Abstract
Fibroblast growth factor receptor 3 (FGFR3) gain-of-function mutations cause dwarfisms, including achondroplasia (ACH) and thanatophoric dysplasia (TD). The constitutive activation of FGFR3 disrupts the normal process of skeletal growth. Bone-growth anomalies have been identified in skeletal ciliopathies, in which primary cilia (PC) function is disrupted. In human ACH and TD, the impact of FGFR3 mutations on PC in growth plate cartilage remains unknown. Here we showed that in chondrocytes from human (ACH, TD) and mouse Fgfr3Y367C/+ cartilage, the constitutively active FGFR3 perturbed PC length and the sorting and trafficking of intraflagellar transport (IFT) 20 to the PC. We demonstrated that inhibiting FGFR3 with FGFR inhibitor, PD173074, rescued both PC length and IFT20 trafficking. We also studied the impact of rapamycin, an inhibitor of mammalian target of rapamycin (mTOR) pathway. Interestingly, mTOR inhibition also rescued PC length and IFT20 trafficking. Together, we provide evidence that the growth plate defects ascribed to FGFR3-related dwarfisms are potentially due to loss of PC function, and these dwarfisms may represent a novel type of skeletal disorders with defective ciliogenesis.
Introduction
Fibroblast growth factor receptors (FGFRs) are a family of five receptor tyrosine kinases and are important regulators of skeleton development during endonchondral and membranous ossification throughout embryonic and postnatal development (1). FGFR3 is expressed in both bone and cartilage (2) and FGFR3 signaling regulates a variety of biologic events during skeletal development. FGFR3 gain-of-function mutations are responsible for a family of chondrodysplasias (3) namely, achondroplasia (ACH) the most common form of dwarfism and a lethal form of dwarfism thanatophoric dysplasia (TD) (4–8). Pathogenic dominant FGFR3 mutations also cause craniosynostosis. Muenke syndrome is the most common craniosynostosis syndrome (9) and Crouzon syndrome associated with acanthosis nigricans is a rare syndrome (10).
FGFR3 plays a significant role in growth plate cartilage, acting to inhibit both the rate of proliferation and the initiation of the chondrocyte hypertrophy. Intricate networks of regulatory proteins spatially govern cartilage growth plate including SRY box 9, Indian hedgehog, parathyroid hormone-like protein, bone morphogenetic protein, Wingless-type MMTV integration site family members (Wnts) (3) and mammalian target of rapamycin (mTOR) (11). The impairment of the FGF signaling disrupts growth development and the columnar arrangement of chondrocytes into the growth plate cartilage thus disrupting the elongation of long bones.
FGFR3-related chondrodysplasias (12,13) share clinical features with skeletal ciliopathies (14). The clinical phenotypes of various ciliopathies such as Jeune syndrome (15), Ellis-van Creveld syndrome (16,17) and short-rib polydactyly (18) include shortened limbs and ribs, polydactyly, and sometimes craniofacial malformations (19,20).
Skeletal ciliopathies arise from defects in the formation, maintenance and function of the primary cilia (PC). The PC is chemosensitive and controls diverse signaling pathways important in cartilage growth plate development and tissue homeostasis (21,22), including hedgehog (23), Wnt/β-catenin (24), Pdgf (25), cAMP/PKA signaling and cell cycle progression (26). PC is also a mechanosensor leading to the downregulation of mTOR pathway necessary for the control of cell size in ciliopathy polycystic disease (27,28).
The biogenesis of the PC and certain signaling events mediated by the PC are controlled by the trafficking of intraflagellar transport (IFT) proteins along the axoneme, a structure of aligned microtubules emanating from the basal body (29,30). PC elongation involves the recruitment of Golgi-derived and IFT20-enriched ciliary vesicles to the basal body region of the PC (31,32). Signaling pathways rely on the trafficking of proteins into or out of the cilium by numerous IFT proteins and motor proteins such as kinesin and dynein (29,33,34). Along the axoneme, IFT-B proteins control anterograde transport and IFT-A proteins control retrograde transport. Inactivation of IFT-A and IFT-B lead to major defects in ciliogenesis (34,35).
To date, the role of PC in FGFR3-related dwarfisms remains unknown. However, given the overlaps in clinical features between FGFR3-related dysplasias and skeletal ciliopathies, we hypothesized that the PC contributes to the defective endochondral ossification process in ACH and TD.
In this study, we evaluated the relationships between FGFR3 gain-of-function mutations, PC morphology and signaling pathways involved in PC and in cartilage homeostasis. We show that the length of PC is significantly smaller in both ACH and TD human fetal chondrocytes and Fgfr3Y367C/+ murine embryonic chondrocytes. We identify an accumulation of IFT20 at the basal body of the PC in human and murine chondrocytes. We also show that inhibiting the tyrosine kinase activity of FGFR3 or mTOR signaling rescue the length of the PC and rescue the localization of IFT20 in chondrocytes. We propose that FGFR3 has an essential role in ciliogenesis, and its constitutive activation disrupts ciliogenesis, which in turn results in defective cartilage growth plate organization and bone development in ACH and TD.
Results
Fibroblast growth Fgfr3 mouse model exhibits phenotypic traits of ciliopathy
Radiographic analyses of skeletons from the FGFR3-related dwarfisms (ACH and TD), show a narrow thorax, short ribs and limbs, brachydactyly, platyspondyly and macrocephaly (4,7,8). In these conditions, bone growth is characterized by an abnormal endochondral ossification process (36). These features overlap with the skeletal malformations previously described for skeletal ciliopathy (14,37) (Table 1). Mouse models of skeletal ciliopathy (based on conditional gene knockouts; Col2a-ift88fl/fl and Col2a-Kif3afl/fl mice) (20,22,38) exhibit a dwarf phenotype with small long bones and ribs and narrow thorax. Similar to what is observed in these skeletal-ciliopathy models, in the Fgfr3Y367C/+mouse model reproducing ACH, the whole skeleton and long bones were smaller size than those of control Fgfr3+/+ mice (Fig. 1A) (39). We evaluated the rib cage volume (normalized to the naso-anal length) in 4 weeks old Fgfr3Y367C/+ and Fgfr3+/+ mice. The volume of the rib cage of Fgfr3Y367C/+mice was significantly lower (−29.9%) than that in Fgfr3+/+ mice (Fig. 1B and C). The organization of growth plate cartilage, which consists of stacked columns of chondrocytes parallel to the longitudinal bone axis in Fgfr3+/+ mice, was disrupted in the Fgfr3Y367C/+mice (Fig. 1D and E). Interestingly, these growth plate anomalies were also observed in various mouse model of ciliopathies (20,22,38), in these conditions, cilia protein or related-cilia protein mutations affected PC formation.
Phenotypic comparison of FGFR3-related chondrodysplasias with others skeletal and craniofacial ciliopathies
| Skeletal phenotype | FGFR3-related chondrodysplasias | Skeletal ciliopathies | Craniofacial ciliopathies |
|---|---|---|---|
| Short–long bones | + | + | |
| Short ribs | + | + | |
| Short chest | + | + | |
| Brachydactyly | + | + | |
| Polydactyly | + | ||
| Cone-shape epiphysis | + | ||
| Hair-nail teeth anomalies | + | ||
| Hyper/hypotelorism | + | ||
| Micrognathia | + | ||
| Cleft lip and/or palate | + | ||
| Craniosynostoses | + | + |
| Skeletal phenotype | FGFR3-related chondrodysplasias | Skeletal ciliopathies | Craniofacial ciliopathies |
|---|---|---|---|
| Short–long bones | + | + | |
| Short ribs | + | + | |
| Short chest | + | + | |
| Brachydactyly | + | + | |
| Polydactyly | + | ||
| Cone-shape epiphysis | + | ||
| Hair-nail teeth anomalies | + | ||
| Hyper/hypotelorism | + | ||
| Micrognathia | + | ||
| Cleft lip and/or palate | + | ||
| Craniosynostoses | + | + |
Phenotypic comparison of FGFR3-related chondrodysplasias with others skeletal and craniofacial ciliopathies
| Skeletal phenotype | FGFR3-related chondrodysplasias | Skeletal ciliopathies | Craniofacial ciliopathies |
|---|---|---|---|
| Short–long bones | + | + | |
| Short ribs | + | + | |
| Short chest | + | + | |
| Brachydactyly | + | + | |
| Polydactyly | + | ||
| Cone-shape epiphysis | + | ||
| Hair-nail teeth anomalies | + | ||
| Hyper/hypotelorism | + | ||
| Micrognathia | + | ||
| Cleft lip and/or palate | + | ||
| Craniosynostoses | + | + |
| Skeletal phenotype | FGFR3-related chondrodysplasias | Skeletal ciliopathies | Craniofacial ciliopathies |
|---|---|---|---|
| Short–long bones | + | + | |
| Short ribs | + | + | |
| Short chest | + | + | |
| Brachydactyly | + | + | |
| Polydactyly | + | ||
| Cone-shape epiphysis | + | ||
| Hair-nail teeth anomalies | + | ||
| Hyper/hypotelorism | + | ||
| Micrognathia | + | ||
| Cleft lip and/or palate | + | ||
| Craniosynostoses | + | + |
Fgfr3 gain-of-function mutation is associated with a ciliogenesis defect in the growth plate. (A) Radiographs of Fgfr3+/+ and Fgfr3Y367C/+ mice (4 weeks old). (B) Representative 3D-reconstruction of rib cages from Fgfr3+/+ and Fgfr3Y367C/+ mouse (4 weeks old). Scale bar: 1 cm. (C) Graph showing rib-cage 3D volume from Fgfr3+/+ (n = 4) and Fgfr3Y367C/+ (n = 3) mice. (D) Hematoxylin–eosin (H&E) staining of growth plate from Fgfr3+/+ and Fgfr3Y367C/+ mouse (E16.5), black arrow showing the absence of columns in growth plate. Scale bar: 60 µm. HY, hypertrophic zone. (E) Model of growth plate cartilage in Fgfr3-related disorders. In control Fgfr3+/+ growth plate, the chondrocytes in the proliferative and hypertrophic zones form stacked columns of cells. In the Fgfr3Y367C/+ growth plate, the chondrocyte division misalignments result in the failure to form columns. HY, hypertrophic zone. (F) PC immunolabeling in growth plate chondrocytes of Fgfr3+/+ and Fgfr3Y367C/+ mice (5 days old). Confocal analysis of γ-tubulin (green), Arl13b (red) and nuclei with DAPI (blue) in frozen sections. These data are representative of three independent experiments. Scale bar: 20 µm. (G) Height of collagen type X (Col X) zone in the growth plates of Fgfr3+/+ and Fgfr3Y367C/+ mice (5 days old). These data are representative of five independent experiments. (H) Two-dimensional-projected PC length in growth plate chondrocytes of Fgfr3+/+ mice in collagen type X negative areas (n=1323) and Fgfr3Y367C/+ mice (n=1119) (5 days old). Data represent mean±S.E.M. Two-tailed unpaired t-test. ns, not significant, *p<0.05, **p<0.01, ****p<0.0001.
Thus, we investigated whether Fgfr3 mutations in the growth plate chondrocytes from mice on postnatal day 5, affect PC formation. PC was detected by immunolabeling of frozen growth plate sections with anti-γ-tubulin and anti-Arl13b antibodies as markers of the basal body and axoneme, respectively (Fig. 1F). In Fgfr3+/+ mice, the longitudinal axis of the PC aligned with the columnar arrangement of proliferative chondrocytes (Fig. 1F). Remarkably, in the Fgfr3Y367C/+ mice, PC positioning was not parallel to the longitudinal axis of the growth plate cartilage thus suggesting that PC has disturbed the chondrocyte rotation and the organization of the proliferative chondrocytes in the cartilage (Fig. 1F).
The hypertrophic zone of the growth plate cartilage can be distinguished from the proliferative zone by immunolabeling of collagen type X (postnatal day 5). The hypertrophic zone in the growth plates of Fgfr3Y367C/+ mice were smaller compared with Fgfr3+/+ mice (Fig. 1H), as reported previously (3). In the proliferative zone (collagen type X negative zone), the mean length of PC was 1.13 ± 0.01 µm in Fgfr3Y367C/+ chondrocytes and was marginally smaller (by 6%) than that in Fgfr3+/+ chondrocytes (1.20 ± 0.01 µm; Fig. 1I). These results indicate that dwarfism due to FGFR3-gain-of-function mutations was associated with shorter PC in the growth plate at postnatal day 5.
Together these data suggest that defective ciliogenesis is responsible, in part of the alteration of the chondrocyte orientation, the disorganization of the growth plate architecture and finally the impairment of the long bone elongation.
The length of the primary cilium in the chondrocyte is controlled by FGFR3
To confirm the defective ciliogenesis observed in the growth plate cartilage tissue in Fgfr3Y367C/+ mice, we examined the PC in the proliferative chondrocytes. Differences in PC lengths associated with FGFR3 gain-of-functions mutations were also observed in primary cultures of mouse and human fetal proliferative chondrocytes, and in immortalized fetal human chondrocytes (40). In primary cultures of mouse fetal chondrocytes (E16.5), the proportion of ciliated cells from Fgfr3Y367C/+ mice (87.4 ± 2.2%, n = 4) was similar to those from Fgfr3+/+ mice (92.6 ± 3.7%, n = 4) (Fig. 2A and B). The mean PC length was smaller in Fgfr3Y367C/+ chondrocytes (2.46 ± 0.03 µm) compared with Fgfr3+/+ chondrocytes (2.82 ± 0.05 µm, n > 700) (Fig. 2C and D). Similarly, in primary cultures of human ACH and TD chondrocytes, the proportions of ciliated cells were comparable to that in control human chondrocytes (ACH, 94.0 ± 2.6%; TD, 95.0 ± 0.6%; control, 96.3 ± 0.9%, all n = 3) (Fig. 2E and F;Supplementary Material, Table S1). As with the mouse chondrocytes, the mean PC lengths were smaller by 20% in human ACH chondrocytes (3.14 ± 0.05 µm, n > 500), and by 22% in human TD chondrocytes (3.45 ± 0.06 µm, n > 500) (Fig. 2G and H;Supplementary Material, Table S1) in comparison with human control chondrocytes (3.91 ± 0.05 µm, n > 500). Shorter PC lengths were also observed in human immortalized ACH and TD chondrocyte cell lines in comparison with human control chondrocyte cell lines (Supplementary Material, Fig. S1A and B).
Fgfr3 gain-of-function mutation is associated with a ciliogenesis defect in chondrocytes from the Fgfr3Y367C/+ mouse, and ACH and TD tissue. (A) Representative image of confocal analysis of Arl13b (red), γ-tubulin (green), DAPI (blue) in ciliated chondrocytes in Fgfr3+/+ and Fgfr3Y367C/+ mice. (B) Graphical representation of the number of ciliated chondrocytes in Fgfr3+/+ and Fgfr3Y367C/+ mice. These data are representative of five independent experiments. (C) Representative image of confocal analysis of Arl13b (red), acetylated α-tubulin (orange), γ-tubulin (green), DAPI (blue) in PC in Fgfr3+/+ and Fgfr3Y367C/+ mouse chondrocytes. (D) Graphical representation of the PC length of Fgfr3Y367C/+ and Fgfr3+/+ chondrocytes (n>700). (E) Representative image of confocal analysis of Arl13b (red), γ-tubulin (green), DAPI (blue) in human ACH and TD ciliated chondrocytes. (F) Graphical representation of the number of ciliated cells in human control (nos 6–8; see Supplementary Material, Table S1), ACH (nos 9–11; see Supplementary Material, Table S1) and TD (nos 12, 13, 15 and 16; see Supplementary Material, Table S1). These data are representative of three independent experiments. (G) Representative images of confocal analysis of Arl13b (red), acetylated α-tubulin (orange), γ-tubulin (green), DAPI (blue) in PC in human control, ACH and TD chondrocytes. (H) Graphical representation of length of PC in human control (nos 6–8; see Supplementary Material, Table S1), ACH (nos 9–11; see Supplementary Material, Table S1) and TD chondrocytes (nos12, 13, 15 and 16; see Supplementary Material, Table S1) (n>700). Scale bar: 1 µm. (I) Specific ciliary localization motifs observed in FGFR3 protein sequence are 4 VxPx motifs (yellow), 2 FR motifs (green) and 3 FK (blue) motifs, where ‘x’ represents any amino acid. (J) Confocal analysis of Arl13b (red), Fgfr3 (green), DAPI (blue) in PC from Fgfr3+/+ and Fgfr3Y367C/+ mice, the data are representative of 5 independent experiments. (K) Representative quantification of FGFR3-positive PC in mouse chondrocytes (n>150). Data represent mean±S.E.M. Two-tailed unpaired t-test. ns, not significant; **p<0.01; ****p<0.0001. n is the number of PC analyzed (see also Supplementary Material, Fig. S1 and Table S1).
Given that our results indicated that the PC length is reduced in FGFR3-related conditions, we next examined whether FGFR3 protein is localized or not at the basal body or in the axoneme of the PC. In the FGFR3 protein sequence, we identified 9 putative motifs for ciliary localization (Fig. 2I) such as VxPx, FR and FK motifs, where ‘x’ represents any amino acid (41,42) and found by immunolabeling that the FGFR3 protein is colocalized with Arl13b in the axoneme of the PC, irrespective of genetic background in mouse and human chondrocytes (i.e. in control or FGFR3 gain-of-function mutations) (Fig. 2J and K).
Taken together, these results show that FGFR3 gain-of-function mutations in both Fgfr3 mouse model and ACH and TD human might have a role in the regulation of PC length in the chondrocytes.
The short-primary-cilium defect in mouse Fgfr3Y367C/+ and human ACH primary chondrocytes is rescued by a tyrosine-kinase inhibitor
Since the Fgfr3 mutations lead to PC length defect and FGFR3 protein is present in PC of mouse and human chondrocytes, we used pharmacologic approach as a second method to test whether the shortened length of the PC is directly due to FGFR3 constitutive-kinase activity. Mouse femur cultures and mouse and human primary chondrocytes were treated with a FGFR-specific tyrosine-kinase inhibitor (TKI; PD173074). The efficacy of PD173074 on FGFR3 signaling was confirmed in mouse femurs (postnatal day 5) and primary cultures of Fgfr3Y367C/+ chondrocytes (E16.5), because PD173074 treatments (100 nM) rescued the size of Fgfr3Y367C/+ femurs (data not shown), the growth plate cartilage organization and abrogated the FGF2-induced phosphorylation of the two effector kinases (Erk1–2) of the MAP-kinase pathway, which is the canonical FGFR3 signaling pathway (Supplementary Material, Fig. S2A–C).
To test the hypothesis that the constitutive activation of FGFR3 signaling pathway play a critical role in ciliogenesis, we analyzed the effect of PD173074 (100 nM) on PC length by using in vitro chondrocyte cultures isolated from embryonic rib cages (E16.5). In control primary chondrocytes from Fgfr3+/+ mice, PD173074 treatment had no effect on PC length (Fig. 3A and B). By contrast, in chondrocytes from Fgfr3Y367C/+mice, PD173074 treatment significantly rescued PC length to 96% of the length observed in PC from Fgfr3+/+ chondrocytes (Fig. 3A and B). As expected, the number of ciliated cells was not modified with PD173074 treatment in Fgfr3Y367C/+ or Fgfr3+/+ primary chondrocytes (data not shown). A similar effect of PD173074 treatment was observed in human fetal chondrocytes. PD173074 treatment had no effect on PC length in human control fetal chondrocytes, whereas PD173074 treatment significantly rescued PC length in human ACH chondrocytes to 91% of that observed in control chondrocytes, and in human fetal TD chondrocytes to 94% of that observed in control chondrocytes (Fig. 3C and D).
The FGFR-specific tyrosine-kinase inhibitor PD173074 corrects the primary-cilium length defect and growth plate anomalies. (A) Representative image of confocal analysis of Arl13b (red), acetylated α-tubulin (orange), γ-tubulin (green), DAPI (blue) of the increased length of PC in Fgfr3Y367C/+ mouse chondrocytes after PD173074 treatment. Scale bar: 1 µm. (B) Graphical representation of length of PC in Fgfr3Y367C/+ and Fgfr3+/+ chondrocytes with and without PD173074 treatment (n>300). (C) Representative image of confocal analysis of Arl13b (red), acetylated α-tubulin (orange), γ-tubulin (green), DAPI (blue) of the length of PC in human ACH, TD and control chondrocytes treated with PD173074. Scale bar: 1 µm. (D) Graphical representation of PC length in human control (nos 6–8; see Supplementary Material, Table S1), ACH (nos 9–11; see Supplementary Material, Table S1) and TD chondrocytes (nos 12, 13, 15 and 16; see Supplementary Material, Table S1) treated with PD173074 (nos 6 and 7; ACH: nos 9 and 10; TD: nos 12, 13 and 15; see Supplementary Material, Table S1) (n>200). Data represent mean±S.E.M., two-tailed unpaired t-test. ns, not significant; *p<0.05, ****p<0.0001. n is the number of PC analyzed (see also Supplementary Material, Figs S2 and S3).
To confirm that FGFR3 plays a critical role in ciliogenesis, the measurements of the PC length were performed using chondrocyte cultures isolated from Fgfr3 knockout mouse model (Fgfr3 KO). These analyses revealed that the number and the length of the Fgfr3 KO PC (2.78 ± 0.08 µm, n = 117) were similar to Fgfr3+/+PC (2.79 ± 0.04 µm, n = 472; Supplementary Material, Fig. S3). These data indicate that the absence of FGFR3 protein did not impair the ciliogenesis.
Altogether, these results confirm that FGFR3 constitutive kinase activity disrupts the length of PC in mouse and human fetal chondrocytes. These alterations in the length of the PC strongly suggest a mechanism in which FGFR3 signaling affects the elongation process of the PC in FGFR3-related dwarfisms.
Constitutively active FGFR3 affects ciliogenesis by disrupting the trafficking of IFT20-enriched vesicles
Ciliogenesis can be promoted by inhibition of actin polymerization (43–45). Cytochalasin D, an inhibitor of actin polymerization promotes ciliogenesis and increases the ciliary recruitment of specific vesicles containing IFT proteins (43,46). We thus examined the effect of cytochalasin D in the primary cultures of mouse chondrocytes. Cytochalasin D treatment increased PC length by a greater degree in Fgfr3Y367C/+ chondrocytes (by 80%) than in Fgfr3+/+ chondrocytes (by 28%) (Fig. 4A and B). Moreover, cytochalasin D treatment similarly increased PC length in both primary cultures of ACH and TD human chondrocytes (by 36%, ACH versus 9%, control; and by 29%, TD versus 9%, control) (Fig. 4C and D). This suggested that trafficking to the PC of vesicles containing IFT proteins might be partially inhibited in chondrocytes with FGFR3 gain-of-function mutations.
Constitutively active FGFR3 disrupts the localization of IFT20 into the primary cilium. (A) Representative images of confocal analysis of Arl13b (red), acetylated α-tubulin (orange), γ-tubulin (green), DAPI (blue) of PC length in Fgfr3+/+ and in Fgfr3Y367C/+ mouse primary chondrocytes after cytochalasin D treatment. Scale bar: 1 µm. (B) Graphical representation of PC length in Fgfr3+/+ and in Fgfr3Y367C/+ mouse primary chondrocytes after cytochalasin D treatment (n>100). (C) Representative images of confocal analysis of Arl13b (red), acetylated α-tubulin (orange), γ-tubulin (green), DAPI (blue) of PC length in control, human ACH and TD chondrocytes after cytochalasin D treatment. Scale bar: 1 µm. (D) Graphical representation of PC length in control (nos 6–8; see Supplementary Material, Table S1), human ACH (nos 9–11; see Supplementary Material, Table S1) and TD chondrocytes (nos 12, 13, 15 and 16; see Supplementary Material, Table S1) after cytochalasin D treatment (n>100). (E) Representative images of the localization of IFT20 (green) in Fgfr3+/+ and in Fgfr3Y367C/+ mouse primary chondrocytes analyzed by high-resolution STimulated Emission Depletion (STED) nanoscopy. PC was immunodetected with acetylated α-tubulin (red). (F) Quantification of IFT20 area per cilia at the basal body of mouse PC in Fgfr3+/+ chondrocytes (n=11), Fgfr3Y367C/+ chondrocytes (n=18) Fgfr3Y367C/+ chondrocytes (n=12) after PD173074 treatment. (G) Representative STED-nanoscopy images of the localization of IFT20 (green) in human TD primary chondrocytes with and without PD173074 treatment. PC was immunodetected with acetylated α-tubulin (red). (H) Quantification of recovery surface of IFT20 at the basal body of human TD PC with (n = 14) and without (n = 15) PD173074 treatment (nos 14 and 17; see Supplementary Material, Table S1). Data represent mean±S.E.M., two-tailed unpaired t-test. ns, not significant; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; n is the number of PC analyzed (see also Supplementary Material, Figs S4 and S5).
Therefore, we examined major components of PC formation including IFT-B complex: IFT88, which is involved in actin cytoskeletal organization in chondrocytes (47); IFT74, which is required to stabilize the IFT-B complex (48); and IFT20, the smallest IFT-B protein (33) which contributes to trafficking between the Golgi apparatus and the cilium-basal body (49). Confocal microscopy was used to detect IFT88 and IFT74, and in Fgfr3+/+ chondrocytes and Fgfr3Y367C/+ chondrocytes, IFT88 and IFT74 were consistently localized to the cilium axoneme (Supplementary Material, Fig. S4A and B). High-resolution STimulated Emission Depletion (STED) nanoscopy was used to detect IFT20 (and associated vesicles of 50 nm range). In Fgfr3Y367C/+ chondrocytes, IFT20 was localized in punctate structures (assumed to be membrane-associated structures) proximal to the cilium-basal body, whereas in Fgfr3+/+ chondrocytes, IFT20 was localized primarily in the cilium axoneme. By quantifying the membrane-surface-associated levels of IFT20, the amount of IFT20 was 2.2-fold greater in punctate structures proximal to the basal bodies of the PC in Fgfr3Y367C/+ chondrocytes than the amount observed in Fgfr3+/+ chondrocytes (Fig. 4E and F). In human fetal TD chondrocytes, an important amount of IFT20 in punctate structures was also observed proximal to the basal bodies of the PC (Fig. 4G and H). PD173074 treatment of Fgfr3Y367C/+ mouse (E16.5) and human TD chondrocytes rescued localization of IFT20 to the axoneme and lowered the accumulation of IFT20 punctate structures proximal to the basal body of PC by 2.3-fold in Fgfr3Y367C/+ mouse chondrocytes and by 7.5-fold in human fetal TD chondrocytes compared with the respective controls (Fig. 4G and H).
It has been previously shown that IFT20 functions at the Golgi apparatus to regulate PC length (50,51) and that constitutive-activity of FGFR3 can be detected at the Golgi apparatus (52). We confirmed here that IFT20 was colocalized with GM130, a marker of cis-Golgi apparatus in Fgfr3Y367C/+ chondrocytes (Supplementary Material, Fig. S5). Taken together, these data suggested that the punctate structures proximal to the basal body of the PC included the Golgi apparatus and vesicles proximal to the Golgi apparatus. Hence constitutively active FGFR3 may have disrupted the sorting and trafficking of IFT20-enriched vesicles from the Golgi apparatus to the PC.
mTOR regulates ciliogenesis in FGFR3-related disorders
Ciliogenesis is controlled by autophagy (53) and the induction of autophagy is suppressed by the mTOR pathway, a sensor of cellular metabolism. Hyperactivation of the mTOR pathway has been shown to inhibit chondrocyte differentiation and results in chondrodysplasia in mice (11), and autophagy is inhibited by constitutively active FGFR3 in a mouse model of ACH (54). Given these relationships, we investigated the role of mTOR pathway in our Fgfr3 cell-culture models.
With respect to ciliogenesis, treatment with the mTOR inhibitor rapamycin rescued PC length in Fgfr3Y367C/+ chondrocytes (to 92% of that observed in Fgfr3+/+ chondrocytes) (Fig. 5A and B) and in human TD chondrocytes (to 90% of that observed in human control chondrocytes) (Fig. 5C and D). Rapamycin treatment did not affect PC length in Fgfr3+/+ chondrocytes also suggesting that mTOR pathway activity is associated with the constitutive activity of FGFR3 (Fig. 5A and B). We next wondered whether IFT20 localization was modified under inhibition of mTOR signaling. Interestingly, similar to what we previously observed with PD173074 treatment, we found that rapamycin treatment also rescued the localization of IFT20 to the axoneme and lowered the accumulation of IFT20 in the proximity of the basal-body in Fgfr3Y367C/+ chondrocytes (by 2.9-fold) (Fig. 5E and F). Overall, these results suggested the short-PC defect associated with constitutively active FGFR3 is mediated by increased activity in the mTOR pathway.
mTOR inhibitor rapamycin corrects the primary-cilium length defect. (A) Representative image of PC length in Fgfr3Y367C/+ mouse chondrocytes after rapamycin treatment. Confocal analysis of PC with γ-tubulin (green), acetylated α-tubulin (orange) and nuclei with DAPI (blue) in Fgfr3+/+ and Fgfr3Y367C/+ mouse chondrocytes. Scale bar: 1 µm. (B) Graphical representation of PC length in Fgfr3+/+ and Fgfr3Y367C/+ chondrocytes after rapamycin treatment (n>100). (C) Representative image of the length of PC in human TD chondrocytes after rapamycin treatment. Confocal analysis of PC with γ-tubulin (green), acetylated α-tubulin (orange) and nuclei with DAPI (blue) in TD chondrocytes. Scale bar: 1 µm. (D) Graphical representation of length of PC in human TD chondrocytes after rapamycin treatment (no 18 and 19; see Supplementary Material, Table S1) (n>100). (E) Representative images of the localization of IFT20 (green) in Fgfr3+/+ and in Fgfr3Y367C/+ mouse primary chondrocytes analyzed by high-resolution STimulated Emission Depletion (STED) nanoscopy. PC was immunodetected with acetylated α-tubulin (red). (F) Quantification of recovery surface of IFT20 at the basal body of Fgfr3Y367C/+ mouse PC with (n=32) and without (n=15) rapamycin treatment (nos 18 and 19; see Supplementary Material, Table S1). Scale bar: 1 µm. Data represent mean±S.E.M., two-tailed unpaired t-test. ns, not significant, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
Taken together, these data suggest a model in which the delivery of IFT20-enriched vesicles to the PC is also mediated by mTOR signaling pathway. These results identify mTOR signaling as one of the regulator of ciliogenesis in FGFR3-related disorders.
Discussion
Longitudinal bone growth is severely affected in ACH (55). FGFR3 gain-of-function mutations affect the proliferation and differentiation of chondrocytes resulting in a large disorganization of the growth plate cartilage and severe impairment of longitudinal-bone growth. Bone-growth anomalies have been identified in skeletal ciliopathies, in which PC function is disrupted (56). Many skeletal ciliopathies arise from mutations in genes encoding for proteins involved in IFT complex (57–61). Mouse models of these skeletal ciliopathies display abnormal bone growth. In the cartilage, the PC formation and orientation are disrupted, the organization of the growth plate is altered and the chondrocyte differentiation is abnormal (20). Today, the precise role of IFT and PC in skeletal disorders remains unclear.
Here, we report that constitutive activation of FGFR3 disrupt PC length in fetal ACH and TD human and Fgfr3Y367C/+ mouse chondrocytes in vitro, and in mouse Fgfr3Y367C/+ growth plate cartilage ex vivo. Our data show that the PC defect in FGFR3-related dwarfisms impact the organization of the chondrocyte columns, the chondrocyte orientation and finally the long bone elongation. These growth plate cartilage anomalies observed in achondroplasia and TD resemble to a wide variety of skeletal phenotypes due to mutations in ciliary proteins (Table 1). It is well known that PC is involved in the development and function of the cartilage and bone, but how PC regulates cellular events is just beginning to be understood. Hedgehog (HH) signaling appears to be essential for PC structure and function as disruption of HH pathways leads to widespread ciliopathy. However, defects in HH signaling cannot account for all phenotypes (20). Others signaling pathways are involved in skeletal development but it remains unknown how these pathways are regulated by PC in cartilage cells. Here, our data show that the activation of FGF pathway is involved in the regulation of PC elongation in a family of FGFR3-related dwarfism.
Our study extends other observations that show that FGFR signaling (FGFR1 and FGFR2) can regulate PC length (62). In these specific models, knock-down of FGFR1 shortened PC length in Xenopus epithelial cells in the otic vesicles and pronephric ducts (62), and knock-down of FGFR2 shortened PC length in epithelial cells in zebrafish embryos associated with the establishment of left–right asymmetry (63).
Our study provides insights on how constitutively active FGFR3 plays a role in disrupting PC elongation. Based upon the fact that IFT20 are found in both secretory vesicles and PC, we demonstrated that the short-PC defect was associated with a loss of IFT20-positive vesicles in the axoneme of the PC and the accumulation IFT20-positive vesicles at the base of the PC or retention of IFT20 in the Golgi apparatus. The short-PC defect and mislocalization of IFT20 were rescued by the FGFR3 TKI (PD173074). We therefore propose that FGFR3 pathway plays a potential direct role of FGFR3 related to disrupting the correct function of IFT20. Constitutively active FGFR3 was detected in the PC axoneme and has been suggested to have tyrosine-kinase activity in the Golgi (52), which would correspond the sites where the IFT20 trafficking appeared to be disrupted. Interestingly, it was recently demonstrated that, in addition to regulating PC elongation, IFT20 may also regulate the intracellular trafficking of collagen in cranial neural crest cells and thus play a role in craniofacial skeletal development (64). An important remaining question is whether IFT20 trafficking is also affected in craniofacial ciliopathies (Table 1).
We also investigated the role of the mTOR pathway in the FGFR3-related dwarfisms. The implication of mTOR in the cartilage growth plate has recently started to emerge. Recent study demonstrated that mTORC1 promotes early stage chondrocyte proliferation and prevents later stage terminal differentiation (11). mTOR, a well-known regulator of the cell growth, cell volume, cell proliferation and cell migration, is implicated in the ciliogenesis (27,28). In our study, the mTOR pathway inhibitor rapamycin also rescued the short-PC defect and mislocalization of IFT20 in Fgfr3Y367C/+ mouse and fetal TD human chondrocytes. Therefore, our results suggested a model in which the short-PC defect associated with constitutively active FGFR3 is mediated by mTOR pathway activity and defective trafficking of IFT20 from the cis-Golgi to the PC.
We found that the inhibition of mTOR restores the ciliogenesis in FGFR3-related chondrodysplasias. In the kidney deletion of the PC by knock-out of Kif3a caused cell enlargement as a result of abnormal upregulation of the mTORC1 pathway (27,65). Interestingly, mice treated with rapamycin showed a reduced number of proliferating and prehypertrophic chondrocytes (11), our data suggest that the activation of mTOR in FGFR3-related disorders could participate to the defective chondrocyte differentiation in the growth plate cartilage. This suggested that the heightened activity in the mTOR pathway was an indicator of the disrupted terminal chondrocyte differentiation. Whether this heightened mTOR activity inhibited autophagy-related processes that regulate PC-elongation in these chondrocytes (66–69) remains to be investigated. The plausibility of such a mechanism is suggested from observations which show that the targeted genetic ablations in chondrocytes of autophagy-related genes, either Atg5 or Atg7, suppress autophagy in the cartilage, resulting in mild growth retardation (70,71), and that constitutively active FGFR3 and FGF have been shown to inhibit autophagy in growth plate chondrocytes (54,72). Rapamycin treatment also suppressed the effect of constitutively active FGFR3 on the anterograde IFT velocity of IFT20. Our data suggest that this effect of FGFR3 gain-of-function on PC length and on IFT20 could be mediated by mTOR activity.
During the skeletal development, many developmental signaling pathways including Wnt, TGF-β, Notch, mTOR and FGF may also depend on PC. The spectrum of signaling pathways interfering with FGFR3 need to be investigated further to understand this impairment of the ciliogenesis.
In conclusion, we provide evidence that the over-activation of FGFR3 disturbs the formation of the PC and the trafficking of cilia-related protein, in particular IFT20. Therefore, our results support the hypothesis that human ACH and TD are skeletal disorders with a defective ciliogenesis and the PC-defects can explain certain overlaps in the clinical features with skeletal ciliopathies.
Moreover, by considering ACH as a new type of ‘skeletal ciliopathy’, the scope should be widened for the development of new therapies for ACH and others FGFR3-related chondrodysplasias.
Materials and Methods
Animals
Details of the Fgfr3Y367C/+ and Fgfr3fl/fl (KO) mouse models have been described previously (39). All mice were on a C57BL/6 strain background. Chondrocyte cultures and cartilage growth plate analyses were performed in embryonic E16.5 male and female mice. Genomic DNA was isolated from tail tips by proteinase-K digestion and extracted with NucleoSpin® Tissue according to the manufacturer’s instructions (Macherey-Nagel, Germany). PCR was performed using primers as previously described (39). Experimental animal procedures and protocols were approved by the French Animal Care and Use Committee.
X-radiography and computed tomography scanner and 3D reconstruction
Radiography was performed on whole mice to reveal skeletons and ribs (Faxitron). Miro-computed tomography (µCT) analyses were performed using Quantum FX instrument (PerkinElmer) and the following settings: integration time: 2 min; 90 E(kVp); 160 µA; (Field of View) FOV24 (50 µm) and FOV73 (148 µm). The 3D volume of a rib cage was reconstructed by using Osirix® to convert files into .obj format followed by Rhinocéros® to draw and reconstruct the form of the rib cage and thereby determine its volume.
PC elongation
Optimal serum deprivation for PC elongation was determined as a function of the maximal length attainable without cell death. This was determined independently in primary mouse, human and immortalized chondrocytes and was determined as 24, 48 and 72 h, respectively.
Human primary and immortalized cells
Human primary and immortalized chondrocytes and mouse chondrocytes were obtained and collected as previously described (36,40). Primary human chondrocytes used in the experiments expressed heterozygous p.Gly380Arg, p.Arg248Cyst, p.Ser249Cyst, pTyr373Cys or pGly370Cys FGFR3 mutations (Supplementary Material, Table S1).
Immunohistologic analyses of mouse explants
Fetal femur (E16.5) explants were fixed in 4% paraformaldehyde, decalcified with EDTA 0.4 M and embedded in paraffin as previously described (73). Serial sections of 5 µm were stained with hematoxylin–eosin using standard protocols Images were captured with an Olympus PD70-IX2-UCB microscope (Olympus, Tokyo, Japan). Postnatal femurs (P5) were fixed in MeOH at 4 °C (chilled at −20 °C) for 5 h. After incubation in EDTA 0.5 M, pH 8 for 72 h, femurs were placed in sucrose 30% for 24 h, then in optimum cutting temperature compound at room temperature and frozen in isopentane −45 °C. Fifty micrometers of tissue sections were permeabilized with Triton X-100 0.3% for 30 min and immunolabeled with rabbit anti-Arl13b (Proteintech #17711-1-AP, IF 1:100), mouse IgG1 anti-γ-tubulin (Sigma-Aldrich #T6557, 1:100) primary antibodies. Primary antibodies were detected with the following secondary antibodies: goat anti-mouse IgG1 coupled with AlexaFluor 488 (LifeTechnologies, 1:1000); goat anti-mouse IgG2b coupled with AlexaFluor 568 (LifeTechnologies, 1:1000) and anti-rabbit coupled with AlexaFluor 647 (LifeTechnologies, 1:1000). Tissue sections were mounted with DAPI-Fluoromount G and analyzed under spinning disk microscope.
Immunocytochemistry
Cultured chondrocytes were fixed at room temperature for 10 min in methanol (chilled at −20 °C), and then washed with PBS. Samples were permeabilized for 10 min with PBS containing 0.1% Triton-X100 (Sigma-Aldrich), then washed three times for 5 min. Samples were incubated with PBS containing 10% goat serum (Biowest) for 1 h at room temperature. Primary antibodies were incubated at 4 °C overnight. The following primary antibodies were used: mouse IgG1 anti-γ-tubulin (Sigma-Aldrich #T6557, 1:1000), mouse IgG2b anti-acetylated α-tubulin (Sigma-Aldrich #T6793, 1:2000), rabbit anti-Arl13b (Proteintech #17711-1-AP, 1:500), rabbit anti-IFT20 (Proteintech #13615-1-AP, 1:500), rabbit anti-IFT88 (Proteintech #13967-1-AP, 1:200) and goat anti-IFT74 (Abnova #PAB6611, 1:200). Cells were washed with PBS, and then incubated with goat anti-mouse IgG1 coupled with AlexaFluor 488 (Life Technologies, 1:1000); goat anti-rabbit Alexa Fluor 532 (Life Technologies, 1:100), goat anti-mouse IgG2b coupled with AlexaFluor 568 (Life Technologies, 1:1000), donkey anti-goat AlexaFluor 555 (Life Technologies, 1:100) and/or anti-rabbit coupled with AlexaFluor 647 (Life Technologies, 1:1000) secondary antibodies for 2 h at room temperature in the dark. Samples were extensively washed and mounted with a solution of DAPI-Fluoromount-G® (SouthernBiotech) containing DAPI (4′,6′-diamidino-2-phenylindole) for nuclear staining.
Image acquisition and analysis
Images were captured using a Zeiss LSM700 confocal microscope equipped with a 63× 1.4 numerical aperture oil objective. In order to compare measured data, all confocal experiments showing PC length was acquired in same conditions using slice thickness of 0.33 μm and a pixel size of 60 nm. Post-treatment analyses were performed with Imaris v8.3 software and FIJI (Fiji Is Just ImageJ; NIH) using FigureJ plugin.
Spinning disk confocal microscopy
Three-dimensional images of growth plate were obtained using a spinning disc confocal. The system is composed by a Yokogawa CSU-X1 spinning disk scanner coupled to a Zeiss Observer Z1 inverted microscope and controlled by Zen Blue software. Tile images were acquired with a Plan Apochromat 63× oil immersion objective (NA 1.46) through a Hamamatsu Orca Flash 4.0 sCMOS Camera.
Time gated continuous wave STED nanoscopy
Image acquisitions were performed with a 100× oil immersion objective (NA 1.4) through gCW STED imaging (TCS SP8-3X; Leica Microsystems) with optimized parameters for Alexa Fluor 532 and 568 detection. Samples (zoom 4, pixel size = 14 nm) were excited sequentially with a 575- and a 528-nm wavelength of a supercontinuum laser and a 660-nm laser for depletion. For Alexa Fluor 568, 20% AOTF, conventional scanner (400 Hz, Line Average 2, Accumulation 4) and 100% of depletion lasers were used. Fluorescence (585–640 nm) was collected with a hybrid detector (Gain 30%) in the gated mode (1–6 ns) and a pinhole for 1 Airy Unit. For Alexa Fluor 532, 50% AOTF, conventional scanner (400 Hz, Line Average 2, Accumulation 4) and 100% of depletion lasers were used. Fluorescence (538–570 nm) was collected with a hybrid detector (Gain 30%) in the gated mode (1.5–6 ns) and a pinhole for 1 Airy Unit. Deconvolution of raw data from STED imaging was obtained through image processing with Huygens professional 4.5.1 software. Recovery surface analysis of IFT20 area was performed cell-by-cell using ImageJ software. Image acquisition and image analysis were performed at Institute Imaging Facility.
Imaging post-treatment analysis
The PC lengths were analyzed using two different methods based on confocal Z-slices. Steps size was established to 0.2 µm to abolish measure errors. The first method was used for primary and immortalized cultured cell to measure PC in 3D by Imaris v8.3 software (Bitplane). Serum deprivation allowed a maximal PC-stimulated elongation. This method allowed to normalized genotypes and offer precise value of PC lengths at a time given.
The second method was used to measure PC lengths in cartilage tissue. Due to the absence of PC-stimulated elongation and possibilities of non-specific staining in cartilage tissue, we performed 30 images per Z-stack and realized a Z-projection with maximal intensity. We next measured in 2D the PC lengths by IMARIS software (Bitplane). This method allowed us to measure at least 90% of PC in each analyzed growth plate.
In both cases, we double-checked PC length through staining with Arl13b and α-acetylated tubulin.
Protein analysis
Whole-chondrocyte lysates were prepared using RIPA buffer supplemented with protease-inhibitor cocktail tablets (cOmplete Mini, EDTA-free, Roche). Proteins were incubated and gently mixed on wheel-type tube rotator at 11 rpm for 2 h at 4 °C. Proteins were isolated by centrifugation at 11 000 rpm for 20 min at 4 °C, then denatured with β-mercaptoethanol 2.5% at 95 °C 10 min. Extracted proteins were subjected to NuPAGE 4–12% (Life Technologies) or 15% bis–tris acrylamide gels (Life Technologies), and transferred onto PVDF membranes (Amersham). Using standard protocols, blots were probed with primary antibodies such as rabbit anti-phospho-Erk1–2 (Cell Signaling #4370, 1:1000), mouse anti-total Erk1–2 (Cell Signaling #9107, 1:1000) and mouse anti-actin (Millipore, MAB1501, 1:10 000) antibodies. Proteins of interest were detected with horseradish peroxidase (HRP)-conjugated and visualized with the ECL (GE Healthcare) or with goat anti-mouse IRDye® 680RD (925-68070) and goat anti-rabbit IRDye® 800CW (925-32211) when Li-Cor system was used. The obtained immunoreactivities were determined and calculated using ImageJ software using the ‘Gels and Plot lanes’ plug-in.
Pharmacologic reagents
The FGFR-specific TKI, PD173074 (100 nM), cytochalasin-D (500 nM) and rapamycin (200 nM) were obtained from Sigma-Aldrich and used in the in vitro experiments.
Data analysis
Differences between experimental groups were assessed using analysis of variance (ANOVA) or Mann–Whitney test. The significance threshold was set at p ≤ 0.05. Statistical analyses were performed using GraphPad PRISM 6. All values are shown as mean ± S.E.M. Paired-Student’s t-test was performed when comparing the same cell population with two different treatments. Unpaired-Student’s t-test was performed when comparing two groups of mice or different primary chondrocyte preparations.
Supplementary Material
Supplementary Material is available at HMG online.
Acknowledgements
The authors thank Alban Mallet for the 3D reconstruction. We thank Drs Anne Lise Delezoide, Marie Gonzales, Philippe Loget, Bettina Bessières, Brigitte Leroy and Jelena Martinovič for providing fetal cartilage samples and Lofti Slimani for help in µCT acquisition at PIPA (plateforme d’imagerie du petit animal), Paris Descartes.
Conflict of Interest statement. None declared.
Funding
This program received a state subsidy managed by the National Research Agency under the ‘Investments for the Future’ program bearing the reference ANR-10-IHU-01. Some of the work presented here was funded by the European Community’s Seventh Framework Programme Under grant agreement no. 602300 (Sybil). We thank the Association des Personnes de Petites Tailles for their financial support.
