Mkx-deficient Mice Exhibit Hedgehog Signaling Dependent Ectopic Ossification in the Achilles Tendons

Abstract

Heterotopic ossification is the abnormal formation of mineralized bone in skin, muscle, tendon, or other soft tissues. Tendon ossification often occurs from acute tendon injury or chronic tendon degeneration, of which current treatment relies heavily on surgical removal of the ectopic bony tissues. Unfortunately, surgery creates additional trauma, which often causes recurrence of heterotopic ossification. The molecular mechanisms of heterotopic ossification are not well understood. Previous studies demonstrate that Mkx is a transcription factor crucial for postnatal tendon fibril growth. Here we report that Mkx^−/− mutant mice exhibit ectopic ossification in the Achilles tendon within one month after birth and the tendon ossification deteriorates with age. Genetic lineage labeling revealed that the tendon ossification in Mkx^−/− mice resulted from aberrant differentiation of tendon progenitor cells. Furthermore, tissue-specific inactivation of Mkx in tendon cells postnatally resulted in similar ossification phenotype, indicating that Mkx plays a key role in tendon tissue homeostasis. Moreover, we show that Hedgehog signaling is ectopically activated at early stages of tendon ossification and that tissue-specific inactivation of Smoothened, which encodes the obligatory transducer of Hedgehog signaling, in the tendon cell lineage prevented or dramatically reduced tendon ossification in Mkx^−/− mice. Together, these studies establish a new genetic mouse model of tendon ossification and provide new insight into its pathogenic mechanisms.

Introduction

Heterotopic ossification (HO) is the pathological formation of bone in soft tissues such as muscle, tendon, and ligaments ⁽¹⁾. Clinically, the most common form of HO occurs following traumatic injuries or orthopedic surgeries, of which ossification is often not detected clinically until months after the traumatic events although the pathogenic processes likely started soon after injury and the underlying pathogenic mechanisms are not understood ^(2–4). Patients with HO experience a wide range of problems, including chronic pain, limitation in range of motion, joint ankyloses, and skin ulceration ⁽⁴⁾. Currently, the primary treatment modality for HO, besides conservative management approaches including physical therapy and use of systematic non-steroidal anti-inflammatory drugs, is surgical excision of the ectopic bone. However, the efficacy of surgical excision is controversial since recurrences have been reported in many patients ⁽³⁾. In addition to trauma-induced HO, two rare genetic disorders, fibrodysplasia ossificans progressive (FOP) (OMIM 135100) and progressive osseous heteroplasia (POH) (OMIM 166350), are clinically characterized by extensive and progressive HO ^(5–7). FOP is caused by gain-of-function mutations in the ACVR1 gene, which encodes a type-I receptor for both Activin and BMP family proteins ⁽⁶⁾, whereas POH is caused by loss-of-function mutations in GNAS, encoding the Gαs subunit involved in G-protein coupled receptor signaling ^(5,8). FOP patients develop painful inflammatory nodules within the fascia, skeletal muscle, tendons, and ligaments, and start to form heterotopic bone through endochondral ossification before the age of five years, with ectopic bone accumulating over time and eventually causing nearly complete immobility ⁽⁷⁾. Although heterotopic bone formation in POH patients is also episodic and progressive, HO in POH patients is not associated with inflammation and develops mainly through intramembranous ossification that initiates within the dermis and progresses to the underlying deep connective tissues ⁽⁷⁾. Whereas HO in FOP patients is associated with aberrant Activin and BMP signaling ^(9,10), studies in mutant mouse models suggest that Gαs functions as a negative regulator of Hedgehog (Hh) signaling and that HO in POH patients might be due to increased Hh signaling ⁽⁸⁾. The episodic and progressive nature of HO in both FOP and POH suggests that elevated basal BMP or Hh signaling sensitizes soft tissues for ectopic osteoblast differentiation when combined with additional osteogenic signals provided by the local environment. However, the local environmental factors involved are not well understood. In addition, although studies in various animal models of trauma-induced HO suggest that BMP signaling might be involved in the pathogenesis of acquired HO ^(11,12), no genetic correlation with either the BMP or Hh signaling pathway has been found in patients with trauma-induced HO ⁽¹³⁾.

HO often affects tendon tissues. Tendon connects muscle to bone and transmits forces generated by muscle to bone, thus confer the integrity and moving capability of the musculoskeletal system. In trauma-induced tendon ossification, ectopic bony structures often form at sites of acute tendon rupture, or sites of surgical repair. In addition, tendon ossification is a major histological feature of tendinopathy at later stages ⁽¹⁴⁾. Reports have shown that tendon ossification occurs prevalently in rotator cuff tendons ⁽¹⁵⁾, as well as Achilles, biceps brachii, and quadriceps tendons ^(16,17). Recently, Agarwal et al. (2017) took advantage of previous findings that the Scleraxis (Scx) gene, encoding a basic helix-loop-helix transcription factor, is highly specifically expressed in all tendon cells in embryonic and postnatal stages ^(18,19), and demonstrated that postnatal Scx-expressing tendon progenitor cells contribute to ectopic ossification in both muscle and tendon tissues in mouse HO models ⁽²⁰⁾. Furthermore, tissue-specific activation of expression of a constitutively active form of the type I BMP receptor ACVR1 in the tendon lineage caused ectopic ossification in joints and the Achilles tendon in the transgenic mice in the absence of trauma, indicating that persistent excessive activation of BMP signaling can induce resident tendon progenitor cells to undergo osteogenic differentiation in vivo ⁽²⁰⁾. Interestingly, genetic inactivation of Mkx, which encodes a transcription factor important in tendon fibrillogenesis, caused Achilles tendon ossification in neonatal rats ⁽²¹⁾, but tendon ossification has not been reported in Mkx^−/− mutant mice. Whereas cultured tendon-derived cells from Mkx^−/− mutant rats showed accelerated chondrogenic and osteogenic differentiation compared with similar cells from wildtype rats, the cellular and molecular mechanisms underlying Achilles tendon ossification in Mkx^−/− rats are not understood. Here we report that Mkx^−/− mutant mice exhibit complete penetrance of Achilles tendon ossification by two months of age. We demonstrate, using conditional knockout and genetic lineage tracing studies, that Mkx is required postnatally in tendon cells to maintain Achilles tendon homeostasis. Furthermore, we uncover a crucial role for Hh signaling in Achilles tendon ossification.

Materials and Methods

Animals

The Mkx^−/− mutant mice and Mkx^c/c conditional mice have been described previously ⁽²²⁾. The Smo^c/c conditional mice ⁽²³⁾, Gli1^lacZ/+ knock-in mice ⁽²⁴⁾, and R26R-tdTomato reporter mice ⁽²⁵⁾ were obtained from the Jackson Laboratory (Bar harbor, ME). The Scx-Cre and Scx-Cre^ER mice ⁽²⁶⁾ and Scx-GFP transgenic mice ⁽¹⁹⁾ were generously provided by Dr. Ronen Schweitzer (Shriners Hospitals for Children, Portland, Oregon). All mouse strains were maintained by crossing to CD-1 wildtype mice (Charles River, Stock# 022), or by intercrossing between siblings. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at Cincinnati Children’s Hospital Medical Center (CCHMC). The mice were housed in AAALAC accredited barrier housing conditions at CCHMC. Autoclaved ventilated microisolator cages were changed every two weeks, and the feed (Lab diet 5010) and bedding (corn cob) were also autoclaved. Automatic water system was used, with reverse osmosis water that was filtered and UV sterilized. Light cycle was 14 hours light and 10 hours dark. Sentinel animals were tested quarterly and the mouse colony was maintained free of a list of viral, bacterial, and parasitic agents. Animals from other institutions were quarantined and tested negative or re-derived using blastocyst transfer before entering into the barrier housing facility. The facility requires that all animals are handled in certified cage change stations and that all clinical cases are appropriately followed up by veterinary staff. No veterinary clinical concern occurred in this study mouse colony. Prior to tissue collection and analysis, mice were euthanized by carbon dioxide asphyxiation for 10 minutes at 3 L/min, followed by cervical dislocation.

Whole mount skeletal preparations

Mice were euthanized at desired stage. Skin tissues were manually removed from hindlimbs. Hindlimb samples were fixed in 100% ethanol for 24 hours, stained in Alcian blue staining solution (15mg Alcian blue reagent dissolved in 80 ml 95% ethanol plus 20ml acetic acid) for 24 hours, fixed in 100% ethanol again for 24 hours, stained in Alizarin red staining solution (50 mg Alizarin red reagent dissolved in 1L 2% Potassium Hydroxide) for 24 hours, briefly washed, then cleared and stored in clearance solution (benzyl-alcohol: glycerol: 70% ethanol, 1:2:2).

Radiographic analysis

Animals were euthanized at desired stages, and imaged by X-ray with a Faxitron MS-20 specimen radiography system for 5 seconds at 25 kV. All images were taken at the same setting and with the same magnification. The areas of the ectopic bones were calculated using Microsoft photoshop to compare the severity of ectopic ossification.

Histology and X-gal staining

Mice were euthanized at predetermined stage and the hindlimbs were dissected, fixed in 4% paraformaldehyde (PFA) at 4 °C overnight, washed in PBS three times, and de-calcified in 14% EDTA solution (for Safranin-O staining) for 3 to 7 days. De-calcified samples were washed in running water overnight, then processed into 30% Sucrose solution, and embed in O.C.T. Cryo-section was performed at 12 to 20 µm thickness. For histology analysis, sections were stained with hematoxylin and Safranin-O. X-gal staining was performed as previously described ⁽²⁷⁾ to detect LacZ expression.

Immunofluorescent staining

Hindlimb samples were dissected and fixed in 4% PFA at 4°C overnight. After fixation, samples were washed three times, 10 minutes each in PBS. The fixed hindlimb samples were then soaked in 30% sucrose solution for 2 hours and embedded in O.C.T. Cryo-section was performed at 12 μm thickness in the sagittal orientation. Immunofluorescent staining was performed as previously described ⁽²⁸⁾. The primary antibodies used include anti-phospho-Smad1/5 (Cell Signaling Technologies, catalog# 13820, 1:50 dilution) and anti-Sox9 (Santa Cruz, catalog# sc-20095, 1:25 dilution). The secondary antibody was Alexa Fluor 568-conjugated goat anti-rabbit IgG (H+L) (Thermo Fisher, catalog# A11011, 1:500 dilution).

Tamoxifen treatment for inducible Cre activity

Tamoxifen stock solution was prepared by dissolving Tamoxifen (Sigma, catalog# T5648) in corn oil (Sigma, catalog# C8267) at 5 mg/ml, aliquoted into sterile 1.5 ml microcentrifuge tubes, and stored at −20 ^oC. An aliquot is thawed at room temperature shortly immediately before use and administered to both control and mutant animals by oral gavage once every 24 hours from postnatal day (P) 3 to P5 at a dosage of 75 mg/kg bodyweight per treatment. Mice were euthanized at the desired stages for skeleton staining.

Sample numbers and statistical analysis

For the X-ray radiography scanning, we scanned at least 4 male mice and 4 female mice of each genotype (wildtype, heterozygous, and homozygous mutants) at each stage (1 month, 2 months, 4 months and 6 months). Since the ectopic bone formed either unilaterally or bi-laterally in Achilles tendons, we considered each limb from the same mouse as an independent sample.

For skeleton analysis of the Mkx/Smo compound mutant limbs, we collected 16 Mkx^−/−; Smo^c/c samples at P30, of which 14 showed ectopic bone and were used in the statistical analysis. We collected 8 Mkx^−/−;Smo^c/c samples at P60, of which all showed ectopic bone and were used in the statistics analysis. We collected 22 Mkx^−/−; Smo^c/c;Scx-Cre samples at P30, of which 7 showed ectopic bone and were used in the statistical analysis. We collected 12 Mkx^−/−;Smo^c/c;Scx-Cre samples at P60, of which 8 showed ectopic bone and were used in statistical analysis.

For all other experiments, we analyzed at least 5 samples for each group per genotype per stage, and the actual numbers of samples were indicated in the relevant figure panels in the Results section. Samples from both sexes were included in analysis.

A method similar as previously described for quantifying bone formation from radiographs ^(29,30) was used to quantify the size of ectopic bone in the Achilles tendon. Briefly, all photographs for samples in each experiment were taken using the exact same settings, and the ectopic bone area was circled manually and pixel numbers recorded using Adobe Photoshop. When multiple separate pieces of ectopic bone were present in a single sample, a sum of the total pixels of the ectopic bones were used for that sample. Total pixel number of each sample was used to represent the relative size of the ectopic bone.

Statistical analyses were carried out using tools in the GraphPad Prism 5.0 software package (https://www.graphpad.com) using parametric assumptions. Normality test of ectopic bone size data distribution was performed using the Shapiro-Wilk W-test included in GraphPad Prism. To compare the ectopic bone size between male and female samples of the same genotype at any given stage, two-tailed Student’s t-test was used. Power analysis was performed using SAS9.4 (https://www.sas.com/en_us/software/sas9.html). To compare the ectopic bone size between Mkx^−/−;Smo^c/c and Mkx^−/−;Smo^c/c;Scx-Cre mice, two-tailed Student’s t-test was also used. However, for comparison of the ectopic bone formation in the Achilles tendon in wildtype, Mkx^+/−, and Mkx^−/− samples, data were analyzed by using one-way ANOVA, followed by Turkey’s multiple comparison post test. P value less than 0.05 was considered statistically significant and marked as *, P value less than 0.01 was marked as **, P value less than 0.001 was marked as *** when applicable.

Results

Mkx^−/− mutant mice exhibit ectopic ossification in the Achilles tendons

We first noticed that Mkx^−/− mutant mice walked abnormally by six months of age. Upon skeletal preparation and examination, we discovered that Mkx^−/− mutant mice had ectopic ossification in the Achilles tendons (Fig. 1). By one month of age, some Mkx^−/− mutant mice exhibit a small piece of ectopic ossification in the Achilles tendon, localized close to, but separate from, the tendon enthesis at the calcaneus (Fig. 1D). At later stages, multiple pieces of ectopic ossification appear to form independently in the midsubstance of the Achilles tendon and gradually grow and fuse into large pieces of bone by six months of age (Fig. 1E and 1F). Whereas Mkx^−/− mutant mice exhibit reduced collagen fibril growth in both limb and tail tendons, we have not detected any ectopic ossification in tendons outside of the hindlimbs (Supplemental Figure 1).

Fig. 1.

Mkx mutant mice exhibited ectopic ossification in tendons. (A–F) Skeleton staining of hindlimbs from wildtype (A–C) and Mkx^−/− mutant mice (D–F) at 1 month (A, D), 2 months (B, E), and 6 months of age (C, F). Six independent samples of each genotype at each stage were examined. Arrows point to ectopic ossification. ca, calcaneus; ti, tibia.

Suzuki et al. recently reported that Mkx-deficient rats display ectopic ossification in the Achilles tendons as early as postnatal day 3 but they did not detect any tendon ossification in their independently-generated Mkx knockout mouse line ⁽²¹⁾. To further examine the ectopic tendon ossification phenotype in Mkx^−/−mice, we analyzed Mkx^−/− mutant mice and Mkx^+/−heterozygous and wildtype littermates at different ages, from one month to six months, by X-ray radiography. As shown in Fig. 2, 60% (12 of 20) of the Mkx^−/− mutant mouse Achilles tendons had a clearly detectable ectopic radio-dense structure at one month of age (Fig. 2I). By two months of age, all Mkx^−/− mutant mice showed one or more pieces of radio-dense areas in each of their Achilles tendons (Fig. 2J–L), with 16 or more samples of each genotype examined at each stage. Remarkably, we found that Mkx^+/− mice also exhibited ectopic ossification detectable by two months of age, with the penetrance of ectopic ossification in the Achilles tendon increasing from less than 20% (3 of 18) at two months to 82% (18 of 22) by six months of age (Fig. 2F–H). The ossified areas increased in size with age in both Mkx^−/− and Mkx^+/− mice, with the ectopic bones being significantly larger in size in the Mkx^−/−mutant mice than in Mkx^+/− littermates at any given age (Fig. 2A–M). We also compared the ectopic bone sizes in the male and female mice at each stage but have not detected any obvious difference between the sexes, although further examination of a larger sample size is needed to rule out any small but significant difference between the sexes (Supplemental Figure 2). In addition, we found that a low percentage of wild-type mice also showed small radio-dense areas in the Achilles tendon at 6 months and later stages, although the size of the radio-dense areas was very small and did not rapidly form ectopic bone (Fig. 2D, M). The increases in size and percentage of the ectopic bones in Mkx^−/− and Mkx^+/− mice, in comparison with wild-type littermates, indicate that Mkx plays a crucial role in tendon maintenance and that the Mkx-deficient mice provide an excellent animal model for studying the cellular and molecular mechanisms of tendon ossification.

Fig. 2.

Achilles tendon ossification in Mkx^−/− mutant mice increased in severity with age. (A–L) X-ray radiography of hindlimbs from wild-type control (A–D), Mkx^+/− (E–H), and Mkx^−/− mutant (I–L) mice. White arrows point to the site of ectopic ossification in the Achilles tendons. The ratios at the bottom of each panel indicate the number of samples with ectopic ossification in the Achilles tendon over the total number of samples examined for that genotype at that stage. (M) Quantification of the ectopic bone size. The Y axis indicates the quantitative values for the ectopic bone area of each sample. Shapiro-Wilk Normality Test was performed and the ectopic bone size data for Mkx^−/− samples showed normal distribution at each stage examined. The ectopic size data for Mkx^+/− samples at four and six months also showed normal distribution. One-way ANOVA analysis followed by Turkey’s multiple comparison post test was performed on the measurements. Significant differences between Mkx^−/− and Mkx^+/− mice as well as between Mkx^−/− and wildtype mice are indicated (*, p<0.05; **, p<0.01; ***, p<0.001). ca, calcaneus; ti, tibia.

Ectopic bones in the Achilles tendons in Mkx^−/− mice develop from aberrant differentiation of the tendon lineage cells through endochondral ossification

It has been suggested that heterotopic bone formation could occur through either intramembranous or endochondral ossification ⁽⁷⁾. We performed Safranin-O staining of sagittal sections of the Achilles tendons from Mkx^−/−and control littermates at early postnatal stages and found that Safranin-O positive cartilage nodules were detectable in Mkx^−/− mutant Achilles tendons by P14 (4/10 Mkx^−/−mutants) (Fig. 3A–F and Supplemental Figure 3), indicating that the heterotopic bones develop through endochondral ossification.

Fig. 3.

Ectopic bones form through endochondral ossification in the Achilles tendons in Mkx^−/− mutant mice. (A–F) Saphranin-O staining of sagittal sections through the Achilles tendons of wildtype (A–C) and Mkx^−/− (D–F) littermates at P7 (A, D), P14 (B, E), and P21 (C, F). Arrow points to chondral nodules. Numbers at the lower-right corner of each panel indicate the number of samples with the phenotype shown over the total number of samples analyzed for that genotype at that stage (i.e., no abnormal chondral nodule in A – D whereas the indicated ratio of Mkx^−/− samples had chondral nodules in the Achilles tendons as shown in E and F). at, Achilles tendon; ca, calcaneus.

We next investigated the cellular process of tendon ossification in the Mkx^−/− mutant mice. Scx is a bHLH transcription factor expressed in tendon progenitor cells during embryonic development and continues expressed in tenocytes up to four months postnatally ^(19,31). Taking advantage of the transgenic Scx-GFP reporter mice ⁽¹⁹⁾, we compared Scx-GFP expression in Achilles tendons in Mkx^−/− mutant and control littermates. At P7, the Achilles tendons in wildtype mice exhibit uniformly robust Scx-GFP expression (Fig. 4A). In contrast, the Mkx^−/− mice exhibit reduced Scx-GFP levels in the midsubstance of the Achilles tendons (Fig. 4C) and some Mkx^−/− mice started to have Scx-GFP-negative nodules in the Achilles tendons (Fig. 4E). By P14, 50% of the Mkx^−/−mutant Achilles tendons (5 of 10 samples) exhibited at least one Scx-GFP-negative nodule (Fig. 4F).

Fig. 4.

Ectopic ossification in the Achilles tendons in Mkx^−/− mutant mice is accompanied by loss of Scx expression. (A–F) Frozen sections through the Achilles tendons of wildtype (A, B) and Mkx^−/− (C, D, E, F) mice showing Scx-GFP transgenic reporter fluorescence at P7 (A, C, E) and P14 (B, D, F), respectively. Arrow points to midsubstance of the Achilles tendon. Arrowhead points to chondral nodule. Numbers at the lower-right corner of each panel indicate the number of samples with the Scx-GFP pattern shown (no Scx-GFP negative nodule in A – D and with Scx-GFP negative nodule in E and F) over the total number of samples analyzed for that genotype at that stage. ca, calcaneus.

To investigate whether the Scx-GFP-negative nodules result from aberrant differentiation of tendon cells or from other cell types, we performed Cre/loxP-mediated genetic lineage tracing using Scx-Cre and R26R-tdTomato transgenic mice. The Scx-Cre transgenic mice express the Cre recombinase in all embryonic tendon progenitor cells. In Scx-Cre;R26R-tdTomato double transgenic mice, Cre-mediated recombinase activates tdTomato red fluorescence protein expression in all embryonic tendon cells and their progeny. As shown in Supplemental Figure 4, almost all of the Scx-GFP+ cells in the Achilles tendon in Mkx^+/+ mice were also tdTomato-positive (Supplemental Figure 4A). Remarkably, the Scx-GFP-negative nodule cells in the Mkx^−/− mutant Achilles tendons were also tdTomato-positive (Supplemental Figure 4B), indicating that these cells are derived from the tendon cell lineage. To further investigate the cell fate change during ectopic ossification in the Achilles tendons in Mkx^−/− mice, we examined expression of the Sox9 protein, a master regulator of chondrogenic differentiation ^(32–34), by immunofluorescent detection. Remarkably, whereas no Sox9-positive cell was detected in the midsubstance of the Achilles tendons in wildtype littermates (Fig. 5A, B), many cells in the Scx-GFP negative nodules in the Achilles tendons of P14 Mkx^−/− mice exhibited nuclear localized Sox9 protein (Fig. 5C, D, and Supplemental Figure 5C–F). In some P14 Mkx^−/− mice that had not formed Scx-GFP negative nodules but a subset of individual tendon cells in the midsubstance of the Achilles tendon had already down-regulated Scx-GFP expression (Supplemental Figure 5J), Sox9 protein was not detected in those Scx-GFP negative tendon cells (Supplemental Figure 5J). On the other hand, multiple Scx-GFP/Sox9 double positive cells were detected at the periphery of the Scx-GFP negative nodules that had already formed in the Achilles tendons in P14 Mkx^−/− mice (Fig. 5D and Supplemental Figure 5D, F), indicating that a subset of tendon cells could directly change their fate to undergo osteochondrogenic differentiation during ectopic ossification in the Achilles tendons in Mkx^−/− mice.

Fig. 5.

Analysis of Sox9 protein and Scx-GFP distribution in the Achilles tendons in wildtype and Mkx^−/− littermates at P14. (A, B) No immunofluorescent staining of Sox9 (red) was detected in the Achilles tendon tissues in wildtype mice. Scx-GFP is shown in green color and the cell nuclei were counterstained with DAPI (blue). B shows high magnification view of the region of the Achilles tendon marked by the white square in A. (C, D) Most cells in the chondral nodule in the Achilles tendon in Mkx^−/− mice showed Sox9 immunofluorescence (red). D shows a high magnification view of the region marked by the white square in C. Arrowheads point to Sox9/Scx-GFP double positive cells at the periphery of the chondral nodule. Five samples of each genotype were examined, with the patterns of Sox9 protein distribution in the other Mkx^−/− mutant samples shown in Supplemental Figure 5. ca, calcaneus.

Inactivation of Mkx postnatally and specifically in tendon cells results in tendon ossification

Mkx is expressed in tendon cells from E14.5, and Mkx^−/− mutant mice show hypoplastic tendon fibril growth postnatally ⁽²²⁾. To further investigate whether Mkx plays a primary role in tendon homeostasis postnatally, we used a Tamoxifen-inducible Cre mouse line, Scx-Cre^ER mouse, to conditionally inactivate Mkx in tendon cells in postnatal mice. Tamoxifen was given once a day for three days from postnatal day 3 to 5 and treated mice were examined by skeletal staining at 2 or 3 months (Fig. 6). By 2 months, ectopic ossification was detected in Achilles tendons at multiple spots in Mkx^c/c;Scx-Cre^ER mice, with 100% penetrance (n=8) (Fig. 6C), and the ectopic bones further enlarged by 3 months (n=8) (Fig. 6D). These results indicate that Mkx is required postnatally to maintain tendon homeostasis.

Fig. 6.

Conditional inactivation of Mkx in the postnatal tendon lineage led to ectopic ossification in the Achilles tendons. (A–D) Skeletons of hindlimbs from Mkx^c/c;Scx-Cre^ER conditional mutant mice (C, D) and Mkx^c/c control littermates (A, B) at 2 months (A, C) and 3 months (B, D) are shown. Six Mkx^c/c control and eight Mkx^c/c;Scx-Cre^ER samples were examined at each stage. Arrow points to ectopic bone. ca, calcaneus; ti, tibia.

Hedgehog signaling is ectopically activated at the onset of and plays an important role in tendon ossification in Mkx^−/−mutant mice

Aberrant activation of BMP signaling has been implicated in both hereditary and acquired forms of HO ⁽³⁵⁾. To investigate whether BMP signaling was ectopically activated during Achilles tendon ossification in the Mkx^−/− mutant mice, we performed immunofluorescent staining to detect phosphorylated Smad1/5 (pSmad1/5), which accumulates upon activation of BMP signaling ^(36–39). We did not detect any pSmad1/5 positive cells in the midsubstance of the Achilles tendon in Mkx^−/− mutant mice prior to aberrant nodule formation or in the wildtype control littermates at P7 (Supplemental Figure 6A, D). Even in the well-formed Scx-GFP negative nodules in the Achilles tendons in P14 Mkx^−/− mutant mice, only a small subset of the cells within the nodule were positive for pSmad1/5 (Supplemental Figure 6E, F). Furthermore, no increase in pSmad1/5 accumulation was detected at the periphery of the chondral nodules where Scx-GFP/Sox9 double positive cells were located in the Achilles tendons in Mkx^−/− mice (compare Supplemental Figure 6F with Fig. 5D). These results suggest that Bmp signaling is unlikely to be the major driving force of Achilles tendon ossification in Mkx^−/− mice although it is activated in a subset of cells within the chondral nodules during the ectopic ossification.

Hedgehog signaling is crucial in endochondral ossification and involved in proper mineralization of the fibrocartilage cells within the enthesis, the tendon to bone insertion site ^(40–44). Gli1 is a transcription factor downstream of Hedgehog signaling ^(45,46). To investigate whether Hedgehog signaling activity is involved in Achilles tendon ossification in Mkx^−/− mice, we crossed mice carrying the Gli1^LacZ knock-in allele with Mkx^−/− mice and subsequently analyzed LacZ reporter activity in the Mkx^−/− mutant and control littermates. LacZ was not detected in the tendon midsubstance in Mkx^+/+;Gli1^lacZ/+ mice at any stage up to four months of age (Fig. 7A, B, and Supplemental Figure 7). In contrast, we found that LacZ is activated in a subset of cells in the Achilles tendon midsubstance in Mkx^−/−;Gli1^LacZ/+ mice as early as P7 (Fig. 7C). Upon formation of chondral nodules in the Achilles tendon in P14 Mkx^−/−;Gli1^LacZ/+ mice, LacZ was expressed throughout the nodules except in the hypertrophic chondrocytes in the center of the nodules (Fig. 7D). These results indicate that Hedgehog signaling is activated during tendon ossification in Mkx^−/− mutant mice.

Fig. 7.

Hedgehog signaling was activated at the onset of ectopic ossification in the Achilles tendons in Mkx^−/− mutant mice. (A–D) X-gal staining detection of Gli1-LacZ expression in the Achilles tendons in Mkx^+/+;Gli1^lacZ/+ mice (A, B) and Mkx^−/−;Gli1^lacZ/+ littermates (C, D) at P7 (A, C) and P14 (B, D), respectively. Numbers at the lower-right corner of each panel indicate the number of samples with the X-gal expression pattern shown over total number of samples analyzed for that genotype at that stage. Arrows points to X-gal stained region in the Achilles tendons only detected in a subset of Mkx^−/−;Gli1^lacZ/+ mutants at these stages. at, Achilles tendon; ca, calcaneus.

To further investigate whether Hedgehog signaling plays a crucial role in tendon ossification in the Mkx^−/− mutant mice, we generated and analyzed Mkx^−/− mice with tendon tissue-specific deletion of Smo. In contrast to Mkx^−/−;Smo^c/c littermates, of which more than 70% (14 out of 16 hindlimbs) showed obvious bony tissues in the Achilles tendon by P30 (Fig. 8A), only about 30% of Mkx^−/−;Smo^c/c;Scx-Cre compound mutant mice (7 out of 22 hindlimbs) had barely detectable ossification in the Achilles tendon at P30 (Fig. 8C). By P60, 100% of Mkx^−/−;Smo^c/c mice (8 of 8) showed ectopic bones in the Achilles tendon (Fig. 8D), whereas about 67% of Mkx^−/−; Smo^c/c;Scx-Cre mice (8 of 12) had small ossified tissues (Fig. 8F) and the other 33% Mkx^−/−; Smo^c/c;Scx-Cre mice had no detectable tendon ossification at all (Fig. 8E). Quantitative comparison of the ectopic bones in the Mkx^−/−;Smo^c/c;Scx-Cre with the Mkx^−/−;Smo^c/c mice showed a significant reduction in ectopic ossification in the Mkx^−/−;Smo^c/c;Scx-Cre mice (Fig. 8G). Considering that the Cre-mediated deletion of Smo in the tendon tissues is likely incomplete, the dramatic reduction in both frequency and size of the ossified tissues in the Achilles tendon in Mkx^−/−;Smo^c/c;Scx-Cre mice, in comparison with the Mkx^−/−;Smo^c/c siblings, indicate that Smo-mediated Hedgehog signaling plays a crucial role in tendon ossification in Mkx^−/− mice.

Fig. 8.

Achilles tendon ossification in Mkx^−/− mutant mice was dependent on Hedgehog signaling activation. (A–G) Skeletal preparations of hindlimbs from Mkx^−/−;Smo^c/c (A, D) and Mkx^−/−;Smo^c/c;Scx-Cre (B, C, E, F) mice at P30 (A–C) and P60 (D–F) are shown. Arrow points to ectopic bone. Numbers at the lower-right corner of each panel indicate the number of samples with the phenotype shown over the total number of samples analyzed for that genotype at that stage (i.e., Panels A, C, D, F show ratio of samples with the ectopic bone phenotype whereas Panels B and E show the ratio of samples without ectopic bone). (G) Quantitative analysis of the ectopic bone. The Y axis indicates the quantitative values for the ectopic bone area of each sample, with the Mkx^−/−;Smo^c/c control samples shown as filled circles and the Mkx^−/−;Smo^c/c;Scx-Cre mutant samples shown as open circles. Two-tailed Student’s t-test was used to compare the relative sizes of the ectopic bones in Mkx^−/−;Smo^c/c and Mkx^−/−;Smo^c/c; Scx-Cre mice at P30 and P60, respectively. Significant differences are indicated. **, p<0.01; ***, p<0.001. ca, calcaneus; ti, tibia.

Discussion

Tendons are integral parts of the musculoskeletal system and have an essential role in transmitting contractile forces from muscle to bone to generate body movement. Due to this role, tendons withstand considerable loads during locomotion, often at several times of the body weight. In addition to tendon tissue rupture caused by athletic or other activities, repetitive loading contributes to overuse injuries. Damaged tendon tissue heals very slowly and rarely attains the structural integrity and mechanical strengths of the preinjury tendon tissue, resulting in tissue degeneration with age and the common clinical problem called tendinopathy ^(14,47). A major histopathological feature of tendinopathy, particularly at later stages, is ectopic ossification in the tendon tissues ^(14,48). The pathogenic mechanisms underlying tendinopathy and tendon ossification are not well understood. Suzuki et al. (2016) reported that ectopic ossification occurred in the Achilles tendons in 20% and 100% of the Mkx^−/− homozygous mutant rats at three and five weeks of age, respectively. However, they did not detect ossification in the Achilles tendons in P0 and 4-week-old Mkx homozygous mutant mice ⁽²¹⁾. In this study, we show that 40% of Mkx^−/− mice had chondral nodules in the Achilles tendon by P14 and 100% of the Mkx^−/− mice showed endochondral bone in the Achilles tendons by two months of age. Although there might be phenotypic differences between the two Mkx mutant mouse strains used in these studies, possibly due to the differences in the targeted mutations, with Suzuki et al. studying mice carrying a venus GFP expression cassette at the ATG codon of the Mkx gene ⁽⁴⁹⁾ whereas our mouse Mkx⁻ allele has an exon-3 deletion that not only removes the homeodomain-coding region but also disrupt the reading frame of all alternatively spliced forms of the gene product ⁽²²⁾, our finding that 100% of Mkx^−/− mice exhibit Achilles tendon ossification by two months of age and 80% of Mkx^+/− mice develop Achilles tendon ossification by six months of age, together with the prior report of Achilles tendon ossification in Mkx^−/− rats, indicate that disruption of Mkx function genetically predisposes the animals to Achilles tendon ossification.

What are the cellular and molecular mechanisms involving Mkx function in tendon ossification? Suzuki et al. (2016) suggested that the Mkx^−/− rats exhibit Achilles tendon ossification due to failed tendon differentiation ⁽²¹⁾. However, most, if not all, tendons were phenotypically abnormal in the Mkx^−/− rats and Mkx^−/− mice ^(21,22,49), ectopic ossification mainly occurred in the Achilles tendons in these mutant animals ⁽²¹⁾ (and this study). In addition, whereas significant reduction in tendon extracellular matrix molecules have been detected in Mkx^−/− mice prior to birth, we demonstrate that conditional inactivation of Mkx in Scx-expressing tendon cells in postnatal mice still caused ectopic ossification in the Achilles tendons. Moreover, no detectable defect in tendon differentiation has been reported in Mkx^+/− heterozygous mice, but we detected ectopic ossification in the Achilles tendons in a high percentage of those animals as well. Thus, although the effects of loss of Mkx function on tendon differentiation likely contributes to the predisposition to tendon ossification, together with our data showing that ossification occurs from Scx-expressing tendon progenitor cells that subsequently lose Scx expression in the Mkx^−/− mice, our data suggest that Mkx plays a major role in maintaining tendon homeostasis against aberrant osteogenic differentiation.

Previous studies suggest that the tendon ECM molecules, particularly biglycan (Bgn) and Fibromodulin (Fmod), form the niche that controls self-renewal and differentiation of tendon stem/progenitor cells (TSPCs) in vivo ⁽⁵⁰⁾. Mice lacking both Bgn and Fmod exhibit ectopic ossification in Achilles and patellar tendons as early as two months after birth ^(50,51). Both Mkx^−/− rats and Mkx^−/− mice had significant reduction in expression of Fmod and several other tendon ECM components in the early postnatal stages ^(21,22,49). It is possible that an important role for Mkx in maintaining TSPC cell fate is through regulation of expression of tendon ECM molecules including Fmod. Bi et al. (2007) showed that TSPCs isolated from mice lacking both Bgn and Fmod showed enhanced signaling activity in response to BMP2 treatment than TSPCs from wildtype mice ⁽⁵⁰⁾. Suzuki et al. (2016) reported that expression of several BMP pathway genes, including Bmpr1a, Bmpr2, Smad1, and Smad5 was elevated in the Achilles tendons in two-week-old Mkx^−/− rats ⁽²¹⁾. However, it has not been reported whether TSPCs had increased BMP signaling in vivo in the Achilles tendons in any of these mutant animals prior to the onset of tendon ossification. We detected pSmad1/5 protein in a subset of Scx-negative cells in the chondral nodules, but not prior to nodule formation, in the Mkx^−/− Achilles tendon. On the other hand, we found that Mkx^−/− mutant mice had localized activation of Gli1 expression in the midsubstance of the Achilles tendon as early as P7, suggesting that Hh signaling is locally activated prior to tendon ossification in these mutant mice. Furthermore, we demonstrate that tendon-specific inactivation of Smo significantly blocked tendon ossification in the Mkx^−/− mutant mice, providing definitive evidence that Hh signaling activation plays a crucial role in tendon ossification in these mutant mice. Since Hh signaling has not previously been implicated in HO pathogenesis other than in POH, further studies are warranted to understand whether Hh signaling plays an important role in trauma-induced HO or tendon ossification in late stages of degenerative tendinopathy and how Hh signaling is activated during tendon ossification.

Whereas studies of patient specimens have suggested that ectopic bone formation in tendons frequently involved endochondral ossification ^(14,17,52), the sources of cells contributing to the development and progression of tendon ossification in patients are not known. In a recent study, Agarwal et al. (2017) showed that Scx-lineage cells contribute to in both trauma-induced and ectopic BMP signaling induced HO in muscle and tendons in mice ⁽²⁰⁾. In this study, we demonstrate that the chondral nodules that formed in the Achilles tendons in Mkx^−/− mice were all derived from Scx-expressing tendon lineage cells. Whereas Sox9 protein was only detected after downregulation of the Scx-GFP reporter in most of the cells forming the nodules, many cells at the periphery of the nodules expressed both Scx-GFP and Sox9 protein. These results suggest that the formation of chondral nodules during Achilles tendon ossification in Mkx^−/− mice bears some similarity to the formation of bone eminences during development of tendon-bone attachment sites. Previous studies showed that the tendon-bone attachment unit is formed modularly from a distinct pool of Scx⁺/Sox9⁺ progenitor cells and that both Scx⁺/Sox9⁺ and Scx⁺/Sox9⁻ progenitor cells contribute to tenocytes ^(53,54). However, in those studies, the Scx⁺/Sox9⁺ progenitor cells were only detected in embryonic stages. In this study, we show that conditionally inactivating Mkx in the postnatal tendon cells also lead to ectopic ossification in the Achilles tendon, similar as in the Mkx^−/− mice. Moreover, we could only detect Sox9 protein in the chondrogenic nodules but not prior to nodule formation in the midsubstance of the Achilles tendons in Mkx^−/− mice, indicating that the ectopic nodules formed from postnatal tendon cells that failed to maintain tendon fate rather than from a preexisting Scx⁺/Sox9⁺ progenitor pool. The Scx-GFP⁺/Sox9⁺ cells detected at the periphery of the chondral nodules in the Achilles tendons in Mkx^−/− mice most likely arose locally from Scx⁺/Sox9⁻ tendon cells in the tendon midsubstance. Further investigation whether similar molecular mechanisms are involved in the induction of the Scx⁺/Sox9⁺ progenitor cells during formation of tendon-bone attachment units and induction of the Scx⁺/Sox9⁺ cells during ectopic tendon ossification will provide new insight into the mechanisms of tendon ossification.

A potential limitation of the Mkx mutant mice as a model for understanding the pathogenic mechanisms of tendon ossification is that these mice mainly develop ectopic ossification in the Achilles tendon whereas tendon ossification in humans have been reported in various tendons and is more common in other regions such as the rotator cuff tendons ⁽⁵⁵⁾. A clear difference between the Achilles tendon ossification in Mkx^−/− mice and most reported tendon ossification pathologies in humans is that the vast majority of the cases of tendon ossification in humans occurred subsequently to trauma or surgery or preexisting tendinopathy ⁽⁵⁵⁾. A few patients with extensive Achilles tendon ossification without known predisposing factors have been reported ^(56–58) and the possibility of a hereditary component for this entity has been raised but has not been proven ⁽⁵⁶⁾. Given that both Mkx^−/− rats as well as Mkx^+/− and Mkx^−/− mice develop ectopic bones in the Achilles tendons, it is possible that rare variants in the MKX gene might be a genetic predisposing factor for Achilles tendon ossification in humans. Recent studies have shown that Mkx expression in tendon cells is regulated by mechanical strain through the Gtf2ird1 transcription factor and that mechanical stimulation caused cultured Mkx^−/− mutant, but not wildtype, tendon cells to undergo chondrogenic differentiation ^(21,59). Furthermore, it has been shown that the inflammatory cytokine IL-1β strongly suppressed MKX gene expression in cultured human anterior cruciate ligament (ACL) derived cells and that MKX expression was significantly reduced in degenerated ACL in osteoarthritis patients ⁽⁶⁰⁾. Thus, it is possible that loss of Mkx function might enhance trauma- or surgery-induced pathogenesis of tendon ossification. Further investigation of the roles of Mkx in trauma-induced ectopic ossification using the Mkx mutant mouse model will significantly improve the understanding of the pathogenic mechanisms of tendon ossification.