Ultrastructural Properties in Cortical Bone Vary Greatly in TwoInbred Strains of Mice as Assessed by Synchrotron Light Based

Micro- and Nano-CT

Philipp Schneider,1 Martin Stauber,1 Romain Voide,1 Marco Stampanoni,2 Leah Rae Donahue,3 and Ralph Müller1

ABSTRACT: Nondestructive SR-based µCT and nano-CT methods have been designed for 3D quantificationand morphometric analysis of ultrastructural phenotypes within murine cortical bone, namely the canalnetwork and the osteocyte lacunar system. Results in two different mouse strains, C57BL/6J-Ghrhrlit/J andC3.B6-Ghrhrlit/J, showed that the cannular and lacunar morphometry and their bone mechanics were fun-damentally different.

Introduction: To describe the different aspects of bone quality, we followed a hierarchical approach andassessed bone tissue properties in different regimens of spatial resolution, beginning at the organ level andgoing 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 homozygousmutants C3.B6-Ghrhrlit/J (C3.B6-lit/lit) were measured with global SR �CT and local SR nano-CT (nCT) atnominal resolutions ranging from 3.5 �m to 700 nm, respectively. For volumetric quantification, morphometricindices were determined for the cortical bone, the canal network, and the osteocyte lacunar system usingnegative 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 ultrastruc-tural 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 inC3.B6-lit/lit, so that lacuna volume density was found to be constant and therefore independent of mousestrain and sex. The mechanical properties in C3.B6-lit/lit were generally improved compared with B6-lit/litspecimens. For C3.B6-lit/lit, we observed a sex specificity of the mechanical parameters, which could not beexplained by bone morphometry on an organ level. However, there is evidence that for C3.B6-lit/lit, the largercortical 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 pheno-types, namely the canal network and the osteocyte lacunar system. Nondestructive global and local SR-basedCT methods have been designed for 3D quantification and subsequent morphometric analysis of these phe-notypes. Results in the two different mouse strains C57BL/6J-Ghrhrlit/J and C3.B6-Ghrhrlit/J showed that thecannular and lacunar morphometry and the biomechanical properties were fundamentally different.J Bone Miner Res 2007;22:1557–1570. Published online on July 2, 2007; doi: 10.1359/JBMR.070703

Key words: bone ultrastructure, canal network, osteocyte lacunae, µCT, nano-CT, synchrotron radiation,negative imaging, biomechanics

INTRODUCTION

OSTEOPOROSIS IS PRIMARILY a disease of bone fragilityresulting from decreased bone mass. In addition, al-

tered architectural arrangement of bone tissue and im-

paired bone quality leads to decreased skeletal strength andincreased fracture risk.(1) BMD, a measure of bone mass,has been identified in several epidemiological studies asbeing the most important risk factor for osteoporotic frac-tures.(2,3) Nevertheless, the idea that BMD is the best pre-dictor of fracture risk has become a controversial issue.(4)

On an individual basis, density alone accounts for 10–90%The authors state that they have no conflicts of interest.

1Institute for Biomedical Engineering, University and ETH Zürich, Zürich, Switzerland; 2Swiss Light Source (SLS), Paul ScherrerInstitut (PSI), Villigen, Switzerland; 3The Jackson Laboratory, Bar Harbor, Maine, USA.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 22, Number 10, 2007Published online on July 2, 2007; doi: 10.1359/JBMR.070703© 2007 American Society for Bone and Mineral Research

1557

JO612799 1557 1570 October

of the variation in strength of trabecular bone.(5) This alsomeans that 90–10% of the variation in strength cannot beexplained by BMD.(6) Recent data have shown that pre-dicting trabecular bone strength can be greatly improved byincluding microarchitectural parameters in the analysis.(7,8)

Further improvement in predicting bone strength can beachieved by considering additionally the cortical bone com-partment. Whereas the focal point of hip fracture studieswas on trabecular bone for the last decades, cortical bonecontributes significantly to the mechanical strength ofbone.(9,10) Moreover, regarding the stiffness of corticalbone, the influence of small changes in the amount or den-sity of bone tissue is even more pronounced than similarchanges would exert in trabecular bone.(11) Eventually, lossof cortical rather than loss of trabecular bone predominatesin cases of proximal femur fracture,(12) which is among themost devastating of all osteoporotic fractures.(1) On thisaccount, cortical bone tissue has been studied in more detailand cortical bone strength has been related primarily toBMD and other parameters, which describe cortical geom-etry.(13,14) However, BMD was reported to be related onlyweakly to the mechanical properties of cortical bone.(15) Onthe other hand, cortical geometry and particularly intracor-tical porosity has been shown already before to be linked tostiffness and strength of cortical bone specimens from hu-man donors(16–18) and from different vertebrates.(16,19,20)

Furthermore, intracortical porosity has been associatedwith fracture risk of patients with femoral neck frac-tures.(21) Changes in human femoral intracortical porosityhave accounted for 76% of the reduction in bonestrength,(22) and several other studies revealed the negativeinfluence of increased intracortical porosity on fracture re-sistance.(23–25)

We therefore suggest studying cortical bone morphologyin more detail as an important aspect of bone quality. Tothis end, we followed a hierarchical approach and assessedcortical bone tissue properties in different regimens of spa-tial resolution, beginning at the organ level and going downto the cellular domain. We especially focused on the poros-ity within cortical bone using synchrotron radiation (SR)-based �CT and nano-CT (nCT). In general, �CT is a con-venient technique to study trabecular bone architecture andmodeling because it allows nondestructive measurement ofbone microstructure and quantitative evaluation of its 3Dmorphometric parameters.(26,27) With the advent of third-generation SR sources, CT in the micrometer range hasbecome feasible and has been used to analyze trabeculararchitecture and local bone tissue properties.(28–30) Recentresults show that the canal network is a major contributorto local tissue porosity,(31–33) and therefore, can directly belinked to measures of bone tissue quality and thus, to themechanical properties of bone. Additionally, osteocyte la-cunae are believed to act as stress concentrators within tis-sue of compact bone.(34)

The purpose of this study was to establish a strategy tosubdivide murine intracortical porosity into 3D ultrastruc-tural phenotypes, namely the canal network and the osteo-cyte lacunar system, and to provide methods to volumetri-cally quantify these phenotypes for subsequentmorphometric analysis. In a second step, we applied this

framework to the femoral mid-diaphyses of two geneticallydistinct mouse strains. Moreover, Beamer et al.(35) showedthat there is a significant difference in BMD and BMCbetween the two inbred mouse strains C57BL/6J (B6) andC3H/HeJ (C3H). Finally, these two mouse strains showeddifferences in several skeletal characteristics(35,36) and inresponse to mechanical loading.(37,38) On this account, B6and C3H inbred mouse strains offer a good model for theanalysis of the different phenotypes of bone and their in-fluence on bone quality. Based on this knowledge, we per-formed mechanical tests on our mouse model to determinepotential mouse strain and, for the first time, sex differ-ences. This also allowed studying the relations betweenbone mechanics and morphometry on an organ and ultra-structural level.

In the long run, we believe that the morphometric analy-sis of the ultrastructural phenotypes and the study of bonephenotypes at different hierarchy levels, in conjunctionwith bone mechanics, will provide new insights in the as-sessment of bone quality.

MATERIALS AND METHODS

Mouse model

To partition bone quality into its regulatory pathways, wechose a mouse model called little (C57BL/6J-Ghrhrlit/J),wherein regulation of bone morphometry is independent ofgrowth hormone (GH) and insulin-like growth factor-I(IGF-I).(39) More precisely, the little mouse strain carries amutation in the growth hormone–releasing hormone recep-tor (Ghrhr), resulting in undetectable circulating GH andserum IGF-I that is fixed at low, but detectable levels. Forthis study, we used homozygous mutant mice from the in-bred strain C57BL/6J-Ghrhrlit/J (B6-lit/lit) and the congenicstrain C3.B6-Ghrhrlit/J (C3.B6-lit/lit) that was created bybackcrossing the little mutation from the original B6 back-ground to C3H. The result of introgressing the little muta-tion onto the C3H background is that C3.B6-lit/lit micehave higher BMD than B6-lit/lit.(39) All mice were raised atThe Jackson Laboratory, and all animal procedures wereapproved by the local authorities. On necropsy, femora weredissected and were stored in ethanol for further analysis.

Hierarchical bone model

Within this study, we followed a hierarchical descriptionof the cortical bone matrix. For these purposes, we distin-guished three levels on different length scales. On the organlevel, cortical bone was considered to be compact and wascharacterized by its radial extension and volume distribu-tion. On the tissue level and cellular level, we described theultrastructure of the intracortical porosity. Whereas the tis-sue level was described by the cannular network, the osteo-cyte lacunar system constituted the cellular level. The can-nular network will be introduced and defined properly inthe subsection 3D quantitative morphometry that follows.

SR CT

To acquire cortical bone tissue properties in distinct regi-mens of length scale (organ, tissue, and cellular level), dif-

SCHNEIDER ET AL.1558

ferent CT techniques were applied. In general, the organand the tissue level encompasses the wide range of corticaland trabecular bone structure, which can be assessed bycommercially available desktop �CT scanners at a spatialresolution in the order of 10 �m. Whereas the resolution ofdesktop �CT systems is sufficient to acquire the microstruc-ture of trabecular bone, the ultrastructure of cortical boneis currently not accessible with this technique. For this rea-son, SR-based CT was adapted and further developed toassess the cortical bone matrix on the organ, the tissue, andthe cellular level. Although it would be sufficient to assessthe organ level by conventional desktop �CT systems, werestricted our study to the technique of SR CT to guaranteeoptimal conditions for comparison between the differenthierarchy levels.

To study the organ and tissue level of the cortical bone,we scanned femoral mid-diaphyses from 12 B6-lit/lit (6 fe-male 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 �CTsetup at a nominal resolution of 3.5 �m (Fig. 1). This cor-tical compartment started at 56% of the whole femur length(calculated from the greater trochanter) and contained 388slices resulting in a stack height of 1.56 mm (Table 1). Tostudy the cellular level, we applied local SR CT at a sub-micron resolution (700-nm nominal resolution); a techniquewe call nano-CT (nCT). In a local CT setup, the specimenis bigger than the recorded field of view (FOV) perpendicu-lar to the rotation axis and therefore only a portion of thewhole sample is assessed. This allows for high-resolutionCT without destruction of the specimen. Because of thelimited access to SR facilities, only two B6-lit/lit (one femaleand one male) and two C3.B6-lit/lit (one female and onemale) femora were scanned locally, each at the anterior,

posterior, lateral, and medial site of the mid-diaphysis. TheFOV within the cortical bone assessed by the local SR nCTmeasurements was a subset of the FOV of the global SR�CT setup, so that the same cortical regions were imagedand corresponding ultrastructural morphometric param-eters could be assessed.

The SR �CT and nCT measurements were performed inair at the X-ray Tomographic Microscopy (XTM) station ofthe Materials Science (MS) beamline at the Swiss LightSource (SLS).(40,41) For both setups, in total, 1001 projec-tions were acquired over a range of 180° at a photon energyof 17.5 keV. The data were reconstructed using filteredbackprojection. The parameters of the different CT tech-niques applied are summarized in Table 1.

Image processing

A new sinogram-based algorithm was devised to elimi-nate ring artifacts, which arose from defects on the scintil-lator of the optical system at the XTM station and whichwere clearly visible in the reconstructed slices. To partiallysuppress noise within the reconstructed tomograms, aGaussian filter was applied. Subsequent iterative globalthresholding(42) provided binarized data sets separatingbone matrix from soft tissue and air.

For quantitative analysis of the bone tissue on the organlevel, component labeling and morphologic operationswere applied to the binarized data to obtain a solid repre-sentation of the mid-diaphysis. The technique of negativeimaging was applied in this study to assess the porositywithin cortical bone on the tissue and on the cellular level.In this context, negative imaging denotes the technique tofirst measure the matrix of a porous structure using CT, and

TABLE 1. DIFFERENT HIERARCHY LEVELS AND CORRESPONDING IMAGING TECHNIQUES USED

Hierarchy level Technique Resolution FOV 5 FOV 0 #(B6/C3.B6)-lit/lit

Organ Global �CT 3.5 �m 1.36 mm 3.58 mm 12/12Tissue Global �CT 3.5 �m 1.36 mm 3.58 mm 12/12Cell Local nCT 700 nm 0.72 mm 0.72 mm 2/2

Resolution, isotropic nominal image resolution as assessed from voxel size; FOV, field of view parallel (5) and perpendicular (0) to the femoral long axis,which was aligned with the rotation axis; #(B6/C3.B6)-lit/lit, number of mice included in the analysis.

FIG. 1. Murine femur of B6-lit/lit (A) andC3.B6-lit/lit (B) mouse. The mid-diaphysealcompartment (indicated by a box) that wasmeasured at the Swiss Light Source (SLS)started at 56% of the whole femur length(calculated from the greater trochanter), ac-cording to Kohler et al.(66) The two geneti-cally distinct inbred mouse strains displaydifferent morphology and architecture al-ready at the organ level.

ULTRASTRUCTURAL PROPERTIES IN MURINE CORTICAL BONE 1559

subsequently, to extract the enclosed porosity as a negativeimprint of the surrounding matrix. In this study, the extrac-tion of the cortical void spaces comprised different image

processing procedures (IPL; Scanco Medical, Bassersdorf,Switzerland), including morphological operators. Theywere optimized to extract the canal network and the osteo-cyte lacunar system within cortical bone as two separatephases. Additionally, the data sets of the canal network asassessed by global SR �CT 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 themorphometric indices describing the canal network. Fur-thermore, this supplementary partition was essential forcorrelation analysis between the canal network obtainedfrom global SR �CT measurements, which contained theentire cortical shaft and the osteocyte lacunar system ac-quired from local SR nCT measurements, which were re-stricted to a single anatomical site, because of the limitedexperimental FOV (Table 1).

3D quantitative morphometry

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

Tissue level: On the tissue level, the void spaces withincortical bone obtained by negative imaging formed cannu-lar structures (Fig. 2). We believe that these cannular ele-ments represent the living space of the vasculature and/orbone remodeling units (including osteoclast lacunae) withincortical bone. We name this phenotype canal network andwe precede each corresponding parameter the designationcanal or cannular. For quantification of the canal network,we introduced cannular indices, corresponding to standardnomenclature of bone morphometry(44) for trabecular boneand similar to the naming of cannular parameters moti-vated by Cooper et al.(33) Hence, we defined number ofcanals (N.Ca), canal number density (N.Ca/Ct.TV), canalvolume (Ca.V), canal volume density (Ca.V/Ct.TV), andcanal spacing (Ca.Sp), where Ct.TV designates the corticaltotal volume. These parameters describe the overall canalnetwork and were quantified by standard morphometry asintroduced for 3D trabecular bone analysis.(43) Moreover,the canal network was spatially decomposed into single el-ements and analyzed subsequently by using element-basedmorphometry as recently introduced.(45) For this reason,we further defined corresponding element-based indices,which are given as mean values over the total number ofelements (marked with brackets ⟨ ⟩). Among those aver-aged indices were mean canal volume (⟨Ca.V⟩), mean canaldiameter (⟨Ca.Dm⟩), mean canal length (⟨Ca.Le⟩), andmean canal orientation (⟨Ca.�⟩) in relation to the femorallong axis. An illustration on how these parameters are de-fined is given in Fig. 3. All cannular indices were computedin 3D space without any model assumptions. In addition tothe analysis for the complete mid-diaphyseal cortex, thecannular indices were computed separately for the anterior,posterior, lateral, and medial sites as well. These data werea requisite for later correlation analysis between the can-nular and lacunar morphometric outcome.

FIG. 2. 2D reconstructed raw data (A) and 3D mid-diaphysealcanal network (B and C) extracted by negative imaging for thetwo mouse strains B6-lit/lit (left) and C3.B6-lit/lit (right), wherethe inner and outer rims shown in B and C represent the extent ofthe cortical bone. The data have been assessed by global SR �CTat 3.5-�m nominal resolution.

SCHNEIDER ET AL.1560

Cellular level: On the cellular level, the negative imprintof the cortical bone formed—additionally to the canal net-work—a cellular system (Fig. 4). Figure 4 identifies the cel-lular void spaces as osteocyte lacunar system. A number ofosteocyte lacunar indices were introduced in analogy tostandard morphometry, including number of lacunae(N.Lc), lacuna number density (N.Lc/Ct.TV), lacuna vol-ume (Lc.V), and lacuna volume density (Lc.V/Ct.TV).Moreover, we introduced the element-based mean lacunavolume as the ratio ⟨Lc.V⟩ � Lc.V/N.Lc.

Biomechanical tests

For mechanical testing, we included an additional set offemoral mid-diaphyses from 12 B6-lit/lit (6 female and 6male) and 12 C3.B6-lit/lit (6 female and 6 male) 4-mo-oldmice. Before testing, the bones were rehydrated for 24 h inPBS. After a preload of 1 N, the mouse femora were loadedto failure in the anterior-posterior direction by three-pointbending, using a custom-made loading device with a sup-port distance of 5 mm, which was integrated in a materialstesting machine (1456; Zwick, Ulm, Germany). The femorawere positioned so that the load was applied at 56% of thewhole femur length (calculated from the greater trochan-ter), which corresponds to the femoral height where the SRCT measurements were performed. During three-pointbending, the femora were lying freely and load-displacement curves were recorded at a cross-head speed of0.5 mm/s.(36) Ultimate force (Fu), work to failure (U), andstiffness (S) were calculated from the load-displacement

curve as described elsewhere.(46) Fu indicates the strengthof the bone, whereas S reflects resistance to elastic defor-mation, 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 SRCT measurements on the organ and tissue hierarchy levelcomprised B6-lit/lit (6 female and 6 male) and C3.B6-lit/lit(6 female and 6 male) mice and consequently, our experi-mental setup represented a 2 × 2 factorial design, implyingthe two-level independent variables mouse strain (B6-lit/lit,C3.B6-lit/lit) and sex (female, male). A two-way ANOVAwas performed to test for significances (p < 0.05). Addition-ally, to examine and discuss the behavior of the cannularindices, for which the interaction of the independent vari-ables mouse strain and sex was not clear, they were com-pared by interaction plots. Finally, to inspect if the cannularindices were dependent on the anatomical site, a two-wayANOVA was applied to the factors site (anterior, posterior,lateral, and medial) and sex (female, male), and Tukey’shonestly significant difference (HSD) procedure was per-formed subsequently for pairwise comparisons among sitemeans as a posthoc analysis. On the cellular level, two B6-lit/lit (one female and one male) and two C3.B6-lit/lit (onefemale and one male) femora were scanned at the anterior,posterior, lateral, and medial sites of the mid-diaphysis. Theprimary osteocyte lacunar indices (N.Lc, Lc.V, and corre-sponding Ct.TV) were summed up over all sites (anterior,posterior, lateral, and medial) for the calculation of thelacunar ratios N.Lc/Ct.TV, Lc.V/Ct.TV, and ⟨Lc.V⟩ �Lc.V/N.Lc. No further statistics were performed for the cel-lular level because only one animal per mouse strain andsex was measured on this hierarchy level. Linear regressionanalysis was performed between all morphometric indicesto test for linear relationships. Finally, for three-point bend-ing, the biomechanical parameters Fu, U, and S were com-pared between the two mouse strains (B6-lit/lit and C3.B6-lit/lit) and among the two sexes (female and male) withinthe mouse strains using multiple unpaired t-tests, wheredifferences were considered statistically significant at p <0.05 after Bonferroni correction.

RESULTS

3D ultrastructural imaging andquantitative morphometry

Following our hierarchical approach on different regimesof spatial resolution, we present in a first step the morpho-metrical results for the cortical bone compartment (Table2), for the canal network (Tables 3 and 4), and for theosteocyte lacunae (Table 5). In a second step, correlationsbetween the morphometric parameters on the organ, tissue,and cellular level will be discussed (Tables 6 and 7).

Organ level: The cortical indices summarized in Table 2represent the morphometry results on the organ level.ANOVA of these indices identified the mouse strain as theonly significant effect, with p < 0.001. One exception wasCt.BV/TV, where sex had a significant influence on the ab-

FIG. 3. Illustration of the cannular parameters, which were in-troduced 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 ori-entation (⟨Ca.�⟩).

ULTRASTRUCTURAL PROPERTIES IN MURINE CORTICAL BONE 1561

solute values. Groupwise unpaired t-tests with Bonferronicorrection located a significant difference (p < 0.05) be-tween female and male C3.B6-lit/lit, whereas Ct.BV/TV ofB6-lit/lit specimens were not affected by sex. We observedthat the morphometry of the two mouse strains was dissimi-lar already on the organ level or more precisely, that thefemoral mid-diaphysis of C3.B6-lit/lit animals was larger incomparison 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 signifi-cantly (p < 0.001) larger for C3.B6-lit/lit compared withB6-lit/lit (Table 2).

Tissue level: On the tissue level, negative imaging allowedvisualizing the canal network within cortical bone (Fig. 2).The morphometric results for the canal network aresummed up in Table 3. Canal spacing (Ca.Sp) and mean

FIG. 4. The top row shows reconstructedraw data of the lateral cortical mid-diaphysisof a C3.B6-lit/lit mouse femur in a transver-sal (A) and sagittal view (B). The bottomrow represents the canal network (tubes inred) and osteocyte lacunae (prolate ellip-soids in yellow) within the same lateral cor-tical bone (semitransparent shell in blue)that was extracted from raw data (A and B),using negative imaging. The data have beenassessed by local SR nCT at 700-nm nominalresolution.

TABLE 2. CORTICAL INDICES OF B6-LIT/LIT AND C3.B6-LIT/LIT MICE ON THE ORGAN LEVEL

Parameter

B6-lit/lit C3.B6-lit/lit ANOVA

Female Male Difference (%) Female Male Difference (%) Strain Sex Interaction

TV (mm3) 1.47 ± 0.07 1.56 ± 0.10 6 1.15 ± 0.07 1.18 ± 0.11 3 <0.001 0.098 0.409Ct.BV (mm3) 0.60 ± 0.04 0.61 ± 0.03 3 0.76 ± 0.06 0.75 ± 0.08 −2 <0.001 0.982 0.494Ct.BV/TV (%) 41 ± 1 39 ± 2 −3 66 ± 2 63 ± 2 −5 <0.001 <0.01 0.153Ct.Th (�m) 134 ± 7 133 ± 4 −1 222 ± 12 210 ± 15 −6 <0.001 0.124 0.197J (mm4) 0.13 ± 0.02 0.14 ± 0.02 10 0.10 ± 0.01 0.11 ± 0.02 3 <0.001 0.286 0.491

Values are mean ± SD.Interaction within ANOVA denotes interaction between the factors mouse strain and sex.Total tissue volume (TV) including 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).

SCHNEIDER ET AL.1562

Fig 4 live 4/C

canal diameter (⟨Ca.Dm⟩) were significantly higher inC3.B6-lit/lit compared with B6-lit/lit. In agreement with thelarger Ca.Sp, canal number density (N.Ca/Ct.TV) waslower in C3.B6-lit/lit mice; however, this was not significant(p � 0.060). For these three indices, only the factor mousestrain had significant explanation power. All other cannularindices revealed a significant and disordinal interaction be-tween mouse strain and sex (Table 3). For canal volumedensity (Ca.V/Ct.TV), mean canal volume (⟨Ca.V⟩), andmean canal length (⟨Ca.Le⟩), mouse strain and sex wereboth significant factors. On closer examination, these threeindices showed the same interaction behavior, as it is ex-emplified 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/litmice (cf. Fig. 5). Thus, there is no real cross-over in theinteraction plots, and accordingly, the interaction shall beclassified as ordinal. To abstract these results, we can statethat first Ca.V/Ct.TV, ⟨Ca.V⟩, and ⟨Ca.Le⟩ were larger forC3.B6-lit/lit compared with B6-lit/lit mice, and second, thatfor B6-lit/lit specimens, these indices were independent ofsex. In contrast, the mean canal orientation (⟨Ca.�⟩), whichdescribes the spatial arrangement of the cannular branches,was found to be dependent on mouse strain and sex (trulydisordinal interaction), meaning that it must always be con-sidered separately for mouse strain and sex. Canal branch-ing occurred only rarely and bifurcation of the cannularnetwork was the exception, rather than the normal case.

TABLE 3. CANNULAR INDICES OF B6-LIT/LIT AND C3.B6-LIT/LIT MICE ON THE TISSUE LEVEL

Parameter

B6-lit/lit C3.B6-lit/lit ANOVA

Female Male Mean Female Male Mean Strain Sex Interaction

N.Ca/Ct.TV (mm−3) 197 ± 45 190 ± 48 193 ± 44 184 ± 21 142 ± 27 163 ± 32 0.060 0.126 0.256Ca.V/Ct.TV (‰) 1.7 ± 0.6 1.8 ± 0.8 1.8 ± 0.6 5.7 ± 0.8 3.8 ± 1.5 4.8 ± 1.5 <0.001 <0.05 <0.05Ca.Sp (�m) 130 ± 5 129 ± 4 129 ± 5 179 ± 7 184 ± 7 181 ± 7 <0.001 0.498 0.217⟨Ca.V⟩ (103·�m3) 10 ± 1 10 ± 3 10 ± 2 28 ± 4 20 ± 1 24 ± 5 <0.001 <0.01 <0.001⟨Ca.Dm⟩ (�m) 7.9 ± 0.7 7.8 ± 0.9 7.8 ± 0.8 9.3 ± 0.6 8.9 ± 0.3 9.1 ± 0.5 <0.001 0.315 0.624⟨Ca.Le⟩ (�m) 71 ± 8 78 ± 9 74 ± 9 173 ± 19 128 ± 12 150 ± 28 <0.001 <0.01 <0.001⟨Ca.�⟩ (deg) 62 ± 6 55 ± 8 — 36 ± 4 40 ± 3 — <0.001 0.602 <0.05

Values are mean ± SD.Interaction within ANOVA denotes interaction between the factors mouse strain and sex.Mean value for ⟨Ca.�⟩ is not given because the interaction was significant and truly disordinal.Canal number density (N.Ca/Ct.TV), canal volume density (Ca.V/Ct.TV), canal spacing (Ca.Sp), mean canal volume (⟨Ca.V⟩), mean canal diameter

(⟨Ca.Dm⟩), mean canal length (⟨Ca.Le⟩), and mean canal orientation (⟨Ca.�⟩).

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)

Parameter

B6-lit/lit C3.B6-lit/lit

Anterior Posterior Lateral Medial Anterior Posterior Lateral Medial

N.Ca/Ct.TV (mm−3) 313 ± 114 60 ± 37 136 ± 52 244 ± 82 127 ± 48 145 ± 52 213 ± 45 178 ± 48Ca.V/Ct.TV (‰) 2.3 ± 1.0 0.6 ± 0.4 1.1 ± 0.5 2.5 ± 1.4 2.8 ± 1.6 4.0 ± 2.0 6.5 ± 1.7 5.0 ± 2.0Ca.Sp (�m) 322 ± 67 320 ± 277 443 ± 122 305 ± 67 409 ± 104 275 ± 78 201 ± 33 274 ± 60⟨Ca.V⟩ (103·�m3) 8 ± 2 9 ± 4 9 ± 3 9 ± 2 21 ± 6 19 ± 7 23 ± 5 25 ± 7⟨Ca.Dm⟩ (�m) 7.5 ± 1.1 7.9 ± 1.0 8.1 ± 1.1 7.2 ± 0.8 8.9 ± 1.0 8.6 ± 0.9 8.8 ± 0.5 9.2 ± 0.8⟨Ca.Le⟩ (�m) 60 ± 9 87 ± 70 58 ± 13 80 ± 11 136 ± 24 126 ± 37 154 ± 32 149 ± 32⟨Ca.�⟩ (deg) 65 ± 11 47 ± 17 56 ± 8 58 ± 9 45 ± 7 31 ± 5 33 ± 7 46 ± 7

Values are mean ± SD.Values statistically significant different to another site (p < 0.05) according to Tukey’s HSD test are shown in bold.See legend for Table 3 for further information concerning nomenclature of the listed morphometric indices.

TABLE 5. OSTEOCYTE LACUNAR INDICES OF B6-LIT/LIT AND C3.B6-LIT/LIT MICE ON THE CELL LEVEL

Parameter

B6-lit/lit C3.B6-lit/lit

Female Male Mean Female Male Mean

N.Lc/Ct.TV (mm−3) 68,257 63,473 65,865 48,611 51,146 49,879Lc.V/Ct.TV (%) 1.4 1.2 1.3 1.3 1.4 1.3⟨Lc.V⟩ (�m3) 209 191 200 265 274 269

Primary indices number of lacunae (N.Lc), lacuna volume (Lc.V), and corresponding cortical total volume Ct.TV, averaged over all sites (anterior,posterior, lateral, and medial).

Ratios lacuna number density (N.Lc/Ct.TV), lacuna volume density (Lc.V/Ct.TV), and mean lacuna volume (⟨Lc.V⟩ � Lc.V/N.Lc).

ULTRASTRUCTURAL PROPERTIES IN MURINE CORTICAL BONE 1563

TA

BL

E6.

CO

RR

EL

AT

ION

BE

TW

EE

NC

OR

TIC

AL

,CA

NN

UL

AR

,AN

DL

AC

UN

AR

MO

RP

HO

ME

TR

ICIN

DIC

ES

FO

RC

3.B

6-L

IT/L

ITM

ICE

Cor

tical

Can

nula

rL

acun

ar

Ct.T

hC

t.BV

Ca.

VC

a.V

/Ct.T

VN

.Ca

N.C

a/C

t.TV

�Ca.

V�

Lc.

VL

c.V

/Ct.T

VN

.Lc

N.L

c/C

t.TV

�Lc.

V�

Cor

tica

lC

t.Th

0.82

0.76

0.67

0.71

0.55

0.27

0.48

−0.

100.

51−

0.03

−0.

21C

t.BV

0.79

0.61

0.80

0.55

0.24

0.46

−0.

190.

43−

0.18

0.01

Can

nula

rC

a.V

0.96

0.92

0.82

0.50

0.44

−0.

140.

50−

0.03

−0.

30C

a.V

/Ct.T

V0.

850.

860.

550.

45−

0.06

0.54

0.08

−0.

39N

.Ca

0.93

0.33

0.48

−0.

140.

51−

0.06

−0.

22N

.Ca/

Ct.T

V0.

310.

51−

0.08

0.56

0.02

−0.

28⟨C

a.V

⟩-0

.45

−0.

81−

0.47

−0.

790.

15L

acun

arL

c.V

0.78

0.96

0.74

−0.

08L

c.V

/Ct.T

V0.

750.

95−

0.09

N.L

c0.

81−

0.37

N.L

c/C

t.TV

−0.

40⟨L

c.V

⟩

Stat

isti

cally

sign

ific

ant

rva

lues

are

show

nin

bold

(p<

0.05

).Se

ele

gend

for

Tab

les

2,3,

and

5fo

rfu

rthe

rin

form

atio

nco

ncer

ning

nom

encl

atur

eof

the

indi

cate

dm

orph

omet

ric

indi

ces.

TA

BL

E7.

CO

RR

EL

AT

ION

BE

TW

EE

NC

OR

TIC

AL

,CA

NN

UL

AR

,AN

DL

AC

UN

AR

MO

RP

HO

ME

TR

ICIN

DIC

ES

FO

RB

6-L

IT/L

ITM

ICE

Cor

tical

Can

nula

rL

acun

ar

Ct.T

hC

t.BV

Ca.

VC

a.V

/Ct.T

VN

.Ca

N.C

a/C

t.TV

�Ca.

V�

Lc.

VL

c.V

/Ct.T

VN

.Lc

N.L

c/C

t.TV

�Lc.

V�

Cor

tica

lC

t.Th

Ct.B

V0.

88C

annu

lar

Ca.

V0.

680.

58C

a.V

/Ct.T

V0.

490.

390.

96N

.Ca

0.66

0.56

0.89

0.88

N.C

a/C

t.TV

0.36

0.25

0.74

0.84

0.92

⟨Ca.

V⟩

0.08

0.11

0.23

0.23

−0.

07−

0.14

Lac

unar

Lc.

V0.

470.

450.

11-0

.03

−0.

13−

0.22

−0.

25L

c.V

/Ct.T

V−

0.15

−0.

65−

0.23

-0.0

8−

0.16

−0.

01−

0.38

0.38

N.L

c0.

370.

660.

17-0

.06

−0.

21−

0.37

−0.

180.

83−

0.01

N.L

c/C

t.TV

−0.

26−

0.52

−0.

23−

0.15

−0.

34−

0.23

−0.

400.

390.

840.

29⟨L

c.V

⟩0.

13−

0.42

−0.

060.

090.

210.

33−

0.12

0.17

0.63

−0.

400.

11

Stat

isti

cally

sign

ific

ant

rva

lues

are

show

nin

bold

(p<

0.05

).Se

ele

gend

for

Tab

les

2,3,

and

5fo

rfu

rthe

rin

form

atio

nco

ncer

ning

nom

encl

atur

eof

the

indi

cate

dm

orph

omet

ric

indi

ces.

SCHNEIDER ET AL.1564

Moreover, the canals delineated a cone around the femorallong axis described by ⟨Ca.�⟩, whereas single canals runningparallel or perpendicular were observed only rarely. Fi-nally, cannular indices were calculated for the different ana-tomical sites (anterior, posterior, lateral, and medial) andare listed in Table 4. Because a two-way ANOVA for thefactors site and sex did not show any interaction, femaleand male specimens were pooled for the different sites. Thequestion if the cannular indices do depend on the anatomi-cal site was decided by Tukey’s HSD pairwise comparisonsamong site means for the two mouse strains B6-lit/lit andC3.B6-lit/lit. Those site means, which were significantly dif-ferent (p < 0.05) from at least one other site, are printed inbold in Table 4. According to Table 4, the element-basedindices ⟨Ca.V⟩, ⟨Ca.Dm⟩, and ⟨Ca.Le⟩ are independent ofthe site, whereas the mean canal orientation (⟨Ca.�⟩) wassite-dependent for C3.B6-lit/lit and partly for B6-lit/lit. Thisis shown in Fig. 6, where the canal network within the an-terior (A) and posterior (B) cortical bone of a C3.B6-lit/litmouse femur is displayed. The cannular branches are bentmore toward the femurs longitudinal axis for the posteriorsite compared with the anterior site. With the exception ofCa.Sp for B6-lit/lit, the overall morphometric indices wereall dependent on the site, including N.Ca/Ct.TV and Ca.V/Ct.TV. This is particularly pronounced for B6-lit/lit, wherethe maximal N.Ca/Ct.TV and Ca.V/Ct.TV identified at theanterior site were a multiple of the minimal cannular den-sities detected at the posterior site. In contrast, only mod-erate site differences in N.Ca/Ct.TV and Ca.V/Ct.TV werefound for C3.B6-lit/lit.

Cellular level: Figure 4 shows the canal network and theosteocyte lacunar system within the murine cortical femoralbone as assessed by the nCT approach. The abundance ofcell spaces in the cortex deserves closer attention. Theseosteocyte lacunae were mostly plate-like ellipsoids, ori-ented mainly in the longitudinal direction of the femur. Themorphometric analysis for the lacunar indices is given inTable 5. For the cell space analysis, it must be noted that inC3.B6-lit/lit, the mean lacuna volume (⟨Lc.V⟩) was largerthan in B6-lit/lit. On the other hand, lacuna number density(N.Lc/Ct.TV) was proportionally smaller in C3.B6-lit/lit, sothat there is more bone matrix per osteocyte lacuna com-pared with B6-lit/lit mice, whereas sex seemed to have noinfluence on the actual numbers. This led to a lacuna vol-ume density (Lc.V/Ct.TV) that was constant at 1.3% forboth mouse strains (and both sexes). This is in contrast tovolumetric canal density (Ca.V/Ct.TV), which was muchlarger in C3.B6-lit/lit compared with B6-lit/lit.

Correlations: Correlation (Tables 6 and 7) showed that allparameters describing the overall cannular morphometryscaled significantly with bone size, i.e., with cortical thick-ness (Ct.Th) and cortical bone volume (Ct.BV). Amongthose 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 basicstructural elements such as mean canal volume (⟨Ca.V⟩)and mean lacuna volume (⟨Lc.V⟩) were unaffected by sizeand rather constant. Nevertheless, they were unique permouse strain. This was corroborated by the finding thatstrong correlations were detected between number of la-cuna (N.Lc) and lacuna volume (Lc.V) (R � 0.83) as well asbetween N.Ca and Ca.V (R � 0.89) for both B6-lit/lit andC3.B6-lit/lit mice. In general, no linear relations betweenthe morphometric indices describing the osteocyte lacunaeand bone size have been observed.

FIG. 5. Interaction plots for mean canalvolume (⟨Ca.V⟩) of B6-lit/lit and C3.B6-lit/litmice. Because the ⟨Ca.V⟩ is similar for fe-male and male B6-lit/lit mice, there is no realcross-over, and the significant interaction be-tween the factors mouse strain and sex wasqualified as ordinal. The interaction plots ofcanal volume density (Ca.V/Ct.TV) andmean canal length (⟨Ca.Le⟩) behave simi-larly.

FIG. 6. Canal network within the anterior (A) and posterior (B)cortical bone of a C3.B6-lit/lit mouse femur, where the borderlayers represent the extent of the cortical bone. The morphomet-ric parameters describing the 3D extension and orientation of thecanal network vary for different sites (A and B). Note in particu-lar the behavior of the canal penetration within cortical bone fordifferent cortical thicknesses. The data have been assessed byglobal SR �CT at 3.5-�m nominal resolution.

ULTRASTRUCTURAL PROPERTIES IN MURINE CORTICAL BONE 1565

Biomechanics

The mechanical parameters assessed from three-pointbending are given in Table 8. First, Fu, U, and S were allsignificantly different for the two different mouse strains asis shown representatively for Fu in Fig. 7. Second, therewere no significant differences found between female andmale specimens for B6-lit/lit mice (Table 8). However, inthe case of C3.B6-lit/lit, ultimate force (Fu) and work tofailure (U) were significantly increased in male specimenscompared with female mice (Fig. 7). The same trend wasobserved for the stiffness (S) of C3.B6-lit/lit mice, althoughthe sex difference was not statistically significant.

DISCUSSION

3D ultrastructural imaging andquantitative morphometry

Organ level: Femoral mid-diaphysis of C3.B6-lit/lit islarger compared with B6-lit/lit mice.

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

Tissue level: Cannular indices for C3.B6-lit/lit are gener-ally larger compared with B6-lit/lit mice.

On the tissue level, the cortical bone ultrastructure hasbeen often characterized by intracortical porosity. The in-tracortical femoral porosity of rats at different sites wasfound to be increased by several factors, such as immobili-zation,(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 his-tomorphometry(50) and accounted for 0.8–1.2% of the cor-tical bone extension. In this study, we showed that the canalnetwork and the osteocyte lacunar system together (i.e., theintracortical porosity accessible with our techniques) occu-pied 1.5–1.8% of the cortical bone extension (Tables 3 and5). However, we always disentangled the intracortical po-rosity into the canal network and the osteocyte lacunar sys-tem for the whole study and quantified these two ultrastruc-tural phenotypes separately. To our knowledge, the canalnetwork within cortical human bone was quantified onlyonce in 3D by Cooper et al.,(33) using a �CT system at10-�m nominal spatial resolution and subsequent negativeimaging for the canal extraction. Using a combination of SRCT measurements, negative imaging, and direct morpho-metric 3D methods, Martín-Badosa et al.(51) quantified theporosity and the pore diameter within the distal femoraldiaphysis of B6 and C3H mice. Because these authors fixedthe nominal resolution at 6.65 �m, osteocyte lacunae werebeyond the resolution limit, and the porosity was exclu-sively a manifestation of the canal network. Given that,according to our results, the mean canal diameter(⟨Ca.Dm⟩) of femoral murine bone is <10 �m (cf. Table 3),

TABLE 8. BIOMECHANICAL PARAMETERS FOR B6-LIT/LIT AND C3.B6-LIT/LIT MICE

Parameter

B6-lit/lit C3.B6-lit/lit

Female Male Difference (%) Female Male Difference (%)

Fu (N) 9 ± 2 9 ± 1 −2 16 ± 1 18 ± 1 17*U (mJ) 1.6 ± 0.3 1.7 ± 0.2 6 1.7 ± 0.2 2.1 ± 0.2 19*S (N/mm) 85 ± 15 72 ± 20 −16 153 ± 19 178 ± 18 16

Values are mean ± SD.Unpaired t-test with Bonferroni correction: * p(sex) < 0.05.Ultimate force (Fu), work to failure (U), and stiffness (S).

FIG. 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 andmale specimens showed the same strength, male C3.B6-lit/lit re-vealed a significantly increased Fu compared with females. p val-ues of unpaired t-tests with Bonferroni correction are indicatedwith *p < 0.05 and ***p < 0.001. Comparable results were foundfor work to failure (U) and stiffness (S).

SCHNEIDER ET AL.1566

the canal network assessed in the study of Martín-Badosa etal. was partly disconnected and was reduced to bigger ca-nals only. In this study, the cannular indices for C3.B6-lit/litwere generally larger compared with B6-lit/lit mice.Whereas for B6-lit/lit, all cannular indices were indepen-dent of sex, Ca.V/Ct.TV, ⟨Ca.V⟩, and ⟨Ca.Le⟩ each de-pended similarly and significantly on sex for C3.B6-lit/lit, asshown representatively for ⟨Ca.V⟩ in Fig. 5. It is interestingto note that, although Ct.Th and Ct.BV vary for the differ-ent anatomical sites within the diaphysis (data not shown),⟨Ca.V⟩, ⟨Ca.Dm⟩, and ⟨Ca.Le⟩ describing basic elements ofthe canal network within the cortical bone are independentof site, and therefore, do not scale with bone size (cf. Fig. 6).This is consistent with our correlation results where param-eters describing basic elements, such as mean canal volume(⟨Ca.V⟩) and mean lacuna volume (⟨Lc.V⟩), were unaf-fected by size. On the other hand, ⟨Ca.�⟩ and the cannulardensities N.Ca/Ct.TV and Ca.V/Ct.TV were site-dependent. Especially, the respective 5-fold and 4-fold dif-ference in N.Ca/Ct.TV and Ca.V/Ct.TV between the ante-rior and posterior site in B6-lit/lit specimens is noteworthy.An analog site-dependency in Ca.V/Ct.TV or nonlacunarporosity was discovered by Skedros et al.(52,53) in mid-diaphyseal ulnas of immature turkeys, where no secondaryosteons were detected. There, the cortical octants with thelowest nonlacunar porosity corresponded to the bone regionwhere compression occurs during the normal wing-flappingcycle.(54) No information on habitual loading of the mousefemur diaphysis in vivo is available. Nevertheless, if we as-sume that the strain distributions within the femoral mid-diaphysis in mice and dogs(55) are comparable, we can con-clude that for B6-lit/lit the anterior site with predominanttensile strains and the posterior site with prevalent com-pressive strains correspond to regions with high and lowcannular densities for N.Ca/Ct.TV and Ca.V/Ct.TV, respec-tively.

Cellular level: The lacuna volume density is independentof mouse strain and sex.

As for the canal network, the known facts relating toosteocytes and osteocyte lacunae are mainly based on mi-croscopical analysis of histological bone sections and aretherefore intrinsically 2D. Quantitative information re-specting the 3D osteocyte lacunar system within murinecortical bone is rare and has been assessed thus far usingconfocal laser scanning microscopy (CLSM).(56) Neverthe-less, quantitative analysis in 3D remains challenging be-cause CLSM introduces image artifacts, such as signal at-tenuation with increasing depth or image distortion causedby the sample refractive index mismatch.(57) However, os-teocytes and osteocyte lacunae have been characterizedmanifold since Frost developed the methodology for dy-namic bone histomorphometry. For example, Mullender etal. 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 fo-cused on murine specimens, we located N.Lc/Ct.TV be-tween 50,000·mm−3 and 70,000·mm−3 (Table 5). We ob-served that the osteocyte lacunae are mostly plate-likeellipsoids. Similarly, but based on CLSM measurements inhuman femoral bone tissue specimens, McCreadie et al.(60)

reported that an ellipsoidal model is appropriate for mod-eling lacunae. The reported values we found in the litera-ture for osteocyte lacuna size of different species are mainlyinterpolations of 2D data but correspond in the order ofmagnitude to the values of ∼200–300 �m3 for the meanlacuna volume of our murine bone specimens (Table 5). Incomparison, Wang et al.(61) reported on mean lacuna volu-mina in the range of 200–700 �m3 within the tibial proximaldiaphysis of B6 mice, where CLSM in conjunction with aprolated spheroid model for osteocyte lacunae has beenused for quantitative morphometry.

In our experiments, the volumetric density of the mid-diaphyseal intracortical canal network (Ca.V/Ct.TV) waspositively 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, atvarying mean lacuna volume (⟨Lc.V⟩) and lacuna numberdensity (N.Lc/Ct.TV) (Table 5). No consistent relation be-tween the lacunar indices and anatomical site could be es-tablished. Because only two mice from each mouse strainwere analyzed at each single site on the cellular level, wewere not able to conclusively answer the question whetherthere 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 lowerlacuna density, similar to findings of Jordan et al.(62) andBorton et al.(63) for mice and men, respectively. This mightimply that individual osteoblasts formed more bone beforebeing incorporated into the mineralized tissue. In general, itwas proposed that bone resorption and bone remodelingmaintain the cell density within a normal range(64) and that,under circumstances where cell and lacuna size and numberare not well controlled, lacuna density may have a signifi-cant effect on bone matrix stiffness(65) and risk of bonefracture.(64)

Scaling rule: Overall cannular morphometry scales withbone size and lacunar morphometry does not.

We derived a scaling rule for the canal network and theosteocyte 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 andCt.BV). In contrast, ⟨Ca.V⟩, ⟨Ca.Dm⟩, and ⟨Ca.Le⟩ charac-terizing basic element dimensions of the cannular network,as well as all morphometric indices quantifying the osteo-cyte lacunae did not scale with bone size.

Biomechanics

In contrast to the femoral midshaft of the wildtype mousestrains B6 and C3H,(36) ultimate force (Fu), work to failure(U), and stiffness (S) of the little specimens were all reducedbecause of an overall reduction in bone size. Nevertheless,C3.B6-lit/lit showed statistically significant increased me-chanical properties versus B6-lit/lit specimens, very similarto the increased mechanical parameters in C3H comparedwith B6.(36) In the same way, reduced bone size and density,represented by smaller Ct.BV, Ct.Th and Ct.BV/TV (Table2), can account for lowered strength, rigidity, and work tofailure, or in particular, can explain why B6-lit/lit specimens

ULTRASTRUCTURAL PROPERTIES IN MURINE CORTICAL BONE 1567

are mechanically inferior to C3.B6-lit/lit mid-diaphysealfemoral bones. Because for C3.B6-lit/lit mice, Ct.BV/TVwas significantly reduced for males compared with females(Table 2), we would now expect decreased mechanicalcharacteristics for male C3.B6-lit/lit. However, this couldnot be seen in our three-point bending results, where allmechanical parameters were found to be increased in malespecimens compared with female mice. Consequently, thebiomechanical results cannot be explained by consideringexclusively bone phenotypes on the organ level. Here, wesuggest including the tissue level, or in our case the canalnetwork, for the discussion of the mechanical results.Where for B6-lit/lit, all cannular indices were independentof sex, the morphometric indices Ca.V/Ct.TV, ⟨Ca.V⟩, and⟨Ca.Le⟩ were greater for females in comparison with malesin the C3.B6-lit/lit mouse strain (cf. Fig. 5). Consequently,there is evidence that for C3.B6-lit/lit, the larger corticalbone mass is counterbalanced or even outweighed by thelarger canal network in the female mice, leading to in-creased porosity and eventually to decreased mechanicalparameters.

Study limitations

As a caveat for this study, it must be mentioned thatrodent bones, unlike those of higher mammals, rarely un-dergo intracortical remodeling. Therefore, direct compari-son between the canal network in mice cortices and that inhuman bone might be misleading. In the mouse, corticalbone is considered to be unchanged since the time the bonewas originally formed, whereas in humans, cortical bone isextensively remodeled over time. Although these resultsshowed extensive canal void spaces in the cortical diaphysisof the mouse, this model clearly fails to show osteonal struc-tures as we can find them in humans and other larger mam-mals. However, we are convinced that the insight gainedinto the ultrastructure and biomechanics of murine boneprovides the requisite framework for genetic studiess inmice dedicated to a better understanding of the hierarchicalstructural organization in the cortex and its contribution tothe mechanical behavior of bone.

Conclusions

To our knowledge, this is the first study investigatingquantitatively the ultrastructural cortical bone morphom-etry of mice in 3D and in a fully nondestructive fashion.Murine intracortical porosity has been disentangled into thecanal network and the osteocyte lacunar system, as twoultrastructural phenotypes of (murine) cortical bone. Glo-bal and local SR-based CT methods have been designed forquantification and morphometric analysis of these pheno-types. The advantage of SR CT is primarily its high spatialresolution, opening the ultrastructure of bone for nonde-structive and 3D assessment, currently not accessible withconventional �CT desktop systems. New morphometric in-dices have been devised and applied for quantitative char-acterization of the canal network and the osteocyte lacunarsystem in the mid-diaphyseal femoral cortex of the twogenetically 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 sexdifferences independent of GH and IGF-I. Their bone me-chanics has been described and successfully related to themorphometry on the organ and the ultrastructural level.We believe that the morphometric analysis of the ultra-structural phenotypes and the study of the relationshipsbetween phenotypes of bone at different hierarchy levelswill provide new insights in the assessment of bone qualityon all levels of bone hierarchy. In particular, we hypoth-esize that the prediction of bone mechanics can be im-proved in certain diseases and genetic predispositions byincluding ultrastructural bone tissue properties, such as thecanal network and the osteocyte lacunar system.

ACKNOWLEDGMENTS

The authors thank Stefan Heinzer for help during thepreparation of the manuscript. This study was supported bythe Swiss National Science Foundation through the SNFProfessorship 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. NIH 2001 Osteoporosis prevention, diagnosis, and therapy.JAMA 285:785–795.

2. Hui SL, Slemenda CW, Johnston CC Jr 1988 Age and bonemass as predictors of fracture in a prospective study. J ClinInvest 81:1804–1809.

3. Black DM, Cummings SR, Genant HK, Nevitt MC, Palermo L,Browner W 1992 Axial and appendicular bone-density predictfractures in older women. J Bone Miner Res 7:633–638.

4. Nielsen SP 2000 The fallacy of BMD: A critical review of thediagnostic use of dual X-ray absorptiometry. Clin Rheumatol19:174–183.

5. Ciarelli MJ, Goldstein SA, Kuhn JL, Cody DD, Brown MB1991 Evaluation of orthogonal mechanical properties and den-sity of human trabecular bone from the major metaphysealregions with materials testing and computed tomography. JOrthop Res 9:674–682.

6. Müller R 2003 Bone microarchitecture assessment: Currentand future trends. Osteoporos Int 14(Suppl 5):89–99.

7. Turner CH, Cowin SC, Rho JY, Ashman RB, Rice JC 1990The fabric dependence of the orthotropic elastic constants ofcancellous bone. J Biomech 23:549–561.

8. Müller R, Hannan MK, Smith SY, Bauss F 2004 Intermittentibandronate preserves bone quality and bone strength in thelumbar spine after 16 months of treatment in the ovariecto-mized cynomolgus monkey. J Bone Miner Res 19:1787–1796.

9. Mazess RB 1990 Fracture risk—a role for compact-bone. CalcifTissue Int 47:191–193.

10. Spadaro JA, Werner FW, Brenner RA, Fortino MD, Fay LA,Edwards WT 1994 Cortical and trabecular bone contributestrength to the osteopenic distal radius. J Orthop Res 12:211–218.

11. Schaffler MB, Burr DB 1988 Stiffness of compact bone: Effectsof porosity and density. J Biomech 21:13–16.

12. Ferretti JL, Frost HM, Gasser JA, High WB, Jee WS, JeromeC, Mosekilde L, Thompson DD 1995 Perspectives on osteopo-rosis research: Its focus and some insights from a new para-digm. Calcif Tissue Int 57:399–404.

13. Augat P, Reeb H, Claes LE 1996 Prediction of fracture load atdifferent skeletal sites by geometric properties of the corticalshell. J Bone Miner Res 11:1356–1363.

14. Louis O, Boulpaep F, Willnecker J, Van den Winkel P, Os-

SCHNEIDER ET AL.1568

teaux M 1995 Cortical mineral content of the radius assessedby peripheral QCT predicts compressive strength on biome-chanical testing. Bone 16:375–379.

15. Snyder SM, Schneider E 1991 Estimation of mechanical prop-erties of cortical bone by computed tomography. J Orthop Res9:422–431.

16. Saha S, Hayes WC 1977 Relations between tensile impactproperties and microstructure of compact bone. Calcif TissueRes 24:65–72.

17. Currey JD 1979 Changes in the impact energy absorption ofbone with age. J Biomech 12:459–469.

18. Dickenson RP, Hutton WC, Stott JR 1981 The mechanicalproperties of bone in osteoporosis. J Bone Joint Surg Br63:233–238.

19. Currey JD 1988 The effect of porosity and mineral content onthe Young’s modulus of elasticity of compact bone. J Biomech21:131–139.

20. Martin RB, Ishida J 1989 The relative effects of collagen fiberorientation, porosity, density, and mineralization on bonestrength. J Biomech 22:419–426.

21. Barth RW, Williams JL, Kaplan FS 1992 Osteon morphometryin females with femoral neck fractures. Clin Orthop 283:178–186.

22. McCalden RW, Mcgeough JA, Barker MB, Courtbrown CM1993 Age-related changes in the tensile properties of corticalbone. The relative importance of changes in porosity, miner-alization, and microstructure. J Bone Joint Surg Am 75:1193–1205.

23. Yeni YN, Brown CU, Wang Z, Norman TL 1997 The influenceof bone morphology on fracture toughness of the human femurand tibia. Bone 21:453–459.

24. Bell KL, Loveridge N, Power J, Garrahan N, Meggitt BF,Reeve J 1999 Regional differences in cortical porosity in thefractured femoral neck. Bone 24:57–64.

25. Jordan GR, Loveridge N, Bell KL, Power J, Rushton N, ReeveJ 2000 Spatial clustering of remodeling osteons in the femoralneck cortex: A cause of weakness in hip fracture? Bone26:305–313.

26. Feldkamp LA, Goldstein SA, Parfitt AM, Jesion G, Kleereko-per M 1989 The direct examination of three-dimensional bonearchitecture in vitro by computed tomography. J Bone MinerRes 4:3–11.

27. Rüegsegger P, Koller B, Müller R 1996 A microtomographicsystem for the nondestructive evaluation of bone architecture.Calcif Tissue Int 58:24–29.

28. Bonse U, Busch F, Gunnewig O, Beckmann F, Pahl R, DellingG, Hahn M, Graeff W 1994 3D computed X-ray tomography ofhuman cancellous bone at 8 microns spatial and 10(-4) energyresolution. Bone Miner 25:25–38.

29. Peyrin F, Salome M, Cloetens P, Laval-Jeantet AM, Ritman E,Rüegsegger P 1998 Micro-CT examinations of trabecular bonesamples at different resolutions: 14, 7 and 2 micron level. Tech-nol Health Care 6:391–401.

30. Salome M, Peyrin F, Cloetens P, Odet C, Laval-Jeantet AM,Baruchel J, Spanne P 1999 A synchrotron radiation microto-mography system for the analysis of trabecular bone samples.Med Phys 26:2194–2204.

31. Jones AC, Sheppard AP, Sok RM, Arns CH, Limaye A, Aver-dunk H, Brandwood A, Sakellariou A, Senden TJ, MilthorpeBK, Knackstedt MA 2004 Three-dimensional analysis of cor-tical bone structure using X-ray micro-computed tomography.Physica A (Amsterdam) 339:125–130.

32. Bousson V, Peyrin F, Bergot C, Hausard M, Sautet A, LaredoJD 2004 Cortical bone in the human femoral neck: Three-dimensional appearance and porosity using synchrotron radia-tion. J Bone Miner Res 19:794–801.

33. Cooper DM, Turinsky AL, Sensen CW, Hallgrimsson B 2003Quantitative 3D analysis of the canal network in cortical boneby micro-computed tomography. Anat Rec 274B:169–179.

34. Currey JD 2003 The many adaptations of bone. J Biomech36:1487–1495.

35. Beamer WG, Donahue LR, Rosen CJ, Baylink DJ 1996 Ge-netic variability in adult bone density among inbred strains ofmice. Bone 18:397–403.

36. Turner CH, Hsieh YF, Müller R, Bouxsein ML, Baylink DJ,Rosen CJ, Grynpas MD, Donahue LR, Beamer WG 2000 Ge-netic regulation of cortical and trabecular bone strength andmicrostructure in inbred strains of mice. J Bone Miner Res15:1126–1131.

37. Akhter MP, Cullen DM, Pedersen EA, Kimmel DB, ReckerRR 1998 Bone response to in vivo mechanical loading in twobreeds of mice. Calcif Tissue Int 63:442–449.

38. Kodama Y, Dimai HP, Wergedal J, Sheng M, Malpe R,Kutilek S, Beamer W, Donahue LR, Rosen C, Baylink DJ,Farley J 1999 Cortical tibial bone volume in two strains of mice:Effects of sciatic neurectomy and genetic regulation of boneresponse to mechanical loading. Bone 25:183–190.

39. Donahue LR, Guido VE, Rosen CJ, Horton LG, Ackert-Bicknell CL, Bouxsein ML, Shultz KL, Beamer WG 2003 GH/IGF-I independent genetic effects on BMD and skeletal mor-phology are both gender dependent and independent. J BoneMiner Res 18:S123.

40. Stampanoni M, Borchert G, Wyss P, Abela R, Patterson B,Hunt S, Vermeulen D, Rüegsegger P 2002 High resolutionX-ray detector for synchrotron-based microtomography. NuclInstrum Methods Phys A Res 491:291–301.

41. Patterson BD, Abela R, Auderset H, Chen Q, Fauth F, GozzoF, Ingold G, Kuhne H, Lange M, Maden D, Meister D, Patti-son P, Schmidt T, Schmitt B, Schulze-Briese C, Shi M, Stam-panoni M, Willmott PR 2005 The materials science beamline atthe swiss light source: Design and realization. Nucl InstrumMethods Phys A Res 540:42–67.

42. Leung CK, Lam FK 1996 Performance analysis for a class ofiterative image thresholding algorithms. Pattern Recogn29:1523–1530.

43. Hildebrand T, Laib A, Müller R, Dequeker J, Rüegsegger P1999 Direct three-dimensional morphometric analysis of hu-man cancellous bone: Microstructural data from spine, femur,iliac crest, and calcaneus. J Bone Miner Res 14:1167–1174.

44. Parfitt AM, Drezner MK, Glorieux FH, Kanis JA, Malluche H,Meunier PJ, Ott SM, Recker RR 1987 Bone histomorphom-etry: Standardization of nomenclature, symbols, and units. Re-port of the ASBMR Histomorphometry Nomenclature Com-mittee. J Bone Miner Res 2:595–610.

45. Stauber M, Muller R 2006 Volumetric spatial decomposition oftrabecular bone into rods and plates - A new method for localbone morphometry. Bone 38:475–484.

46. Turner CH, Burr DB 1993 Basic biomechanical measurementsof bone: A tutorial. Bone 14:595–608.

47. Jozsa L, Reffy A, Jarvinen M, Kannus P, Lehto M, Kvist M1988 Cortical and trabecular osteopenia after immobilization.A quantitative histological study of the rat knee. Int Orthop12:169–172.

48. Lauritzen DB, Balena R, Shea M, Seedor JG, Markatos A, LeHM, Toolan BC, Myers ER, Rodan GA, Hayes WC 1993 Ef-fects of combined prostaglandin and alendronate treatment onthe histomorphometry and biomechanical properties of bonein ovariectomized rats. J Bone Miner Res 8:871–879.

49. Ibbotson KJ, Orcutt CM, D’Souza SM, Paddock CL, ArthurJA, Jankowsky ML, Boyce RW 1992 Contrasting effects ofparathyroid hormone and insulin-like growth factor I in anaged ovariectomized rat model of postmenopausal osteoporo-sis. J Bone Miner Res 7:425–432.

50. Aguirre JI, Plotkin LI, Stewart SA, Weinstein RS, Parfitt AM,Manolagas SC, Bellido T 2006 Osteocyte apoptosis is inducedby weightlessness in mice and precedes osteoclast recruitmentand bone loss. J Bone Miner Res 21:605–615.

51. Martin-Badosa E, Amblard D, Nuzzo S, Elmoutaouakkil A,Vico L, Peyrin F 2003 Excised bone structures in mice: Imagingat three-dimensional synchrotron radiation micro CT. Radiol-ogy 229:921–928.

52. Skedros JG, Hunt KJ, Hughes PE, Winet H 2003 Ontogeneticand regional morphologic variations in the turkey ulna diaphy-

ULTRASTRUCTURAL PROPERTIES IN MURINE CORTICAL BONE 1569

sis: Implications for functional adaptation of cortical bone.Anat Rec Part A 273:609–629.

53. Skedros JG, Hunt KJ 2004 Does the degree of laminarity cor-relate with site-specific differences in collagen fibre orientationin primary bone? An evaluation in the turkey ulna diaphysis. JAnat 205:121–134.

54. Rubin CT, Lanyon LE 1985 Regulation of bone mass by me-chanical strain magnitude. Calcif Tissue Int 37:411–417.

55. Shahar R, Banks-Sills L, Eliasy R 2003 Stress and strain dis-tribution in the intact canine femur: Finite element analysis.Med Eng Phys 25:387–395.

56. Wang CC, Chahine NO, Hung CT, Ateshian GA 2003 Opticaldetermination of anisotropic material properties of bovine ar-ticular cartilage in compression. J Biomech 36:339–353.

57. Jones CW, Smolinski D, Keogh A, Kirk TB, Zheng MH 2005Confocal laser scanning microscopy in orthopaedic research.Prog Histochem Cytochem 40:1–71.

58. Mullender MG, van der Meer DD, Huiskes R, Lips P 1996Osteocyte density changes in aging and osteoporosis. Bone18:109–113.

59. Mullender MG, Huiskes R, Versleyen H, Buma P 1996 Osteo-cyte density and histomorphometric parameters in cancellousbone of the proximal femur in five mammalian species. J Or-thop Res 14:972–979.

60. McCreadie BR, Hollister SJ, Schaffler MB, Goldstein SA 2004Osteocyte lacuna size and shape in women with and withoutosteoporotic fracture. J Biomech 37:563–572.

61. Wang L, Wang Y, Han Y, Henderson SC, Majeska RJ, Wein-baum S, Schaffler MB 2005 In situ measurement of solutetransport in the bone lacunar-canalicular system. Proc NatlAcad Sci USA 102:11911–11916.

62. Jordan GR, Loveridge N, Power J, Clarke MT, Parker M,Reeve J 2003 The ratio of osteocytic incorporation to bonematrix formation in femoral neck cancellous bone: An en-hanced osteoblast work rate in the vicinity of hip osteoarthritis.Calcif Tissue Int 72:190–196.

63. Borton AJ, Frederick JP, Datto MB, Wang XF, Weinstein RS2001 The loss of Smad3 results in a lower rate of bone forma-tion and osteopenia through dysregulation of osteoblast differ-entiation and apoptosis. J Bone Miner Res 16:1754–1764.

64. Hernandez CJ, Majeska RJ, Schaffler MB 2004 Osteocyte den-sity in woven bone. Bone 35:1095–1099.

65. Yeni YN, Vashishth D, Fyhrie DP 2001 Estimation of bonematrix apparent stiffness variation caused by osteocyte lacunarsize and density. J Biomech Eng 123:10–17.

66. Kohler T, Beyeler M, Webster D, Muller R 2005 Compart-mental bone morphometry in the mouse femur: Reproducibil-ity and resolution dependence of microtomographic measure-ments. Calcif Tissue Int 77:281–290.

Address reprint requests to:Ralph Müller, PhD

Institute for BiomechanicsETH Zürich

Wolfgang-Pauli-Strasse 108093 Zürich, Switzerland

E-mail: ram@ethz.ch

Received in original form December 19, 2006; revised form June12, 2007; accepted June 26, 2007.

SCHNEIDER ET AL.1570