Fgf-9 is required for angiogenesis and ... - Semantic Scholar

Download PDF
2 downloads 0 Views 1MB Size Report
Comment
Jun 29, 2010 - lethal phenotypes, finally exhibit rhizomelia, i.e., a shortening of proximal skeletal elements (11). An additional hint for the regu- latory in vivo ...
Fgf-9 is required for angiogenesis and osteogenesis in long bone repair Björn Behra,b, Philipp Leuchta,c, Michael T. Longakera,1, and Natalina Quartoa,d,1 a Children’s Surgical Research Program, Department of Surgery and cDepartment of Orthopedic Surgery, Stanford University School of Medicine, Stanford, CA 94305; bBG-Unfallklinik Ludwigshafen, Department of Plastic and Hand Surgery, University of Heidelberg, Heidelberg 39120, Germany; and dDepartment of Structural and Functional Biology, University of Naples Federico II, Complesso M. S. Angelo, Naples 80125, Italy

Bone healing requires a complex interaction of growth factors that establishes an environment for efficient bone regeneration. Among these, FGFs have been considered important for intrinsic bonehealing capacity. In this study, we analyzed the role of Fgf-9 in long bone repair. One-millimeter unicortical defects were created in tibias of Fgf-9+/− and wild-type mice. Histomorphometry revealed that half-dose gene of Fgf-9 markedly reduced bone regeneration as compared with wild-type. Both immunohistochemistry and RT-PCR analysis revealed markedly decreased levels of proliferating cell nuclear antigen (PCNA), Runt-related transcription factor 2 (Runx2), osteocalcin, Vega-a, and platelet endothelial cell adhesion molecule 1 (PECAM-1) in Fgf-9+/− defects. μCT angiography indicated dramatic impairment of neovascularization in Fgf-9+/− mice as compared with controls. Treatment with FGF-9 protein promoted angiogenesis and successfully rescued the healing capacity of Fgf-9+/− mice. Importantly, although other pro-osteogenic factors [Fgf-2, Fgf-18, and bone morphogenic protein 2 (Bmp-2)] still were present in Fgf-9+/− mice, they could not compensate for the haploinsufficiency of the Fgf-9 gene. Therefore, endogenous Fgf-9 seems to play an important role in long bone repair. Taken together our data suggest a unique role for Fgf-9 in bone healing, presumably by initiating angiogenesis through Vegf-a. Moreover, this study further supports the embryonic phenotype previously observed in the developing limb, thus promoting the concept that healing processes in adult organisms may recapitulate embryonic skeletal development. tibia

| regeneration | tissue

B

one healing is an efficient regenerative process resulting in newly formed bone equivalent to the original tissue. It involves the interplay of numerous factors over the well-characterized cascade of phases, including inflammation, callus formation, and remodeling (1), which recapitulate aspects of skeletal development (2, 3). Among these factors, several members of the FGF ligands, including FGF-2, -9, and -18, as well their corresponding receptors FGFR1–3, have been identified previously as having a major role during skeletal development (4–12). Although their contribution during skeletal development is well established, less is known about FGFs in fracture healing. The recent identification of different expression patterns of FGF ligands and receptors during fracture repair (13) indicates potential functions in bone regeneration. Within the group of pro-osteogenic FGF ligands, Fgf-9 has been reported to play a role during the development of skeletal vascularization (11). Skeletal vascularization of the injury site is a key step in facilitating successful bone regeneration. Angiogenesis is achieved by two hormonal pathways: the angiopoietin- and VEGF-mediated pathways (2, 14). Vegf was identified previously as a coordinator of chondrocyte growth and bone formation in growth plates of juvenile mice (15). Moreover, Vegf is released physiologically in fracture hematoma (16). Current knowledge about the interaction between Fgf-9 and Vegf arose mostly from embryonic studies. In a study by Hung et al. (11), femurs of Fgf-9−/− day 15.5 (E15.5) and day 16.5 (E16.5) embryos showed delayed vascular invasion and decreased expression of Vegf in hypertrophic chonwww.pnas.org/cgi/doi/10.1073/pnas.1003317107

drocytes as compared with wild-type mice (11). Conversely, exogenous FGF-9 did induce Vegf expression and vascular invasion in organ-cultured forelimbs of heterozygotic Vegf-LacZ embryos. Moreover, during limb skeletal development, Fgf-9 was shown to be expressed in the proximal limbs, which showed delayed skeletogenesis at E15.5 and E16.5. Fgf-9–null mice, which develop lethal phenotypes, finally exhibit rhizomelia, i.e., a shortening of proximal skeletal elements (11). An additional hint for the regulatory in vivo function of Fgf-9 in angiogenesis was derived from studies on lung tissue (17). In the lung mesenchyme, Fgf-9 was shown to regulate Vegf-a expression. By investigating the effects of loss of function (Fgf-9−/−) and gain of function (Fgf-9dox(48)) in embryonic mice, Fgf-9 (in concert with Sonic Hedgehog) was found to be necessary for the branching of the distal capillary plexus of the lung (17). The role of Fgf-9, in part mediating vascularization and growth in bones, suggested that Fgf-9 might be important during various stages of bone healing. In the current study, we hypothesized that haploinsufficiency of Fgf-9 might repress crucial stages in bone repair, and therefore we investigated bone repair in Fgf-9+/− and wild-type mice. Results Tibias of Fgf-9+/− and Wild-Type Mice Showed No Differences in Phenotype. As a first step we investigated whether macroscopic

differences between tibias of Fgf-9+/− and wild-type mice existed. CT scanning revealed no obvious differences between the two strains (Fig. 1A). Moreover, bone mineral densities were similar in Fgf-9+/− and wild-type mice (Fgf-9+/−: 816 ± 181 mg/cm2; wild-type: 774 ± 128 mg/ cm2) (Fig. 1B). Importantly, quantitative real-time PCR analysis of typical osteogenic markers such as Runtrelated transcription factor 2 (Runx2), alkaline phosphatase, and osteocalcin, as well as proliferative and pro-osteogenic genes, such as Pcna and Bmp-2, did not reveal significant differences in the diaphysis of Fgf-9+/− and wild-type tibia (Fig. 1C). Tibial Bone Healing Is Impaired in Fgf-9+/− Mice. To test how phenotypical unremarkable tibias with reduced levels of Fgf-9 would regenerate, 1-mm unicortical defects were created (Fig. 1D), and histomorphometry was performed postoperatively. Aniline bluestained slides (for identification of new osteoid matrix within the defect) of Fgf-9 and wild-type mice were analyzed 5 and 7 d after injury, times that represent the stages of callus formation and remodeling in this model (Fig. 2A). In Fgf-9+/− mice, bone regeneration was reduced dramatically, by 90.1% at day 5 and by 79.6% at day 7 compared with wild-type mice. In defects in wild-

Author contributions: B.B., P.L., and N.Q. designed research; B.B. and P.L. performed research; B.B., M.T.L., and N.Q. analyzed data; and B.B. and N.Q. wrote the paper. The authors declare no conflict of interest. *This Direct Submission article had a prearranged editor. 1

To whom correspondence may be addressed. E-mail: [email protected] or quarto@ unina.it.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1003317107/-/DCSupplemental.

PNAS | June 29, 2010 | vol. 107 | no. 26 | 11853–11858

DEVELOPMENTAL BIOLOGY

Edited* by Clifford J. Tabin, Harvard Medical School, Boston, MA, and approved May 24, 2010 (received for review March 16, 2010)

Fig. 1. (A) CT scans of tibia in Fgf-9+/− and wild-type mice showed no apparent differences in phenotype. (B) Bone marrow density (BMD) measurements were in the same range, and no difference could be observed. (C) Real-time PCR analysis did not reveal significant differences in the expression profile of osteogenic and proliferative markers. (D) Model of tibia injury. Unicortical tibia injuries were performed by drilling a 1-mm hole in one cortex, leaving the opposite cortex intact. Alk ph, alkaline phosphatase; bm, bone marrow; cb, cortical bone; is, injury site; Oc, osteocalcin.

type mice, regeneration occurred not only between the two cortical bones but also medially, in the bone marrow. The histomorphometry data demonstrated impaired healing of tibial defects in Fgf-9+/− mice. We then asked whether proliferation was affected in defects in Fgf-9+/− mice. Immunohistochemistry for PCNA revealed numerous proliferating cells in the defects in wild-type mice; markedly fewer cells were positive for PCNA in Fgf-9+/− mice at day 3, the stage of inflammation and angiogenesis (Fig. 2B). Immunohistochemistry for Runx2, an early osteogenic marker, showed intense nuclear staining in the defect area of wild-type mice at day 3 (Fig. 2B). In Fgf-9+/− mice, only faint staining was detected.

Moreover, immunohistochemistry for osteocalcin revealed staining in defects in wild-type mice at day 5, but staining was almost absent in defects in Fgf-9+/− mice (Fig. 2B). Analysis of the time course of gene expression in the tissue harvested from the defect sites mirrored the results obtained from immunohistochemistry (Fig. 2C). Throughout the time course, defects in wild-type mice elaborated higher levels of Runx2. Expression of Osteocalcin peaked at day 7 in wild-type mice but was undetectable in Fgf-9+/− mice (Fig. 2C). As the next step, we wanted to pinpoint whether the impaired osteogenesis was unique to Fgf-9 and whether compensatory up-regulations of other osteogenic Fgf-family members or Bmp-2 would occur. Real-time PCR expression profiles of three pro-osteogenic Fgf family members (Fgf-2, Fgf-9, and Fgf-18) in wild-type and Fgf-9+/− mice revealed lower expression levels of all Fgfs in uninjured Fgf-9+/− tibias (Fig. 2D). Of note, Fgf-9 expression was not detectable in Fgf-9+/− tibia. In tibias of Fgf-9+/− mice, Fgf-2 increased to levels similar to those in tibias of wild-type mice at days 5 and 7, but Fgf-9 did not. Moreover, expression of Fgf-18 increased to levels comparable to those in wild-type mice throughout the injury response. Interestingly, expression of Bmp-2, the prototypical osteogenic Bmp, was not impaired in tibial injuries of Fgf-9+/− mice (Fig. 2D). Neovascularization Is Impaired During Tibial Regeneration of Fgf-9+/− Mice. The severe impairment of bone regeneration led us in-

vestigate whether the reduced bone regeneration was caused by impaired neovascularization. To verify this possibility, we performed immunohistochemistry for platelet endothelial cell adhesion molecule 1 (PECAM-1), an endothelial marker (Fig. 3A). During the phase of inflammation and angiogenesis 3 d after injury, only faint staining was observed for PECAM-1 in the tibial defects of Fgf-9+/− mice. In contrast, in the defect area of wild-type mice, specific and strong staining was observed in endothelial cells of presumably regenerating vessels. Immunohistochemistry for VEGF-A, a factor that might be an effector protein of Fgf-9, was performed. Congruent with data obtained for PECAM-1, immunostaining for VEGF-A was faint in the vicinity of tibial defects in Fgf-9+/ −mice, but strong staining for VEGF-A was observed in defects in wild-type mice after 3 d (Fig. 3A). Real-time PCR analysis performed on tibial defects at days 3 and 5 revealed less expression of Vegf-a in Fgf-9+/− mice than in wild-type mice.

Fig. 2. Bone healing was impaired in Fgf-9+/− mice. (A) (Upper) Aniline blue staining of unicortical tibial defects at days 5 (Left) and 7 (Right). At both time points, bone regeneration was markedly reduced in Fgf-9+/− mice compared with wild-type mice. The injury site is marked in orange and is segregated in the Insets. (Lower) The amount of new bone formation was quantified with histomorphometry performed on aniline blue-stained slides, revealing a remarkable drop in bone regeneration in Fgf-9+/− mice. ***P < 0.0005. (B) Immunohistochemistry for PCNA, Runx2, and osteocalcin in Fgf-9+/− and wild-type mice. PCNA staining revealed fewer proliferating cells in Fgf-9+/− mice. For Runx2, only faint staining was observed at day 3 in defects of Fgf-9+/− mice, whereas strong nuclear staining was observed in defects in wild-type mice. At day 7, no staining for osteocalcin was observed in Fgf-9+/− mice, whereas staining was observed in wild-type mice. Dashed lines indicate the cortical bone. (Scale bars, 200 μm in A and 50 μm in B.) (C) RT-PCR analysis of Runx2 and osteocalcin (Oc) performed on regenerating bone tissue harvested from Fgf-9+/− and wild-type mice. (D) Real-time PCR analysis for Fgf-2, Fgf-9, Fgf-18, and Bmp-2. *P < 0.05, **P < 0.005 calculated using Student’s t test.

11854 | www.pnas.org/cgi/doi/10.1073/pnas.1003317107

Behr et al.

performed μCT angiography (Fig. 3C). These angiographies revealed marked impairment of neovascularization in tibial defects in Fgf-9+/− mice compared with wild-type mice. The vessel volume in the defect area was 118 μm3 in Fgf-9+/− mice, decreased by 3-fold compared with the volume of 355 μm3 in wild-type mice (P < 0.05). Moreover, the vessel surface was decreased in Fgf-9+/− mice (7.85 mm2) compared with wild-type mice (14.75 mm2) (P < 0.05). These data revealed that at day 7 Fgf-9+/− mice had a severely impaired angiogenic response to injury.

during the injury response of Fgf-9+/− and wild-type mice. Next we evaluated whether there were any significant changes in bone marrow-derived osteoclastogenesis. Tartrate-resistant acid (TRAP) staining revealed impairment of osteoclast recruitment in the defects in Fgf-9+/− mice compared with defects in wild-type mice at day 7 (Fig. 4). In wild-type mice, TRAP-positive cells were found predominantly in the medial aspect of the callus. This result is suggestive of tissue remodeling. In contrast, only a few TRAP-positive cells were found at the bony edges of defects in Fgf-9+/− mice. However, it must be pointed out that bone formation was decreased in Fgf-9+/− mice; thus, the decreased TRAP staining in Fgf-9+/− mice also could be secondary to diminished bone formation. In addition, matrix metallopeptidase 9 (MMP-9) staining, a marker for chondroclastic and osteoclastic cells, revealed less staining in the injury site of Fgf-9+/− mice 7 d postoperatively. Of note, bone marrow stained strongly for MMP-9. Defects in Fgf-9+/− Mice Could Be Rescued with FGF-9 and VEGF-A.

Fig. 3. Neovascularization was impaired in bone regeneration of Fgf-9+/− mice. (A) Immunohistochemistry for PECAM-1 and VEGF-A in Fgf-9+/− and wildtype mice. At day 3, only faint PECAM-1 staining was observed in defects of Fgf-9+/− mice, whereas intense PECAM-1 staining was detected in wild-type mice. Arrows indicate blood vessels. Immunohistochemistry for VEGF-A showed less staining in Fgf-9+/− than in wild-type mice at day 3. Dashed lines indicate cortical bone. (B) Real-time PCR time-course analysis of regenerating bone tissue from Fgf-9+/− and wild-type mice for Vegf-a, VegfR1, and VegfR2. *P < 0.05; **P < 0.005; ***P < 0.0005 calculated using Student’s t test. (C) μCT angiography of defects in Fgf-9+/− and wild-type mice at day 7. Angiography revealed impaired neovascularization in defects of Fgf-9+/− mice. The histogram shows that both vessel volume and surface were decreased in defects of Fgf-9+/− mice. *P < 0.05. (Scale bars, 50 μm in A and 200 μm in C.)

However, at day 7, expression of Vegf-a in Fgf-9+/− mice was similar to that in wild-type mice (Fig. 3B). Expression of Vegf receptor 1 (VegfR1) showed a similar trend: Expression was highest at day 7 in both Fgf-9+/− and wild-type mice (Fig. 3B). Vegf receptor 2 (VegfR2) expression was not up-regulated in defects in Fgf-9+/− mice, whereas an up-regulation occurred in defects in wild-type mice at days 3 and 5. Immunohistochemistry and real-time PCR data provided some clues for the diminished healing (e.g., decreased neovascularization in Fgf-9+/− mice). To investigate this evidence more thoroughly, we Behr et al.

Given that tibial defects in Fgf-9+/− mice healed significantly less than in wild-type mice, presumably because of impaired angiogenesis, we wanted to determine whether we could rescue defects in Fgf-9+/− mice with recombinant FGF-9. First, we applied FGF-9 or PBS (as a control) on a collagen sponge and investigated the angiogenic response after 3 d (Fig. 5A). PECAM-1 staining revealed vessel formation in Fgf-9+/− mice treated with FGF-9, whereas no vessels were observed in the control. A similar trend was detected for VEGF-A. Having demonstrated that angiogenesis in Fgf-9+/− mice could be rescued by application of FGF-9, we next asked whether osteogenesis could be rescued by application of FGF-9 protein. Bone formation was assessed by aniline blue staining at day 7 (Fig. 5B). Healing with FGF-9 was accelerated by 401% as compared with PBS-treated defects in Fgf-9+/− mice (P < 0.0005). Here, new bone formation was found predominantly in the periphery of the collagen sponge. To understand, whether this effect was caused mainly by the increased angiogenesis elicited by FGF-9 or by osteogenesis, we also rescued Fgf-9+/− tibia with FGF-2, VEGF-A, and a combination of VEGF-A and FGF-9. FGF-2 did enhance healing by 55% as compared with PBS; however, this value was not statistically significant. Importantly, significant acceleration of healing could be achieved with VEGF-A (251%; P < 0.005) but was lower than with FGF-9 treatment (Fig. 5C). The combination of FGF-9 and VEGF-A further increased the healing rate by 535% and showed a trend to enhance healing further (P < 0.0005), although the improvement was statistically nonsignificant compared with FGF-9 treatment alone (Fig. 5C). However, combined treatment significantly increased healing compared with VEGF-A treatment alone (P < 0.05). PECAM-1 staining in the defect area at day 7 revealed vessel formation in all treated groups but not in the control group (Fig. 5B, Right column), further indicating the importance of FGF-9 in angiogenesis. Exogenous FGF-9 Accelerates Bone Repair in Wild-Type Mice. Having demonstrated the importance and requirement for Fgf-9 during bone repair, we asked whether bone regeneration already is maximal in normal physiological conditions. We therefore apPNAS | June 29, 2010 | vol. 107 | no. 26 | 11855

DEVELOPMENTAL BIOLOGY

Osteoclast Recruitment and Bone Remodeling Impairment in Fgf-9+/− Mice. To this point, we had focused on differences in osteogenesis

Fig. 4. Osteoclastogenesis was impaired in Fgf-9+/− mice. Bone regeneration was impaired in Fgf-9+/− compared with wild-type mice at day 7 (Top row, aniline blue staining,). On adjacent sections, faint TRAP staining was observed in defects of Fgf-9+/− compared with wild-type mice at day 7 (Second row). Immunohistochemistry for MMP-9 revealed less staining in the regenerating bone of Fgf-9+/− mice than in wild-type mice (Third row). Boxed areas are enlarged in the bottom row. Dashed lines indicate the cortical bone. Arrowheads indicate MMP-9–positive cells. bm, bone marrow. (Scale bars, 200 μm for aniline blue and TRAP staining and 100 μm for MMP-9 staining.)

plied exogenous FGF-9 to defects in wild-type mice and evaluated the amount of newly formed bone. After 7 d, we found that in wild-type mice treatment with FGF-9 resulted in an increase in bone regeneration of 58.2% as compared with untreated defects (P < 0.005) and 118.1% (P < 0.0005) as compared with defects treated with PBS (Fig. S1). Collagen sponges did impair the endogenous bone regeneration capacity of defects in wildtype mice, although the impairment was not statistically significant (29% decrease in bone regeneration as compared with untreated defects in wild-type mice). Discussion In this study, we demonstrate the importance of Fgf-9 in bone healing and provide data showing that Fgf-9 contributes functionally to both angiogenic and osteogenic processes during bone repair. Uninjured tibias of Fgf-9+/− mice appeared normal and had no obvious phenotype; however; the half-dose gene of Fgf-9 was sufficient to induce a severely impaired healing phenotype. We also found that Fgf-9+/− mice had impaired angiogenesis, suggesting that Fgf-9 is important for inducing an angiogenic response during injury repair. The decreased levels of Runx2 and osteocalcin observed strongly corroborated impaired bone regeneration in Fgf-9+/− mice. Moreover, effects such as mitogenic activity of FGF-9 might explain decreased cell proliferation in 11856 | www.pnas.org/cgi/doi/10.1073/pnas.1003317107

PCNA-stained defects in Fgf-9+/− mice. Importantly, although other osteogenic factors such as Fgf-18 or Bmp-2 were present at the injury site, they could not compensate for the lack of Fgf-9. Although bone is a highly vascularized organ, the importance of angiogenesis for a successful bone repair is only partially elucidated (15, 18, 19). Decreased angiogenesis in tibial defects of Fgf-9+/− mice might account for decreased recruitment of osteoprogenitor cells that participate in bone regeneration. One of the major players in angiogenesis is VEGF, which, interestingly, itself increased osteogenesis (20, 21). In the context of bone regeneration, exogenous Vegf accelerates bone regeneration in vivo and induces differentiation in osteoblasts in vitro (20–22), whereas inhibition of Vegf impaired fracture healing (20). Besides adequate levels of Vegf ligands, it has been proposed that both receptors for Vegf-a— VegfR1 and VegfR2—are important for bone regeneration during distraction osteogenesis and that mesenchymal cell recruitment fails to progress in the absence of VegfR signaling (23). Interestingly, expression of VegfRs correlated with expression of Vegf (15, 24, 25), a finding that is consistent with our data because we observed delayed up-regulation of Vegf-a and VegfR1 in defects in Fgf-9+/− mice. The fact that VegfR2 gene expression was not elevated in defects of Fgf-9+/− mice could reflect the fact that VegfR2 expression primarily affects endothelial cell growth (26). In our study, we propose Fgf-9 as an additional player for angiogenesis during bone repair. Our hypothesis is supported by real-time PCR, immunohistochemistry, and μCT angiography data obtained from defects in Fgf-9+/− mice. Importantly, angiogenesis could be rescued in Fgf-9+/− mice by application of exogenous FGF-9. Taken together, our data suggest that FGF-9 not only has direct effects on osteoprogenitor proliferation and osteogenesis but also has profound effects on angiogenesis. It is tempting to hypothesis that FGF-9 is upstream of VEGF-A. Our hypothesis is supported by a previous embryonic developmental study, which demonstrated that Vegf, VegfR1, and VegfR2 expression was decreased in Fgf-9−/− mice (11). Conversely, exogenous FGF-9 did increase Vegf expression (11). Furthermore, expression of Fgf-9 has been identified in an angiogenic collagen gel culture model, coinciding with the formation of capillary-like tubes (27). The data obtained from our injury model indicated that Fgf-9 regulates vascularization in long bone repair by triggering Vegf expression. A similar observation has been reported previously for FGF-2, which induces VEGF in endothelial cells, thereby executing its angiogenic activity (28). In this context, it is worth mentioning that in our analysis Vegf-a expression in tibia of Fgf-9+/− mice equaled that in wild-type mice only at day 7, coinciding with similar levels of Fgf-2 expression observed in mutant and wild-type mice at days 5 and 7. It must be pointed out that Fgf-2 does not have a signal peptide and therefore is poorly released (29, 30). Upon tibia injury, Fgf-2 might be released, but at an early time point (day 3) the expression level of Fgf-2 in tibia of Fgf-9+/− mice was lower than in wild-type mice. This finding could explain the lack of a compensatory effect by FGF-2 at an early time point. Conversely, the up-regulation of Fgf-2 at later time points might elicit a compensatory effect by triggering Vegf-a expression in Fgf-9+/− tibia. Another interesting finding was that similar expression levels of Fgf-18 in Fgf-9+/− and wild-type mice upon injury did not compensate for the loss of Fgf-9, even though Fgf-18 and Fgf-9 share some functional redundancy during embryonic limb development (11): Fgf-9 and Fgf-18 both regulate chondrocyte differentiation and skeletal vascularization and mineralization during endochondral ossification. The reason that Fgf-18 does not compensate upon injury in Fgf-9+/− mice might be that, in our model, bone healing occurs through intramembranous ossification. Previous research for factors controlling VEGF in the context of bone healing focused on BMPs. For instance, studies using musclederived stem cells suggest that BMP4 acts synergistically with VEGF to enable bone healing (31). Interestingly, it also has been Behr et al.

DEVELOPMENTAL BIOLOGY

Fig. 5. Defects in Fgf-9+/− mice can be rescued with FGF-9 and VEGF-A. (A) PECAM-1 and VEGF-A immunohistochemistry of defects in Fgf-9+/− mice treated with PBS or 2 μg FGF-9 revealed increased vessel formation and VEGF-A staining in the FGF-9–treated defects at day 3. (B) Aniline blue staining of defects in Fgf-9+/− mice treated with PBS, 2 μg FGF-9, 2 μg FGF-2, 2 μg VEGF-A, or a combination of 2 μg FGF-9 and 2 μg VEGF-A. The injury site is marked in orange (Left column) and is segregated in the middle column. Accompanying PECAM-1 staining performed at day 7 revealed vessel formation (arrows) in the treated groups as compared with the PBS control (Right column). (C) Histomorphometry revealed rescue of the defects with all treatments, but FGF-9 and FGF-9 in combination with VEGF-A were the most efficient. *P < 0.05 for combined FGF-9 and VEGF-A treatment vs. VEGF-A treatment alone; **P < 0.005 and ***P < 0.0005 for treatment groups vs. PBS. (Scale bars, for PECAM and 200 μm for aniline blue.)

shown that VEGF interacts synergistically with BMP2 to induce bone formation (32), and this interaction could be another potential explanation for the impaired bone healing in Fgf-9+/− mice that exhibited low levels of Vegf-a and unaffected Bmp-2 levels in our work. Another interesting aspect of our study was the decreased TRAP staining in defects in Fgf-9+/− mice, as is consistent with data obtained from Fgf-9−/− bones at embryonic stages (11). This decreased staining could be caused either by the lack of Fgf-9 or by impaired Vegf expression. Stimulation of osteoclast recruitment also has been demonstrated previously for Fgf-2 (33). In addition, VEGF can act as a chemoattractant for osteoclasts (34). Moreover, it has been reported that osteoclast numbers were decreased in the perichondrium surrounding the radius of VEGF120/120 mice and that MMP-9 expression was reduced in the hypertrophic region and the perichondrium (35). Behr et al.

Studies that identified single factors essential for fracture healing included those using mice deficient for Bmp-2 (36), Osteopontin (37), and bone sialoprotein (38). Similar to observations in Bmp2c/c mice (36), we also found the presence of other osteogenic factors in the bone injury sites in Fgf-9+/− mice which could not compensate for the lack of Fgf-9. The fact that it was possible to rescue the healing phenotype of Fgf-9+/− tibial injuries with FGF-9 protein further emphasizes the importance of this ligand. It is remarkable, that only a half-dose of Fgf-9 gene had such a profound effect on bone repair. This effect probably results from impaired angiogenesis in bone repair. Given that Fgf-9 previously was associated with endochondral ossification during skeletal development, our results are even more remarkable, because we uncovered the healing phenotype in a model of intramembranous long bone healing model (i.e., a stabilized fracture). In this model, PNAS | June 29, 2010 | vol. 107 | no. 26 | 11857

bone remodeling can be observed as early as day 7. Although it is possible that defects in Fgf-9+/− mice eventually might heal, the rescue experiments in Fgf-9+/− mice, as well as the increased healing of FGF-9–treated defects in wild-type mice, emphasize the importance of Fgf-9 in bone regeneration. Intramembranous ossification in the healing of long bone fractures is achieved by rigid fixation, because the default healing mode in long bone healing is endochondral ossification, which recapitulates aspects of embryonic skeletal development. From a translational perspective, local administration of a stimulatory protein with angiogenic properties, such as FGF-9, may accelerate fracture healing. In conclusion, this study demonstrates a postnatal phenotype for Fgf-9 haploinsufficient mice, highlighting the essential role of FGF-9 during bone repair, potentially by controlling angiogenesis. In this context, it will be of interest in future studies to investigate the relationship between FGF-9 and perivascular stem cells. Our work furthermore supports the embryonic phenotype previously observed in developing limbs, specifically in the stylopod (11), supporting the concept that healing processes in adult organisms may recapitulate the embryonic skeletal development.

Ornitz (Washington University, St. Louis, MO) and were described previously (39). Genotyping was performed by PCR analysis on genomic DNA. For this study, a previously established murine model of tibial bone repair was used (40, 41). After mice were deeply anesthetized, the right legs were shaved, and the skin was disinfected. Then an incision was made over the proximal medial diaphysis, and the anterior tibial muscle was divided. The medial surface of the tibia was exposed, and the periosteum was preserved. A unicortical defect was created using a 1-mm drill bit under constant irrigation. In this model, the healing response is equivalent to that in a stabilized fracture and occurs through intramembranous ossification. The anterior tibial muscle was reapproximated, the skin was closed, and mice were allowed to recover. Mice were killed after 3, 5, and 7 d to cover the timing of inflammation, callus formation, and remodeling, respectively. For rescue or gain-of-function experiments, defects were treated with a 1-mm-diameter collagen sponge (Helistat; Integra LifeSciences Corporation) soaked in 2 μg of recombinant human FGF-9 (hFGF-9) and/or 2 μg human VEGF (hVEGF 165) (R&D Systems), 2 μg FGF-2 (Santa Cruz Biotechnologies), or PBS as a control. Collagen sponges were inserted in the defect and filled out the generated bone marrow space. Analysis of histology, immunohistochemistry, CT data, RT-PCR, and realtime PCR are described in detail in SI Materials and Methods.

Materials and Methods Skeletal Injuries. All experiments using animals were performed in accordance with Stanford University Animal Care and Use Committee Guidelines. Ageand sex-matched 10- to 12-wk-old Fgf-9+− and C57/Bl6 littermate control mice were used for all studies. Fgf-9+− mice were kindly provided by David

ACKNOWLEDGMENTS. We thank Dr. David Ornitz (Washington University, St. Louis, MO) for providing Fgf-9+/− mice and Michael Sorkin for expert assistance in QRT-PCR experiments. This work was supported by the Oak Foundation and Grants R21DE019274 from the National Institutes of Health (to M.T.L.) and DFG BE 4169-1 from the German Research Foundation (to B.B.).

1. Einhorn TA (1998) The cell and molecular biology of fracture healing. Clin Orthop Relat Res 46 (355, Suppl)S7–S21. 2. Gerstenfeld LC, Cullinane DM, Barnes GL, Graves DT, Einhorn TA (2003) Fracture healing as a post-natal developmental process: Molecular, spatial, and temporal aspects of its regulation. J Cell Biochem 88:873–884. 3. Schindeler A, McDonald MM, Bokko P, Little DG (2008) Bone remodeling during fracture repair: The cellular picture. Semin Cell Dev Biol 19:459–466. 4. Eswarakumar VP, et al. (2002) The IIIc alternative of Fgfr2 is a positive regulator of bone formation. Development 129:3783–3793. 5. Deng C, Wynshaw-Boris A, Zhou F, Kuo A, Leder P (1996) Fibroblast growth factor receptor 3 is a negative regulator of bone growth. Cell 84:911–921. 6. Montero A, et al. (2000) Disruption of the fibroblast growth factor-2 gene results in decreased bone mass and bone formation. J Clin Invest 105:1085–1093. 7. Coffin JD, et al. (1995) Abnormal bone growth and selective translational regulation in basic fibroblast growth factor (FGF-2) transgenic mice. Mol Biol Cell 6:1861–1873. 8. Ohbayashi N, et al. (2002) FGF18 is required for normal cell proliferation and differentiation during osteogenesis and chondrogenesis. Genes Dev 16:870–879. 9. Quarto N, Behr B, Li S, Longaker MT (2009) Differential FGF ligands and FGF receptors expression pattern in frontal and parietal calvarial bones. Cells Tissues Organs 190:158–169. 10. Ornitz DM, Itoh N (2001) Fibroblast growth factors. Genome Biol 2 (3):S3005. 11. Hung IH, Yu K, Lavine KJ, Ornitz DM (2007) FGF9 regulates early hypertrophic chondrocyte differentiation and skeletal vascularization in the developing stylopod. Dev Biol 307:300–313. 12. Liu Z, Xu J, Colvin JS, Ornitz DM (2002) Coordination of chondrogenesis and osteogenesis by fibroblast growth factor 18. Genes Dev 16:859–869. 13. Schmid GJ, Kobayashi C, Sandell LJ, Ornitz DM (2009) Fibroblast growth factor expression during skeletal fracture healing in mice. Dev Dyn 238:766–774. 14. Keramaris NC, Calori GM, Nikolaou VS, Schemitsch EH, Giannoudis PV (2008) Fracture vascularity and bone healing: A systematic review of the role of VEGF. Injury 39 (Suppl 2): S45–S57. 15. Gerber HP, et al. (1999) VEGF couples hypertrophic cartilage remodeling, ossification and angiogenesis during endochondral bone formation. Nat Med 5:623–628. 16. Street J, et al. (2000) Is human fracture hematoma inherently angiogenic? Clin Orthop Relat Res (378):224–237. 17. White AC, Lavine KJ, Ornitz DM (2007) FGF9 and SHH regulate mesenchymal Vegfa expression and development of the pulmonary capillary network. Development 134: 3743–3752. 18. Vu TH, et al. (1998) MMP-9/gelatinase B is a key regulator of growth plate angiogenesis and apoptosis of hypertrophic chondrocytes. Cell 93:411–422. 19. Rowe NM, et al. (1999) Angiogenesis during mandibular distraction osteogenesis. Ann Plast Surg 42:470–475. 20. Street J, et al. (2002) Vascular endothelial growth factor stimulates bone repair by promoting angiogenesis and bone turnover. Proc Natl Acad Sci USA 99:9656–9661. 21. Midy V, Plouët J (1994) Vasculotropin/vascular endothelial growth factor induces differentiation in cultured osteoblasts. Biochem Biophys Res Commun 199:380–386. 22. Geiger F, et al. (2005) Vascular endothelial growth factor gene-activated matrix (VEGF165-GAM) enhances osteogenesis and angiogenesis in large segmental bone defects. J Bone Miner Res 20:2028–2035. 23. Jacobsen KA, et al. (2008) Bone formation during distraction osteogenesis is dependent on both VEGFR1 and VEGFR2 signaling. J Bone Miner Res 23:596–609.

24. Barleon B, et al. (1997) Vascular endothelial growth factor up-regulates its receptor fms-like tyrosine kinase 1 (FLT-1) and a soluble variant of FLT-1 in human vascular endothelial cells. Cancer Res 57:5421–5425. 25. Shen BQ, et al. (1998) Homologous up-regulation of KDR/Flk-1 receptor expression by vascular endothelial growth factor in vitro. J Biol Chem 273:29979–29985. 26. Prewett M, et al. (1999) Antivascular endothelial growth factor receptor (fetal liver kinase 1) monoclonal antibody inhibits tumor angiogenesis and growth of several mouse and human tumors. Cancer Res 59:5209–5218. 27. Nagatoro T, Fujita K, Murata E, Akita M (2003) Angiogenesis and fibroblast growth factors (FGFs) in a three-dimensional collagen gel culture. Okajimas Folia Anat Jpn 80:7–14. 28. Seghezzi G, et al. (1998) Fibroblast growth factor-2 (FGF-2) induces vascular endothelial growth factor (VEGF) expression in the endothelial cells of forming capillaries: An autocrine mechanism contributing to angiogenesis. J Cell Biol 141:1659–1673. 29. Mignatti P, Morimoto T, Rifkin DB (1992) Basic fibroblast growth factor, a protein devoid of secretory signal sequence, is released by cells via a pathway independent of the endoplasmic reticulum-Golgi complex. J Cell Physiol 151:81–93. 30. Moscatelli D, Quarto N (1989) Transformation of NIH 3T3 cells with basic fibroblast growth factor or the hst/K-fgf oncogene causes downregulation of the fibroblast growth factor receptor: Reversal of morphological transformation and restoration of receptor number by suramin. J Cell Biol 109:2519–2527. 31. Peng H, et al. (2002) Synergistic enhancement of bone formation and healing by stem cell-expressed VEGF and bone morphogenetic protein-4. J Clin Invest 110:751–759. 32. Peng H, et al. (2005) VEGF improves, whereas sFlt1 inhibits, BMP2-induced bone formation and bone healing through modulation of angiogenesis. J Bone Miner Res 20: 2017–2027. 33. Collin-Osdoby P, et al. (2002) Basic fibroblast growth factor stimulates osteoclast recruitment, development, and bone pit resorption in association with angiogenesis in vivo on the chick chorioallantoic membrane and activates isolated avian osteoclast resorption in vitro. J Bone Miner Res 17:1859–1871. 34. Engsig MT, et al. (2000) Matrix metalloproteinase 9 and vascular endothelial growth factor are essential for osteoclast recruitment into developing long bones. J Cell Biol 151:879–889. 35. Zelzer E, et al. (2002) Skeletal defects in VEGF(120/120) mice reveal multiple roles for VEGF in skeletogenesis. Development 129:1893–1904. 36. Tsuji K, et al. (2006) BMP2 activity, although dispensable for bone formation, is required for the initiation of fracture healing. Nat Genet 38:1424–1429. 37. Duvall CL, Taylor WR, Weiss D, Wojtowicz AM, Guldberg RE (2007) Impaired angiogenesis, early callus formation, and late stage remodeling in fracture healing of osteopontin-deficient mice. J Bone Miner Res 22:286–297. 38. Monfoulet L, et al. (2010) Bone sialoprotein, but not osteopontin, deficiency impairs the mineralization of regenerating bone during cortical defect healing. Bone 46:447–452. 39. Colvin JS, White AC, Pratt SJ, Ornitz DM (2001) Lung hypoplasia and neonatal death in Fgf9-null mice identify this gene as an essential regulator of lung mesenchyme. Development 128:2095–2106. 40. Minear S, Leucht P, Miller S, Helms JA (2010) rBMP represses Wnt signaling and influences skeletal progenitor cell fate specification during bone repair. J Bone Miner Res 25:1196–1207. 41. Kim JB, et al. (2007) Bone regeneration is regulated by wnt signaling. J Bone Miner Res 22:1913–1923.

11858 | www.pnas.org/cgi/doi/10.1073/pnas.1003317107

Behr et al.