Animals

All experiments involving animals were performed in compliance with the Guide for the Care and Use of Laboratory Animals and were approved by the Harvard Medical Area Institutional Animal Care and Use Committee. The mice were housed in cages with sterilized paper bedding in a 14:10-hour light/dark cycles with ad libitum access to sterilized water and chow. Mice containing loxP sites flanking exons 2 and 3 of Vlk and a selection NEO cassette (Vlk ^flox;neo) were kindly donated by Dr Aimee Zuniga.⁽⁹⁾ Vlk^{flox neo} mice were crossed with Flp1 mice (129S4/SvJaeSor-Gt(ROSA)26Sor^tm1(FLP1)Dym/J, JAX Strain no. 003946) to produce offspring without the NEO cassette (FLP1; Vlk^flox;neo mice), which were then mated with C57BL/6J mice (Jax Strain no. 000664) to remove the FLP1 transgene. The resulting heterozygous progeny (Vlk^flox/+) were subsequently mated together to obtain homozygous Vlk^flox/flox mice, herein referred to as Vlk^fl/fl. Vlk^fl/fl mice were crossed to Prx1-Cre (Cg-Tg(Prrx1-cre)1Cjt/J, Strain no. 005584) and Ubiquitin (Ubq)-Cre-ERt (Cg-Ndor1^{Tg(UBC-cre/ERT2)1Ejb}/1J, Strain no. 007001) deleter strains. For embryonic data collection, mice were time-mated to control fertilization. To induce recombination using the Ubq-Cre-ERt strain, tamoxifen was resuspended in corn oil and injected intraperitoneally at 80 μg/g of body weight. Figures contain data combining both male and female mice before 10 days of age. After 10 days of age, male data are presented in figures unless otherwise specified.

Quantitative PCR (qPCR)

RNA samples were isolated using TriZOL reagent (Thermo Fisher Scientific, Waltham, MA, USA), and 500 ng of RNA was used for conversion into cDNA with the High-Capacity cDNA Transcription Kit (Thermo Fisher Scientific). qPCR was performed using the FastStart Universal SYBR Green (Sigma-Aldrich, St. Louis, MO, USA) with mouse primers to detect the following mRNAs Vlk: F: CAAGCTGCTCAAAGAGATGGT, R: TGGTAGCAATAGCCATAGAGCTG; Rankl: F: AGCCATTTGCACACCTCAC, R: CGTGGTACCAAGAGGACAGAGT; Opg: F: GTTTCCCGAGGACCACAAT, R: CCATTCAATGATGTCCAGGAG; Sox9: F: CAGCAAGACTCTGGGCAAG, R: TCCACGAAGGGTCTCTTCTC; Mmp9: F: CTGGACAGCCAGACACTAAAG, R: CTCGCGGCAAGTCTTCAGAG; Mmp13: F: GCCAGAACTTCCCAACCAT, R: TCAGAGCCCAGAATTTTCTCC; Col2a1: F: CGGTCCTACGGTGTCAGG, R: TTATACCTCTGCCCATTCTGC; Col10a1: F: GAGGAAGCCAGGAAAGCTG, R: CCATGAACCAGGGTCAAGAA; Opn: F: CCCGGTGAAAGTGACTGATT, R: TTCTTCAGAGGACACAGCATTC. β-actin was used as housekeeping gene with the following sequences: F: GTGTACGACCAGAGGCATAC, R: AAGGCCAACCGTGAAAAGAT.

Femur fracture

Fractures were performed on 12-week-old male mice using sterile techniques and tools as previously described.⁽¹³⁾ Animals remained under anesthesia using isoflurane for the whole surgical procedure and an analgesic (slow-release norbuprenorphine) was injected subcutaneously for post-surgery pain management. Legs were shaved and disinfected with 70% EtOH and wiped with iodine to prep for surgery. A small 1 cm incision was made on the thigh, and the underlying fascia was cut using fine small scissors. The thigh muscle was pushed to the side using forceps to expose the femur. Small, curved forceps were inserted below the femur to maintain muscle separation and the femur exposed. A battery-powered DREMEL 7700 saw equipped with a circular THIN-FLEX blade was used on low power to make a straight transverse cut through the femur. Once the femur was separated into a proximal section (attached to the hip) and a distal section (attached to the tibia), a guide needle was used to create a path for the stabilizing pin. The guide needle (23G, 0.6 × 25 mm) was inserted into the exposed marrow cavity of the distal section, pushed through the knee joint, and removed from the distal section. The same needle was then inserted in the exposed marrow cavity of the proximal section and pushed through the hip until the tip of the needle emerged from the skin. The guide needle was left in place inside the proximal femur section. The stabilizing pin (needle, 27G, 0.4 × 30 mm) was placed into the end of the guide needle and the whole assembly was then pushed back so that the stabilizing pin went through the hip, into the femur, and exited through the marrow cavity with the help of the guide needle. The guide needle was discarded and the stabilizing pin was inserted into the marrow cavity of the distal section and pushed through the skin of the knee. We ensured that both sections of the femur were aligned and came back together using forceps. Both ends of the stabilizing pin were twisted using clamps to secure the femur sections together and blunt the ends of the pin. The wound opening was closed using wound clips. For contralateral sham operations, the same steps were followed for opening and closing the wound, but no cut was made to the femur. The animals were placed in a clean cage on top of a warm pad until complete recovery from anesthesia. We monitored the animals daily for 5 days post-surgery for any signs of infection or pain. Wound clips were removed 10 days post-surgery.

Histology

Histology was performed on formalin-fixed tissues embedded in paraffin. Sections (7 μm) were deparaffinized with xylene and rehydrated through a gradient of ethanol. All staining solutions were purchased from Sigma-Aldrich. For Toluidine blue staining, sections were dipped in a 0.1% solution Toluidine blue in 0.1% NaCl and 70% ethanol. For Safranin O staining, sections were stained with hematoxylin for 7 minutes, then a 0.08% Fast Green solution for 3 minutes. Sections were dipped in 1% acetic acid, incubated 5 minutes in 0.1% Safranin O, and dipped in 0.5% acetic acid. For Von Kossa, sections were incubated in 3% silver nitrate for 1 hour under strong light, then washed with 5% sodium thiosulfate for 2 minutes. Sections were then counterstained with Fast Red Counterstain for 5 minutes. Tartrate-resistant acid phosphatase (TRAP) staining was performed with Acid Phosphatase Leukocytes Staining kit (Sigma-Aldrich) according to the manufacturer's guidelines, and sections were counterstained with 0.2% Fast Green solution. For Alcian blue/Orange G staining, sections were incubated in 1% Alcian blue in hematoxylin for 40 minutes, then quickly differentiated in 1% HCl in 70% ethanol and then in 0.5% ammonium hydroxide. Sections were then stained in 0.07% Phloxine B, 0.06% Orange G in Eosin Y. Skeletal preparations were performed following previously published protocols.⁽¹⁴⁾ In situ hybridization with digoxigenin-labeled Collagen Type II probe was carried out on P0 hindlimb sections as previously described.⁽¹⁵⁾ For Gli1 immunohistochemistry, deparaffinized and rehydrated sections were microwaved in Target Retrieval Solution until boiling was reached, cooled for 10 minutes, and blocked with Dual Endogenous Enzyme Block for 30 minutes using the Dual Endogenous Enzyme Block Dako kit (Agilent, Santa Clara, CA, USA). Slides were then incubated in blocking solution composed of 5% goat serum (Vector Laboratories, Burlingame, CA, USA) in PBS + 0.1% Tween 20 (Sigma-Aldrich), incubated in a primary rabbit polyclonal antibody for Gli1 (NBP1-78259SS, Novus Biologicals, Littleton, CO, USA) at a dilution of 1:1000 in blocking solution overnight at 4 °C, and incubated in a goat anti-rabbit biotin secondary (B2770, Thermo Fisher Scientific) at a dilution of 1:2000 in blocking solution. Amplification was performed per company protocol using the TSA Biotin System Kit (Akoya Biosciences, Marlborough, MA, USA). Slides were then incubated in a Streptavidin Alexa Fluor 555 conjugate (S21381, Thermo Fisher Scientific) at a dilution of 1:250 in blocking solution and mounted with media containing DAPI (Electron Microscopy Sciences, Hatfield, PA, USA) to visualize nuclei. Incubations were performed for 30 minutes at ambient temperature unless noted otherwise. Quantification: growth plate (GP) length: length of the entire GP, proliferative (PZ), and hypertrophic (HZ) zones of the femoral distal GP were measured at P0 using ImageJ (version 1.53k). Specifically, the midpoint of the growth plate was identified and the length of each region was measured along the midpoint axis, with the top of the resting zone and bottom of the HZ serving as landmarks for total growth plate length. Quantifications from three consecutive sections were averaged for each sample (n = 3/group). Gli1: Gli1⁺ and Gli1⁻ cells were counted on images covering the anterior, middle, and posterior part of the HZ of the femoral distal GP at P0 and P10. Mean counts from three consecutive sections per sample (n = 2–4/group) were used to represent the proportion of cells covering the HZ. Adipocyte density: Osteomeasure (version 7.0, Osteometrics Inc, Decatur, GA, USA) was used to count adipocytes in the proximal metaphyseal and mid-diaphyseal regions of tibias from P10 mice. Adipocytes were counted on TRAP-stained sections and the data are reported as number of adipocytes per total area of the region of interest. Cartilage callus quantification: ImageJ (version 1.53k) was used to delineate and measure area occupied by the cartilage on fracture sections stained with Alcian blue/Orange G. The entire area of the callus was also measured and data are reported as the proportion of cartilage in the total callus area. For static histomorphometry, analysis was performed using the Osteomeasure software on sections of tibia 200 μm from the growth plate stained with Toluidine blue (osteoblast and adipocyte counts) or TRAP (osteoclast count). Standard nomenclature was used for the report.

X-ray and micro–computed tomography (μCT)

X-rays were performed on femurs and tibias using In Vivo MS FX PRO (Bruker Biospin, Billerica, MA, USA) using a 20-second exposure time, 35 kVP, and X-ray filter of 0.4 mm. μCT was performed on formalin-fixed femurs stored in 70% ethanol. A high-resolution desktop micro-tomographic imaging system (μCT35, Scanco Medical AG, Brüttisellen, Switzerland) was used to scan samples. Scans were acquired using a 7 μm isotropic voxel size, 55 kVp peak X-ray tube intensity, 145 μA, 8 W X-ray tube current, and were subjected to Gaussian filtration and segmentation. Image acquisition and analysis protocols adhered to guidelines for the use of μCT for the assessment of bone microstructure in rodents.⁽¹⁶⁾ Trabecular bone was evaluated in the distal femoral metaphysis in a 1400-μm-long region beginning 210 μm above the peak of the distal growth plate and extending proximally. Trabecular bone was segmented from soft tissue using thresholds of 365 mgHA/cm³. The Scanco Evaluation Program Trabecular Morphology script was used to measure trabecular bone volume fraction (BV/TV, %), trabecular thickness (Tb. Th, mm), trabecular number (Tb.N, mm⁻¹), trabecular separation (Tb.Sp, mm), connectivity density (Conn.Density, mm⁻³), and small model index (SMI). Cortical bone architecture was evaluated in a 504-μm-long region at the femoral mid-diaphysis. Cortical bone was segmented using a threshold of 700 mgHA/cm³ and then evaluated using the Scanco Mid-shaft Evaluation script to measure cortical bone area fraction (Ct.Ar/Tt.Ar, %) and thickness (Ct.Th, mm). Mineralized calluses were delineated by hand around the cortical bone and the parameters set for the Trabecular Morphology Script were used to obtain a bone volume. Fracture gap was measured as the average distance between the cortical broken sections measured on the μCT slices.

Statistics

The data are expressed as the mean ± standard error of the mean (SEM). Results were analyzed for statistically significant differences using Prism 8 statistical software (GraphPad Software, Inc., La Jolla, CA, USA). We performed Student’s t test or 2-way ANOVA with multiple comparisons with statistical significance set at p < 0.05. For mice cohorts, we powered (β = 0.95) the study for n = 5 mice per experimental group for μCT data collection to complete all assays with a standard deviation of 1% in the μCT measurement tool for an α (p < 0.01). Numbers of mice (n) used in specific experiments are indicated in each figure.

Deletion of Vlk in Prx1⁺ progenitors recapitulates the skeletal phenotype of Vlk^−/− mice

To understand whether the skeletal alterations previously reported in the Vlk^−/− mice were caused by a loss of Vlk in skeletal progenitors, we removed Vlk using Prx1-Cre (Vlk-Prx1 cKO) to target the limb bud mesenchyme. To verify that Vlk was successfully removed in bones, we performed qPCR on cortical bones of postnatal day 14 (P14) Vlk-Prx1 cKO mice. Vlk expression was markedly decreased in the femur shafts of Vlk-Prx1 cKO mice compared with that of controls (Fig. 1A). At E13.5, the cartilage template of the humerus was formed in Vlk-Prx1 cKO mice, but initiation of the primary ossification center had not yet begun, indicating delayed progression of endochondral ossification (Fig. 1B). this resulted in limbs from the Vlk-Prx1 cKO mice being shorter compared with those of Vlk^fl/fl at E15.5 (Fig. 1C). Von Kossa staining of the humeri revealed that the primary ossification center was still absent and mineralization had failed to occur in Vlk-Prx1 cKO bones at E15.5 (Fig. 1C). However, a primary ossification center was established in Vlk-Prx1 cKO limbs by the time the mice were born (Fig. 1D). At P0, both femurs and tibias of Vlk-Prx1 cKO mice remained shorter than those of controls and contained Collagen Type II cartilage remnants in the marrow cavity (Fig. 1D). The tibias of the Vlk-Prx1 cKO mice exhibited a prominent curvature at their mid-section that was not present in the femora (Fig. 1D). At P0, the overall width and length of the growth plate were significantly decreased in the femurs of Vlk-Prx1 cKO mice (Fig. 1E). Although the proliferative zone was markedly shorter in the femurs of Vlk-Prx1 cKO mice, the hypertrophic zone was significantly longer (Fig. 1E). These data suggest that the deletion of Vlk in Prx1⁺ skeletal progenitors recapitulates the embryonic limb phenotype of Vlk^−/− mice and that Vlk produced by skeletal progenitor cells is active during endochondral ossification, where it promotes the progression of chondrogenesis.

Vlk-Prx1 cKO mice survive after birth, unlike Vlk^−/− mice

Vlk^−/− mice die shortly after birth due to respiratory failure.⁽⁸⁾ However, Vlk-Prx1 cKO mice survived, grew, and bred. The postnatal skeletal phenotype in Vlk-Prx1 cKO mice was consistent between males and females, and figures show data from male mice unless otherwise specified. At P7, the tibia curvature observed in Vlk-Prx1 cKO appeared less pronounced than at P0 (Fig. 2A). In addition, the cartilage remnants observed at P0 in Vlk-Prx1 cKO mice were no longer present at P7 (Fig. 2A).

To test whether resorption played a role in resolving the tibia curvature, we performed TRAP staining on P10 tibias to label osteoclasts. We found that TRAP staining was restricted to the endocortical surface of the tibia diaphysis in control mice but was detected on the exterior of the cortical bone in the periosteal region in Vlk-Prx1 cKO mice (Fig. 2B). Interestingly, this was observed only on the anterior side of the tibia in the direction of curvature in the Vlk-Prx1 cKO tibia (Fig. 2B). Additionally, adipocytes accumulated at the site of the curvature in Vlk-Prx1 cKO (Fig. 2B). The marrow fat was only significantly increased in the mid-diaphysis region of the tibia of Vlk-Prx1 cKO mice but was not found to be different in the metaphyseal region (Fig. 2C). When we performed qPCR on the tibial marrow, we found that RANKL and OPG were significantly increased and decreased, respectively, in bones from Vlk-Prx1 cKO mice (Fig. 2D).

As Vlk has previously been reported to interact with Hedgehog signaling,⁽⁹⁾ we performed immunofluorescence for Gli1, a Hedgehog signaling target, on P0 and P10 femurs. At P0, Gli1 expression was mostly present in the distal femoral growth plate in Vlk^fl/fl mice and in a few cells in the marrow and subchondral area. In Vlk-Prx1 cKO mice, Gli1 staining was stronger and located throughout the entire subchondral area, the growth plate, and the marrow (Supplemental Fig. SS1A). At P10, Gli1 staining became restricted to the hypertrophic zone of the growth plate in both Vlk^fl/fl and Vlk-Prx1 cKO mice but persisted in the surrounding area of the secondary ossification center in Vlk-Prx1 cKO mice (Supplemental Fig. SS1A). At both P0 and P10, there were more Gli1⁺ cells in the hypertrophic zone of the growth plate of Vlk-Prx1 cKO mice compared with that of Vlk^fl/fl mice (Supplemental Fig. SS1B). Taken together, these results indicate that deletion of Vlk in skeletal progenitors leads to changes in the growth plate and the bone marrow microenvironment, resulting in mishappened bones after birth.

At skeletal maturity, bone mass is decreased in Vlk-Prx1 cKO mice

We performed X-rays of long bones in adult mice to determine if deletion of Vlk driven by Prx1-Cre resulted in changes in the shape or size of skeletal elements. At 10 weeks of age, the curvature of the Vlk-Prx1 cKO tibia that was present at birth and at P7 was completely resolved (Fig. 3A). Although femur length was significantly shorter when compared with Vlk^fl/fl male mice (femur length Vlk^fl/fl = 16.55 mm; Vlk-Prx1 cKO = 15.21 mm, 8% difference), total body weight and length were not changed in Vlk-Prx1 cKO mice (Fig. 3B). At 16 weeks of age, we performed μCT on the femurs to assess changes in bone mass and architecture. We found that cortical fraction and thickness were both markedly decreased in Vlk-Prx1 cKO male mice (Fig. 3C). Trabecular bone volume and trabecular number were also significantly decreased in the femurs of Vlk-Prx1 cKO male mice, resulting in an increase in trabecular separation and reduced connectivity density in these bones (Fig. 3D). However, there was no difference in trabecular thickness and small model index between Vlk^fl/fl and Vlk-Prx1 cKO male mice (Fig. 3D). By X-ray, the marrow cavity of Vlk-Prx1 cKO male mice appeared wider (Fig. 3A), and the marrow volume measured by μCT was significantly increased (Fig. 3D). However, bone volume of Vlk-Prx1 cKO male mice was still significantly lower than that of Vlk^fl/fl before dividing by total (marrow) volume to obtain trabecular BV/TV (Fig. 3D). We did not detect differences in the number of osteoblasts, osteoclasts, or adipocytes in the proximal tibias of the Vlk^fl/fl and Vlk-Prx1 cKO mice (Fig. 3E). Female Vlk-Prx1 cKO mice also had shorter femurs (femur length Vlk^fl/fl = 15.72 mm; Vlk-Prx1 cKO = 14.68 mm, 7% difference) but no change in body length and weight at 10 weeks of age (Supplemental Fig. SS2A). However, unlike males, cortical measurements were not different in female Vlk-Prx1 cKO mice compared with Vlk^fl/fl (Supplemental Fig. S2B). The trabecular phenotype observed in male Vlk-Prx1 cKO mice was also present in female Vlk-Prx1 cKO mice (Supplemental Fig. SS2C). Again, we found no difference in the number of osteoblasts, osteoclasts, or adipocytes in the proximal tibia of Vlk^fl/fl and Vlk-Prx1 cKO female mice (Supplemental Fig. S2D). These data indicate that embryonic loss of Vlk in skeletal progenitors results in decreased bone mass acquisition in adult mice.

Loss of Vlk delays the initial fracture repair response

As the chondrogenic phase of skeletogenesis was altered with loss of Vlk, and Prx1⁺ cells are known to form the early callus after fracture,^{(17, 18)} we analyzed callus formation in control and Vlk-Prx1 cKO mice after femur fracture in 12-week-old male animals. In control mice, periosteal expansion and Vlk mRNA expression were detected in the forming callus 5 days post fracture (dpf; Fig. 4A). This response was not detected in sham-operated contralateral controls (Supplemental Fig. SS3). This early healing response appeared blunted in Vlk-Prx1 cKO mice (Fig. 4B). However, at 10 dpf, large cartilage calluses were observed in controls and Vlk-Prx1 cKO mice and both went on to mineralize and initiate bridging at 28 dpf (Fig. 4B). The cartilage area present inside the callus was significantly decreased in Vlk-Prx1 cKO mice at 5 dpf (Fig. 4C). From 5 to 10 dpf, the cartilage area was reduced in control mice but remained the same in Vlk-Prx1 cKO mice (Fig. 4C). Volume of mineralized callus and the fracture gap were not different between Vlk^fl/fl and Vlk-Prx1 cKO mice at 28 dpf (Fig. 4D). We investigated the expression of cartilage markers and genes known to be involved in fracture repair at 5 dpf. There was no difference in gene expression of Mmp9, Mmp13, Col2a1, Col10a1, Sox9, and Opn in calluses obtained from Vlk^fl/fl and Vlk-Prx1 cKO mice at 5 dpf (Fig. 4E). These data suggest that Vlk produced by Prx1⁺ cells plays a role in the initial phase of fracture callus formation.

Since the Prx1-Cre deleter strategy we chose removed Vlk during embryogenesis and resulted in an observable skeletal phenotype, it was unclear whether the early delay in fracture repair we found in the absence of Vlk expression was due to embryonic loss of Vlk or by the subsequent absence of Vlk immediately after injury. To address this question, we used a global inducible knockout strategy to remove Vlk in all tissues before inducing fracture (Vlk-Ubq iKO). We determined that five injections of tamoxifen were sufficient to delete Vlk in tissues of 12-week-old mice (Fig. 5A). To test if removing Vlk could affect bone mass in adult mice, we induced the deletion of Vlk in 12-week-old female and male mice and collected femurs for μCT analysis a month after the last tamoxifen injection. We found that there was no difference in femur length or cortical and trabecular parameters in male or female Vlk-Ubq iKO mice when compared with Vlk^fl/fl (Supplemental Fig. S4A–C). We then injected 12-week-old Vlk^fl/fl and Vlk-Ubq iKO mice daily with tamoxifen for 5 consecutive days and allowed them to recover for 3 days before performing femur fracture. Femurs were then harvested 5 and 28 dpf (Fig. 5B), and qPCR was performed to confirm Vlk deletion in bones and calluses (Fig. 5B). At 5 dpf, a cartilage callus was visible in fractured femurs of control mice, but the cartilage area was considerably decreased in Vlk-Ubq iKO mice (Fig. 5C, D). No fracture or periosteal expansion were detected in sham-operated controls (Supplemental Fig. S3). At 28 dpf, mineralized calluses were present in both Vlk^fl/fl and Vlk-Ubq iKO mice (Fig. 5C) and were of equivalent volume by μCT (Fig. 5E). These results support our findings from Vlk-Prx1 cKO mice and further emphasize that Vlk expression is important for the proper onset of fracture repair.

Osteopontin phosphorylation is altered in the absence of Vlk

Bordoli and colleagues previously identified OPN as a potential target for Vlk phosphorylation.⁽⁶⁾ OPN production and phosphorylation are known to occur in the growth plate during endochondral ossification.^(19-21) Therefore, we selected OPN as a potential candidate for Vlk-dependent phosphorylation in skeletal tissue. As a first step, we validated the tyrosine phosphorylation of OPN by Vlk in vitro using K4 cells expressing wild-type Vlk, kinase-dead Vlk, or GFP. Conditioned media was collected and subjected to immunoprecipitation to pull down OPN. Immunoprecipitates were treated with or without λ phosphatase (λPP) as a technical control for Vlk-dependent phosphorylation. Immunoprecipitates were then subjected to Western blot analysis with anti-phosphotyrosine (pTyr) or anti-OPN antibodies. In cells transfected with GFP or kinase-dead Vlk, we did not detect phosphorylation of OPN (Fig. 6A). However, in samples from cells transfected with wild-type Vlk, we detected the presence of phosphorylated OPN that was absent upon λPP treatment (Fig. 6A). To examine if the tyrosine phosphorylation of OPN occurred in vivo, we used protein lysates from tibias of 2-month-old Vlk-Ubq iKO and Vlk^fl/fl mice. Protein lysates were immunoprecipitated with anti-OPN antibody followed by Western blot analysis with anti-pTyr and anti-OPN antibodies. In samples extracted from Vlk-Ubq iKO mice, we found decreased levels of phosphorylated OPN when compared with samples from Vlk^fl/fl mice (Fig 6B, C). Together, these results suggest that Vlk can phosphorylate skeletally produced OPN.