Enhancer hijacking at the ARHGAP36 locus is associated with connective tissue to bone transformation

Abstract

Heterotopic ossification is a disorder caused by abnormal mineralization of soft tissues in which signaling pathways such as BMP, TGFβ and WNT are known key players in driving ectopic bone formation. Identifying novel genes and pathways related to the mineralization process are important steps for future gene therapy in bone disorders. In this study, we detect an inter-chromosomal insertional duplication in a female proband disrupting a topologically associating domain and causing an ultra-rare progressive form of heterotopic ossification. This structural variant lead to enhancer hijacking and misexpression of ARHGAP36 in fibroblasts, validated here by orthogonal in vitro studies. In addition, ARHGAP36 overexpression inhibits TGFβ, and activates hedgehog signaling and genes/proteins related to extracellular matrix production. Our work on the genetic cause of this heterotopic ossification case has revealed that ARHGAP36 plays a role in bone formation and metabolism, outlining first details of this gene contributing to bone-formation and -disease.

Fig. 1. Extreme case of heterotopic ossification.

a Computed tomography (CT) scan at the age of 5 years shows the muscle-to-bone transformation in the proband. b Array-CGH detected an 820 kb duplication on chr2. c Whole genome sequencing mapped the duplication to the chrX (orange: chrX; blue: chr2). d Sanger sequencing mapped the breakpoints at the base pair level. e Schematic representation of the duplicated genes from chr2 and the insertion at the ARHGAP36/IGSF1 locus in chrX.

Fig. 2. Formation of a novel chromatin domain in the proband Hi–C.

a Hi–C map showing ectopic signal in the “chr2–chrX” trans-map of the proband. b Hi–C cis-map shows the topologically associating domain (TAD) containing ARHGAP36 and IGSF1. In the proband, the insertion breakpoint (vertical dashed line) reduces the chromatin interaction within this chromatin domain. c Customized “chrX–chr2–chrX” map in control shows blank spaces between chr2–chrX contacts. Two putative novel chromatin domains left (#1) and right breakpoints (#2), contain no mapped Hi–C reads in control, as expected. On the other hand, proband custom-map shows Hi–C reads in #1, indicating not only physical proximity between chr2–chrX but also the formation of a new chromatin domain (Shuffled-TAD). A weak Hi–C signal is observed at the right breakpoint (#2). d Schematic representation of the rearranged der(X). Putative enhancers located on the ANTXR1 gene body may ectopically activate ARHGAP36 in a cell-type specific manner.

Fig. 3. Chromatin activity and gene expression evaluation of der(X).

a Schematic representation of the customized “chrX–chr2–chrX” 3D genomic landscape (top). Active and inactive chromatin domains were classified based on histone mark signals from MSCs (middle) (Fig. S2c). MSC enhancers were called by the enhancer tool CRUP and showed enrichment at the ANTXR1 gene body. RNA-seq in fibroblasts in the proband and controls (bottom). Purple color represents the proband sample, and gray represents the controls. Expression data is VST normalized. Pb: proband; Ct: control. b ARHGAP36 is upregulated (activated) in the proband in comparison to controls, while IGSF1 is not expressed in all tested fibroblasts. Statistical significance within the indicated groups was calculated using the Wald test (DESeq2) and Benjamini-Hochberg multiple comparisons test with a 95% confidence interval of the fitted general linear model; p-value *<0.05; **<0.01; ***<0.001. c Western blot confirmed activation of ARHGAP36 in the proband fibroblasts (n = 2 technical replicates). Source data are provided as a Source Data file.

Fig. 4. RNA-seq in MSCs upon ARHGAP36 and GFP transfection.

a Schematic representation of ARHGAP36 and GFP transfection in MSCs. RNA-seq was performed in samples collected at 1 and 4 days after transfection. b Four clusters (K1, K3, K4, and K5) show high variability in ARHGAP36-transfect samples and minor variance among GFP ones. c Gene ontology and KEGG analysis revealed enrichment for TGFβ-BMP pathways in K1, regulation of osteogenesis in K3, and tight junction (cell-adhesion) in K4.

Fig. 5. ARHGAP36 function in TGFβ pathway in mouse and human cell lines.

a Schematics of transfection protocol in NIH/3T3 (fibroblast) and C2C12 (myoblast-like) mouse lines (top). Murine cells were transfected with an empty plasmid (as control), hARHGAP36 (purple), and mIgsf1 (as positive control; blue) to evaluate TGFβ pathway activity. These cells were co-transfected with CAGA₁₂MLP-Luc plasmid (TGFβ-sensitive reporter). Two different TGFβ1 ligand concentrations were used in this assay (0.1 and 0.2 nM, plus MOCK). TGFβ1 in NIH/3T3 cells: Mock and 0.1 nM, n = 4 technical replicates; TGFβ1 in C2C12 cells: Mock and 0.1 nM, n = 5; 0.2 nM, n = 4 technical replicates. NIH/3T3 cells showed a reduction of TGFβ activity after TGFβ1 induction at two concentrations (bottom). TGFβ inhibition is more dramatic in C2C12 cells at higher TGFβ1 concentrations. mIgsf1, here used as a positive control, only inhibits TGFβ in myoblast-like cells. Statistical significance within the indicated groups was calculated using two-way ANOVA and Dunnett’s multiple comparisons tests; p-value: *<0.05, **<0.01. Relative Luminescence Units (RLU) are expressed as mean fold induction ±SD over unstimulated transfected control cells. b Schematic representation of the TGFβ experiment in proband fibroblasts and controls (top). Cells were seeded for 24 h in fibroblast media; on the next day, media was replaced by media without serum for 5 h. Cells were induced with TGFβ1 at 0.2 nM and collected at three-time points for SMAD3 phosphorylation analysis. c Proband cells show decreased pSMAD3 levels by western blot at 30 and 60 min after TGFβ1 induction (bottom) (n = 1 proband, n = 3 controls; 6 technical replicates of each sample). Densitometric quantification of pSMAD3 and pSMAD1/5 relative to GAPDH levels expressed as mean fold induction ±SD in arbitrary units. Source data are provided as a Source Data file.

Fig. 6. Osteogenic differentiation in fibroblasts from the patient and controls.

a Schematic representation of the osteogenic protocol. Cells (n = 1 proband, n = 3 controls; 3 technical replicates of each sample) were stimulated with glycerophosphate (10 mM), ascorbic acid (50 µM), and dexamethasone (100 nM), were collected in six-time points (0–5 weeks) and stained for alizarin red (AR) for calcium deposition. b Patient samples showed a faster osteogenic differentiation after two weeks of stimulation, lasting 3-4 weeks. The control values at each time point were set to 1.0. The translucent band shows the confidence interval for each replicate for the respective time point. c Schematic representation of HH inhibition by GANT61 (1 and 5 µM) during the osteogenic differentiation. d GANT61 (5 µM) reduces by half GLI1 expression in both patient and control samples after 48 h of molecule exposure (n = 1 proband, n = 1 control). Pb: proband; Ct: controls. RT-qPCR quantification is expressed as mean fold ±SD in arbitrary units. Source data are provided as a Source Data file. e HH inhibition by GANT61 (5 µM) revealed a minor decrease of 10% of AR staining in all samples. The median score is represented by the horizontal line in the center. The 25th and 75th percentile values are indicated by the lower and upper limits of the box. Source data are provided as a Source Data file.