Prx1-expressing cells contributing to fracture repair require primary cilia for complete healing in mice

Abstract

Bone is a dynamic organ that is continuously modified during development, load-induced adaptation, and fracture repair. Understanding the cellular and molecular mechanisms for natural fracture healing can lead to therapeutics that enhance the quality of newly formed tissue, advance the rate of healing, or replace the need for invasive surgical procedures. Prx1-expressing cells in the periosteum are thought to supply the majority of osteoblasts and chondrocytes in the fracture callus, but the exact mechanisms for this behavior are unknown. The primary cilium is a sensory organelle that is known to mediate several signaling pathways involved in fracture healing and required for Prx1-expressing cells to contribute to juvenile bone development and adult load-induced bone formation. We therefore investigated the role of Prx1-expressing cell primary cilia in fracture repair by developing a mouse model that enabled us to simultaneously track Prx1 lineage cell fate and disrupt Prx1-expressing cell primary cilia in vivo. The cilium KO mice exhibited abnormally large calluses with significantly decreased bone formation and persistent cartilage nodules. Analysis of mRNA expression in the early soft callus revealed downregulation of osteogenesis, Hh signaling, and Wnt signaling, and upregulation of chondrogenesis and angiogenesis. The mutant mice also exhibited decreased Osx and Periostin but increased αSMA protein expression in the hard callus. We further used a Gli1^LacZ reporter and found that Hh signaling was significantly upregulated in the mutant callus at later stages of healing. Interestingly, altered protein expression and Hh signaling did not correlate with labeled Prx1-lineage cells, suggesting loss of cilia altered Hh signaling non-autonomously. Overall, cilium KO mice demonstrated severely delayed and incomplete fracture healing, and our findings suggest Prx1-expressing cell primary cilia are necessary to tune Hh signaling for proper fracture repair.

1. Introduction

Bone is a dynamic organ that is continuously modified during development, load-induced adaptation, and fracture repair. When a bone is broken, coagulation is activated, inflammation processes commence, and a soft callus forms to provide an initial structure. Eventually the cartilaginous tissue in the soft callus mineralizes into a hard callus and newly formed bone tissue is remodeled over time to establish complete repair. The fracture healing process therefore involves coordination of several signaling pathways and recruitment of skeletal progenitor cells in distinct stages to conduct this complex operation. Depending on the severity of the break, bone may undergo this process to heal naturally or require invasive surgical intervention. Understanding the cellular and molecular mechanisms for natural fracture healing can lead to important insights into bone regeneration and inspire therapeutics that enhance the quality of newly formed tissue, advance the rate of healing, or replace the need for invasive surgical procedures.

Recent studies suggest that skeletal progenitors derived from the periosteum predominantly contribute to fracture repair. The periosteum is a thin coating on the exterior surfaces of bone and is composed of an outer fibrous layer and inner cambium layer, which contains a wealth of osteochondroprogenitors. In response to fracture, the periosteum thickens and progenitors differentiate into osteoblasts and chondrocytes that populate the fracture callus [1]. Bone marrow mesenchymal stem cells (BMSCs) were initially suspected to be the source of chondrocytes and osteoblasts in the fracture callus; however, Zhang et al. demonstrated that nearly 70% of osteogenesis in an implanted graft was due to donor periosteal progenitors [2]. Moreover, Colnot et al. determined that fracture callus osteoblasts originate from both periosteum and bone marrow but the overwhelming majority of chondrocytes are periosteum-derived [3]. The periosteum has since become the primary focus for recapitulating natural bone repair [4–7], but whether a specific periosteal cell population uniquely drives callus and bone formation has yet to be determined. Several markers have been proposed for periosteum derived-osteochondroprogenitors including Sox9 [8], alpha smooth muscle actin (αSMA) [9], Dermo-1 [10], bone morphogenic protein 2 (BMP-2) [11,12], and Paired-related homeobox 1 (Prx1) [13–17]. Prx1 is a particularly attractive marker because Prx1 expression appears to be restricted to the periosteum in adult mice [13,17–19], and periosteum-derived Prx1-expressing cells are required for adult load-induced bone formation [19]. A recent study suggests that rare periosteal Prx1-expressing cells are the main contributor to cells populating the fracture callus in a Periostin-dependent manner [15]. More importantly, a second recent publication proposes that BMP-2/CXCL2 signaling regulates Prx1 expression during fracture and proposes an FDA-approved drug can tune this pathway for enhanced repair [17]. We therefore investigated Prx1-expressing cells in this fracture repair study.

The primary cilium is a sensory organelle present in skeletal cells and known to be important in bone development and maintenance. Several skeletal disorders can be tied directly to the primary cilium [20] and a large body of in vitro and in vivo evidence suggests mesenchymal stem cells, osteoblasts, osteocytes, and periosteal osteochondroprogenitors require primary cilia to participate in osteogenesis [21–28]. Additionally, the primary cilium serves as a signaling nexus [20,26,29] and is known to mediate several signaling pathways involved in fracture repair, including but not limited to Hedgehog (Hh) [30,31], TGF-β/BMP [32–34], and Wnt [35,36]. The structure and function of the primary cilium can be disrupted via deletion of intraflagellar transport protein 88 (Ift88). We previously used this approach to demonstrate that primary cilia are required for Prx1-expressing cells to contribute to normal juvenile skeletal development [18] and adult load-induced bone formation [28]. However, the role of Prx1-expressing cell primary cilia has yet to be determined in the context of fracture repair.

The objective of this study is to demonstrate that Prx1-expressing cells require functional primary cilia for complete fracture repair. We utilized our cilium knockout (KO) mouse model [18] to track Prx1-expressing cells and conditionally disrupt their primary cilia, and performed complete femoral fractures on these mice to examine healing at early, intermediate, and late stages [37]. We determined that Prx1-expressing cells still populate the fracture callus without primary cilia, but fracture healing is severely delayed in cilium KO mice due to persistent Hh signaling that encourages chondrogenesis over osteogenesis. This work highlights the utility of Prx1-expressing cells for bone regeneration and, for the first time, demonstrates that the primary cilium is necessary for these crucial osteochondroprogenitors to perform their role during fracture repair.

2. Materials and Methods

2.1. Animals and Tamoxifen injections

Prx1CreER-GFP and Ift88^fl/fl mice were gifted from Christopher Jacobs’ laboratory at Columbia University in New York, NY. Rosa26^tdTomato (Stock #007914) and Gli1^LacZ (Stock #008211) reporter mice were obtained from The Jackson Laboratory. Prx1CreER-GFP;Rosa26^tdTomato;Ift88^fl/+ males were bred with Ift88^fl/+ or Gli1^LacZ;Ift88^fl/+ females to generate Prx1CreER-GFP;Rosa26^tdTomato;Ift88^fl/+ or Prx1CreER-GFP;Rosa26^tdTomato;Gli1^LacZ;Ift88^fl/+ control and Prx1CreER-GFP;Rosa26^tdTomato;Ift88^fl/fl or Prx1CreER-GFP;Rosa26^tdTomato;Gli1^LacZ;Ift88^fl/fl experimental animals. 12 week-old mice received subcutaneous injections of 100 mg/kg tamoxifen (Sigma-Aldrich, T5648) + 10% ethanol in corn oil for 5 consecutive days to induce tdTomato expression and primary cilium ablation. All mice are on a C57BL/6 background and were housed and cared for in accordance with IACUC standards in an AAALAC accredited facility. Both non-bred male and female mice were used for experiments to address potential sex differences.

2.2. Fracture surgery

Three days after the final tamoxifen injection, mice were anesthetized via intraperitoneal injection of 40 mg/kg Ketamine (Henry Schein) + 5 mg/kg Xylazine (Akorn) in sterile PBS, and fractures were generated per a published protocol [12,38]. The hindlimb was shaved and wiped with betadine solution (Santa Cruz, sc-394616) followed by 70% ethanol 3 times to sterilize the incision site. An incision was made parallel to the femur and the fascia were cut and pulled apart using surgical scissors. Forceps were used to gently peel the muscle to expose the femur without disrupting the periosteum. The femur was cut in half using an 18 mm diameter diamond cutting wheel (Lukcase, B07BNKM1XH) attached to a rotary tool (Dremel 3000) at low speed to prevent heating and minimize bone debris. The area was washed with sterile PBS to remove any bone debris and the bone segments were “pinned” together to provide support. Specifically, a 23-gauge thin-walled needle was threaded through the marrow canal to poke holes in both epiphyseal ends. A 27-gauge needle was then threaded into the 23-gauge needle and through both epiphyseal ends. Needle holders were used to bend the ends of the needle to secure it in place and the sharp ends were removed with wire cutters. The muscle was folded back into place and the incision site was closed with stainless steel wound clamps (Braintree Scientific). A single dose of 25 μg/kg slow release Buprenorphine (ZooPharm) was administered as analgesic. Mice did not exhibit signs of pain and their ambulation was not restricted due to the pin. All surgeries were conducted using sterile technique and animals were monitored per IACUC regulations.

2.3. Visualizing tdTomato expression

Femurs were dissected and fixed in 4% paraformaldehyde (Sigma-Aldrich, P6148) at 4°C overnight then decalcified in 15% EDTA (VWR, BDH4616) for 4 weeks. Decalcified femurs were cryosectioned in 10 μm increments, thawed at room temperature, hydrated in PBS, and mounted with media containing a nuclear stain (Electron Microscopy Sciences, 17985–50). Images were acquired using an inverted fluorescent microscope (ECHO Revolve 4). Images were collected at 2X magnification and manually stitched together in Adobe Photoshop to visualize tdTomato expression throughout the entire femur and callus.

2.4. Alcian Blue/Orange G and H&E staining

Femurs were fixed and decalcified as explained above, then embedded in paraffin and sectioned in 5–8 μm increments. Prior to staining, paraffin sections were baked for 45 minutes at 65°C then deparaffinized and rehydrated. For Alcian Blue/Orange G stains, sections were first submerged in acid alcohol (4% HCL in 70% Ethanol) for 30 s then 10 μg/mL Alcian Blue (Sigma-Aldrich, A5268) in Mayer’s Hematoxylin (Sigma-Aldrich, GHS232) solution for 25 min. Slides were washed in distilled water for 20 min, quickly dipped in acid alcohol followed by 1 μg/mL sodium bicarbonate (Sigma-Aldrich, S5761), washed in distilled water for 10 min, and soaked in 95% ethanol for 1 min. Fresh Eosin/Orange G stock solution was prepared using 18.5 mL of 1% Phloxine B (Sigma-Aldrich, 18472-87-2), 8 mL of 2% Orange G (Sigma-Aldrich, 1936-15-8) and 250 mL of Eosin solution (ThermoFisher Scientific, 6766008). Slides were transferred directly from 95% ethanol to the Eosin/Orange G solution for 90 s then dehydrated in 95% ethanol, 100% ethanol, and Xylene at 1 min per change. To measure the presence of cartilage in the fracture callus, paraffin sections within 10 sections of the femur midpoint were stained and the area of cartilage was manually traced and measured using a pixel to mm conversion in ImageJ. Each sample measurement is the average of 4 consecutive sections. For H&E stains, rehydrated slides were incubated in Mayer’s Hematoxylin for 10 min, rinsed in running tap water for 15 min, incubated in Eosin for 30 s, washed in distilled water for 10 min, and dehydrated as explained above. Slides were mounted with Sub-X mounting medium (Electron Microscopy Sciences, 13519) and sealed. Images were collected at 2X magnification and stitched together using a Keyence BZ-X710 microscope and its associated software.

2.5. Immunohistochemistry

Paraffin and frozen sections were prepared as described previously. Sections were permeabilized in 0.5% Triton X-100 (Sigma-Aldrich, T8787) in PBS for 5 min, blocked in 10% goat serum (MP Biomedicals, 0219135680) for 1 h at ambient temperature, and incubated in primary antibody overnight at 4°C. The following primary antibodies were used at a 1:500 dilution in PBS: rabbit polyclonal Osx (abcam, ab22552), rabbit polyclonal αSMA (Abclonal, A1011), and rabbit polyclonal Periostin (Sino-Biological, 50450-RP02). The PECAM-1 mouse monoclonal primary antibody (BioLegend, 102502) was used at a 1:100 dilution. To detect primary cilia, we used a 1:10 dilution of an anti-acetylated α-tubulin mouse monoclonal antibody produced in house using a C3b9 hybridoma (Sigma-Aldrich, 00020913). Following the primary incubation, slides were washed in PBS for 30 min, incubated in a fluorescent secondary for 2 h at ambient temperature, washed for 30 min, and mounted in media containing a nuclear stain (Electron Microscopy Sciences, 17985–50). The secondary antibodies used are Alexa Fluor 568 goat anti-rabbit (Life Technologies, A11011) and Alexa Fluor 488 goat anti-mouse (BioLegend, 405319). Images were acquired using an inverted fluorescent microscope (ECHO Revolve 4). To quantify primary cilium incidence in the fracture callus, cryosections within 10 sections of the femur midpoint were stained and 4 frames were selected near the fracture site. Each frame contained 40–50 labeled cells and the percentage of ciliated tMt+ cells was manually counted using ImageJ. This process was repeated on 3 consecutive sections and all 12 frames were averaged for each sample.

2.6. X-gal staining

Femurs were dissected and fixed in 1% formaldehyde + 0.2% glutaraldehyde + 2 mM MgCl₂ + 5 mM EGTA + 0.02% NP-40 overnight at 4°C overnight. Specimens were washed with PBS for 15 min and incubated in 5 mM K₃Fe(CN)₆ + 5 mM K₄Fe(CN)₆ + 2 mM MgCl₂ + 0.01% NaDeoxycholate + 0.02% NP-40 + 1 mg/mL X-gal (MP Biomedicals, 150001) for 2 h at 37°C protected from light. Specimens were then washed again and transferred to 15% EDTA for 4 weeks. Decalcified femurs were embedded in paraffin and sectioned in 5–8 μm increments, deparaffinized, and mounted with Sub-X media. Images were collected at 2X magnification and stitched together using a Keyence BZ-X710 microscope and its associated software.

2.7. MicroCT

Upon sacrifice, fracture and sham femurs were carefully dissected to not disturb the callus and stored in 70% ethanol for up to a week. Each specimen was imaged by microCT (Scanco Medical μcT 35) at 6 μm isotropic resolution using scan settings of 55 kVp, 145 μA, and 300 ms integration time. Scanco Medical analysis software was used to distinguish the fracture callous from bone and determine total volume (TV), bone surface (BS), bone volume (BV), bone volume fraction (BV/TV), bone surface density (BS/TV), and specific bone surface (BS/BV). Specifically, the callus and femur were manually contoured in each section to include the entire fracture callus but exclude cortical bone and fragments for analysis. This was performed twice on each sample to ensure the manual contouring was consistent.

2.8. RTqPCR

Femurs were dissected and the surrounding skin and muscle were carefully removed to expose the soft tissue directly surrounding the fractured bone. The hematoma, soft callus, and/or periosteum were peeled off the fractured cortical bone under a dissecting microscope, placed into RNAzol RT solution (Sigma-Aldrich, R4533), and broken down with a handheld tissue homogenizer (Omni International) at low speed. RNA was isolated per the company’s standard protocol and converted to cDNA using the High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific, 4368814). qPCR was performed using Power SYBR Green Master Mix (ThermoFisher Scientific, 4368577) and StepOnePlus Real-Time PCR System (Applied Biosciences). mRNA values were normalized to GAPDH – a housekeeping gene constitutively expressed at high levels – to account for general variability in mRNA expression between samples. Experimental groups are expressed as a fold change in relation to controls normalized to a value of “1”.

2.9. Statistics

Animals were randomly assigned to groups depending on genotype, and researchers were blinded to all data analysis. No sex-dependent differences were identified according to a twoway ANOVA so males and females were grouped together for RTqPCR and microCT analysis. Differences between control and experimental animals were determined using a two-tailed student’s t test. Values are reported as mean ± SEM, with p < 0.05 considered statistically significant. The sample size was selected to achieve a power of at least 80%. Statistical analysis was conducted using GraphPad Prism (San Diego, CA).

3. Results

3.1. Mineralization is compromised in mice lacking Prx1-expressing cell primary cilia

In order to examine the role of Prx1-expressing cell primary cilia in complete fracture healing, animals received tamoxifen injections to genetically remove IFT88 (Supplementary Fig. 1) and then underwent surgery where the femur diaphysis was cut in half and stabilized with a pin. Femurs were dissected 7, 14, and 21 days post-fracture (dpf) in order to examine early, intermediate, and late stages of healing [37]. Fracture calluses in mice containing a Prx1-driven primary cilium deletion were noticeably larger in size than their control counterparts at 14 and 21 dpf (Supplementary Fig. 1A), which we confirmed histologically (Fig. 1). At 14 dpf, the two control fracture ends mostly fused together into a cohesive callus but mutant fracture ends were distinct and typically pulled apart if the pin was removed before fixation. We then examined fracture callus composition at various stages of healing by staining for cartilage and bone. One week after fracture, control mice exhibited deep purple staining throughout the callus indicating high levels of GAG and cartilage formation, as expected (Fig. 1). Mice with Prx1-driven primary cilium deletion had significantly less cartilage formation at this stage. In control mice the amount of cartilage in the callus decreased in favor of bone tissue at the intermediate stage of healing, and cartilage was almost completely absent at the late stage of healing. Interestingly, we observed more cartilage in the fracture callus at 14 dpf and persistent cartilage nodules near the fracture interface at 21 dpf. We then performed microCT analysis to quantify callus volume and mineralization at 14 and 21 dpf. Consistent with our histological observations, mutant calluses were larger in size but contained significantly less mineralized tissue (Fig. 2). Specifically, mutant calluses contained less bone in terms of volume, surface, and density. Furthermore, closer examination of the calluses in 3D revealed that mineralized tissue was more tightly confined to the fracture surface in control mice. Mutant calluses exhibited sporadic, unconnected mineralized regions surrounded by unmineralized tissue throughout the callus. Control mice demonstrated a clear non-mineralized line where the two fracture ends meet at 14 dpf, but this interface was completely mineralized by 21 dpf. This non-mineralized interface was less distinct in mutants compared to controls at 14 dpf and remained at 21 dpf, consistent with the persistent cartilage nodules we observed in Figure 1.

Figure 1. Mice lacking Prx1-expressing cell primary cilia exhibit delayed fracture repair.

Sections were stained with hematoxylin (dark blue), Alcian blue (light blue - dark purple), Orange G (pinkish orange), and eosin (pink) to identify nuclei, cartilage, bone, and all other tissues, respectively. Images were collected at 2X and stitched together using a light microscope and associated software. Black scale bar indicates 250 μm. n > 5 for each group.

Figure 2. Cilium KO mice demonstrate attenuated mineralization in the fracture callus.

Three dimensional reconstructions of μCT scans indicating sham control, mineralized control callus (green), and mineralized mutant callus (red). Bone properties were quantified for the callus region only. Black scale bar shown with sham control indicates 2 mm. n > 8 for each group.

3.2. Prx1-expressing cells and their progeny populate the callus independent of primary cilia

Prx1-expressing cells and their progeny have been shown to populate the fracture callus, but whether this behavior requires the primary cilium is unknown. We therefore generated a tamoxifen-inducible fluorescent reporter model to track the fate of these cells with and without primary cilia at various phases of healing. We first confirmed primary cilia disruption up to 21 dpf using immunohistochemistry to determine that fewer tMt⁺ cells contained primary cilia in newly formed bone in the mutant callus compared to controls (Supplementary Fig. 1C). Labeled cells were present in the calluses at all stages of healing but the number and spatial location was altered with primary cilium deletion (Fig. 3). In the early stages of healing, labeled cells were rampant throughout the calluses of control mice (Fig. 3a), but very few labeled cells were found in mutant calluses (Fig. 3b). At 14 dpf, labeled cells were still more prevalent in control calluses (Fig. 3g) but significantly more labeled cells were present in the mutant calluses (Fig. 3h) compared to 7 dpf. At this stage both cartilaginous and mineralized tissue are present. In both groups, more labeled cells were present in the mineralized regions (Fig. 3i–j) compared to cartilaginous regions (Fig. 3k–l), but control calluses always had more labeled cells. In fact, little to no labeled cells were present in some of the cartilaginous regions of the mutant calluses. At 21 dpf, the number of labeled cells declines in the control calluses (Fig. 3m) at highly mineralized (Fig. 3o) and less mineralized (Fig. 3q) regions. In contrast, 21 dpf mutant calluses (Fig. 3n) contain more labeled cells compared to 21 dpf control and 14 dpf mutant calluses (Fig. 3h). These cells also appear to be more evenly distributed throughout the callus than at earlier timepoints.

Figure 3. Prx1-expressing cells with disrupted cilia and their progeny still populate the fracture callus.

Prx1-expressing cells and their progeny (red) in the fracture callus, bone marrow, and newly formed cartilaginous and mineralized tissues at early (A,B), intermediate (G, H), and late (M, N) stages of healing. Lower panels are higher magnification representative images collected to closely examine regions with high (C, D, I, J, O, P) and low tMt expression (E, F, K, L, Q, R). Nuclei are depicted in blue. Upper panel images were collected with a fluorescent microscope at 2X and stitched together using software. White boxes outline the location of the lower panel images that were collected at 20X. White scale bars in the upper and lower panels represent 250 μm and 100 μm, respectively. n > 5 for all groups.

3.3. Chondrogenesis and angiogenesis are favored over osteogenesis in the mutant callus

In the days following fracture, several signaling pathways are upregulated to facilitate the initial soft callus and eventual hard callus formation. The primary cilium is known to mediate many of these pathways so we dissected the hematoma and soft callus at 3 and 7 dpf [37], respectively, and quantified mRNA expression of downstream targets. First, we confirmed that Ift88 expression is decreased in cilium KO tissues compared to controls, demonstrating our tamoxifen injections were effective (Supplementary Fig. 1B). We then quantified mRNA expression of reported markers for periosteum-derived osteochondroprogenitors involved in fracture repair (Periostin [15], αSMA [9], and BMP-2 [12] and mesenchymal stem cell markers for bone marrow-derived cells with osteochondrogenic potential (PDGFRα [39,40] and CD105 [40,41]). Interestingly, Periostin and PDGFRα were downregulated and CD105 and αSMA were upregulated in mutant tissues (Fig. 4A). Expression levels of chondrogenic (Sox9, Col2a, Col10a) and vascular (VEGF, PECAM-1) markers were increased (Fig. 4B) but expression levels of osteogenic markers (Osx, Runx2, Osc, BSP, SOST, ALP) were noticeably decreased (Fig. 4C) in cilium KO mice compared to controls. We also evaluated downstream targets of two signaling pathways known to be mediated by the primary cilium and involved in fracture repair: Hh and Wnt signaling. Indeed, expression of Hh (Gli1, Ptch1, Hip) and Wnt targets (Axin2, Lef1, Tcf1) were decreased in the mutant tissues compared to controls (Fig. 4D). We also examined changes in expression for the three Hh ligands and noticed a significant decrease in Ihh in cilium KO mice. Shh and Dhh demonstrated poor amplification in both groups, suggesting a lack of involvement for these ligands in the early stages of fracture repair. Overall, the trend for changes in expression was consistent between 3 and 7 dpf for every gene, but the disparity between control and mutant calluses did vary slightly as a function of time for some groups. For example, Gli1 was more severely downregulated at 3 dpf compared to 7 dpf (Fig. 4D), Col10a was upregulated to a greater degree at 7 dpf compared to 3 dpf (Fig. 4B), and PECAM-1 expression was less pronounced at 7 dpf compared to 3 dpf (Fig. 4B). Since the mutant calluses were larger but contained less bone, we further investigated the presence of vasculature at later stages of healing. Indeed, PECAM-1 expression was elevated in the mutant calluses compared to controls at both 14 and 21 dpf (Fig. 5). Cilium KO calluses also exhibited increased PECAM-1 expression detected via IHC at 7 dpf, which is consistent with our molecular analysis (Fig. 4B). Furthermore, we observed significant overlap between tMt+ cells and PECAM-1+ cells indicating that Prx1-expressing cell progeny contributed to the increased vasculature.

Figure 4. Signaling pathways associated with early fracture healing are altered in cilium KO mice.

(A) mRNA expression of genes thought to label cells involved in fracture repair. (B) mRNA expression of genes associated with chondrogenesis (Sox9, Col2a, Col10a) and angiogenesis (VEGF, PECAM-1). (C) mRNA expression of genes associated with osteogenic differentiation and bone formation. (D) mRNA expression of Hh ligand Ihh and signaling pathway targets (Gli1, Ptch1, Hip) and Wnt signaling pathway targets (Axin2, Lef1, Tcf1). Expression for each gene is normalized to the housekeeping gene, GAPDH, as an internal control. n > 4 for each group.

Figure 5. Cilium KO fracture calluses are more vascularized at all stages of healing.

Immunohistochemistry for PECAM-1 in the fracture callus at 7, 14, and 21 days post-fracture is depicted in green and Prx1-expressing cells and their progeny (tMt) are depicted in red. Yellow indicates overlap of PECAM-1 and tMt. Nuclei are depicted in blue. Representative images were collected at 20X using a fluorescent microscope. White scale bars indicate 50 μm. n > 3 for all groups.

3.4. Protein expression is altered in the mutant callus at later stages of healing

We then used immunohistochemistry to further examine a subset of targets from Figure 4 that are also known to be important in the later stages of healing. In control calluses, Periostin expression was observed in mineralized tissue at both 14 and 21 dpf (Fig. 6A, 7A) and to a lesser degree in cartilaginous tissue at 14 dpf (Fig. 6B). In contrast, cilium KO calluses exhibited very little Periostin expression in mineralized tissue (Fig. 6C) and no expression in cartilaginous regions at 14 dpf (Fig. 6D). At 21 dpf, Periostin was rarely observed in cartilage and essentially absent from bone tissue in cilium KO calluses compared to controls (Fig. 7B,C). αSMA was moderately expressed in mineralized regions of control calluses at both timepoints (Fig. 6E, 7D), but absent from cartilage at 14 dpf (Fig. 6F). We did observe consistent localization of αSMA⁺ cells at the bone/ cartilage interface, though (Fig. 6F). At both timepoints, αSMA expression was rampant throughout mineralized regions (Fig. 6G, 7E) in cilium KO calluses and prevalent in the cartilaginous regions (Fig. 6H, 7F). While the expression of αSMA remained relatively consistent between 14 and 21 dpf in control mice, the number of αSMA⁺ cells was noticeably increased from 14 to 21 dpf in cilium KO mice, especially in bone tissue. Osx expression was prevalent in mineralized tissue (Fig. 6I, 7G) at both timepoints but absent from cartilage (Fig. 6J) in control calluses, as expected. Osx was also absent from cartilage in cilium KO calluses (Fig. 6L, 7I) and significantly decreased in mineralized tissue compared to controls at both 14 and 21 dpf (Fig. 6K, 7H). The number of Osx⁺ cells in the mutant calluses did increase from 14 dpf to 21 dpf. Interestingly, we observed consistent localization of Osx⁺ cells at the bone/ cartilage interface in cilium KO mice at 21 dpf (Fig. 7I), similar to what we observed with αSMA⁺ cells in 14 dpf control calluses (Fig. 6F). To determine whether these changes in protein expression are due to Prx1-expressing cells and their progeny directly or because of non-autonomous cell behavior, we examined overlap of tMt⁺, Osx⁺ and αSMA⁺ cells (Fig. 8). In control mice, we observed some overlap with tMt⁺ and αSMA⁺ cells but these populations were mostly distinct at both stages of healing (Fig. 8A,E). Cilium KO mice exhibited the most overlap at 21 dpf (Fig. 8F) of any group, but tMt⁺ and αSMA⁺ cells were still generally distinct. Some cells co-expressed Osx and tMt at 14 dpf in control calluses (Fig. 8C) and this overlap was slightly more noticeable at 21 dpf (Fig. 8G). In cilium KO calluses, we observed zero correlation between Osx⁺ and tMt⁺ cells at 14 dpf (Fig. 8D) but nearly all Osx⁺ cells also expressed tMt at 21 dpf (Fig. 8H).

Figure 6. Protein expression is altered in the cilium KO fracture callus at the intermediate stage of healing.

Immunohistochemistry for Periostin (A–D), αSMA (E–H), and Osx (I–L) expression in the fracture callus at 14 days post-fracture, all depicted in red. White dashed lines separate bone tissue and cartilaginous regions. Nuclei are depicted in blue. Representative images were collected at 20X using a fluorescent microscope. White scale bars indicate 50 μm. n > 4 for all groups.

Figure 7. Protein expression is altered in the cilium KO fracture callus at the late stage of healing.

Immunohistochemistry for Periostin (A–C), αSMA (D–F), and Osx (G–I) expression in the fracture callus at 21 days post-fracture, all depicted in red. White dashed lines separate bone tissue and cartilaginous regions, which were both present only in the cilium KO group. Nuclei are depicted in blue. Representative images were collected at 20X using a fluorescent microscope. White scale bars indicate 50 μm. n > 4 for all groups.

Figure 8. Labeled cells in the fracture callus do not significantly overlap with αSMA- and Osx-expressing cells.

Immunohistochemistry for αSMA (A,B,E,F) and Osx (C,D,G,H) expression in the fracture callus is depicted in green and Prx1-expressing cells and their progeny (tMt) are depicted in red. Yellow cells indicate overlap of Osx/αSMA and tMt. White dashed lines separate bone tissue and cartilaginous regions in the 14 days post-fracture groups (A–D). Bone tissue is examined at 21 days post-fracture (E–H). Nuclei are depicted in blue. Representative images were collected at 20X using a fluorescent microscope. White scale bars indicate 50 μm. n > 4 for all groups.

3.5. Hh signaling is upregulated in the periosteum and callus of cilium KO mice at late stages of healing

The primary cilium is a key mediator of Hh signaling, a pathway known to be upregulated during fracture healing. Hh signaling is downregulated in mutant mice during early fracture callus formation (Fig. 4D), so we crossed our cilium KO model with a Gli1LacZ reporter to observe Hh-responsive cells in vivo at later stages of healing. At 14 dpf, we observed a slight difference in the pattern of Hh-responsive cells between mutant and control mice. Specifically, in both control and mutant groups Hh-responsive cells were located at the epiphyseal ends in articular and growth plate cartilage and within the callus at the fracture site in the mid-diaphysis (Fig. 9A). Closer inspection of the fracture callus showed slight expansion of Hh-responsive cells in the mutant periosteum compared to controls (Fig. 9B). Cilium KO mice also exhibited Hh-responsive cells at the outer edge of the callus but few to no cells were present in control calluses. At 21 dpf cilium KO mice contained significantly more Hh-responsive cells. The expression pattern for 21 dpf control mice is similar to the 14 dpf samples, but with more faint staining in the callus near the fracture site. However, Hh-responsive cells were rampant in the mutant fracture callus and surrounding tissue in addition to the epiphyseal ends and fracture site (Fig. 9A). At 21 dpf control mice contained very few Hh-responsive cells, which were tightly localized to the periosteum nearest the callus surface (Fig. 9B, arrows). In contrast, Hh-responsive cells were even further expanded in the mutant periosteum than they were at 14 dpf, but fewer labeled cells were present in the mutant fracture callus. We then investigated whether these Hh-responsive cells originate from Prx1-expressing cells by observing overlap of LacZ⁺ and tMt⁺ cells at 14 dpf. In both groups we observed no overlap between these cell populations. In control calluses, tMt⁺ cells were restricted to the newly formed bone in the fracture callus and Hh-responsive cells were localized to the periosteum (Fig. 9C). Mutant calluses deviated slightly from this pattern: few tMt⁺ cells were present in the periosteum and several LacZ⁺ cells were located in the callus, as seen before (Fig. 9B).

Figure 9. Hh signaling is upregulated in the fracture callus at the later stage of healing in cilium KO mice.

(A) X-gal staining of whole femurs to detect Hh-responsive cells (blue) in the fracture callus and bone. (B) H&E stains of fracture callus from samples in (A) to visualize Hh-responsive cells (blue). (C) Brightfield images of Hh-responsive cells (dark grey) overlayed with fluorescent images of Prx1-expressing cells and their progeny (red) at 14 dpf. Black arrows indicate LacZ⁺ cells and white arrows indicate tMt⁺ cells. Images were collected at 20X using light and fluorescent microscopes. Black scale bars indicate 50 μm. n = 3 for all groups.

4. Discussion

This work demonstrates for the first time that the primary cilium is important for fracture repair. Building evidence suggests that bone cells require primary cilia in order for natural bone formation and maintenance to occur. Primary cilia have been shown to be essential for osteoblasts and Prx1-expressing periosteal cells to participate in load-induced bone formation in adult mice [19,27]. Osteocytes, osteoblasts, and Prx1-expressing periosteal cells employ primary cilia to sense fluid shear and initiate osteogenic behavior [19,20,42]. In addition to mechanically-mediated bone maintenance, osteoblast and Prx1-expressing cell primary cilia are critical for normal skeletal development at both embryonic and juvenile stages [18,43–46]. In all the above instances, bone formation, osteocyte mechanosensing, and skeletal development were not completely abolished but were severely attenuated, indicating the primary cilium is a critical component but not the sole mechanism. It is therefore perhaps not surprising that Prx1-expressing cell primary cilia are required for complete fracture healing but some bone still forms in their absence. However, we did observe severely delayed healing, a lack of newly formed bone, and persistent cartilaginous nodules that are undoubtedly problematic for long-term bone health in cilium KO mice.

We further determined that Prx1-expressing cell primary cilia are most important for the transition from soft callus to newly formed bone. We detected changes in mRNA expression of angiogenic, chondrogenic, and osteogenic markers in the hematoma and soft callus at 3 and 7 dpf, respectively (Fig. 4). In summary, osteogenesis was downregulated while chondrogenesis and angiogenesis were upregulated in cilium KO mice. These changes are reflected in the lack of mineralized tissue in the mature callus (Fig. 2), increased expression of PECAM-1 (Fig. 5), absence of Osx expression in the later stages of healing (Fig. 6,7), and persistent cartilage nodules in place of new bone at 21 dpf (Fig. 1) we observed in cilium KO mice. This is consistent with our previous work showing that adult cilium KO mice form significantly less new bone in response to mechanical loads [28], juvenile cilium KO mice form shorter, thinner limbs [18], and Prx1-expressing cells exhibit severely attenuated fluid shear-induced osteogenesis when their primary cilia are disrupted [19]. We further speculate that the increased αSMA mRNA expression in the early stages of repair (Fig. 4) may explain the high levels of αSMA protein in cilium KO calluses and new bone at later stages of healing (Fig. 6,7). αSMA⁺ cells have been identified as immature periosteal progenitors that populate the fracture callus [9], so it is also possible these cells persist because they cannot properly differentiate when primary cilia are disrupted. Surprisingly, at 3 and 7 dpf we identified slight decreases in mRNA expression of Col10a and Osteocalcin, which are typically upregulated at later stages of fracture healing [37]. This perhaps suggests the primary cilium preemptively coordinates late-stage healing in the early stages of fracture repair, and disrupting primary cilia early ultimately resulted in the incomplete healing we observed in cilium KO mice at 21 dpf.

Collectively, our data suggest the Prx1-expressing cell primary cilium functions as a signaling nexus to simultaneously mediate various pathways during fracture healing. The primary cilium is known to directly mediate Wnt [36], TGFβ [33,34], and Hh [31] signaling. Canonical Wnt signaling is critical for fracture repair and thought to be upregulated to facilitate osteogenesis in general [47]. We observed that mRNA expression of Wnt signaling targets was decreased in the early stages of healing (Fig. 4). β-catenin influences the ratio of osteoblasts and chondrocytes in the fracture callus [48] and is important for osteoblast differentiation and bone matrix deposition [16] at the early and later stages of healing, respectively. Inhibiting Wnt resulted in callus enlargement in wildtype mice [49] so disruptions in Wnt signaling may explain the enlarged callus and lack of mineralization phenotypes we observed in cilium KO mice (Fig. 2). Administering a Hh agonist to elderly mice in one study resulted in larger callus size, but this was due to enhanced osteogenesis [50] so we speculate Wnt signaling is the predominant mechanism here. Non-canonical Wnt signaling is also thought to be important in fracture repair [51] and is critical for BMP-2-mediated osteoblast differentiation [52]. Disrupting the primary cilium may also directly impact BMP-2’s critical role in fracture repair [12] since BMP receptors localize to the primary cilium [34], which facilitates TGFβ signaling [33,34] and has been shown to influence BMP-2 levels in chondrocytes [53]. We observed a decrease in BMP-2 mRNA expression in the early healing stages (Fig. 4), which likely contributed to the delayed healing in cilium KO mice since BMP-2 is critical for initiating the healing process and progenitor differentiation [12,17]. In this study we also demonstrate for the first time that primary cilium function influences Periostin expression. Specifically, mRNA expression was decreased in cilium KO hematoma and soft callus (Fig. 4) and protein expression was almost completely abolished at the later stages of healing (Fig. 6,7). Mounting evidence suggests Periostin promotes osteogenesis [54] and one study demonstrated that overexpressing Periostin enhanced bone formation rate and overall bone mass in rats [55]. More importantly, Periostin was recently shown to be critical for Prx1-expressing cells to engage in fracture repair in adult mice [15]. More work is required to establish a direct link between the primary cilium and Periostin, but our data indicate a strong correlation between primary cilium function and Periostin expression in the fracture callus.

Interestingly, Hh signaling was downregulated during the initial stages of repair but upregulated during advanced healing in cilium KO mice. The primary cilium is central to transduction of Hh signaling in vertebrates because associated receptors and transcription factors are localized to the ciliary membrane and tip [56]. Hh signaling is upregulated in the periosteum during early stages of fracture repair to facilitate periosteal progenitor differentiation into osteoblasts and chondrocytes [1]. It is therefore not surprising that mRNA expression of Hh signaling targets was decreased in cilium KO mice at the early stages of healing (Fig. 4). Pharmacological inhibition of Hh signaling in young mice [57] and naturally blunted upregulation of Hh signaling in older rodents results in delayed fracture healing [58,59], so this may contribute to the delayed healing we observe in cilium KO mice. However, our Gli1LacZ reporter data show Hh signaling is upregulated in the periosteum surrounding the fracture callus at later stages of healing (Fig. 9A,B). Gli1 is known to be partially activated in the cilium’s absence [60,61] and primary cilia are thought to be important for turning Hh signaling both on and off [31]. We therefore speculate the persistent activation of Hh signaling in cilium KO mice is a combination of ciliary-independent Gli1 activity and the inability to suppress periosteal cell Hh signaling without functional primary cilia. This is supported by our data showing the periosteum remains thick around cilium KO fracture calluses and LacZ⁺ chondrocytes persist in said calluses (Fig. 9B) perhaps due to improper, continual differentiation. Moreover, we further determined through lineage tracing that the increased Hh signaling in the periosteum did not correlate with Prx1-expressing cells and their progeny in the callus and new bone tissue (Fig. 9C), suggesting cell non-autonomous behavior. We also observed weak overlap of both Osx⁺ and αSMA⁺ cells with tMt expression, further indicating primary cilium disruption negatively affects cell non-autonomous behavior. The primary cilium is a potent chemosensor thought to mediate intercellular signaling through vesicle release [62] but more work is needed to establish if and how the primary cilium mediates intercellular signaling.

Additionally, our lineage tracing data indicate that Prx1-expressing cells and their progeny are prevalent in the fracture callus, but other cells also participate in fracture repair. Evidence suggests the vast majority of cells in the fracture callus are thought to originate from the periosteum, with a potential contribution from BMSCs [3,6,15]. Other groups have demonstrated that all cells in the murine fracture callus were derived from Prx1-expressing cells [6,15]. In our control mice the majority of fracture callus cells were tMt⁺ but many cells were not labeled, indicating a source other than Prx1-expressing cells. This discrepancy either demonstrates that the contribution of non-Prx1-expressing cells is greater than originally anticipated or, more likely, highlights inconsistencies in our experimental approaches. We utilized a complete femoral fracture model with pinning for stability, whereas the aforementioned groups conducted tibial fracture without stabilization [6,15]. It is possible that insertion of the pin alone [17] or enabling the mice to bear weight on the fractured limb encouraged a greater contribution from BMSCs or other periosteal cells. Furthermore, we utilized a tamoxifen-inducible fluorescent reporter and the other groups used GFP or LacZ driven by Prx1Cre. This raises two important considerations. First, we may visualize fewer labeled cells due to lower efficiency with Cre recombination or insufficient tamoxifen administration. Second, Prx1 is expressed in a variety of cells throughout development [13,18,63] but becomes more restricted to the periosteum as skeletal maturity is approached [18,19]. Thus, using a Prx1CreER model perhaps results in fewer labeled cells compared to the Prx1Cre models. Furthermore, we utilize 12-week-old mice so Prx1-expression is likely more restricted to the periosteum compared to the 5 to 8-week-old mice used in the other studies [17,64]. Despite this inconsistency, it is important to note that many of our other results align with these publications and our approach still effectively demonstrates the importance of Prx1-expressing primary cilia in fracture repair.

The primary cilium is an attractive therapeutic target that could be manipulated to enhance fracture repair. Primary cilia function as both mechano- and chemosensors and fracture repair involves physical loading and paracrine signaling, so sensitizing cilia could potentially have an effective dual impact. The primary cilium is a fluid structure, altering its presence and length to tune its response to external stimuli [21,65]. Studies indicate that increasing the length and flexibility of primary cilia can enhance the cellular response to fluid shear [21,65]. Indeed, fluid shear-induced osteogenesis is enhanced when osteocyte primary cilia are lengthened by overexpressing ciliary proteins [66,67] or treating with lithium or fenoldopam in vitro [68]. Although lengthening bone cell primary cilia is a promising strategy to promote osteogenesis, proper in vivo experiments are needed to determine the effectiveness of sensitizing Prx1-expressing cell primary cilia to enhance the rate and quality of fracture healing.

5. Conclusion

In this work we demonstrated that Prx1-expressing cell primary cilia are important for complete fracture healing in mice. Animals lacking Prx1-expressing primary cilia exhibited abnormally large fracture calluses that contained more cartilage and vasculature but less bone compared to control mice. Primary cilia disruption did not influence the presence of Prx1-expressing cells and their progeny in the callus, but osteogenesis and signaling pathways associated with early-stage fracture healing (Wnt, Hh, and BMP) were impaired. We also determined that Hh signaling was upregulated in the periosteum at late stages of healing when the mutant callus failed to generate a fully mineralized callus. We therefore conclude that in the context of fracture, primary cilia are important for proper differentiation of Prx1-expressing cells populating the callus and coordinating the various signaling pathways required for complete healing.

Supplementary Material

1

Supplementary Figure 1. Confirmation of Ift88 knockout. (A) Dissected femurs 21 days post-fracture demonstrating a visible phenotype in the cilium KO group. (B) Ift88 mRNA expression in the fracture callus 3 and 7 days post fracture. Ift88 expression is normalized to GAPDH mRNA expression as an internal control. n > 4 for all groups. (C) Immunohistochemistry to detect and quantify primary cilia in newly-formed bone in the fracture callus 14 days post-fracture. Prx1-expressing cells and their progeny are depicted in red, nuclei are depicted in blue. White arrows indicate primary cilia. Images were collected at 20X using a fluorescent microscope. Lower panels are magnifications of white dashed boxes in the upper panels. White scale bar indicates 100 μm.

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Animals
PrxlCreER	Christopher Jacobs, Columbia University New York, NY	The Jackson Laboratory Stock No. 029211
Ift88tl/fl	Christopher Jacobs, Columbia University New York, NY	The Jackson Laboratory Stock No. 022409
Rosa26tdTomato	The Jackson Laboratory	Stock No. 007914
G1i1LacZ	The Jackson Laboratory	Stock No. 008211
Antibodies
Acetylated α-tubulin	Sigma-Aldrich	Cat. No. 00020913
αSMA	Abclonal	Cat. No. A1011
Osx	abcam	Cat. No. ab22552
PECAM-1	BioLegend	Cat. No. 102502
Periostin	Sino-Biological	Cat. No. 50450-RP02

Highlights.

• Primary cilia are sensory organelles important for bone maintenance and healing

• Fracture healing is severely delayed in mice lacking Prx1-expressing cell primary cilia

• Mutant fracture calluses contain more cartilage and vasculature but less bone

• Primary cilia function is critical for several signaling pathways associated with healing