Intrafibrillar Mineralization Deficiency and Osteogenesis Imperfecta Mouse Bone Fragility

Abstract

Osteogenesis imperfecta (OI), a brittle bone disease, is known to result in severe bone fragility. However, its ultrastructural origins are still poorly understood. In this study, we hypothesized that deficient intrafibrillar mineralization is a key contributor to the OI induced bone brittleness. To test this hypothesis, we explored the mechanical and ultrastructural changes in OI bone using the osteogenesis imperfecta murine (oim) model. Synchrotron X-ray scattering experiments indicated that oim bone had much less intrafibrillar mineralization than wild type bone, thus verifying that the loss of mineral crystals indeed primarily occurred in the intrafibrillar space of oim bone. It was also found that the mineral crystals were organized from preferentially in longitudinal axis in wild type bone to more randomly in oim bone. Moreover, it revealed that the deformation of mineral crystals was more coordinated with collagen fibrils in wild type than in oim bone, suggesting that the load transfer deteriorated between the two phases in oim bone. The micropillar test revealed that the compression work to fracture of oim bone (8.2±0.9 MJ/m³) was significantly smaller (p<0.05) than that of wild type bone (13.9±2.7 MJ/m³), while the bone strength was not statistically different (p>0.05) between the two genotype groups. In contrast, the uniaxial tensile test showed that the ultimate strength of wild type bone (50±4.5 MPa) was significantly greater (p<0.05) than that of oim bone (38±5.3 MPa). Furthermore, the nanoscratch test showed that the toughness of oim bone was much less than that of wild type bone (6.6±2.2 GJ/m³ vs. 12.6±1.4 GJ/m³). Finally, in silico simulations using a finite element model of sub-lamellar bone confirmed the links between the reduced intrafibrillar mineralization and the observed changes in the mechanical behavior of OI bone. Taken together, these results provide important mechanistic insights into the underlying cause of poor mechanical quality of OI bone, thus pave the way toward future treatments of this brittle bone disease.

1. Introduction

Frequent bone fragility fractures emerge as a result of bone tissue losing its normal capacity needed to withstand routine physiological loads (Turner, 2002), which, in turn, results in poor quality of life of patients, accompanying large socio-economic burdens to the society (Palazzo et al., 2014; Yelin et al., 2016). One severe form of bone fragility manifests itself in Osteogenesis Imperfecta (OI) or the so-called brittle bone disease. OI is a genetic disorder which originates from abnormal collagen synthesis, but gives rise to a much wider range of complications at different hierarchies of bone (Dickson et al., 1975; Sillence et al., 1979). Due to structural and mechanical deteriorations in bone (Forlino and Marini, 2016), OI leads to frequent bone fractures and pathologic skeletal deformities (Rauch and Glorieux, 2004; Albert et al., 2017).

To date, several animal models have been developed to help discern the complex pathways leading to the brittleness of OI bone and to assess the efficacy of clinical therapies in treating the disease (Stacey et al., 1988; Bonadio et al., 1990; Rao et al., 2008; Enderli et al., 2016). Among them, osteogenesis imperfecta murine (oim) model was originally introduced as a model with phenotypic and biochemical features reminiscent to that of observed in human patients with mild to severe OI (Chipman et al., 1993). Since then, this popular model has been extensively exploited in the OI-related research (Bonadio et al., 1993; Landis, 1995; Saban et al., 1996; Misof et al., 1997; Miller et al., 2007; Cundy, 2012; Coleman et al., 2012; Lim et al., 2017).

In oim model, a mutation by G deletion at nucleotide 3983 of the Cola-2 gene results in homotrimeric collagen synthesis absent in pro-alpha2(I) chain, as opposed to normal heterotrimeric collagen with two pro-alpha1(I) chains and one pro-alpha2(I) chain. The homozygous oim/oim animal mimics the severe type of OI exhibiting frequent fractures, progressive skeletal deformities, cortical thinning, osteopenia, and reduced body weight (Chipman et al., 1993). The heterozygous oim/+ animal contains both types of homotrimeric and heterotrimeric collagen fibrils in bone making it suitable for studying the mild type of OI in human (McBride et al., 1997).

Previous studies have shown that OI alters the biomechanical response and architecture of bone across different hierarchies. First, it has been reported that OI results in reduced elastic modulus, ultimate strength (Gautieri et al., 2008, 2009; Chang et al., 2012), and toughness of collagen fibrils (Misof et al., 1997; Carriero et al., 2014), often attributed to the structural changes in collagen molecules (Forlino et al., 1998; Chang et al., 2012) and collagen fibrils (Bonadio et al., 1990; McBride et al., 1997; Camacho et al., 1999; Miles et al., 2002; Sims et al., 2003; Gautieri et al., 2009; Wallace et al., 2011; Coleman et al., 2012; Carriero et al., 2014; Li et al., 2016; Kłosowski et al., 2017). In addition, OI collagen fibril has been shown to suffer from a reduced attraction force between collagen helices causing weakened intermolecular adhesion (Kuznetsova et al., 2001) partially related to containing more bound water with a lower ability to absorb unbound water (Andriotis et al., 2015). These changes were described to accompany negative mechanical consequences for the collagen phase in oim bone, in one way or another.

Second, OI has been found to also inflict impacts to the mineral phase in bone. Early studies showed that the mineral phase in oim bone had reduced thickness, crystallinity (Fratzl et al., 1996; Camacho et al., 1999; Grabner et al., 2001; Rodriguez-Florez et al., 2015), and spatial alignment (Grabner et al., 2001). It was also reported that oim bone had altered chemical composition (Phillips et al., 2000) such as reduced carbonate content (Camacho et al., 1999) and increased calcium content (Grabner et al., 2001). Moreover, the mineral crystallites in oim bone were found to be more tightly packed exhibiting a disorganized manner (Vanleene et al., 2012) having enhanced extrafibrillar mineralization (Grabner et al., 2001). Since the mineral phase is a major contributor to mechanical competence of bone, these negative alterations would inevitably work against the remarkable mechanical properties of the otherwise normal (healthy) bone.

Third, the ultrastructure of oim bone, that is the intricate nanocomposite of intertwined mineral and collagen phases (Maghsoudi Ganjeh et al., 2020), is observed to markedly differ from that of healthy bone. The ultrastructure of oim bone was described as a malformed material of loosely packed, randomly distributed collagen fibrils, often bent or twisted, devoid of the well-known banding pattern (Nadiarnykh et al., 2007; Kłosowski et al., 2017) and discontinuous mineral crystals with largely dissimilar size, shape, location, alignment and orientation (Landis, 1995). This altered ultrastructure unavoidably cascades to further downstream changes in mechanical properties and remodeling of the tissue at higher hierarchies beyond the nanoscale.

Finally, the ramifications of OI is reported to exceedingly propagate to larger length scales at tissue and whole bone levels. Research has shown that OI bone demonstrates poor tissue mechanical properties such as reduced modulus, ultimate strength, and toughness (Bonadio et al., 1990; Camacho et al., 1999; Carriero et al., 2014; Dong et al., 2010; Islam et al., 2012; Jepsen et al., 1996; Miller et al., 2007; Misof et al., 1997; Saban et al., 1996). Bone mineral content and density have been reported to be decreased in oim bone (Phillips et al., 2000). At the whole bone scale, reductions in length, moment of inertia, cortical thickness, porosity, trabecular number and thickness have been observed in oim bone (McBride et al., 1998; Camacho et al., 1999; Coleman et al., 2012; Yao et al., 2013). Therefore, changes originally initiating at the collagen molecules level in OI would translate all the way up to the whole organ level.

Although a growing body of experimental results mount on oim bone brittleness, most of them are either phenomenological or correlative, lacking rigorous mechanistic explanations for oim bone brittleness, particularly at ultrastructural levels. To fill in this knowledge gap, we hypothesized in this study that reduction of intrafibrillar mineralization is a key player in OI bone brittleness. To test the hypothesis, we used several advanced techniques, including synchrotron X-ray scattering, focused ion beam (FIB) created micropillar compression test, high resolution scanning electron microscopy (SEM) imaging, and nanoscratch test, to investigate differences in the mechanical and structural properties between oim and wild type bone at microstructural and ultrastructural levels. Furthermore, cohesive finite element simulations were conducted to verify the hypothesis in silico. Overall, the results obtained would help better understand the OI bone disease at the ultrastructural level.

2. Material and methods

2.1. Specimen preparation

The OI murine (oim) model used in this study was not created through the process of genetic engineering. Due to the similar phenotypes, however, the homozygous oim mice (oim/oim) have been used as a model for OI III in humans (Chipman et al., 1993). In addition, the heterozygous oim mouse (oim/+) has been used as a model of OI I in humans (Saban et al., 1996). Bone specimens were prepared from freshly frozen femurs and tibias of seventeen wild type and oim/oim mice for micropillar compression test, uniaxial tensile test concurrently with synchrotron X-ray scattering measurements, and nanoscratch test.

For micropillar compression test, tibias from three mice of each genotype group were cleaned of soft tissue, embedded in PELCO epoxy resin, and sectioned transversely at the mid-diaphysis with a low-speed precision saw (PICO 155, PACE Technologies) under continuous irrigation of deionized water. The cross-section of the specimens was ground and polished with successive grades of sandpaper (from P800 to P4000) and alumina suspension (particle size: 1.0, 0.3, and 0.05 μm). Specimens were generously cleaned with distilled water in an ultrasound bath between each grinding or polishing steps and were left air-dried for 48 hrs. Then, they were coated with a 30-nm thick Au film using a sputter coater (Cressington 108 Auto) employed to alleviate charging drifts caused by ion or electron beams of the FIB workstation (Zeiss Cross Beam 340) used for fabrication of micropillars. Specimens were attached to carbon-taped aluminum specimen holders with their lateral surfaces painted with conductive graphite (Ted Pella DAG-T-502). The gallium ion source of the FIB machine was operated at acceleration voltage of 30 kV with a 3.0 nA probe selected for initial coarse milling of micropillars of about 6.0 μm in diameter. The coarse milled micropillars were subsequently milled down to 4.0 μm using a 500-pA probe, and finally to 3.0 μm using a fine 100-pA probe. These sequential steps were carried out to minimize the undesirable ion beam radiation damage to the bone specimens. The height of final micropillars was around 9 pm in average. From the wild type and oim/oim mice samples, fourteen micropillars oriented in axial direction of long bones were fabricated in posterior quadrants of mid-diaphysis sections from the tibias of the two genotype groups.

For the uniaxial tensile test concurrently with synchronized X-ray measurements, eight femur specimens from each genotype group were CNC machined and polished to reach the gauge dimensions of approximately 4.0 mm, 2.0 mm, and 0.2 mm in length, width, and thickness, respectively. To ensure reliable gripping of test specimens, both ends of the specimens were embedded in a plastic resin holder.

In the nanoscratch test, specimens from six femurs of two genotype groups were prepared. Each femur was freshly embedded in the plastic resin (PMMA) without going through dehydration process and the cross section of mid-diaphysis of the femur was sliced, polished, and mounted on a custom designed specimen holder (Islam et al., 2012). To ensure wet condition and tissue viability, a hole was drilled laterally to the embedded specimen to allow PBS solution fill into the intra-medullar cavity and maintain the bone specimens hydrated throughout the testing process.

2.2. Tensile test with concurrent synchrotron X-ray scattering measurements

The tensile tests were conducted concurrently with X-ray scattering measurements at the 1-ID beam line, Advanced Photon Source, Argonne National Laboratory (ANL), Argonne IL. An MTS servo hydraulic load frame (MTS 858) was used for loading the specimens in tension. A force transducer attached to the load frame was used to record the load needed to calculate the stress applied to the specimens. Specimens were kept hydrated by continuously dripping phosphate buffered saline (PBS) solution onto them. Since the mechanical test frame could not provide sufficiently low loading rates, load-displacement data were collected only at a very limited number of points and only ultimate tensile strength was estimated for specimens. Concurrently, a monochromatic X-ray beam with a cross section of 100 μm × 100 μm was incident at specimen’s center. Four GE area detector panels were used to record WAXS diffraction patterns. The beam then passed a center hole to simultaneously collect SAXS signals with a Bruker 6500 CCD placed further downstream (Samuel et al., 2016). The amount of intrafibrillar mineral crystals and internal strain of both mineral and collagen phases in the tensile tests were measured using both small and wide-angle scattering signals (WAXS and SAXS) obtained from the synchrotron X-ray scattering measurements (Advanced Photos Source, ANL). For WAXS measurements, intensity values at each pixel in relative arbitrary units were considered to be roughly correlated with the amount of mineral crystallites participating in diffraction (Fratzl et al., 1992). The area detector captured volume average diffraction from the participating crystals leading to formation of Debye rings corresponding to different lattice planes on the area detector. The obtained diffraction images were divided into angular bins of 10° azimuthal (η) intervals and integrated to obtain line profiles at each azimuthal bin. The area under the diffraction peak observed in the normalized SAXS 2-D line profile along the longitudinal axis of bone was measured following standard SAXS data reduction techniques (Samuel et al., 2016). The integrated area under 3rd order harmonic peak, which is associated with 67.6 nm periodic spacing of mineral-filled gap zones within the collagen fibrils, was considered as an indicator of the intrafibrillar mineral content.

The preferential alignment or orientation distribution of mineral crystallites in bone was assessed based on the shape of SAXS contours for wild type and oim/oim bone specimens. In general, the more circular the contours become, the more random the crystals are arranged, indicating a lack of preferential alignment of mineral crystals in bone (Fratzl et al., 1992). SAXS images taken from different genotype groups were plotted using the SAXS image processing software IRENA (Ilavsky and Jemian, 2009).

In order to measure the internal strain of mineral and collagen phases at the ultrastructural level, the shift in the diffraction peaks from WAXS and SAXS measurements were used to quantify the in situ strain in mineral crystals and collagen fibrils, respectively, following the procedure reported by Samuel et al. (2016). Briefly, as the specimen was being deformed, the lattice spacing of mineral crystals and collagen fibrils in bone changed, thus causing peak shift in WAXS and SAXS profiles associated with the deformation, which was to approximate the in situ strain of mineral crystals and collagen fibrils in bone.

2.3. Micropillar compression tests

The micropillar compression tests were performed under dehydrated condition on Nano Indenter XP System (MTS) equipped with a flat-end diamond punch of ~10 μm in diameter (Micro Star Technologies, Huntsville, TX). Micropillars were first located by the built-in optical microscope of the nanoindentation apparatus, then the nanoindenter and micropillar specimen were brought to the near touching position where sufficient dwelling time was allowed in order to compensate for instrument drift. Once a stable contact was established, the flat-end indenter was lowered down at a 5.0 nm/s loading rate to slowly compress the micropillar beneath it. Although the nanoindentation machine used here was inherently a load-controlled instrument, it was operated in a displacement-control mode by augmenting the default operating procedure through employing the software-enabled feedback system of the apparatus helping to control the displacement instead. This technique was, however, effective up to the moment when ultimate stress point of the specimens achieved, after which further compression resulted in abrupt strain bursts. Therefore, the load and displacement data were frequently collected up to the moment when ultimate loading was achieved. A preload cut-off of 0.5 mN was applied to the original load-displacement data to correct for the initial toe region, a typical hallmark of the early mechanical engagement of diamond punch with the micropillar. Engineering stress was calculated by dividing the loading force by the cross-sectional area at top, and average strain was obtained by dividing punch displacement by the initial height of the micropillar, respectively. The sink-in of the base extracellular substrate supporting the micropillar was assumed to be negligible. All micropillars were imaged before (to measure initial dimensions) and after each test (to observe failure modes) using the high-resolution FIB/SEM microscope (Zeiss Cross Beam 340) operated at 5kV, with the specimen stage tilted by 50°.

2.4. Nanoscratch tests

In order to obtain a measure of in situ toughness of wild type and oim/oim bone, nanoscratch tests were performed following the procedure developed in our group in the past (Islam et al., 2012). Briefly, the Nano Indenter XP system furnished with a cube-corner diamond tip was used to scratch the surface of polished bone specimens mounted on the custom-designed holders used to ensure a hydrated condition. The built-in optical microscope was used to detect central (middle) cortex region of femoral mid-diaphysis sections for performing batches of nanoscratch tests. First, a fixed penetration load of 5 mN was applied to the indenter, then it was moved laterally to create nanoscratch paths of length 20 μm while the normal and lateral scratching forces were being recorded. The nanoscratch toughness was then quantified as the scratch work per unit volume of material removed by the scratch using the equation $u_t=\frac{F_t}{w\times d_r}$ as proposed in Wang et al., 2007, in which u_t is the in situ toughness of bone, F_t is scratch force, w and d_r are width and depth of the scratch groove, respectively. Several scratches were performed on each specimen using the batch mode, and then the measurements were averaged.

2.5. Finite element (FE) simulations

To further verify the hypothesis of this study, finite element (FE) simulations of compression test using sub-lamellar bone models (Fig. 1) with different degree of intrafibrillar mineral volume fraction (15%, 10%, and 5%) were performed using the FE package ABAQUS. The extrafibrillar mineralization was kept constant across the models to single out the effect of the intrafibrillar mineralization only. A recently proposed two-dimensional (2-D) ultrastructural model of bone (Maghsoudi-Ganjeh et al., 2019), which is representative of an ordered array of mineralized collagen fibrils and extrafibrillar matrix subunits, considered as basic building blocks of bone, was employed for simulations. The mineralized collagen fibril consisted of intrafibrillar mineral platelets embedded in the gap regions of collagen matrix following the well-known staggered pattern (Jäger and Fratzl, 2000). The extrafibrillar matrix was comprised of mineral nanocrystals bounded by a thin layer of non-collagenous proteins modeled using cohesive elements. Cohesive elements were also placed between intrafibrillar mineral platelets and collagen matrix to properly account for the mineral-collagen interactions (Maghsoudi-Ganjeh et al., 2020). The detailed description of model development (including determination of constituent material properties), verification, and simulation set-up are provided elsewhere (Maghsoudi-Ganjeh et al., 2019; Ahsan, 2017). The FE models were loaded in uniaxial compression mode to predict elastic modulus, ultimate compressive strength, and compression work to fracture (measured as the area under stress-strain curve). The models were loaded in tension to calculate strain rations between mineral/collagen and fibril/tissue.

Fig. 1:

FE model used to study the effect of intrafibrillar mineralization on mechanical properties of oim/oim bone. Model consisted of different units of mineralized collagen fibrils and extrafibrillar matrix considered as the building blocks. Extrafibrillar matrix minerals were bound by non-collagenous proteins modeled as cohesive elements. Intrafibrillar mineralization content was varied to cover 5%, 10%, and 15% overall intrafibrillar mineralization. The model was tested under compression to study failure modes, and under tension to calculate average mineral and collagen strain.

3. Results

3.1. SAXS results

The images of SAXS scattering (Fig. 2a & 2b) demonstrated a shift of patterns from peanut-shaped contours for wild type bone to near circular-shaped contours for the oim/oim bone. The circular contours indicated a random distribution of mineral crystals in oim/oim bone, suggesting that the degree of preferential alignment of mineral crystals in normal bone, which exhibited peanut-shaped SAXS scattering contours, was diminished in the oim/oim bone. In addition, the integrated intensity under the 3^rd order peak in the SAXS diffraction profile was found to be significantly lower in the oim/oim (p<0.05) compared to that of the wild type bone (Fig. 2c), indicating a pronounced decrease in intrafibrillar mineralization in the gap regions of the oim/oim collagen fibrils.

Fig. 2.

SAXS contour maps shown for (a) wild type and (b) oim/oim bone. The shape of contours shifted from peanut-shaped contours for wild type to near-circular contours for oim/oim bone, indicating a lack of preferential alignment of mineral crystals in oim/oim bone; (c) The integrated intensity defined as the area under the peak of 3^rd order SAXS diffraction peak was considered as a measure of the amount of intrafibrillar mineralization present in bone. Oim/oim bone showed significant reduction in intrafibrillar mineralization.

The linear regression analysis on the relationship between the mineral and collagen fibril strain based on the results of the uniaxial tensile tests revealed two major differences between oim/oim and wild type bone (Fig. 3). First, the strain ratio between the mineral crystals and collagen fibrils in wild type bone indicated a much stronger linear correlation (R²=0.89) than that of the oim/oim bone (R²=0.45). Second, the slope of the fitted line for wild type bone (slope=0.76) was two folds greater than that of oim/oim bone (slope=0.34). These results suggested that the load transfer mechanism between the collagen and mineral phases was significantly deteriorated in oim/oim bone.

Fig. 3.

The in situ mineral and fibril strain measured from x-ray scattering measurements for (a) wild type and (b) oim/oim bone. The fitted line was obtained from the linear regression model. The correlation between mineral and collagen strain was much weaker in oim/oim bone. In addition, the ratio between mineral to collagen strain was smaller in oim/oim bone. These results were indicative of compromised cooperative deformation mechanisms among bone constituents in oim/oim bone.

3.2. Micropillar compression tests

Stress-strain curves obtained from micropillar compression tests demonstrated that the wild type bone specimens in general experienced more inelastic deformation prior to failure compared to the oim/oim bones (Fig. 4). The elastic modulus and ultimate compressive strength were not statistically different between the oim/oim and wild type (p>0.05) (Table 1). Based on the post-test high-resolution SEM images shown in Fig. 5 wild type micropillars were found to fail by development of shear failure planes or localized plastic deformation on top regions of micropillars. In contrast, the dominant failure mode for oim/oim micropillars was a combination of cracking, axial splitting and shattering modes, typically observed in brittle materials. The compression work to fracture of oim/oim bone (8.2±0.9 MJ/m³) was significantly smaller than that of wild type bone (13.9±2.7 MJ/m³) (p < 0.05) (Fig. 6a). The ratio of failure strain to yield strain for oim/oim bone (1.12±0.005) was also significantly smaller (p<0.001) compared to that of wild type bone (1.60±0.026), suggesting less plastic deformation taking part in oim/oim bone (Fig. 6b).

Fig. 4.

Stress-strain response of wild type and oim/oim bone obtained from testing FIB-created micropillar samples under uniaxial compression loading. The curves represent the measurements taken up to the moment when peak force achieved after which strain bursts followed due to the abrupt failure of specimens, which, in turn, inhibited a smooth collection of load displacement data. Overall, wild type bones exhibited an appreciable amount of inelastic deformation prior to failure, whereas oim/oim bones largely failed in a brittle manner.

Table 1:

Materials properties (mean±std) obtained from micropillar compression, nanoscratch, and tissue tensile tests for wild type and oim/oim bone samples.

Test modes	Material property	wild type	oim/oim	p-value
Micropillar test	Compression work to fracture [MJ/m3]	13.9±2.7	8.2±0.9	<0.05
	Elastic modulus [GPa]	24±6.8	21±3.1	>0.05
	Ultimate stress [MPa]	584±157	521±104	>0.05
Tensile test	Ultimate stress [MPa]	50±4.5	38±5.3	<0.001
Nanoscratch test	Toughness [1012 J/m3]	12.6±1.4	6.6±2.2	<0.001

Fig. 5.

Representative post and before test high-resolution SEM images of wild type and oim/oim micropillar bone samples. Wild type micropillars demonstrated shearing and local plastic deformation, whereas oim/oim micropillars primarily failed in a brittle manner showing splitting or shattering.

Fig. 6.

Mechanical properties of wild type (wt) and oim/oim bone obtained from experiments and simulations: (a) Toughness (rigorously, work to fracture in compression) calculated as area under stress strain curve obtained from micropillar tests; (b) The ratio of failure strain to yield strain as a measure of the degree of plastic deformation for micropillar specimens; (c) The ultimate strength in axial tensile tests at the cortical tissue scale; (d) The in situ nanoscratch toughness. In all plots, error bar represents standard deviation. The statistical significance was tested using t-test with p < 0.05.

3.3. Mechanical properties obtained from other tests

At the tissue scale, the uniaxial testing showed that the ultimate tensile strength of oim/oim bone (38±5.3 MPa) was much less than that of wild type bone (50±4.5 MPa) (Fig. 6c). The in situ toughness of oim/oim bone obtained from the nanoscratch tests was also significantly smaller than that of wild type bone (Fig. 6d). These two tests performed at the two distinct length scales confirmed that the material properties and mechanical performance of the oim/oim bone were decreased at different length scales.

3.4. FE simulations results

Under compressive load, the model demonstrated multiple failure modes (Fig. 7). Several shear cracks appeared within the extrafibrillar matrix because of damage occurring in the cohesive elements representing the thin layer of non-collagenous protein binding the mineral crystals within the extrafibrillar matrix. The shear cracks resembled the cross-hatch failure patterns often observed as crack formation mode of bones tested in compression (Ebacher et al. 2007). These shear cracks further induced debonding between the adjacent mineralized collagen fibrils and extrafibrillar matrix subunits. Localized buckling was detected for the mineralized collagen fibrils, more strongly at lower levels of intrafibrillar mineralization. At somewhat higher scale, i.e., at the sub-lamellar scale, the shear nanocracks aligned to form inclined shear failure planes consistent with the micropillar tests results obtained for wild type bone. In the wild type model with higher levels of intrafibrillar mineralization, e.g. 15%, collagen fibrils demonstrated a greater amount of resistance to shear plane formation by undergoing more severe deformations. In contrast, in the model with lower amount of intrafibrillar mineralization, e.g. 5%, the formation of shear plane was less resisted by weakened mineralized collagen fibrils, thus shear cracking of extrafibrillar matrix more readily led debonding and collapse of the bone model. The mechanical properties obtained from simulating the models with different degrees of intrafibrillar mineralization were summarized in Table 2. The reduction in intrafibrillar mineralization, mimicking the case of oim/oim bone, clearly resulted in significant loss of bone mechanical competence, such as reduced elastic modulus, ultimate strength, and compression work to fracture.

Fig. 7.

Deformation and failure pattern obtained from FE model of sub-lamellar bone with different levels of intrafibrillar mineralization analyzed under compression loading in vertical direction. As the amount of intrafibrillar mineralization was diminished the resistance of mineralized collagen fibrils were decreased, resulting in a weaker behavior for oim/oim bone having smaller intrafibrillar mineral content.

Table 2:

Mechanical properties obtained from FE simulation of the ultrastructural bone model with different amounts of intrafibrillar mineralization.

Mechanical property	Intrafibrilar mineralization content
	5%	10%	15%
Elastic modulus E [GPa]	19.5	21.0	23.1
Compressive strength σu [MPa]	822	880	967
Compression work to fracture UT [MJ/m3]	21.6	27.6	31.7
Mineral/collagen strain ratio	0.16	-	0.33
Collagen/tissue strain ratio	1.1	-	1.8

It was observed that the average strain of intrafibrillar mineral platelets and collagen matrix were 0.0018 and 0.011 for the FEM model with 5% intrafibrillar mineralization, 0.0046 and 0.018 for the FEM model with 15% intrafibrillar mineralization, respectively, when these models were loaded under the applied tensile strain of 0.01 (Fig. 8). Using these values, the mineral/fibril and fibril/tissue strain ratios were calculated as 0.16 and 1.1 for the FEM model with 5% intrafibrillar mineralization, 0.33 and 1.8 for the FEM model with 15% intrafibrillar mineralization, respectively. The simulation results indicated that the value of mineral/fibril strain ratio increased markedly with increasing intrafibrillar mineralization (from 0.16 to 0.33). This result was consistent with the in situ strain measurements from X-ray scattering data. Since the mineral platelets were larger in lengths for 15% intrafibrillar mineralization, more load would be transferred to them through the soft collagen matrix, hence increasing the intrafibrillar mineral strain compared to 5% intrafibrillar mineralization.

Fig. 8.

Strain distribution in models with different amounts of intrafibrillar mineralization. The applied tensile strain to the model was 0.01 in the vertical direction. For the ease of strain visualization in mineral and collagen phases, cohesive elements were masked. Increased strain in intrafibrillar region of the model with higher intrafibrillar mineralization was observed. The ratio of average mineral/collagen and collagen/tissue strain were different for two models (see values reported in Table 2). The mineral/collagen strain ratio was higher for model with 15% intrafibrillar mineralization, in line with our synchrotron X-ray measurements of the in situ strain values.

4. Discussion

This study was performed to determine whether the intrafibrillar mineralization diminishes in OI bone and whether the reduced intrafibrillar mineralization contributes to OI induced deterioration of bone mechanical properties at both ultrastructural and tissue levels. The in situ and tissue level mechanical properties, failure modes, intrafibrillar mineralization status, and ultrastructural organization of wild type and oim/oim bones were studied via a combination of several advanced techniques, i.e., micropillar compression test, bulk uniaxial test associated with synchrotron X-ray diffraction measurement, nanoscratch test, and in silico simulation. The obtained results verified the hypothesis of this study, showing that the diminished intrafibrillar mineralization induced by OI was associated with significant deterioration in the strength and toughness of bone at different length scales. Previous studies report that oim bone has an increased mineral density at the tissue level, while it shows a lower mineral density at the whole bone level. We speculate that at the tissue level, the extrafibrillar matrix is more mineralized, whereas collagen fibrils are packed more loosely in OI bone, thus leading to an increase in the overall mineralization at the tissue level.

The synchrotron X-ray scattering results demonstrate that the intrafibrillar mineralization diminishes drastically in the oim/oim bone (Fig. 2). The results of this study suggest that the diminished intrafibrillar mineralization could contribute to the reduced mechanical properties of oim/oim bone, including the strength and damage-tolerance of the tissue (Table 2), similar with the results reported in the previous studies (Buehler, 2007; Nair et al., 2014; Fielder and Nair, 2019). The reduction in intrafibrillar mineralization has also been reported in dentinogenesis imperfecta teeth specimens (Kinney et al., 2001). The diminished intrafibrillar mineralization in oim/oim bone might be associated with the following changes in oim/oim bone: (i) Abnormal lateral packing of collagen fibrils (McBride et al., 1997; Chang et al., 2012), which might limit the initial nucleation and subsequent growth of mineral crystallites; (ii) Insufficient amount of non-collagenous proteins (Dickson et al., 1975; Marini et al., 2014) that regulate the mineralization process from nucleation to mature growth of crystals (Mbuyi-Muamba et al., 1989; Al-Qtaitat and Aldalaen, 2014); (iii) Altered hydration capacity of OI collagen fibrils (Andriotis et al., 2015) given that bound water is indispensable in stabilizing mineral-collagen interactions in bone; (iv) Unbalanced bone formation-resorption and accelerated turnover in oim/oim bone (Camacho et al., 1999; Grabner et al., 2001) suggested by less carbonated mineral crystals (less mature bone) and increased number of largely branched vascular canals in oim/oim bone (Carriero et al., 2014).

It is noted that the mineral crystals alignment in oim/oim bone appears to be uniform in all orientations (Fig. 2), whereas the mineral crystals in normal bone are preferentially aligned in the longitudinal axis of bone. The poor alignment of mineral crystals in oim/oim bone may be due to two possible scenarios. One is most likely related to the reduced intrafibrillar mineralization in oim/oim bone. It is known that the mineral crystals in collagen fibrils are aligned in the longitudinal axis of the collagen fibrils, whereas the mineral crystals in the extrafibrillar matrix of bone are more randomly oriented (Grünewald et al., 2020; Samuel, 2018). Thus, reduction in intrafibrillar mineralization would lead to fewer mineral crystals that are aligned in the preferential direction. The other may pertain to the disorganized arrangement of collagen fibrils in oim/oim bone, thus increasing the randomness of mineral crystal orientation in the tissue (McBride et al., 1997; Forlino et al., 1998). The disorganization of mineralized collagen fibrils may be a lesser factor in compromising the strength of oim/oim bone at the ultrastructural level since no significant difference was reported in previous studies between strength of bone samples obtained from ordered and disordered regions using the similar micropillar compression technique (Tertuliano and Greer, 2016). This disorganization could however affect the toughness of bone, as bone toughness may be sensitive to fiber orientations.

The reduced intrafibrillar mineralization in oim/oim bone may contribute to the thinner oim/oim collagen fibrils reported in the previous studies (Li et al., 2016; Kłosowski et al., 2017). Since the diameter of mineralized collagen fibrils expands as the number of mineral crystals embedded in the collagen fibrils increases, a reduction of mineralization in collagen fibrils would likely result in a decrease in the diameter of fibrils.

The mechanical properties of oim/oim bone evaluated in the different mechanical tests of this study exhibit significant deterioration at both tissue and ultrastructural levels. The micropillar compression tests revealed that oim/oim bone lacks plastic deformation and exhibits more brittle failure modes compared to the wild type bone. (Fig. 4 and Fig. 5). While the wild type bone accommodates local plastic deformation (mainly at top regions of the micropillar) and develops shear failure planes, the oim/oim bone, on the other hand, fails in either splitting mode or shatters into many pieces. The plasticity and shear deformation of the wild type bone is conceivably conferred to it by the deformability of normally synthesized collagen phase (Fratzl, 2008), stiffening of collagen fibrils through sufficient intrafibrillar mineralization, and the energy dissipating denaturation of non-collagenous proteins binding the extrafibrillar minerals in wild type bone (Wang et al., 2018). Previous micropillar compression studies on healthy bone (Schwiedrzik et al., 2017, 2014) have reported failure modes similar to our findings, however, to the best of our knowledge, there are no report of micropillar compression test on oim/oim mice bone in the literature.

The micropillar compression tests offer a great advantage over the common nanoindentation tests (Vanleene et al., 2012; Rodriguez-Florez et al., 2015) to evaluate the mechanical behavior of bone at microscopic levels. Nanoindentation tests do not directly provide the material properties from the measured load-displacement data due to the complex stress states during indentation, varying deformation response of the material beneath the indenter, and the dependence of data interpretation processes on the analytical fitting models frequently used to interpret the data (Rodriguez-Florez et al., 2013). In contrary, the uniaxial micropillar compression tests give rise to less approximation required to derive material properties at microscopic levels. The micropillar compression test can also rule out the effect of structural factors at higher architectural levels, such as the shape, size, and porosity of bone samples (McBride et al., 1998; Camacho et al., 1999). Therefore, this test is more suitable for examining the mechanical properties of bone at ultrastructural levels.

While the compression work to fracture of oim/oim bone measured from the micropillar test is significantly lower than the wild type bone, its compressive strength is approximately equal to the wild type bone (Fig. 6). This finding is similar to the results reported for wild type and osteoporotic bones tested using micro-sized samples (Jimenez-Palomar et al., 2015), suggesting that perhaps bone pathologies affect the toughness more than the compressive strength of bone. Tissue toughness is an important property defining the capacity of bone to resist crack/damage initiation and propagation. On the other hand, our results indicate that the oim/oim bone is significantly weaker in tensile tests at millimeter length scales. This discrepancy may be due to the different behavior of bone between loading modes (tension vs. compression) and bone hierarchies involved. In fact, bone is known to display largely different responses when tested in tension compared to compression (Nyman et al., 2009). Furthermore, the behavior of bone is also strongly dependent on the size of bone specimens, which represent different bone hierarchies (Currey, 2002). Our results of tensile test is in line with previous results of oim/oim bone obtained from tests at similar length scales (Miller et al., 2007; Carriero et al., 2014).

The in situ mineral strain in oim/oim bone is less coordinated with the collagen fibril strain compared to that of the wild type bone (Fig. 3), which is in line with findings reported in the literature (Carriero et al., 2014). This could be due to three reasons. First, the reduced intrafibrillar mineralization and poor spatial order of mineral crystals may negatively affect the cooperative deformation of mineral and collagen phases in oim/oim bone, thus suppressing the otherwise synergistic load-bearing function of these two phases. Second, the inherent difference in the ultrastructure of mineral crystals between oim/oim and wild type bone may contribute to this phenomenon. In fact, the mineral crystallites in oim/oim bone appear relatively thick, bulky and elongated, largely dissimilar in size, shape, location, alignment and orientation with the thin flat platelets usually observed in normally mineralized connective tissues (Landis, 1995; Fratzl et al., 1996; Grabner et al., 2001). This, in turn, changes the strain distribution in the stiff mineral phase and the compliant collagen matrix. Third, the abnormal mineral-collagen interactions in oim/oim bone (Miller et al., 2007) can also play a significant role in changing the strain distribution in mineral and collagen phases, further exacerbating the fragility of oim bone. In fact, enhancing strength and energy absorption capacity of bone via mineral-collagen interactions is known as a fundamental mechanism governing the ultrastructure behavior of bone (Stock, 2015).

While this study suggests that diminished intrafibrillar mineralization plays a role in the increased fragility of OI bone, the brittleness of OI bone may further be related to the following factors: (1) Inherent mechanical weakness of collagen phase (Misof et al., 1997; Chang et al., 2012; Gautieri et al., 2008, 2009) due to the orientation disorder of collagen fibrils (Nadiarnykh et al., 2007), the reduced amount of healthy collagen fibrils (Camacho et al., 1999), and diminished intermolecular interactions of collagen helices in oim bone (Kuznetsova et al., 2001). In addition, kinking of the homotrimeric type I collagen helix in oim bone may interfere with normal helix folding and the interaction of collagen fibrils with non-collagenous extracellular matrix proteins (Forlino et al., 1998). Moreover, reduction in cross-linking due to increased separation of the collagen molecules in the more hydrated fibers in oim bone compared to normal tissue could work against the mechanical properties of oim bone as well (Miles et al., 2002; Sims et al., 2003; Gautieri et al., 2009); (2) Abnormalities in types and density of collagen cross-links (Misof et al., 1997; Miles et al., 2002; Sims et al., 2003; Gautieri et al., 2009; A. Carriero et al., 2014), exacerbated by lack of critical bonds in homotrimeric collagen molecules in oim bone (McBride et al., 1997); (3) Changes in the lateral crystalline packing and axial order in the oim collagen fibrils which could potentially hinder proper mineralization and cooperative deformation of mineral and collagen phases (Misof et al., 1997; Gautieri et al., 2009); (4) Negative alterations in the bone mineral content, mineral density, chemical composition (Camacho et al., 1999), and increased turnover rate (Kalajzic et al., 2002); (5) Negative modifications in interaction behavior between mineral-collagen phases (Miller et al., 2007). Given all these possible factors contributing to the brittleness of OI bone, further comprehensive and multiscale studies are still needed to fully understand this complicated disease.

The FE simulations show that reducing the intrafibrillar mineralization in bone results in weakening the mechanical properties (Table 2 and Fig. 7). This is in line with findings reported in literature where intrafibrillar mineralization was described a critical element to mechanical properties of bone (Balooch et al., 2008; Nair et al., 2014). In fact, proper levels of intrafibrillar mineralization can dramatically enhance the toughness and strength of bone (Buehler, 2007). On the other hand, poor collagen mineralization could also result in reduced mechanical properties as shown from the results obtained in our work. It should be mentioned that the FE simulations performed here were solely intended to act as a comparative study, in which all the parameters were kept unchanged except for the degree of intrafibrillar mineralization. In addition, the FE model was only a simple 2D representation of the much more complex 3D architecture of bone, thus leading to some discrepancies in the results between FE simulations and the micropillar experiments. Therefore, the predicted values for some properties such as ultimate compressive strength may not necessarily match with that of our micropillar compression test. Particularly, one discrepancy between the FE simulations and micropillar tests was that while simulations predicted different modulus and strength for normal and oim bone, micropillar tests found no statistically significant difference between modulus and strength of the two genotype groups. However, the influence of intrafibrillar mineralization on compression work to fracture is clearly verified by the results of the FE simulations.

There are several limitations associated with this study: (1) We mainly focused on mechanisms only at ultrastructural levels, leaving out the effect of other hierarchies such as geometrical factors and porosity. However, the ultrastructural changes may be the origins where bone fragility is initiated, thus directing to potential targets for effective treatments. For instance, therapeutic interventions for promoting mineralization in collagen fibrils may alleviate the adverse effect of OI induced intrafibrillar mineral loss on the mechanical competence of bone tissues. (2) Micropillar test of this study did not provide the entire stress-strain curve due to the constraints of the test device. Nonetheless, the compression work to fracture measurements up until the maximum load were still able to detect brittleness of oim bone at the sub-micron scale. (3) The computational model used to study the effect of diminished intrafibrillar mineralization was a simple 2D model, which may not fully capture the 3D complex ultrastructure of the bone. Another limitation with the model is that collagen fibrils were considered to be fully aligned, which may not be exactly the case in the oim bone. Future studies are planned to create the randomness of collagen fibril orientation in the model such that its effect could also be investigated. (4) The micropillar tests were conducted under the dehydrated condition, which may not fully represent the physiological behavior of the tissue. However, the comparison between the wild type and oim/oim bone is still valid since both are treated in the same way. (5) A potential drawback associated with the micropillar test is the possible influence of sample preparations, in which ion beams might have impacts on the virgin tissue properties. This can be mitigated if proper sequential protocols are employed to use progressively milder ion beam current density in preparing the micropillar samples. (6) In this study, the possible changes in the persistence length of collagen molecules induced by OI are not considered since our focus is on changes in mineral phase and its effect on the mechanical behavior of oim bone. Nonetheless, previous molecular dynamics simulations indicate that changes in the persistence length may lead to a long toe region of stress-strain curve of collagen molecules in oim bones (Chang et al., 2012).

5. Conclusions

This study examined a possible ultrastructural origin of OI bone fragility, i.e., diminished intrafibrillar mineralization, using the oim model and a combination of advanced techniques including micropillar test, tensile test associated with synchrotron X-ray scattering measurements, nanoscratch test, and cohesive FE simulations. The results indicate that diminished intrafibrillar mineral content may lead to deleterious changes in load transfer between the mineral and collagen phases and orientation disorder of mineral crystals in the tissue, thus resulting in the significantly deteriorated mechanical properties and brittle failure mode of oim/oim bone.