Morphological and mechanical characterization of bone phenotypes in the Amish G610C murine model of osteogenesis imperfecta

Osteogenesis imperfecta (OI) is a hereditary bone disease where gene mutations affect Type I collagen formation resulting in osteopenia and increased fracture risk. There are several established mouse models of OI, but some are severe and result in spontaneous fractures or early animal death. The Amish Col1a2G610C/+ (G610C) mouse model is a newer, moderate OI model that is currently being used in a variety of intervention studies, with differing background strains, sexes, ages, and bone endpoints. This study is a comprehensive mechanical and architectural characterization of bone in G610C mice bred on a C57BL/6 inbred strain and will provide a baseline for future treatment studies. Male and female wild-type (WT) and G610C mice were euthanized at 10 and 16 weeks (n = 13–16). Harvested tibiae, femora, and L4 vertebrae were scanned via micro-computed tomography and analyzed for cortical and trabecular architectural properties. Femora and tibiae were then mechanically tested to failure. G610C mice had less bone but more highly mineralized cortical and trabecular tissue than their sex- and age-matched WT counterparts, with cortical cross-sectional area, thickness, and mineral density, and trabecular bone volume, mineral density, spacing, and number all differing significantly as a function of genotype (2 Way ANOVA with main effects of sex and genotype at each age). In addition, mechanical yield force, ultimate force, displacement, strain, and toughness were all significantly lower in G610C vs. WT, highlighting a brittle phenotype. This characterization demonstrates that despite being a moderate OI model, the Amish G610C mouse model maintains a distinctly brittle phenotype and is well-suited for use in future intervention studies.


Introduction
Osteogenesis imperfecta (OI) is a hereditary bone disease in which gene mutations affect the formation of Type 1 collagen, leading to weak and brittle bones which can cause severe skeletal deformity and increased fracture risk. OI presentation can vary in severity and is traditionally classified as: Type I, non-deforming (most common); Type II, perinatally lethal; Type III, Committee (Protocol #SC296R). Mice used in this study were bred in house and maintained on a C57BL/6J background strain (Jackson Laboratory, Bar Harbor, ME). Male Col1a2 G610C/+ (G610C) were housed with pairs of wildtype (WT) females. To genotype WT and G610C offspring, DNA was extracted from ear notches by first incubating each tissue sample in 50 uL of lysis solution (25 mM NaOH/0.2 mM EDTA) at 90˚C for 1 hour, then neutralizing with 50 uL of neutralization buffer (40 mM Tris HCl). Target genes were amplified with PCR using primers for the G610C forward gene (5'-TCC CTG CTT GCC CTA GTC CCA AAG ATC CTT-3') and reverse gene (5'-AAG GTA TAG ATC AGA CAG CTG GCA CAT CCA-3'), then run on a 10% agarose gel and imaged with a G8100 E-Gel™ Power Snap Electrophoresis Device (Invitrogen, Carlsbad CA). Mice from each sex and genotype (n = 15-16/group) were euthanized at 10 and 16 weeks of age. Femora, tibiae, and the fourth lumbar spine vertebrae (L4) were harvested, stripped of soft tissue, and frozen in saline-soaked gauze at -20˚C until needed. Some bones were accidentally fractured during harvesting, so testing groups varied by bone (n = [13][14][15], as detailed in each table.

Microcomputed tomography (μCT) & architectural analysis
To analyze cortical and trabecular architecture, bones were scanned using an isotropic voxel size of 10 μm (Skyscan 1172, Bruker). The right femur, right tibia, and L4 vertebrae from each mouse were scanned through a 0.5 mm Al filter (V = 60kV, I = 167μA) with a 0.7-degree angle increment and two frames averaged. Images were reconstructed (nRecon) and rotated (Data Viewer) before calibrating to hydroxyapatite-mimicking phantoms (0.25 and 0.75 g/cm3 Ca-HA). A 1 mm trabecular region of interest (ROI) was selected at the distal metaphysis of femora and extended proximally from the most proximal portion of the growth plate, and at the proximal metaphysis of the tibia and extending distally from the most distal portion of the growth plate. Trabecular bone in the L4 vertebrae was examined over the entire length of the bone. Cancellous architecture in each ROI was quantified using CT Analyzer (CTAn). A 0.1 mm cortical ROI was selected at approximately 50% length of the femur or tibia, then analyzed with a custom MATLAB (MathWorks, Inc. Natick, MA) program [26]. For visualization, cross-sectional renderings of representative tibia, femora, and vertebrae were created with Drishti Volume Exploration tool [27].

Mechanical testing
Each femur was tested to failure in three-point bending (support span at 7.5 mm for 16 week bones; 6.5 mm for 10 week bones), with the posterior surface in tension. Right tibiae were tested to failure in four-point bending (lower span at 9mm; upper span at 3mm), with the medial surface in tension. Bones were loaded at a displacement control rate of 0.025 mm/s while the sample remained hydrated with PBS. Cross-sectional cortical properties at the fracture site were obtained from μCT images as described above. These properties were used to map load-displacement data into stress-strain data using standard engineering equations as previously reported to estimate tissue level properties [28][29][30].

Statistical analysis
At each age, data were statistically analyzed using Two-Way ANOVA for main effects of sex and genotype. If a significant interaction (sex X genotype) occurred, the ANOVA was followed by a Tukey post-hoc test. Analysis was performed using GraphPad Prism (v.8) with a significance level at α = 0.05. Table 1 shows that G610C mice and their bones were consistently smaller than WT mice, with main effects of genotype for body weight and long bone length at both ages. In contrast, sexbased differences between bone lengths lessened as the mice reached skeletal maturity, although body weight differences between sexes persisted.

Cortical bone phenotype in femora and tibiae
G610C mice had less, more highly mineralized cortical bone than their sex-and age-matched WT counterparts. With the exception of cortical thickness, there was a main effect of genotype (p � 0.0062 for all) for every cortical measure in both male and female mice, at both age points, and in both femora and tibiae (Tables 2 and 3). G610C bones were smaller overall as compared to WT bones (Fig 1, S1 and S2 Figs), highlighted by smaller total area, bone area, maximum moment of inertia (Imax), and minimum moment of inertia (Imin) as compared to WT mice. Tissue mineral density (TMD) was higher in G610C bones versus WT. There was also a significant effect of sex (p � 0.0215) for almost every property investigated, except for 16-week tibial TMD and femoral cortical thickness, as the male mice tended to be larger than the female mice. There were significant interactions in only two measures: 10-week tibial marrow area (p = 0.0054) and 10-week femoral marrow area (p = 0.0466), with post-hoc analyses showing the same behavior of WT mice having larger marrow area than G610C mice.

Trabecular phenotype
G610C mice had less trabecular bone than their sex-and age-matched WT counterparts. There was a significant main effect (p � 0.0362) of genotype for nearly every trabecular measure investigated. (Tables 4 and 5). OI mice had sparse, small trabeculae (Fig 2, S3-S5 Figs), as evidenced by lower bone volume fraction (BV/TV), trabecular thickness (Tb.Th), and trabecular number (Tb.N), and higher trabecular spacing (Tb.Sp), as compared to WT mice. In contrast to the cortical phenotype, the TMD phenotype in trabecular bone flipped between 10 and 16-week mice. At 10 weeks, female G610C had lower TMD than WT mice but by 16 weeks of age, G610C trabecular TMD was higher than WT in both sexes. As with cortical bone, there was a significant (p � 0.0233) main effect of sex for most trabecular properties, with male mice generally having denser, more mineralized trabecular bone than female mice. There was a  genotypes than male mice. Similar results were also seen in trabecular bone of L4 vertebrae (S1 Table).

Mechanical properties
Bending tests demonstrated that the G610C bones exhibit brittle behavior in comparison to WT bones, a phenotype that was consistent in both bones and age groups (Figs 3 and 4). There was a main effect (p � 0.0309) of genotype for most structural mechanical properties. The smaller G610C bones sustained a lower yield force (only in tibiae), ultimate force, displacement to yield (only in femora), postyield displacement, total displacement, stiffness, and work, as compared to WT bones. However, trends in estimated tissue-level properties differed. There was a significant (p � 0.0263) main effect of genotype for total strain and toughness in the tibiae, while a main effect of genotype was seen for all properties investigated in the femur except for 10-week strain and work to yield. There were significant (p � 0.0489) main effects of sex for many mechanical properties as well, including yield force, ultimate force, work to yield, ultimate stress, and strain to yield, with males consistently having higher properties than females (Tables 6 and 7). There was a single interaction effect in 16-week femoral toughness (p = 0.044), but there were no specific post-hoc differences in comparisons of interest.

Discussion
It is well established that fracture risk is a function of both bone size and tissue quality. Intuitively, small bones generally fracture at lower loads than larger bones. A hallmark of OI bone is brittle mechanical behavior, reflected in reduced post-yield deformation and toughness. These properties have less to do with bone size and strength, and more to do with quality and deformability of the tissue. OI is characterized by bones that are not only small, but also suffer from inferior tissue quality due to mutant collagen being incorporated into the bone matrix thereby inhibiting normal cross-linking, mineralization, and hydration [31]. A representative model of this disease should not only have smaller bones with reduced structural mechanical properties, but also tissue-level mechanical inferiorities. The G610C model of OI is well suited for investigating bone mechanical deficits and potential interventions. Several other OI models exist, but most suffer from limitations that make them less suitable for studies looking at mechanical stimulation (as bones may break in vivo),    or where mechanical endpoints are planned (because of sample loss due to spontaneous fracture and the need for large sample sizes to overcome variability in mechanical analyses). The popular (and commercially available) homozygous oim model exhibits a severe phenotype, leading to mice often sustaining spontaneous long bone fractures during life and an increased risk of specimen loss during tissue harvesting and handling [32]. While mechanical properties are markedly deficient, even more so than G610C vs. WT, this combination of poor health and brittleness makes it difficult to properly power loading intervention studies. The Brtl IV model, another common OI model, has a moderate bone phenotype and mechanical analyses have been successfully performed with minimal specimen loss [33]. However, these mice can be difficult to acquire as they are not commercially available. Although additional models exist (e.g. CRTAP [34], Swaying [35], Jrt [36], Col1a1 ±365 [37]), the G610C model may be preferable as it is commercially available, is reflective of a mutation found in humans, and is severe enough to be relevant but moderate enough to be practical. Results from the current study show that G610C mice bred on a B6 background have small, highly mineralized bones with lower structural strength and decreased deformability. Tissuelevel analyses demonstrate that this skeletal weakness is not only attributable to the smaller body and bone sizes of the diseased mice, but also to inferior tissue quality. G610C bones exhibited brittle behavior, fracturing at lower displacements and strains than their WT counterparts (Figs 3 and 4), despite showing largely similar tissue-level strength versus WT bones (as measured by ultimate stress). This finding was true for both male and female mice, and at ages reflecting adolescence through skeletal maturity. Brittle behavior is expected of an OI phenotype due to a lack of normal collagen and subsequent aberrant mineralization which can manifest as increased or decreased tissue mineral density, demonstrating that the observed structural weakness is caused, in part, by inferior tissue quality in addition to reduced bone quantity. G610C tissue consistently showed higher mineralization than WT tissue (as measured by TMD which is reflective of the mineralization of the tissue itself without the confounding factor of bone mass), a trend which has been seen before in G610C [6,12], as well as oim mice [38]. Importantly, no spontaneous in vivo fractures were observed in this study.
Together, these findings demonstrate that this model is a faithful representation of mild to moderate OI and a solid test bed for intervention studies aimed at mitigating defects in quantity and quality.
The results presented here agree with and expand upon previously published data of the G610C mechanical phenotype, despite the use of diverse methodologies. Jeong et al. performed torsion tests on femora from 16 wk male and female femora and saw the same general trend of G610C bones having lower ultimate strength and performing lower total work than WT [25]. Mertz et al. performed 4-pt bending tests on 16 wk male femora and saw a significant decrease in total work, post-yield work, and post-yield displacement, and significant increase in displacement to yield [17]. Omosule et al. and Bi et al. both tested 16 wk femora from female mice in 3-pt bending, and saw a significant decrease in ultimate force, post-yield work, total work, and post-yield displacement [21,23], as well as ultimate stress and modulus [21]. Omusule et al. also analyzed femora from male mice and saw higher ultimate force values in G610C than in WT, however this is inconsistent with our results and other published data. Bateman et al. tested femora and tibiae from 8 wk male mice in 4-pt bending, and saw similar results to the 10 wk data reported here, with ultimate force, work to yield, post-yield work, and stiffness being or trending significantly lower in G610C mice in both bones, with the exception of a reverse trend for tibial work-to-yield (inconsistent with our results) [18]. Lee et al. performed 4-pt bending with tibiae from 15 wk female mice, and also reported a significant decrease in ultimate force and work to yield [20]. Finally, Little et al. performed 4-pt bending on tibiae from 9 wk female mice and saw a significant decrease in ultimate force and post-yield work [19]. In summary, ultimate force and post-yield work are consistently lower in G610C mice than in WT, regardless of age or sex. Stiffness and work to yield either do not change between genotypes or are slightly reduced in G610C, with femora and tibiae behaving slightly differently. Importantly, no study to date has looked at the mechanical behavior of G610C in both hindlimb bones, both sexes, or two different age points.
Mechanical studies of G610C have been performed with mice bred on a variety of background strains other than C57BL/6, but since phenotype severity is strongly dependent on background strain, as shown by Daley et al. [6], results from these studies cannot be directly compared with the data presented here. As has been briefly summarized, previous studies with mice bred on a B6 background have only reported basic structural mechanical properties (with the exception of Bi et al. [21]), and are often statistically limited by smaller sample sizes (e.g. n = 5). By estimating tissue-level defects in addition to structural deficiencies and increasing statistical power by using larger groups, the characterization reported here provides an expanded analysis of the underlying causes of skeletal weakness and brittleness in the G610C model of OI.
This study is limited by its scope, which focused on the architecture and mechanics of commonly analyzed long bones (hindlimbs) at two important developmental age points. It is not known how the phenotype might evolve as these mice continue to age, nor if there are similar phenotypes in other skeletal and non-skeletal locations. The impacts of cellular processes were also not examined, nor was consideration made for other measures of tissue quality, including characterization of tissue composition (beyond TMD) or collagen structure. These experiments were beyond the scope of the current investigation but will be considered in future work.
In conclusion, the Amish Col1a2 G610C/+ model of OI exhibits a distinctly weak, brittle phenotype that similar in both males and female mice, and at two distinct and important developmental ages. Importantly, no spontaneous fractures were noted in any bones harvested in the current study. This characterization demonstrates that despite being a moderate OI model, the Amish G610C mouse model bred on a B6 background strain maintains a distinctly brittle phenotype and is well-suited for use in future intervention studies.
Supporting information S1 Table.