Osteocyte Lacuna Population Densities in Sheep, Elk and Horse Calcanei

Original Paper

Cells Tissues Organs 2005;181:23–37 Accepted after revision: August 18, 2005 DOI: 10.1159/000089966

John G. Skedros

Department of Orthopaedic Surgery, University of Utah, Salt Lake City, Utah , USA

Key Words versus plantar comparisons with the highest Ot.Lc.N/ Artiodactyl calcaneus Bone adaptation Osteocyte B.Ar in dorsal cortices of sheep and elk (p ! 0.0001); but Osteocyte lacunae Osteons this was a statistical trend in the equine calcanei (p = 0.14). There were no consistent transcortical (pericortical to endocortical) differences, and Ot.Lc.N/B.Ar in neutral Abstract axes was not consistently different from dorsal/plantar Osteocytes, the most prevalent cell type in bone, appear cortices. Correlations of Ot.Lc.N/B.Ar with structural and to communicate via gap junctions. In limb-bone diaphyses, it has been hypothesized that these cellular networks have the capacity to monitor habitual strains, which can differ signifi cantly between cortical locations Abbreviations used in this paper of the same bone. Regional differences in microdamage associated with prevalent/predominant strain mode Ar area CFO predominant collagen fi ber orientation (tension, compression, or shear) and/or magnitude may D dorsal represent an important ‘variable’ detected by this net- En endocortical aspect of the cortex work. This hypothesis was indirectly addressed by ex- L lateral amining bones subjected to habitual bending for correla- Lc-Lc estimated lacuna-lacuna distance (m) tions of osteocyte lacuna population densities (n/mm 2 M medial Mi middle aspect of the cortex bone area, Ot.Lc.N/B.Ar) with locations experiencing MC3s third metacarpals high and low strain, and/or prevalent/predominant ten- On osteons sion, compression, and shear. We examined dorsal On.Ar mean area of complete secondary osteons (‘compression’), plantar (‘tension’), and medial/later- On.Ar/T.Ar fractional area of secondary osteonal bone ! 100 al (‘shear’ or neutral axis) cortices of mid-diaphyseal (%, includes central canals in total area) On.N/T.Ar secondary osteon population density (n)/total area sections of calcanei of adult sheep, elk, and horses. (mm2), includes central canals in total area Ot.Lc.N/B.Ar data, quantifi ed in backscattered electron Ot.Lc.N/B.Ar osteocyte lacuna population density (n)/bone area images, were also evaluated in a context of various ad- (mm2), excludes central canals ditional structural and material variables (e.g. % ash, cor- Pe pericortical aspect of the cortex P plantar tical thickness, porosity, and secondary osteon popula- T total tion). Results showed signifi cant differences in dorsal

© 2005 S. Karger AG, Basel John G. Skedros, MD 1422–6405/05/1811–0023$22.00/0 Utah Bone and Joint Center Fax +41 61 306 12 34 5323 South Woodrow Street, Suite 202 E-Mail [email protected] Accessible online at: Salt Lake City, UT 84107 (USA) www.karger.com www.karger.com/cto Tel. +1 801 713 0606, Fax +1 801 713 0609, E-Mail [email protected] material parameters were also poor and/or inconsistent [Skedros et al., 2004]. A goal of the present study was to within or between species. These results provide little or determine if this relationship in deer calcanei and equine no evidence that the number of osteocyte lacunae has a radii is consistent in similarly loaded regions of other functional role in mechanotransduction pathways that bones. Additionally we further evaluate relationships of are typically considered in bone adaptation. Although Ot.Lc.N/B.Ar with characteristics of habitual strain envi- dorsal/plantar differences may be adaptations for preva- ronments, and regional metabolism/remodeling-related lent/predominant strain modes and/or associated micro- activities. This was accomplished by examining Ot.Lc.N/ damage, it is also plausible that they are strongly infl u- B.Ar in calcanei of mature sheep, elk, and horses. enced by differences in the bone formation rates that Cortices within the same mid-diaphyseal transverse produced the tissue in these locations. cross-sections of these calcanei have marked structural Copyright © 2005 S. Karger AG, Basel and material heterogeneity that is correlated with a habitual dorsal/plantar (‘compression/tension’) strain distribution produced during functional loading typical of Introduction weight-bearing activities [Su et al., 1999] (fi g. 1). It has been hypothesized that specifi c features of the non-uni- Results of a companion study of osteocyte lacuna pop- form strain distributions of these bones are important ulation densities per bone area (Ot.Lc.N/B.Ar, n/mm 2 ) in causal infl uences in the attainment of this heterogeneous multiple locations of mid-diaphyseal equine radii and construction [Skedros et al., 2001a, 2004]. These features third metacarpals (MC3s) showed poor correlations with include, among others, strain mode (i.e. tension, com- many mechano-biological and morphological character- pression, and shear) and the corresponding strain magni- istics, including variations in strain magnitude/mode, tude differences, where the highest absolute values of lon- marrow proximity, cortical thickness, mineral content (% gitudinal strains are compressive and dominate in the ash), porosity, secondary osteon population densities, dorsal cortex, and tensile strains dominate in the plantar and mean osteon cross-sectional areas [Skedros et al., cortex. Longitudinal strains are lower in the medial and 2005]. These data also reject the hypothesis that osteo- lateral cortices; however, neutral axis regions generally cytes might be distributed in a manner refl ecting their undergo more prevalent/predominant shear strains [i.e., putative role as ‘operational’ networks that sense micro- where shear strains are not generally predominated by damage, which can differ between ‘tension’ and ‘com- longitudinal (i.e. normal strains) tension or compression] pression’ areas in formation frequency and/or other char- and maximal strains that are 20–30° oblique to the long acteristics (e.g. differences in microcrack size and shape). axis of the bone. [While shear strains exist throughout the This is because differences in Ot.Lc.N/B.Ar, although sta- bone cross-section, this is the prevalent strain mode along tistically signifi cant between the ‘tension’ and ‘compres- the neutral axis since bending loads, superimposed on sion’ areas, were opposite in these two equine bones (i.e. torsional loading, would eclipse the shear strains in the relatively higher Ot.Lc.N/B.Ar in the ‘compression’ area compression/tension (dorsal/plantar) regions.] In adult of equine radii, and relatively lower Ot.Lc.N/B.Ar in the mule deer calcanei the dorsal cortex is signifi cantly thick- ‘compression’ area of equine MC3s). However, it is pos- er, and has greater population density of secondary os- sible that the habitual loading environment of the MC3s teons, smaller osteon cross-sectional areas, osteons with is suffi ciently complex to obscure tension versus compres- quasi-circular to circular shapes, more oblique-to-trans- sion Ot.Lc.N/B.Ar differences. This complexity is most verse collagen fi ber orientation (CFO), more highly cross- obvious in the equine MC3 when increased torsional linked collagen fi bers, lower porosity, and lower fraction- stresses associated with fast gait speeds cause the neutral al area of secondary bone compared to the plantar ‘ten- axis to rotate 30–45° [Gross et al., 1992; Skedros et al., sion’ cortex [Gunasekaran et al., 1991; Skedros et al., 1996; Nunamaker, 2001]. It is possible that correlations 1994a, 1997, 2001a, 2001b, 2004; Skedros and Brady, of Ot.Lc.N/B.Ar with regional strain, histology and/or 2001]. In contrast to the dorsal and plantar cortices, the other morphologic or mechano-biological characteristics medial and lateral cortices (‘neutral axis’ regions) exhibit might be consistent and signifi cant in other bones that are fewer osteons, which are also often oval in cross-section relatively more simply loaded. This possibility appears to (suggesting oblique osteon orientation) with highly trans- be supported by data in calcanei of mature wild deer verse CFO, suggesting adaptation to prevalent shear showing signifi cantly greater Ot.Lc.N/B.Ar in the dorsal strains and obliquely oriented principal strains [Petrtýl et (‘compression’) cortex than the ‘tension’ (plantar) cortex al., 1996; Su et al., 1999]. The organization of the plantar

24 Cells Tissues Organs 2005;181:23–37 Skedros

Fig. 1. Deer calcaneus in lateral view showing typical loading along the Achilles ten- Achilles don (large arrow) producing compression tendon (C ) and tension (T ) on dorsal and plantar Superficial cortices, respectively. This loading regime digital flexor is similar in all of the calcanei. The cross- tendon section shows image locations and the Tibia C 0% oblique neutral axis (NA), the location of C which is based on empirical data derived Dorsal from multiple triple stacked-rosette strain T gauges as reported in an ex vivo study of Pericortical 50% Middle functionally loaded sub-adult and adult Endocortical deer calcanei [Su et al., 1999]. This study used multiple loading regimes to simulate a variety of functional loading conditions, in- 100% cluding turning and running. The dorsal/ Medial Lateral Plantar plantar (compression/tension) strain distri- Metatarsal ligament bution was found to be highly consistent in NA these various loading conditions, probably reﬂ ecting the high degree of constraint im- Plantar posed by the talo-calcaneal articulation. It 1 cm 1 cm is not known if the neutral axis is similarly T obliquely oriented in the calcanei of the other species.

‘tension’ cortex of deer calcanei is also distinct, with its tially adaptable characteristics that are linked with bone increased resorption spaces, more longitudinal CFO, and remodeling (e.g. osteon population densities, porosity, larger, younger (lower mineralization), and more irregu- mineral content); (2) strong correlation of spatial varia- larly shaped secondary osteons – suggesting an increased tions in Ot.Lc.N/B.Ar with the increased renewal activity remodeling rate compared to the other cortices 1 , in addi- or metabolic demands associated with the plantar liga- tion to adaptation for tension strain mode (i.e. longitudi- ment insertion (pericortical margin of the plantar cortex) nal CFO). Most of these differences are also present in and marrow proximity (endocortical regions); (3) in- calcanei of adult sheep, elk, and horses. creased Ot.Lc.N/B.Ar in the actively renewing plantar In view of these comprehensive data in deer calcanei, cortices or, by contrast, Ot.Lc.N/B.Ar will be highest in we hypothesized that sheep, elk, and horse calcanei would the dorsal cortex, refl ecting the putative role of these cells show: (1) strong correlation of spatial variations in osteo- in detecting microdamage differences between tension cyte lacuna population densities (Ot.Lc.N/B.Ar) with re- (also lower strain) and compression (also higher strain) modeling activity and one or more of the myriad, poten- environments [Skedros et al., 2005], and (4) increased Ot.Lc.N/B.Ar in endocortical regions (compared to other intra-cortical regions) of each cortex, perhaps in associa- tion with bioactive molecules from the nearby marrow, which may activate regional remodeling [Frost, 1998; Martin, 2000]. 1 Although this interpretation seems most parsimonious in view of available Correlations of Ot.Lc.N/B.Ar with regional variations data, it is constrained by the lack of dynamic measures of remodeling activity (e.g. fl uorochrome labels). For example, it is plausible that small second- in mean osteon cross-sectional areas were also examined. ary osteons are indicative of a history of high activation frequency [Martin et This inquiry stems from observations in a previous study al., 1998]. Resorption spaces only refl ect the most recent remodeling events. of developing mule deer calcanei showing increased Therefore, at one time the activation frequency may have actually been higher in the dorsal as opposed to the plantar cortex. Nevertheless, data available from Ot.Lc.N/B.Ar and signifi cantly reduced osteon cross-sec- the adult tissues examined in the present study suggest that even if the dorsal tional areas in the dorsal cortex [Skedros et al., 2004]. cortex at one time exhibited a higher activation frequency, the plantar cortex was more actively remodeling (higher resorption spaces and new remodeling Suggested mechanisms that might explain this relation- events) at the time of necropsy. ship are also discussed.

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 25 Calcanei Sheep 100% 60% 0% D

M L ST

100% Elk 60% 0% D

M L

100% 60% 0%

Horse D

M L

Fig. 2. Lateral view of calcanei of a sheep, elk, and horse: the cross-sections of each ST bone show the locations (in ‘gray’) where microscopic analyses were conducted. ST = P Sustentaculum talus (at the medial aspect of each bone); 0% = distal ‘end’ of each 1 cm bone.

Material and Methods a highly consistent distribution of net compression in the dorsal cortex and net tension in the plantar cortex during 80% of stance The Calcaneus Model, Specimens, Specimen Preparation and phase. There are biomechanical data also suggesting that in vivo Analysis strain distributions are similar at equivalent anatomical locations Available in vivo strain gauge data suggest that, during func- of sheep, deer, and horse calcanei [Badoux, 1987; Skedros et al., tional loading, the artiodactyl (sheep) calcaneus (fi g. 2) behaves like 1997]. a short-cantilevered beam with longitudinal compression and ten- One calcaneus was obtained from each of seven Standardbred sion strains predominating in opposing dorsal and plantar cortices, horses with no history of racing or race training, seven wild-shot respectively [Lanyon, 1974; Su et al., 1999]. Corroborative ex vivo North American elk (Cervus elaphus), and seven domesticated analyses in deer calcanei reported in detail in previous studies [Su, sheep ( Ovis aries, breed is crossed Suffolk/Hampshire and Ram- 1998; Su et al., 1999] include a variety of loading regimes to esti- bouillet; fi g. 2). All bones were from adult animals that had been mate strain distributions. These data were obtained in both sub- used in a previous study [Skedros et al., 1997]. The bones were adult and adult deer, and used up to seven stacked triple-rosette randomly selected, resulting in an approximately even number of strain gauges, including one beneath the plantar ligament. Off-axis left and right bones. The elk were males, had shed the periosteal loading (with the Achilles tendon loaded 5° medially to the sagittal cover of their antlers, and were obtained from their natural habitat plane of the bone and 5° lateral to the sagittal plane of the bone) has (northern Utah) in a fall hunting season. The equine calcanei were also been accomplished in order to estimate affects of turning and from mixed sexes that had been set to pasture. Although specifi c jumping during ambulation [Su, 1998]. These results demonstrate ages of several of the horses were not known, the animals were be-

26 Cells Tissues Organs 2005;181:23–37 Skedros

tween 2 and 9 years old. The sheep were females and approximately 2 years old. The sheep and horses had been kept in large pastures. Skeletal maturity of all of the bones was confi rmed by gross exam- ination showing co-ossifi cation of the sagitally cut distal calcaneal epiphyseal growth plate. None of the animals had evidence of skeletal disease and all were sacrifi ced for reasons other than limb lame- ness. Unstained, undecalcifi ed transverse segments, 4–5 mm thick, were cut from each calcaneus at 60% of shaft length. These segments were embedded in polymethyl methacrylate using conven- tional methods [Emmanual et al., 1987]. The proximal face of each segment was ground, polished, and prepared for backscattered electron imaging using a scanning electron microscope (JEOL, Pea- body, Mass., USA) [Skedros et al., 1993a, 1993b]. One 50 ! BSE image representing 3.41 mm 2 (2.23 ! 1.53 mm) was obtained in each transcortical ‘region’ [pericortical (Pe), middle cortical (Mi), endocortical (En)] of the dorsal cortices of each specimen. In many cases only two images could be obtained in the c thinner plantar cortices. These images were taped together and sub- b divided into Pe, Mi, and En regions. The images excluded the circumferential lamellar bone in the dorsal and plantar cortices. One image was also taken within each medial and lateral ‘neutral axis’ a cortex. Each backscattered electron micrograph was developed on ax + bx + cx 3-D Lc-Lc distance = high-resolution Polaroid fi lm. Osteocyte lacunae were manually 26 counted in these images by trained technicians and confi rmed by a = 6 the principal investigator who independently analyzed a large sub- b = 8√3 set of images. The images were randomly assorted and the techni- c = 12√2 1 cians were blinded to cortical location and to the hypotheses of the x = 2-D Lc-Lc distance = study. Intra- and interobserver error was below 81.5%. In each √Ot.Lc.N/B.Ar image, the population density of osteocyte lacunae per bone area (Ot.Lc.N/B.Ar; n/mm 2 ) was calculated by dividing total number of lacunae by bone area (excluding central canals and other vascular Fig. 3. Three-dimensional lattice used to estimate three-dimen- canals and non-lacuna porosity). Lacunae with viable and non-vi- sional (3-D) mean osteocyte Lc-Lc distances from two-dimension- able osteocytes could not be determined. al (2-D) Ot.Lc.N/B.Ar data. The equations used for these estimates In each image, a mean distance between osteocyte lacunae was are also shown. also estimated. This was accomplished by using the formula shown in fi gure 3, which allowed for both estimates of two-dimensional and three-dimensional lacuna-lacuna (Lc-Lc) distances. This method produced similar values for distances between two neighboring imity), and (2) for correlations with On.Ar/T.Ar, which would be lacunae reported by Weinbaum et al. [1994]. expected if remodeling produces bone with different cell densities Additional characteristics were also used in correlation analy- than the antecedent primary bone, and if there are differences in ses. These characteristics, reported in previous studies of the same the amount of remodeling. Alternatively, remodeling may produce bones [Skedros et al., 1997], include: (1) secondary osteon popula- regional differences in osteocyte spatial heterogeneity if secondary tion density (On.N/T.Ar), (2) fractional area of secondary bone osteon ‘types’ differ between cortical locations. This is plausible in (On.Ar/T.Ar), (3) porosity, (4) estimated mean cross-sectional area these calcanei since cell densities are known to differ signifi cantly of osteons in each image (calculated from mean dorsal and plantar between parallel-fi bered secondary osteons (prevalent in the ‘ten- osteon diameters and assuming circularity) and minimum-to-max- sion’ cortices of these bones) and lamellar (‘alternating’) osteons imum chord ratio which may indicate osteon obliquity with respect (prevalent in the ‘compression’ cortices of these bones). (As dis- to the long axis of the bone, (5) population density of new remodel- cussed below, these osteon distinctions are consistent with the ing events [number of resorption spaces and newly forming osteons scheme of Marotti [1996, and pers. commun.].) Support for this per total (T) square millimeter area (Ar, n/mm 2 , includes central possibility derives from a study where Ferretti et al. [1999] quanti- canals in T.Ar)], (6) cortical thickness of the dorsal, plantar, me- fi ed signifi cant differences in osteocyte densities in these two sec- dial, and lateral cortices, and (7) percentage of ash of adjacent dor- ondary osteon types in limb-bone cortices of adult animals of vari- sal, plantar, medial, and lateral segments. ous species (frog, sheep, dog, cow, horse, and human). Spatial heterogeneity of osteocyte lacunae was estimated by cal- A multiple-comparison ANOVA model was used for statistical culating the coeffi cient of variation for each intracortical region in analysis. Pair-wise comparisons between cortical locations (D = each species. This was accomplished by dividing the standard de- dorsal, P = plantar, M = medial, L = lateral) and transcortical re- viation by the mean [Vashishth et al., 2000]. Spatial heterogeneity gions (endocortical, middle cortical and pericortical) of each species was examined: (1) in the context of strain mode and transcortical were assessed for statistical signifi cance ( ^ 0.05) using Fisher’s region (which correlates with strain magnitude and marrow prox- post-hoc protected least-signifi cance difference test [Sokal and

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 27 Calcanei 800 52 53 Horse Elk 750 Sheep 56 59 56 700

) 58 2 650

600

550 Ot.Lc.N/B.Ar (n/mm 500

450

400 Dorsal Plantar Medial Lateral a Cortex

0.18 Horse Elk Sheep 0.14

0.10

Fig. 4. Histograms of Ot.Lc.N/B.Ar ( a ) and Spatial heterogeneity 0.06 spatial heterogeneity of osteocyte lacunae ( b) in cortices of all three species. The numbers above the dorsal and plantar data bars 0.02 in a are mean 3-D Lc-Lc distances in mi- Dorsal Plantar Medial Lateral crons. In a the Ot.Lc.N/B.Ar data are shown b Cortex as means and standard deviations.

Rohlf, 1995] (StatView, version 5.0, SAS Institute, Cary, N.C., table 2). The cortex-to-cortex comparisons showed sig- USA). Pearson correlation coeffi cients (r values) for various com- nifi cant differences or statistical tendencies in Ot.Lc.N/ parisons were determined within each group. B.Ar in all but the plantar (‘tension’) versus medial cortex (p 6 0.3). Comparisons among intracortical regions (Pe vs. Mi Results vs. En, in D and P cortices) showed no statistically signifi cant differences in the equine calcanei (table 3). Equine (Horse) Calcanei There were no statistically signifi cant differences in Mean osteocyte lacuna population densities (Ot.Lc.N/ the spatial heterogeneity of osteocyte lacunae between the B.Ar; n/mm 2 ) for each of the cortical locations of each four cortical locations (D, P, M, L; fi g. 4b). These data species are shown in fi gures 4a and 4 b and table 1. In the reject the hypothesis that the prevalence of lamellar os- equine calcanei, Ot.Lc.N/B.Ar is higher in the dorsal teons in the ‘compression’ area would result in lower spa- ‘compression’ cortex when compared to the plantar ‘ten- tial heterogeneity. Similarly, there were no signifi cant dif- sion’ cortex, but in contrast to sheep and elk (see below) ferences in transcortical spatial heterogeneity between this difference was not statistically signifi cant (p = 0.14, the Pe, Mi, and En regions in all possible comparisons.

28 Cells Tissues Organs 2005;181:23–37 Skedros

Correlations exceeding r = 0.500 are shown in ta- Table 1. Osteocyte lacuna population density data: cortex and ble 4; Ot.Lc.N/B.Ar did not exhibit any correlations ex- transcortical region data (means 8 standard deviations) ceeding this absolute r value in the equine calcanei ( ta- Cortex Ot.Lc.N/B.Ar, n/mm2 ble 4). Spatial heterogeneity of osteocyte lacunae corre- horse elk sheep lated with r 1 0.5 only with On.Ar/T.Ar data from medial and lateral cortices. The estimated three-dimen- Dorsal 640871 732855 710865 sional mean distance between two neighboring lacunae Plantar 5998110 644862 609864 was 56 8 3 m in the dorsal and 59 8 7 m in the plan- Medial and lateral 536869 661834 578845 tar cortex (p = 0.11; fi g. 4a). Medial 562897 679821 546854 Lateral 510853 643850 610855 Cervine (Elk) Calcanei Dorsal In the cervine calcanei, Ot.Lc.N/B.Ar was signifi cant- Pericortical 656859 760845 737861 ly higher in the dorsal ‘compression’ cortex than in the Middle 660869 728814 705864 8 8 8 plantar ‘tension’ cortex (p ! 0.0001; fi g. 4a, table 2). Sig- Endocortical 605 79 709 27 688 69 nifi cant differences were also found between the dorsal Plantar cortex and each of the medial and lateral cortices. There Pericortical 6248155 6548103 574870 8 8 8 were no signifi cant differences between the plantar, me- Middle 595 86 637 41 668 21 Endocortical 580886 640824 611851 dial, and lateral cortices. Statistically signifi cant differences are shown in tables 2 and 3.

Table 2. One-way ANOVA (p values of paired comparisons) of Ot.Lc.N/B.Ar data: dorsal (‘compression’), plantar (‘tension’), medial, and lateral cortices

Dorsal Plantar vs. plantar vs. medial vs. lateral vs. medial vs. lateral

Horse <0.14 (6.8%)a <0.04 (12.2%)* 0.001 (20.3%)* 0.3 (6.2%) <0.02 (14.9%)* Elk <0.0001 (13.7%)* <0.03 (7.2%)* 0.0003 (12.2%)* 0.13 (–5.4%) >0.5 (0.2%) Sheep <0.0001 (16.6%)* <0.0001 (23.1%)* 0.0004 (14.1%)* 0.02 (10.3%)* >0.5 (–0.2%)

a Numbers in parentheses are the percent differences of the paired comparisons; percent differences are calculated as: [(compression – medial or lateral)/(compression)] !100; [(tension – medial or lateral)/(tension)] !100. Statistically signiﬁ cant values are italicized.

* p ^ 0.05.

Table 3. One-way ANOVA (p values of paired comparisons) of regional Dorsal (‘compression’) regions Plantar (‘tension’) regions Ot.Lc.N/B.Ar data pericortical vs. middle vs. pericortical vs. middle vs. middle endocortical middle endocortical

Horse >0.5 >0.16 >0.5 >0.5 Elk >0.3 >0.5 >0.5 >0.5 Sheep >0.4 >0.5 >0.02 >0.12

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 29 Calcanei Table 4. Pearson’s correlation coeffi cients > 0.5 lence of lamellar osteons in the ‘compression’ area would result in lower spatial heterogeneity. Similarly, there were Location Ot.Lc.N/B.Ar Cortical thickness no signifi cant differences in transcortical spatial hetero- n/mm2 cm geneity between the Pe, Mi, and En regions in all possible Horse comparisons. All none On.N/T.Ar (0.62)* Correlation analyses of Ot.Lc.N/B.Ar with the other Dorsal, plantar none % ash (0.76)* characteristics showed two high and several moderate On.N/T.Ar (0.72)* correlations (table 4). Spatial heterogeneity of osteocyte Rs.N/T.Ar (–0.51) porosity (–0.78)* lacunae showed several moderate-to-high correlations 1 On.Ar (–0.57)* (r 0.5 ) with the microstructural variables (– indicates Dorsal none none a negative correlation, * p ! 0.05): On.Ar/T.AR (*), po- Plantar none none rosity, On.N/T.Ar (–, *), Rs.N/T.Ar in the dorsal cortex; Medial, lateral none none porosity, On.N/T.Ar (–, *), On.Ar (*) in the plantar cor- Elk tex; On.Ar/T.Ar in the medial and lateral cortices; All cortical thickness (0.63)* Ot.Lc.N/B.Ar (0.63)* On.N/T.Ar (–, *) in all cortices. The estimated mean Dorsal, plantar % ash (0.63)* % ash (0.90)* three-dimensional distance between neighboring lacunae cortical thickness (0.82)* On.N/T.Ar (0.77)* 8 8 On.Ar/T.Ar (–0.60)* was 52 2 m in the dorsal and 56 3 m in the plan- porosity (–0.65)* tar cortex (p ! 0.0001; fi g. 4a). On.Ar (–0.66)* Ot.Lc.N/B.Ar (0.82)* Ovine (Sheep) Calcanei Dorsal On.N/T.Ar (0.57)* none In the ovine calcanei, Ot.Lc.N/B.Ar was signifi cantly On.Ar/T.Ar (0.65)* Plantar On.N/T.Ar (–0.64)* none higher in the dorsal ‘compression’ cortex than in the plan- On.Ar (0.72)* tar ‘tension’ cortex (p ! 0.0001; fi g. 4a, table 2). Signifi - Medial, lateral none none cant differences were found in all other comparisons, ex- Sheep cept plantar (‘tension’) versus lateral (p = 0.5). All cortical thickness (0.70)* Ot.Lc.N/B.Ar (0.70)* Comparisons among intracortical regions (Pe vs. Mi Dorsal, plantar On.Ar/T.Ar (–0.69)* % ash (0.72)* vs. En, in D and P cortices) showed only one statistically porosity (–0.53) On.Ar/T.Ar (–0.71)* signifi cant difference in the ovine calcanei: Pe vs. Mi in On.Ar (–0.58)* porosity (–0.77)* the P ‘tension’ cortex (table 3). cortical thickness (0.65)* Ot.Lc.N/B.Ar (0.65)* Dorsal On.Ar/T.Ar (–0.51)* none There were no statistically signifi cant differences in On.Ar (–0.61)* the spatial heterogeneity of osteocyte lacunae between the Plantar none none four cortical locations (D, P, M, L) in the ovine calcanei Medial, lateral Rs.N/T.Ar (–0.68)* On.N/T.Ar (–0.60)* (fi g. 4 b). These data reject the hypothesis that the preva- porosity (–0.64)* porosity (–0.53) lence of lamellar osteons in the ‘compression’ area would On.Ar (0.67)* result in lower spatial heterogeneity. Similarly, there were

* p < 0.05. Rs.N/T.Ar = Number of resorption spaces and new- no signiﬁ cant differences in transcortical spatial hetero- ly forming osteons per T.Ar (n/mm2, includes central canals in geneity between the Pe, Mi, and En regions in all possible T.Ar). comparisons. Correlation analyses of Ot.Lc.N/B.Ar with the other characteristics showed two high and several moderate correlations (table 3). Spatial heterogeneity of osteocyte lacunae showed only a few moderate-to-high correlations

(r 1 0.5 ) with some of the microstructural variables, but Comparisons among intracortical regions (Pe vs. Mi these were not statistically signifi cant (– indicates a nega- vs. En, in D and P cortices) showed no statistically sig- tive correlation): On.Ar/T.Ar (–), On.N/T.Ar (–) in the nifi cant differences (table 3). dorsal cortex; On.Ar/T.Ar, On.N/T.Ar, Rs.N/T.Ar in the There were no statistically signifi cant differences in plantar cortex. The estimated mean three-dimensional the spatial heterogeneity of osteocyte lacunae between the distance between neighboring lacunae was 53 8 2 m in four cortical locations (D, P, M, L) in the cervine calcanei the dorsal and 58 8 3 m in the plantar cortex (p ! (fi g. 4b). These data reject the hypothesis that the preva- 0.0001; fi g. 4a).

30 Cells Tissues Organs 2005;181:23–37 Skedros

Interspecies Differences in Osteocyte Lacuna gions), active renewal (plantar cortex) and ligament inser- Densities tion proximity (pericortical region of plantar cortex); Interspecies differences are apparent in the statistical- (3) issue constraints (primary vs. secondary bone) and ly signiﬁ cant differences between them (p ! 0.01; means spatial heterogeneity of osteocyte lacunae; (4) correla- 8 SD of On.Ar/T.Ar data from all regions analyzed: tions with mean cross-sectional osteon areas and (5) cor- sheep 640 8 85, elk 681 8 68, horse 599 8 97). tical thickness and the concept of ‘stressed volume’.

Strain-Mode- and Magnitude-Related Associations; Discussion Load Predictability and Habitual Bending In all three species, strain mode was correlated with Results of this study typically showed inconsistent Ot.Lc.N/B.Ar – the dorsal ‘compression’ cortices having correlations between Ot.Lc.N/B.Ar and the various higher Ot.Lc.N/B.Ar than the plantar ‘tension’ cortices. structural and material characteristics. Spatial heteroge- Although a similar tension vs. compression pattern is re- neity of osteocyte lacunae was also inconsistently corre- ported in horse radii in a companion study [Skedros et lated with On.Ar/T.Ar. Also in confl ict with one of our al., 2005], it is not known if this pattern can be general- hypotheses (No. 3), the plantar cortices, where remodel- ized to other similarly loaded bones. Additionally, in ing is more active, exhibited lower Ot.Lc.N/B.Ar than these equine radii the medial and lateral cortices, where the dorsal cortices. This dorsal/plantar difference was longitudinal strains are presumably low and shear strains statistically signifi cant in sheep and elk calcanei, but was are prevalent/predominant, had lacuna densities that are only a statistical tendency in the equine bones (p = 0.14). similar to the plantar ‘compression’ cortices of these This lack of statistical signifi cance may be the result of bones. The medial/lateral (‘neutral axis’) cortices of the inadequate statistical power, and/or due to our inability calcanei had lowest Ot.Lc.N/B.Ar in only two of the spe- to control the age range of the horses. This latter sugges- cies (table 1). Furthermore, the companion study showed tion implies that age-related changes in lacuna density unexpected (i.e. opposite) differences between ‘tension’ might increase regional variations (as shown in human and ‘compression’ regions of equine third metacarpals bone by Vashishth et al. [2000]), hence making it diffi cult (MC3s). But this result may be confounded by the com- to discern signifi cant differences with the sample size paratively more prevalent torsional loading of this bone, (n = 7). Another limitation of this study is our inability which may, while enhancing shear-related adaptations, to determine the percentage of lacuna with viable osteo- tend to obscure or eliminate what might be considered cytes. However, our specimens were generally from rela- obvious adaptations for tension and compression in tively young adults, and our observations did not reveal bones that are relatively more simply loaded [Skedros et regions with lacunae that were plugged with hypermin- al., 2003c; Skedros and Hunt, 2004]. eralized tissue, which has been associated with aging, Although in the calcanei it was found that Ot.Lc.N/B.Ar ischemia, or necrosis [Frost, 1960; Currey, 1964; Korn- also positively correlates with areas that receive higher blum and Kelly, 1964; Jowsey, 1966; Stout and Sim- strains (i.e. dorsal cortex) and lower densities were ob- mons, 1979; Parfi tt, 1993]. In human bone, it has also served in areas that receive lower strains (i.e. plantar cor- been suggested that extensive death of osteocytes may tex), other fi ndings are inconsistent with the hypothesis not occur in osteons because of their younger bone age that Ot.Lc.N/B.Ar correlates with strain magnitudes. For when compared to interstitial/primary bone [Qiu et al., example, as noted above, the neutral axes (i.e. medial and 2003]. These observations, coupled with the estimated lateral cortices) did not consistently have the lowest Ot.Lc. ‘younger’ ages of the animals used in the present study, N/B.Ar. Additionally pericortical-to-endocortical (high- suggest that there are a relatively small percentage of to-low strain, respectively) strain-related variations in dead osteocytes. Ot.Lc.N/B.Ar were also not detected. (Absence of associa- In order to consider the possibility that the Ot.Lc.N/ tions with marrow proximity and plantar ligament inser- B.Ar differences shown between some of the cortical location are considered below.) The results of the companion tions have signifi cant mechano-biological implications, study also did not reveal consistent relationships with data from the present study are considered in the follow- transcortical variations in strain magnitude in equine radii ing contexts: (1) predominant strain mode, magnitude and MC3s. These inconsistent correlations with strain and putative associations with microdamage morphol- mode and/or magnitude question the reliability of using ogy incidence; (2) marrow proximity (endocortical re- Ot.Lc.N/B.Ar for inferring a history of habitual bending.

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 31 Calcanei Microdamage Implications and Mechanical and her overview of previous studies suggesting that Relevance osteocyte lacunae can infl uence microdamage forma- In the companion study we considered the hypothesis tion differently in specifi c strain modes [Reilly, 2000, of Vashishth et al. [2000] that regional/strain-related p. 1131]: variations in microdamage incidence and/or other micro- “Until recently, most research on damage created in damage characteristics might require differences in the tension concerned microcracks easily visible under bright- osteocyte ‘networks’, communicating via gap junctions, fi eld light microscopy [Huja et al., 1999]. These types of for localizing microdamage. In this context, regional vari- microcracks have been called ‘linear’ by some authors ations in osteocyte densities, if present, might refl ect en- [Vashishth et al., 2000]. They have been shown to form hancements in their capacity to detect the microdamage preferentially at cement lines and interlamellar bound- produced in different strain environments. Although aries [Carter and Hayes, 1976; Schaffl er et al., 1989; there are data suggesting that this interpretation is plau- Courtney et al., 1996; Vashishth et al., 1997] where one sible [Donahue, 2000; Tami et al., 2003], the rate of bone might expect planes of weakness. Microdamage created apposition during modeling and remodeling may more in torsion [i.e. , prevalent shear] has also been shown to strongly infl uence regional osteocyte densities, suggesting form at planes of weakness [Jepsen et al., 1999]. In previ- that variations in Ot.Lc.N/B.Ar are epiphenomena in ous microdamage investigations only compression crack- some cases. This possibility is discussed further in the ing has been shown to be infl uenced by (osteocyte) lacu- companion study [Skedros et al., 2005]. However, a ha- nae [Carter and Hayes, 1976].” bitual non-uniform strain distribution must be accom- Thus, habitual compression environments may be modated by what we have called ‘strain-mode-specifi c’ where modifi cations in Ot.Lc.N/B.Ar are most benefi cial. adaptations of the local bone matrix, since failure to do This suggestion has been corroborated in recent mechan- so could result in deleterious increases in microdamage ical testing studies of equine MC3 diaphyseal cortices [Skedros et al., 2003a]. It has been hypothesized that del- loaded in strain-mode-specifi c tension and compression eterious increases are avoided by modeling/remodeling- [Skedros et al., 2003d]. mediated enhancements in fatigue resistance, toughness and/or energy absorption (e.g. by introducing cement Marrow Proximity and Active Renewal, and lines and/or reorienting predominant CFO) [Skedros et Ligament Insertion Proximity al., 2003a, 2004]. Among various modifi able material Contrary to one of our hypotheses (No. 2), there were characteristics, predominant CFO appears to be most no correlations between Ot.Lc.N/B.Ar and marrow prox- consistently associated with a locally prevalent/predomi- imity (i.e. endocortical regions). In turn, there was also nant strain mode in various bones that have been studied no correlation between Ot.Lc.N/B.Ar and active renewal [Mason et al., 1995; Skedros et al., 1996, 2000, 2003c, (i.e. population densities of new remodeling events) in the 2004]. In turn, studies of cortical bone organization in plantar cortex compared to the other cortical locations. developing mule deer calcanei suggest that Ot.Lc.N/B.Ar However, the endocortical regions examined in this study and predominant CFO are not correlated [Skedros et al., may not be analogous to the area of active renewal that 2004]. can occur along the endocortical regions of limb bones of Data from mechanical tests suggest that the dorsal- humans with osteoporosis or advanced age [Smith and plantar differences in predominant CFO and Ot.Lc.N/ Walker, 1964; Atkinson, 1965; Ruff and Hayes, 1988]. B.Ar that occur in mule deer calcanei are mechanically This is because, according to criteria used to defi ne the adaptive. For example, predominantly longitudinal CFO endocortical region in the present study, larger vascular enhances material properties of the plantar cortex in ten- spaces were avoided – along endosteal margins of osteo- sile loading (i.e. its habitual loading mode) [Skedros et al., porotic or aged bones such ‘larger’ vascular spaces are 2003b]. By contrast, Ot.Lc.N/B.Ar and osteonal charac- prevalent [Amtmann, 1971; Feik et al., 2000; Bousson et teristics (e.g. mean area and shape of secondary osteons, al., 2001]. and On.N/Ar) relatively more strongly infl uence material The absence of signifi cant Ot.Lc.N/B.Ar differences properties of the dorsal cortex in compressive loading (i.e. between the endocortical region and other regions of the its habitual loading mode). This difference may refl ect plantar cortex in all species also rejects the hypothesis that fundamental differences in microdamage formation/ increased osteocyte densities are associated with the in- morphology in these two loading modes. Support for this sertion of the plantar ligament. But this does not preclude possibility is suggested by Reilly’s [2000] data in rat ulnae the possibility that the generally more active osteonal re-

32 Cells Tissues Organs 2005;181:23–37 Skedros

newal in the plantar cortex is a product of biomechanical femoral cortical bone. In turn, the clustering of osteocytes stimuli associated with this insertion, including enhanced that this represents could adversely affect the sensitivity blood supply and associated fl uid-fl ow dynamics, and of osteocyte networks within regions (e.g. some regions lower ambient strains [Skedros et al., 1994b, 2001b]. failing to detect a microcrack) [Qiu et al., 2005]. This theoretical possibility, however, cannot be implied by the Tissue Constraints (Primary vs. Secondary Bone) and results of the regional analyses conducted in present Spatial Heterogeneity of Osteocytes study, including the plantar cortices where the rate might When all cortical locations were considered, none of be expected to show the highest regional spatial heteroge- the species showed correlations between Ot.Lc.N/B.Ar neity. Furthermore, spatial variations in Ot.Lc.N/B.Ar and On.Ar/T.Ar. However, the variation in primary ver- also did not consistently correlate with any of the other sus secondary bone may not have been suffi cient for de- regional characteristics of the spatially heterogeneous tecting differences in these calcanei – in all three species morphologic organization. This is not surprising since the majority of the sampled areas were predominately these characteristics are also not strongly correlated with secondary osteonal bone (typically 170% of the imaged the local differences in strain magnitude or the preva- area) [Skedros et al., 1997]. The possibility that broader lence/predominance of a specifi c strain mode [Skedros, variations in the relative percentages of secondary bone 2001]. might show strong correlations with Ot.Lc.N/B.Ar is sug- Although in each species there were statistically sig- gested by Remaggi et al. [1998] in their histomorphomet- nifi cant regional differences in Lc–Lc, the range was only ric study of mature limb-bone diaphyseal cortices in var- on the order of 3–5 m. These small differences are prob- ious species (frog, chick, rabbit, cow, horse, dog, and hu- ably not biomechanically signifi cant. This is consistent man). They found that (p. 152): with suggestions of the companion study [Skedros et al., “… the shape, size and density of osteocyte bodies and 2005] wherein we hypothesized that a relatively broad cytoplasmic processes … strictly depend on the spatial range of osteocyte densities reported in various species organization of collagen fi bers and not on the time of bone primarily refl ects constraints imposed by corresponding deposition (both primary and secondary bone may be histology/growth rates and/or other species-related differ- composed of woven-fi bered, parallel-fi bered or lamellar ences. Consequently, the minimal requirements of a com- bone); …” 2 munication network are probably readily satisfi ed at cell In a follow-up study, Ferretti et al. [1999, p. 127] found densities that are lower than those required for other that the distribution of osteocytes is more regular in lamel- mechano-biological demands. As discussed in the com- lar bone, but Ot.Lc.N/B.Ar is higher in parallel-fi bered panion study, it is also possible that there are differences bone. However, this is opposite of what we found in the in cell level adaptations (e.g. cell matrix tethering) be- deer calcanei: higher Ot.Lc.N/B.Ar in dorsal cortex where tween populations of osteocytes that modify their respon- there are lamellar osteons, and lower Ot.Lc.N/B.Ar in siveness in signifi cantly different regional strain environ- plantar cortex where there are parallel-fi bered osteons. ments. Results of the present study also did not show consis- Signifi cant differences in Ot.Lc.N/B.Ar were found be- tent dorsal versus plantar differences in spatial heteroge- tween species. Factors that infl uence these interspecies neity of osteocyte lacunae (fi g. 4b). This parameter was variations might be most important in affecting osteocyte examined since regional differences in the coeffi cient of lacuna density. For example, some factors might include variation of osteocyte lacunae would indicate increased animal activity, animal age, or degree of osteocyte death. heterogeneity in the spatial organization of osteons. For One limitation of this study is, as noted, our inability to example, Vashishth et al. [2000] showed that this coeffi - control for these factors. cient of variation increased linearly with age in human Correlations with Mean Cross-Sectional Osteon Areas It was hypothesized that Ot.Lc.N/B.Ar would be associated with osteon size (i.e. high Ot.Lc.N/B.Ar and

2 small osteon cross-sectional areas). This seems to be sup- This refl ects Marotti and co-worker’s [Marotti, 1990, 1996; Marotti and Muglia, 1992; Ferretti et al., 1999] scheme of osteon morphologies that, con- ported in the dorsal and plantar cortices of each species trary to the classical view that ascribes a lamellar structure to all secondary os- since, compared to the plantar cortices, the dorsal cortices teons, describes the existence of three differently structured secondary osteons, namely, woven fi bered, parallel-fi bered and lamellar. In other words, all three have: (1) signifi cantly lower mean cross-sectional osteon types of bone tissue may be present inside secondary osteons. areas and signifi cantly higher On.N/T.Ar [Skedros et al.,

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 33 Calcanei 1997], and (2) higher Ot.Lc.N/B.Ar (present study). An are highest in neutral axis regions, correlate with region- explanation supporting the possibility that these differ- ally increased Ot.Lc.N/B.Ar. This correlation was not ences in Ot.Lc.N/B.Ar are biomechanically relevant and found in the present study. A possible explanation for this might infl uence local osteon size and/or population den- may be that in some instances cell populations are con- sity is that spatial variations in load intensity (e.g. strain strained by growth rate and histologic organization more magnitude variations) cause differences in osteo blast re- so than by associated strain gradients. For example, the cruitment and/or percent survival. In turn, the relatively smaller and more numerous osteons in the cranial corti- higher densities of the osteocytes that remain in some ar- ces compared to the larger, oval, and less numerous os- eas could have a repressive effect on subsequent osteo- teons in the medial and lateral cortices may refl ect stron- clast activity. A high-stress-related repressive effect on ger infl uences associated with tissue construction on os- osteoclast vigor has been offered as an explanation for the teoblast recruitment/osteocyte survival than putative smaller diameter osteons in femora and tibiae between infl uences of local strain gradients. An additional con- groups of Pleistocene hominids compared to recent pop- founding variable in the calcanei examined in this study, ulations [Abbott et al., 1996]. which is not present in equine MC3s and radii, includes Bioactive non-collagenous proteins within bone ma- the perturbation of fl uid fl ow dynamics that might be as- trix may help explain relationships that exist between sociated with the plantar ligament – thus rendering com- Ot.Lc.N/B.Ar, osteon size, and other features of the local parisons diffi cult with other bones that do not exhibit a histology. For example, studies in 6-month-old rat tibiae similar enthesis. loaded in vivo have shown that levels of transforming growth factor (TGF- ) increase with loading and be- Cortical Thickness and the Concept of come incorporated in the matrix [Raab-Cullen et al., ‘Stressed Volume’ 1994]. Mice and rat cell culture experiments suggest that We had also speculated that Ot.Lc.N/B.Ar would pos- TGF- suppresses osteoclastic vigor and differentiation itively correlate with cortical thickness, possibly refl ect- [Heino et al., 2002]. In turn this may stimulate osteoblas- ing a requirement for detecting higher prevalence of mi- tic vigor, increasing osteoblast proliferation and matrix crodamage in larger ‘stressed volumes’ [Skedros et al., formation [Baylink et al., 1993; Geiser et al., 1998; Bone- 2005]. This interpretation, if correct, might be supported wald, 2002]. Studies in mice femoral diaphyses have also in calcanei of elk and sheep when all cortices were con- shown that the overexpression of TGF- increases the sidered (r = 0.63 and 0.70, respectively). However, this rate of osteoblast differentiation, subsequently increas- correlation was !0.4 in the equine calcanei. This issue is ing osteocyte population density [Erlebacher et al., 1998; also confounded by the presence of cancellous bone in the Hamrick et al., 2004]. Extrapolating these data to larger marrow cavity, which tends to be more prevalent in the mammals, Susan Pfeiffer [pers. commun.] has hypoth- cross-sections of the equine calcanei when compared to esized that suppressed osteoclastic vigor, mediated by the section locations in the other two species (fi g. 2). TGF- , could lead to the production of secondary osteons with reduced cross-sectional areas. The possibility Distinguishing Osteocyte Cell Bodies from Dendrites that this can be extrapolated from small mammals (that in Affecting Bone Matrix Modifi cation lack secondary osteons) to larger mammals (that have In view of other investigators’ fi ndings that have shown secondary osteons) is indirectly supported by recent ex at least a portion of remodeling is targeted to the repair vivo studies of cortical bone from adult human tibiae of microdamage, it is possible that other mechanotrans- [Yeni et al., 2004] demonstrating that mechanical load- duction pathways might predominate, leading to the re- ing can mobilize TGF- from the mineralized matrix. modeling and osteocyte distributions observed in the However, this relationship may not be generally appli- bones of the present study. These include innervation of cable since inconsistent relationships between mean os- bone, and changes in the dendritic processes of osteocytes teon cross-sectional areas and prevalent/predominant within canaliculi that are not associated with osteocyte strain mode or magnitude have been observed in tibiae death or changes in osteocyte lacunar density [Colopy et of sheep, radii of sheep and horses, and MC3s of horses al., 2004; Burt-Pichat et al., 2005; Szczesniak et al., 2005]. [Martin et al., 1996; Skedros et al., 1996; Skedros, 2001, For example, using in vivo loading of rat ulnae, Colopy and unpubl. data]. et al. [2004] report that remodeling (‘quasi-remodeling’ Data in the companion study of equine MC3s and ra- in the rat) distant from linear microcracks may be tar- dii suggest that circumferential strain gradients, which geted by osteocytes with disrupted cell processes rather

34 Cells Tissues Organs 2005;181:23–37 Skedros

than osteocytes with disruption of the cell body. Their evidence that the number of osteocyte lacunae has a func- results showed that while matrix injury and disruption of tional role in mechanotransduction pathways that are dendritic cell processes appear at least partially revers- typically considered in bone adaptation. Lack of consis- ible, disruption of osteocyte cell bodies does not. These tent correlations among Ot.Lc.N/B.Ar and the structural results suggest that cell dendrites can have important bio- and material characteristics also reject the possibility that mechanical functions that might not be recognized when regional variations in Ot.Lc.N/B.Ar are mechanically only cell bodies (hence osteocyte lacunae) are considered. ‘synergistic’ with the signifi cant regional differences in This bespeaks potentially important, but controversial mineral content, porosity, On.N/T.Ar, and cross-sec- [Parfi tt, 1977], aspects of osteocyte physiology – osteo- tional osteon area, for example, that have been described cytes are thought to have the capacity to modify the ma- within the same cross-sections of these bones. Addition- trix of bone [Belanger et al., 1967], and the architecture ally, regional variations in Ot.Lc.N/B.Ar do not appear of canaliculi containing the dendritic cell processes of os- suffi ciently reliable for inferring a history of habitual teocytes is known to be quite dynamic [Palumbo et al., bending. The mechanisms that govern the production 2004] and to potentially change over time [Okada et al., and maintenance of species/site-specifi c histogenesis 2002]. might be the most important factors governing the distribution of these cells. In turn, these tissue level ‘adaptations’ appear to require strain and fatigue-damage sen- Summary and Conclusions sors, which may be the function of osteocyte networks.

Results of this study showed signifi cant differences or tendencies (elk) in comparisons of plantar ‘tension’ vs. Acknowledgments dorsal ‘compression’ cortices, with highest Ot.Lc.N/B.Ar in dorsal cortices of all species. Although these relatively This study was supported by the National Science Foundation, the Department of Veteran Affairs Medical Research Funds, the small differences might be adaptations for prevalent/pre- Utah Bone and Joint Center, and Orthopaedic Research and Edu- dominant strain characteristics and/or associated micro- cation Foundation grant No. 01-024. The author thanks Roy Bloe- damage, it is also plausible that they are strongly infl u- baum for laboratory support, Betty Rostro and Mark Gunst for enced by corresponding differences in the bone formation their technical work, Steve Su and Todd Grunander for assistance rates that produced the tissue in these locations. Bone with statistical analyses, Todd Grunander and Heidi Jackson for independently verifying the method used in fi gure 3, Kerry Matz growth rates and other factors (e.g. animal activity and for illustrations, and Christian Sybrowsky, Scott Sorenson and Teri age) might help to explain the interspecies differences. Rosenbaum for their critical reviews of the manuscript. Consequently, the results of this study provide little or no

References

Abbott, S., E. Trinkaus, D.B. Burr (1996) Dynam- Belanger, L.F., C. Belanger, T. Semba (1967) Tech- Carter, D.R., W.C. Hayes (1976) Fatigue life of ic bone remodeling in later Pleistocene fossil nical approaches leading to the concept of os- compact bone. I. Effects of stress amplitude, hominids. Am J Phys Anthropol 99: 585– teocytic osteolysis. Clin Orthop 54: 187–196. temperature and density. J Biomech 9: 27– 601. Bonewald, L.F. (2002) Transforming growth fac- 34. Amtmann, E. (1971) Mechanical stress, functional tor-beta; in Bilezikian, J.P., L.G. Raisz, G.A. Colopy, S.A., J. Benz-Dean, J.G. Barrett, S.J. Sam- adaptation and the variation structure of the Rodan: Principles of Bone Biology. San Diego, ple, Y. Lu, N.A. Danova, V.L. Klalscheur, R. human femur diaphysis. Ergeb Anat Entwick- Academic Press, vol 2, pp 903–918. Vanderby, Jr., M.D. Markel, P. Muir (2004) lungsgesch 44: 1–89. Bousson, V., A. Meunier, C. Bergot, E. Vicaut, Response of the osteocyte syncytium adjacent Atkinson, P.J. (1965) Changes in resorption spaces M.A. Rocha, M.H. Morais, A.M. Laval-Jean- to and distant from linear microcracks during in femoral cortical bone with age. J Pathol Bac- tet, J.D. Laredo (2001) Distribution of intra- adaptation to cyclic fatigue loading. Bone 35: teriol 89: 173–178. cortical porosity in human midfemoral cortex 881–891. Badoux, D.M. (1987) Some biomechanical aspects by age and gender. J Bone Miner Res 16: 1308– Courtney, A.C., W.C. Hayes, L.J. Gibson (1996) of the structure of the equine tarsus. Anat Anz 1317. Age-related differences in post-yield damage in 164: 53–61. Burt-Pichat, B., M.H. Lafage-Proust, F. Duboeuf, human cortical bone – experiment and model. Baylink, D.J., R.D. Finkelman, S. Mohan (1993) N. Laroche, C. Itzstein, L. Vico, P.D. Delmas, J Biomech 29: 1463–1471. Growth factors to stimulate bone formation. J C. Chenu (2005) Dramatic decrease of inner- Currey, J.D. (1964) Some effects of ageing in hu- Bone Miner Res 8: S565–S572. vation density in bone after ovariectomy. En- man Haversian systems. J Anat Lond 98: 69– docrinology 146: 503–510. 75.

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 35 Calcanei Donahue, H.J. (2000) Gap junctions and biophys- Lanyon, L.E. (1974) Experimental support for the Remaggi, F., V. Canè, C. Palumbo, M. Ferretti ical regulation of bone cell differentiation. trajectorial theory of bone structure. J Bone (1998) Histomorphometric study on the osteo- Bone 26: 417–422. Joint Surg 56B: 160–166. cyte lacuno-canalicular network in animals of Emmanual, J., C. Hornbeck, R.D. Bloebaum Marotti, G. (1990) The original contribution of the different species. I. Woven-fi bered and paral- (1987) A polymethyl methacrylate method for scanning electron microscope to the knowledge lel-fi bered bones. Ital J Anat Embryol 103: large specimens of mineralized bone with im- of bone structure; in Bonucci, E., P.M. Motta: 145–155. plants. Stain Tech 62: 401–410. Ultrastructure of Skeletal Tissue. Boston, Klu- Ruff, C.B., W.C. Hayes (1988) Sex differences in Erlebacher, A., E.H. Filvaroff, J.Q. Ye, R. Derynck wer Academic Publisher, pp 19–39. age-related remodeling of the femur and tibia. (1998) Osteoblastic responses to TGF-beta Marotti, G. (1996) The structure of bone tissues J Orthop Res 6: 886–896. during bone remodeling. Mol Biol Cell 9: and the cellular control of their deposition. Ital Schaffl er, M.B., E.L. Radin, D.B. Burr (1989) Me- 1903–1918. J Anat Embryol 101: 25–79. chanical and morphological effects of strain Feik, S.A., C.D. Thomas, R. Bruns, J.G. Clement Marotti, G., M.A. Muglia (1992) The structure of rate on fatigue of compact bone. Bone 10: 207– (2000) Regional variations in cortical model- primary and secondary osteons studied with 214. ing in the femoral mid-shaft: sex and age dif- the scanning electron microscope. Bone 13: Skedros, J.G. (2001) Collagen fi ber orientation: a ferences. Am J Phys Anthropol 112: 191– A22. characteristic of strain-mode-related regional 205. Martin, R.B. (2000) Does osteocyte formation adaptation in cortical bone. Bone 28: S110– Ferretti, M., M.A. Muglia, F. Remaggi, V. Canè, cause the nonlinear refi lling of osteons. Bone S111. C. Palumbo (1999) Histomorphometric study 26: 71–78. Skedros, J.G., R.D. Bloebaum, K.N. Bachus, T.M. on the osteocyte lacuno-canalicular network in Martin, R.B., D.B. Burr, N.A. Sharkey (1998) Skel- Boyce (1993a) The meaning of graylevels in animals of different species. II. Parallel-fi bered etal Tissue Mechanics. New York, Springer, pp backscattered electron images of bone. J and lamellar bones. Ital J Anat Embryol 104: 1–392. Biomed Mater Res 27: 47–56. 121–131. Martin, R.B., V.A. Gibson, S.M. Stover, J.C. Skedros, J.G., R.D. Bloebaum, K.N. Bachus, T.M. Frost, H.M. (1960) In vivo osteocyte death. Am J Gibeling, L.V. Griffi n (1996) Osteonal struc- Boyce, B. Constantz (1993b) Infl uence of min- Orthop 42-A: 138–143. ture in the equine third metacarpus. Bone 19: eral content and composition of gray levels in Frost, H.M. (1998) On rho, a marrow mediator, 165–171. backscattered electron images of bone. J and estrogen: their roles in bone strength and Mason, M.W., J.G. Skedros, R.D. Bloebaum Biomed Mater Res 27: 57–64. ‘mass’ in human females, osteopenias, and os- (1995) Evidence of strain-mode-related corti- Skedros, J.G., R.D. Bloebaum, M.W. Mason, D.M. teoporosis – insights from a new paradigm. J cal adaptation in the diaphysis of the horse ra- Bramble (1994a) Analysis of a tension/com- Bone Miner Metab 16: 113–123. dius. Bone 17: 229–237. pression skeletal system: possible strain-spe- Geiser, A.G., Q.Q. Zeng, M. Sato, L.M. Helvering, Nunamaker, D. (2001) Bucked shins in horses; in cifi c differences in the hierarchical organiza- T. Hirano, C.H. Turner (1998) Decreased bone Burr, D.B., C. Milgrom: Musculoskeletal Fa- tion of bone. Anat Rec 239: 396–404. mass and bone elasticity in mice lacking the tigue and Stress Fractures. Boca Raton, CRC Skedros, J.G., J.H. Brady (2001) Ontogeny of can- transforming growth factor- 1 gene. Bone 23: Press, pp 203–219. cellous bone anisotropy in a natural ‘trajecto- 87–93. Okada, S., S. Yoshida, S.H. Ashrafi , D.E. Schrauf- rial structure’: genetics or epigenetics? J Bone Gross, T.S., K.J. McLeod, C.T. Rubin (1992) Char- nagel (2002) The canalicular structure of com- Miner Res 16(suppl 1): S331. acterizing bone strain distribution in vivo us- pact bone in the rat at different ages. Microsc Skedros, J.G., M.R. Dayton, C.L. Sybrowsky, R.D. ing three triple rosette strain gauges. J Biomech Microanal 8: 104–115. Bloebaum, K.N. Bachus (2003a) Are uniform 25: 1081–1087. Palumbo, C., M. Ferretti, G. Marotti (2004) Osteo- regional safety factors an objective of adaptive Gunasekaran, B., B.R. Constantz, M. Monjazeb, cyte dendrogenesis in static and dynamic bone modeling/remodeling in cortical bone? J Exp J.G. Skedros, R.D. Bloebaum (1991) An effec- formation: an ultrastructural study. Anat Rec Biol 206: 2431–2439. tive way of assessing crosslinks of collagenous 278A: 474–480. Skedros, J.G., T.R. Grunander, M.W. Hamrick proteins in biomaterials and tissues. Trans Soc Parfi tt, A.M. (1977) The cellular basis of bone turn- (2005) Spatial distribution of osteocyte lacu- Biomaterials XIV: 139. over and bone loss. Clin Orthop 127: 236– nae in equine radii and third metacarpals: con- Hamrick, M.W., C. Pennington, D. Newton, 247. siderations for cellular communication, micro- D. Xie, C. Isales (2004) Leptin defi ciency pro- Parfi tt, A.M. (1993) Bone age, mineral density, and damage detection, and metabolism. Cells duces contrasting phenotypes in bones of the fatigue damage. Calcif Tissue Int 53( suppl 1): Tissues Organs 180: 215–236. limb and spine. Bone 34: 376–383. S82–S85; discussion S85–S86. Skedros, J.G., K.J. Hunt (2004) Does the degree of Heino, T.J., T.A. Hentunen, H.K. Vaananen Petrtýl, M., J. Hert, P. Fiala (1996) Spatial organi- laminarity mediate site-specifi c differences in (2002) Osteocytes inhibit osteoclastic bone re- zation of haversian bone in man. J Biomech collagen fi ber orientation in primary bone? An sorption through transforming growth factor- 29: 161–169. evaluation in the turkey ulna diaphysis. J Anat beta: enhancement by estrogen. J Cell Biochem Qiu, S., D.P. Fyhrie, S. Palnitkar, D.S. Rao (2003) 205: 121–134. 85: 185–197. Histomorphometric assessment of Haversian Skedros, J.G., K.J. Hunt, K.N. Bachus (2001a) Huja, S.S., M.S. Hasan, R. Pidaparti, C.H. Turner, canal and osteocyte lacunae in different-sized Strain-mode-specifi c loading of cortical bone L.P. Garetto, D.B. Burr (1999) Development osteons in human rib. Anat RecA Discov Mol reveals an important role for collagen fi ber ori- of a fl uorescent light technique for evaluating Cell Evol Biol 272A: 520–525. entation in energy absorption. Trans Orthop microdamage in bone subjected to fatigue Qiu, S., D. Sudhaker Rao, D.P. Fyhrie, S. Palnit- Res Soc 26: 519. loading. J Biomech 32: 1243–1249. kar, A.M. Parfi tt (2005) The morphological as- Skedros, J.G., K.J. Hunt, R.D. Bloebaum (2004) Jepsen, K.J., D.T. Davy, D.J. Krzypow (1999) The sociation between microcracks and osteocyte Relationships of loading history and structural role of the lamellar interface during torsional lacunae in human cortical bone. Bone 37: 10– and material characteristics of bone: Develop- yielding of human cortical bone. J Biomech 32: 15. ment of the mule deer calcaneus. J Morphol 303–310. Raab-Cullen, D.M., M.A. Thiede, D.N. Petersen, 259: 281–307. Jowsey, J. (1966) Studies of Haversian systems in D.B. Kimmel, R.R. Recker (1994) Mechanical Skedros, J.G., K.J. Hunt, M.R. Dayton, R.D. Bloe- man and some animals. J Anat 100: 857–864. loading stimulates rapid changes in periosteal baum, K.N. Bachus (2003b) The infl uence of Kornblum, S.S., P.J. Kelly (1964) The lacunae and gene expression. Calcif Tissue Int 55: 473– collagen fi ber orientation on mechanical prop- Haversian canals in tibial cortical bone from 478. erties of cortical bone of an artiodactyl calca- ischemic and non-ischemic limbs. A compara- Reilly, G.C. (2000) Observations of microdamage neus: implications for broad applications in tive microradiographic study. J Bone Joint around osteocyte lacunae in bone. J Biomech bone adaptation. Trans Orthop Res Soc 28: Surg Am 46: 797–810. 33: 1131–1134. 411.

36 Cells Tissues Organs 2005;181:23–37 Skedros

Skedros, J.G., K.J. Hunt, P.E. Hughes, H. Winet Skedros, J.G., C.L. Sybrowsky, M.R. Dayton, R.D. Szczesniak, A.M., R.W. Gilbert, M. Mukhida, G.I. (2003c) Ontogenetic and regional morphologic Bloebaum, K.N. Bachus (2003d) The role of Anderson (2005) Mechanical loading modu- variations in the turkey ulna diaphysis: impli- osteocyte lacuna population density on the me- lates glutamate receptor subunit expression in cations for functional adaptation of cortical chanical properties of cortical bone. Trans Or- bone. Bone 37: 63–73. bone. Anat Rec 273A: 609–629. thop Res Soc 28: 414. Tami, A.E., P. Nasser, M.B. Schaffl er, M.L. Knothe Skedros, J.G., M.W. Mason, R.D. Bloebaum Skedros, J.G., Y. Takano, C.H. Turner (2000) Me- Tate (2003) Noninvasive fatigue fracture mod- (1994b) Differences in osteonal micromor- chanical loading patterns determine collagen el of the rat ulna. J Orthop Res 21: 1018– phology between tensile and compressive cor- and mineral orientation. J Bone Miner Res 15: 1024. tices of a bending skeletal system: indications S348. Vashishth, D., J.C. Behiri, W. Bonfi eld (1997) of potential strain-specifi c differences in bone Smith, R.W., R.R. Walker (1964) Femoral expan- Crack growth resistance in cortical bone: con- microstructure. Anat Rec 239: 405–413. sion in aging women: implications for osteopo- cept of microcrack toughening. J Biomech 30:

Skedros, J.G., M.W. Mason, R.D. Bloebaum rosis and fractures. Science 145: 156–157. 763–769. (2001b) Modeling and remodeling in a devel- Sokal, R.R., F.J. Rohlf (1995) Biometry: The Prin- Vashishth, D., O. Verborgt, G. Divine, M. Schaff- oping artiodactyl calcaneus: a model for evalu- ciples and Practice of Statistics in Biological ler, D.P. Fyhrie (2000) Decline in osteocyte la- ating Frost’s mechanostat hypothesis and its Research, ed 3. New York, W.H. Freeman, cunar density in human cortical bone is associ- corollaries. Anat Rec 263: 167–185. p 887. ated with accumulation of microcracks with Skedros, J.G., M.W. Mason, M.C. Nelson, R.D. Stout, S., D.J. Simmons (1979) Use of histology in age. Bone 26: 375–380. Bloebaum (1996) Evidence of structural and ancient bone research. Yearbook of Physical Weinbaum, S., S.C. Cowin, Y. Zeng (1994) A mod- material adaptation to speciﬁ c strain features Anthropology, vol 22, pp 228–249. el for the excitation of osteocytes by mechani- in cortical bone. Anat Rec 246: 47–63. Su, S. (1998) Microstructure and mineral content cal loading-induced bone ﬂ uid shear stresses. J Skedros, J.G., S.C. Su, R.D. Bloebaum (1997) Bio- correlations to strain parameters in cortical Biomech 27: 339–360. mechanical implications of mineral content bone of the artiodactyl calcaneus. Salt Lake Yeni, Y.N., B. Zhang, D.P. Fyhrie, G.J. Gibson and microstructural variations in cortical bone City, University of Utah, p 64. (2004) Damaging mechanical loads release se- of horse, elk, and sheep calcanei. Anat Rec 249: Su, S.C., J.G. Skedros, K.N. Bachus, R.D. Bloe- questered transforming growth factor beta-1 297–316. baum (1999) Loading conditions and cortical from human cancellous and cortical bone ma- bone construction of an artiodactyl calcaneus. trix. Transactions of the Orthopedic Research J Exp Biol 202 (Pt 22): 3239–3254. Society, vol 29, p 73.

Osteocyte Lacuna Population Densities in Cells Tissues Organs 2005;181:23–37 37 Calcanei