Recessive osteogenesis imperfecta caused by missense mutations in SPARC

Abstract

Secreted protein, acidic, cysteine-rich (SPARC) is a glycoprotein that binds to collagen type I and other proteins in the extracellular matrix. Using whole-exome sequencing to identify the molecular defect in two unrelated girls with severe bone fragility and a clinical diagnosis of osteogenesis imperfecta type IV, we identified two homozygous variants in SPARC (GenBank: NM_003118.3; c.497G>A [p.Arg166His] in individual 1; c.787G>A [p.Glu263Lys] in individual 2). Published modeling and site-directed mutagenesis studies had previously shown that the residues substituted by these mutations form an intramolecular salt bridge in SPARC and are essential for the binding of SPARC to collagen type I. The amount of SPARC secreted by skin fibroblasts was reduced in individual 1 but appeared normal in individual 2. The migration of collagen type I alpha chains produced by these fibroblasts was mildly delayed on SDS-PAGE gel, suggesting some overmodification of collagen during triple helical formation. Pulse-chase experiments showed that collagen type I secretion was mildly delayed in skin fibroblasts from both individuals. Analysis of an iliac bone sample from individual 2 showed that trabecular bone was hypermineralized on the material level. In conclusion, these observations show that homozygous mutations in SPARC can give rise to severe bone fragility in humans.

Figure 1

Phenotypic Findings in Individuals 1 and 2 (A and B) Anteroposterior and lateral views of the spine of individual 1 (age: 6 years), demonstrating severe scoliosis, kyphosis, and vertebral compression fractures. (C) Right hip and proximal femur of individual 1 (age: 14 years). The femoral neck has a very small diameter. An intramedullary rod has been inserted to treat repeated fractures and femoral bowing. (D) Hand and wrist radiograph of individual 1 (age: 14 years), showing metaphyseal lines from prior intravenous pamidronate infusions. The skeletal maturation (bone age) corresponds to chronological age. (E and F) Anteroposterior and lateral views of the spine of individual 2 (age: 6 years), showing mild scoliosis, kyphosis, and severe vertebral compression fractures. (G) The femurs of individual 2 (age: 6 years) have a small diameter and thin cortices, but are straight. A callus is visible on the right femur after a diaphyseal fracture. Metaphyseal lines resulting from pamidronate infusions are also visible. (H) Hand and wrist radiograph of individual 2 (age: 5 years), showing thin cortices of metacarpal bones but no dysplastic features. (I) Forearm peripheral quantitative computed tomography of individual 1 at the age of 14 years, compared to a healthy control of the same age and sex. The size of the two forearm cross-sections is similar, but the outer size of the radius (marked by arrows) is much lower in individual 1 (total cross-sectional area of the radius: 57 mm²; age- and sex-matched Z score: −3.9) than in the control individual (118 mm², Z score +0.9). The percentage of fat (dark gray area, marked by an F) in the forearm cross-section is increased (48%), compared to the control (28%). The reverse is true for muscle (light gray area, marked by an M). (J) Bone mineralization density distribution in cancellous bone of individual 2. The curve labeled in gray represents pediatric reference data (mean and 95% confidence interval), as established previously in our laboratory.

Figure 2

Confirmation of the Homozygous SPARC Mutations and Prediction of Their Impact on SPARC (A) Sanger sequencing results of individuals 1 and 2 and their parents. Individual 1 (II-1 in family 1) is homozygous for the c.497G>A variant, and both parents are heterozygous. Individual 2 (II-1 in family 2) is homozygous for the c.787G>A variant, and both parents are heterozygous. (B) Schematic representation of the coding exons of SPARC cDNA (GenBank: NM_003118.3) and of SPARC protein (GenBank: NP_003109.1). The locations of the exons are aligned relative to the regions of the SPARC protein that each exon encodes. The positions of the mutations reported in the present study are indicated. Abbreviations are as follows: S, signal peptide; Acidic, acidic domain; FS, follistatin-like domain; EC, extracellular collagen binding domain. (C) The mutated SPARC residues Arg166 and Glu263 and several adjoining residues at each site are perfectly conserved among species. (D) Modeling of the interaction between SPARC and collagen type I. Top: SPARC EC domain with the triple helical collagen type I alpha chains shown in light blue (alpha 1 chain) and dark blue (alpha 2 chain). Bottom: the critical regions of SPARC and collagen type I are magnified. The Phe residue of the alpha 1 chain is surrounded by the “Phe pocket” of SPARC. Arg166 and Glu263 form a salt bridge that is critical for maintaining the Phe pocket. When Arg166 or Glu263 are substituted by other residues, the loop conformation of the Phe pocket can not be maintained and collagen binding is impaired. (E) Immunoblot of SPARC protein in conditioned medium of skin fibroblasts from a control individual (C) and from individuals 1 (I1) and 2 (I2). The amount of SPARC in the medium is clearly lower in individual 1. Ponceau staining (bottom) shows equal loading in the control and individual 1, but somewhat lower loading for individual 2.

Figure 3

Collagen Type I Studies in Skin Fibroblasts (A) Biochemical analysis of collagen type I in the cell layer. Skin fibroblasts of individuals 1 and 2 (P) synthesize collagen type I alpha chains that have a very mild delay in migration, as compared to a control sample (C). (B) Pulse-chase analysis of collagen type I secretion. The bands in each set represent the collagen trimers that migrated into the gel under nonreducing conditions. In individuals 1 and 2, there is a delay in secretion of the type I collagen trimer.