Skip to main content

National Institutes of Health

Eunice Kennedy Shriver National Institute of Child Health and Human Development

2020 Annual Report of the Division of Intramural Research

Genetic Disorders of Bone and Extracellular Matrix

Joan Marini
  • Joan C. Marini, MD, PhD, Chief, Section on Heritable Disorders of Bone and Extracellular Matrix
  • Aileen M. Barnes, MS, Biologist
  • Ying Liu, PhD, Biologist
  • Gali Guterman-Ram, PhD, Postdoctoral Fellow
  • Milena Jovanovic, PhD, Postdoctoral Fellow
  • Heeseog Kang, PhD, Postdoctoral Fellow
  • Allahdad Zarei, PhD, Postdoctoral Fellow
  • Sonali Gupta, BS, Postbaccalaureate Intramural Research Training Award Fellow
  • Mia Hancock, BS, Postbaccalaureate Intramural Research Training Award Fellow
  • Ariana Taj, BA, Postbaccalaureate Intramural Research Training Award Fellow

In an integrated program of laboratory and clinical investigation, we study the molecular biology of the heritable connective tissue disorders collectively known as osteogenesis imperfecta (OI). Our objective is to elucidate the mechanisms by which the primary gene defect causes skeletal fragility and other connective tissue symptoms and then to apply this knowledge to the treatment of children with these conditions. We identified several key genes causing recessive and X-linked OI. Discoveries of defects in collagen modification have generated a new paradigm for OI as a collagen-related disorder of matrix. We established that structural defects in collagen cause dominant OI, while deficiency of proteins that interact with collagen for folding, post-translational modification, or processing cause recessive OI. Our challenge now is to understand the cellular and biochemical mechanisms of recessive OI. We also generated a knock-in murine model for OI with a classical collagen mutation as well as a murine model for recessive type IX OI and X-linked type XVIII OI, and we are using these models to study disease pathogenesis and the skeletal matrix of OI, the effects of pharmacological therapies, and approaches to gene therapy. Our clinical studies involve predominantly children with types III and IV OI, who form a longitudinal study group enrolled in age-appropriate clinical protocols for the treatment of their condition.

We are also investigating melorheostosis, a very rare bone dysostosis, which is characterized by radiographic patterns of either “dripping candle wax” or endosteal bone overgrowth. We recently identified mosaic mutations in the oncogene MAP2K1 as the cause “dripping candle wax” melorheostosis and somatic mutations in SMAD3 as causative of endosteal melorheostosis. In each gene, the causative mutations occur at a hot spot and result in gain-of-function. We are now developing animal models for studies of melorheostosis pathophysiology and treatment.

Mechanism of rare forms of osteogenesis imperfecta

Recessive null mutations in SERPINF1, which encodes pigment epithelium–derived factor (PEDF), cause OI type VI. PEDF is well-known as a potent anti-angiogenic factor. Type VI OI patients have no serum PEDF, elevated alkaline phosphatase (ALPL) as children, and bone histology with broad unmineralized osteoid and a fish-scale pattern. However, we identified a patient with severe atypical type VI OI, whose osteoblasts displayed minimal secretion of PEDF, but whose SERPINF1 sequences were normal despite typical type VI OI bone histology. Surprisingly, exome sequencing on this proband and family members yielded a de novo mutation in IFITM5 (the gene encoding interferon-induced transmembrane protein 5, which is mutated in type V OI) in one proband allele, causing a p.S40L substitution in the intracellular domain of BRIL, the encoded protein. The IFITM5 transcript and BRIL were normal in proband fibroblasts and osteoblasts. SERPINF1 expression and PEDF secretion were reduced in proband osteoblasts. In contrast, osteoblasts from a typical case of type V OI have elevated SERPINF1 expression and PEDF secretion during osteoblast differentiation. Together, the data suggest that BRIL and PEDF have a relationship that connects the genes for types V and VI OI and their roles in bone mineralization.

The endoplasmic reticulum (ER)–resident procollagen 3-hydroxylation complex is responsible for the 3-hydroxylation of type I collagen alpha1(I) chains. Deficiency of components of the collagen P3H (prolyl 3-hydroxylase) complex causes recessive OI. For type VIII OI, we investigated bone and osteoblasts. Collagen has near-absent 3-hydroxylation from both bone and dermis, demonstrating that P3H1 is the unique enzyme responsible for collagen 3-hydroxylation. Bone histomorphometry revealed patches of increased osteoid, although the overall osteoid surface was normal. Quantitative backscattered electron imaging (qBEI) showed increased mineralization of cortical and trabecular bone, as in other OI types. However, the proportion of bone with low mineralization was higher in type VIII bone than in type VII, consistent with patchy osteoid occurring only in type VIII.

The third member of the complex, cyclophilin B (CyPB), encoded by PPIB, is an ER–resident peptidyl-prolyl cis-trans isomerase (PPIase). CyPB is the major PPIase catalyzing collagen folding. We characterized the first patient with deficiency in PPIB, which causes recessively inherited type IX OI. Our group generated a Ppib knock-out (KO) mouse model that recapitulates the type IX OI phenotype. Intracellular collagen folding occurs more slowly in CyPB null cells, supporting the enzyme's role as the rate-limiting step of folding. However, treatment of KO cells with the cyclophilin inhibitor cyclosporin A caused further delay in folding, providing support for the existence of a further collagen PPIase. We found that CyPB supports collagen lysyl hydroxylase 1 (LH1) activity, demonstrating significantly reduced hydroxylation of the helical crosslinking residue K87, which directly affects both the extent and type of collagen intermolecular crosslinks in bone. However, CyPB deficiency results in increased hydroxylation at telopeptide crosslinking sites in tendon, with moderate increase in glycosylation. In our recent collaboration with Mitsuo Yamauchi and colleagues [Reference 2], we explored the role of CyPB in post-translational modifications of collagen in skin. As in bone, hydroxylation of collagen crosslinking sites was almost absent in mutant mice, and the key cross-linking residue alpha1(I)K87 was under-glycosylated. Absence of CyPB led to the occurrence of two novel types of collagen crosslinks that are not present in normal skin. Atomic force microscopy showed that this was associated with a lower nano-indentation modulus in KO than in normal skin. The studies underscore the tissue-dependent effects of CyPB, which have common effects on cross-linking and mechanical properties.

In collaboration with Vorasuk Shotelersuk and Cecilia Giunta, we identified a new OI–causative gene on the X-chromosome. This is the first type of OI with X-linked inheritance, and it causes a moderate to severe bone dysplasia with pre- and postnatal fractures of ribs and long bone, bowing of long bones, low bone density, kyphoscoliosis and pectal deformities, and short stature. Affected individuals have missense mutations in MBTPS2, which encodes the protein S2P. S2P is a transmembrane protein in the Golgi and is a critical component of regulated intramembrane proteolysis (RIP). In RIP, regulatory proteins are transported from the ER membrane to the Golgi in times of cell stress or sterol depletion, where they are sequentially cleaved by S1P/S2P to release activated N-terminal fragments, which enter the nucleus and activate gene transcription. Mutant S2P protein is stable but has impaired RIP functioning, with deficient cleavage of the ER–stress transducers OASIS, ATF6, and SREBP. Furthermore, hydroxylation of the collagen residue K87 is reduced by half in proband bone, consistent with reduced lysyl hydroxylase in proband osteoblasts. Reduced collagen crosslinks presumably undermine bone strength. The mutations in MBTPS2 demonstrate that RIP plays a fundamental role in bone development.

C-propeptide cleavage-site mutations increase bone mineralization.

Type I procollagen is processed to mature collagen by the removal of both N- and C-terminal propeptides. The C-propeptide is cleaved at the Ala-Asp peptide bond between the telopeptide and the C-propeptide of each chain by procollagen C-proteinase (also known BMP-1 or bone-morphometric protein). Probands with substitutions at any of the four cleavage-site residues have a high-bone-mass form of OI, first reported by our lab in collaboration with Katarina Lindahl. The patients have elevated bone-density DEXA Z-scores and, in bone histology, patchy unmineralized osteoid. The processing of the C-propeptide from collagen secreted by proband cells is delayed. Using bone mineralization density distribution (BMDD), we investigated mineralization to show that, in the alpha2(I) cleavage site mutation, the bone had a uniformly higher mineral density, while in the alpha1(I) mutation, the average mineral density was markedly heterogeneous, with areas of either very high or low bone density.

To investigate the role of the C-propeptide in bone mineralization and developmental progression, we developed a knock-in murine model with a COL1A1 (the gene encoding the pro-alpha1 chain of type 1 collagen) cleavage site mutation. Bone collagen fibrils showed a “barbed-wire” appearance consistent with the presence of the processing intermediate pC-collagen, which was detected in extracts of bone from mutant mice, and with impaired collagen processing in vitro. Impaired C-propeptide processing affects skeletal size and biomechanics. The mice are small, and their femora exhibit extreme brittleness on mechanical testing, as well as reduced fracture load. BMDD measurement on their femora show significantly higher mineralization than in wild-type (WT) mice, which continues to increase in HBM (high bone mass) mice even after WT mineralization plateaus at 6 months. PINP and TRAP, serum markers of bone remodeling, are significantly elevated in HBM mice. Osteocyte density is reduced but lacunar area is increased.

Mutations in the COL1A1 C-propeptide

The C-propeptide of type I collagen (COL1A1 C-propeptide) is processed after collagen is secreted from the cell and before it is incorporated into matrix. Interestingly, mutations in the C-propeptide are present in about 6% of OI patients. Our investigation of the biochemical consequences of C-propeptide mutations in comparison with collagen helical mutations revealed both intra- an extracellular differences [Reference 3]. Immunofluorescence microscopy indicated that procollagen with C-propeptide defects was mis-localized to the ER lumen, in contrast to the ER membrane localization of normal procollagen and to helical mutations. Second, although the mutations were not located in the processing site itself, pericellular cleavage of the C-propeptide was defective in both pericellular processing assays and in cleavage assays with purified C-proteinase, consequences that are expected to contribute to abnormal osteoblast differentiation and matrix function, respectively.

Insights from the Brtl mouse model for OI

The Brtl mouse model for OI, generated by our lab, is a knock-in mouse that contains a Gly349Cys substitution in the alpha1(I) collagen chain. Brtl was modeled on a type IV OI child and accurately reproduces features of type IV OI. Brtl has provided important insights into OI mechanism and treatment.

We collaborated with Kenneth Kozloff's group to investigate a potential anabolic therapy, sclerostin antibody (Scl-AB), which stimulates osteoblasts via the canonical Wnt pathway. Scl-AB stimulated bone formation in young Brtl mice and increased bone mass and load-to-fracture. Treatment with Scl-AB caused no detrimental change in Brtl bone material properties. Nano-indentation studies indicated unchanged mineralization, unlike the hyper-mineralization induced by bisphosphonate treatment. In addition, Scl-AB was successfully anabolic in adult Brtl mice, and may be a therapy for adult patients who have fewer treatment options. Because Scl-AB is a short-acting drug, we recently investigated sequential Scl-AB/bisphosphonate treatment. The study showed that administration of a single dose of bisphosphonate after cessation of Scl-AB treatment preserved anabolic gains from Scl-AB. Alternatively, a single low dose of bisphosphonate concurrent with Scl-AB treatment facilitated the anabolic action of Scl-AB by increasing availability of trabecular surfaces for new bone formation. Because a lifelong deficiency of sclerostin leads to patterns of excessive cranial bone growth and nerve compression, we undertook dimensional and volumetric measurements of the skulls of Brtl mice treated with Scl-AB. Treated mice showed calvarial thickening but minimal effects on cranial morphology and anatomic landmarks. Narrowing of vascular but not neural foramina was seen.

Brtl mice provided important information on the cytoskeletal organization in OI osteoblasts and their potential role in phenotypic variability. We observed abnormal cytoskeletal organization involving vimentin, strathmin, and cofilin-1 in lethal pups. Decreased vimentin (an intermediate filament) can lead to cytoskeletal collapse, and increased strathmin (a regulatory factor that promotes microtubular disassembly) and cofilin-1 (an inducer of actin depolymerization) work in concert to disrupt cytoskeletal cellular functions. The alterations affected osteoblast proliferation, collagen deposition, integrin, and TGF-beta signaling. The data suggest that cytoskeletal elements present novel OI treatment targets. Another potential novel treatment may be 4-PBA, a chemical chaperone. When the drug is used to treat OI cells, it enhances autophagy, as opposed to apoptosis, of the cells and stimulates protein secretion. Interestingly, the enhanced protein secretion reflects a broad range of cellular proteins rather than simply the retained mutant collagen and relieves the ER stress along the PERK pathway.

Two basic insights have emerged from Brtl studies. The first concerns hyper-mineralization of OI bone, which was previously thought to be a passive process. Altered levels for osteocyte transcripts involved in bone mineralization, such as Dmp1 and Sost, demonstrated, however, the presence of an actively directed component. We used acoustic transmission microscopy to characterize the properties of Brtl cortical bone. The periodically oriented collagen organization in periosteal cortex of Brtl bone was markedly lower than in normal bone. Young’s modulus and ER sound velocity were significantly increased in Brtl cortex. The data demonstrate that the mutant collagen of Brtl mice affects the mechanical behavior of bone predominantly in the endosteal region by altering collagen orientation. Second, the osteoclast is important for the OI phenotype, with elevated numbers of osteoclasts. Co-culture experiments with Brtl and wild-type (WT) mesenchymal stem cells (MSCs) and osteoclast precursors yielded elevated osteoclast numbers from WT or Brtl precursors grown with Brtl MSCs, but not with WT MSCs. The results indicate that an osteoblast product is necessary to elevate osteoclast numbers.

Natural history and bisphosphonate treatment of children with types III and IV OI

We recently published the cardiopulmonary aspects of our natural history study on types III and IV OI. Longitudinal evaluations were completed in 23 children with type III OI and 23 children with type IV OI, who had pulmonary function tests every 1–2 years. Compared with size-matched children, our patients showed a significant decline over time in pulmonary function, including in lung volumes and flow rates. The decline was worse in the 36 children with scoliosis (average curve 25 degrees) but also occurred in 20 participants without scoliosis, who had declining function with restrictive disease, suggesting that the pulmonary dysfunction of OI is attributable to a primary defect in the lung related to structurally abnormal collagen. The studies are important because pulmonary issues are the most prevalent cause of morbidity and mortality in OI. Affected individuals should seek anticipatory evaluation and treatment.

Previously, OI–specific growth curves were not available, despite the fact that short stature is one of the cardinal features of OI. We assembled longitudinal length, weight, head circumference, and body mass index (BMI) data on 100 children with types III and IV OI with known mutations in type I collagen, to generate sex- and type-specific growth curves for OI. We examined effects of gender, OI type, and pathogenic variant, using multilevel modeling, and we constructed OI–specific centile curves, using a generalized additive model for location, scale, and shape (GAMLSS). The data show that gender and OI type, but not the collagen chain in which the causative mutation is located, have significant effects on height in OI. Boys are taller than girls, and type IV OI boys and girls are taller than type III. In both genders, length curves for types III and IV OI overlap, and the type IV 95th centile curve overlaps the lower U.S. Centers for Disease Control and Prevention (CDC) curves for the general population. A pubertal growth spurt is generally absent or blunted in types III/IV OI. The BMI 50th and 95th centile curves are distinctly shifted above respective CDC curves in both genders. Weight differs by OI type, but not by gender or mutant collagen chain. Interestingly, head circumference does not differ by gender, OI type, or collagen mutation. Imposition of OI height curves on standard CDC curves reveals an overlapping of type III and IV percentiles and the absence of a growth spurt in type III OI. Standard growth curves for OI will be of great value to primary caregivers and families and will provide a baseline for treatment trials.

Our randomized controlled trial of bisphosphonate in children with types III and IV OI was the first randomized bisphosphonate trial for OI in the United States. It examined direct skeletal and secondary gains reported in uncontrolled trials. For skeletal outcomes, we found increased BMD (bone mineral density) Z-scores and improved vertebral geometry. We noted that vertebral BMD improvement tapered off after two years' treatment. Our treatment group did not experience fewer long-bone fractures, coinciding with the lack of improvement or equivocal improvement in fractures in other controlled trials. This controlled trial did not support the claims for improvement in ambulation level, lower-extremity strength, or alleviation of pain, suggesting these were placebo effects in observational trials. Our current recommendation is for treatment for 2–3 years, with subsequent follow-up of bone status. We are now engaged in a dose-comparison trial, comparing the dose from our first trial with a lower dose, achieved by increasing the cycle interval at the same dose/kg/cycle. Our preliminary analysis indicates that OI children obtain comparable benefits from lower and higher doses of pamidronate.

Melorheostosis: genetic and clinical delineation

Melorheostosis is a very rare sporadic bone dysostosis that is characterized by metabolically active bone in the appendicular skeleton, which leads to asymmetric bone overgrowth, seen radiographically as “dripping candle wax,” functional impairment, and pain. Skin overlying the bone lesion sometimes has a hyperpigmented, vascular lesion. Given that attempts to identify germline mutations causing melorheostosis were unsuccessful, we proposed somatic mutations. Our collaborative team (with Tim Bhattacharyya, Richard Siegel, and Nadja Fratzl-Zelman) was the first to look directly at bone samples. Fifteen patients with melorheostosis had paired biopsies of both affected and contralateral unaffected bone.

DNA extracted from each patient’s two bone samples was subjected to whole-exome sequencing (WES); sequences from each individual patient were compared, and secondarily compared among the set of patients. We identified two genes causing somatic mutations in melorheostotic lesions. Each gene was associated with one of the radiographic forms of melorheostosis, and the bone lesions had distinct histology and mechanism along the TGFß pathway. In affected but not unaffected bone or blood, eight of the 15 patients had somatic mutations for MAP2K1, located in two adjacent residues of the negative regulatory domain and would be expected to increase MEK1 activity. Increased MAPK activity along the non-canonical TGFβ pathway leads to increased phosphorylation and activation of ERK1/2, accounting for the mosaic pattern of increased p-ERK1/2 in osteoblasts on immunohistochemistry of affected bone. Osteoblasts cultured from affected bone constitute two populations with distinct p-ERK1/2 levels by flow cytometry, enhanced ERK1/2 activation, and elevated cell proliferation. However, the MAP2K1 mutations inhibit BMP2–mediated osteoblast mineralization and differentiation in vitro, underlying the markedly increased osteoid detected in affected bone histology. Our data show that the MAP2K1 oncogene is important in human bone formation and implicate MAP2K1 inhibition as a potential treatment avenue for melorheostosis.

Four patients were determined to have causative somatic mutations in SMAD3 [Reference 4], a component of the canonical TGFβ pathway. SMAD3 phosphorylation was increased in affected bone, and downstream target genes of TGFβ signaling had increased expression. The mutations were associated with an endosteal radiographic pattern. Cultured osteoblasts from affected bone exhibited decreased proliferation in vitro, increased expression of osteoblast differentiation markers, and increased mineralization. However, the constitutive activation of the SMAD3 dampened the activity of BMP2, because addition of BMP2 to culture media decreased osteoblast differentiation and mineralization in vitro. Bone lesions with SMAD3 mosaicism did not show increased cellularity or osteoid accumulation and were more highly mineralized.

Melorheostotic bone from both MAP2K1–positive and SMAD3–positive patients showed two zones of distinct morphology. In MAP2K1–positive melorheostosis [Reference 5], the inner osteonal zone is intensively remodeled and has increased osteoid, which is covered by a zone containing compact multi-layered lamellae. The remodeling zone has lower than normal bone mineralization and a high porosity, reflecting increased tissue vascularity. The lamellar portion is less mineralized than the remodeling zone, indicating a younger tissue age. Nano-indentation was not increased in the lamellar zone, indicating that the surgical hardness of this bone reflects its lamellar structure. We propose that the genetically induced deterioration of bone micro-architecture in the remodeling zone triggers a periosteal reaction.

Additional Funding

  • NICHD DIR Director’s Award


  1. Cabral WA, Fratzl-Zelman N, Weis MA, Perosky JE, Alimasa A, Harris R, Kang H, Makareeva EN, Barnes AM, Roschger P, Leikin S, Klaushofer K, Forlino A, Backlund PS, Eyre DR, Kozloff KM, Marini JC. Substitution of murine type I collagen A1 3-hydroxylation site alters matrix structure but does not recapitulate osteogenesis imperfecta bone dysplasia. Matrix Biol 2020;90:20–39.
  2. Terijima M, Taga Y, Cabral WA, Nagasawa M, Sumida N, Kayashima Y, Chandrasekaran P, Han L, Maeda N, Perdivara I, Hattori S, Marini JC, Yamauchi M. Cyclophilin B control of lysine post-translational modification of skin type I collagen. PLoS Genet 2019;15:e1008196.
  3. Barnes AM, Ashok A, Makareeva EN, Brusel M, Cabral WA, Weis MA, Moali C, Bettler E, Eyre DR, Cassella JP, Leikin S, Hulmes DJ, Kessler E, Marini JC. COL1A1 C-propeptide mutations cause ER mislocalization of procollagen and impair C-terminal procollagen processing. Biochim Biophys Acta Mol Basis Dis 2019;1865:2210-2223.
  4. Kang H, Jha S, Ivovic A, Fratzl-Zelman N, Deng Z, Mitra A, Cabral WA, Hanson EP, Lange E, Cowen EW, Katz J, Roschger P, Klaushofer K, Dale RK, Siegel RM, Bhattacharyya T, Marini JC. Somatic SMAD3-activating mutations cause melorheostosis by up-regulating the TGF-ß/SMAD pathway. J Exp Med 2020;217:e20191499.
  5. Fratzl-Zelman N, Rochger P, Kang H, Jha S, Roschger A, Blouin S, Deng Z, Cabral WA, Ivovic A, Katz J, Siegel RM, Klaushofer K, Fratzl P, Bhattacharyya T, Marini JC. Melorheostotic bone lesions caused by somatic mutations in MAP2K1 have deteriorated microarchitecture and periosteal reaction. J Bone Miner Res 2019;34:883-895.


  • Patricia Becerra, PhD, Laboratory of Retinal Cell and Molecular Biology, NEI, Bethesda, MD
  • Timothy Bhattacharyya, MD, Clinical and Investigative Orthopedics Surgery Unit, NIAMS, Bethesda, MD
  • Anne De Paepe, MD, PhD, Universitair Ziekenhuis Gent, Ghent, Belgium
  • David Eyre, PhD, University of Washington, Seattle, WA
  • Antonella Forlino, PhD, Università degli Studi di Pavia, Pavia, Italy
  • Nadja Fratzl-Zelman, PhD, Ludwig Boltzmann-Institut für Osteologie, Hanusch Krankenhaus der WGKK und Unfallkrankenhaus Meidling, Vienna, Austria
  • Cecilia Giunta, PhD, Kinderspital Zürich, Zürich, Switzerland
  • Wolfgang Högler, MD, DSc, FRCPCH, Birmingham Children's Hospital NHS Foundation Trust, Birmingham, United Kingdom
  • Kenneth Kozloff, PhD, University of Michigan, Ann Arbor, MI
  • Sergey Leikin, PhD, Section on Physical Biochemistry, NICHD, Bethesda, MD
  • Katarina Lindahl, MD, Uppsala Universitet, Uppsala, Sweden
  • Scott Paul, MD, Rehabilitation Medicine, NIH Clinical Center, Bethesda, MD
  • Cathleen L. Raggio, MD, Weill Medical College of Cornell University, New York, NY
  • Vorasuk Shotelersuk, MD, FABMG, King Chulalongkorn Memorial Hospital, Bangkok, Thailand
  • Richard Siegel, MD, PhD, Autoimmunity Branch, NIAMS, Bethesda, MD
  • Mitsuo Yamauchi, PhD, University of North Carolina, Chapel Hill, NC
  • Joshua Zimmerberg, MD, PhD, Section on Integrative Biophysics, NICHD, Bethesda, MD


For more information, email or visit

Top of Page