Mechanical load promotes bone formation as illustrated by the trabecular pattern of the proximal femur and by increased bone mass in the active arm of tennis players. Load also enhances fracture healing. However, whether these observations reflect a direct effect of load on the cytoskeleton of bone cells via for example canalicular flow,1 or an indirect effect via other mediators is unknown. The latter would imply a system sensitive and responsive to load and capable of regulating local bone turnover. Plausible candidates are various signaling pathways of the nervous system. Evidence suggests that neuronal mediators are involved in skeletal development and local bone turnover.2–8 Patients and animals with hemiparesis or systemic neuropathies exhibit decreased bone density and impaired fracture healing characterized by excessive callus formation, decreased mechanical strength, and nonunion.9–12
Over the last decade a number of neuropeptides,3, 5, 6, 13, 14 have been implicated in the regulation of local bone turnover in addition to nociception, inflammation, angiogenesis, and cellular proliferation. Notably, the sensory neuropeptide substance P (SP) was identified in the most active areas of osteogenesis, such as growth plate, periosteum, and bone marrow.14–16 Moreover, the SP receptor, neurokinin 1 (NK-1), was demonstrated on chondrocytes,17, 18 osteocytes,19 osteoblasts,4, 20 and osteoclasts.19–21
The role of SP in local bone turnover is paradoxical. In vitro studies suggest that SP stimulates osteoclastic activity,3, 21, 22 while others demonstrate a stimulatory effect on osteoblastic activity.3, 4 The in vivo role of SP in bone is unknown. Its occurrence has not been explored in relation to load during fracture healing. In the present study, the temporal and spatial occurrence of SP during different phases of bone repair was compared in angulated and straight fractures in rat tibias representing healing under different loading conditions.
MATERIALS AND METHODS
Fifty-five male Sprague–Dawley rats (10-week old, mean body weight of 200 g) were housed three per cage with free access to standard rat chow and tap water in the animal laboratory at Karolinska Hospital with controlled temperature in a 12-h light/dark cycle. All experiments were performed with approval from the Ethics Committee for Animal Research, Stockholm North. The rats were assigned into three groups: (1) 25 rats for tibial fracture in straight (normal) alignment; (2) 25 rats for tibial fracture in anterior angulation; and (3) 5 rats aged 13 weeks with intact tibias used as control.
Rats were anesthetized by an intraperitoneal injection (i.p.) of fentanyl-fluanisone (Hypnorm1 0.5 ml/kg body weight; Janssen Pharmaceutica, Beerse, Belgium). The right tibia was first weakened at the mid-diaphysis by percutaneous drilling with an 18-G needle. A small anteromedial skin incision was made to the proximal tibial tubercle, and a 17-G cannula needle was inserted through the proximal metaphyseal cortex into the medullary canal.23 The tibia was then fractured by manual three-point bending and fixed in either 40° anterior angulation or straight alignment; both cases included the normal 13° anatomical curve of the diaphysis. Post-operative X-rays confirmed the adequacy of fracture position and fixation. Healing was monitored by lateral X-rays under Hypnorm anesthesia at days 7, 21, 35, 56, and 84 post-fracture (Fig. 1A and B). All rats regained weight bearing within a few hours and gait returned to normal around day 7.
Rats were anaesthetized with sodium pentobarbitone (Apoteksbolaget, Umeå, Sweden, 60 mg/kg, i.p.). Five rats in each treated group were euthanized at different times (7, 21, 35, 56, and 84 days post-fracture), while all five rats in the control group were killed at day 21. In vivo intra-aortic perfusion was carried out with phosphate buffered saline (PBS), followed by fixation with Zamboni's solution. The right tibia was dissected and demineralized at room temperature in a solution of 7% AlCl3, 5% formic acid, and 8.5% HCl for about 8 h. The samples were soaked for ≥2 days in 20% sucrose. The tibia was sagittally divided in half. The medial half of the diaphysis was sectioned sagittally at a thickness of 15 µm using a Leitz (Leitz, Wetzlar, Germany) cryostat (Fig. 1C). Thus, the regions undergoing endochondral and intramembranous ossification were all included in the same section.
The frozen sections were mounted directly on SuperFrost® Plus slides (Menzel-Glaser, Braunschweig, Germany). The serial sections were numbered consecutively from the middle to the medial aspect. Two interval sections, that is, one close to the middle and one close to the medial part of each tibia at each time point, were chosen for immunostaining according to the biotin–avidin system (Vector Laboratories Inc., Burlingame, CA). Thus, the sections were hydrated in 0.01 M PBS for 10 min, and then incubated at room temperature for 30 min with 10% normal goat serum. Subsequently, the sections were incubated at room temperature over night in a humid atmosphere with polyclonal antibody to SP (1:10,000, Peninsula Laboratories, Belmont, CA). After rinsing in PBS (2 × 5 min), the sections were incubated with secondary biotinylated goat anti-rabbit antibodies (1:250; Vector Laboratories Inc., Burlingame, CA; for 40 min). Finally, the fluorochrome Cy3-conjugated avidin (1:2,000, Amersham Life Science Inc., Arlington Heights, IL) was used for visualization of the immunoreaction. After completing the staining step for SP with the visualization of the immunoreaction of fluorochrome Cy3-conjugated avidin (1:1,000, Amersham), the sections were incubated for double staining for 15 min with avidin blocking solution followed by biotin blocking solution. The sections were incubated at room temperature for 30 min with 10% normal horse serum, then kept at room temperature over night in humid atmosphere with mouse monoclonal antibody to GAP-43 (1:2,000, Boehringer Mannheim Biochemicals, Mannheim, Germany). Subsequently, the sections were incubated with biotinylated anti-mouse antibodies (1:250, Vector Labs) for 40 min. The fluorochrome Cy2-conjugated avidin (1:1,000, Amersham) was used for visualization of the immunoreaction. An epifluorescence microscope (Eclipse E800™, Nikon, Yokohama, Japan) was used for fluorescence microscopic analysis (objective 20×). Through a video camera (DEI 750, Optronics Engineering, Goleto, CA), 16 microscope images (8 on the concave side and others on the convex side) were chosen and stored on a computer for later morphological and semi-quantitative analysis (Fig. 1C).
The whole fracture area was examined including the callus and the intact bone proximal and distal to the fracture. The analysis focused on the periosteal area on the anterior and posterior sides (Fig. 1C). Nerve fiber shape, location, and direction were compared.
The nerve fiber number in and adjacent to the fracture area on both anterior and posterior sides was quantified. The mean value of nerve fiber numbers based on the 16 images was determined to obtain a measure of fiber density for comparison in each tibia.
The results were presented as mean ± SEM. Statistical analysis was based on the mean of the values from five rats at each time point in each group. A nonparametric approach was applied in the semi-quantitative analysis of the number of SP-positive fibers. To compare the differences of SP fiber density between the anterior (convex) and posterior (concave) sides within the same group, the Wilcoxson test was utilized. To compare the differences of SP fiber density between time points and groups of individual side, Mann–Whitney tests were applied. A p-value <0.05 was considered significant.
Radiographic healing was noted at day 35 post-fracture in the angulated and at day 21 in the straight fractures. At day 21, peak callus thickness (anterior + posterior side) was observed in both fracture types; in the angulated fractures, 4 mm on the convex, and 10 mm on the concave side (Fig. 1A), whereas the straight fractures exhibited equal callus thickness of 7 mm on both sides (Fig. 1B). In the remodeling phase, during days 21–56, the angulated fractures exhibited a significant reduction in callus thickness on the convex side, while no significant change was observed on the concave side. The straight fractures displayed a side symmetrical reduction in callus (Fig. 1 A and B).
Histological Bone Repair
Hematoxylin eosin staining demonstrated cortical bridging at day 21 on the concave side (Fig. 3A) and at day 35 on the convex side (Fig. 4A) of angulated fractures. In the straight fractures, cortical bridging was observed on both sides at day 21.
Semi-Quantification of Sensory Neuropeptide Substance P
A significant increase in the number of nerve fibers immunoreactive to SP was observed both in straight (twofold) and angular (threefold) fractures compared to intact tibia over the experimental period. In straight fractures, a peak increase in SP immunoreactivity was observed side symmetrically at day 21, followed by a gradual reduction. No significant side difference was observed over the experimental period (Fig. 5A). Notably, in the angulated fractures, the peak increase in SP-positive fibers was seen at day 21 on the concave side, while at day 35 on the convex side. The SP peak was followed by a gradual decrease in SP expression. Comparative side-specific analysis demonstrated significantly higher SP density on the convex compared to the concave side at day 56 (p = 0.041), during the late remodeling phase (Fig. 5B).
To further explore whether SP fiber density on the convex side of angular fractures was higher, we also individually compared the concave and convex side between two fractures (Fig. 5C and D). No significant change of SP fiber was found on the concave (posterior) side between two fractures (Fig. 5C). Notably, the SP fiber density on the convex (anterior) side of angular fractures was significantly higher than that on the convex side of straight fractures at day 35 (p = 0.009), 56 (p = 0.009), and 84 (p = 0.016; Fig. 5D).
Immunohistochemical Analysis of Immunoreaction to SP
The immunohistochemical analysis strengthened the semi-quantitative results and demonstrated the specific localization of SP-positive nerves. By day 3 post-fracture, non-vascular SP-positive nerve fibers were identified in the fracture hematoma, and at day 7 SP fibers were found sprouting into fibrocartilage callus close to chondroid cells (Fig. 2A) and sprouting towards to the woven bone (Fig. 2B). Abundant SP fibers were seen in the hypertrophic periosteum around new bone at days 21 (Fig. 2C) and 35 (Fig. 2D). On day 21, more SP fibers were noted on the callus surface on the concave compared to the convex side of angulated fractures (Fig. 3B and C). On the contrary, on day 35 more SP fibers were observed sprouting into the callus from the periosteum on the convex compared to the concave side (Fig. 4B and C). On day 56, abundant SP nerves were still noted on the convex side sprouting into new bone from the periosteum, while on the concave side SP fibers were withdrawing. At day 84, only occasional SP fibers were seen in the bone tissue, while most nerve fibers had retracted from the remodeled callus. Comparative analysis demonstrated that the distribution of SP-positive fibers in straight fractures was quite similar to that of the concave side of angulated fractures with no significant side differences.
Double staining experiments revealed neuronal co-localization of SP with a nerve growth marker, GAP-43, at the fracture site in both groups at days 3–21. Co-localization of GAP-43 and SP was seen in the hematoma at day 3 (Fig. 6A), connective tissue, periosteum, and woven bone at day 7 (Fig. 6B), and in new bone at day 21 (Fig. 6C and D). From days 35 to 84, only sporadic signs of co-existence between GAP-43 and SP was detected.
Our study suggests that a sensory neuronal mechanism is involved in regulating fracture healing. Although the temporal and spatial occurrence of the sensory neuropeptide SP may be secondary to and elicited by mechanical loading factors, this does not preclude a role of SP in controlling bone formation and resorption at the fracture site. Thus, a close relationship exists between the occurrence of SP and the different phases of radiographic healing and modeling in both angulated and straight fractures. Nerve ingrowth in response to injury may be required for delivery of different neuromediators including SP to initiate and regulate fracture healing, as suggested for tendon healing.8, 24
The ingrowth of sprouting nonvascular SP fibers in the fracture hematoma by day 3 coincides with the early inflammatory phase of fracture healing (Fig. 6A). This complies with reports suggesting that SP has proinflammatory effects by recruiting inflammatory cells. Sprouting nonvascular SP fibers were also noted in cartilage callus and new woven bone at day 7 (Figs. 2A and 6B). The nonvascular fibers emerged from the deep layers of periosteum with ramifications in close proximity to chondrocytes (Fig. 2A). This observation complies with our previous finding that axon growth in early chondrogenesis occurs before or independently of angiogenesis.8 Notably, exogenous SP administration during chondrogenesis causes a significant increase in both chondrocyte proliferation and glycosaminoglycan production.18 In the later stage of chondrogenesis, subsequent ingrowth of vessels may trigger ossification and thereby loss of chondroid phenotype. Given that mature hyaline cartilage is normally devoid of both vessels and nerve fibers, this peculiar neuro-vascular deficiency is probably a prerequisite for maintaining a high degree of cartilage differentiation.
As for bone formation, the peak SP occurrence coincided with radiographic healing (Fig. 1A and B) and cortical bridging (Figs. 3A, 4A, and 6C and D). On the concave side of angular fractures (Figs. 3C and 5B), as on both sides of straight fractures (Figs. 1B and 5A), the SP peak was noted at day 21 concomitantly with maximum callus thickness, while on the convex side of angular fractures not until day 35 (Figs. 4B and 5B). The observed relationship between peak SP occurrence and bone healing may well be secondary to loading. Nonetheless, SP stimulates fibroblast proliferation and synthesis of different cytokines and growth factors.25–27 SP also promotes osteoblast differentiation, bone colony formation, and second messenger cyclic adenosine monophosphate production in osteoblasts.4 Therefore, our observations comply with a role of SP in orchestrating bone formation.
After the peak occurrence of SP coinciding with the time of fracture healing, the SP-positive nerve fibers gradually disappeared from the fracture site with one exception. Thus, from days 35 to 84, corresponding to the remodeling phase, an abundance of SP fibers remained on the convex side of the angulated fractures (Fig. 4B), where predominantly resorption took place contributing to correction of the angular deformity (Figs. 1A and 4A).23, 28 Although this seems to contradict the assumed stimulatory role of SP in bone formation, SP under certain circumstances might promote bone resorption. Indeed, studies corroborate a role for SP in bone resorption.21, 29, 30 Thus, SP in vitro stimulates osteoclast formation and suppresses osteoblast differentiation.31, 32 Therefore, the strikingly higher occurrence of SP fibers on the convex side during angular fracture remodeling (Figs. 4B and 5B) compared to the concave side, and on both sides of straight fractures (Fig. 5A) at the same time-points could be explained by a resorptive role of SP and by the significantly high expression of SP fibers occurring only on the convex side of angular fractures as compared with straight fractures at days 35, 56, and 84 (Fig. 5D).
The dual, opposite roles of SP in bone turnover are not easily understood. A plausible explanation pertains to different loading conditions on the sides of an angulated fracture. SP under loading on the concave side might stimulate bone formation, and on the non-loaded convex-side promote bone resorption. Notably, the SP receptor NK-1, identified on a variety of cells involved in repair and break down of bone tissue, such as fibroblasts,33 chondrocytes,17, 18 osteoblasts, and osteoclasts,4, 19–21 is involved in chondrocyte mechanotransduction in human articular cartilage.18
In fracture repair, the receptor NK-1 probably exhibits a time- and site-dependent variation in its expression on different target cells. NK-1 presumably orchestrates the various types of cellular responses elicited by SP.3, 34, 35 Moreover, mechanical loading affects cellular expression of NK-1.18, 36 Thus, 2 weeks of loading following connective tissue injury leads to a threefold up-regulation of NK-1 in healing tissue, while immobilization down-regulates NK-1 expression to levels equal to or below intact tissue levels.36 Whether the SP receptor elicits different responses in different cell types depending on the loading conditions after fracture has yet to be investigated.
This in vivo study only offers morphological and semi-quantitative findings on SP occurrence without evidence of a causal relationship to fracture healing and modeling. We have not adjusted the degrees of freedom for multiple comparisons in order not to inflate this risk any further. Since all statistical analyses were planned and theoretically based, and not exploratory, we have not corrected the analyses for multiple testing (e.g., by Tukey HSD or Bonferroni), which would have reduced the risk for type I errors but increased the risk for type II errors.
In the future, the regulatory role of SP in bone turnover should be explored by interventional in vivo studies permitting analysis of different cell types at different stages of fracture healing and remodeling, instead of exclusively relying on in vitro experiments.
This study was supported by grants from the AO/ASIF, Foundation (99-K31) and the Swedish Medical Research Council (13107).