Dysregulation of hedgehog signalling predisposes to synovial chondromatosis
Abstract
Synovial chondromatosis is a condition affecting joints in which metaplastic cartilage nodules arise from the synovium, causing pain, joint dysfunction, and ultimately joint destruction. Because dysregulation of hedgehog signalling is a feature of several benign cartilaginous tumours, expression of the hedgehog target genes PTC1 and GLI1 was examined in this study in samples from human synovial chondromatosis. Significantly higher expression levels were found in synovial chondromatosis than in the synovium, from which it arises. To determine if hedgehog-mediated transcription predisposes to synovial chondromatosis, the extra-toes mutant mouse, which harbours a heterozygous mutation in the hedgehog transcriptional repressor, Gli3, resulting in decreased expression of Gli3 protein, was studied. The extra-toes mutant mouse has a phenotype consistent with overactive hedgehog signalling, suggesting that Gli3 acts as a transcriptional repressor of limb development. Eighty-five per cent of Gli3 mutant mice developed synovial chondromatosis at 18 months of age, compared with 30% of wild-type littermates (p < 0.05). Three of the ten Gli3 mutant mice treated with triparanol, which blocks hedgehog signalling upstream of the Gli transcription factors, developed synovial chondromatosis, compared with eight of ten control mice. These data demonstrate that hedgehog signalling plays an important role in the development of synovial chondromatosis and suggest that blockade of hedgehog signalling may be a potential treatment for this disorder. Keywords: synovial chondromatosis; hedgehog; triparanol; Gli genes; mouse models; chondrocytes; synovium; immunohistochemistry Introduction Synovial chondromatosis is a condition in which meta- plastic cartilage nodules arise from synovial tissues. Affected joints become painful and there can be destruction of the articular surface. Symptoms and joint destruction are caused by motion of the cartilage nodules within the joint [1– 4]. Although it is gen- erally considered to be a benign disorder, malignant transformation has been reported in a few cases [5– 9]. Surgical excision is the usual treatment but the results of surgery are less than satisfactory, with residual pain and joint dysfunction being not uncommon out- comes [10,11]. Familial occurrence is reported [12,13], although linkage studies have not been performed. The demonstration of chromosomal alterations in the lesions suggests that it is a clonal process [14]. Despite this, the cause of synovial chondromatosis remains unknown, and both neoplastic and reactive aetiologies have been proposed. Enchondromas and osteochondromas are benign cartilage tumours that occur near the ends of long bones and can be caused by mutations that dereg- ulate Indian hedgehog (IHH) signalling, a pathway that plays a crucial role regulating growth plate chon- drocyte differentiation [15,16]. Long bones form by enchondral ossification, a process in which mesenchy- mal cells condensate and differentiate into chondro- cytes, forming a cartilaginous skeleton, which is sub- sequently transformed into a calcified skeleton. Later in this process, cartilage persists at the ends of the bone, forming the joint and the growth plate, or physis, which is responsible for growth of the long bones. Growth plate chondrocytes undergo an ordered pro- cess of differentiation, in which resting chondrocytes proliferate, hypertrophy, and undergo terminal differ- entiation as they progress away from the epiphyseal to the metaphyseal side of the growth plate. Pre- hypertrophic chondrocytes express IHH, which stim- ulates the production of parathyroid hormone-related protein (PTHrP). PTHrP, in turn, inhibits the terminal differentiation of growth plate chondrocytes, and also down-regulates expression of IHH. In this way, IHH and PTHrP act in a feedback loop regulating physeal chondrocyte differentiation [17– 22]. Hedgehog ligands, including IHH, signal through Patched-1 (PTC1) and Smoothened (SMO). Hedgehog binding releases the otherwise constitutive inhibition of PTC1 on SMO, ultimately activating members of the GLI family of transcription factors [23]. GLI genes are the mammalian homologues of the Drosophila zinc-finger gene cubitus interruptus. In the absence of hedgehog, cubitus interruptus is cleaved by the pro- teosome, and the carboxy-terminus fragment translo- cates to the nucleus, where it acts as a transcriptional repressor. When hedgehog ligand is present, this pro- tein modification is inhibited and the full-length form acts as a transcriptional activator. There are three mammalian GLI proteins, some of which have been shown to be regulated by hedgehog-mediated cleav- age. Although the precise functions of the GLI tran- scription factors are still being elucidated, each has a different function in transcriptional regulation. Genetic and biochemical data suggest that GLI1 and GLI2 are positive regulators and GLI3 is a negative regu- lator of hedgehog signalling in a variety of cell types during development [24– 27]. Studies of mutant mice suggest that Gli1 is not essential for transduction of the hedgehog signal [28]. Transcriptional activation by hedgehog signalling results in the up-regulation of downstream genes, including the hedgehog recep- tor, PTC1, and the transcription factor, GLI1 [29,30] (Figure 1). A variety of agents block hedgehog signalling: most of these were discovered as teratogens [31] that cause holoprosencephaly. This is a disorder in which there is malformation of the central portions of the brain and head, and can also be caused by mutations in sonic hedgehog [32]. In vitro data demonstrate that such agents block hedgehog signalling upstream of the Gli transcription factors. One such agent, triparanol, was utilized as an anti-cholesterol agent in humans until side effects were discovered [33]. Another hedgehog blocking agent, cyclopamine, was found to inhibit cell growth in medulloblastoma cell cultures and xenografts, tumours that exhibit hedgehog signalling activation [34,35]. Dysregulation of hedgehog signalling is present in a variety of benign cartilage tumours. Enchondromato- sis can be caused by a mutation in the receptor for PTHrP, resulting in activation of hedgehog signalling [16]. Osteochondromas can be caused by mutations in EXT genes [15,36], the mammalian homologues of Drosophila tout-velu, which regulates the diffusion of hedgehog ligand [37]. Because we previously found that overexpression of Gli2 in the murine growth plate caused an enchondromatosis-like phenotype [16], we investigated mice deficient in Gli3 for the development of cartilaginous tumours. Surprisingly, we found that these mice developed synovial chondromatosis. There- fore, in this study, we investigated the role of hedge- hog transcriptional dysregulation in synovial chondro- matosis. Human samples were examined for expres- sion of the hedgehog target genes, and mice in which hedgehog transcriptional repression is inhibited due to a mutation in Gli3 were investigated further. Figure 1. The hedgehog signalling cascade. (A) In the absence of hedgehog ligand, the receptor Patched-1 (PTC1) inhibits the transmembrane protein smoothened (SMO). This inhibition keeps the hedgehog-mediated transcription factors GLI1, GLI2, and GLI3 associated with a microtubule complex, and proteins including fused (FU), suppressor of fused (SUFU), and costal-2 (COS2). SMO inhibition is also associated with repressor forms of the GLI transcription factors entering the nucleus. The repressor forms are cleaved versions of the full-length transcription factors, and act to inhibit expression of hedgehog target genes, such as PTC1 and GLI1. In mammals, genetic evidence suggests that GLI1 is functionally redundant, GLI2 acts primarily as an activator of transcription, and GLI3 acts primarily as an inhibitor of transcription. (B) In the presence of hedgehog ligand (HH), PTC1 no longer inhibits SMO; the GLI transcription factors are released from the microtubule-associated complex; and full-length forms enter the nucleus, activating transcription. This results in the expression of the target genes PTC1 and GLI1. The up-regulation of components of its own signalling cascade (PTC1 and GLI1) by hedgehog ligand is thought to act as a feedback loop to regulate the level of activation Materials and methods The human and animal studies reported in this paper were reviewed and approved by the human research and animal research review boards of The Hospital for Sick Children, University of Toronto. Human samples Case material from two patients who underwent surgi- cal treatment for synovial chondromatosis was inves- tigated. Excised metaplastic cartilage as well as nor- mal synovium from the same joint was available for analysis. Normal synovial tissue from three addi- tional patients was obtained at the time of surgery for procedures unrelated to synovial pathological pro- cesses as additional controls. Tissues were cryopre- served as soon as possible after surgical excision. RNA was extracted from the cryopreserved tissues using TRIZOL reagent (Gibco-BRL). The relative level of expression of PTC1 and GLI1 in the samples was analysed using real-time and semi-quantitative RT- PCR, compared with the expression of GAPDH using primers as previously reported [16]. Mice The extra-toes mouse harbours a 3r deletion in the Gli3 gene, resulting in decreased expression of Gli3 [38]. Heterozygous Gli3 mutant mice have a normal lifespan, exhibiting post-axial polydactyly and cranio- facial anomalies, a phenotype similar to that found in Grieg cephalopolysyndactyly. Heterozygous Gli3 mutant mice were bred on a C57BL/6J background to produce Gli3 heterozygous mice and wild-type litter- mates. Mutant mice were identified by the polydactyly phenotype, and the genotype of the mice was con- firmed.Sixty heterozygous Gli3 mutant mice (half males and half females) and 60 wild-type littermates (half males and half females) were studied. Twenty mice from each group (ten males and ten females) were sacrificed at 3, 9, or 18 months of age. The limbs were examined by radiography using high-resolution radiographic film. Synovial chondromatosis was identified by observing ectopic calcification around a joint. The limbs were fixed in paraformaldehyde for histol- ogy with safranin-O and for immunohistochemistry. Immunohistochemistry was performed as previously reported [39] using a goat anti-PTC antibody raised against mouse PTC amino acids 18– 36 (1 : 100; Santa Cruz). Staining was blindly assessed by three indi- viduals and was scored by the percentage of cells exhibiting staining, the intensity of staining using a three-point scale, and the location of staining within 3 weeks of age, using orogastric gavage. Twenty mice were studied, with ten receiving triparanol (500 mg/kg administered three times each week; Hoechst Mar- ion Roussel, Cincinnati, OH, USA) and the remain- der receiving only the carrier, olive oil. Because tri- paranol decreases serum cholesterol [33], this was measured as a surrogate to determine if the mice absorbed the drug, using a fluorometric assay [40]. The mice were sacrificed at 18 months of age. All of the limbs of the mice were analysed for the development of synovial chondromatosis after staining with safranin-O and were observed using high-resolution radiography. After sacrifice, RNA from the proximal end of the humerus (the growth plate) was used to determine the level of expres- sion of Ptc expression compared with the control gene Gapdh, using previously reported primers and techniques [16]. Results Human synovial chondromatosis expresses the hedgehog target genes PTC1 and GLI1 To determine if there is hedgehog transcriptional activation in human synovial chondromatosis, the expression level of hedgehog target genes was com- pared between synovial chondromatosis and the nor- mal synovium. The synovial chondromatosis samples exhibited a level of expression of PTC1 that was eight-fold higher than the normal synovium from both the same patient and other patients. The level of GLI1 expression was elevated six-fold in the syn- ovial chondromatosis samples compared with nor- mal synovium. Similar levels of increased expression were found using semi-quantitative RT-PCR analysis (Figure 2). Hedgehog signalling blockade To determine if hedgehog signalling upstream of the Gli transcription factors is important in the for- mation of synovial chondromatosis, additional male heterozygous Gli3 mutant mice were treated with the hedgehog blocking agent triparanol starting at We studied the phenotype of heterozygous Gli3 mutant mice and their wild-type littermates to deter- mine if hedgehog-mediated transcriptional activation could cause synovial chondromatosis. Heterozygous Gli3 mutant mice, which exhibit decreased expres- sion of the hedgehog transcriptional repressor Gli3, have a relative increase in hedgehog-mediated tran- scriptional activation [38]. Neither the Gli3 mutant mice nor their wild-type littermates showed evidence of synovial chondromatosis at birth. However, there was a dramatically increased incidence of the lesions developing in the Gli3 mutant mice with time. At 3 months of age, four of the Gli3 mutant mice devel- oped synovial chondromatosis, while none of the wild- type mice developed this phenotype. When the mice were 18 months of age, 17 of the 20 mutant mice (nine male and eight female) developed synovial chondro- matosis, while only six of the 20 wild-type littermates (four male and two female) developed this pheno- type (Figure 3, p < 0.01 using Fisher’s exact test). Although synovial chondromatosis can form in the absence of a genetic alteration decreasing repression of hedgehog-regulated transcription, lack of the hedgehog transcriptional repressor Gli3 contributes significantly to its development. Figure 2. Expression of hedgehog target genes in human synovial chondromatosis. Semi-quantitative RT-PCR for PTC1, GLI1, and GAPDH from two cases of synovial chondromatosis (lanes labelled ‘C’), normal synovium from the same patients (lanes labelled ‘S’), and normal synovium from a knee without synovial chondromatosis (labelled ‘NS’), showing substantially greater levels of expression of PTC1 and GLI1 in synovial chondromatosis compared with the normal synovium. Products from both primer pairs were amplified together in a single reaction and were separated by electrophoresis in each lane. The murine lesions were located exclusively at the knees. These lesions showed a histological appearance similar to that in humans [1– 4], with benign-appearing chondrocytes in a cartilaginous matrix in lesions that arise from the synovium. In many of the lesions there was calcification. Similar to the findings in advanced human cases, degenerative changes devel- oped in the articular cartilage (Figure 4). The degen- erative changes in the Gli3 mutant mice occurred adjacent to the synovial lesions, while regions of the articular cartilage in locations remote to the synovial lesions did not show degenerative changes. Observa- tion of sections stained using Safranin-O showed nor- mal staining of the articular surface in regions remote from the lesions, suggesting a normal distribution of proteoglycans. While it is possible that synovial chon- dromatosis occurs secondary to degenerative changes in articular cartilage, with loose cartilaginous frag- ments acting as a nidus for the formation of a synovial chondroma, the relationship of the chondromas to the articular degenerative changes suggests that the devel- opment of synovial lesions precedes and causes the degenerative changes in the adjacent articular carti- lage. This notion is supported by our finding of a lack of degenerative changes in regions distant to the syn- ovial lesions. Figure 3. The incidence of synovial chondromatosis in mice over time, as detected using high-resolution radiography. ‘Gli3 mutant’ indicates heterozygous Gli3 mutant mice (extra-toes mouse), and ‘wild type’ indicates wild-type littermates. The percentage of mice developing synovial chondromatosis at each time point (3, 9, and 18 months) is given. Synovial chondromatosis develops over time and the lesions are present in a significantly higher proportion of Gli3 mutant mice than in wild-type littermates Ptc1 is expressed in murine synovial chondromatosis at a similar level to the growth plate To analyse for evidence of hedgehog-mediated tran- scriptional activation in murine synovial chondromato- sis, we examined the expression of the hedgehog tar- get gene Ptc1 at the protein level, using immunohis- tochemistry. Since synovial chondromatosis and the growth plate are located near each other, the level of protein expression could be compared between the lesions and the growth plate on the same histological section. Although not all cells in synovial chondro- matosis expressed Ptc1, the intensity of staining for Ptc1 in the cells expressing this protein was similar to that observed in cells in the pre-hypertrophic zone of the growth plate (Figure 5). The pre-hypertrophic zone of the growth plate is composed of chondrocytes that have not undergone terminal differentiation and con- tinue to proliferate. Hedgehog signalling is known to play a crucial role in these cells. Ptc1 expression was previously identified in these pre-hypertrophic cells, using the same immunohistochemical technique and antibody that we used in this study [39]. There was no difference in the proportion of cells exhibiting stain- ing for Ptc1 or in the intensity of cellular staining between lesions that arose in wild-type or in Gli3 mutant mice [54% 8% standard deviation (SD) in lesions in Gli3 mutant mice versus 49% 9% SD in lesions in wild-type mice, with an intensity of staining 2 in all of the lesions]. However, there was a dif- ference in the proportion of cells in the growth plate exhibiting Ptc1 staining between wild-type and Gli3 mutant mice, with the Gli3 mutant mice exhibiting a larger proportion of positively staining cells (24% of cells ±8% SD in wild-type mice versus 36% ± 5% SD, p < 0.05, n 20 at age 18 months, Figure 5). This finding is consistent with the notion of Gli3 act- ing as a repressor of hedgehog signals, and thus its deficiency resulting in increased expression of hedge- hog signalling targets, such as Ptc1. Figure 4. The histological and radiographic appearance of synovial chondromatosis in mice. Panels A and B are high-resolution radiographic images of a mouse knee, with A showing synovial chondromatosis as indicated by the extra-articular calcifications [most dramatic in the bottom (posterior) portion of the joint]. The femur is on the left, the tibia on the right, and the patella at the top of the radiograph. Panel B indicates the appearance of an uninvolved knee, lacking the extra-articular calcifications. Panels C, D, and E are histology sections, with D showing a knee from a wild-type mouse and C showing the knee from a Gli3 mutant mouse, which developed synovial chondromatosis (C and D at 100 magnification). Panel C shows significant extra-articular cartilage (involving most of the posterior part of the joint), and associated joint destruction. Panel D shows an uninvolved joint and the tissue between the tibia and the femur is the normal meniscus. Panel E shows a high magnification view of the synovial chondromatosis lesions (1000× magnification) Hedgehog signalling upstream of the transcription factors plays a crucial role in the development of synovial chondromatosis To determine if hedgehog signalling upstream of the Gli transcription factors plays a role in the develop- ment of synovial chondromatosis, Gli3 heterozygous mutant mice were treated with the hedgehog block- ing agent triparanol. Since triparanol also decreases serum cholesterol, we measured the serum choles- terol level as a surrogate marker for the absorption of the drug. Serum cholesterol levels in the mice treated with triparanol were half those in control mice. The level of expression of Ptc1 in the proximal humeral growth plate was used to determine if triparanol was indeed acting to decrease hedgehog-mediated tran- scriptional activity. Mice treated with triparanol had a 50% decline in the level of Ptc1 expression as measured using real-time PCR. Immunohistochemical analysis of the growth plates showed a decline in the number of growth plate chondrocytes expressing Ptc1 (36% of cells 5% SD for control mice versus 22% of cells 9% SD in mice treated with tripara- nol, p < 0.05, n 10 at age 18 months, Figure 6). Treatment with triparanol resulted in a substantial decrease in the chance of mice developing synovial chondromatosis, with only three of the ten treated mice developing synovial chondromatosis, while eight of the ten mice treated with the carrier alone devel- oped the lesions (Figure 6, p < 0.01). Treatment with triparanol did not seem to alter the histology or the radiographic appearance of synovial chondromatosis, when it occurred. Discussion Here we demonstrate that dysregulation of hedgehog signalling predisposes to the development of synovial chondromatosis. The expression levels of the hedge- hog target genes PTC1 and GLI1 are substantially higher in human synovial chondromatosis than in nor- mal synovial tissues. Ptc1 expression in murine syn- ovial chondromatosis was found to be at a level com- parable to that found in growth plate chondrocytes, suggesting that Gli-mediated transcriptional activation is present at levels similar to those found in normal physeal development, and to those found in proliferat- ing growth plate cells. The high incidence of synovial chondromatosis in mice deficient in Gli3 (a repres- sor of hedgehog-induced transcription in limb devel- opment) provides additional evidence that hedgehog transcriptional activation plays a crucial role in their formation. Figure 6. Incidence of synovial chondromatosis in Gli3 mutant mice treated with triparanol. The percentages of Gli3 mutant mice developing synovial chondromatosis at 18 months of age are shown: mice treated with triparanol are labelled ‘ ’ and those treated with control carrier are labelled ‘ ’. There is a substantial decrease in the number of mice developing synovial chondromatosis in the group treated with triparanol. Figure 5. Ptc1 expression in synovial chondromatosis. Immunohistochemistry for Ptc1 shows the expected localization in a wild-type growth plate in the pre-hypertrophic zone (A). Gli3 mutant mice show a slightly larger zone of staining for Ptc1, consistent with lack of the transcriptional repressor Gli3, and thus greater hedgehog-mediated transcriptional activity (B). In Gli3 mutant mice treated with triparanol, the level of Ptc staining is similar to that found in wild-type mice (C). Panel D shows synovial chondromatosis that formed in a Gli3 mutant mouse, demonstrating Ptc1 staining. All are at 300 magnification, except the inset in D, which shows a higher-power (1100 ) view of the lesion. Synovial chondromatosis in Gli3 mutant mice shares nearly identical features with the disease in humans. In both humans and mice, the lesions develop with age and are more common in males. There is a similar pathological progression, with cartilage rests developing from the synovium, which subsequently undergo calcification. Defects in the joint surface are found in both species, associated with advanced pro- gression. The knee is the most common site for occur- rence in humans, and the only site in which the process was identified in mice. The reason for the predomi- nance in the knee is unclear, although the large amount of synovial tissue in the knee, compared with other joints, could be partly responsible for the high inci- dence at this location. Synovial chondromatosis has been hypothesized to arise secondary to loose cartilaginous fragments in joints, with the cartilage fragment acting as a nidus for the development of the chondroma. We found that the degenerative changes in the Gli3 mutant mice occurred adjacent to the synovial lesions, while regions of the articular cartilage in locations remote to the synovial lesions did not show any degenerative changes. This pattern suggests that the development of synovial lesions preceded and caused the degenerative changes in the adjacent articular cartilage. Activation of hedgehog signalling upstream of Gli3 is crucial in the formation of synovial chondromatosis, as triparanol treatment substantially attenuates the rate of formation of the lesions in the extra-toes mouse. Tri- paranol treatment, however, only partially reduced the chance of mice developing synovial chondromatosis. This may be due to only partial blockade of hedgehog signalling, demonstrated by only partial reduction in Ptc1 expression. We utilized a relatively low dose of triparanol, so that side effects, such as cataract for- mation, did not occur. In addition, it is likely that more complete blockade of hedgehog signalling is not compatible with normal development, as hedgehog signalling is required for the normal growth and pat- terning of a variety of tissues in addition to the growth plate. Although a lack of repression of hedgehog tran- scriptional activation predisposes to the development of synovial chondromatosis, it is not required for the formation of synovial chondromatosis, as synovial chondromatosis developed in six of 20 wild-type litter- mates. Interestingly, the Gli3 mutant mice treated with triparanol developed synovial chondromatosis at a rate comparable to that of wild-type mice, suggesting that the effect of triparanol on modulating hedgehog sig- nalling balances the lack of transcriptional repression caused by the relative deficiency of Gli3. Although tri- paranol also has other effects, eg lowering the choles- terol level, which cannot be excluded as a cause of the decreased incidence of synovial chondromatosis, the ability to counteract the effect of Gli3 deficiency suggests that, in this instance, it is acting to decrease hedgehog-mediated signalling. The metaplastic cartilage in synovial chondromato- sis is thought to arise from the synovium [1,41,42]. Cell culture studies have shown that synovial cells can produce cartilage and that treatment with growth factors, such as transforming growth factor beta, may enhance this ability [43]. The location of the cartilage lesions in the mice supports the notion that the lesions arise from the synovial cells. Our data suggest that hedgehog signalling activation plays a crucial role in the ability of synovial cells to undergo metaplasia to cartilage matrix-producing cells. Recently, it was found that transgenic mice express- ing the SV40 T antigen developed synovial chon- dromatosis, as well as cartilaginous metaplasia of other tissues on a C57BL/6J background [44]. It was suggested that rearrangement in the transgene with breeding in the C57BL background altered SV40 T antigen expression, causing the cartilage phenotype. These mice, as well as our mice, are on the C57BL strain, which may be predisposed to developing such cartilaginous phenotypes. Indeed, the C57BL mouse is genetically predisposed to developing cartilaginous metaplasia in the aorta [45]. In addition, strains of rodents often exhibit an increased frequency of disor- ders such as synovial chondromatosis with prolonged inbreeding [46]. Thus, it is likely that the C57BL strain is predisposed to developing synovial chondromatosis due to other genetic factors. Despite the likelihood that the strain of our mice predisposed to the development of synovial chondro- matosis, a crucial role for hedgehog signalling in the pathogenesis of this lesion is provided by the substan- tially greater chance of developing the lesions in the Gli3 mutant mice, and the ability of triparanol treat- ment to decrease this occurrence significantly. This also shows that although Gli3 deficiency predisposes to synovial chondromatosis, it is not required to cause this condition. This finding, and the similar histologi- cal and radiographic appearance to lesions that develop in the Gli3 mutant mice and their wild-type litter- mates, suggests that hedgehog signalling activation plays a role primarily in the formation of the lesions, rather than in the progression of the disease, at least in mice. Our data do not resolve other controversies over the aetiology of the lesion, and the small size and calcified nature of murine synovial chondromato- sis make additional investigations such as clonality studies impractical. The demonstration that hedgehog activation predis- poses to the development of synovial chondromatosis, along with previous studies showing that the benign cartilage tumours, enchondromas and osteochondro- mas, can be caused by mutations dysregulating hedge- hog signalling, implicates hedgehog activation as a common occurrence in cartilaginous lesions. A phar- macological approach, utilizing agents such as tripara- nol, which block hedgehog signalling,MER-29 has the potential to be developed into an effective therapy for such lesions.