SEARCH

SEARCH BY CITATION

Keywords:

  • bone ultrastructure;
  • canal network;
  • osteocyte lacunae;
  • μCT;
  • nano-CT;
  • synchrotron radiation;
  • negative imaging;
  • biomechanics

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Nondestructive SR-based μCT and nano-CT methods have been designed for 3D quantification and morphometric analysis of ultrastructural phenotypes within murine cortical bone, namely the canal network and the osteocyte lacunar system. Results in two different mouse strains, C57BL/6J-Ghrhrlit/J and C3.B6-Ghrhrlit/J, showed that the cannular and lacunar morphometry and their bone mechanics were fundamentally different.

Introduction: To describe the different aspects of bone quality, we followed a hierarchical approach and assessed bone tissue properties in different regimens of spatial resolution, beginning at the organ level and going down to cellular dimensions. For these purposes, we developed different synchrotron radiation (SR)-based CT methods to assess ultrastructural phenotypes of murine bone.

Materials and Methods: The femoral mid-diaphyses of 12 C57BL/6J-Ghrhrlit/J (B6-lit/lit) and 12 homozygous mutants C3.B6-Ghrhrlit/J (C3.B6-lit/lit) were measured with global SR μCT and local SR nano-CT (nCT) at nominal resolutions ranging from 3.5 μm to 700 nm, respectively. For volumetric quantification, morphometric indices were determined for the cortical bone, the canal network, and the osteocyte lacunar system using negative imaging. Moreover, the biomechanics of B6-lit/lit and C3.B6-lit/lit mice was determined by three-point bending.

Results: The femoral mid-diaphysis of C3.B6-lit/lit was larger compared with B6-lit/lit mice. On an ultrastructural level, the cannular indices for C3.B6-lit/lit were generally bigger in comparison with B6-lit/lit mice. Accordingly, we derived and showed a scaling rule, saying that overall cannular indices scaled with bone size, whereas indices describing basic elements of cannular and lacunar morphometry did not. Although in C3.B6-lit/lit, the mean canal volume was larger than in B6-lit/lit, canal number density was proportionally smaller in C3.B6-lit/lit, so that lacuna volume density was found to be constant and therefore independent of mouse strain and sex. The mechanical properties in C3.B6-lit/lit were generally improved compared with B6-lit/lit specimens. For C3.B6-lit/lit, we observed a sex specificity of the mechanical parameters, which could not be explained by bone morphometry on an organ level. However, there is evidence that for C3.B6-lit/lit, the larger cortical bone mass is counterbalanced or even outweighed by the larger canal network in the female mice.

Conclusions: We established a strategy to subdivide murine intracortical porosity into ultrastructural phenotypes, namely the canal network and the osteocyte lacunar system. Nondestructive global and local SR-based CT methods have been designed for 3D quantification and subsequent morphometric analysis of these phenotypes. Results in the two different mouse strains C57BL/6J-Ghrhrlit/J and C3.B6-Ghrhrlit/J showed that the cannular and lacunar morphometry and the biomechanical properties were fundamentally different.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Osteoporosis is primarily a disease of bone fragility resulting from decreased bone mass. In addition, altered architectural arrangement of bone tissue and impaired bone quality leads to decreased skeletal strength and increased fracture risk.(1) BMD, a measure of bone mass, has been identified in several epidemiological studies as being the most important risk factor for osteoporotic fractures.(2,3) Nevertheless, the idea that BMD is the best predictor of fracture risk has become a controversial issue.(4) On an individual basis, density alone accounts for 10–90% of the variation in strength of trabecular bone.(5) This also means that 90–10% of the variation in strength cannot be explained by BMD.(6) Recent data have shown that predicting trabecular bone strength can be greatly improved by including microarchitectural parameters in the analysis.(7,8)

Further improvement in predicting bone strength can be achieved by considering additionally the cortical bone compartment. Whereas the focal point of hip fracture studies was on trabecular bone for the last decades, cortical bone contributes significantly to the mechanical strength of bone.(9,10) Moreover, regarding the stiffness of cortical bone, the influence of small changes in the amount or density of bone tissue is even more pronounced than similar changes would exert in trabecular bone.(11) Eventually, loss of cortical rather than loss of trabecular bone predominates in cases of proximal femur fracture,(12) which is among the most devastating of all osteoporotic fractures.(1) On this account, cortical bone tissue has been studied in more detail and cortical bone strength has been related primarily to BMD and other parameters, which describe cortical geometry.(13,14) However, BMD was reported to be related only weakly to the mechanical properties of cortical bone.(15) On the other hand, cortical geometry and particularly intracortical porosity has been shown already before to be linked to stiffness and strength of cortical bone specimens from human donors(16–18) and from different vertebrates.(16,19,20) Furthermore, intracortical porosity has been associated with fracture risk of patients with femoral neck fractures.(21) Changes in human femoral intracortical porosity have accounted for 76% of the reduction in bone strength,(22) and several other studies revealed the negative influence of increased intracortical porosity on fracture resistance.(23–25)

We therefore suggest studying cortical bone morphology in more detail as an important aspect of bone quality. To this end, we followed a hierarchical approach and assessed cortical bone tissue properties in different regimens of spatial resolution, beginning at the organ level and going down to the cellular domain. We especially focused on the porosity within cortical bone using synchrotron radiation (SR)-based μCT and nano-CT (nCT). In general, μCT is a convenient technique to study trabecular bone architecture and modeling because it allows nondestructive measurement of bone microstructure and quantitative evaluation of its 3D morphometric parameters.(26,27) With the advent of third-generation SR sources, CT in the micrometer range has become feasible and has been used to analyze trabecular architecture and local bone tissue properties.(28–30) Recent results show that the canal network is a major contributor to local tissue porosity,(31–33) and therefore, can directly be linked to measures of bone tissue quality and thus, to the mechanical properties of bone. Additionally, osteocyte lacunae are believed to act as stress concentrators within tissue of compact bone.(34)

The purpose of this study was to establish a strategy to subdivide murine intracortical porosity into 3D ultrastructural phenotypes, namely the canal network and the osteocyte lacunar system, and to provide methods to volumetrically quantify these phenotypes for subsequent morphometric analysis. In a second step, we applied this framework to the femoral mid-diaphyses of two genetically distinct mouse strains. Moreover, Beamer et al.(35) showed that there is a significant difference in BMD and BMC between the two inbred mouse strains C57BL/6J (B6) and C3H/HeJ (C3H). Finally, these two mouse strains showed differences in several skeletal characteristics(35,36) and in response to mechanical loading.(37,38) On this account, B6 and C3H inbred mouse strains offer a good model for the analysis of the different phenotypes of bone and their influence on bone quality. Based on this knowledge, we performed mechanical tests on our mouse model to determine potential mouse strain and, for the first time, sex differences. This also allowed studying the relations between bone mechanics and morphometry on an organ and ultrastructural level.

In the long run, we believe that the morphometric analysis of the ultrastructural phenotypes and the study of bone phenotypes at different hierarchy levels, in conjunction with bone mechanics, will provide new insights in the assessment of bone quality.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Mouse model

To partition bone quality into its regulatory pathways, we chose a mouse model called little (C57BL/6J-Ghrhrlit/J), wherein regulation of bone morphometry is independent of growth hormone (GH) and insulin-like growth factor-I (IGF-I).(39) More precisely, the little mouse strain carries a mutation in the growth hormone–releasing hormone receptor (Ghrhr), resulting in undetectable circulating GH and serum IGF-I that is fixed at low, but detectable levels. For this study, we used homozygous mutant mice from the inbred strain C57BL/6J-Ghrhrlit/J (B6-lit/lit) and the congenic strain C3.B6-Ghrhrlit/J (C3.B6-lit/lit) that was created by backcrossing the little mutation from the original B6 background to C3H. The result of introgressing the little mutation onto the C3H background is that C3.B6-lit/lit mice have higher BMD than B6-lit/lit.(39) All mice were raised at The Jackson Laboratory, and all animal procedures were approved by the local authorities. On necropsy, femora were dissected and were stored in ethanol for further analysis.

Hierarchical bone model

Within this study, we followed a hierarchical description of the cortical bone matrix. For these purposes, we distinguished three levels on different length scales. On the organ level, cortical bone was considered to be compact and was characterized by its radial extension and volume distribution. On the tissue level and cellular level, we described the ultrastructure of the intracortical porosity. Whereas the tissue level was described by the cannular network, the osteocyte lacunar system constituted the cellular level. The cannular network will be introduced and defined properly in the subsection 3D quantitative morphometry that follows.

SR CT

To acquire cortical bone tissue properties in distinct regimens of length scale (organ, tissue, and cellular level), different CT techniques were applied. In general, the organ and the tissue level encompasses the wide range of cortical and trabecular bone structure, which can be assessed by commercially available desktop μCT scanners at a spatial resolution in the order of 10 μm. Whereas the resolution of desktop μCT systems is sufficient to acquire the microstructure of trabecular bone, the ultrastructure of cortical bone is currently not accessible with this technique. For this reason, SR-based CT was adapted and further developed to assess the cortical bone matrix on the organ, the tissue, and the cellular level. Although it would be sufficient to assess the organ level by conventional desktop μCT systems, we restricted our study to the technique of SR CT to guarantee optimal conditions for comparison between the different hierarchy levels.

To study the organ and tissue level of the cortical bone, we scanned femoral mid-diaphyses from 12 B6-lit/lit (6 female and 6 male) and 12 C3.B6-lit/lit (6 female and 6 male) 4-mo-old mice using a conventional (i.e., global) SR μCT setup at a nominal resolution of 3.5 μm (Fig. 1). This cortical compartment started at 56% of the whole femur length (calculated from the greater trochanter) and contained 388 slices resulting in a stack height of 1.56 mm (Table 1). To study the cellular level, we applied local SR CT at a submicron resolution (700-nm nominal resolution); a technique we call nano-CT (nCT). In a local CT setup, the specimen is bigger than the recorded field of view (FOV) perpendicular to the rotation axis and therefore only a portion of the whole sample is assessed. This allows for high-resolution CT without destruction of the specimen. Because of the limited access to SR facilities, only two B6-lit/lit (one female and one male) and two C3.B6-lit/lit (one female and one male) femora were scanned locally, each at the anterior, posterior, lateral, and medial site of the mid-diaphysis. The FOV within the cortical bone assessed by the local SR nCT measurements was a subset of the FOV of the global SR μCT setup, so that the same cortical regions were imaged and corresponding ultrastructural morphometric parameters could be assessed.

Table Table 1.. Different Hierarchy Levels and Corresponding Imaging Techniques Used
Thumbnail image of
thumbnail image

Figure Figure 1. Murine femur of B6-lit/lit (A) and C3.B6-lit/lit (B) mouse. The mid-diaphyseal compartment (indicated by a box) that was measured at the Swiss Light Source (SLS) started at 56% of the whole femur length (calculated from the greater trochanter), according to Kohler et al.(66) The two genetically distinct inbred mouse strains display different morphology and architecture already at the organ level.

Download figure to PowerPoint

The SR μCT and nCT measurements were performed in air at the X-ray Tomographic Microscopy (XTM) station of the Materials Science (MS) beamline at the Swiss Light Source (SLS).(40,41) For both setups, in total, 1001 projections were acquired over a range of 180° at a photon energy of 17.5 keV. The data were reconstructed using filtered backprojection. The parameters of the different CT techniques applied are summarized in Table 1.

Image processing

A new sinogram-based algorithm was devised to eliminate ring artifacts, which arose from defects on the scintillator of the optical system at the XTM station and which were clearly visible in the reconstructed slices. To partially suppress noise within the reconstructed tomograms, a Gaussian filter was applied. Subsequent iterative global thresholding(42) provided binarized data sets separating bone matrix from soft tissue and air.

For quantitative analysis of the bone tissue on the organ level, component labeling and morphologic operations were applied to the binarized data to obtain a solid representation of the mid-diaphysis. The technique of negative imaging was applied in this study to assess the porosity within cortical bone on the tissue and on the cellular level. In this context, negative imaging denotes the technique to first measure the matrix of a porous structure using CT, and subsequently, to extract the enclosed porosity as a negative imprint of the surrounding matrix. In this study, the extraction of the cortical void spaces comprised different image processing procedures (IPL; Scanco Medical, Bassersdorf, Switzerland), including morphological operators. They were optimized to extract the canal network and the osteocyte lacunar system within cortical bone as two separate phases. Additionally, the data sets of the canal network as assessed by global SR μCT was divided into four sections, corresponding to four main anatomical sites, the anterior, posterior, lateral, and medial quadrant of the mid-diaphysis, to analyze potential site-dependencies of the morphometric indices describing the canal network. Furthermore, this supplementary partition was essential for correlation analysis between the canal network obtained from global SR μCT measurements, which contained the entire cortical shaft and the osteocyte lacunar system acquired from local SR nCT measurements, which were restricted to a single anatomical site, because of the limited experimental FOV (Table 1).

3D quantitative morphometry

Organ level:

For morphometry on the organ level, standard 3D algorithms were used(43) to compute total tissue volume (TV) enclosing both the medullary cavity and the cortical bone tissue, cortical bone volume (Ct.BV), cortical bone volume density (Ct.BV/TV), cortical thickness (Ct.Th), and polar area moment (J).

Tissue level:

On the tissue level, the void spaces within cortical bone obtained by negative imaging formed cannular structures (Fig. 2). We believe that these cannular elements represent the living space of the vasculature and/or bone remodeling units (including osteoclast lacunae) within cortical bone. We name this phenotype canal network and we precede each corresponding parameter the designation canal or cannular. For quantification of the canal network, we introduced cannular indices, corresponding to standard nomenclature of bone morphometry(44) for trabecular bone and similar to the naming of cannular parameters motivated by Cooper et al.(33) Hence, we defined number of canals (N.Ca), canal number density (N.Ca/Ct.TV), canal volume (Ca.V), canal volume density (Ca.V/Ct.TV), and canal spacing (Ca.Sp), where Ct.TV designates the cortical total volume. These parameters describe the overall canal network and were quantified by standard morphometry as introduced for 3D trabecular bone analysis.(43) Moreover, the canal network was spatially decomposed into single elements and analyzed subsequently by using element-based morphometry as recently introduced.(45) For this reason, we further defined corresponding element-based indices, which are given as mean values over the total number of elements (marked with brackets 〈 〉). Among those averaged indices were mean canal volume (〈Ca.V〉), mean canal diameter (〈Ca.Dm〉), mean canal length (〈Ca.Le〉), and mean canal orientation (〈Ca.θ〉) in relation to the femoral long axis. An illustration on how these parameters are defined is given in Fig. 3. All cannular indices were computed in 3D space without any model assumptions. In addition to the analysis for the complete mid-diaphyseal cortex, the cannular indices were computed separately for the anterior, posterior, lateral, and medial sites as well. These data were a requisite for later correlation analysis between the cannular and lacunar morphometric outcome.

thumbnail image

Figure Figure 2. 2D reconstructed raw data (A) and 3D mid-diaphyseal canal network (B and C) extracted by negative imaging for the two mouse strains B6-lit/lit (left) and C3.B6-lit/lit (right), where the inner and outer rims shown in B and C represent the extent of the cortical bone. The data have been assessed by global SR μCT at 3.5-μm nominal resolution.

Download figure to PowerPoint

thumbnail image

Figure Figure 3. Illustration of the cannular parameters, which were introduced in the study for morphometry of the canal network. From top to bottom, the windows explain mean canal length (〈Ca.Le〉), mean canal diameter (〈Ca.Dm〉), and mean canal orientation (〈Ca.θ〉).

Download figure to PowerPoint

Cellular level:

On the cellular level, the negative imprint of the cortical bone formed—additionally to the canal network—a cellular system (Fig. 4). Figure 4 identifies the cellular void spaces as osteocyte lacunar system. A number of osteocyte lacunar indices were introduced in analogy to standard morphometry, including number of lacunae (N.Lc), lacuna number density (N.Lc/Ct.TV), lacuna volume (Lc.V), and lacuna volume density (Lc.V/Ct.TV). Moreover, we introduced the element-based mean lacuna volume as the ratio 〈Lc.V〉 = Lc.V/N.Lc.

thumbnail image

Figure Figure 4. The top row shows reconstructed raw data of the lateral cortical mid-diaphysis of a C3.B6-lit/lit mouse femur in a transversal (A) and sagittal view (B). The bottom row represents the canal network (tubes in red) and osteocyte lacunae (prolate ellipsoids in yellow) within the same lateral cortical bone (semitransparent shell in blue) that was extracted from raw data (A and B), using negative imaging. The data have been assessed by local SR nCT at 700-nm nominal resolution.

Download figure to PowerPoint

Biomechanical tests

For mechanical testing, we included an additional set of femoral mid-diaphyses from 12 B6-lit/lit (6 female and 6 male) and 12 C3.B6-lit/lit (6 female and 6 male) 4-mo-old mice. Before testing, the bones were rehydrated for 24 h in PBS. After a preload of 1 N, the mouse femora were loaded to failure in the anterior-posterior direction by three-point bending, using a custom-made loading device with a support distance of 5 mm, which was integrated in a materials testing machine (1456; Zwick, Ulm, Germany). The femora were positioned so that the load was applied at 56% of the whole femur length (calculated from the greater trochanter), which corresponds to the femoral height where the SR CT measurements were performed. During three-point bending, the femora were lying freely and load-displacement curves were recorded at a cross-head speed of 0.5 mm/s.(36) Ultimate force (Fu), work to failure (U), and stiffness (S) were calculated from the load-displacement curve as described elsewhere.(46)Fu indicates the strength of the bone, whereas S reflects resistance to elastic deformation, and U is the required energy to initiate a fracture.

Statistical analysis

For statistical analysis, the GNU statistical package R (Version 2.4.1, http://www.r-project.org) was used. The SR CT measurements on the organ and tissue hierarchy level comprised B6-lit/lit (6 female and 6 male) and C3.B6-lit/lit (6 female and 6 male) mice and consequently, our experimental setup represented a 2 × 2 factorial design, implying the two-level independent variables mouse strain (B6-lit/lit, C3.B6-lit/lit) and sex (female, male). A two-way ANOVA was performed to test for significances (p < 0.05). Additionally, to examine and discuss the behavior of the cannular indices, for which the interaction of the independent variables mouse strain and sex was not clear, they were compared by interaction plots. Finally, to inspect if the cannular indices were dependent on the anatomical site, a two-way ANOVA was applied to the factors site (anterior, posterior, lateral, and medial) and sex (female, male), and Tukey's honestly significant difference (HSD) procedure was performed subsequently for pairwise comparisons among site means as a posthoc analysis. On the cellular level, two B6-lit/lit (one female and one male) and two C3.B6-lit/lit (one female and one male) femora were scanned at the anterior, posterior, lateral, and medial sites of the mid-diaphysis. The primary osteocyte lacunar indices (N.Lc, Lc.V, and corresponding Ct.TV) were summed up over all sites (anterior, posterior, lateral, and medial) for the calculation of the lacunar ratios N.Lc/Ct.TV, Lc.V/Ct.TV, and 〈Lc.V〉 = Lc.V/N.Lc. No further statistics were performed for the cellular level because only one animal per mouse strain and sex was measured on this hierarchy level. Linear regression analysis was performed between all morphometric indices to test for linear relationships. Finally, for three-point bending, the biomechanical parameters Fu, U, and S were compared between the two mouse strains (B6-lit/lit and C3.B6-lit/lit) and among the two sexes (female and male) within the mouse strains using multiple unpaired t-tests, where differences were considered statistically significant at p < 0.05 after Bonferroni correction.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

3D ultrastructural imaging and quantitative morphometry

Following our hierarchical approach on different regimes of spatial resolution, we present in a first step the morphometrical results for the cortical bone compartment (Table 2), for the canal network (Tables 3 and 4), and for the osteocyte lacunae (Table 5). In a second step, correlations between the morphometric parameters on the organ, tissue, and cellular level will be discussed (Tables 6 and 7).

Table Table 2.. Cortical Indices of B6-lit/lit and C3.B6-lit/lit Mice on the Organ Level
Thumbnail image of
Table Table 3.. Cannular Indices of B6-lit/lit and C3.B6-lit/lit Mice on the Tissue Level
Thumbnail image of
Table Table 4.. Cannular Indices of B6-lit/lit and C3.B6-lit/lit Mice on the Tissue Level at Different Anatomical Sites (Anterior, Posterior, Lateral, and Medial)
Thumbnail image of
Table Table 5.. Osteocyte Lacunar Indices of B6-lit/lit and C3.B6-lit/lit Mice on the Cell Level
Thumbnail image of
Table Table 6.. Correlation Between Cortical, Cannular, and Lacunar Morphometric Indices for C3.B6-lit/lit Mice
Thumbnail image of
Table Table 7.. Correlation Between Cortical, Cannular, and Lacunar Morphometric Indices for B6-lit/lit Mice
Thumbnail image of
Organ level:

The cortical indices summarized in Table 2 represent the morphometry results on the organ level. ANOVA of these indices identified the mouse strain as the only significant effect, with p < 0.001. One exception was Ct.BV/TV, where sex had a significant influence on the absolute values. Groupwise unpaired t-tests with Bonferroni correction located a significant difference (p < 0.05) between female and male C3.B6-lit/lit, whereas Ct.BV/TV of B6-lit/lit specimens were not affected by sex. We observed that the morphometry of the two mouse strains was dissimilar already on the organ level or more precisely, that the femoral mid-diaphysis of C3.B6-lit/lit animals was larger in comparison with B6-lit/lit mice (cf. Fig. 1). This finding was supported by the fact that cortical bone volume density (Ct.BV/TV) and cortical thickness (Ct.Th) were significantly (p < 0.001) larger for C3.B6-lit/lit compared with B6-lit/lit (Table 2).

Tissue level:

On the tissue level, negative imaging allowed visualizing the canal network within cortical bone (Fig. 2). The morphometric results for the canal network are summed up in Table 3. Canal spacing (Ca.Sp) and mean canal diameter (〈Ca.Dm〉) were significantly higher in C3.B6-lit/lit compared with B6-lit/lit. In agreement with the larger Ca.Sp, canal number density (N.Ca/Ct.TV) was lower in C3.B6-lit/lit mice; however, this was not significant (p = 0.060). For these three indices, only the factor mouse strain had significant explanation power. All other cannular indices revealed a significant and disordinal interaction between mouse strain and sex (Table 3). For canal volume density (Ca.V/Ct.TV), mean canal volume (〈Ca.V〉), and mean canal length (〈Ca.Le〉), mouse strain and sex were both significant factors. On closer examination, these three indices showed the same interaction behavior, as it is exemplified for 〈Ca.V〉 in Fig. 5. In particular, Ca.V/Ct.TV, 〈Ca.V〉, and 〈Ca.Le〉 were independent of sex for B6-lit/lit mice, but were different for female and male C3.B6-lit/lit mice (cf. Fig. 5). Thus, there is no real cross-over in the interaction plots, and accordingly, the interaction shall be classified as ordinal. To abstract these results, we can state that first Ca.V/Ct.TV, 〈Ca.V〉, and 〈Ca.Le〉 were larger for C3.B6-lit/lit compared with B6-lit/lit mice, and second, that for B6-lit/lit specimens, these indices were independent of sex. In contrast, the mean canal orientation (〈Ca.θ〉), which describes the spatial arrangement of the cannular branches, was found to be dependent on mouse strain and sex (truly disordinal interaction), meaning that it must always be considered separately for mouse strain and sex. Canal branching occurred only rarely and bifurcation of the cannular network was the exception, rather than the normal case. Moreover, the canals delineated a cone around the femoral long axis described by 〈Ca.θ〉, whereas single canals running parallel or perpendicular were observed only rarely. Finally, cannular indices were calculated for the different anatomical sites (anterior, posterior, lateral, and medial) and are listed in Table 4. Because a two-way ANOVA for the factors site and sex did not show any interaction, female and male specimens were pooled for the different sites. The question if the cannular indices do depend on the anatomical site was decided by Tukey's HSD pairwise comparisons among site means for the two mouse strains B6-lit/lit and C3.B6-lit/lit. Those site means, which were significantly different (p < 0.05) from at least one other site, are printed in bold in Table 4. According to Table 4, the element-based indices 〈Ca.V〉, 〈Ca.Dm〉, and 〈Ca.Le〉 are independent of the site, whereas the mean canal orientation (〈Ca.θ〉) was site-dependent for C3.B6-lit/lit and partly for B6-lit/lit. This is shown in Fig. 6, where the canal network within the anterior (A) and posterior (B) cortical bone of a C3.B6-lit/lit mouse femur is displayed. The cannular branches are bent more toward the femurs longitudinal axis for the posterior site compared with the anterior site. With the exception of Ca.Sp for B6-lit/lit, the overall morphometric indices were all dependent on the site, including N.Ca/Ct.TV and Ca.V/Ct.TV. This is particularly pronounced for B6-lit/lit, where the maximal N.Ca/Ct.TV and Ca.V/Ct.TV identified at the anterior site were a multiple of the minimal cannular densities detected at the posterior site. In contrast, only moderate site differences in N.Ca/Ct.TV and Ca.V/Ct.TV were found for C3.B6-lit/lit.

thumbnail image

Figure Figure 5. Interaction plots for mean canal volume (〈Ca.V〉) of B6-lit/lit and C3.B6-lit/lit mice. Because the 〈Ca.V〉 is similar for female and male B6-lit/lit mice, there is no real cross-over, and the significant interaction between the factors mouse strain and sex was qualified as ordinal. The interaction plots of canal volume density (Ca.V/Ct.TV) and mean canal length (〈Ca.Le〉) behave similarly.

Download figure to PowerPoint

thumbnail image

Figure Figure 6. Canal network within the anterior (A) and posterior (B) cortical bone of a C3.B6-lit/lit mouse femur, where the border layers represent the extent of the cortical bone. The morphometric parameters describing the 3D extension and orientation of the canal network vary for different sites (A and B). Note in particular the behavior of the canal penetration within cortical bone for different cortical thicknesses. The data have been assessed by global SR μCT at 3.5-μm nominal resolution.

Download figure to PowerPoint

Cellular level:

Figure 4 shows the canal network and the osteocyte lacunar system within the murine cortical femoral bone as assessed by the nCT approach. The abundance of cell spaces in the cortex deserves closer attention. These osteocyte lacunae were mostly plate-like ellipsoids, oriented mainly in the longitudinal direction of the femur. The morphometric analysis for the lacunar indices is given in Table 5. For the cell space analysis, it must be noted that in C3.B6-lit/lit, the mean lacuna volume (〈Lc.V〉) was larger than in B6-lit/lit. On the other hand, lacuna number density (N.Lc/Ct.TV) was proportionally smaller in C3.B6-lit/lit, so that there is more bone matrix per osteocyte lacuna compared with B6-lit/lit mice, whereas sex seemed to have no influence on the actual numbers. This led to a lacuna volume density (Lc.V/Ct.TV) that was constant at 1.3% for both mouse strains (and both sexes). This is in contrast to volumetric canal density (Ca.V/Ct.TV), which was much larger in C3.B6-lit/lit compared with B6-lit/lit.

Correlations:

Correlation (Tables 6 and 7) showed that all parameters describing the overall cannular morphometry scaled significantly with bone size, i.e., with cortical thickness (Ct.Th) and cortical bone volume (Ct.BV). Among those are canal volume (Ca.V), number of canals (N.Ca), and their corresponding densities Ca.V/Ct.TV and N.Ca/Ct.TV. On the other side, parameters describing basic structural elements such as mean canal volume (〈Ca.V〉) and mean lacuna volume (〈Lc.V〉) were unaffected by size and rather constant. Nevertheless, they were unique per mouse strain. This was corroborated by the finding that strong correlations were detected between number of lacuna (N.Lc) and lacuna volume (Lc.V) (R ≥ 0.83) as well as between N.Ca and Ca.V (R ≥ 0.89) for both B6-lit/lit and C3.B6-lit/lit mice. In general, no linear relations between the morphometric indices describing the osteocyte lacunae and bone size have been observed.

Biomechanics

The mechanical parameters assessed from three-point bending are given in Table 8. First, Fu, U, and S were all significantly different for the two different mouse strains as is shown representatively for Fu in Fig. 7. Second, there were no significant differences found between female and male specimens for B6-lit/lit mice (Table 8). However, in the case of C3.B6-lit/lit, ultimate force (Fu) and work to failure (U) were significantly increased in male specimens compared with female mice (Fig. 7). The same trend was observed for the stiffness (S) of C3.B6-lit/lit mice, although the sex difference was not statistically significant.

Table Table 8.. Biomechanical Parameters for B6-lit/lit and C3.B6-lit/lit Mice
Thumbnail image of
thumbnail image

Figure Figure 7. Ultimate force (Fu) of B6-lit/lit and C3.B6-lit/lit mice. The strength of the bone described by Fu is doubled for C3.B6-lit/lit compared with B6-lit/lit. Whereas for B6-lit/lit, female and male specimens showed the same strength, male C3.B6-lit/lit revealed a significantly increased Fu compared with females. p values of unpaired t-tests with Bonferroni correction are indicated with *p < 0.05 and ***p < 0.001. Comparable results were found for work to failure (U) and stiffness (S).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

3D ultrastructural imaging and quantitative morphometry

Organ level:

Femoral mid-diaphysis of C3.B6-lit/lit is larger compared with B6-lit/lit mice.

In this study, quantitative morphometry on the organ level showed that the femoral mid-diaphysis of C3.B6-lit/lit was larger compared with B6-lit/lit mice, reflecting the increased volume density (Ct.BV/TV) and cortical thickness (Ct.Th) for C3.B6-lit/lit specimens. Similar differences were previously shown by Beamer et al.(35) between the two inbred mouse strains C3H and B6, where C3H mice were reported to exhibit greater total femur density and thicker mid-diaphyseal cortex. Therefore, B6-lit/lit and C3.B6-lit/lit appear to be good models for a high and low bone mass mouse, respectively, similar to B6 and C3H. Moreover, we showed—in line with our previous work(39)—that Ct.BV/TV of females was significantly increased for C3.B6-lit/lit in comparison with males, which is in contrast to B6-lit/lit mice, where all morphometric indices on the organ level were independent of sex. In other words, we could verify a sex specificity of Ct.BV/TV for our little mouse strain, whereas for other mouse models, the direct effect of sex steroids on bone, which is independent of IGF-I, can in general not easily be discerned.(39)

Tissue level:

Cannular indices for C3.B6-lit/lit are generally larger compared with B6-lit/lit mice.

On the tissue level, the cortical bone ultrastructure has been often characterized by intracortical porosity. The intracortical femoral porosity of rats at different sites was found to be increased by several factors, such as immobilization,(47) ovariectomy,(48) and treatment with IGF-I.(49) For wildtype Swiss Webster mice, the intracortical porosity of vertebral bone has been assessed recently by bone histomorphometry(50) and accounted for 0.8–1.2% of the cortical bone extension. In this study, we showed that the canal network and the osteocyte lacunar system together (i.e., the intracortical porosity accessible with our techniques) occupied 1.5–1.8% of the cortical bone extension (Tables 3 and 5). However, we always disentangled the intracortical porosity into the canal network and the osteocyte lacunar system for the whole study and quantified these two ultrastructural phenotypes separately. To our knowledge, the canal network within cortical human bone was quantified only once in 3D by Cooper et al.,(33) using a μCT system at 10-μm nominal spatial resolution and subsequent negative imaging for the canal extraction. Using a combination of SR CT measurements, negative imaging, and direct morphometric 3D methods, Martín-Badosa et al.(51) quantified the porosity and the pore diameter within the distal femoral diaphysis of B6 and C3H mice. Because these authors fixed the nominal resolution at 6.65 μm, osteocyte lacunae were beyond the resolution limit, and the porosity was exclusively a manifestation of the canal network. Given that, according to our results, the mean canal diameter (〈Ca.Dm〉) of femoral murine bone is <10 μm (cf. Table 3), the canal network assessed in the study of Martín-Badosa et al. was partly disconnected and was reduced to bigger canals only. In this study, the cannular indices for C3.B6-lit/lit were generally larger compared with B6-lit/lit mice. Whereas for B6-lit/lit, all cannular indices were independent of sex, Ca.V/Ct.TV, 〈Ca.V〉, and 〈Ca.Le〉 each depended similarly and significantly on sex for C3.B6-lit/lit, as shown representatively for 〈Ca.V〉 in Fig. 5. It is interesting to note that, although Ct.Th and Ct.BV vary for the different anatomical sites within the diaphysis (data not shown), 〈Ca.V〉, 〈Ca.Dm〉, and 〈Ca.Le〉 describing basic elements of the canal network within the cortical bone are independent of site, and therefore, do not scale with bone size (cf. Fig. 6). This is consistent with our correlation results where parameters describing basic elements, such as mean canal volume (〈Ca.V〉) and mean lacuna volume (〈Lc.V〉), were unaffected by size. On the other hand, 〈Ca.θ〉 and the cannular densities N.Ca/Ct.TV and Ca.V/Ct.TV were site-dependent. Especially, the respective 5-fold and 4-fold difference in N.Ca/Ct.TV and Ca.V/Ct.TV between the anterior and posterior site in B6-lit/lit specimens is noteworthy. An analog site-dependency in Ca.V/Ct.TV or nonlacunar porosity was discovered by Skedros et al.(52,53) in mid-diaphyseal ulnas of immature turkeys, where no secondary osteons were detected. There, the cortical octants with the lowest nonlacunar porosity corresponded to the bone region where compression occurs during the normal wing-flapping cycle.(54) No information on habitual loading of the mouse femur diaphysis in vivo is available. Nevertheless, if we assume that the strain distributions within the femoral mid-diaphysis in mice and dogs(55) are comparable, we can conclude that for B6-lit/lit the anterior site with predominant tensile strains and the posterior site with prevalent compressive strains correspond to regions with high and low cannular densities for N.Ca/Ct.TV and Ca.V/Ct.TV, respectively.

Cellular level:

The lacuna volume density is independent of mouse strain and sex.

As for the canal network, the known facts relating to osteocytes and osteocyte lacunae are mainly based on microscopical analysis of histological bone sections and are therefore intrinsically 2D. Quantitative information respecting the 3D osteocyte lacunar system within murine cortical bone is rare and has been assessed thus far using confocal laser scanning microscopy (CLSM).(56) Nevertheless, quantitative analysis in 3D remains challenging because CLSM introduces image artifacts, such as signal attenuation with increasing depth or image distortion caused by the sample refractive index mismatch.(57) However, osteocytes and osteocyte lacunae have been characterized manifold since Frost developed the methodology for dynamic bone histomorphometry. For example, Mullender et al. estimated the cortical osteocyte number density (N.Lc/Ct.TV) in humans(58) and rats(59) to be around 10,000·mm−3 and 90,000·mm−3, respectively. In this study, which is focused on murine specimens, we located N.Lc/Ct.TV between 50,000·mm−3 and 70,000·mm−3 (Table 5). We observed that the osteocyte lacunae are mostly plate-like ellipsoids. Similarly, but based on CLSM measurements in human femoral bone tissue specimens, McCreadie et al.(60) reported that an ellipsoidal model is appropriate for modeling lacunae. The reported values we found in the literature for osteocyte lacuna size of different species are mainly interpolations of 2D data but correspond in the order of magnitude to the values of ∼200–300 μm3 for the mean lacuna volume of our murine bone specimens (Table 5). In comparison, Wang et al.(61) reported on mean lacuna volumina in the range of 200–700 μm3 within the tibial proximal diaphysis of B6 mice, where CLSM in conjunction with a prolated spheroid model for osteocyte lacunae has been used for quantitative morphometry.

In our experiments, the volumetric density of the mid-diaphyseal intracortical canal network (Ca.V/Ct.TV) was positively related with bone size (both Ct.Th and Ct.BV) for both mouse strains (Tables 6 and 7). On the other hand, lacuna volume density (Lc.V/Ct.TV) was constant at 1.3% and consequently independent of mouse strain and sex, at varying mean lacuna volume (〈Lc.V〉) and lacuna number density (N.Lc/Ct.TV) (Table 5). No consistent relation between the lacunar indices and anatomical site could be established. Because only two mice from each mouse strain were analyzed at each single site on the cellular level, we were not able to conclusively answer the question whether there were any site-dependencies of the lacunar indices. However, it is noteworthy that mice with the larger cortices (Ct.Th of C3.B6-lit/lit compared with B6-lit/lit) have a lower lacuna density, similar to findings of Jordan et al.(62) and Borton et al.(63) for mice and men, respectively. This might imply that individual osteoblasts formed more bone before being incorporated into the mineralized tissue. In general, it was proposed that bone resorption and bone remodeling maintain the cell density within a normal range(64) and that, under circumstances where cell and lacuna size and number are not well controlled, lacuna density may have a significant effect on bone matrix stiffness(65) and risk of bone fracture.(64)

Scaling rule:

Overall cannular morphometry scales with bone size and lacunar morphometry does not.

We derived a scaling rule for the canal network and the osteocyte lacunar system. To state explicitly, Ca.V, N.Ca, and their corresponding densities Ca.V/Ct.TV and N.Ca/Ct.TV, all scaled positively with bone size (Ct.Th and Ct.BV). In contrast, 〈Ca.V〉, 〈Ca.Dm〉, and 〈Ca.Le〉 characterizing basic element dimensions of the cannular network, as well as all morphometric indices quantifying the osteocyte lacunae did not scale with bone size.

Biomechanics

In contrast to the femoral midshaft of the wildtype mouse strains B6 and C3H,(36) ultimate force (Fu), work to failure (U), and stiffness (S) of the little specimens were all reduced because of an overall reduction in bone size. Nevertheless, C3.B6-lit/lit showed statistically significant increased mechanical properties versus B6-lit/lit specimens, very similar to the increased mechanical parameters in C3H compared with B6.(36) In the same way, reduced bone size and density, represented by smaller Ct.BV, Ct.Th and Ct.BV/TV (Table 2), can account for lowered strength, rigidity, and work to failure, or in particular, can explain why B6-lit/lit specimens are mechanically inferior to C3.B6-lit/lit mid-diaphyseal femoral bones. Because for C3.B6-lit/lit mice, Ct.BV/TV was significantly reduced for males compared with females (Table 2), we would now expect decreased mechanical characteristics for male C3.B6-lit/lit. However, this could not be seen in our three-point bending results, where all mechanical parameters were found to be increased in male specimens compared with female mice. Consequently, the biomechanical results cannot be explained by considering exclusively bone phenotypes on the organ level. Here, we suggest including the tissue level, or in our case the canal network, for the discussion of the mechanical results. Where for B6-lit/lit, all cannular indices were independent of sex, the morphometric indices Ca.V/Ct.TV, 〈Ca.V〉, and 〈Ca.Le〉 were greater for females in comparison with males in the C3.B6-lit/lit mouse strain (cf. Fig. 5). Consequently, there is evidence that for C3.B6-lit/lit, the larger cortical bone mass is counterbalanced or even outweighed by the larger canal network in the female mice, leading to increased porosity and eventually to decreased mechanical parameters.

Study limitations

As a caveat for this study, it must be mentioned that rodent bones, unlike those of higher mammals, rarely undergo intracortical remodeling. Therefore, direct comparison between the canal network in mice cortices and that in human bone might be misleading. In the mouse, cortical bone is considered to be unchanged since the time the bone was originally formed, whereas in humans, cortical bone is extensively remodeled over time. Although these results showed extensive canal void spaces in the cortical diaphysis of the mouse, this model clearly fails to show osteonal structures as we can find them in humans and other larger mammals. However, we are convinced that the insight gained into the ultrastructure and biomechanics of murine bone provides the requisite framework for genetic studiess in mice dedicated to a better understanding of the hierarchical structural organization in the cortex and its contribution to the mechanical behavior of bone.

Conclusions

To our knowledge, this is the first study investigating quantitatively the ultrastructural cortical bone morphometry of mice in 3D and in a fully nondestructive fashion. Murine intracortical porosity has been disentangled into the canal network and the osteocyte lacunar system, as two ultrastructural phenotypes of (murine) cortical bone. Global and local SR-based CT methods have been designed for quantification and morphometric analysis of these phenotypes. The advantage of SR CT is primarily its high spatial resolution, opening the ultrastructure of bone for nondestructive and 3D assessment, currently not accessible with conventional μCT desktop systems. New morphometric indices have been devised and applied for quantitative characterization of the canal network and the osteocyte lacunar system in the mid-diaphyseal femoral cortex of the two genetically distinct mouse strains B6-lit/lit and C3.B6-lit/lit. Moreover, these little mouse strains seem to be a good model for high and low bone mass, and in addition, for sex differences independent of GH and IGF-I. Their bone mechanics has been described and successfully related to the morphometry on the organ and the ultrastructural level. We believe that the morphometric analysis of the ultrastructural phenotypes and the study of the relationships between phenotypes of bone at different hierarchy levels will provide new insights in the assessment of bone quality on all levels of bone hierarchy. In particular, we hypothesize that the prediction of bone mechanics can be improved in certain diseases and genetic predispositions by including ultrastructural bone tissue properties, such as the canal network and the osteocyte lacunar system.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The authors thank Stefan Heinzer for help during the preparation of the manuscript. This study was supported by the Swiss National Science Foundation through the SNF Professorship in Bioengineering (FP 620–58097.99 and PP-104317/1) and by the U.S. Department of the Army (DAMD 17–01–1–0808).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES