Development of the post-natal growth plate requires intraflagellar transport proteins

Abstract

In the post-natal growth plate, chondrocytes are arranged in columns parallel to the long axis of the bone. Chondrocytes divide perpendicular to this axis and then move into position one on top of another in a process called “rotation” that maintains columnar organization. Primary cilia are non-motile microtubule base appendages extending from the surface of almost all vertebrate cells. Primary cilia were described on chondrocytes almost 40 years ago but the function of these structures in cartilage biology is not known. Intraflagellar transport (IFT) is the process by which primary cilia are generated and maintained. This study tested the hypothesis that IFT plays an important role in post-natal skeletal development. Kif3a, a subunit of the Kinesin II motor complex, that is required for intraflagellar transport and the formation of cilia, was deleted in mouse chondrocytes via Col2a-Cre-mediated recombination. Disruption of IFT resulted in subsequent depletion of cilia and postnatal dwarfism due to premature loss of the growth plate likely a result of reduced proliferation and accelerated hypertrophic differentiation of chondrocytes. Cell shape and columnar orientation in the growth plate were also disrupted suggesting a defect in the process of rotation. Alterations in chondrocyte rotation were accompanied by disruption of the actin cytoskeleton and alterations in the localization of activated FAK to focal adhesion-like structures on chondrocytes. This is the first report indicating a role for IFT and primary cilia in development of the post-natal growth plate. The results suggest a model in which IFT/cilia act to maintain the columnar organization of the growth plate via the process of chondrocyte rotation.

Introduction

Primary cilia are non-motile microtubule based appendages extending from the surface of most vertebrate cells (Davenport and Yoder, 2005; Rosenbaum and Witman, 2002; Scholey, 2003). The process of Intraflagellar Transport (IFT) is responsible for building and maintaining the structure and function of primary cilia (Pazour and Rosenbaum, 2002; Rosenbaum and Witman, 2002; Scholey, 2003; Sloboda, 2002). Motor proteins, including Kinesin and Dynein, are required to move necessary components into and out of the cilium axoneme. Specific cargoes bind to specialized IFT particles, which include the protein IFT88/Tg737/Polaris, and are transported into cilia where they dissociate after arriving at their destination. By this means, the growth and maintenance of cilia structure and function is achieved. Disruption of Kif3a, a component of the Kinesin-II motor complex, disables anterograde IFT and leads to failure in the formation and maintenance of cilia (Hirokawa, 2000). Likewise, the absence of IFT88/Tg737/Polaris, a core component of the IFT particle, results in the loss of cilia (Murcia et al., 2000; Yoder et al., 2002). It has been proposed that cilia can act as a mechanosensors in various cell types (Jensen et al., 2004; Joly et al., 2003; Nauli et al., 2003). In addition, IFT and cilia are required to convey signals for various secreted growth factors during early embryonic development (Corbit et al., 2005; Haycraft et al., 2005; Huangfu and Anderson, 2005; Huangfu and Anderson, 2006; Huangfu et al., 2003; Liu et al., 2005; Park et al., 2006; Ross et al., 2005; Schneider et al., 2005).

Defects in the normal structure or function of primary cilia result in various congenital human diseases including polycystic kidney disease (PKD) (Lin et al., 2003; Lina and Satlinb, 2004; Nauli et al., 2003; Ou et al., 2005), situs-inversus (Marszalek et al., 1999), and retinal degeneration(Marszalek et al., 2000) as well as pleiotrophic syndromes such as Senior-Loken syndrome [Online Mendelian Inheritance in Man (OMIM) 266900, http://www.ncbi.nlm.nih.gov/omim], Jeune syndrome (OMIM 208500) and Bardet-Biedl syndrome (OMIM 209900), which often include skeletal defects (Pazour and Rosenbaum, 2002; Rosenbaum and Witman, 2002 Rosenbaum and Witman, 2003; Zhang et al., 2005). Specific skeletal defects noted include short limb dwarfism, chest deformity resulting from shortened ribs, and polydactyly (Tarnok et al., 2003; Verma, 2004). Primary cilia were first described on chondrocytes almost 40 years ago (Meier-Vismara et al., 1979; Scherft and Daems, 1967; Wilsman et al., 1980). Nevertheless, the role of IFT and cilia in skeletal development remains unclear.

Most of the bones in the body develop by a process called endochondral bone formation (Provot and Schipani, 2005). First, embryonic mesenchymal cells condense at the site where the skeletal element will form. Through the co-ordination of various growth factors, the chondrocytes proliferate and differentiate to form the anlagen of the future bone. Terminally differentiated chondrocytes are called the hypertrophic chondrocytes. These cells undergo cell cycle arrest and secrete a unique extra-cellular matrix (ECM) containing Type X collagen (Col10). Hypertrophic chondrocytes attract and promote invasion of vasculature and osteogenic cells then undergo apoptosis and invading osteoblasts replace the cartilage matrix with bone tissue. After birth, cartilage in the growth plate allows longitudinal growth of the bones. The post-natal growth plate has a highly organized structure in which cells are arranged into columns parallel to the long axis of the bone (Dodds, 1930). During cell division, chondrocytes divide perpendicular to the long axis of the bone and then undergo a series of cell shape changes and movements called rotation that result in the careful positioning of cells one on top of the other (Dodds, 1930). Very little is known about how this process is regulated.

Since IFT and primary cilia are important for the early development of many tissues, germ-line deletion of IFT proteins in the mouse results in mid-gestation embryonic lethality (Murcia et al., 2000; Takeda et al., 1999). Therefore, a targeted strategy of IFT deletion is necessary to study post-natal skeletal development. Previously we showed that deletion of Kif3a or IFT88/Tg737/Polaris in early embryonic limb mesenchyme via Prx1-Cre mediated recombination results in alterations in embryonic endochondral bone formation that are likely a result of alterations in Ihh signaling (Haycraft et al. 2007). In this study, deletion of Kif3a or IFT88/Tg737/Polaris was targeted to chondrocytes via Col2a-Cre mediated recombination in mice. The expression pattern of Cre recombinase under the control of the Col2a1 promoter was previously described (Ovchinnikov et al., 2000). Cre activity was first detected in the cartilage of the limbs between E11.5 and E12.5 days of gestation. Cre activity was not detected in the perichondrium. Here deletion of Kif3a or IFT88/Tg737/Polaris in chondrocytes resulted in dwarfism by 30 days after birth that was accompanied by complete loss of the growth plate. Long bones appeared normal in Kif3a deleted embryonic and newborn mice. Defects in the columnar organization of the growth plate were observed in Kif3a deleted mice between post-natal day 7 (P7) and P10. Alterations included reduced cell division and accelerated hypertrophic differentiation. Cell shape and the orientation of the cells relative to the long axis of the bone were disrupted suggesting defects in the process of chondrocyte rotation. Alterations in the organization of the growth plate were accompanied by disruption of the actin cytoskeleton and alterations in the localization of activated FAK to focal adhesion-like structures. The results indicate that IFT and primary cilia on post-natal chondrocytes mediate cell shape, orientation, growth, and differentiation.

Materials and Methods

Identification of transgenic mice

Generation of Kif3a^fl/fl mice was described previously (Lin et al., 2003). Mice with LoxP sites flanking the IFT88/Tg737/Polaris gene were recently described (Haycraft et al., 2007). Transgenic mice expressing Cre under the control of the Type II collagen promoter (Col2a-Cre) were obtained from Jackson Laboratories (Bar Harbor, Maine; (Ovchinnikov et al., 2000)). For timed pregnancies, experimental pairs (Col2a-Cre;Kif3a^fl/wt X Kif3a^fl/fl) were set up in the afternoon and checked for vaginal plugs the next morning. Noon on the day of the vaginal plug was considered E0.5 days of gestation. The genotype of transgenic mice was determined by PCR analyses of genomic DNA isolated from mouse tail (post-natal) or liver (embryos). The Col2a-Cre mice were identified by a pair of primers, Primer1: 5′ TGC TCT GTC CGT TTG CCG 3′; Primer2: 5′ ACT GTG TCC AGA CCA GGC 3′. The loxP alleles or Kif3a were identified using three primers, Primer1: 5′ TCT GTG AGT TTG TGA CCA GCC 3′; Primer2: 5′ AGG GCA GAC GGA AGG GTG G 3′; and Primer 3: 5′ TGG CAG GTC AAT GGA CGC AG 3′. IFT88/Tg737/Polaris mice were identified using the following primers: BY598 GCC TCC TGT TTC TTG ACA ACA GTG; BY919 GGT CCT AAC AAG TAA GCC CAG TGT T; BY956 CTG CAC CAG CCA TTT CCT CTA AGT CAT GTA. Col2a-Cre;Kif3a^fl/fl and Col2a-Cre;Ift88^fl/fl mice were generated at the expected Mendelian ratios.

Whole mount skeletal preparation

Skeletons from E18.5 day fetal mice and newborn mice were double stained for cartilage and bone with Alcian Blue and Alizarin Red (Kimmel and Trammell, 1981; McLeod, 1980). Whole mount skeletal preparations of P30 day mice were prepared according to Selby (Selby, 1987). Images were taken using an Olympus SZX12 dissecting microscope and Magnafire digital camera.

X-ray imaging

Four and eight week old mice were anesthetized by intra-peritoneal injection of Ketamine and Xylasine (1ml of 100mg/ml of Ketamine and 100μl of 100ng/ml of Xylasine diluted in PBS in a total of 10ml, inject 10μl per gram body weight) before taking the X-ray images. The images were taken using a Faxitron MX-20 X-ray machine and processed in Specimen software managed by the Center for Metabolic Bone Disease, UAB.

Histology, immunohistochemistry, and in situ hybridization

Long bones from E15.5, newborn, P7, P10, and P15-day old mice were fixed in 4% paraformaldehyde overnight at 4ºC. Bones from post-natal mice were taken out of the paraformaldehyde and rinsed with Diethylpyrocarbonate (DEPC) treated water and put into decalcification buffer containing 0.1M Tris, pH 7.5, 0.1% DEPC, 10% EDTA-4 Na, and 7.5% polyvinyl pyrolidione (PVP). The samples were kept in 4ºC with shaking for 30 days. The decalcification solution was changed at 14 days. Specimens were then rinsed with DEPC treated water, dehydrated, and embedded in paraffin. Sections were cut at a thickness of 5μm and mounted on Superfrost Plus slides (Fisher). For routine histological analysis, sections were stained with hematoxylin and eosin. For immunofluorescent staining, anti-GM130 monoclonal antibody (BD Biosciences, Cat No. 610822), rabbit anti-aggrecan polyclonal antibody (Chemicon, Cat No. AB1031), anti-acetylated tubulin monoclonal antibody (Sigma, Clone 6-11B-1, Cat No. T 6793), rabbit anti-phospho-Histone H3 (Ser10) polyclonal antibody (Upstate, Cat No. 06-570) or rabbit anti-FAK[pY⁸⁶¹] phosphospecific antibody (Biosource, Cat No. 44-626G) were used. Biotinylated secondary antibodies (Vector laboratories) and Cy3 conjugated streptavidin (Vector laboratories) were applied to yield the signal. The negative control was a slide incubated without the primary antibody followed by incubation with the secondary antibody and Cy3 conjugated biotin (not shown). Cytoskeletal F-actin was visualized using Alexa fluor 594 conjugated phalloidin (Invitrogen, Cat No. A12381). Frozen sections were used for phalloidin stains. Slides were counter stained with YO-PRO^®-3 iodide (612/631) (Invitrogen, Cat No Y3607) or 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI) (Invitrogen, Cat No. D1306). Vector M.O.M basic kit (Vector laboratories, Cat No. BMK-2202) was used when staining with monoclonal antibodies. FragEL^™ DNA Fragmentation Detection Kit (Colorimetric-TdT enzyme, Calbiochem, Cat No. QIA33) was used to detect apoptotic cells in the paraffin-embedded sections of P0, P7 and P10 control and mutant growth plates. Images of the sections were taken using an Olympus BX-51 upright microscope equipped with a Magnafire digital camera (Olympus). Confocal images were taken using a Leica confocal scanning microscope within the UAB high resolution imaging facility.

In situ hybridization was performed as described (Pelton et al., 1990). Briefly, sections were hybridized to ³⁵S-labeled anti-sense probes of Collagen type X (Col10a1), Indian Hedgehog (Ihh), Patched 1 (Ptc1), or Gli1. The Col10a1 probe was previously described (Apte et al., 1992). Ihh probe was kindly provided by Dr. Andrew McMahon, Harvard University. Ptc1 and Gli1 probe were previously described (Goodrich et al., 1996). Slides were exposed to KODAK Autoradiography Emulsion Type NTB (Cat. No. 889 5666, Kodak) at 4ºC for 4 days (Col10a1), 7 days (Ihh), 10 days (Ptc1), and 21 days (Gli1) and then developed by KODAK Developer D-19 (Cat. No. 146 4593, Kodak). Sections were counterstained with 0.2% toluidine blue. Images were obtained using an Olympus BX51 microscope and Magnafire digital camera. Images were taken under bright field (for toluidine blue) and dark field illumination (³⁵S signal).

Semi-Quantitative reverse transcription PCR

Chondrocytes from P10 control and Kif3a mutant mice were isolated from the epiphyses of the long bones. The tissue was homogenized with a PowerGen 125 homogenizer (Fisher Scientific). The homogenized tissue was processed in TRIzol Reagent (Invitrogen, Cat No. 15596-018) to extract RNA. cDNA was generated by reverse transcription PCR (RT-PCR). Conditions to allow DNA amplification in the linear range were determined by altering the amount of cDNA and cycle number for each primer pair tested. Final PCR reactions were carried out with 20ng of cDNA for 20 cycles for 18S and 200ng of cDNA for 30 cycles for Ptc1.

Results

Conditional deletion of Kif3a results in the depletion of chondrocyte cilia

To confirm the loss of cilia in Col2a-Cre;Kif3a^fl/fl mice, immunofluorescent staining using anti-actetylated α-tubulin antibody was performed on sections from growth plates of Col2a-Cre;Kif3a^fl/fl, Col2a-Cre;Kif3a^fl/wt, and Kif3a^fl/fl mice at E15.5, P0, and P7 days (Figure 1 and data not shown). At post-natal stages in control mice, cilia were seen as a concentrated red spot of acetylated α-tubulin at the center of the columnar cells forming a line along the longitudinal axis of the bone (Figure 1C). Staining was similar in Col2a-Cre;Kif3a^fl/wt and Kif3a^fl/fl mice suggesting one allele of Kif3a is sufficient for the formation and maintenance of cilia structure (data not shown). Cilia were also detected on control chondrocytes at E15.5 days of gestation (Figure 1G). Since the cells in the embryonic stages do not yet have a strict columnar orientation, the cilia were not preferentially localized along any particular axis within the cartilage. As expected, cilia were detected on only very few chondrocytes in Col2a-Cre;Kif3a^fl/fl mice at any of the time points examined (Figure 1A,E and data not shown). Based on this staining and our previous work with this Cre line, we estimated greater than 80% reduction in the number of chondrocytes with cilia in Col2a-Cre;Kif3a^fl/fl mice relative to the Col2a-Cre;Kif3a^fl/wt and Kif3a^fl/fl controls. In contrast, cilia on perichondrial cells in Col2a-Cre;Kif3a^fl/fl mice at all stages examined were intact (Figure 1B,D,F,H). This is consistent with previous reports indicating that this Cre-expressing strain does not effectively target the perichondrium (Ovchinnikov et al., 2000).

Figure 1. Depletion of cilia in Col2a-Cre;Kif3a^fl/fl chondrocytes.

Sections from growth plates of newborn (A–D) or E15.5 day (E–H) Col2a-Cre;Kif3a^fl/fl(A,B,E, F) and control (C,D,G, H) littermates were stained using anti-acetylated α-tubulin antibody (red) to show the cilia on chondrocytes (A,C,E,G) or perichondrial cells (B,D,F,H). The cilia are seen as condensed red dots and short lines (examples denoted with yellow arrows). Some diffuse staining is also seen in the cytoplasm. The nucleus is stained green with Yo-Pro. Many cilia were visible on control chondrocytes and perichondrial cells at all stages examined (C,D,F,G). Very few cilia were found on mutant chondrocytes (A,E) while cilia on perichondrial cells were intact in mutant mice (B,F). The longitudinal axis of the bone is indicated by the white arrows.

Col2a-Cre;Kif3a^fl/fl mice develop post-natal dwarfism

Phenotypic differences between Col2a-Cre;Kif3a^fl/fl and Kif3a^fl/fl controls were obvious by P30 days (Figure 2A,B). The mutant mice developed post-natal dwarfism rendering them smaller (32% reduced crown to rump and limb length) than their control littermates. Whole mount Alizarin Red skeletal preparations (Figure 2B) and X-ray images (Figure 2D,E) confirmed this observation. Additional skeletal defects including failure of fusion in the dorsal vertebrae (spina bifida occulta; Figure 2C), sporadic fusion of thoracic vertebrae (Figure 2 H-K), as well as small and deformed ribs; most likely a result of the loss of the costal growth plate (data not shown) were also observed in Col2a-Cre;Kif3a^fl/fl mice. X-ray images also revealed that the growth plates in the long bones, while clearly visible in control Kif3a^fl/fl mice, were missing in Col2a-Cre;Kif3a^fl/fl mice (Figure 2 F, G). In agreement with the normal cilia staining observed in Col2a-Cre; Kif3a^fl/WT mice, dwarfism was not detected in these mice suggesting no haplo-insufficiency effects due to deletion of one allele of Kif3a. The results suggest that dwarfism in Col2a-Cre;Kif3a^fl/fl mutant mice is due to alterations in the development of the post-natal growth plate. Dwarfism was also observed in Col2a-Cre;Ift88^fl/fl mice at P30 days (data not shown).

Figure 2. Col2a-Cre;Kif3a^fl/fl mice develop post-natal dwarfism.

(A) Pictures of P30 day Col2a-Cre;Kif3a^fl/fl (left) and control (right) littermates. (B) Alizarin red staining of the skeletons of P30 day Col2a-Cre;Kif3a^fl/fl (left) and control (right) littermates. (C) High magnification images from the Alizarin red stained samples of Col2a-Cre;Kif3a^fl/fl (left) and control mice (right) indicating the location of spina bifida occulta in the mutant mice and closed vertebrae in the control mice (black arrows). (D,E) Whole body X-rays of P30 day Col2a-Cre;Kif3a^fl/fl (D) and control (E) mice. (F,G) X-rays of the knee joint of Col2a-Cre;Kif3a^fl/fl (F) and control (G) mice. The translucent line at the epiphysis of the proximal tibia of the control mouse indicates the growth plate, which is absent in the mutant mouse (black arrows). (H, I) X-rays of the spine from the dorsal perspective of Col2a-Cre;Kif3a^fl/fl (H) and control (I) mice revealed fusion in some of the thoracic vertebrae of the Col2a-Cre;Kif3a^fl/fl mouse (black arrow, H). (J, K) X-rays of the spine from a lateral perspective of Col2a-Cre;Kif3a^fl/fl (J) and control (K) mice. The black arrow in J indicates an area where the mutant vertebrae appear fused.

Next, we determined the age at which the defects in skeletal development could first be seen. Even though the Col2a-cre transgene is expressed in limb chondrocytes by E12.5 days (Ovchinnikov et al., 2000) and there was a dramatic reduction in cartilage cilia by E15.5 days (Figure 1), differences in the morphology or organization of Col2a-Cre;Kif3a^fl/fl and control skeletons were not detected in E18.5 or P0 day mice as determined by Alizarin Red and Alcian Blue staining (data not shown). Differences in the length of long bones or levels of ossification were not observed. P0 day Col2a-Cre;Ift88^fl/fl mice also appeared normal. Hematoxylin and Eosin (H&E) staining of sections of long bones from Col2a-Cre;Kif3a^fl/fl and control mice at E15.5 and P0 days further confirmed this observation (Figure 3 A–D). The size and organization of the epiphysis was similar in mutant and control mice. The total length of the growth plate, proliferating (white double arrows) and hypertrophic (black double arrows) zones were also similar. In contrast, we recently showed that deletion of Kif3a or IFT88/Tg737/Polaris using Prx-1-cre mediated recombination results in defects in embryonic endochondral bone formation including a dramatic reduction in bone length and accelerated hypertrophic differentiation (Haycraft et al., 2007). Prx1- Cre is expressed in early limb mesenchyme starting at E9.5 days of gestation and thus recombination is targeted to chondrocytes and cells in the perichondium (Logan et al., 2002). The differences in the phenotypes observed in the two mouse models are likely due to the localization and timing of Cre-mediated recombination. Col2a-Cre is expressed later in the limbs, after cells have committed to the chondrocyte lineage, and the perichondrium is not efficiently targeted (Baffi et al. 2004; Ovchinnikov et al., 2000).

Figure 3. Alterations in the structure the Col2a-Cre;Kif3a^fl/fl growth plate.

Sections of the proximal tibia from Col2a-Cre⁺;Kif3a^fl/fl (A, C, E, G, I, K), Col2a-Cre; Polaris^fl/fl (M) and control (B, D, F, H, J, L, N) mice at E15.5 (A,B), P0 (C, D), P7 (E,F), P10 (G–J), and P15 days (K–N) were stained with hematoxylin and eosin to visualize growth plate structure. The length of the epiphysis to the beginning of the hypertrophic zoned was comparable in E15.5 day control and mutant mice (white double arrows in A,B). Likewise, the length of the zone of flat proliferating cells (C,D, white double arrows) and the length of the hpertrophic zone (C,D, black double arrows) were comparable in P0 day control and mutant mice. In P7 mice the length of the zone of flat proliferating cells was reduced in growth plates from mutant mice (E, white double arrow) relative to control (F, white double arrow) mice. The length of the hypertrophic zones (E,F, black double arrows) were comparable at P7 days. At P10 days, alterations in the organization of the cells within the growth plate were observed (G–J). The white arrowheads in H and J denote the sharp boundary between the prehypertrophic and hypertrophic zones of the growth plate in control mice, which is not clear in mutant mice (white arrowhead G and I). In addition, cells in the proliferating zone are flat and aligned into columns in control growth plates (J, white arrow). In contrast, the cells are not aligned into columns in the mutant growth plate (I, white arrow). At P15 days, control mice also demonstrate a sharp boundary between prehypertrophic and hypertrophic zones of the growth plate (white arrowheads L,N). The growth plate is essentially absent in Col2a-Cre⁺;Kif3a^fl/fl (K), Col2a-Cre; Polaris^fl/fl (M) mice.

At P7 days, subtle changes in the Col2a-Cre;Kif3a^fl/fl growth plate relative to littermate controls were observed (Figure 3 E, F). The total length of the epiphysis from the top of the bone to the bottom of the hypertrophic zone was reduced. This was due to a reduction in the distance between the top of the bone and the beginning of the zone of flat chondrocytes as well as a reduction in the length of the zone of flat chondrocytes (Figure 3E,F white double arrows). The length of the hypertrophic zone (black double arrows) was similar in control and mutant mice. By P10 days, alterations of the columnar organization of the Col2a-Cre;Kif3a^fl/fl growth plates relative to controls were observed (Figure 3 G–J). In control mice, the secondary ossification center was established and the growth plate was visible across the entire width of the skeletal element (Figure 3H). The round cells of the resting zone, flat cells of the proliferating zone (white arrows), and the hypertrophic zone were clearly demarcated (Figure 3J). The borders for each zone were sharp across the width of the growth plate (Figure 3H, white arrow heads). In contrast, the growth plate in Col2a-Cre;Kif3a^fl/fl mice was discontinuous across the width of the skeletal elements (Figure 3G). The primary and secondary ossification centers were continuous in areas where the growth plate was missing. The overall length of the epiphysis was reduced. The separate zones of cartilage were not clearly demarcated within the growth plate (white arrowhead Figure 3I). Hypertrophic cells had invaded the prehypertrophic zones, connecting the hypertrophic zone of the primary and secondary ossification centers (Figure 3G,I white arrowheads). Cells within the growth plate were not aligned into columns (white arrow; Figure 3I) and many of the cells within the growth plate were round instead of flat (Figure 3I). At P15 days, the growth plate structure in Col2a-Cre;Kif3a^fl/fl mice was almost completely abrogated. The primary and secondary ossification centers had merged and the distance from the articular surface to the trabecular bone was reduced relative to controls (Figure 3 K,L). Mice with targeted deletion of IFT88/Tg737/Polaris presented with a similar phenotype at P15 days (Figure 3 M,N). The results suggest that IFT in chondrocytes is required for normal progression of the growth plate between P7 and P15 days.

Reduced proliferation and accelerated hypertrophic differentiation in Col2a-Cre;Kif3a^fl/fl growth plate

To determine if differentiation of chondrocytes within the growth plate was altered, we looked at the expression of the ECM protein Aggrecan, a cartilage specific proteoglycan, and Col10a1, which is specifically secreted by hypertrophic chondrocytes (Figure 4). As expected, localization of Aggrecan and Col10a1 in newborn Col2a-Cre;Kif3a^fl/fl and control growth plates was not altered (data not shown), consistent with the histology data. At P10 and P15 days, Aggrecan was strongly localized to the ECM of all cartilage, including hypertrophic cartilage (white double arrows), in control mice (Figure 4 B, D). The columnar and organized nature of the growth plate was clear. Strong staining was also observed in the articular cartilage. There was strong Aggrecan staining in the Col2a-Cre;Kif3a^fl/fl growth plate, but disorganization of the growth plate was apparent (Figure 4A, C). At P10 days, staining in the hypertrophic zone was not uniform and columnar proliferating chondrocytes were difficult to define (Figure 4A). Clusters of morphologically hypertrophic cells were seen in the “prehypertrophic” area (Figure 4A, white arrow). Aggrecan staining was maintained in the articular cartilage in Col2a-Cre;Kif3a^fl/fl mice (Figure A,C). At P15 days, due to the loss of the growth plate, very little Aggrecan staining was detected (Figure 4C). Some staining at the lateral sides of the skeletal element suggested a residual growth plate (Figure 4C, white arrow).

Figure 4. Aggrecan and Type X collagen expression.

Immunofluorescent staining of Aggrecan (red; counterstain is green) in the growth plate of the proximal tibia from Col2a-Cre⁺;Kif3a^fl/fl (A,C) and control (B,D) mice at P10 (A, B) and P15 days (C, D). The hypertrophic zones are marked with white double arrows in A–D. The large white arrow in A indicates areas of histologically hypertrophic cartilage in the prehypertrophic zone of mutant growth plates at P10 days. The large white arrows in C represent the residual growth plate present in mutant mice at P15 days. In situ hybridization demonstrating Col10a1 expression in sections from growth plates of Col2a-Cre⁺;Kif3a^fl/fl (E) and control (F) P10 day tibia. Merged dark field and phase contrast images are shown. The black arrow in E represents an area in the mutant growth plate where hypertrophic cells are observed within the prehypertrophic region.

Next, the expression of Col10a1 mRNA was examined by radioactive in situ hybridization (Figure 4 E,F). At P10 days, Col10a1 mRNA was expressed at high levels in the hypertrophic zone of control mice (Figure 4F). The border between expressing and non-expressing cells was clearly demarcated. In contrast, Col10a1 mRNA levels were variable within the hypertrophic zone of Col2a-Cre;Kif3a^fl/fl growth plates (Figure 4E). In addition, the Col10a1 expression boundaries were not clearly demarcated and in some cases Col10a1 staining was mixed in with the prehypertrophic zones (Figure 4E, black arrow). At P15 days, Col10a1 mRNA was only detected in the residual growth plate at the lateral edge of the epiphysis and directly underneath the articular cartilage (data not shown). The results suggest that accelerated hypertrophic differentiation may in part contribute to the premature loss of the growth plate.

Since there was an overall reduction in the length of the growth plate, we compared the level of cell division in sections from control and mutant growth plates by anti- Phospho-Histone H3 staining at P0, P7, and P10 days (Figure 5). Phosphorylation of Histone H3 on serine 10 has previously been used as a marker for mitosis (Hendzel et al., 1997; Wei et al., 1999). The number of Phospho-Histone H3 positive cells was not statistically different in sections from control and mutant mice at P0 days (Figure 5 A,B, red cells and Figure 5G graph). In sections from P7 day mice the number of cells undergoing mitosis in the mutant growth plate was significantly reduced (86% reduced; p-value 8.77 X 10⁻⁶) relative to littermate controls (Figure 5C, D, G). At P10 days, the number of cells undergoing mitosis in the mutant growth plate was also reduced (78% reduced; p-value 0.001) relative to controls (Figure 5 E, F, G). By P15 days, essentially no staining was observed in the residual cells of the growth plates in Col2a-Cre;Kif3a^fl/fl mice (data not shown). Alterations in apoptosis were not detected by TUNEL staining in mutant chondrocytes at P0, P7, or P10 days compared to controls (data not shown). These data suggest that reduced cell division in chondrocytes may in part contribute to the premature loss of the growth plate in Col2a-Cre;Kif3a^fl/fl mice.

Figure 5. Proliferation is reduced in the Col2a-Cre⁺;Kif3a^fl/fl growth plate.

P0 (A,B), P7 (C,D), and P10 (E,F) Col2a-Cre⁺;Kif3a^fl/fl (A, C, E) and control (B, D, F) chondrocytes were immunostained using an anti-Phospho-Histone H3 anti-body (red). The bright red staining shows cells that are dividing. Cells were counterstained with YoPro, a green nuclear stain. The percentage of cells in mitosis was determined as the number of the red stained cells divided by the total number of cells (red plus green). Cells in 7 to 10 microscopic fields were counted and the mean, standard deviation, and T-test P-value were calculated and graphed (G). Significance was determined as a P-value < 0.05 (*).

Ihh signaling in Col2a-Cre;Kif3a^fl/fl growth plates

It has been shown that IFT proteins are required for normal Hedgehog (Hh) signaling in early embryonic development at a step downstream of Smoothened (Smo) and upstream of Gli proteins (Huangfu and Anderson, 2005; Park et al., 2006). Moreover, Smo and Gli proteins have been shown to localize to the cilia and IFT proteins are required for Gli processing (Corbit et al., 2005; Haycraft et al., 2005). A role for Hh signaling in the post-natal growth plate has not yet been demonstrated; however, Indian Hedgehog (Ihh) is known to regulate chondrocyte proliferation and differentiation during embryonic skeletal development (St-Jacques et al., 1999). Recently, we showed that deletion of Kif3a or IFT88/Tg737/Polaris in early limb mesenchyme via Prx1-Cre mediated recombination resulted in reduced Ihh signaling in embryonic cartilage as measured by the expression of Ptc1 and Gli1, direct targets of Hh signaling (Haycraft et al., 2007). We also show here that E15.5 day Col2a-Cre;Kif3a^fl/fl embryos demonstrate a dramatic reduction in Ptc1 expression in prehypertrophic chondrocytes (Figure 6 A–D). Ptc1 expression was maintained in only a few prehypertrophic cells, potentially representing the few cells in which cilia remain intact (Figure 6 C, D, small arrows). In contrast to what is observed in Prx1Cre cilia depleted mice, Ptc1 expression is maintained in the perichondrium of E15.5 day Col2a-Cre;Kif3a^fl/fl embryos consistent with the localization of Cre expression described previously. The results suggest cilia are involved in the regulation of Ptc expression during embryonic skeletal development.

Figure 6. Ptc1 expression in embryonic and post-natal Col2a-Cre⁺;Kif3a^fl/fl mice.

Sections of the elbow joint of E15.5 day Col2a-Cre⁺;Kif3a^fl/fl (A,C) and control (B, D) mice were hybridized to an anti-sense probe for Ptc1 mRNA (A–D). High magnification insets focusing on the proximal ulna of A and B are shown in C and D. Bright field (1) and dark field (2) images are shown. Larger arrows in A–D indicate staining in the perichondrium. The brackets in A and B indicate the prehypertrophic chondrocytes and asterisks denote staining in the mineralizing area of the bone. Smaller arrows in A and C denote the few cells in the mutant embryos that express Ptc1. Humerus = h, Radius = r, Ulna = u. Sections of proximal tibia of Col2a-Cre⁺;Kif3a^fl/fl (E, G) and control (F, H) P10 (E,F) and P15 day (G,H) mice were hybridized to an anti-sense probe for Ptc1 mRNA (E–H). Bright field (1) and dark field (2) images are shown. Arrows indicate the boundary of Ptc1 expression in the perichondrium; brackets in G and H indicate Ptc1 expression in prehypertrophic chondrocytes. The relative levels of Ptc1 expression were determined by semi-quantitative RT-PCR (I). Ptc1 was amplified for 30 cycles from cDNA generated from 200 ng RNA isolated from P10 day control Cre⁻ ;Kif3a^fl/wt (−;−/+), Cre+;;Kif3a^fl/wt (+;−/+), and experimental Cre+;;Kif3a^fl/fl growth plates. Amplification of 18S (20 ng RNA at 20 cycles) was used as a control. Conditions were shown to be in the linear range for product formation.

Since post-natal chondrocytes in Col2a-Cre;Kif3a^fl/fl mice demonstrate reduced proliferation and accelerated hypertrophic differentiation, it was of interest to investigate whether Ptc1 expression was affected in post-natal Col2a-Cre;Kif3a^fl/fl cartilage. To this end, expression of Ihh and Ptc1 mRNA was determined by radioactive in situ hybridization in sections from P0, P7, P10, and P15 day control and mutant growth plates (Figure 6 and data not shown). In P10 day control and mutant mice, Ihh expression was similar and the mRNA was localized to cells in the lower prehypertrophic and upper hypertrophic zones (data not shown). In control mice, Ptc1 was expressed in cells in the perichondrium, articular cartilage, and at the osteochondral junction. In contrast to what was observed in embryonic skeletons, only low levels of diffuse Ptc1 expression were detected in prehypertrophic chondrocytes (Figure 6 F, H and data not shown). Differences in the level or pattern of Ptc1 expression were not observed in control or mutant mice at P0, P7, or P10 days (Figure 6 E, F and data not shown). Although there were not many chondrocytes present at P15 days in mutant growth plates, after accounting for differences in the structure of the bone, Ptc1 expression was expressed at a similar pattern and level within chondrocytes to that seen in the control mice (Figure 6 G,H). Since the growth plate was essentially gone at this late stage, perichondrium and perichondrial staining for Ptc1 were not detected (Figure 6 G, H). Finally, to better determine the relative levels of Ptc1 mRNA expression in control versus mutant mice, RNA isolated from growth plates dissected from P10 day control and Col2a-Cre;Kif3a^fl/fl mice was used in a semi-quantitative RT-PCR assay (Figure 6 I). Conditions to detect Ptc1 and 18S PCR product formation within the linear range were determined first. Within this range, the relative levels of Ptc1 mRNA were comparable in control and mutant growth plates. Amplification of 18S was used as an internal control. In contrast to what we observed in the embryonic skeleton, the results suggest that Ptc1 expression is not altered in the post-natal growth plate of Kif3a deficient mice relative to controls.

Col2a-Cre;Kif3a^fl/fl chondrocytes demonstrated alterations in cell shape and orientation

In the post-natal growth plate, chondrocytes are arranged in columns parallel to the axis of the long bone. The cells divide perpendicular to this axis and the daughter cells move into a position one on top of another in a process called “rotation” to maintain columnar organization. Since the columnar organization of the growth plate was lost, we compared the polarity and orientation of growth plate chondrocytes in control and mutant mice. It has been shown that the relationship of the Golgi apparatus to the nucleus defines the polarity of chondrocytes. That is, the Golgi side of the cell can be defined as the “apical” side and polarity is determined by the location of the Golgi relative to the nucleus. In contrast, the orientation of cells is determined by the relationship of cells to each other and to the longitudinal axis of the bone (Holmes and Trelstad, 1980). To determine if there were alterations in the polarity or orientation of the cells in the growth plate of mutant mice, immunofluorescent staining with antibodies directed to a Golgi specific protein, GM-130 (red), in combination with a nuclear counter-stain (green) was used (Figure 7A–H). Polarity of the cells, as defined by which side of the cell contained the Golgi, alternated within the proliferating cells within columns in control mice (Figure 7 D,H). This strict polarity is not maintained through the entire growth plate. Golgi staining can be seen all around the nucleus in cells in the prehypertrophic to hypertrophic transition in control mice (Figure 7H inset) indicating that polarity is not normally maintained in mature chondrocytes. The polarity of Col2a-Cre;Kif3a^fl/fl cells appeared normal at P10 days since the Golgi was still restricted to one side of the cell (Figure 7C). By P15 days, staining of the Golgi in mutant cells was all around the cell rather than limited to one side (Figure 7G) indicating a lack of polarity. Since there is accelerated maturation in the mutant growth plate, the lack of polarity in the few cells left by P15 days is likely a consequence of the maturation state of these cells and not a direct effect on polarity.

Figure 7. Orientation and shape of chondrocytes is altered in the Col2a-Cre⁺;Kif3a^fl/fl growth plate.

Sections of Col2a-Cre⁺;Kif3a^fl/fl (A, C, E, G) and control (B, D, F, H) growth plates at P10 (A–D) and P15 days (E–H) were stained with anti-GM130 antibody to show the Golgi apparatus (red). The nucleus was stained with YoPro (green). Low (A, B, E, F) and high (C, D, G, H) magnification images of the same field are shown. The white arrow in A denotes the area that is magnified in B. Double white arrows denote the direction of the longitudinal axis of the bone. The white lines through representative cells denote the orientation of the cell. The inset in H shows control cells in the transition zone between the prehypertrophic and hypertrophic zones. (I, J) Cells stained with Phospho-histone H3 (red) were used to determine the plane of chondrocyte division. Examples of dividing mutant (I) and control (J) cells are shown. All of the cells in the growth plate that were stained divided perpendicula to the long axis of the bone. Double white arrows denote the direction of the longitudinal axis of the bone. Small white arrows indicate the direction of cell division.

Chondrocytes in the columnar zone of the growth plate divide along an axis perpendicular to the long axis of the column. We could define the axis of division in high magnification images of cells stained for Phospho Histone H3 (examples in Figure 7 I,J, axis of the long bone indicated by double white arrows). In both control and mutant cells, division occurred along the perpendicular axis in every stained cell that was observed. No stained cells were observed within circular or disorganized groups of cells but cells just outside these areas were stained and could be seen dividing along the perpendicular axis. The results again suggest that alterations in the polarity of Col2a-Cre;Kif3a^fl/fl chondrocytes are not the primary defect.

To determine orientation of cells within the column, a line can be drawn through the center of the Golgi and the nucleus of a particular cell and the angle of that line relative to the long axis of the bone can be determined (Holmes and Trelstad, 1980). In sections from control P10 and P15 day growth plates, the line through the Golgi and nucleus of a particular cell was essentially always perpendicular to the longitudinal axis of the bone (Figure 7D,H; small white lines relative to double white arrows). Furthermore, the cells in the columns were flat. In contrast, the orientation of the cells in the growth plates of Col2a-Cre;Kif3a^fl/fl mice was not always perpendicular to the long axis of the bone, in some cases resulting in circular complexes of cells instead of columns (Figure 6C; small white lines relative to the double white arrows). In addition, the cells demonstrated a round morphology and there were many cells that were sitting side-by-side rather than one on top of the other. The alterations in cell shape and orientation observed suggest that there are defects in the process of rotation, which is required to maintain the columnar structure of the growth plate. We propose that the polarity of the cells is intact but alterations in cell shape and movement that occur during rotation lead to disorientation of the chondrocytes so that they do not form columns. This may explain in part why the epiphysis is very broad but the bones are short (Figure 2).

Localization of Phosho-FAK and actin organization are altered in the Col2a-Cre;Kif3a^fl/fl growth plate

Very little is known about the process of chondrocyte rotation. It has been shown that mice with conditional deletion of β1 integrin in Col2a expressing cells have, among multiple other defects, alterations in the shape and orientation of cells that would suggest defects in chondrocyte rotation (Aszodi et al., 2003; Bengtsson et al., 2005; Grashoff et al., 2003; Terpstra et al., 2003). Integrins are localized to both the primary cilia and the cell body (McGlashan et al., 2006), so it is of interest to know if disruption of IFT and cilia result in alterations in integrin signaling. Fak activation and phosphorylation are tightly linked to integrin signaling and cell adhesion so we looked at the localization of phosphorylated Focal adhesion kinase (pFAK) within the columnar chondrocytes in sections from P7 and P10 control and mutant mice using immunofluorescent staining (Figure 8A–D). Most of the pFAK observed by immunofluorescent staining appeared to be intracellular; however, upon closer inspection, we saw that pFAK was also distributed in a punctate pattern around control cells, likely representing areas of focal adhesion (Figure 8B,D small white arrows). In contrast, only a few spots of pFAK staining were detected in focal adhesions in mutant chondrocytes while intracellular pFAK staining was unaltered (Figure 8A,C). Alterations in pFAK localization were observed in mutant columnar cells at P7 days (Figure 8A) and in mutant cells at P10 days that were still in columns (Figure 8C1) as well as cells in the disorganized circular complexes (Figure 8C2) suggesting alterations in pFAK localization occur prior to changes in cell orientation and morphology. Next, we looked at the organization of the actin cytoskeleton by staining with phalloidin (Figure 8E,F). In control mice, even cortical actin was observed (Figure 8F). In contrast, actin was unorganized in chondrocytes from Kif3a deleted mice (Figure 8E). The results indicate that deletion of IFT and cilia result in alterations in chondrocyte rotation that correlate with disruptions in the organization of the actin cytoskeleton and alterations in the localization of activated FAK within the cells of the post-natal growth plate.

Figure 8. Alterations in phospho-FAK and actin cytoskeleton in Col2a-Cre⁺;Kif3a^fl/fl mice.

Sections from the proximal tibia of Col2a-Cre⁺;Kif3a^fl/fl (A, C, E) and control (B, D, F) P7 (A, B) and P10 (C–F) day mice were immunostained with an antibody directed to pFAK followed by a biotinylated secondary antibody and Cy3 conjugated avidin (A–D) or Alexfluor 594 conjugated phalloidin (E,F). pFAK staining was visible in focal adhesions in the columnar chondrocytes from control mice (B, D1, D2). A few examples of focal adhesions are indicated with arrows. There was a dramatic reduction in pFAK in staining in the focal adhesions in the Kif3a deleted samples (A, C1, C2). Staining in a few focal adhesions remained (small arrows). Chondrocytes in control mice demonstrated uniform cortical actin staining (F1–F3, three samples are shown). Mutant chondrocytes also demonstrated some cortical actin staining but, in most cells, actin was disorganized and uneven (E1–E3, three samples are shown).

Discussion

In this study, proteins required for IFT were deleted in chondrocytes using Col2a-Cre mediated recombination. The results demonstrate an important role for IFT in the process of post-natal skeletal development. Conditional deletion of Kif3a, a subunit of the Kinesin-II motor complex, resulted in significant depletion of chondrocyte cilia as early as E15.5 days of gestation while cilia were maintained on cells in the perichondrium. Deletion of IFT and cilia resulted in post-natal dwarfism due to premature loss of the growth plate. Chondrocyte proliferation was reduced in the mutant growth plate and hypertrophic differentiation accelerated by P7 and P10 days respectively. In addition to alterations in proliferation and differentiation, chondrocytes demonstrated alterations in cell shape and rotation, the process by which chondrocytes maintain columnar organization. Prehypertrophic chondrocytes maintained their polarity but failed to maintain their strict orientation within the growth plate. Defects in chondrocyte rotation correlated with alterations in localization of activated FAK and disruption of the actin cytoskeleton. The results suggest that IFT/cilia are required to maintain the growth plate by regulating chondrocyte proliferation, differentiation, and rotation.

Recently, it was shown that signaling by Hh proteins require normal cilia structure and function (Corbit et al., 2005; Haycraft et al., 2005; Huangfu et al., 2003; Liu et al., 2005). Smo, an important membrane bound signaling protein for Hh, as well as Gli1, Gli2 and Gli3, intracellular signaling molecules, are localized in the cilia (Corbit et al., 2005; Haycraft et al., 2005). Mice with germline deletion of the Ihh gene demonstrate defects in embryonic development of the long bones that are apparent as early as E13.5 days of gestation (St-Jacques et al., 1999). Specifically, mice demonstrate reduced chondrocyte proliferation and accelerated hypertrophic differentiation. Similar results were seen in mice with targeted deletion of Ihh in Col2a-Cre expressing cells (Razzaque et al., 2005). Nevertheless, Col2aCre;Kif3a^fl/fl mice did not demonstrate an embryonic phenotype even though it is known that Col2a-Cre is expressed at E12.5 days and we demonstrated significant depletion of cilia in chondrocytes by E15.5 days. Cilia were maintained on cells in the perichondrium of Col2aCre;Kif3a^fl/fl mice consistent with previous reports indicating that this Cre-expressing strain does not effectively target the perichondrium (Baffi et al., 2004; Ovchinnikov et al., 2000). Furthermore, Ptc1 expression was dramatically reduced in prehypertrophic cells but was maintained in the perichondrium. In contrast, we recently showed that deletion of Kif3a or IFT88/Tg737/Polaris using Prx-1 mediated recombination results in defects in embryonic endochondral bone formation that include a dramatic reduction in bone length and accelerated hypertrophic differentiation with similarities to the phenotypes of Ihh-null and Ihh/Gli3 double null mice (Haycraft et al., 2007). Deletion of Kif3a or IFT88/Tg737/Polaris in early limb mesenchyme via Prx1-Cre mediated recombination also resulted in reduced Ihh signaling in embryonic cartilage and perichondrium as measured by the expression of Ptc1 and Gli1 (Haycraft et al., 2007). The results indicate that IFT is required for Ihh signaling in the embryonic skeleton. Prx1- Cre is expressed in early limb mesenchyme starting at E9.5 days of gestation and thus recombination is targeted to chondrocytes and perichondium (Logan et al., 2002). The differences in the phenotypes observed between the two mouse models are likely due to the localization and timing of Cre-mediated recombination. Col2a-Cre is expressed later in the limbs, after cells have committed to the chondrocyte lineage and the perichondrium is not efficiently targeted. There is significant evidence for Ihh signaling in the perichondrium during embryonic development (Alvarez et al., 2002; Long et al., 2001; Vortkamp et al., 1996) and this may in part explain the lack of an embryonic phenotype in Col2aCre;Kif3a^fl/fl mice. Alternatively, the few cells that maintain cilia and Ptc1 expression in Col2aCre;Kif3a^fl/fl mice may be sufficient to mediate all the biological effects of Ihh required for skeletal development.

Although Col2aCre;Kif3a^fl/fl mice demonstrated reduced proliferation and accelerated hypertrophic differentiation in the post-natal growth plate, we did not observe alterations in the expression or localization of Ptc1 mRNA. Ihh has not been specifically deleted from post-natal chondrocytes so it is not clear there is a biological role for Ihh signaling directly in post-natal chondrocytes. The role of Ihh in post-natal cartilage may be distinct from that during embryonic development. Furthermore, most Ptc1 expression in the post-natal growth plate was localized to the perichondrium with only very low levels of staining observed in pre-hypertrophic chondrocytes. It is possible that the level of staining observed in prehypertrophic cells represents a basal, unactivated level of expression and perichondrial cells, which are not targeted in Col2aCre;Kif3a^fl/fl mice, are the primary Ihh responding cells in the post-natal growth plate. It is also a formal possibility that, in contrast to what has been shown in embryonic development, Ihh signaling occurs in the absence of IFT and cilia in post-natal cartilage. Additional mouse and cell culture models will be required to distinguish between these possibilities.

Post-natal proliferating zone chondrocytes are discoid in shape and orient themselves into columns of 6 to 8 cells. Although the cells form columns along the longitudinal axis of the bone, the cells demonstrate specific polarity and divide along the perpendicular axis. After the cells divide they flatten and migrate so that one daughter cell is under the other in the column. This process has been termed rotation and very little is known about how it is regulated. The columnar organization of the growth plate is lost in Col2aCre;Kif3a^fl/fl mice suggesting IFT/cilia mediate the process of chondrocyte rotation. Specifically, the polarity of the cells and axis of cell division is normal but the shape and orientation of the cells relative to each other and relative to the long axis of the bone is disrupted. Many of the aspects of chondrocyte rotation are similar to the process of convergent extension that occurs during embryonic gastrulation (Michael Schubert, 2003). The establishment of cell polarity is the first step. Cells then flatten, migrate to the midline, and intercalate forming columns of cells. It has been shown that Wnts, signaling through a non-canonical pathway referred to as Planar Cell Polarity (PCP), regulate cell polarity, migration, and orientation of cells during convergent extension. Recently, it was shown that defects in cilia disrupt PCP and convergent extension-like movements in many organs, including the Meckel’s cartilage in Xenopus embryos (Dabdoub and Kelley, 2005; Park et al., 2006; Ross et al., 2005). It is possible that PCP-like signaling is altered in the Col2aCre;Kif3a^fl/fl growth plate leading to defects in rotation.

Communication with the ECM is also required during convergent extension and chondrocyte rotation (Aszodi et al., 2003; Marsden and DeSimone, 2003). Integrins have been localized to primary cilia as well as the cell body in chondrocytes and other cells (McGlashan et al., 2006; Praetorius et al., 2004). Conditional deletion of β1 integrin in chondrocytes results in defects in the process of rotation that include alterations in cell shape and orientation with some similarities to that seen in mice with disrupted cilia (Aszodi et al., 2003). Furthermore, it was recently shown that deletion of Pkhd1, a cilia associated protein, in cultured epithelial cells resulted in the loss of primary cilia as well as alterations in cell adhesion and activation of FAK suggesting a link between the primary cilia and integrin mediated cell-ECM interactions (Mai et al., 2005). In this report, we show that disruption of IFT and cilia results in alterations in chondrocyte rotation that are accompanied by alterations in the localization of pFAK to focal adhesion-like structures on chondrocytes and disruption of the actin cytoskeleton. Based on the combined evidence, we propose a model in which the loss of cilia results in alterations in cell-ECM interactions, which in turn result in defects in chondrocyte rotation. The establishment of direct cause and effect relationships proposed in this model will have to await experiments using more elaborate mouse models as well as cell culture systems.

It is thought that primary cilia on kidney epithelial cells are sensitive to mechanical stimuli and act as sensors for fluid flow (Pazour and Witman, 2003; Praetorius and Spring, 2003). Polycystin-2 (Pkd2) and Polycystin-1 (Pkd1), two proteins involved in human polycystic kidney disease, are located in the cilia and are thought to function as a mechanosensitive cation channel and a channel regulatory protein, respectively. It is known that mechanical stress plays an important role in bone formation and adaptation. Recently, it was shown that mice with an inactivating mutation in Pkd1 demonstrate osteopenia as a result of reduced osteoblast function (Xiao et al., 2006). Cilia on osteoblasts/osteocytes were suggested as candidates for the elusive mechanosensor in bone. The optimum amount of mechanical force imposed on a long bone will also promote the proliferation and differentiation of growth plate chondrocytes while excessive force will cause cessation of growth and cell death (Frost, 1990). Very little is known about how mechanical force is received by chondrocytes. It was recently suggested that chondrocyte cilia function as mechanosensors (Jensen et al., 2004). Studies of the ultrastructure of cilia in primary cultures of sternal chondrocytes showed that the cilia interdigitate among collagen fibers and proteoglycans. Bending patterns of the cilia were also examined using confocal microscopy. Bending patterns suggesting response to shear stress were observed as well as patterns suggesting deflection by ECM contact. It was suggested that cilia could transmit mechanical force through their interaction with the surrounding ECM. It is interesting to note that although cilia are dramatically reduced in the cartilage of Col2a-Cre;Kif3a^fl/fl mice as early as E15.5 days, we do not detect alterations in the organization of the growth plate until seven to ten days after birth. This is the time when the young mice are starting to become subjected to significant mechanical load. It was previously shown that proliferation and differentiation of rat growth plate chondrocytes are affected by microgravity (Duke and Montufar-Solis, 1999). The effects were visible only during a short window of development with older rats being resistant to the effects of unloading. It is possible that the delay between the loss of cilia and disorganization of the growth plate is related at least in part to the onset of mechanical force and the inability of the cells to respond at a critical window in development.

In summary, this is the first report to demonstrate the function of the IFT/cilia in the post-natal growth plate. The results presented here suggest a model in which primary cilia regulate proliferation, differentiation, and organization of the growth plate.