and klf2b, respectively; n = 4–5 replicates per each group; 8–10 embryos were collected for each replicate). Statistical testing is based on student’s T-test between mutants and corresponding wild-type siblings (s.e.m. bars indicated). (H–K) Representative images of monolayers of HUVECs silenced for control (siCT) (H, H’), CCM2 (I, I’), co-silencing of CCM2 and CBX7 (J, J’), and co-silencing of CCM2 and CBX2 (K, K’). Phenotypes of CCM2-depleted HUVECs show an elongated morphology with VE-cadherin at cell-cell junctions and increased stress fiber formation (I, I’) as compared to controls (H, H’). Restoration of cell morphology with cortical actin and reduced stress fibers is observed upon knock-down of CBX7 (J, J’), but not CBX2 (K, K’). (L) Quantification of the normalized impedance of HUVEC monolayers 24 h after serum starvation. Knockdown of CCM2 in HUVECs causes a reduction of barrier function, which is restored upon co-depletion of CBX7, but not with co-depletion of CBX2. There are three replicates per each group and statistical testing is based on one-way ANOVA with Tukey’s multiple comparisons test (s.e.m bars indicated). (M–R) Representative images of cbx7a expression detected by whole-mount in situ hybridization in the zebrafish heart at 56 hpf. (M) cbx7a expression is low in wild-type and (N) expanded throughout the entire endocardium in ccm2^m201 mutants. (O) Higher levels of cbx7a mRNA in ccm2^m201 mutants are reduced in ccm2^m201; klf2a^sh317; klf2b^pbb42 triple mutants with normal heart morphology (n = 8/10 triple mutants with reduced cbx7a mRNA expression) and elevated in triple mutants in which the cardiac morphology has not been restored (P). In Tg(fli1a:Gal4FF)^ubs3;Tg(UAS:klf2a)^ig1 double transgenic embryos [Tg(fli1a>klf2a)], in which blood flow is present, mRNA levels of cbx7a remain low (Q). Upon endothelial-specific overexpression of Klf2a and concomitant loss of blood flow, cbx7a mRNA levels are elevated (R). (S) Quantification of cbx7a mRNA levels by qRT-PCR in Tg(fli1a>klf2a) embryos that also lack blood flow (tnnt2a MO against sarcomeric protein Tnnt2a) or with normal cardiac function (standard morpholino, std MO). There are five replicates per each group and each replicate is based on 8–10 pooled embryos. Statistical testing is based on student’s T-test (s.e.m. bars indicated). AVC atrioventricular canal, endo endocardium, myo myocardium, OFT outflow tract, Ven ventricle, Atr atrium. Red arrowheads indicate the location of the AVC and white arrows indicate the presence of ALCAM expression in AVC endocardial cells. Experiments in zebrafish were done in three biological replicates. Scale bars are (M–R) 100 µm, (C–E, H–K, H’–K’) 50 µm, (C’–E’) 20 µm. Source data are available online for this figure.

Cbx7a/CBX7 contributes to pathological gene expression in CCM, impacting cardiovascular defects in zebrafish mutants.

(A) Functional clustering of gene ontology terms for deregulated genes identified in RNA-sequencing datasets from ccm2^m201 in comparison with wild-type (wt) and ccm2^m201;cbx7a^pbb62 double mutants in comparison with ccm2^m201 mutants. Differentially-expressed genes were identified by using DEseq2 (Galaxy Version 2.11.40.7) and a Wald statistic for pairwise comparisons. (B) Schematic overview of candidate genes related to CCM pathology gene ontology terms. Shown are fold-changes of expression levels in wild-type, ccm2^m201, and ccm2^m201;cbx7a^pbb62 double mutants. (C) Quantification of wnt9b and wnt9a mRNA levels in ccm2^m201 and krit1^ty219c extracted hearts. wnt9b is upregulated in both mutants while wnt9a is upregulated only in krit1^ty219c mutants (n = 3–4 replicates per each group, and each replicate is a pool of 50–100 hearts). Statistical testing is based on one-way ANOVA with Tukey’s multiple comparisons testing (s.e.m. bars indicated). (D, E) Representative images of whole-mount in situ hybridizations of wnt9b expression in the 56 hpf zebrafish heart. In wild-type, wnt9b is expressed in a defined domain confined to the atrioventricular canal (AVC) (D). In Tg(fli1a:Gal4FF)^ubs3;Tg(UAS:klf2a)^ig1 double transgenic embryos [Tg(fli1a>klf2a)] with an endothelial-specific overexpression of Klf2a, the expression of wnt9b is expanded into the cardiac chambers (E). (F–I) Functional rescue of endocardial phenotypes through genetic depletion of Wnt9b in zebrafish ccm2^m201 mutant embryos. Shown are maximum projections of confocal image z-stacks of zebrafish hearts at 56 hpf, with the endocardium being marked by Tg(kdrl:EGFP)^s843 expression. (F) The AVC region forms correctly in wild-type hearts (red arrowhead). The expression of ALCAM in endocardial AVC cells (F’; white arrows) is lost in ccm2^m201mutants (G’). Upon depletion of wnt9b using an antisense oligo morpholino, the overall morphology of ccm2^m201 mutant hearts is improved, the AVC region restored (H; red arrowhead), and ALCAM expression at the endocardial AVC is restored (H’; white arrows). Knock-down of wnt9b alone does not affect AVC formation (I; red arrowhead) and ALCAM expression (I’; white arrows). (J–M) Representative fluorescence microscopy images of CCM2-deficient (CCM2-KO) and wild-type (WT) iPSC-derived ECs immunostained for F-Actin (J, K), and VE-Cadherin (L, M). (N) Schematic overview of WNT9A and WNT9B mRNA levels in primary human lesion material from familial and sporadic CCM patients as compared to healthy brain material and in human iPCS-derived endothelial cells depleted of CCM2 as compared to non-edited cells. In all cases, WNT9A and WNT9B are upregulated. (O–Q) Quantifications based on qRT-PCR of angpt1 and tek/tie2 mRNA levels in ccm2^m201 and ccm2^m201;cbx7a^pbb62 double mutants (O) and in cbx7a antisense morpholino-injected Tg(fli1a>klf2a) (P, Q) (n = 4–5 replicates and each replicate contains 8–10 pooled embryos). Statistical testing is based on student T-test (s.e.m. bars indicated). (R, X) Shown are maximum projections of confocal image z-stacks of zebrafish hearts at 48 hpf in which the endocardium is marked by Tg(kdrl:EGFP)^s843 expression. Wild-type embryos treated with DMSO form a normal AVC (R; red arrowhead) and express ALCAM in endocardial cells at the AVC (R’; white arrows). Endocardial defects in ccm2^m201 mutants include the loss of the AVC constriction (S) and ALCAM expression (S’). (T) ccm2^m201 mutant embryos treated with the Tie1/2 signaling inhibitor BAY826 have a normalized cardiac morphology, show a restoration of the AVC region (T; red arrowhead) and endocardial ALCAM expression (T’; white arrows). (U) Treatment of wild-type embryos with BAY826 has only a mild effect on cardiac morphology as AVC (U; red arrowhead) and expression of ALCAM in endocardial AVC cells (U’; white arrows) are present. (V–X) The genetic depletion of zebrafish tek and tie1 genes suppresses the cardiac ballooning phenotype in krit1 morphants. At 56 hpf, wild-type embryos injected with an antisense morpholino against krit1 lose that AVC constriction and exhibit strongly ballooned hearts, which is a characteristic feature of the cardiac phenotype in zebrafish ccm mutants (W). The genetic depletion of tek/tie2 and tie1 in krit1 morphants restores the AVC constriction and suppresses cardiac ballooning (X, red arrowhead). (Y) Quantifications of the krit1 morphant cardiac phenotype in different tek^hu1667;tie1^bns208 mutant combinations. The penetrance of krit1 morphant cardiac phenotypes decreases with lowered copy numbers of wild-type tek/tie2 and tie1 alleles and is most strongly suppressed by the combined loss of both Tie1/2 receptors. Numbers in boxes are embryos with a rescued heart / total number of embryos with the respective genotype. AVC atrioventricular canal, endo endocardium, myo myocardium, Ven ventricle, Atr atrium. Red arrowheads indicate AVC, while white arrows indicate the presence of ALCAM expression in AVC endocardial cells. Experiments in zebrafish were done in three biological replicates. Scale bars are (D, E, J–M) 100 µm, (F–I; R–U; V–X) 50 µm, (F’–I’; R’–U’) 20 µm. Source data are available online for this figure.

The pharmacological inhibition of PRC1 protein Cbx7a/CBX7 suppresses CCM-associated cardiac phenotypes in zebrafish ccm2^m201 mutants and lesion burden in a murine pre-clinical CCM3 disease model.

(A) Proteins of the PRC2 are involved in the tri-methylation of histone H3 at Lys27 (H3K27me3). This epigenetic modification recruits CBX7 and other PRC1 proteins to these genomic sites. The RING1A/B ubiquitin ligases of the PRC1 catalyze H2AK119ub modifications, which results in transcriptional downregulation of genes. The two small compound drugs MS351 and MS37452 inhibit the activity of CBX7. (B–D) The preventive pharmacological treatment of zebrafish ccm2^m201 mutants with CBX7 inhibitors suppresses the development of a cardiac ballooning phenotype and the loss of AVC constriction. Shown are maximum projections of confocal image z-stacks of zebrafish hearts at 56 hpf. The endocardium is marked by the expression of Tg(kdrl:EGFP)^s843. The ccm2^m201 mutant heart is ballooned (B) and lacks expression of ALCAM in atrioventricular canal (AVC) endocardial cells. Treatment with the CBX7 inhibitor MS351 (C, C’) or MS37452 (D,D’) strongly improves cardiac morphology (C, D) and normalizes the endocardial expression of ALCAM in AVC endocardial cells (C’,D’; arrows). (E–J) The inhibition of CBX7 with MS37452 in CCM3^iECKO (n = 4) and control mice (n = 10). (E) Treatment with MS37452 does not produce any adverse effects on the brains of the treated mice. (F) CCM3^iECKO animals treated with DMSO show hemorrhages in the region of the cerebellum (arrows). These are markedly reduced in mice treated with MS37452 (H). (G, H) Isolectin B4 (ISOB4) immunostainings reveal fewer lesions in CCM3^iECKO mice that were treated with MS37452, reducing markedly the number of lesions (G, H, arrows). (I) Quantifications of lesion numbers and (J) total lesioned area. Statistical testing is based on unpaired student’s T-test (s.e.m bars indicated). (K) Model figure depicting the control of vasoprotective and pathophysiological gene expression through the interplay of the proteins of the CCM complex, the PRC1 complex protein CBX7, the transcription factor KLF2, and blood flow as a biophysical modulator. Under physiological conditions, CBX7 expression is negatively controlled by blood flow and the CCM complex downregulates KLF2 expression. The loss of CCM becomes detrimental in low-shear stress regions due to the activation of KLF2 expression. In addition, CBX7 expression becomes upregulated due to the lack of blood flow. Together this renders pathophysiological targets, such as WNT9B, TEK, and ANGPT1, available for KLF2-dependent activation. AVC atrioventricular canal, endo endocardium, myo myocardium, Ven ventricle, Atr atrium. Experiments in zebrafish were done in three biological replicates. Scale bars are (B–D) 50 µm; (B’–D’) 20 µm; (G, H) 500 µM. Source data are available online for this figure.