Lnk-dependent axis of SCF–cKit signal for osteogenesis in bone fracture healing

We first attempted to compare the frequency of BM HSC/EPC fraction identified as cKit⁺SL (KSL) cells at pre-fracture and 7 d after fracture in Lnk KO mice and age-matched WT mice (C57BL/6). The percentage of BM KSL fraction was significantly higher in post-fracture bone than in pre-fracture bone, regardless of mouse phenotype; however, it was significantly high in both pre- and post-fracture bone in Lnk KO mice compared with WT mice (pre-fracture: Lnk KO, 19.3 ± 0.5, WT, 5.6 ± 0.2; post-fracture: Lnk KO, 23.7 ± 0.5, WT, 8.3 ± 0.4%, respectively; P < 0.05 for Lnk KO vs. WT in both pre- and post-fracture bone; pre- vs. post-fracture bone in both Lnk KO and WT; n = 5; Fig. 1 a). Also, the number of KSL cells in 10⁵ of PB mononuclear cells (MNCs) was significantly greater in post-fracture bone than in pre-fracture bone, regardless of mouse phenotype, and it was significantly great in both pre- and post-fracture PB in Lnk KO mice compared with WT mice (prefracture: Lnk KO, 1,141.4 ± 192.0; WT, 581.9 ± 97.7, post-fracture: Lnk KO, 1,467.1 ± 235.2; WT, 941.9 ± 300.0, respectively; P < 0.01 for Lnk KO vs. WT in both pre- and post-fracture; P < 0.05 for pre- vs. post-fracture in both Lnk KO and WT; n = 5; Fig. 1 b).

We next investigated which cell populations were mobilized into PB under fracture stress. Because the KSL cell population is extremely small in PB, there was no significant difference between the number and the percentage in PB at pre-fracture and 7 d post-fracture (unpublished data). We therefore attempted to make a comparison of the frequency of PB HSC/EPC-enriched fraction identified as cells at prefracture, 1, 4, 7, and 14 d after fracture in Lnk KO and WT mice. The percentage of post-fracture HSC/EPC-enriched fraction in PB was significantly higher, peaking at day 1 in both Lnk KO and WT mice, than that of pre-fracture (pre-fracture: Lnk KO, 39.7 ± 3.1, WT, 44.9 ± 0.8; 1 d post-fracture: Lnk KO, 56.2 ± 1.6, WT, 59.3 ± 3.1; 4 d post-fracture: Lnk KO, 53.5 ± 3.0, WT, 57.7 ± 1.1; 7 d post-fracture: Lnk KO, 43.8 ± 4.1, WT, 55.0 ± 1.2; 14 d post-fracture: Lnk KO, 41.0 ± 1.9, WT, 52.6 ± 2.4%; P < 0.01 for pre-fracture vs. 1 d post-fracture in both Lnk KO and WT; P < 0.05 for pre-fracture vs. 4 d post-fracture in both Lnk KO and WT; n = 5). The number of post-fracture SL cells per 1 ml of PB was significantly high, peaking at day 1 compared with pre-fracture (pre-fracture: Lnk KO, 15.1 ± 1.0, WT, 7.2 ± 1.1; 1 d post-fracture: Lnk KO, 35.6 ± 3.7, WT, 22.1 ± 6.1; 4 d post-fracture: Lnk KO, 29.5 ± 2.8, WT, 15.6 ± 9.7; 7 d post-fracture: Lnk KO, 25.3 ± 5.4, WT, 13.7 ± 1.9; 14 d post-fracture: Lnk KO, 24.8 ± 5.4, WT, 8.4 ± 1.9 × 10⁴ cells/ml; P < 0.01 for pre-fracture vs. 1 d post-fracture in both Lnk KO and WT; P < 0.05 for pre-fracture vs. 4 d post-fracture in both Lnk KO and WT, Lnk KO vs. WT at each time point; n = 5; Fig. 1 c). These results indicate that both fracture stress and lack of Lnk induce the mobilization of SL cells, including KSL cells from BM, resulting in a high SL/KSL cell number in PB in Lnk KO fractured mice compared with that in WT mice.

Based on the aforementioned evidence, we looked for BM-derived Tie2⁺ stem/progenitor cells in sites of fracture using a mouse BM transplantation (BMT) model. First, we generated Tie2/LacZ Lnk KO mice, in which the LacZ gene is expressed under the regulation of promotor Tie2 (lacking the Lnk gene), by crossing Tie2/LacZ transgenic and Lnk KO mice. We then performed BMT from Tie2/LacZ Lnk KO mice to Lnk KO mice and from Tie2/LacZ WT mice to WT mice. Fracture was induced 4 wk after BMT, after reconstitution of BM with Tie2/LacZ Lnk KO cells in Lnk KO and WT mice, and granulation tissue was analyzed histologically in sites of fracture 7 d after surgery. Double-fluorescent immunostaining was performed for β-galactosidase (β-gal), which can detect BM-derived Tie2⁺ cells and CD31, an EC marker, or for β-gal and osteocalcin (OC). The β-gal⁺ and CD31 double-positive ECs were frequently observed in Tie2/LacZ Lnk KO-BMT Lnk KO mice, whereas the double-positive cells were rarely identified in Tie2/LacZ WT-BMT WT mice (Fig. 2 a). The number of double-positive cells was significantly increased in Tie2/LacZ Lnk KO-BMT Lnk KO mice compared with that in Tie2/LacZ WT-BMT WT mice (Lnk KO, 410.0 ± 37.6 vs. WT, 230.0 ± 56.7/mm²; P < 0.05; n = 3; Fig. 2 b). The β-gal⁺ and OC double-positive OBs were frequently observed as lining cells along with newly formed bone surface in Tie2/LacZ Lnk KO-BMT Lnk KO mice, whereas only a few double-positive cells were identified in Tie2/LacZ WT-BMT WT mice. (Fig. 2 a) The number of double-positive cells was significantly increased in Tie2/LacZ Lnk KO-BMT Lnk KO mice compared with that in Tie2/LacZ WT-BMT WT mice (Lnk KO, 195.0 ± 31.0 vs.WT, 105.0 ± 21.5/mm²; P < 0.05; n = 3; Fig. 2 c). These results suggest that enhanced recruitment of mobilized Tie2⁺ stem/progenitor cells to fracture sites contributes to vasculogenesis and osteogenesis, leading to accelerated fracture healing in Lnk KO mice.

To explore whether SL cells have multilineage differentiation potential, specifically to ECs, OBs, and adipocytes (ADs), BM-derived SL cells were cultured in differentiation induction medium for each lineage, and each marker expression was examined by real-time RT-PCR and immunocytochemistry after culture for 7 and 14 d, respectively. The mRNA expression of EC markers (CD31: 243.8 ± 7.3 vs. 372.1 ± 7.0, P < 0.05, n = 3; vWF: 49.1 ± 1.0 vs. 396.0 ± 24.1, P < 0.01, n = 3; Fig. 2 d), OB markers (OC: 6.7 ± 0.8 vs. 182.6 ± 9.5, P < 0.001, n = 3; Collagen 1-A1: 784.6 ± 107.8 vs. 22,667 ± 2,114, P < 0.001, n = 3; Fig. 2 e), and AD markers (LPL: 1.4 ± 0.1 vs. 642.0 ± 17.2, P < 0.0001, n = 3; PPARγ: 7.1 ± 0.3 vs. 162.7 ± 6.7, P < 0.0001, n = 3; Fig. 2 f) were significantly up-regulated by culture with each lineage induction medium. These lineage marker expressions were further confirmed by immunopositivities for CD31, vWF, Flk-1 (Fig. 2 g), OC (Fig. 2 h), and adiponectin (Fig. 2 i) in SL cells. The alizarin red and oil red O staining also show characteristics of OBs and ADs, respectively (Fig. 2, h and i).

To explore specific gene expression differences under fracture stress between Lnk-deficient and WT mice, angiogenesis (96)– and osteogenesis (96)–related genes of the tissue at the fracture site spotted on cDNA microarrays were hybridized with biotin-labeled cDNA probes according to the manufacturer's instructions. In angiogenesis gene array, relative expression levels of 12 genes (ANG, Fisp12 [Ctgf], VEGFR[Flt1], HGF, IL-10, Gelatinase B [Mmp9], SR-A [Msr1], Restin [Rsn], BM40 [Sparc], ALK-5 [Tgfbr1], THBS1, and VCAM-1) of Lnk-deficient fractured mice increased by >1.5-fold, but 5 genes (Ephrin B4 [Ephb4], IFNr [Ifng], NOS3, Osteopontin [Spp1], and THBS2) decreased by >2-fold compared with WT fractured mice (Fig. 3 a). In osteogenesis gene array, 24 genes (annexin A5 [Anxa5], OC [Bglap1], Bgn, Alk-6 [Bmpr1a], BMPR2, Col11a1, Col1a1, Col1A2, Col4a6, Col6a2, Ctsk, Dcn, FLG [Fgfr1], Smad2 [Madh1], Smad3 [Madh3], Collagenase-3 [Mmp13], MMP8, Gelatinase B [Mmp9], Osf2/Cbfa1 [Runx2], SPARC, TGFb3, ALK-5 [Tgfbr1], VCAM-1, and Vdr) of Lnk-deficient fractured mice increased by >1.5-fold compared with WT fractured mice (n = 3; Fig. 3 b).

Real time RT-PCR analysis of the tissue that was RNA isolated from the peri-fracture site was also performed in the expression of several key genes as a quantitative analysis. The results demonstrated a significantly higher expression of EC markers (CD31, VE-cadherin [VE-cad], and KDR/Flk1) in Lnk KO group compared with WT group (CD31: Lnk KO, 5.803 ± 0.667, WT, 1.171 ± 0.029, P < 0.01; VE-cad: Lnk KO, 7.693 ± 0.602, WT, 1.244 ± 0.006, P < 0.01; KDR: Lnk KO, 11.082 ± 1.036, WT, 1.415 ± 0.035, respectively, P < 0.01; n = 3; Fig. 3 c). The expression of bone-related markers (OC, collagen1A1 [Col1A1], and Core binding factor 1 [Cbfa1]) were significantly enhanced in Lnk KO group compared with WT group (OC: Lnk KO, 29.707 ± 3.116, WT, 1.294 ± 0.056, P < 0.01; Col1A1: Lnk KO, 46.189 ± 0.870, WT, 1.225 ± 0.225, P < 0.01; Cbfa1: Lnk KO, 13.586 ± 1.767; WT, 0.999 ± 0.049, P < 0.01; n = 3; Fig. 3 d). These results indicate the enhancement of osteogenesis, as well as angiogenesis, at fracture sites in Lnk KO mice.

Enhanced angiogenesis and osteogenesis were further confirmed by immunohistochemistry. Vascular staining with CD31 (marker for mouse EC) 7 d after fracture demonstrated enhanced neovascularization around the endochondral ossification area in Lnk KO mice compared with that in WT mice (Fig. 4 a). Neovascularization assessed by capillary density was significantly enhanced in Lnk KO mice compared with that in WT mice (Lnk KO, 1,101.8 ± 97.7 vs. WT, 628.7 ± 28.5/mm²; P < 0.01; n = 3; Fig. 4 b).

OB staining with OC (marker for mouse OB) 7 d after fracture also revealed the augmentation of osteogenesis in newly formed bone area in Lnk KO mice compared with WT mice (Fig. 4 c). Osteogenesis assessed by OB density was significantly enhanced in Lnk KO mice compared with WT mice (Lnk KO, 111.7 ± 13.2, WT, 66.7 ± 6.5/mm²; P < 0.05; n = 3; Fig. 4 d). Callus formation was monitored radiographically to evaluate fracture healing process, and the relative callus areas detected by radiography were quantified at week 1 in Lnk KO and WT group (Fig. 4 e). The relative callus area was significant larger in the Lnk KO group compared with the WT group (Lnk KO, 22.3 ± 2.0, WT, 10.0 ± 3.0; P < 0.05; n = 3; Fig. 4 f). These results indicate the morphological enhancement of angiogenesis and osteogenesis in Lnk KO mice compared with WT mice in fracture-induced environment.

Blood perfusion and morphological fracture healing in each group was evaluated by laser Doppler perfusion imaging (LDPI) system and radiological examinations at subsequent time points after fracture. LDPI analysis (Fig. 5 a) demonstrated a severe loss of blood flow at the fracture sites 1 h after fracture creation in both groups. However, the blood flow of the fractured limb increased until 2 wk post-fracture, and then decreased slowly in KO mice. In WT mice, the blood flow showed a mild increase until 1 wk post-fracture, and then decreased. In both groups, the ratio of fractured/intact (contralateral) blood flow significantly increased by week 1 (Fig. 5 b). There was no significant difference in the ratio 1 h after fracture creation between the groups, whereas the ratio at weeks 1, 2, and 3 were significantly higher in Lnk KO mice compared with WT mice (0 wk: Lnk KO, 0.753 ± 0.067, WT, 0.707 ± 0.045; 1 wk: Lnk KO, 1.370 ± 0.081, WT, 1.144 ± 0.088; 2 wk: Lnk KO, 1.425 ± 0.103, WT, 1.110 ± 0.077; 3 wk: Lnk KO, 1.241 ± 0.064, WT, 0.972 ± 0.030; P < 0.01 for week 0 vs. weeks 1–3 in both Lnk KO and WT; P < 0.05 for Lnk KO vs. WT in week 1–3; n = 4 mice/group/timepoint; Fig. 5 b).

In 70% (14 out of 20) of Lnk KO mice at week 2 and 100% (20 out of 20) at week 3, the fracture healed with bridging callus formation radiographically, whereas in WT mice only 35% (7 out of 20) had healed at week 2 and 90% (18 out of 20) at week 3 (P < 0.05 in week 2; Fig. 5 c), which is consistent with the data in previous studies demonstrating the natural course in this animal model (Manigrasso and O'Connor, 2004). To evaluate bone remodeling, bone absorption was monitored radiographically and relative callus density was quantified at week 3 and 4 in Lnk KO and WT group. Relative callus density was significantly lower in Lnk KO group compared with that in WT group (3 wk: Lnk KO, 58.1 ± 2.4 vs. WT, 62.2 ± 6.8, ns; 4 wk: Lnk KO, 22.5 ± 2.8 vs. WT, 40.2 ± 5.3, P < 0.05, n = 20; Fig. 5 d).

Morphological and functional fracture healing in each group was further evaluated by micro-computed tomography (micro-CT), histological analysis, and biomechanical examination 28 d after fracture surgery. The micro-CT results showed striking trabecular bone formation in Lnk KO mice compared to that in WT mice (Fig. 5 e). Quantitative analysis for bone formation was performed with micro-CT images and expressed as bone density (Fig. 5 f, left) and trabecula number (Fig. 5 f, right). Both parameters were significantly high in Lnk KO mice compared with that in WT mice (bone density: Lnk KO, 10.4 ± 1.5 vs. WT, 5.0 ± 0.3, P < 0.05, n = 4; number of trabecular: Lnk KO, 2.7 ± 0.3 vs. WT, 1.3 ± 0.03, P < 0.01, n = 4; Fig. 5 f). Biomechanical examinations by three-point bending test also showed significantly increased ultimate stress ratio (Lnk KO, 1.2 ± 0.02 vs. WT, 0.8 ± 0.03, P < 0.001, n = 4), failure energy ratio (Lnk KO, 2.2 ± 0.1 vs. WT, 1.2 ± 0.1, P < 0.001, n = 4), and extrinsic stiffness ratio (Lnk KO, 1.1 ± 0.1 vs. 0.4 ± 0.04, P < 0.01, n = 4; Fig. 5 g).

Histological evaluation with toluidine blue staining demonstrated the enhanced endochondral ossification consisting of numerous chondrocytes and newly formed trabecular bone at week 1, bridging callus formation at week 2, and complete union at week 3 in Lnk KO mice. In contrast, although a callus formation was observed at week 1, bridging callus formation was rarely found at week 2 in WT mice (Fig. 5 h). The degree of fracture healing assessed by Allen's classification (Allen et al., 1980) was significantly higher in the Lnk KO group compared with the WT group at week 1 and 2 (1 wk: Lnk KO, 1.67 ± 0.21 vs. WT, 0.50 ± 0.22; 2 wk: Lnk KO, 2.67 ± 0.21 vs. WT, 1.50 ± 0.22, P < 0.01, n = 6; 3 wk: Lnk KO, 3.67 ± 0.21 vs. WT, 3.17 ± 0.17, ns, n = 6; Fig. 5 i). These results indicate that Lnk KO mice have a potential for prompt fracture healing evaluated by not only radiographical analysis but also biomechanical functional examination.

Although the aforementioned results clearly demonstrated enhanced vasculogenesis/angiogenesis and osteogenesis via SL or Tie2⁺ cell mobilization into PB and recruitment to sites of fracture in Lnk KO mice, the precise mechanism remains unclear. To explore the possible mechanism, we focused on the role of the stem/progenitor cell chemokine SCF, which is a ligand for the receptor cKit, in Lnk KO mice as well as in WT mice. Accordingly, mouse (m) SCF plasma levels were measured before and after fracture at subsequent time points in both types of mice. The plasma SCF levels were decreased after fracture surgery regardless of Lnk gene deficiency in mice; however, the extent of decrease was significantly limited during the fracture healing process and returned to over the baseline level at day 14 in Lnk KO mice compared with that in WT mice (prefracture: Lnk KO, 263.4 ± 9.4 vs. WT, 251.0 ± 35.6 pg/ml; ns and 1 d post-fracture: Lnk KO, 194.1 ± 11.7 vs. WT, 166.8 ± 12.7 pg/ml; 4 d post-fracture: Lnk KO, 197.7 ± 7.2 vs. WT, 181.5 ± 28.9 pg/ml; 7 d post-fracture: Lnk KO, 214.7 ± 20.6 vs. WT, 190.7 ± 24.0 pg/ml; and 14 d post-fracture: Lnk KO, 284.2 ± 19.7 vs. WT, 207.3 ± 16.6 pg/ml; P < 0.01 for each comparison; n = 5; Fig. 6 a).

Next, the gain and loss of function test of SCF for SL cell mobilization in PB was performed by FACS analysis. We intraperitoneally injected 20 µg/kg mSCF, soluble SCF receptor (sKit, antagonist of SCF; Turner et al., 1995; Nakamura et al., 2004), or PBS into unfractured Lnk KO or WT mice for 5 d and examined SL cell kinetics in PB. The number of SL cells significantly increased, peaking at day 7 after SCF stimulation in Lnk KO mice, but not in WT mice, and no SL cell number increase was observed in the other treatment groups (sKit treatment in Lnk KO mice and PBS treatment in Lnk KO mice and in WT mice; prestimulation: Lnk KO-PBS, 13.0 ± 1.3; Lnk KO-SCF, 13.1 ± 0.7; Lnk KO-sKit, 15.3 ± 0.2; WT-PBS, 8.1 ± 0.9; WT-SCF, 8.7 ± 0.6 × 10⁴ cells/ml, ns, n = 5; 7 d post-stimulation: Lnk KO-SCF, 73.4 ± 4.1 vs. Lnk KO-PBS, 15.3 ± 2.5; Lnk KO-sKit, 10.0 ± 0.8; WT-PBS, 8.3 ± 0.3; and WT-SCF, 12.8 ± 2.4 × 10⁴ cells/ml, P < 0.05, n = 5; and 14 d post-fracture: Lnk KO-SCF, 39.2 ± 6.2 vs. Lnk KO-PBS, 13.6 ± 9.7; Lnk KO-sKit, 12.0 ± 0.3; WT-PBS, 7.9 ± 0.9; and WT-SCF, 10.0 ± 0.9 × 10⁴ cells/ml, P < 0.01, n = 5; Fig. 6 b). Next, to identify the synergistic effect of fracture stress and SCF stimulation on SL cell mobilization, SCF with or without sKit and PBS was injected into fractured WT mice, and sKit alone was injected into fractured Lnk KO mice in the same way. In the fractured WT mice, the number of SL cells was significantly increased, peaking at day 7 post-stimulation by SCF, but not by PBS, whereas no response of SL cell number was observed in nonfractured WT mice by SCF, and the effect of SCF on SL cell mobilization was completely cancelled by coinjection of sKit. The inhibitory effect of sKit on SL cell mobilization in PB was also observed partially in Lnk KO mice with fracture (prestimulation: Lnk KO-sKit-Fx, 16.7 ± 2.1 vs. WT-PBS-Fx, 9.0 ± 2.1; WT-SCF-Fx, 8.3 ± 0.7; and WT-SCF-sKit-Fx, 9.8 ± 0.7 × 10⁴ cells/ml, P < 0.05, n = 5; 7 d post-stimulation: WT-SCF-Fx, 34.1 ± 1.4 vs. WT-PBS-Fx, 12.9 ± 3.3 and WT-SCF-sKit-Fx, 8.0 ± 1.0 × 10⁴ cells/ml, P < 0.01, n = 5, [Lnk KO-sKit-Fx, 22.1 ± 3.4 × 10⁴]; 14 d post-fracture: WT-SCF-Fx, 17.9 ± 2.0 vs. WT-PBS-Fx, 8.5 ± 1.9 and WT-SCF-sKit-Fx, 7.7 ± 1.3 × 10⁴ cells/ml, P < 0.05, n = 5, [Lnk KO-sKit-Fx, 15.0 ± 4.8 × 10⁴]; Fig.6 c).

We further confirmed that SCF-induced stem/progenitor cell mobilization mediated enhanced angiogenesis and osteogenesis in sites of fracture by SCF–cKit signaling inhibition study in vivo. The mSCF (20 µg/kg/day), sKit (20 µg/kg/day), or PBS was injected intraperitoneally into mice with fracture for 5 d, and angiogenesis/osteogenesis was evaluated by immunofluorescent staining for CD31 and OC 7 d after surgery (Fig. 6 d). As we expected, SCF stimulation significantly increased both CD31⁺ capillary density (WT-SCF⁺/sKit⁻, 830.4 ± 55.7 vs. WT-SCF⁻/sKit⁻, 578.6 ± 55.2/mm², P < 0.05, n = 5; Fig. 6 e) and OC⁺ OB density (WT-SCF⁺/sKit⁻, 495.8 ± 52.2 vs. WT-SCF⁻/sKit⁻, 350.0 ± 25.8/mm², P < 0.05, n = 5; Fig. 6 f) around the endochondral ossification area compared with PBS injection (control: SCF⁻/sKit⁻) in WT mice, and the effect of SCF on the increase of capillary density (WT-SCF⁺/sKit⁺, 601.8 ± 60.0 vs. WT-SCF⁺/sKit⁻, 830.4 ± 55.7/mm², P < 0.05, n = 5; Fig. 6 e) and OB density (WT-SCF⁺/sKit⁺, 344.6 ± 20.7 vs. WT-SCF⁺/sKit⁻, 495.8 ± 52.2/mm², P < 0.05, n = 5; Fig. 6 f) was significantly reversed by coadministration of sKit with SCF. The sKit administration also significantly reduced both capillary density (Lnk KO-SCF⁻/sKit⁺, 653.6 ± 20.4 vs. Lnk KO-SCF⁻/sKit⁻, 1,101.8 ± 97.7/mm², P < 0.01, n = 5; Fig. 6 e) and OB density (Lnk KO-SCF⁻/sKit⁺, 389.3 ± 17.3 vs. Lnk KO-SCF⁻/sKit⁻, 558.9 ± 37.9/mm², P < 0.01, n = 5; Fig. 6 f) in Lnk KO mice to levels similar to those found in WT mice with or without coadministration of SCF and sKit. These findings indicate that enhanced SL cell mobilization into PB and its recruitment to fracture sites in Lnk KO mice is, at least in part, regulated by SCF–cKit signaling pathway.

Although there is no doubt about stimulated stem/progenitor cell recruitment to sites of fracture by SCF–cKit signaling activation in WT and Lnk KO mice, the Lnk-regulated differential potential of progenitors needed to be evaluated, especially regarding osteogenesis. We isolated mouse calvarial OBs from 3–5-d-old Lnk KO mice and WT mice, and compared the differentiation capacity in osteogenic condition medium after 7, 14, and 21 d in culture. Alkaline phosphatase (ALP) staining exhibited larger nodule formation in Lnk KO OBs than in WT OBs until day 14 (Fig. 7 a). In addition, OB mineralized matrix formation assessed by alizarin red staining in Lnk KO OBs was striking compared to that in WT OBs (Fig. 7 a). Quantitative analysis for calcium content in culture medium was also significantly high in Lnk KO OBs than that in WT OBs after 21 d in culture, (Lnk KO-SCF⁻/sKit⁻, 15.9 ± 3.5 vs. WT-SCF⁻/sKit⁻, 0.6± 0.1 mg/ml, P < 0.01, n = 5; Fig. 7 d). However, the number of OB CFUs (CFU-O) showed no statistical differences between the two groups at each time point (unpublished data).

Next, we investigated whether this enhancement of differentiation and mineralized matrix formation in Lnk KO OBs was also regulated by SCF signals. First, we confirmed the significantly higher mRNA expression of SCF in Lnk KO mice than that in WT OBs by real-time RT-PCR analysis (Lnk KO, 1.707 ± 0.345 vs. WT, 0.587 ± 0.111, P < 0.05, n = 3; Fig. 7 b). In WT mice, as we expected, SCF morphologically enhanced mineralized matrix formation after 21 d in culture (WT-SCF⁺/sKit⁻), and the effect of SCF on WT OBs (WT-SCF⁺/sKit⁺) and Lnk KO OBs (Lnk KO-SCF⁻/sKit⁺) was inhibited by sKit (Fig. 7 c). Calcium content in WT OBs with SCF was significantly higher than that in WT OBs without SCF (WT-SCF⁺/sKit⁻, 7.1 ± 1.4 vs. WT-SCF⁻/sKit⁻, 0.6± 0.1 mg/ml, P < 0.05, n = 5), and the effect was reversed by coincubation of SCF and sKit in WT OBs (WT-SCF⁺/sKit⁻, 1.0 ± 0.3 vs. WT-SCF⁺/sKit⁻, 7.1 ± 1.4 mg/ml, P < 0.05, n = 5) and in Lnk KO OBs (Lnk KO-SCF⁻/sKit⁺, 4.8 ± 1.1 vs. Lnk KO-SCF⁻/sKit⁻, 15.9 ± 3.5 mg/ml, P < 0.01, n = 5; Fig. 7 d). We then further investigated the role of SCF–cKit signaling, focusing on an osteogenesis-related key molecule, bone morphogenetic protein (BMP)-2, was examined in Lnk KO OBs and WT OBs with SCF stimulation in the presence or absence of sKit. The significantly high BMP-2 mRNA expression was observed in Lnk KO OBs and SCF-treated WT OBs compared with WT OBs (Lnk KO-SCF⁻/sKit⁻, 2.732 ± 0.551 and WT-SCF⁺/skit⁻, 2.576 ± 0.369 vs. WT-SCF⁻/skit⁻, 1.189 ± 0.161, P < 0.01, n = 3 in each group) and the BMP-2 mRNA up-regulation in Lnk KO OBs and that in SCF-treated WT OBs were significantly inhibited by sKit (Lnk KO-SCF⁻/sKit⁺, 1.391 ± 0.203 vs. Lnk KO-SCF⁻/sKit⁻, 2.732 ± 0.551 and WT-SCF⁻/sKit⁺, 0.911 ± 0.136 vs. WT-SCF⁺/skit⁻, 2.576 ± 0.369, P < 0.01, n = 3; Fig. 7 e). These results indicate that BMP-2 might be involved in the SCF–cKit signaling pathway, which is critical for OB terminal differentiation and mineralized matrix formation leading to osteogenesis enhancement in Lnk KO mice.

Finally, we examined whether Lnk gene deficiency in BM stem/progenitor cells, but not in bone with surrounding tissue. To test this hypothesis, we assessed fracture healing in a murine BMT model. BM cells isolated from either Lnk KO mice or WT mice were transplanted to Lnk KO mice after lethal irradiation, and generated the following chimera mice: 1) Lnk KO mice with Lnk KO BM and 2) Lnk KO mice with WT BM. These chimera mice underwent surgery for fracture and evaluated the fracture healing by micro-CT imaging system and biomechanical analysis 28 d after surgery. The results of micro-CT exhibited striking trabecular bone formation in Lnk KO mice with Lnk KO BM than that in Lnk KO mice with WT BM (Fig. 8 a). Quantitative analysis for bone formation was performed with micro-CT images, and expressed as bone density (Fig. 8 b, left) and trabecula number (Fig. 8 b, right). The both parameters were significantly great in Lnk KO mice with Lnk KO BM compared with that in Lnk KO mice with WT BM (Bone density: Lnk KO mice with Lnk KO BM, 11.3 ± 0.4 vs. Lnk KO mice with WT BM, 8.9 ± 0.8, P < 0.05, n = 4 and Number of trabecular: Lnk KO mice with Lnk KO BM, 3.0 ± 0.2 vs. Lnk KO mice with WT BM, 1.8 ± 0.1, P < 0.05, n = 4). Biomechanical examinations by three-point bending test also showed significantly increased ultimate stress ratio (Lnk KO mice with Lnk KO BM, 1.1 ± 0.05 vs. Lnk KO mice with WT BM, 0.5 ± 0.08, P < 0.05, n = 4), failure energy ratio (Lnk KO mice with Lnk KO BM, 3.9 ± 0.07 vs. Lnk KO mice with WT BM, 1.5 ± 0.05, p<,0.001 n = 4), and extrinsic stiffness ratio (Lnk KO mice with Lnk KO BM, 0.8 ± 0.03 vs. Lnk KO mice with WT BM, 0.2 ± 0.03, P < 0.05, n = 4; Fig. 8 c). These results suggest that Lnk gene deficiency in BM cells, rather than in bone with surrounding tissue, is critical for enhanced fracture healing.

In this paper, we have demonstrated that both vasculogenesis and osteogenesis were enhanced via large-scale HSC/EPC mobilization in circulation and recruitment to sites of fracture, simultaneously resulting in accelerated bone healing in a Lnk-deficient mouse fracture model.

BM SL cells, recognized as an HSC/EPC-enriched fraction (Takahashi et al., 1999), were reported to synthesize ALP, collagen, and OC and form a mineralized matrix in culture over a decade ago (Van Vlasselaer et al., 1994). It has also been reported that BM side population cells, which contain hematopoietic repopulating cells, can also engraft in bone after transplantation (Dominici et al., 2004), and that the nonadherent population of BM cells contains primitive cells able to generate both hematopoietic and osteocytic lineage cells (Olmsted-Davis et al., 2003). Zhang et al. (2003) reported that depleting a receptor of BMP in OBs caused a doubling in both OB and HSC populations in BM niche. Calvi et al. (2003) also found a parallel expansion of the HSCs when the number of OBs was increased by parathyroid hormone infusion. These findings indicate that osteogenesis and hematopoiesis/vasculogenesis closely regulated each other in terms of microenvironmental interaction for regenerative activity in BM. We have also previously reported that mobilization of BM-derived HSCs/EPCs were triggered by fracture onset and recruited to sites of fracture (Matsumoto et al., 2008). In this study, we have provided evidence of enhanced trans-lineage differentiation of Sca1⁺/Tie2⁺ cells into OBs and ECs both in vitro and in vivo, as well as that of mobilization of BM-KSL cells and PB-SL cells in circulation in Lnk-deficient mice, which is consistent with the previous report that BM cells of Lnk-deficient mice are competitively superior in hematopoietic population to those of WT mice (Takaki et al., 2002). These data suggest that a large number of BM-derived EPCs might not only be mobilized from BM but also recruited to fracture site, differentiating into ECs and OBs by lack of Lnk signaling. In addition to the direct contribution of recruited EPCs for fracture repair, indirect contribution of EPCs, i.e., the paracrine effect on bone tissue regeneration via promotion of osteogenesis/angiogenesis with resident cells, would also be critical for fracture healing (Matsumoto et al., 2006). Indeed, cDNA microarray and quantitative real-time RT-PCR analyses also exhibited up-regulation of pro-angiogenic/-osteogenic gene expressions, which is consistent with increased blood perfusion and callus formation, in sites of fracture with recruited EPCs in Lnk-deficient mice. The series of findings suggest that both direct and indirect contribution of BM-derived EPCs to prompt and functional fracture healing with sufficient mineralization in Lnk-deficient mice.

Takaki et al. (2000, 2002) reported that Lnk acts as a negative regulator in the SCF–cKit signaling pathway. Although the role of Lnk system has gradually been clarified in the field of hematology, nothing is known about it's role in skeletal biology and bone regeneration and repair. Here, we have provided evidence that enhanced HSC/EPC mobilization into PB and its recruitment to the fracture site in Lnk-deficient mice are regulated, at least in part, by SCF–cKit signaling pathway, which was proved by gain and loss of function test using SCF and sKit. However, enhanced vasculogenesis and osteogenesis in Lnk-deficient mice was shown to be significantly superior to that of SCF-treated WT mice, suggesting that another mechanism besides SCF–cKit signaling pathway is involved in the Lnk signal lacking accelerated bone healing. Indeed, molecular cDNA array analysis showed that BMPR2 mRNA expression was up-regulated more in Lnk KO mice than in WT mice. Moreover, lack of Lnk or SCF supplement in OBs induced terminal differentiation and mineralized matrix formation with increased BMP2 gene expression, which was blocked by the SCF antagonist sKit. These findings are consistent with previous studies in which cKit expression is shown in OBs and P-2 enhances OB differentiation. (Bilbe et al., 1996; Hassel et al., 2006) Collectively, it is suggested that OB differentiation and maturation is enhanced in Lnk-deficient mice via the mechanism of BMP2-involved SCF–cKit signaling pathway.

Another interesting finding in cDNA microarray data is that expression of vascular cell adhesion molecule (VCAM)-1, a differentiation-predicting marker for osteogenesis (Fukiage et al., 2008), is significantly up-regulated in sites of fracture and EC-related markers in Lnk KO mice compared with WT mice. Fitau et al. (2006) reported that high expression of VCAM-1 at both the mRNA and protein level in Lnk-deficient mice was regulated in TNF-treated ECs by extracellular signal-related kinase (Erk) 1/2 pathways. In addition, Rhee et al. (2006) reported that Erk 1/2 expression was present within mesenchymal precursor cells during distraction osteogenesis and that Erk expression closely correlates with BMP 2/4 expression. Other studies also indicated that the activation of Erk 1/2 pathway in OBs related to mechanical strain and fluid flow (Jessop et al., 2002; Alford et al., 2003; Liu et al., 2008). Based on this evidence and our findings, the Erk1/2- and BMP-2/4–involved VCAM-1 signaling pathway is considered to be another mechanism for accelerated fracture healing via vasculogenesis and osteogenesis in Lnk KO mice, and further investigation will be required to clarify the entire mechanism for bone regeneration in a Lnk-deficient system.

In conclusion, our data provide novel evidence that the Lnk system acts as a negative regulator in the SCF–c-Kit signaling pathway and that Lnk deficiency modulates both vasculogenesis and osteogenesis via SL stem cell mobilization in PB and Tie2⁺ BM cell recruitment to sites of fracture, although Tie2⁺ cells are not necessarily defined as stem cells, but are likely EPCs. In addition, lack of Lnk signaling further enhances BMP2-induced OB matrix mineralization in vitro. These pathophysiological changes led to accelerated bone healing in Lnk-deficient mice. In the field of skeletal regeneration, recently BM-derived mesenchymal stem cells or whole BMs have been used for injured bone (Petite et al., 2000) or osteogenesis imperfecta (Horwitz et al., 1999) in preclinical studies. However, there are several issues to be resolved for open reduction, such as invasiveness in cell transplantation procedure and possible complications caused by infection. The major strength of this study relies on the concept that negative control of Lnk system for bone regeneration leads to clinical feasibility by generating a Lnk-inhibitory compound. This is the first study demonstrating physiological functions of the adaptor protein Lnk in bone regeneration and suggests that inhibition of the Lnk system could be a novel therapeutic application for genetic bone diseases and bone injuries.

Lnk^−/− mice, whose generation and genomic cloning were described previously (Takaki et al., 2000) were backcrossed with C57BL/6 (B6-Ly5.2) >10 times and paired with age-matched WT mice as controls. Mice congenic for the Ly5 locus (B6-Ly5.1) were bred and maintained at the animal facility of Institute of Physical and Chemical Research Center for Developmental Biology, Kobe, Japan. The mice were fed a standard maintenance diet and provided water ad libitum. Male 10–12-wk-old Lnk^−/− and WT mice were used in this study. The institutional animal care and use committees of the Institute of Physical and Chemical Research Center for Developmental Biology approved all animal procedures, including human cell transplantation.

To confirm the kinetics of KSL or SL cells in BM and PB, we detected KSL or SL cells at pre-fracture and 1, 4, 7, and 14 d post-fracture by FACS analysis (n = 3 at each day).

BM cells were obtained by flushing femurs and tibiae or PB cells were aspirated from the hearts of 10-wk-old Lnk KO mice or WT mice with PBS containing 5% FCS. MNCs were obtained by gradient separation onto a Ficoll Histopaque gradient. Separation of Lin⁻ cells was performed by labeling MNCs with a lin⁻ separation kit (BD) containing biotin-conjugated Mac1, B220, CD3e, Ter119, Ly6G, and CD45R antibodies, followed by streptavidin-conjugated magnetic beads and BD IMagnet separation before Lin⁻ MNCs were counted.

Mouse SL cells were isolated from BM MNCs by Lin⁺ cell depletion with MACS system (Miltenyi Biotec) followed by FACS system with an anti–mouse Sca-1 antibody (FACSAria; BD), and cultured in α-modified Eagle's medium (α-MEM; Cambrex Bio Science). After 3 wk in culture, cells (10⁵/well) were placed in a 6-well plate and further cultured only for osteogenic and adipogenic induction.

For endothelial induction, freshly isolated SL cells were cultured in EBM-2 medium supplemented with 10% FBS and EGM-2 Bullet kit (Lonza) in Pronectin F (Sanyo Chemical, Inc.)–coated culture plate for 7 d. For osteogenic and adipogenic induction, cells were cultured in α-MEM supplemented with 10% FBS. 2 mM L-glutamine, 60 µM ascorbic acid, 10 mM β-glycerophosphate and 0.1 µM dexamethasone (Sigma-Aldrich) and α-MEM supplemented with 1 µM dexamethasone, 60 µM indomethacin, and 5 µg/ml insulin (Sigma-Aldrich) for 3 wk, respectively. After the above induction cultures, each lineage differentiation was confirmed by real-time RT-PCR and fluorescent immunocytostaining for the indicated specific markers. Also, calcium deposits were detected by Alizarin red staining and formations of lipid droplets were assessed by Oil Red O staining as characteristics of OB and AD, respectively. Freshly isolated SL cells and those cultured for 3 wk in α-MEM alone were used as controls for endothelial induction and osteogenic/adipogenic induction in real-time RT-PCR analysis, respectively.

Regular flow cytometric profiles were analyzed with a FACSCalibur analyzer and CELLQuest software (BD). The instrument was aligned and calibrated daily using a four color mixture of CaliBRITE beads (BD) with FACSComp software (BD). Dead cells were excluded from the plots beads on propidium iodide (PI) staining (Sigma-Aldrich). Lineage-depleted MNCs were washed twice with HBSS containing 3.0% heat-activated FCS, and incubated with 10 µl of FcR Blocking Regent to increase the specificity of monoclonal antibodies (Miltenyi Biotec) for 20 min at 4°C, and incubated with the monoclonal antibodies for 30 min at 4°C. The stained cells were washed three times with PBS containing 3.0% FCS, and resuspended in 0.5 ml of HBSS/3%FCS/PI, and analyzed by FACScan caliber flow cytometer (BD). The following monoclonal antibodies were used to characterize the lineage-depleted MNCs: APC-conjugated anti-cKit (BD), FITC-conjugated anti-Sca1 (BD), IgG1-PE isotype controls (BD), IgG1-FITC isotype controls (BD), and PI (Sigma-Aldrich).

All surgical procedures were performed under anesthesia and normal sterile conditions. Anesthesia was performed with ketamine hydrochloride (60 mg/kg) and xylazine hydrochloride (10 mg/kg) administered intraperitoneally. We followed Manigrasso's model of closed femur fracture (Manigrasso and O'Connor, 2004). A lateral parapatellar knee incision on the right limb was made to expose the distal femoral condyle. A 2-mm wedge was made using a 27-gauge needle on the intercondyle of the femur and then a 0.5-mm-diam, stainless wire was inserted in a retrograde fashion. The wire was advanced until its proximal end was positioned stable to the greater trochanter and the distal end was cut close to the articular surface of the knee. A transverse femoral shaft fracture was then created in the right femur of each mouse by three-point bending. The wound was then irrigated with 10 cc of sterile saline and skin was closed in layers with 5–0 nylon sutures. Postoperative pain was managed by administration of subcutaneous injection of buprenorphine hydrochloride after surgery. Unprotected weight bearing was allowed immediately after the operation. The left nonfractured femur served as a control.

20 animals were assigned to each group for radiological assessment, and every additional three to five animals in each group were assigned for each study. If the fracture produced was not a stable transverse fracture or if evidence of deep infection developed then animals were excluded from the study and replaced with another animal. Thus, six mice with comminuted fractures were replaced during the experiment. No mice developed infections, as confirmed by radiograph.

Male Lnk KO mice and WT mice (C57BL6/J; CLEA Japan, Inc.) aged 6 wk were used as recipients for BMT. Transgenic mice of B6/N-TgN (Tie-2-LacZ)^182Sato (Jackson ImmunoResearch Laboratories), which were generated by backcrossing FVB/N-TgN (Tie-2-LacZ)^182Sato mice and C56BL6/J mice, were used as donors for the BMT. The procedure of BMT was performed as described previously, with some modifications (Ii et al., 2005). In brief, the recipient mice were lethally irradiated for BM ablation with 12.0 Gy and received 5 million donor BM MNCs. At 4–6 wk after BMT, by which time the BM of the recipient mice was reconstituted, surgery for fracture induction was performed. The granulation tissue of fractured BMT mice were harvested 7 and 28 d after surgery for histological, radiographical, and biomechanical function analyses.

Total RNA was obtained from tissues at the peri-fracture site at day 7 using Tri-zol (Life Technologies) according to the manufacturer's instructions. cDNAs were synthesized using 1 µg total RNA in the presence of Superscript II and Oligo (dT)₁₂₁₈ (both from Invitrogen). PCR was performed in a 20-µl reaction solution containing 2 µl 10 X PCR buffer, 150 nmol MgCl₂, 10 nmol dNTP, 20 pmol primer, 1 µl 10× diluted cDNA, and 1 U RedTag DNA polymerase (Sigma-Aldrich). PCRs were run as follows: 35 cycles of 94°C for 30 s, 55°C for 30 s and 72°C for 30 s, and a final extension for 10 min at 72°C. Primer sequences are shown in the Primer size and sequence section. Nonradioactive GEArray Q series cDNA expression array filters (Mouse Angiogenesis Gene Array [MM-009] and Mouse Osteogenesis Gene Array [MM-026]; SuperArray Inc.) were used to perform focused array analysis as described by the manufacturer. Microarray databases are available at the National Center for Biotechnology Information Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) under accession no. GPL1121 for MM-009 and GPL1143 for MM-026. In brief, biotin dUTP-labeled cDNA probes were generated in the presence of a designed set of gene-specific primers using total RNA (2 µg per filter) and 200 U MMLV reverse transcription (Promega, Madison, WI, USA). The array filters were hybridized with biotin-labeled probes at 60°C for 17 h. Filters were first washed twice with 2× saline sodium citrate buffer (SSC)/1% sodium dodecyl sulfate (SDS) and then twice with 0.1× SSC/1% SDS at 60°C for 15 min each. Chemiluminescent detection steps were performed by subsequent incubation of the filters with ALP-conjugated streptavidin and CDP-Star substrate. The images were captured using ChemoFlure 8900 (Alpha Innotech). For data analysis, positive and negative spots were independently identified and verified by at least two people. Only matched positive and negative results from two independent experiments were used for analysis. For quantification, intensity of spots was first measured by National Institutes of Health image software (ImageJ) and then the mean intensities derived from the blank spots were subtracted. These subtracted intensities were divided by the mean intensities from GAPDH (three spots in each array), to obtain a relative intensity for each spot. These relative intensities were used to compare gene express levels between the control and the stimulated groups.

Total RNA was obtained from tissues of peri-fracture site at day 7 or from cultured OBs using TRIzol (Life Technologies) according to the manufacturer's instructions. After total RNA isolation, we made single-strand cDNA using a reverse transcription kit (Invitrogen) and used it as template for real-time PCR with SYBR Green PCR Master Mix (Applied Biosystems) and gene-specific primers in an Oligo software (Takara Bio Inc.). The mean cycle threshold values from quadruplicate measurements were used to calculate the gene expression, with normalization to GAPDH as an internal control.

mCD31 (224 bp): sense 5′-TCCCCACCGAAAGCAGTAAT-3′; antisense 5′-CCCACGGAGAAGTACTCTGTCTATC-3′, mVE-cad (369 bp): sense 5′-GAGCTAAGAGGACCCTCTGCTACTC-3′; antisense 5′-TGGGCCTCTTTGTGTCTGTATG-3′, mFlk-1 (346 bp): sense 5′-TGGCGTTTCCTACTCCTAATGA-3′; antisense 5′-GAAGCCACAACAAAGCTAAAATACTGAG-3′, mvWF (20 bp): sense 5′-ACGCCATCTCCAGATTCAAG-3′; antisense 5′-AAGCATCTCCCACAGCATTC-3′, mOC (187 bp): sense 5′-CTGACCTCACAGATCCCAAGC-3′; antisense 5′-TGGTCTGATAGCTCGTCACAA-3′, mCol1A1 (107 bp): sense 5′-CAATGGTGAGACGTGGAAAC-3′; antisense 5′-GGTTGGGACAGTCCAGTTCT-3′, mCbfa1 (151 bp): sense 5′-AACGATCTGAGATTTGTGGGC-3′; antisense 5′-CCTGCGTGGGATTTCTTGGTT-3′, mLPL (65 bp): sense 5′-AGCTGGGAGCAGAAACTGTG-3′; antisense 5′-CATGTGGGTTGGTGTTCAGA-3′, mPPARγ (50 bp): sense 5′-CTGGCCTCCCTGATGAATAA-3′; antisense 5′-AATCCTTGGCCCTCTGAGAT-3′, mSCF (72 bp): sense 5′-CCAAAAGCAAAGCCAATTACAAG-3′; antisense 5′-AGACTCGGGCCTACAATGGA-3′, mBMP-2 (73 bp): sense 5′-TCACTTATAGCCGCATTATCTTCTTC-3′; antisense 5′-TTGGTTTATCCATGAGGCTAACTG-3′, and mGAPDH (484 bp): sense 5′-GTGAGGCCGGTGCTGAGTATG-3′; antisense 5′-AGGCGGCACGTCAGATCC-3′.

Mice were euthanized with an overdose of ketamine and xylazine. Bilateral femurs were harvested and embedded in OCT compound, snap frozen in liquid nitrogen, and stored at −80°C for histochemical staining and immunohistochemistry. Rat femurs in OCT blocks were sectioned, and 6-µm serial sections were collected on slides, followed by fixation with 4.0% paraformaldehyde at 4°C for 5 min, and then stained immediately.

Immunohistochemical staining with anti–mouse rat CD31 (Biogenesis) for mouse EC marker or anti–mouse goat OC (Santa Cruz Biotechnology, Inc.) for mouse OB was performed. The secondary antibodies for each immunostaining are as follows: FITC-conjugated anti–rat IgG (H+L; Jackson ImmunoResearch Laboratories) for CD31 and FITC-conjugated anti–goat IgG (H+L; Jackson ImmunoResearch Laboratories) for OC. Capillary or OB density was morphometrically evaluated by histological examination of 5 randomly selected fields of tissue sections recovered from segments of soft tissue in the peri-fracture site. Capillaries were recognized as tubular structures positive for CD31. OB-like cells were recognized as lining or floating cells positive for OC on new bone surface. All morphometric studies were performed by two examiners who were blinded to treatment.

LDPI (Moor Instrument; Wardell et al., 1993; Linden et al., 1995) was used to record serial blood flow measurements over the 3 wk post-fracture course. In these digital color-coded images, red hue indicated regions with maximum perfusion; medium perfusion values are shown in yellow; lowest perfusion values are represented as blue. This was done under anesthesia with the animal supine and both limbs fully fixed.

To detect Tie2⁺ cell-derived neovascularization in Lnk KO mice with Tie2/LacZ Lnk KO BM at the fracture site, double immunohistochemistry was performed with anti–mouse rat CD31 (1:50; BD) or anti–mouse goat OC (1:250; Biogenesis) and anti–rabbit β-gal (1:1,000; Cortex) to detect Tie2⁺ cell-derived ECs or OBs. For characterization of multilineage cell types in SL cells, the following primary antibodies were used: anti–mouse rat CD31 (1:50; BD), anti–mouse goat vWF (1:100; Santa Cruz Biotechnology, Inc.), anti–mouse rat Flk-1 (1:100; Chemicon), anti–mouse goat OC (1:250; Biogenesis), and anti–mouse mouse adiponectin (1:250; Abcam). The secondary antibodies (1:1,000) for each immunostaining are as follows: FITC-conjugated anti–rat IgG (H+L; Jackson ImmunoResearch Laboratories) for CD31 and Flk-1, FITC-conjugated anti–goat IgG (H+L; Jackson ImmunoResearch Laboratories) for OC and vWF, Alexa Fluor 594–conjugated rabbit anti–mouse IgG (H+L) for adiponectin, and Cy3-conjugated goat anti–rabbit IgG (H+L; Jackson ImmunoResearch Laboratories) for β-gal staining. DAPI solution was applied for 5 min for nuclear staining.

Radiographs of the fractured legs were serially taken at weeks 0, 1, 2, 3, and 4 after creation of the fracture. This procedure was done under anesthesia with the animal supine and both limbs fully extended. Fracture union was identified by the presence of bridging callus on two cortices. Radiographs of each animal were examined by three observers who were blinded to treatment. To evaluate the fracture healing process, callus formation was monitored radiographically and relative callus areas detected by radiography were quantified with National Institutes of Health image at week 1 in both groups. To evaluate bone remodeling, callus absorption was monitored radiographically and relative callus density was detected by radiography were quantified with ImageJ at week 3 and 4 in all groups.

For quantification of callus/trabecular bone formation, micro-CT imaging analysis was performed 4 wk after fracture surgery in Kureha Special Laboratory. In brief, bone density and number of trabecula in callus area were calculated with CT intensity in scanned images by single energy x-ray absorptiometry method and averaged.

Biomechanical evaluation was performed with mice at week 4 after fracture in Kureha Special Laboratory. In brief, fractured femurs and the contralateral nonfractured femurs were prepared and intramedullary fixation pins were removed before the bending test. The standardized three-point bending test was performed using load torsion and bending tester “MZ-500S” (Maruto Instrument Co., Ltd.). The bending force was applied with cross-head at a speed of 2 mm/minute until rupture occurred. The ultimate stress (n), the extrinsic stiffness (n/mm) and the failure energy (n × mm) were interpreted and calculated from the load deflection curve. The relative ratio of the fractured (right) femur to nonfractured (left) femur was calculated in each group and averaged.

Histological evaluation (n = 3 in each group) was performed with toluidine blue staining to address the process of endochondral ossification on weeks 1, 2, and 3. The degree of fracture healing was evaluated at week 1, 2, and 3 in each group using a five point scale proposed by Allen et al. (1980). According to this classification system, grade four represents complete bony union, grade three represents an incomplete bony union (presence of a small amount of cartilage in the callus), grade two represents a complete cartilaginous union (well-formed plate of hyaline cartilage uniting the fragments), grade one represents an incomplete cartilaginous union (retention of fibrous elements in the cartilaginous plate), and grade zero indicates the formation of a pseudoarthrosis (most severe form of arrest in fracture repair). All morphometric studies were performed by two orthopedic surgeons who were blinded to the treatment.

SCF plasma levels of mouse were measured at prefracture and 1, 4, 7, and 14 d post-fracture in both groups via ELISA kit (R&D Systems) according to the manufacturer's instructions (n = 3 in each group).

Mice were injected intraperitoneally with 20 µg/kg SCF or PBS for 5 d in un-fractured Lnk KO, WT, and fractured WT mice, and mobilization of SL cells to PB was assessed at prefracture, day 7, and day 14 post-stimulation time point by FACS analysis (n = 5 in each group). For the other series, SCF or PBS-treated WT mice of femur fracture model were sacrificed at day 7, and tissue samples of the fractured limbs were assessed by capillary and OB density as mentioned in Morphometric evaluation of capillary and OB density.

Calvarial cells were isolated from 3–5-d-old mice using a modification of the method described by Wong and Cohn (1975). In brief, after removal of sutures, calvariae were subjected to four sequential 15-min digestions in an enzyme mixture containing 0.05% trypsin (Invitrogen) and 0.1% collagenase P (Boehringer Mannheim) at 37°C on a rocking platform. Cell fractions 2–4 were collected, and enzyme activity was stopped by the addition of an equal volume of DME containing 10% FCS, 100 U/ml of penicillin, and 100 µg/ml of streptomycin (Invitrogen). The fractions were pooled, centrifuged, resuspended in DME containing 10% FCS, and filtered through a 70-µm cell strainer. Cells were plated at a density of 10⁴ cells/well in 35-mm culture plates in DME containing 10% FCS. The medium was changed 24 h later, and 3 d later cultures were fed again. At 1 wk of culture, the medium was changed to a differentiation medium (α-MEM containing 10% FCS, 50 µg/ml of ascorbic acid, and 4 mM of β-glycerophosphate) and thereafter the medium was changed every 2 d. CFU-O was scored at day 7, 14, and 21 of incubation by in situ observation of plates on an inverted microscope. Effects of SCF and sKit on CFU-O formation and mineralization were also investigated in the above assay. Purified recombinant human SCF and sKit were purchased from R&D Systems and supplemented concentrations of cytokines used for culture were as follows: SCF 100 ng/ml and sKit 10 µg/ml (Nakamura, 2004).

Histochemical staining for ALP was performed at day 7 and 14 using a commercially available kit (Muto-Kagaku) according to the manufacturer's instruction. Mineralization was assessed using alizarin red staining method. In brief, the cultures were rinsed twice with PBS, fixed in 100% ethanol for 30 min, and stained wth 1% alizarin red S (Hartman Leddon Co.) in 0.28% ammonia water for 10 min at room temperature. The stained cell layers were washed, rinsed twice with distilled water, and air-dried.

Calcium content was measured by the orthocresolphthalein complexone (OCPC) colorimetric method (Sigma-Aldrich). In brief, cell layers at day 21 were washed with PBS twice, and then incubated with 50 µl of 0.6 N hydrochloric acid overnight at room temperature. Plates were then vortexed and centrifuged for 20 min at 1,600 g at 25°C and 10 µl of sample was added to 100 µl OCPC solution in fresh wells. The plate was incubated at room temperature for 10 min, and then read at 575 nm. Standards were prepared from a CaCl₂ solution and the results were expressed in milligram/milliliter per 10,000 seeded cells.

All values were expressed as mean ± SEM. Paired Student's t tests were performed for comparison of data before and after fracture. The comparisons among groups were made using the one-way analysis of variance. Post hoc analysis was performed by Fisher's PLSD test. A p-value < 0.05 was considered to denote statistical significance.

The authors state that there is no conflict of interest.