Joan C. Marini, MD, PhD, Head, Section on Connective Tissue Disorders
Weizhong Chang, PhD, Staff Scientist
Antonella Forlino, PhD, Guest Scientist
Thomas Uveges, PhD, Postdoctoral Fellowp
Jiandong Yang, PhD, Postdoctoral Fellow
Mona Abukhaled, MSN, CRNP, Senior Research Assistant
Wayne A. Cabral, AB, Chemist
Aileen M. Barnes, MS, Research Associate (Contractor)
Demaris Yeh, BS, Postbaccalaureate Fellow

In an integrated program of laboratory and clinical investigation, we study the molecular biology of the heritable connective tissue disorders osteogenesis imperfecta (OI) and Ehlers-Danlos syndrome (EDS). Our objective is to elucidate the mechanisms by which primary collagen defects cause skeletal fragility and other significant connective tissue symptoms and then apply the knowledge gained from our studies to the treatment of children with these conditions. An understanding of the interactions of mutant collagen molecules with the normal collagenous and non-collagenous components of extracellular matrix will enhance our understanding of normal bone function and may yield insight into the more common forms of osteoporosis. We recently focused on the development of a non-lethal animal model for OI with a classical collagen mutation. This non-lethal knockin mouse, the Brtl mouse, with a glycine substitution mutation in the α1(I) chain, is an excellent model for pharmacological treatment trials, for approaches to gene therapy suitable for dominant disorders, and for investigating the skeletal matrix of OI. Our clinical studies involve children with types III and IV OI, who form a longitudinal study group enrolled in age-appropriate clinical protocols for treatment.

Mechanism of OI/EDS

Cabral, Marini; in collaboration with Colige, Leikin

We have identified the distinct OI/EDS phenotype and, with Sergey Leikin and Alain Colige, identified the etiology as mutations in a distinct region of type I collagen. We further delineated the mechanism by which the mutations exert their effect. Our probands have a combination of OI Type III or IV as well as severe laxity of large and small joints and early progressive scoliosis. All patients have mutations in the first 90 residues of the α1(I) chain. In contrast to normal procollagen and collagen, the thermal stability of the patients' collagen is less than that of procollagen. Differential scanning calorimetry and proteinase digestion assays define a distinct high-stability folding region at the amino end of type I collagen that anchors the helix. The OI/EDS mutations disrupt the stability of the anchor region and cause it to unfold. The unfolding extends into the adjacent N-propeptide cleavage site, with removal of N-propeptide prevented or incomplete. The resulting pN-collagen is incorporated into matrix deposited in culture. In vivo, dermal fibrils of affected patients are distinctly smaller than those of matched controls, as is seen in EDS VII A and B, owing to the absence of the N-proteinase cleavage site from the α1(I) and α2(I) chain, respectively. Thus, the data provide a direct mechanism for the patients' OI as a consequence of their collagen defect and an indirect mechanism for the EDS symptoms in terms of incorporation of pN-collagen into matrix. The fate of the N-propeptide in matrix is unclear and may be the basis of the phenotypic distinctions among α1(I) OI/EDS, α2(I) OI/EDS, and EDS VII.

The Consortium for collagen mutations in OI

Forlino, Cabral, Barnes, Milgrom,1 Marini

We led an international consortium of connective tissue laboratories to assemble and analyze a database of structural mutations in type I collagen causing OI. The consortium contributed 682 glycine substitutions and 150 splicing site defects, for a total of over 830 mutations. Genotype-phenotype modeling revealed different functional relationships for each chain of type I collagen. Glycine substitutions in the α1 chain have a generally more severe outcome, with 36 percent of substitutions resulting in a lethal phenotype. Substitutions by residues with charged or branched side chains have a lethal outcome in the majority of occurrences. Mutations in the amino quarter of the chain are non-lethal, even those involving residues with charged or branched side chains. We observed two stretches of exclusively lethal mutations in the carboxyl quarter of the chain. The two regions coincide with the Major Ligand Binding Regions (MLBR) of several ligands on the collagen monomer, including integrins and fibronectin. Glycine substitutions in the α2 chain have a more moderate outcome on average, with less than 20 percent of occurrences resulting in lethality. Substitutions by residues with charged side chains are predominantly lethal, as in α1, but valine, with a branched side chain, is lethal in 17 percent of occurrences in α2 (as compared with 73 percent in α1). As for α1, occurrences in the amino third of the chain are non-lethal. Thereafter, the lethal mutations occur in eight clusters that are regularly spaced along the chain. The distribution of lethal mutations in α2 continues to follow the pattern we previously described in the Regional Model for this chain; the cluster boundaries predict the lethal versus non-lethal outcome of 86 percent of cases. The lethal regions coincide with the binding regions for matrix proteoglycans on the collagen fibril. Recurrences of glycine substitutions show a strong association with CpG dinucleotides, although many CpG dinucleotides are not the sites of recurrences. Finally, splice site mutations lead to a mild type I OI outcome in a minority of cases. Most splice site mutations, even those in invariant +1,+2 or --1 or --2 positions, lead to significant dysplasia, suggesting that use of alternative donor or acceptor sites generates translatable products that can be incorporated into matrix. Most mutations that lead to simple exon skipping have a severe or lethal outcome. Our previous modeling has provided testable hypotheses for the mechanism of the dysplasia.

Diagnostic testing for OI

Cabral, Barnes, Milgrom,1 Marini

We conducted a systematic study of the sensitivity of the biochemical test commonly used for the diagnosis of OI, for which 90 percent sensitivity has been claimed. We used the fibroblast collagen from OI cases whose collagen mutations we had identified by examining the gel-electrophoretic migration of both media (secreted) and cell layer collagens under a variety of conditions. Collagens were organized linearly according to chain of mutation and position of mutation in a chain. We demonstrated that the biochemical test is variably sensitive for glycine substitutions in the amino third of the α1 and the amino half of the alpha2 chain. We observed no correlation of sensitivity with substituting residue. The same residue may be detected by the biochemical test amino-terminal to a position at which it is undetected, suggesting that folding regions along the chain are involved in variable sensitivity. We estimate that the biochemical test may miss up to 20 percent of mutations and therefore recommend that a negative biochemical test be supplemented by sequencing the exons encoding the amino half of the chains.

Murine models for OI

Uveges, Forlino, Yang, Marini; in collaboration with Goldstein, Leikin

We previously generated the Brtl mouse model for OI. It bears a classical OI-causing mutation (G349C) in one COL1A1 allele. Our investigations suggest that the heterogeneity of the OI matrix may be a substantial contributor to bone dysplasia. Given that OI is a dominant disorder, we expected that mice carrying two copies of the Brtl allele (Brtl/Brtl) would have a more severe or lethal condition. Surprisingly, Brtl/Brtl mice have a milder phenotype, with growth, bone mineral density (BMD), and femoral geometry and mechanical properties intermediate between wild type and Brtl/+. Brtl/Brtl synthesizes homogeneous collagen with mutant α1(I) chains only. The Brtl/Brtl mouse has less matrix insufficiency than the Brtl/+ mouse; we thus sought to determine whether insufficiency was a significant factor. For this purpose, we made a compound mouse that synthesizes only mutant α1 chain and has increased insufficiency; we crossed Brtl/+ with mov/+ mice, which have a null col1a1 allele. Analysis of Brtl/mov indicates that the compound mice also have a milder phenotype than Brtl/+. We are currently generating G349S mice and a COL1A1 knockout mouse to continue our studies of the insufficiency mechanism.

Bisphosphonate treatment of children with types III and IV OI

Abukhaled, Marini; in collaboration with Paul

We undertook the first randomized controlled trials of bisphosphonate in children with types III and IV OI. The aim was to test both the primary (increased vertebral bone density and decreased fractures) and secondary (improved functional level and muscle strength and decreased pain) gains reported from uncontrolled treatment trials. Children in the treatment group received pamidronate for 18--23 months. All children underwent quarterly rehabilitation and physical therapy assessments, including measurements of function, strength, and pain. The controlled trial demonstrated improvement in vertebral parameters of the treatment group, including improved vertebral BMD z-scores, central vertebral height, and total vertebral area. The patients did not experience a decrease in long-bone fractures. Furthermore, we found that the increment in vertebral BMD in the treatment group tapered off after one to two years of treatment.

In the context of maximized physical rehabilitation, we did not see an additional functional effect from bisphosphonate treatment. In contrast to reports from uncontrolled trials, we found no significant changes in ambulation level, lower-extremity strength, or pain in children with OI treated with pamidronate. Some patients reported increased endurance or decreased back pain, but most reported no perceptible changes. The previously reported changes in these parameters appear to have been a placebo effect in the uncontrolled trials. The data from our controlled trial were in accord with the data from two European controlled trials and were the subject of a review of all the controlled trials. We are now recommending that treatment of children with types III and IV OI with pamidronate be limited to one to three years, with subsequent follow-up of bone status. Furthermore, we are currently engaged in a dose comparison trial, using the dose from our first trial and a lower dose. Our hypothesis is that the children will gain a comparable benefit from the lower dose with deceased detrimental effects.

1 Sarah Milgrom, BS, former Postbaccalaureate Fellow

COLLABORATORS

Alain Colige, PhD, Université de Liège, Liège, Belgium
Steven Goldstein, PhD, University of Michigan, Ann Arbor, MI
Sergey Leikin, PhD, Section on Physical Biochemistry, NICHD, Bethesda, MD
Scott Paul, MD, Rehabilitation Medicine, Warren G. Magnuson Clinical Center, NIH, Bethesda, MD
The OI Mutation Consortium
For further information, contact marinij@mail.nih.gov.