The Material Basis for Reduced Mechanical Properties in oim Mice Bones

Authors


Abstract

Osteogenesis imperfecta (OI), a heritable disease caused by molecular defects in type I collagen, is characterized by skeletal deformities and brittle bones. The heterozygous and homozygous oim mice (oim/+ and oim/oim) exhibit mild and severe OI phenotypes, respectively, serving as controlled animal models of this disease. In the current study, bone geometry, mechanics, and material properties of 1-year-old mice were evaluated to determine factors that influence the severity of phenotype in OI. The oim/oim mice exhibited significantly smaller body size, femur length, and moment of area compared with oim/+ and wild-type (+/+) controls. The oim/oim femur mechanical properties of failure torque and stiffness were 40% and 30%, respectively, of the +/+ values, and 53% and 36% of the oim/+ values. Collagen content was reduced by 20% in the oim/oim compared with +/+ bone and tended to be intermediate to these values for the oim/+. Mineral content was not significantly different between the oim/oim and +/+ bones. However, the oim/oim ash content was significantly reduced compared with that of the oim/+. Mineral carbonate content was reduced by 23% in the oim/oim bone compared with controls. Mineral crystallinity was reduced in the oim/oim and oim/+ bone compared with controls. Overall, for the majority of parameters examined (geometrical, mechanical, and material), the oim/+ values were intermediate to those of the oim/oim and +/+, a finding that parallels the phenotypes of the mice. This provides evidence that specific material properties, such as mineral crystallinity and collagen content, are indicative and possibly predictive of bone fragility in this mouse model, and by analogy in human OI.

INTRODUCTION

Osteogenesis imperfecta (OI), a heritable disease of the connective tissues, is characterized by bone fragility, skeletal deformities, and, in severe cases, death. Several studies have established the molecular basis of this heterogeneous disease to be mutations in the genes that encode for type I procollagen.(1-6) This in turn causes synthesis of abnormal proα chains that may not correctly fold or self-assemble into the normal collagen fibril, and can lead to changes in fibril orientation and packing.(7-9) Other cellular-based OI studies have demonstrated changes in total quantity of extracellular matrix components, including reduced synthesis and overmodification of collagen and reduced formation of proteoglycans.(1,10-14) Ultimately, it is hypothesized that the disrupted extracellular matrix serves as an abnormal template for mineralization in OI. In addition to reduced bone density in humans with OI,(15-18) researchers have described bone mineral crystals that exhibit abnormal size, shape, composition, and alignment.(18-21) Thus, changes in both quantity and quality of collagen and mineral have been implicated in OI. Despite the ultrastructural information from researchers characterizing OI bone, questions persist as to which specific factors are important determinants of the severity of the bone fragility (i.e., phenotype) observed in OI.

The recent availability of several animal models of OI has facilitated the study of the effects of mutations in collagen genes on properties such as bone geometry and mineralization.(22-28) Mice with the naturally occurring oim mutation are a particularly convenient model to work with. Homozygous mice, termed oim/oim, contain a G nucleotide deletion in both copies of the Cola2 gene (the murine proα2(I) collagen gene) that results in synthesis of nonfunctional proα2(I) chains and accumulation of α1(I)3 collagen.(22) This results in a moderate-to-severe OI phenotypecharacterized by spontaneous fractures and limb deformities. The heterozygous oim/+ mice contain one copy each of the normal and mutant alleles. Originally, it was reported that these mice did not display an OI phenotype.(22) However, recent studies have noted limb deformities and spontaneous fractures in some of these mice,(29) as well as histologic evidence of abnormal bone,(30) indicative of a mild OI phenotype. Thus, the homozygous and heterozygous oim mice serve as controlled animal models in which various factors that influence the severity of phenotype in OI can be evaluated.

Recent studies have also demonstrated reduced mechanical properties in oim/oim femora compared with wild-type (+/+), and intermediate values for oim/+ bones,(28,30) results consistent with the phenotype of the mice. Although differences in some geometrical properties of the bone were found, differences in material properties were not investigated. It is likely that in addition to the presence of mutant homotrimer collagen, other qualitative and quantitative abnormalities in the collagen and/or mineral phase contribute to the reduced bone mechanical properties in these mice. The current study seeks to investigate the effect of altered geometry, collagen, and mineral characteristics on bone properties in the oim/+ and oim/oim mice, and thereby elucidate important determinants of the severity of bone fragility in OI.

MATERIALS AND METHODS

Animals: Breeding

All animal procedures were performed under an IACUC-approved protocol. Heterozygous oim/+ breeder mice obtained from the Jackson Laboratory (Bar Harbor, ME, U.S.A.) produced average litter sizes of 6–10 pups. It was not possible to distinguish between the oim/+ or homozygous unaffected (+/+) mice prior to genetic analyses, but homozygous oim/oim mice were initially identified by their smaller body size, limb deformities, and skeletal fractures. Since the survival rate of the oim/oim mice is low (50%), additional oim/oimoim/+ breeder pairs were set up to obtain an increased number of oim/oim mice. Mice were kept at 75°F and received Purina rodent chow 5001 (Purina, St. Louis, MO, U.S.A.) and tap water ad libitum. Tissues from 32 mice were analyzed for this study (n = 9 +/+, n = 10 oim/+, n = 13 oim/oim) with data from males and females combined (based on no significant differences in geometry; data not shown).

Animals: Genotype

Pups were weaned at 4 weeks of age and 1-cm-long tail snips were obtained under anesthesia for genotype determination. Genomic DNA was extracted from mouse tail-snip samples using a DNA isolation kit (Invitrogen, San Diego, CA, U.S.A.) and amplified by the polymerase chain reaction technique using specific primers.(29) Polymerase chain reaction products were then analyzed on an ABI Prism 377 automated DNA sequencer (Perkin Elmer, Foster City, CA, U.S.A.).

Tissue preparation

Mice, 11–13 months of age, were sacrificed by methoxyflurane inhalation and terminal exsanguination via cardiac puncture, and body was weight recorded. Vertebrae and long bones (femora, tibiae, and humerii) were dissected, cleaned of soft tissue, and stored in gauze soaked in Tris buffer (0.05 M, pH 7.4) at −20°C until use. High-resolution radiographs of individual femora in the anterior–posterior (AP) and medial–lateral (ML) planes were obtained using a Faxitron machine (Hewlett Packard, Palo Alto, CA, U.S.A.). The femora were utilized for torsion testing and geometrical analysis, the tibia for mineral analyses (ash weight, Ca:P ratio, and Fourier transform infrared [FTIR] analyses), the humerii for collagen analysis, and the vertebrae for mineral characterization by X-ray diffraction.

Geometrical analysis

Femur length and endosteal and periosteal diameters in both the AP and ML planes were measured. Femur length was determined as the distance from the tip of the femoral head to the base of the condyles. The midpoint of the femur was used to measure endosteal and periosteal diameters. Based on the diameteral measurements in the AP and ML planes, the femoral cross-section was estimated to be elliptical. The moment of area, K, a geometrical property that governs torsional behavior, was estimated based on an elliptical model and calculated as follows(31) :

equation image

where 2Dp = periosteal diameter in the AP plane, 2 De = endosteal diameter in the AP plane, 2dp = periosteal diameter in the ML plane, and q = De/Dp.

Mechanical testing

One femur from each mouse was utilized for mechanical testing. For the oim/oim mice, some of the femurs had severe deformities or fractures and were therefore not appropriate for testing. In this case, the contralateral femur was utilized for testing. For the oim/+ and +/+ mice, the femur to be tested was randomly chosen. In preparation for torsion testing, the distal and proximal ends of the femora were potted with epoxy into square molds 4 mm deep with the diaphyses aligned perpendicular to the molds. While being potted and tested, femora were kept moist with gauze soaked in 0.05 M, pH 7.4, Tris buffer. Torsion testing was conducted at room temperature using a Burstein-Frankel torsion tester(32) with a torque cell of 50 in-oz. The torsion tester employs a dead weight pendulum loading system in which torsional failure occurs in < 0.1 s. Femora were tested in external rotation in random order. Torque and angular deformation were recorded on a digital oscilloscope, and the data were transferred to a personal computer for analysis. Failure torque (N-mm) and angular deflection at failure normalized for exposed diaphyseal length (rad/mm) were determined for each specimen. Torsional stiffness (N-mm/ (rad/mm)) was calculated by regression over the linear region of the curve which was defined by a correlation coefficient (R2) > 0.95.

Mineral and collagen analysis

In preparation for mineral (ash weight, Ca:P, FTIR) and collagen (hydroxyproline and amino acid) analyses, the diaphyses from the left and right tibiae and humerii were removed and cleaned of bone marrow. Since X-ray diffraction analysis requires a relatively larger quantity of bone (at least 10 mg), whole vertebrae (thoracic and lumbar, cleaned of soft tissue) were utilized for this assay. All bones were frozen at −70°C, lyophilized, and ground in a liquid nitrogen–cooled freezer mill (Spex Industries, Metuchen, NJ, U.S.A.). Some of the tissues were very small and unfortunately were not recoverable after the grinding process. Therefore, a reduced number of samples are reported for some of the analyses.

Aliquots from each ground tibia were analyzed by FTIR as KBr pellets (2 mg sample:200 mg KBr) using a Mattson Cygnus 25 Infrared Spectrometer (Mattson Instruments, Madison, WI, U.S.A.). Routinely, 256 interferograms were collected at 4 cm−1 resolution, co-added, and the resultant interferogram Fourier transformed. Data analysis was performed using Grams/32 software (Galactic Industries, Salem, NH, U.S.A.). The ratio of the area of the apatite phosphate ν13 absorbance from 900–1200 cm−1 to the area of the collagen amide I absorbance from 1590–1720 cm−1 was calculated to obtain the relative amounts of mineral and protein present (mineral:matrix ratio). The carbonate:mineral ratio was calculated as the ratio of the area of the carbonate ν2 absorbance from 850–890 cm−1 to the area of the phosphate ν13 absorbance. The relative quantities of carbonate incorporated into apatite at distinct sites were obtained from analysis of the intensity of the ν2 carbonate absorbance at specific frequencies.(33) The infrared frequencies assigned to the three types of carbonate are: type A, substituted for apatite hydroxyl, at 879 cm−1; type B, substituted for apatite phosphate, at ∼872 cm−1; and labile (nonlattice) positions at 867 cm−1.(34) The intensities at these positions were then normalized to the integrated area of the ν2 carbonate absorbance.

The remaining ground tibial diaphyseal bone was used to determine percentage mineral content based on ash weight. The samples (1–4 mg) were dried at 110°C, ashed at 600°C for 18 h, and the gravimetric yield of the ash determined in duplicate. Aliquots of the ash were then analyzed for calcium(35) and phosphate(36) to obtain a Ca:P ratio.

Wide-angle X-ray diffraction was performed on the ground vertebral bone using Cu-Kα radiation on a Siemens Powder Diffractometer (Iselin, NJ, U.S.A.). The line broadening measurement of the hydroxyapatite 002 reflection, β002, which is inversely related to crystal size and/or perfection in the crystallographic c-axis direction,(37) was repeated three times for each sample. Average crystal size was obtained from the Scherrer equation:

equation image

where D = crystal size (angstroms [Å]), λ = wavelength incident X-rays (1.54 Å), and θ = X-ray incident angle (25.85°).

Two assays were carried out to determine collagen content of the ground humerii. Hydroxyproline content per milligram of ground bone was determined colorimetrically,(38) with the assumption that the hydroxyproline measured arose solely from type I collagen, the primary protein component of ground bone. In addition, it was necessary to determine hydroxyproline content of collagen molecules for the three genotypes. This was done by high performance liquid chromatography amino acid analysis of the decalcified ground humerii (n = 3 samples/genotype) in the amino acid laboratory of Dr. Daniel Wellner, Cornell University Medical College. A hydroxyproline/collagen ratio (mg/mg) for each of the three genotypes was calculated based on an estimated molecular weight of collagen of 300 kDa.(39) These values, 0.10, 0.12, and 0.11, for the oim/oim, oim/+, and +/+ bones, respectively, were then utilized to determine the weight percentage of collagen in the ground bone. Thus, the values obtained from the colorimetric hydroxyproline analysis of ground bone, in micrograms of hydroxyproline per milligram of ground bone, were converted to a percentage of collagen per bone (mg/mg).

Statistical analysis

Means and SDs were calculated for each parameter measured. Analysis of variance was performed with the Student–Neuman–Keuls test followed by the Student's t-test to determine the probability of differences. Differences were considered significant at p ≤ 0.05. Due to the large SDs for some measurements, it was also appropriate to report values of p = 0.05–0.1 as “a tendency to be different.”

RESULTS

Geometrical analysis

The oim/oim values of body weight, femoral length, AP diameters, cortical thickness, and Kellipse were reduced compared with the majority of the +/+ and oim/+ values (Table 1). Even after normalization to body weight, Kellipse was still significantly smaller for the oim/oim compared with the +/+ bones. The AP endosteal diameter for the oim/+ femurs was significantly smaller than that of the +/+, resulting in a significantly greater cortical thickness of the oim/+ femurs.

Table Table 1.. Body Weights and Femur Geometrical Measurements for oim/oim, oim/+, and +/+ Mice
original image

Mechanical testing

The oim/oim values of failure torque and stiffness were significantly smaller than those of the oim/+ and the +/+ bones, and the oim/oim deflection was significantly less than that of the +/+ bones (Fig. 1). In addition, the oim/+ failure torque was reduced compared with that of the +/+. Although not all were significantly different, it is still interesting to note that for each of the mechanical parameters, the oim/+ value was intermediate to that of the oim/oim and +/+.

Figure FIG. 1..

Failure torque, angular deflection, and torsional stiffness for the oim/oim, oim/+, and +/+ femora tested in torsion. Values are presented as means ± SD. The oim/oim values of failure torque and stiffness were significantly reduced (p < 0.05) compared with those of the +/+ (**) and oim/+ (†) femora. The oim/+ failure torque was also significantly reduced compared with that of the +/+. The oim/oim deflection was significantly less than that of the +/+ femora.

Mineral analyses

Ash weight analysis revealed a small but significant decrease in mineral content in the oim/oim compared with the oim/+ bones (Table 2). The mean ash weight of the +/+ bones was not significantly different from that of the other two genotypes. FTIR spectra confirmed that the mineral present in all three genotypes was apatitic in nature. (Fig. 2A). However, difference spectra, obtained from subtraction of average, normalized oim/+ or oim/oim spectrum from average, normalized +/+ spectrum in the mineral phosphate region, revealed that the type of phosphate species present was not equivalent for the three genotypes (Fig. 2B). Compared with oim/+ and oim/oim mineral, the +/+ mineral contained a greater quantity of components that absorb at ∼1148, 980, and 953 cm−1, and a reduced quantity of components that absorb at ∼1096 and 1039 cm−1. Based on subtraction of the average, normalized oim/oim spectra from the oim/+ spectra, it was also apparent that the oim/+ mineral contained a larger quantity of components that absorb at ∼1096, 1072, and 1039 cm−1 compared with the oim/oim. Although not all of these components have been assigned to specific phosphate species, the 980, 1072, and 1148 cm−1 species have been positively correlated with increasing apatite crystallinity in the c-axis direction, the 1096 cm−1 with decreasing apatite crystallinity in the c-axis direction, and the 1039 cm−1 species with nonstoichiometric apatite.(40) The mineral:matrix values mirrored the ash weight measurements for oim/oim, oim/+, and +/+ (Table 2), but no significant differences were found. FTIR analysis of carbonate revealed a significantly decreased carbonate content (carbonate:mineral) in the oim/oim bones compared with the +/+ bones. No significant difference in type of carbonate substitution or Ca:P were found among the three genotypes (data not shown).

Figure FIG. 2..

(A) FTIR spectrum obtained from KBr pellets (2 mg sample: 200 mg KBr) of +/+, oim/+, and oim/oim ground tibial diaphyseal bone. The absorbances that arise from the collagen amide I, apatite phosphate, and apatite carbonate vibrations are noted. (B) Difference spectra obtained from subtraction of average, normalized oim/+, or oim/oim spectrum from average, normalized +/+ spectrum in the mineral phosphate region. Compared with oim/+ and oim/oim mineral, the +/+ mineral contains a greater quantity of components that absorb at 1148, 976, and 953 cm−1, and a reduced quantity of components that absorb at 1096 and 1039 cm−1.

Table Table 2.. Mineral Characteristics of Tibial Diaphyseal Bone for oim/oim, oim/+, and +/+ Mice
original image

X-ray diffraction analysis of the ground vertebral bone yielded insight into crystallinity changes (Fig. 3). The mineral in both the oim/oim and oim/+ bones was found to be significantly less crystalline in the c-axis direction compared with the +/+ bones.

Figure FIG. 3..

Crystallinity in the c-axis direction as determined by X-ray diffraction line broadening analysis and the Scherrer equation (37) for the oim/oim, oim/+, and +/+ ground, whole vertebra. Values are presented as means ± SD. The oim/oim and the oim/+ values were significantly reduced compared with the +/+ (**p < 0.05).

Collagen content

The collagen content of the oim/oim bones was significantly reduced compared with that of the +/+ bones (Fig. 4). The oim/+ collagen content tended to be less than that of the +/+ and was intermediate to the oim/oim and +/+ values.

Figure FIG. 4..

Mean collagen content (mg of collagen/mg of ground humeral diaphyseal bone) as determined by hydroxyproline assay of ground humerii. Values are presented as means ± SD. The oim/oim collagen content was significantly reduced compared with the +/+ (**p < 0.05), and the oim/+ collagen content tended to be reduced compared with the +/+ (*p = 0.06).

DISCUSSION

The precise molecular basis for the clinical heterogeneity observed in human OI has remained obscure. However, recent studies have begun to link specific types of mutations in the collagen molecule with severity of the disease.(1,10,41) It has been suggested that quantitative anomalies of type I collagen, i.e., reduced amount, lead to mild forms of the disease, and that severity is greatest when abnormalities in collagen structure (qualitative) and quantitative changes are present. This is supported by observations in the heterozygous Mov13 mouse, an animal model that produces ∼50% less type I collagen than normal, whose phenotype resembles a mild form of OI (Sillence type I).(42) The animals examined in the current study, the homozygous and heterozygous oim mice, oim/oim and oim/+, provide controlled models of moderate-to-severe and mild OI, respectively, in which the same collagen mutation is present, i.e., a qualitative anomaly, but in different quantities. Although other studies have previously demonstrated reduced bone mechanical properties in these mice,(28,30) the contribution of abnormalities in the bone mineral and matrix were not considered in detail. The current study sought to systematically examine quantitative and qualitative differences in collagen and mineral, the primary components of bone, in both the mild and severe model of OI and thereby obtain direct evidence of factors critical to bone fragility.

Changes in body size and bone geometry were prominent in the oim/oim mice. The oim/oim mice displayed smaller body size, femoral length, and Kellipse compared with the oim/+ and the +/+ mice. Furthermore, the oim/oim AP endosteal and periosteal diameters tended to be smaller than those of the +/+. These data, in conjunction with the significantly decreased Kellipse normalized to body weight, is suggestive of an abnormality that prevents compensation for external stresses by the bone. Although increased bone turnover has been implicated in human OI,(43) the data from the oim/oim femora in this study are more consistent with a mechanism that involves defective bone formation. The differences between parameters of the oim/+ and the +/+ mice were not as pronounced. However, the heterozygotes were intermediate in femoral length, AP diameter, and Kellipse compared with the other two genotypes, characteristics that parallel the severity of phenotype in these mice. The interesting exception to this pattern was the cortical thickness of the oim/+ femora which was significantly greater than both the oim/oim and +/+, primarily as a result of the decreased endosteal diameter. Similar to the oim/oim, this result is not consistent with increased bone turnover in these animals.

In agreement with other studies, the mechanical properties of the oim/oim mice bones were substantially reduced compared with the oim/+ and +/+. The oim/oim values of failure torque and stiffness were only 40% and 30%, respectively, of the +/+ values, and only 53% and 36% of the oim/+ values. To assess whether the reduced mechanical properties were attributed solely to differences in bone geometry or if there was a contribution from differences in material properties of the bone, normalization of structural properties to Kellipse calculated at the point of fracture was considered. However, two significant limitations of the current study precluded use of this type of analysis. First, since the estimate of Kellipse from two-dimensional radiographs is not exact, errors could be introduced into the calculation. Second, and perhaps more important, it was not possible to determine where the fractures initiated for most of the bones tested. It appeared that the majority of bones fractured close to the mid-diaphyseal region, but the exact initiation site could not be confirmed. Consequently, an alternative method was utilized to address the potential contribution of changes in material properties. Relative differences in the Kellipse values were compared with relative differences in structural properties for the three genotypes. For the oim/oim bones, the average Kellipse value was 65% of the +/+ and 70% of the oim/+ value, not sufficiently reduced to account for their dramatically diminished mechanical properties. Therefore, changes in bone material properties (mineral and collagen) must be considered as contributing factors.

In this study, both quantitative and qualitative abnormalities of mineral and collagen were considered. Despite reports of an osteopenic phenotype,(22) there were no significant differences (by ash weight determination or mineral:matrix) in the mineral content of the oim/oim and +/+ bones. Initially, this result seems contradictory in light of the well-established correlation between bone strength or stiffness and mineral content.(44,45) However, it has also been proposed that the correlation of increased modulus of elasticity of bone with increased mineral content is caused by the end-to-end fusion of apatite crystals.(46) Thus, it is possible that in the oim/oim bone, although there is a normal quantity of mineral, the defective quality of the mineral prohibits normal fusion of apatite crystals into a contiguous structure. In contrast to the similarity in mineral content between the oim/oim and +/+ bones, the mineral content of the oim/+ bones was slightly but significantly greater compared with that of the oim/oim bones. Extrapolating from an earlier study by Currey(45) that demonstrated a 2% increase in bone mineral ash content could result in as much as a 20% increase in ultimate strength and modulus of elasticity, it is possible that the larger mineral content did contribute to the 36–53% difference in bone mechanical properties between these two genotypes. However, the increased mineral content alone would not account for a difference of this magnitude, and it is probable that variations in quality of the mineral and matrix were contributing factors as well.

The mineral quality of the oim/oim and oim/+ bones was investigated by X-ray diffraction, FTIR, and Ca:P analysis. In agreement with recent small angle X-ray scattering and electron microscopic studies,(47,48) the X-ray crystallinity (size and perfection) of the oim/oim mineral, a biologic apatite, was significantly reduced compared with the +/+ apatite. This result is also consistent with X-ray and electron microscopic studies of ground cortical bone from OI patients that demonstrated a reduced apatite crystal size,(18,49) and a reduced quantity of normal lamellar bone with regions of small, unorganized apatite crystals not associated with collagen.(21) The data obtained from difference FTIR spectroscopy further demonstrated the differences in mineral quality among the three genotypes. Compared with oim/+ and oim/oim apatite, the +/+ apatite contained a greater quantity of phosphate components correlated with larger apatite c-axis dimensions (crystallinity), and a reduced quantity of components associated with nonstoichiometric (i.e., acid-phosphate containing) apatite. The differences between the oim/+ and the oim/oim mineral were more complex, with the oim/+ mineral containing phosphate species correlated with both larger and smaller crystals. Based on an earlier in vitro study that demonstrated increased carbonate content in larger apatite crystals,(40) the reduced carbonate:mineral of the oim/oim compared withthe +/+ bones was not surprising. This result was also in agreement with a recent infrared microscopic study.(26) In contrast, the reduced crystallinity of the oim/+ mineral compared with that of the +/+ was not accompanied by a decreased carbonate content. On that account, mineral quality was affected to a lesser extent in the oim/+ compared with the oim/oim bone, another finding consistent with the phenotype of these mice. Earlier animal studies have also linked changes in crystal size and/or carbonate content with abnormal bone mechanical properties,(50,51) but the specific mechanism was not known. The data from this study and from the previously discussed EM studies strongly suggest that noncontiguous apatite crystals, incapable of resisting external stresses, are the consequences of reduced crystallinity and abnormal molecular structure of the apatite.

In addition to the effect of mineral abnormalities on bone properties, changes in collagen have also been linked to reduced bone properties.(52-54) Most of these studies have addressed qualitative changes in collagen, such as structural abnormalities, as opposed to changes in collagen content. In the current study, collagen content was found to be reduced by 20% in the oim/oim compared with the +/+ bones. Since the reduced collagen content occurs in conjunction with a structural abnormality, it is not clear how the reduced content alone would affect the bone properties. However, two studies on mouse models of OI that exhibit reduced collagen content without a specific collagen structural abnormality found increased brittleness in the bones,(23,55) which is also consistent with the phenotype of the oim/oim mouse. Recent studies have investigated the mechanical(56) and structural (57) properties of oim/oim tendons, the primary component being abnormal homotrimer collagen. The ultimate stress and strain for oim/oim collagen were approximately half the values for control mice. According to that and an earlier study by Burstein et al.(44) a similar change in collagen in bone would likely result in reduced stiffness, similar to what was observed in the bones examined in this study. The X-ray diffraction study of oim/oim tendon(57) concluded that the total absence of α2(I) chains in the oim/oim mice resulted in a decrease in the order of fibril axial packing and a loss of crystalline lateral packing, structural changes that are the probable basis for the reduced mechanical properties of the oim/oim tendon. In the heterozygote tendons, both mutant and normal type I collagen molecules were present in the same copolymeric fibrils and crystalline lateral packing was prevented, but to a lesser degree. We would expect similar collagen characteristics in bone, and based on this predict that failure torque, deflection and stiffness of the oim/+ bones would fall midway between the other two genotypes, as was observed. However, the existence of an abnormality in type I collagen would also confer a change to the inherent nature of the collagen-mineral composite that comprises bone, and thus essentially prohibits separation of the effects of either abnormal collagen or mineral alone on bone fragility.

An interesting observation throughout the current study is that for the majority of the parameters studied, i.e., geometrical, mechanical, and material bone properties, the extent of abnormality paralleled the phenotypes of the mice. This suggests that some of the material properties are indicative and even predictive of bone fragility, and may be useful targets for therapeutic intervention. Comparison of bone characteristics for the three genotypes revealed the major differences are in collagen content and mineral quality. Despite the fact that the mineral crystallinity, a measure of crystal size and/or perfection, appears to be related to severity of phenotype, therapy aimed at increasing mineral crystallinity would probably not be effective. It is likely that the reduced crystallinity is a direct consequence of the presence of an abnormal collagen scaffold that cannot support normal crystal growth, and a therapeutic agent directed at increasing crystallinity would not address this fundamental problem. Earlier studies that investigated the use of fluoride in OI,(58) a compound known to increase apatite crystallinity, were not successful. Recent studies have begun to investigate the use of bisphosphonates in OI,(43,59-63) compounds that inhibit bone resorption but do not necessarily effect mineral crystallinity. Although preliminary results have demonstrated increased bone density, the precise mechanism by which this may reduce bone fragility in OI is not clear at this time.

In conclusion, the results from the current and earlier studies suggest a model in which bone fragility in the oim mice is directly related to the quantity of normal collagen present, and the observed mineral changes are reflective of the severity of the collagen abnormality. We therefore predict that therapy aimed at increasing the quantity of normal collagen would serve to improve bone mineral quality, geometry, mechanical properties, and ultimately reduce fragility in this mouse model, and by analogy in human OI.

Acknowledgements

The authors are grateful for helpful discussions with Dr. Timothy Wright, Dr. Margaret Peterson, and Dr. Robert Blank, and appreciate the technical assistance of Dan Weiland, Brad Schenkel, Darlene Grillo, and the HSS Molecular Biology Core Facility. This work was supported by The Children's Brittle Bone Foundation 007F9698 (N.P.C.) and the National Institutes of Health R29 DE-11803 (N.P.C.).

Ancillary