CMG2/ANTXR2 regulates extracellular collagen VI which accumulates in hyaline fibromatosis syndrome

Abstract

Loss-of-function mutations in capillary morphogenesis gene 2 (CMG2/ANTXR2), a transmembrane surface protein, cause hyaline fibromatosis syndrome (HFS), a severe genetic disorder that is characterized by large subcutaneous nodules, gingival hypertrophy and severe painful joint contracture. Here we show that CMG2 is an important regulator of collagen VI homoeostasis. CMG2 loss of function promotes accumulation of collagen VI in patients, leading in particular to nodule formation. Similarly, collagen VI accumulates massively in uteri of Antxr2^-/- mice, which do not display changes in collagen gene expression, and leads to progressive fibrosis and sterility. Crossing Antxr2^-/- with Col6a1^-/- mice leads to restoration of uterine structure and reversion of female infertility. We also demonstrate that CMG2 may act as a signalling receptor for collagen VI and mediates its intracellular degradation.

Figure 1. HFS patient nodules display ECM accumulation and loss of skin adnexa and dermal vessels.

(a) H&E staining of formalin-fixed non-nodular skin and nodule from the ear and the head of a HFS patient. The nodule displays a striking accumulation of ECM, causing secondary loss of skin adnexa, vascular structures and rarefaction of the dermal cellular component. One biopsy was obtained by tissue. A representative image is shown for each. Scale bar, 200 μm. (b) Ultrastructure by EM of non-nodular and nodular tissues of a HFS patient. Representative images of non-nodular or nodule tissues at low magnification (upper panel) and higher magnification (lower panel) are shown. At least two grid per tissue biopsies were prepared. Scale bar, 1 μm. (c,e) Comparison of non-nodular (Nn) and nodule (Nod) tissues from the ear and the head of a HFS patient. Tissue lysates were analysed by SDS–PAGE using 4–12% Bis-Tris gradient gels under reducing condition followed either by staining with Coomassie blue (c) or by western blot for collagen VI, collagen I or fibronectin (e). The prominent 100 kDa band (arrowhead) was excised, trypsin digested and its composition analysed by nano LC-MS/MS. Only the three collagen VI alpha chains were detected with >2 peptides. Migration of the molecular weight markers (in kDa) is indicated on the left. See Supplementary Fig. 7a for the uncropped version of the western blots. (d) Comparison of the complete tissue lysates from head and ear non-nodular and nodule analysed by nano LC-MS. The total number of peptides associated with each collagen protein is indicated. Each lysate was run twice in the nano-LC-MS to account for technical variability and both peptide counts are indicated. (f) Quantitative RT–PCR analysis of mRNAs coding for the α1, α2 and α3 chains of collagen VI in fibroblast cultures from unaffected control and four HFS patients (P1–P4) (error bars represent s.e.m.; n=3).

Figure 2. Structural changes in the uteri of Antxr2^−/−mice.

(a) H&E and SR staining of formalin-fixed uterine tissues from 38-week-old WT (Antxr2^+/+) and Antxr2 knockout (Antxr2^−/−) mice. Antxr2 knockout mice display ECM accumulation with disarray, disruption and loss of the longitudinal and circular myometrial layers of the uterus. Representative image of at least n=7 mice per genotype. Scale bar, 200 μm. (b) Ultrastructure of Antxr2^+/+ and Antxr2^−/− uterine myometrial layer analysed by transmission electron microscopy (TEM). N=2 mice. Scale bar, 5 μm (upper panels) and 500 nm (lower panels). (c) Collagen VI immunogold labelling on the Antxr2^+/+ and Antxr2^−/− uterine myometrial layer analysed by TEM. N=2 mice. Scale bar, 500 nm.

Figure 3. Severe collagen VI accumulation in the uteri of Antxr2^−/−mice.

(a) Uterus lysates (20 μg) from littermate WT and Antxr2 knockout (KO) mice were analysed by SDS–PAGE using 4–12% Bis-Tris gradient gels under reducing condition and western blotting against all the collagen VI alpha chains. The signals obtained varied beyond the dynamic range allowed with western blotting and thus the amplitude of the increase was not evaluated for the various alpha chains. Coomassie staining was used as loading controls. Migration of the molecular weight markers (in kDa) is indicated on the left. N=4 mice. See Supplementary Fig. 7b for the uncropped version of the western blots. (b) Quantitative RT–PCR analysis of genes involved in the TGFβ pathway was performed on uteri from WT (Antxr2^+/+) and Antxr2 knockout (Antxr2^−/−) mice. (Error bars represent s.e.m.; n=3; *P<0.05; two-tailed unpaired t-test). (c,d) Quantitative RT–PCR analysis of mRNAs coding for the α1 and α2 chains of collagens I, IV and VI in WT (WT) and Antxr2 KO (KO) mouse uterus. (Error bars represent s.e.m.; n=6–7; N.S., P>0.05; two-tailed unpaired t-test).

Figure 4. Regulation of MMPs activity on CMG2 loss of function.

(a) Lysates from non-nodular (Nn) and nodule (Nod) tissues from the ear and the head were analysed by SDS–PAGE and western blotting against MMP2 and MMP14. (b) Supernatant of HFS patients or control fibroblasts were normalized to protein concentration and analysed by gelatin zymography. Cell lysates were analysed by SDS–PAGE and western blotting against MT1–MMP, CMG2 and actin as loading control. (c) Uterus lysates from Antxr2^+/+ or Antxr2^−/− mice were analysed by SDS–PAGE and western blotting against MMP2 and MMP14. (d) Quantification of active MMP2 and MMP14 by densitometric analysis. (e) RPE1 cells were treated with control siRNA or siRNA against CMG2 for 72 h. Cell supernatant were then normalized to protein concentration and used for enzyme zymography. Cell lysates were analysed by SDS–PAGE and western blotting against CMG2, MMP14 and actin as a loading control. (f) Quantification of active MMP2 and MMP14 by densitometric analysis. (d,f) Error bars represent s.e.m.; n=3; *P<0.05; **P<0.01; two-tailed unpaired t-test compared to control. See Supplementary Fig. 8a–d for the uncropped version of the western blots.

Figure 5. Deletion of Col6a1 is sufficient to rescue the uterine phenotype of Antxr2^−/−mice.

H&E and SR staining of uterine tissues from 15-week-old Antxr2^+/+::Col6a1^+/+, Antxr2^+/+::Col6a1^−/−, Antxr2^−/−::Col6a1^+/+ and Antxr2^−/−::Col6a1^−/− mice in metestrus. Failure of parturition in Antxr2^−/−::Col6a1^+/+ mice is caused by the same mechanisms described for Antxr2^−/− (ref. 10), and this phenotype is reversed after ablation of collagen VI in Antxr2^−/−::Col6a1^−/− mice. Representative image of at least n=3 mice per genotype. Scale bar, 100 μm.

Figure 6. Collagen VI binding to CMG2 leads to signalling and degradation.

(a) Collagen I, collagen IV and collagen VI, fibronectin, laminin, PA and BSA were coated on 96-well plates. Subsequently, purified GST, WT CMG2 vWA-GST (WT) or D50A mutant CMG2 vWA-GST (D50A) were incubated at 1 μg μl⁻¹ for 2 h at 25 °C and the interaction revealed using an antibody against GST and a horseradish peroxidase-coupled secondary antibody. The reaction was quantified with a plate reader at 450 nm. (b) HeLa cells were transfected for 48 h with a C terminally V5-tagged WT or D50A mutant CMG2 (D50A), and subsequently plated onto different dishes coated or not with either collagen I, collagen IV, collagen VI, PA, laminin or fibronectin. Cells were collected 60 min after plating. V5 immunoprecipitates were then analysed by SDS–PAGE using 4–12% Bis-Tris gradient gels under reducing condition and western blotting for phospho-Tyr and V5. The phospho-Tyr signal after immunoprecipitation against the V5-tag was quantified by densitometric analysis, and the values normalized as a percentage of the average signal. (c) RPE1 cells were plated on dishes coated with collagen I, collagen IV and collagen VI, PA, laminin, fibronectin or on uncoated dishes. Cells were collected 60 min after plating. β-arrestin immunoprecipitates were then analysed by SDS–PAGE using 4–12% Bis-Tris gradient gels under non-reducing condition and western blotting for endogenous CMG2 and β-arrestin. The CMG2 signal was quantified by densitometric analysis. (d–e) CMG2 knockdown RPE1 cells (d) or control and P3 fibroblasts (e) were blocked with 5% bovine serum albumin 30 min before addition of purified collagen VI tetramer at 1 μg ml⁻¹. Cells were collected 1, 3 or 6 h later. Collagen VI degradation was assessed by SDS–PAGE using 4–12% Bis-Tris gradient gels under non-reducing condition and western blotting for collagen VI, endogenous CMG2 and actin as a loading control. (a–e) Error bars represent s.e.m.; n=3; *P<0.05; **P<0.01; ***P<0.001; two-tailed unpaired t-test to BSA (a), uncoated (b,c) or control (d,e). See Supplementary Fig. 8e–h for the uncropped version of the western blots.