Primary cilia mediate mechanotransduction through control of ATP-induced Ca²⁺ signaling in compressed chondrocytes

Abstract

We investigated the role of the chondrocyte primary cilium in mechanotransduction events related to cartilage extracellular matrix synthesis. We generated conditionally immortalized wild-type (WT) and IFT88^orpk (ORPK) mutant chondrocytes that lack primary cilia and assessed intracellular Ca²⁺ signaling, extracellular matrix synthesis, and ATP release in response to physiologically relevant compressive strains in a 3-dimensional chondrocyte culture system. All conditions were compared to unloaded controls. We found that cilia were required for compression-induced Ca²⁺ signaling mediated by ATP release, and an associated up-regulation of aggrecan mRNA and sulfated glycosaminosglycan secretion. However, chondrocyte cilia were not the initial mechanoreceptors, since both WT and ORPK cells showed mechanically induced ATP release. Rather, we found that primary cilia were required for downstream ATP reception, since ORPK cells did not elicit a Ca²⁺ response to exogenous ATP even though WT and ORPK cells express similar levels of purine receptors. We suggest that purinergic Ca²⁺ signaling may be regulated by polycystin-1, since ORPK cells only expressed the C-terminal tail. This is the first study to demonstrate that primary cilia are essential organelles for cartilage mechanotransduction, as well as identifying a novel role for primary cilia not previously reported in any other cell type, namely cilia-mediated control of ATP reception.—Wann, A. K. T., Zuo, N., Haycraft, C. J., Jensen, C. G., Poole, C. A., McGlashan, S. R., Knight, M. M. Primary cilia mediate mechanotransduction through control of ATP-induced Ca²⁺ signaling in compressed chondrocytes.

Physiological joint loading and the associated compression of articular cartilage produces a number of interrelated physicochemical events that influence both chondrocyte metabolism and extracellular matrix (ECM) synthesis (1, 2). However, the process of chondrocyte mechanotransduction is only partially understood (3–7). A prime candidate for cellular mechanosensation is the primary cilium, an organelle present on nearly all eukaryotic cell types, including chondrocytes (8, 9). It consists of a membrane-coated axoneme that projects from the cell surface into the extracellular microenvironment and an intracellular basal body (9–11). Assembly and maintenance of the cilium depend on the bidirectional intraflagellar transport (IFT) system (12). Disruption of genes involved in IFT, such as IFT88 and Kif3a, causes defective ciliogenesis, which results in a wide range of developmental and postnatal abnormalities (13–15). In cartilage, primary cilia are essential for regulation of joint development such that loss of cilia results in skeletal patterning abnormalities, disruption of cell polarity and organization, and ECM defects (16, 17).

Primary cilia are known to function as mechanosensory organelles in several cell types, in particular those subjected to fluid flow, such as cholangiocytes, vascular endothelium, and renal epithelia (18–23). We hypothesize that the primary cilium is also involved in chondrocyte mechanosensation in response to compressive loading. Chondrocyte cilia are ideally placed to sense associated deformations since they are structurally associated with collagen fibers, express collagen anchoring receptors including integrins, and are mechanically deflected by interactions with the ECM (24, 25). Although several studies have examined the role of primary cilia as flow sensors, no studies have examined the role of the cilium in response to other types of physiologically relevant mechanical loads. Here we present data on postnatal chondrocyte primary cilia and show that they are essential for the transduction of mechanically regulated signals in response to compression. Furthermore, we show that this occurs via a novel cilia mechanotransduction pathway involving modulation of ATP-induced Ca²⁺ signaling via purine receptors.

MATERIALS AND METHODS

Antibodies and fluorescent probes

The study used the following antibodies: P2X2, P2X4, P2X7, P2Y1, and P2Y2 (Alomone Laboratories, Jerusalem, Israel); acetylated α-tubulin (clone 6-11B-1, Sigma, Auckland, New Zealand; and C3B9, gift from T. Sherwin, University of Auckland); large T-antigen SV40 (BD Biosciences, Auckland, New Zealand); polycystin-1 (PC1; H-260) and polycystin-2 (PC2; H-280; Santa Cruz Biotechnologies, Santa Cruz, CA, USA); goat anti-rabbit Dylight 488 and goat anti-mouse Dylight 594 (Jackson ImmunoResearch, West Grove, PA, USA); and sheep anti-rabbit horseradish peroxidase (Chemicon, Auckland, New Zealand). Probes: Hoescht 1:500 (Sigma); Fluo-4 AM, calcein AM, and ethidium homodimer-2 (Invitrogen, Auckland, New Zealand).

Generation of ciliated and nonciliated chondrocyte cell lines

Heterozygous Tg737^ORPK mutant mouse lines were generated as described previously (26, 27). Mice were maintained on a mixed genetic background according to approved protocols at the Medical University of South Carolina. Heterozygous Oak Ridge polycystic kidney (ORPK) mice were bred with heterozygous Immortomouse mice (H-2Kb-tsA58), which harbor a temperature-sensitive SV40 large T-antigen transgene under the control of an interferon-γ-inducible H-2Kb promoter (H-2Kb-tsA58) to produce orpk/Immortomouse compound heterozygous mice (28). Heterozygous ORPK females were bred with heterozygous/Immortomouse orpk males. Chondrocytes were isolated from the sterna of 4-d-old mice by digestion with collagenase type II (2 mg/ml) at 37°C for 4 h. All mice were genotyped by PCR from tail biopsy DNA. Western blot analysis was conducted to confirm the expression of SV40 large T antigen in chondrocytes in the presence of IFN-γ at 33°C.

Chondrocytes were cultured in DMEM supplemented with 10% FCS, 88 U/ml penicillin, 90 μg/ml streptomycin, 10 ng/ml INF-γ, and 2.5 mM l-glutamine (Invitrogen). Immortalized cells were grown to 90% confluence in 5% CO₂ at 33°C with 10 nM IFN-γ, then cultured in nonpermissive conditions at 37°C (without IFN-γ) for 5 d before seeding in 5- × 5-mm cylindrical agarose gels (4). Cell-agarose constructs were cultured for up to 24 h before loading experiments. Cell viability was assessed by calcein AM/ethidium homodimer-2 labeling and fluorescence microscopy.

Compressive loading

To analyze the effect of mechanical loading on ECM synthesis and ATP release, we used a loading rig mounted within a cell culture incubator at 37°C in 5% CO₂, driven by a Bose mechanical compressive strain system (Bose Corp., Eden Prairie, MN, USA). This system has proven to be highly accurate and repeatable, as indicated by load values measured from cells beneath the 24-well plates. Each construct was hydrated in culture medium and loaded within individual wells of a 24-well plate. Control constructs were also mounted in the loading rig but without application of cyclic compression. Samples were allowed to rest for 1 h under pins before loading commenced. Loading was conducted in a sinusoidal fashion with a peak compressive strain of 15% at 1 Hz (29–31). To examine mechanically induced Ca²⁺ signaling, a separate microscope-mounted loading system was used, as described previously (32, 33).

Total sGAG and DNA quantification

sGAG concentration was assessed within both constructs and culture medium following a 24-h period of cyclic compression. Constructs were digested, and the total sGAG content was quantified using the 1,9-dimethylmethylene blue (DMB; Sigma, Poole, UK) assay (34). To assess cell proliferation, DNA was quantified using Hoechst 33258 fluorescence as described previously (4).

Gene expression

Expression of two key cartilage-specific ECM molecules, aggrecan and collagen II mRNA, was examined in unloaded cell-agarose constructs and after 1 h cyclic loading. Constructs were placed in QG buffer (Qiagen, Crawley, UK) and frozen in liquid N₂. RNA extraction from agarose constructs was adapted from the Qiagen RNeasy minikit. Genomic DNA cleanup with a DNase kit (Sigma, Poole, UK) was conducted, and reverse transcription reactions were carried out using products from Sigma. Primer sequences were synthesized by Sigma: collagen IIa, forward 5′-GGCAACAGCAGGTTCACATA and reverse 5′-ATGGGTGCGATGTCAATAAT; aggrecan, forward 5′-CACGCTACACCCTGGACTTTG and reverse 5′-CCATCTCCTCAGCGAAGCAGT; GAPDH, forward 5′-GACAAAATGGTGAAGGTCGG and reverse 5′-TCCACGACATACTCAGCACC. For qPCR, a fast 2-step SYBR kit (KAPA Biosystems, Boston, MA, USA) was used with a reference dye, ROX. Fluorescence data, normalized to ROX, were collected during annealing and analyzed using MxPro QPCR 3.0 software (Stratagene, La Jolla, CA, USA). Cycle threshold (C_t) values for each triplicate reaction were averaged. Relative quantification of aggrecan and collagen type II was normalized to GAPDH and to the mean of calibrator (control) samples from the same preparation by a comparative C_t approach. Expression ratios were expressed as log².

Extracellular ATP measurements

Following a 1-h period of cyclic mechanical loading of chondrocyte-agarose constructs, the medium was removed and kept on ice for immediate ATP measurement, using a luciferase ATP assay kit (Sigma, Poole, UK) and a luminescence plate reader (35).

Intracellular calcium signaling

Chondrocyte-agarose constructs were labeled with 5 μM Fluo-4 AM plus 0.1% Pluronic for 1 h at room temperature, followed by a 30-min deesterification period. Constructs were mounted in the compression rig and visualized using confocal fluorescence microscopy. In a separate set of experiments, chondrocyte-agarose constructs were exposed to either 1 or 100 μM ATP (Sigma, Auckland, New Zealand). Images were acquired every 6 s over a 5-min period of either compressive loading (20% strain) or ATP exposure. A subset of constructs was also incubated with 10 μM suramin (Sigma) for 30 min prior to imaging. Control constructs remained unstrained or untreated. The mean fluorescent intensity within individual cells was recorded and plotted against time. Characteristic Ca²⁺ transients were manually identified as in previous studies (32, 33, 36).

Western blotting

Expression of SV40, P2 receptors, PC1, and PC2 was determined by Western blotting. Monolayer chondrocytes were extracted using RIPA buffer containing protease inhibitors (Roche, Auckland New Zealand) for 1 h at 4°C. Total protein was measured using the Bio-Rad DC Protein Assay (Bio-Rad, Gladesville, NSW, Australia). Supernatants were heated to 70°C for 10 min, and protein was separated on NuPAGE 4–12% gels (Invitrogen) or 3–8% Tris-bis gels and transferred onto PVDF membranes (Hybond-P; Amersham Biosciences, Little Chalfont, UK). Membranes were blocked with 5% nonfat milk powder for 1 h at room temperature, then primary antibody overnight at 4°C, followed by a horseradish peroxidase secondary for 2 h at room temperature. Peroxidase activity was visualized using enhanced chemiluminescence (ECL Plus; Amersham Biosciences). Values were normalized to GAPDH.

Immunofluorescence and confocal microscopy

Chondrocyte-agarose constructs were fixed in 3.7% (w/v) paraformaldehyde and processed in paraffin wax, as described previously (37). Sections (12 μm thick) underwent microwave antigen retrieval, followed by dual immunofluorescent labeling with antibodies against either γ-tubulin or PC1 or PC2 and acetylated α-tubulin, as described previously (37). Sections were imaged using a Leica TCS SP2 confocal microscope using a ×63, 1.3-NA lens (Leica Microsystems, Wetzlar, Germany; ref. 37). Monolayer cells grown on glass coverslips were fixed and stained in a similar manner.

Statistics

Statistical analyses involved Student's t tests, Mann-Whitney U tests, or analysis of variance (ANOVA) to examine the differences between two or more groups, respectively. For Ca²⁺ studies, χ² analysis was used to compare proportions of responding cells in each treatment group. In all cases, 2-tailed tests were employed, with statistically significant differences indicated at P < 0.05, P < 0.01, and P < 0.001. In the figures, asterisks (*) are employed to note significance for tests indicating the influence of loading, and plus symbols (+) are employed for comparisons between wild-type (WT) and ORPK groups.

RESULTS

Characterization of WT and ORPK transgenic chondrocytes

Western blots from conditionally immortalized chondrocyte lysates confirmed that under permissive conditions (IFN-γ, 33°C), WT and ORPK cells expressed the SV40 large T antigen, which was no longer detectable when cells were switched to nonpermissive conditions (37°C) for 3 d (Fig. 1A). In monolayer culture, 85 ± 5% of WT cells expressed primary cilia, which was reduced to 20 ± 8% when cells were cultured in agarose constructs for 24 h. In both monolayer and agarose cultures, the hypomorphic mutation of IFT88 in ORPK chondrocytes caused disruption of ciliogenesis, resulting in disorganized cytoplasmic α-acetylated tubulin and stunted or absent cilia (Fig. 1B, C). Both WT and ORPK cells showed a rounded morphology, similar to primary bovine chondrocytes, in 3-dimensional (3D) culture (Fig. 1C), and viability was >80% during all experiments. No statistically significant difference was found in chondrocyte proliferation between WT and ORPK cells cultured for 24 h in agarose, as determined by DNA quantification.

Figure 1.

Deletion of IFT88 results in loss of primary cilia expression in ORPK chondrocytes but does not affect expression of cartilage-specific matrix molecules aggrecan and collagen type II. A) Western blot shows loss of SV40 at 37°C in the absence of INF-γ after 5 d in nonpermissive conditions. B) Acetylated α-tubulin labeling of primary cilia (red) in WT monolayer chondrocytes. Stunted or no cilia were detected in ORPK cells. Nuclei (blue) are labeled with DAPI. C) Acetylated α-tubulin labeling of primary cilia of primary bovine and WT mouse chondrocytes in 3D agarose constructs (red; arrows) and γ-tubulin labeling of centrioles (green; arrowheads). Cilia were either stunted or absent in ORPK cells (red; arrow). D) RT-PCR (gel inset) and real-time qPCR (graph) show aggrecan and collagen type II gene expression in both WT and ORPK cells. ORPK aggrecan mRNA was significantly up-regulated compared to WT cells. Relative expression ratios were logged to the base 2 and plotted as individual points. Error bars = sd. Scale bars = 10 μm. ⁺⁺⁺P < 0.001.

Chondrocyte phenotype, for both WT and ORPK cells, was confirmed by expression of aggrecan and collagen type II mRNA, measured by RT-PCR (Fig. 1D). No significant difference was found in collagen type II expression between WT and ORPK cells, whereas aggrecan expression was 5.2-fold greater in ORPK cells relative to WT (P<0.001; Fig. 1D). In contrast, sGAG synthesis was reduced by 33% in ORPK cells (see Fig. 2B) compared with WT cells (P<0.001).

Figure 2.

Primary cilia are required for mechanosensitive up-regulation of proteoglycan synthesis. Cyclic compression (1Hz, 0–15%) significantly up-regulates aggrecan gene expression (A) and sGAG synthesis (B) in WT but not ORPK chondrocytes. mRNA was quantified after 1 h loading by real-time qPCR and normalized to unloaded controls. sGAG synthesis was measured after 24 h of loading. Bars represent mean ± sd values (n=36 constructs). Experiments were repeated 3 times. *P < 0.05. ⁺⁺⁺P < 0.001 vs. corresponding WT.

Primary cilia are required for mechanically induced up-regulation of matrix synthesis

Following 1 h of cyclic compression, qPCR analysis showed that aggrecan mRNA expression in WT cells significantly increased by 2.2 fold (P<0.05), whereas no increase was found in ORPK cells (Fig. 2A). No significant change was found in collagen type II expression, such that mechanical compression produced mean fold changes of 0.76 and −0.66 in WT and ORPK chondrocytes, respectively. In WT cells, cyclic compression induced a significant 12% increase in sGAG synthesis (P<0.05; Fig. 2B). This response was completely abolished in ORPK cells, indicating that cilia are required for mechanosensitive proteoglycan synthesis (Fig. 2B). No significant difference was found in cell proliferation, as indicated by negligible changes in DNA content with loading of 0.5 and 0.7% in WT and ORPK chondrocytes, respectively.

Mechanical loading activates ATP release in both WT and ORPK chondrocytes

Our previous studies have shown that primary bovine chondrocytes release ATP through unopposed connexin hemichannels when subjected to compressive loading (35), which, in turn, triggers mechanically induced up-regulation of proteoglycan synthesis (3). Therefore, we next examined whether this response was cilia dependent by measuring cumulative ATP release over a 1-h period from agarose constructs seeded with either WT or ORPK chondrocytes. Mean ATP release into the medium for unloaded WT and ORPK cells was 4.63 ± 0.65 and 3.15 ± 0.26 nM/construct, respectively, with no significant difference between the two cell types. Compressive loading induced statistically significant increases in ATP release in both WT and ORPK cells (P<0.01). No statistically significant difference was found in the magnitude of the increase between WT and ORPK cells, with mean increases of 42 and 34%, respectively. This finding suggests that the mechanosensitive ATP release mechanism was not affected by loss of cilia (Fig. 3A).

Figure 3.

Primary cilia are not required for mechanosensitive ATP release but are required for subsequent ATP-induced Ca²⁺ signaling. A) One hour of cyclic compression (1 Hz, 0–15%) up-regulates ATP release in both WT and ORPK chondrocytes (n=36 constructs), B) Compression induces a significant increase in the percentage of cells exhibiting Ca²⁺ transients in WT cells but not in ORPK cells (n=6 from a total of 150 cells/group). Mechanically induced Ca²⁺ signaling was inhibited by the P2 receptor antagonist suramin in WT cells, indicating that the Ca²⁺ response is due to mechanosensitive ATP release. C) WT cells responded to exogenous application of 100 μM ATP with increased Ca²⁺ signaling, whereas ORPK cells were completely unresponsive (n=3 from a total of 150 cells/group). D) Three representative Ca²⁺ traces from an unstimulated WT cell and cells exposed to either load or ATP. Transients in response to load or ATP were significantly shorter, with reduced rise and fall times when compared to spontaneous transients. Bars represent means ± sd. NS, not significant. *P < 0.05, **P < 0.01, ***P < 0.001.

Mechanically induced Ca²⁺ signaling is abolished in ORPK cells

Previous studies have shown that chondrocytes exhibit mechanosensitive Ca²⁺ signaling (32, 33), which is mediated by ATP release (36). We therefore examined whether these events involved the primary cilium. In unloaded agarose constructs, spontaneous intracellular Ca²⁺ transients occurred in 15–25% of both WT and ORPK cells, with no significant difference between the two cell types. Compressive loading significantly increased the percentage of WT cells exhibiting Ca²⁺ transients to 30–40%, (P<0.05; Fig. 3B). However, the number of Ca²⁺ transients in ORPK loaded cells was significantly lower than both unloaded ORPK controls (P<0.05) and WT loaded cells (P<0.001; Fig. 3B). Temporally, transients observed following load were significantly shorter with a reduction in both transient rise and fall time. These data indicate that cilia are essential for mechanosensitive activation of Ca²⁺ in response to compression.

Loss of ATP-mediated Ca²⁺ signaling in ORPK cells

Given that both ciliated and nonciliated cells release ATP in response to load, we then investigated whether the Ca²⁺ response to load was mediated by ATP release and subsequent P2 receptor activation by subjecting WT and ORPK cells to compression in the presence of suramin, a nonspecific P2 receptor antagonist. We found that suramin reversed the mechanosensitive Ca²⁺ response in WT cells (P<0.05) but had no significant effect on ORPK cells (Fig. 3B). We next examined whether WT and ORPK could elicit purinergic-induced Ca²⁺ signaling and found that neither cell type showed any response to the addition of 1 μM ATP. However, WT cells showed a robust Ca²⁺ response to exogenous 100 μM ATP, which was inhibited by suramin. However, ORPK cells did not respond to 100 μM ATP (Fig. 3C), which suggests that loss of chondrocyte primary cilia prevents P2 receptor-activated Ca²⁺ signaling.

ORPK cells show normal P2 receptor expression but disrupted polycystin expression

To investigate the abnormal purine reception in ORPK cells, we examined the expression profiles of human and bovine P2 receptors in chondrocytes (38, 39). Western blots showed that both WT and ORPK expressed similar levels of P2X4, P2X7, P2Y1 and P2Y2 (Fig. 4A, B). P2X2 was also examined but was not detectable in either cell type. To investigate the role of the cilium in P2-mediated calcium signaling further, we also examined the expression profiles of PC1 and PC2, which have been linked functionally to purinergic Ca²⁺ signaling (40). Western blots showed that WT cells expressed the full-length PC1 protein at ∼350 kDa, which was completely absent in ORPK cells. In contrast, ORPK cells expressed a smaller band at ∼35 kDa, most likely to represent the C-terminal tail (Fig. 4C). Immunofluorescence showed that PC1 was located on the cell membrane and within the cytoplasm as puncta in both WT and ORPK chondrocytes (Fig. 4C). However, unlike in epithelial cells, PC1 was not detected on WT primary cilia. This was verified in both monolayer and 3D cultures. Western blots showed that PC2 was expressed in WT and ORPK cells at similar levels (Fig. 4D) and was located on the cell membrane and throughout the cytoplasm in both cell types and on primary cilia of WT cells.

Figure 4.

P2, PC1, and PC2 expression in WT and ORPK cells. A) Western blots of P2 receptors, P2X4, P2X7, P2Y1, and P2Y2. B) Quantification of P2 receptor expression reveals no significant differences between WT and ORPK chondrocytes. Data were normalized to β-actin. Bars represent means ± sd (n=3). C) Western blot shows full-length PC1 in WT cells, but only a small (∼35 kDa) cleaved product is present in ORPK cells. Confocal extended focus images show that PC1 (green) is present as small bright puncta within the cystoplasm and on the cell membrane in both WT and ORPK cells and is not detectable on any WT primary cilia (arrow). D) Western blot shows that PC2 expression is unaffected by ORPK mutation. Immunofluorescent labeling of PC2 (green) shows expression throughout the cell and localization to the primary cilium in WT cells (yellow). In ORPK cells, PC1 and PC2 were detected within the cytoplasm only. In both cases, primary cilia were labeled with anti-acetylated α-tubulin (arrow, red).

DISCUSSION

A wide variety of cell types across evolution exhibit mechanosensitive behavior, and increasing evidence indicates that primary cilia are key mechanosensory organelles (41). In articular cartilage, mechanical loading is essential for development, health, and homeostasis through the control of ECM synthesis and catabolism. Here we show that the chondrocyte primary cilium is a fundamental part of this mechanotransduction mechanism, regulating matrix synthesis in response to the physiological compressive strain.

Chondrocytes derived from Tg737/IFT88 (ORPK) mice provided an appropriate model to investigate the role of chondrocyte primary cilia in response to compressive loading. These chondrocytes exhibit stunted or no cilia, with elevated and disorganized levels of acetylated α-tubulin, as seen in situ (16, 42). By contrast, WT cells exhibited primary cilia with a frequency of 20% in agarose (37). The 3D agarose culture system maintained chondrocytic phenotype in a rounded morphology with expression of aggrecan and collagen type II. Interestingly, the ORPK cells showed a significantly greater level of aggrecan gene expression compared to WT cells, but this was not translated into an increase in sGAG synthesis, which was reduced compared to WT cells. It is unclear how loss of IFT should down-regulate proteoglycan synthesis, although previous studies report a similar loss of cartilage proteoglycan content in Bardet-Biedl syndrome mice that have reduced cilia expression (43). In addition, given the known tubulin defects within ORPK cells (16, 42, 44), it may be due to defects in secretion via the tubulin network (45). The agarose model also enabled the application of cellular compressive strain in the physiological range reported for articular chondrocytes (46, 47). Due to the absence of a fully developed negatively charged ECM in this model, it is unlikely that compression induced changes in osmotic pressure, electrical streaming potentials, or pH, all of which have been linked to cartilage mechanotransduction (1, 2). Similarly, the agarose model does not include the direct linkage between the primary cilium and the pericellular matrix, as shown in situ where integrins are expressed on the cilium (25). Nevertheless, gross cyclic compressive loading, 1 Hz and 0–15% strain, was shown here to stimulate synthesis of proteoglycan in WT cells (Fig. 2B), in agreement with numerous previous studies using primary bovine chondrocytes (3, 4, 30). This response was completely abrogated in ORPK cells, indicating that the primary cilium is required for chondrocyte mechanotransduction.

Using identical cyclic loading of primary chondrocytes, we, and others, have shown that the mechanosensitive stimulation of proteoglycan synthesis is mediated by mechanically induced release of ATP (3, 5, 35, 48). ATP release is an early mechanotransduction event, whereby ATP binds to P2 receptors activating intracellular Ca²⁺ signaling (36). When ATP is not utilized by the cell, it is quickly hydrolyzed to pyrophosphate (48). In the present study, we also show mechanically induced ATP release (Fig. 3A). This purinergic-Ca²⁺ mechanotransduction pathway is similar to that reported for kidney epithelial cells subjected to fluid shear, where the ATP release is mediated by deflection of the primary cilium (22, 49, 50). However, in contrast to epithelial cells, the present study shows that mechanosensitive release of ATP occurs to a similar extent in both WT and ORPK chondrocytes, suggesting that primary cilia are not required for this part of the chondrocyte mechanotransduction pathway. It is clear that ATP-mediated Ca²⁺ signaling occurs within seconds of applied loading; however, the lack of sensitivity of the ATP assay required measurements to be made from medium collected after a 1-h period to allow for cumulative release and diffusion. Therefore, the local concentrations of pericellular ATP in response to load are likely to be much greater that the nanomolar concentrations measured in the surrounding media. This is further supported by the finding that Ca²⁺ signaling in WT cells was activated by 100 μM ATP but 1 μM ATP had no effect.

Primary cilia were necessary, however, to respond to extracellular ATP. Either compression or addition of 100 μM ATP activated Ca²⁺ signaling in WT but not in ORPK chondrocytes. The role of the calcium and the cilium in chondrocytes is in contrast to studies in bone cells, where the cilium is critical to mechanotransduction independently of intracellular Ca²⁺ (51). The involvement of P2 receptors was confirmed by treatment with the P2 receptor inhibitor suramin, which abolished the Ca²⁺ signaling responses to both compression and ATP in WT cells (Fig. 3B, C). Indeed, both apyrase and suramin have previously been shown to abolish downstream mechanically induced proteoglycan synthesis in chondrocyte-agarose constructs (35). The characteristics of both load-induced and ATP-induced Ca²⁺ transients were similar to those reported in previous studies (36). Furthermore, the nature of the load-induced transients has been shown previously to involve both extracellular and intracellular sources of Ca²⁺ (36). This finding suggests involvement of both P2X and P2Y receptors, which facilitate extracellular Ca²⁺ entry and release of intracellular Ca²⁺ stores, respectively. However, considerable crosstalk occurs between Ca²⁺ pathways with the involvement of mechanisms such as calcium-induced calcium release and store-operated calcium channels. Interestingly, WT and ORPK cells expressed the range of P2 receptors (P2X4, P2X7, P2Y1, and P2Y2) found in cartilage (33). Taken together, these data suggest that the primary cilium is necessary for ATP reception but does not modulate the expression of these receptors.

In other cell types, heterologous expression of the cilia protein PC1 in both its full and cleaved forms has also been found to influence Ca²⁺ release (52–55), including ATP-induced Ca²⁺ signaling (40). Moreover, PC1 and PC2 are thought to interact and reciprocally influence cellular and ciliary localization (56, 57). In the present study, we show that while ATP receptors are unaffected, full-size PC1 expression is disrupted in ORPK cells, leaving only a C-terminal cleavage product. Although not examined in this study, this finding may explain the lack of response to ATP, which is of additional significance given the roles for the polycystins in mechanostransduction and cilia-mediated Ca²⁺ signaling in other cells types (52, 58, 59).

The present study therefore demonstrates for the first time that primary cilia are essential for chondrocyte mechanotransduction and the control of ECM secretion in response to physiological compressive strain. In particular, the chondrocyte primary cilium is required for ATP-induced Ca²⁺ signaling. Thus, in summary, although the mechanosensory role of this highly conserved organelle appears to be ubiquitous, the signaling pathway differs significantly from that found in epithelial cells and osteoblasts exposed to pure fluid shear. This finding suggests that primary cilia mechanotransduction is specialized to the tissue-specific biomechanical environment.