Longitudinal bone growth can be suppressed by compressive loading. In this study, we applied three different magnitudes (17, 8.5, and 4N) of compressive force to growing rat ulnas 10 minutes/day for 8 days and investigated the effects on the distal growth plate biology. Further, to investigate growth rate recovery after cessation of loading, we examined rats 7 days after the loading period. Longitudinal growth of the ulna was suppressed in a dose-dependent manner by applied compressive force. In the 17N group, the longitudinal mineralization rate (LMR) at the distal growth plate was suppressed completely by loading and did not recover. However, in the 8.5N and 4N groups, LMR suppression recovered in 1 week. In the 17N group, growth plate height and hypertrophic zone height were significantly greater than control; the number of hypertrophic chondrocytes was increased; and some traumatic changes such as cracks in the growth plate were found. In addition, 17N loading suppressed cartilage mineralization and capillary invasion beneath the growth plate, although the number of chondrocytes synthesizing vascular endothelial growth factor (VEGF) was increased. Our study shows longitudinal growth suppression caused by axial loading of the ulna, which is proportional to the magnitude of load. Only the largest load (17N) caused morphological changes in the distal growth plate cartilage. There was no association found between mineralization and type X collagen localization or capillary invasion and VEGF expression.
It is well known that mechanical forces influence bone formation and adaptation.(1) A number of experimental models have been developed over the years to investigate the effects of mechanical loading on bone formation.(2–4) One interesting loading model involves axial loading of the rat ulna in vivo. This model was developed by Torrance et al.(5) and further characterized by Mosley et al.(6, 7) One consistent effect of the ulnar axial loading model is suppression of longitudinal growth.(5–8) This observation suggests that mechanical loading might suppress endochondral ossification via alteration in chondrocyte metabolism or by the accumulation of damage in the growth plate. Likewise, loads derived from physical exercise influence long bone growth, although these effects vary considerably. Some studies report that exercise suppresses bone growth,(9–12) while others show enhanced growth with exercise.(13–16) Reports regarding the relation between growth plate cartilage and mechanical force also are inconsistent. For example, some exercises that presumably increase the loads placed on the growth plate result in increased growth plate height,(9, 17) whereas increased skeletal loading resulting from centrifugation (to simulate hypergravity) has been shown to reduce growth plate height.(18, 19) Consequently, it is clear that loading can influence longitudinal bone growth but the relation between different magnitudes of mechanical stress and growth plate biology is still unclear.
Previously, using the rat ulna loading model, we investigated the effects of the static and dynamic loading on endocortical and periosteal bone formation and on longitudinal growth.(8) In that experiment, we found that loaded ulnas were shorter than nonloaded ulnas, a result supporting the original observations in the same model.(5–7) A number of changes at the distal growth plate of loaded ulnas were linked to longitudinal growth suppression, including increased growth plate height and a marked retention of hypertrophic chondrocytes. However, it is still unclear whether the growth-suppressive effects of loading can be reversed after the loading protocol is withheld or what cellular mechanisms are involved in growth suppression at the growth plate.
The goal of this study is to elucidate the role of dynamic loading on endochondral ossification, growth plate cartilage morphology, and chondrocyte metabolism. Specifically, we examined the effects of different force magnitudes on the length of growing ulnas and investigated the effects of axial force on the distal growth plate cartilage and on endochondral ossification. To determine whether lost growth during the experiment would recover, we loaded rats for 8 days and allowed a week to recover before death. Histological and histomorphological techniques were used to investigate the changes in growth plate cartilage. We also investigated the changes in localization of type X collagen, which plays a pivotal role in cartilage mineralization(20, 21) and vascular endothelial growth factor (VEGF) expression, a strong inducer of capillary invasion to the growth plate.(22)
We tested the following hypotheses: (1) growth is suppressed by loading in a dose-dependent manner, (2) recovery of the suppressed longitudinal growth rate occurs within 1 week after loading, and (3) the biology of chondrogenesis in the distal growth plate is changed as a result of compressive loading.
MATERIALS AND METHODS
Seventy-two male Sprague-Dawley rats (body weight of 170 ± 10 g) were purchased from Harlan Sprague-Dawley (Indianapolis, IN, USA) 1 week before the start of the experiment and were provided food and water ad libitum. The rats were divided into three treatment groups, which differed in the magnitude of the dynamic axial force applied to the right ulna (17, 8.5, or 4N). Each load magnitude group was divided further into three subgroups. The first subgroup was baseline controls, and was killed at the beginning of the experiment (S1). The second subgroup received 8 days of loading followed by immediate death on the eighth day (S2). The third subgroup received 8 days of loading followed by 7 days without loading before being killed (S3). All rats were injected with fluorochrome bone labels 5 days and 2 days before death. The labels were calcein (7 mg/kg ip; Sigma Chemical Co., St. Louis, MO, USA) given at the first time point and alizarin complexone (14 mg/kg intraperitoneally [ip]; Sigma Chemical Co., St. Louis, MO, USA) given subsequently. Each subgroup comprised eight rats, five of which were used for histomorphometric analysis. The remaining three rats in each subgroup were processed for immunohistochemical analysis. All procedures performed in this experiment were in accordance with the Indiana University Animal Care and Use Committee guidelines.
Dynamic axial compressive loads were applied to the right ulna. The rat ulna loading device was described previously (Fig. 1).(5, 23) All S2 and S3 subgroups received 1200 cycles/day (10 minutes) of a 2 Hz Haversian wave with peak loads of 17, 8.5, or 4N (and an offset load of 0.5N); these loading regimens created peak load rates of 104, 50, and 22N/s, respectively. The left ulnas of these animals were used as nonloaded controls. All experiments were done under the general anesthesia (ethyl ether inhalation). The rats were allowed normal cage activity between the daily loading sessions. In male rats of the size used in the experiment, 17N of axial load engenders a peak compressive strain of 3500 microstrain on the medial surface of the ulnar midshaft.(6) Axial loads of 8.5N and 4N should cause peak compressive strains of 1750 microstrain and 824 microstrain, respectively, at the ulnar midshaft. During running, rats engender about 1200 microstrain on the medial surface of the ulnar midshaft and during more extreme activities, like jumping from a 30-cm platform, the peak ulnar strain is 2500 microstrain.(6) Consequently, the highest load of 17N on the ulna is larger than the rat normally encounters, the load of 8.5N is near the upper limit of the normal activity range for a rat, and 4N loading is less than expected loads caused by running.
After death, left and right ulnas were removed, cleaned of soft tissue, and fixed for 48 h in 10% neutral buffered formalin. After measuring the total length of each ulna with digital calipers, the distal portion was removed from the shaft, dehydrated in graded alcohols, and embedded in polymethylmethacrylate. Four sections (4-6 μm) were cut sagittally through the center of the distal ulna using a microtome (Reichert-Jung 2050; Reichert-Jung, Heidelberg, Germany). One section was mounted unstained on standard microscope slides for reading fluorescent labeling. The remaining three sections per block were stained with toluidine blue, Goldner's trichrome, or McNeal's tetrachrome, respectively. Histomorphometric measurements were collected in the distal growth plate and metaphysis using a Nikon Optiphot microscope system with the Bioquant 98 Image Analysis System (R & M Biometrics, Nashville, TN, USA). The proximal growth plate was not analyzed because our previous study(8) showed that the significant effects of loading on ulnar growth occurred at the distal growth plate but not at the proximal growth plate. The following measurements were taken: (1) interlabel width (Ir.L.Wi), defined as the distance between the double fluorochrome labels in the growth plate metaphyseal junction and measured at approximately 15 different points in each section and averaged; (2) the total area of growth plate (GPl.Ar); (3) total growth plate height (GPl.Ht) and hypertrophic zone height (HpZ.Ht), measured at a minimum of eight different locations within each section and averaged; and (4) number of proliferating chondrocyte lacunae (NPr.Ce.Lc) and number of hypertrophic chondrocyte lacunae (NHp.Ce.Lc) were counted within their respective zones. CVs were determined for the growth plate measurements by completing five repeated measurements on three sections were chosen at random. CVs for GPl.Ar, GPl.Ht, HpZ.Ht, NPr.Ce.Lc, and NHp.Ce.Lc were 1.5, 2.0, 4.1, 6.1, and 5.8%, respectively. From these primary data, the longitudinal mineralization rate (LMR) was calculated as follows: LMR = Ir.L.Wi./3(days) (Table 1).
Table Table 1. Measurements of Longitudinal Growth and Growth Plate Morphology
To quantify the traumatic histological changes of growth plate resulting from loading, histological changes in the growth plate were scored qualitatively based on the amount of capillary invasion and on the number of cracks in the growth plate. Each growth plate was assigned a crack score of 1-3: 1, no cracks; 2, cracks seen on the edge of the growth plate; and 3, cracks throughout the growth plate were seen. Each growth plate also was assigned a capillary invasion score of 1-3: 1, invasion by numerous capillaries was observed beneath the growth plate; 2, invasion by a few capillaries, which were regularly seen just beneath the growth plate; and 3, no regular capillary invasion was observed.
Rats used for immunostaining were perfusion-fixed with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). The ulnas were excised and fixed overnight in the same fixative at 4°C, decalcified in 10% EDTA, dehydrated in graded ethanols, and embedded in paraffin. Five-micrometer-thick sections were cut using a Reichert-Jung 820 microtome. The sections were deparaffinized with xylene and rehydrated using a series of ethanols. Sections were digested with hyaluronidase for 2 h. To inhibit peroxidase activity, slides were placed in freshly prepared 1.5% H2O2 in methanol for 20 minutes. To reduce nonspecific reactions, slides were incubated with 10% normal goat serum in phosphate-buffered saline (PBS) for 1 h. Then, the sections were incubated in a moisture chamber overnight at 4°C with mouse anti-type X collagen antibody and rabbit anti-VEGF antibody. When the reaction with the antibodies was complete, the sections were rinsed thoroughly with PBS and incubated with anti-mouse or anti-rabbit antibodies (Vector Laboratories, Burlingame, CA, USA) for 4 h at room temperature. Then, the sections were rinsed several times with PBS and incubated with the Vectastain avidin-biotin complex (Vector Laboratories) for 30 minutes at room temperature. Immunoreactivity was visualized with diaminobenzidine (DAB). Then, the sections were counterstained with hematoxylin. As controls, sections were processed with primary antibodies replaced by nonimmune serum.
Data are presented as mean ± SE, unless otherwise noted. For histomorphometrical data, we used the following statistical analyses. Right (loaded) versus left (nonloaded) differences were tested for significance using Wilcoxon signed rank tests. Differences among relative subgroup means were tested for significance by Kruskal-Wallis one-way analysis of variance (ANOVA), followed by Mann-Whitney U tests for post hoc comparisons. A linear regression analysis was used to compare the difference in left and right ulnar length with applied load magnitude.
To test the relationship between each subgroup and scores for histological changes, Kruskal-Wallis one-way ANOVA was used. Further, to investigate the correlation between crack score and capillary invasion score, Spearman rank-order correlation coefficient ρs was used.
Among the 72 rats in the study, 3 rats were excluded from analysis because of anesthesia-related death during the experiment. Animal body weight increased significantly during the study periods (S2 gained 16.3 ± 2.3 g and S3 gained 46.7 ± 3.3 g). There was no significant effect of loading protocol on body weight gain within the S2 and S3 subgroups (p = 0.6 and p = 0.3, respectively).
In subgroups 2 (S2, 8 days of loading) and 3 (S3, loading followed by a week of recovery), the right (loaded) ulnas were significantly shorter than the left (nonloaded) ulnas (p < 0.05; Table 2). The difference between right and left ulnar length was proportional to force magnitude (r = 0.72; p < 0.0005 by linear regression). Despite the growth suppression observed at the distal growth plate in the loaded ulnas, the ulnas were longer than those from the baseline control groups, indicating continued growth in the proximal growth plate. The difference in ulnar length (left minus right) in the 17N-S3 subgroup was significantly greater than that in the 17N-S2 subgroup (p < 0.001). However, in the 8.5N and 4N groups, there were no significant differences between the S2 and S3 subgroups. No significant right versus left differences were detected for total length of the ulna or for histomorphometric measurements from the three baseline control subgroups (S1).
Table Table 2. Effects of Axial Loading on Right Ulna Length and LMR Compared With Those of the Left (Nonloaded) Ulna
Double labeling was not detected in the distal metaphyses of the 17N-S2 and 17N-S3 subgroups (although a single label was present), indicating that there was no movement of the mineralization front at the distal growth plate in the right (loaded) ulnas (Table 2). This finding suggests that either longitudinal growth was suppressed completely or mineralization of the growth plate was impaired. The 8.5N-S2 and 4N-S2 subgroups also showed significant differences in distal LMR between the right and left ulnas (p < 0.05). In the 8.5N-S3 and 4N-S3 subgroups, there were no statistical differences between right and left LMR in distal ulna.
In the 17N-S2 and 17N-S3 subgroups, all right (loaded) ulnas exhibited cracks running through their growth plates and all individuals were assigned a crack score of 3. In the 8.5N-S2, all rats showed cracks in the edges of the growth plate and all animals in this group were assigned a crack score of 2. In the 8.5N-S3 subgroup, there were no cracks observed in the growth plate. In 4N-S2, only one out of five rats showed a crack at the edge of growth plate, and no rats in 4N-S3 exhibited cracks (Table 3).
Table Table 3. Histological Assessment of Vascular Invasion Beneath the Growth Plate and Cracks Within the Growth Plate
In the 17N group, no regular capillary invasion was observed beneath the growth plate and, consequently, the majority were assigned a capillary invasion score of 3. In the loaded ulnas of the 8.5N and 4N groups, regular strong capillary invasions were observed beneath the growth plate, resulting in a capillary invasion score of 1 for all rats. From these data there was significantly less capillary invasion into the growth plate and significantly more cracks in the growth plate in the 17N loaded group than in the other loaded groups (8.5N and 4N; p < 0.0001). The correlation between crack score and capillary invasion score was ρs =0.86 (p < 0.001; Table 3).
In the 17N-S2 group, the GPl.Ht in the right (loaded) ulna was significantly greater than the left (nonloaded) growth plate (p < 0.05). The difference (right minus left) in GPl.Ht was increased after 1 week of recovery (17N-S3; p < 0.05). In the 17N-S2 group, the HpZ.Ht in the right ulnas was significantly greater than that in the left ulnas (p < 0.05). Further, in the 17N-S3 group, the difference in HpZ.Ht between right and left ulnas was significantly greater than in the 17N-S2 group (p < 0.05). Similar to the result for HpZ.Ht, NHp.Ce.Lc's in the right ulna were more numerous than in the left ulna among 17N-S2 and 17N-S3 rats (p < 0.05; Figs. 2 and 3). There were no significant left-right differences in the NPr.Ce.Lc's for the 17N-S2 or 17N-S3 groups. In the 8.5N and 4N groups, there were no significant right versus left differences in GPl.Ht, HpZ.Ht, NHp.Ce.Lc, and NPr.Ce.Lc.
MacNeal's staining revealed that the hypertrophic zone of the distal ulnar growth plate was mostly mineralized with the exception of the right ulnar growth plate in the 17N-S2 group, which showed very little mineralized cartilage in the hypertrophic zone (Fig. 4). In the 17N groups, type X collagen was detected in the hypertrophic zone in both right (loaded) and left (nonloaded) ulnar growth plates (Fig. 5). Although the 17N-S2 group exhibited very little mineralized cartilage in right ulna, type X collagen was localized in the entire hypertrophic chondrocyte zone. In the 8.5N and 4N groups, both the right and the left ulnar growth plates showed similar localization of type X collagen in the hypertrophic zone.
In all control groups (S1), VEGF was detected in hypertrophic chondrocytes. In the right growth plates of subgroups 17N-S2 and 17N-S3, VEGF was detected in almost all chondrocytes (Fig. 6). However, in 8.5N-S2, 8.5N-S3, 4N-S2, and 4N-S3 subgroups, VEGF was detected only in hypertrophic chondrocytes and no differences between right and left ulnas were found.
In this study, our objective was to determine the effects of different magnitudes of axial dynamic loading on longitudinal bone growth and growth plate biology. The results from this experiment show that ulnar growth was suppressed in a dose-dependent manner, reflecting the magnitude of applied force. LMR at the distal growth plate was suppressed completely in rats loaded to 17N (highest force). In addition, LMR at the distal growth plate was significantly suppressed by 8.5N and 4N loading but within 1 week after loading, the suppression was not statistically significant. In the 17N group, we found a number of changes in the distal growth plate such as greater growth plate height, which resulted mainly from an increase in the HpZ.Ht, suppression of cartilage mineralization, and an increase in VEGF expression.
The LMR in the distal ulna was abolished completely by 17N of loading, and was suppressed by 17% and 8% in the 8.5N and 4N groups. These results show that loading causes growth suppression, consistent with the report by Mosley et al.,(6) and this is reflected by changes in LMR measured in the distal metaphysis. Previously, we found that axial loading caused histomorphometric changes in the distal ulnar growth plate but not in the proximal growth plate,(8) suggesting that loading effects are concentrated at the distal metaphysis. Therefore, in the 17N-S2 and 17N-S3 subgroups, longitudinal growth of loaded ulnas occurred at the proximal growth plate, although LMR was zero at the distal growth plate.
Frost proposed the chondral growth force response (CGFR) curve to explain the relationship between longitudinal growth and axial force (Fig. 7).(24) The CGFR curve suggests that small compressive loads increase longitudinal growth, but large compressive loads retard or even abolish growth if large enough. The results from our experiment indicate that compressive force suppresses growth rates, even at low (4N) loads. This is contrary to Frost's suggestion that small compressive forces enhance longitudinal growth but is consistent with studies showing that centrifugation of rats suppresses longitudinal growth.(19) Frost's hypothesis is supported by studies showing that light endurance training appears to enhance longitudinal growth.(16, 25) Accordingly, Swissa- Sivan et al. showed that young rats that swam for 20 weeks had longer humeri than controls.(15) However, physical exercises not only involve local mechanical stimuli, but also induce a number of systemic biological changes such as enhanced insulin-like growth factor I (IGF-I) synthesis(26) and altered metabolic activity. In our study, applied loading at 4N, which produces ulnar bone strains of similar magnitude to those measured during normal ambulation,(6) did not enhance longitudinal growth. Moreover, it is unlikely that loading from routine daily activities increases longitudinal growth because normal activities in rats do not increase growth, compared with skeletal unloading such as space flight(27) or hindlimb suspension.(28–30) (Long bone growth is suppressed in disuse states associated with neurological deficits, e.g., poliomyelitis or congenital paralysis,(24) possibly because of lack of proper innervation rather than lack of mechanical load. (31)) These results are consistent with our findings, suggesting that long bone growth occurs normally when the growth plate is unloaded or under physiological pressures, but growth is suppressed when larger compressive forces are applied to the growth plate. This conclusion is consistent with the proposal by Hert(32, 33) (Fig. 7).
We found a number of changes in the distal metaphysis of the 17N groups such as thicker growth plates, increased HpZ.Ht, and increased NHp.Ce.Le in the growth plate. All 17N growth plates exhibited cracks and lacked normal capillary invasion. Ex vivo mechanical tests of growth plate cartilage loaded to failure reveal cracks in the growth plate, especially in the hypertrophic zone and junction between bone and cartilage,(34, 35) which resemble those found in the 17N group from our study. In addition, Vico et al. showed similar damage to the growth plates of rats subjected to centrifugation (to simulate hypergravity).(18) Thus, the force used in the 17N group might be in excess of the yield point for growth plate cartilage, suggesting that the changes shown in the growth plate might be traumatic changes. In a previous study, we showed that static loading of 17N caused the same inhibition of ulnar growth as did 17N dynamic loading.(8) These findings suggest that growth suppression is caused by mainly the magnitude of the applied load, not the rate at which the load is applied, and that load-induced creep of the growth plate cartilage is a plausible damage mechanism.
Interestingly, the changes in chondrocyte populations were localized mainly to the hypertrophic zone. These data suggest that overload affects mineralization. In the hypertrophic chondrocyte zone, not only the cell populations were changed but also metabolism was changed. Immediately after 7 days of loading, type X collagen was seen in the hypertrophic zone but little mineralization was seen. Endochondral ossification begins with type X collagen synthesis in cartilage and then chondrocyte hypertrophy, cartilage matrix mineralization, angiogenesis, and, finally, the mineralized cartilage is remodeled into bone matrix. Type X collagen is present in the hypertrophic chondrocyte zone in the growth plate(36) and is known to have a role in cartilage mineralization.(20, 21) However, in the 17N-S2 group, cartilage mineralization was not associated with type X collagen localization, suggesting disruption in the process of cartilage mineralization. It has been shown previously that local ischemia inhibits mineralization of the growth cartilage.(37) Furthermore, ischemic growth plates exhibit thickened hypertropic chondrocyte zones(37) as was observed after 17N loading in this study. It is quite possible that the impaired growth plate mineralization after loading was the result of poor capillary invasion.
In the growth plate, hypertrophic chondrocytes express VEGF, a coordinator of chondrogenesis and especially angiogenesis in the growth plate.(23, 38) In 17N groups we found that VEGF was expressed highly and was detected not only in hypertrophic chondrocytes but also in most other chondrocytes in the growth plate. However, compromised capillary invasion into the growth plate was found. Previous experiments showed that VEGF is expressed also at the repair site of healing tissues to induce capillary migration.(39) Therefore, the VEGF expressing chondrocytes might increase after loading in the 17N group to promote healing in the growth plate. One week of recovery might be too early to start the capillary migration for healing. However, others have reported that the VEGF receptor colocalizes with VEGF in hypertrophic cartilage, suggesting an autocrine loop in chondrocytes for maintaining chondrocyte survival and controlling cell differentiation.(40) The autocrine loop may cause VEGF levels to increase in the loaded growth plates. Furthermore, several growth factors crucial for cartilage development such as IGF-I or transforming growth factor β (TGF-β) increase VEGF expression.(41) Under compressive forces, these growth factors might be expressed and help chondrocytes to secrete VEGF.
In the 8.5N-S2 group, LMR in the right (loaded) ulna was suppressed by 17% compared with the left (control) ulna. In the 8.5N-S3 (recovery) group, LMR in the right ulna was not significantly different than the left ulna. Likewise, LMR recovered somewhat in the 4N loaded ulna after the recovery period. In the 17N-S2 group, LMR at the distal growth plate was abolished completely and did not change within 1 week after loading (S3). These data suggest that in 1 week, growth plates loaded at 8.5N and 4N initiated the recovery of LMR, but growth plates loaded at 17N did not. Therefore, there might be a correlation between the loading force magnitude (or degree of injury) and the period for the recovery of growth suppression. Spengler et al. showed that 25 days after the completion of a centrifugation experiment, growth rates in growth-suppressed limbs returned to control levels.(27) Thus, for the 17N loaded ulna, a longer time than 1 week might be required to return to normal (nonsuppressed) growth rates. We speculate that if we had followed the rats for an extended period (e.g., 1 month) after cessation of loading, the ulnar growth plates might return to normal endochondral ossification.
We thank Mary Hooser and Diana Jacob for assistance with tissue processing. This work was supported by the National Institute of Arthritis, Musculoskeletal, and Skin Diseases (AR43730).