SEARCH

SEARCH BY CITATION

Keywords:

  • RUNX2;
  • SOX9;
  • PROTEIN DEGRADATION;
  • PROTEASOME;
  • LYSOSOME

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Mesenchymal stem cell differentiation is controlled by the cooperative activity of a network of signaling mechanisms. Among these, RUNX2 and SOX9 are the major transcription factors for osteogenesis and chondrogenesis, respectively. Their expression is overlapped both temporally and spatially during embryogenesis. Here we have demonstrated that RUNX2 and SOX9 physically interact in intact cells and have confirmed that SOX9 can inhibit the transactivation of RUNX2. In addition, RUNX2 exerts reciprocal inhibition on SOX9 transactivity. In analyses of the mechanism by which SOX9 regulated RUNX2 function, we demonstrated that SOX9 induced a dose-dependent degradation of RUNX2. Although RUNX2 is normally degraded by the ubiquitin-proteasome pathway, we found that SOX9-mediated degradation was proteasome-independent but phosphorylation-dependent and required the presence of the RUNX2 C-terminal domain, which contains a nuclear matrix targeting sequence (NMTS). Furthermore, SOX9 was able to decrease the level of ubiquitinated RUNX2 and direct RUNX2 to the lysosome for degradation. SOX9 also preferentially directed β-catenin, an intracellular mediator of canonical Wnt signaling, for lysosomal breakdown. Consequently, the mechanisms by which SOX9 regulates RUNX2 function may underlie broader signaling pathways that can influence osteochondrogenesis and mesenchymal fate. © 2010 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Endochondral ossification is an orchestrated series of events that includes mesenchymal stem cell (MSC) condensation, chondrogenesis of the aggregated MSCs, perichondrium and periosteum formation, and hypertrophy of the chondrocytes and replacement of the cartilage by bone with a mineralized extracellular matrix. It is believed that transcriptional networks tightly control this process. SOX9 is a major transcriptional activator of chondrogenesis1, 2 and can target chondrocyte-specific genes, including Col2a1,3–5Col11a2,6 and aggrecan.7 The malfunction of SOX9 results in campomelic dysplasia (CMD1), which is a semilethal osteochondrodysplasia, characterized by skeletal anomalies that include bending of the long bones and XY sex reversal.8, 9 RUNX2 is the key transcription factor for osteogenesis and can bind to the osteoblast-specific cis-acting element (OSE2) in the promoter of osteocalcin, Col1a1, bone sialoprotein (BSP), and osteopontin.10 RUNX2 protein consists of several functional modules: the glutamine-arginine-rich domain (QA), the DNA-binding runt homology domain (RHD), the nuclear localization signal (NLS), the nuclear matrix target signal (NMTS), and the transducin-like enhancer of split/groucho-interacting C-terminal pentapeptide VWRPY.11 The C-terminus of RUNX2, which is around the NMTS motif, has been reported to contribute the interaction with chaperone protein C-terminus of Hsc70-interacting protein (CHIP).12 Abnormalities in RUNX2 function can lead to cleidocranial dysplasia (CCD), in which there is defective ossification of the cranial bones with large fontanels and delayed closing of the sutures with complete or partial absence of the clavicles.13–19 Mice with homozygous mutations in RUNX2 cannot survive after birth owing to breathing defects caused by the absence of ossification of the ribs.20, 21 The functions of RUNX2 and SOX9 are controlled by transcriptional and posttranscriptional regulation, including ubiquitination, phosphorylation, and acetylation.12, 22–30 The temporal and spatial overlap of RUNX2 and SOX9 expression during embryonic development indicates a close coordination between these two key transcription factors during bone formation.1, 10, 31 Indeed, it has been found that SOX9 exerts a dominant function over RUNX2 in mesenchymal precursors.32 RUNX2 regulates chondrocyte maturation and also chondrocyte hypertrophy,33–35 which suggests that RUNX2 suppresses SOX9 at some stage during skeletal development. Here we determined the nature of RUNX2-SOX9 interactions and the mechanism by which SOX9 regulates RUNX2 function.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

DNA constructs

The full-length cDNA of mouse RUNX2 (NCBI Accession Number NM_009820.2) and mouse SOX9 (NCBI Accession Number NM_011448.3) were amplified by polymerase chain reaction (PCR) from a cDNA library of MC3T3-E1 cells and cloned into pRK5 or pECFP-N1/pEYFP-N1 expression vector. The mutant RUNX2 construct, SA-RUNX2, was generated using a Stratagene QuickChange site-directed mutagenesis kit to replace amino acid 540 (corresponding to serine 472 in human RUNX2 sequence) with arginine. Truncated RUNX2, RUNX2-N, containing amino acids 1 to 429, was generated by PCR and subcloned into pRK5. Human osteocalcin promoter (1–1727) was amplified from genomic DNA of HEK293 cells and cloned into pGL4.21 luciferase reporter vector (Promega, Madison, WI, USA) and named as OC-Luc.

Cell culture and transfections

HEK293 and C3H10T1/2 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) and supplemented with 10% fetal calf serum (FBS), 100 U/mL of penicillin, and 100 µg/mL of streptomycin. The prechondrogenic cell line ATDC5 was maintained in DMEM and Ham's F-12 medium containing 5% FBS, 100 U/mL of penicillin, and 100 µg/mL of streptomycin. To isolate human primary chondrocytes, articular cartilage from the femoral heads retrieved from hip-joint replacements following informed consent were removed and minced followed by digestion for 8 hours in DMEM and Ham's F-12 medium containing 5% FBS and 400 U/mL of type 2 collagenase (17101–015; Invitrogen, Paisley, Renfrewshire, UK). The resulting cell suspension was passed through a 100-µm mesh and collected by centrifugation and resuspended in DMEM and Ham's F-12 medium containing 10% FBS. Chondrocytes were cultured as monolayer or pellet. For the pellet culture, 1-mL aliquots of suspension of chondrocytes at 5 × 105 cells/mL were transferred to 20-mL centrifugation tubes. Pellets were formed by centrifugation at 10,000g for 10 seconds. Loosely capped tubes were transferred to a standard tissue culture incubator. To induce the chondrocytes hypertrophy, the GSK-3β inhibitor 6-bromo-indirubin-3'-oxime (BIO) was applied at 2.5 µM for 7 days. Then cell pellets were embedded in 22-oxa-1,25-dihydroxyvitamin D3 (OCT) and followed by cryosectioning for immunofluorescence (IMF). DNA plasmids were transiently transfected using the lipofectamine 2000 reagents (Invitrogen) for HEK293, C3H10T1/2, and ATDC5. The Neon Transfection System (Invitrogen) was used to transfect primary chondrocytes (1650 V, 20 ms, 1 pulse). Empty vector was used to maintain the total amount of DNA transfected constant.

Immunofluorescence

Monolayers of cells were grown on coverslips or cryosections of pellet culture of chondrocytes collected on slides were fixed in 4% paraformaldehyde (PFA) in PBS for 10 minutes at room temperature and then washed three times in PBS and blocked (PBS containing 0.1% Triton X-100, 10% serum) for 30 minutes. Subsequent washes and antibody dilutions were performed in block solution without Triton X-100. Samples were incubated with primary antibody at 4°C overnight, then were washed in PBS three times, and incubated in fluorescent secondary antibodies for 30 minutes at room temperature, followed by three times washes in PBS, and then they were mounted for confocal microscopy.

Acceptor photobleaching fluorescent energy resonance transfer (FRET)

RUNX2-ECFP and SOX9-EYFP were transfected into C3H10T1/2 cells individually or together in a Lab-Tek chambered coverglass system (155411, Nalge Nunc International, Rochester, NY, USA). After 48 hours, acceptor photobleaching FRET was performed on a Zeiss LSM510 META System (Carl Zeiss, Inc., Welwyn Garden City, Hertfordshire, UK) equipped with a 63 × 1.4 Plan-Apocharomat oil-immersion object. The laser was used to excite ECFP at 405 nm and EYFP at 514 nm, and emission signals were recorded in META detector λ and unmixed detection mode. EYFP signals were bleached using a 514-nm laser at 100% power for 100 iterations, and time-series images were captured for both before (3 times) and after (12 times) bleaching. To construct the FRET tracing curve, the intensities of the ECFP and EYFP after bleaching were normalized by the average prebleached fluorescence intensities of ECFP and EYFP. Multiple cells (5 to 10) from at least three experiments were measured for each construct.

Luciferase assay

The OC-Luc reporter plasmid was cotransfected with RUNX2 expression plasmid and different amounts of SOX9 expression plasmid into HEK293 cells. The 4Col2-Luc reporter plasmid was cotransfected with SOX9 expression plasmid and different amounts of RUNX2 plasmid into HEK293 cells. The 4Col2-Luc reporter plasmid also was cotransfected with increasing amounts of RUNX2 plasmid into ATDC5 cells following application of chondrogenic factors (20 ng/mL of transforming growth factor β3, 50 µg/mL of L-ascorbic acid-2-phosphate, 100 nM of dexamethasone, 40 µg/mL of 1% L-proline ITS+1) for 48 hours. All transfections were performed with phRL-CMV to normalize transfection efficiencies. All data are shown as fold activation relative to the activity obtained with the empty-vector plasmids. Luciferase activity was measured using a Promega Dual Luciferase Reporter Assay Kit (Promega) 48 hours after transfection.

Western blot analysis

Cells were lysed with RIPA buffer containing protease inhibitor cocktail 3 (Calbiochem, Nottingham, UK), and cell lysates were separated electrophoretically on an 8% to 10% SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane (Amersham Biosciences, GE Healthcare Life Sciences, Little Chalfont, UK). The membrane was blocked with 5% milk in Tris-buffered saline Tween-20 (TBST) for 1 hour at room temperature and incubated with anti-RUNX2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-SOX9 (Abcam, Cambridge, UK), anti-GAPDH (Genetex, Nottingham, UK) or anti-β-catenin (Sigma-Aldrich, Dorset, UK) primary antibody for 1 hour at room temperature. The membrane then was washed and incubated with horseradish peroxidase–conjugated antirabbit or antimouse secondary antibody for 1 hour at room temperature. Immunoreactive bands were visualized by ECL (Amersham Biosciences). The intensity of protein bands was quantitated using Image J software 1.43 (NIH, available as freeware from http://rsbweb.nih.gov/ij/).

Ubiquitination assay

HEK293 cells were transfected as indicated, and after 42 hours, the cells were pretreated with 40 µM of MG-132 or 50 mM of NH4Cl for 6 hours before being harvested. Cells were lysed in buffer I (6 M guanidinium/HCl, 93 mM Na2HPO4, 6.8 mM 0.01 M Tris, pH 8), and genomic DNA was disrupted by sonication. His-tagged protein was purified with Ni-NTA beads (GE Healthcare Life Sciences) for 4 hours at room temperature after washing with buffer I and buffer II (8 M Urea, 22 mM Na2HPO4, 78 mM NaH2PO4, 0.01 M Tris, pH 6.3) containing 10 mM of imidazole. The beads were heated with loading buffer and then subjected to Western blot against preferred antibodies.

Statistical analysis

Statistical analysis was performed in SPSS16 (SPSS, Inc., Chicago, IL, USA). Initially variance analysis (ANOVA) was performed, followed by Student's t test (unpaired). Significance was assumed at the level of p < 0.05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

RUNX2 and SOX9 physically associate with each other in intact cells

Direct interaction between RUNX2 and SOX9 was reported previously using glutathione-S-transferase (GST) pull-down assays.32 To determine their interaction in intact cells, we performed acceptor photobleaching FRET in C3H10T1/2 cells. First, to show the localization of overexpressed RUNX2 and SOX9, we transfected EYFP-tagged RUNX2 and SOX9 plasmids into C3H10T1/2 cells and plated the cells on coverslips. The cells were fixed in 4% PFA 48 hours after transfection, and the nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI, 1 µg/mL), followed by confocal microscopy. EYFP rather that ECFP was used to tag both RUNX2 and SOX9 because of the difficulties in distinguishing between ECFP and DAPI, which have the close excitation wavelengths. We found that RUNX2 and SOX9 localized mainly to the nuclear matrix, which is the typical distribution of transcription factors. (Fig. 1A). Then we applied the combination of RUNX2-ECFP and SOX9-EYFP to perform acceptor photobleaching FRET. We found that after photobleaching the acceptor SOX9-EYFP using a 514-nm laser (middle column of Fig. 1B), the emission intensity of donor RUNX2-ECFP increased steadily by excitation with the 405-nm laser (Fig. 1B, left column). The recorded intensities of ECFP and EYFP after photobleaching were normalized by the average of respective intensities before bleaching and were shown as a FRET curve (Fig. 1C). It clearly showed that about 50% of acceptor bleaching caused a 50% to 100% intensity increase in the donor molecule, which strongly indicated the high FRET efficiency and physical association between SOX9 and RUNX2. To exclude overexpression artifacts, photobleaching was performed using combinations of empty ECFP and EYFP vectors as controls; the emission intensity of ECFP was maintained at same level throughout (data not shown).

thumbnail image

Figure 1. Subcelluar localization of SOX9 and RUNX2. (A) SOX9-EYFP and RUNX2-EYFP were transfected, respectively, into C3H10T1/2 cells. Then, 48 hours after transfection, cells were fixed and followed by DAPI staining to show the nuclei by the confocal microscopy. Scale bar = 10 µm. (B) Acceptor photobleaching FRET between RUNX2-ECFP and SOX9-EYFP. Both the expression constructs were transfected into C3H10T1/2 cells separately or together. Then, 24 hours after transfection, acceptor photobleaching FRET measurements were performed with RUNX2-ECFP (donor) and SOX9-EYFP (acceptor). Representative images showing the region of interest (ROI) for photobleaching and the enlarged RIO are presented at the opposite corner to compare the changes of intensities after photobleaching. Scale bar = 10 µm. (C) Quantitative analysis of FRET changes before and after photobleaching during the time course. The fluorescence intensities were normalized by the average intensity before photobleaching. Error bars represent ± SEM. n = 10 in all groups.

Download figure to PowerPoint

RUNX2 and SOX9 decrease transcriptional activation of each other

To confirm that SOX9 decreased the transcriptional activation of RUNX2, SOX9 and RUNX2 were cotransfected with a RUNX2-responsive luciferase reporter, in which the osteocalcin promoter drives firefly luciferase expression, into HEK293 cells that express negligible endogenous SOX936 and do not express RUNX2.37 We confirmed that SOX9 had a strong dose-dependent inhibitory effect on RUNX2-induced reporter activity using a promoter reporter containing one copy of osteoblast-specific cis-acting elements 2 (OSE2)32 (Fig. 2A). To determine reciprocal regulatory effects of RUNX2 on SOX9 function, we used a sensitive 4Col2E-Luc reporter,38 which contained four tandem 48-bp chondrocyte-specific enhancer segments of Col2a1. This demonstrated that RUNX2 had a significant dose-dependent inhibitory effect on SOX9-induced type II collagen promoter activity (Fig. 2B).

thumbnail image

Figure 2. Potent transactivity inhibition between RUNX2 and SOX9. (A) HEK293 cells were transfected with the RUNX2-responsive osteocalcin reporter OC-luc and expression plasmids for RUNX2 and SOX9 in increasing amounts. Then, 48 hours after transfection, luciferase assays were performed, and the fold changes of promoter activities were normalized to the basal activity of the promoter. (B) HEK293 cells were transfected with the SOX9-responsive Collagen2 reporter 4Col2E-luc and expression plasmids for SOX9 and RUNX2 in increasing concentrations. Then, 48 hours after transfection, luciferase assays were performed, and the fold changes of promoter activities were normalized to the basal activity of the promoter. (C) RUNX2 expression plasmid was cotransfected with different amount of SOX9 expression plasmid into HEK293 cells, as shown in part A (* indicate the same amount of SOX9 plasmid transfected). Western blot analysis was performed 48 hours after transfection. (D) SOX9 expression plasmid was cotransfected with different amount of RUNX2 expression plasmid into HEK293 cells, as shown in part B (◊ indicate that same amount of RUNX2 plasmid transfected). Western blot analysis was performed 48 hours after transfection. (E) RUNX2 affects the nuclear expression of endogenous SOX9 in chondrocytes. Primary human articular chondrocytes were transfected with RUNX2-EYFP, and IMF for SOX9 was performed 24 hours after transfection. Nuclei were counterstained with DAPI. Open arrow and closed arrow indicate the expression of SOX9 in the nuclei of cells with or without RUNX2 expression, respectively. Scale bar = 10 µm (F) ATDC5 prechondrogenic cells were transfected with the SOX9 response Collagen2 promoter 4Col2E-luc reporter plasmid with increased amounts of RUNX2 expression plasmid (0.3, 1, and 3 µg). After transfection, chondrogenic factors were applied for 48 hours followed by luciferase assay.

Download figure to PowerPoint

Reciprocal regulation of RUNX2 and SOX9 protein stability by different mechanisms

To identify the mechanism of transactivity inhibition on each other, Western blot analyses were performed to determine protein expression levels of RUNX2 and SOX9 when coexpressed or expressed alone in parallel with the promoter reporter assays. Using a threefold increased concentration of SOX9 plasmids and the same amount of RUNX2 plasmid, we found that SOX9 strongly promoted the degradation of RUNX2 in HEK293 cells in a dose-dependent manner (Fig. 2C; also compare lane 5 with lane 6 in Fig. 2D for RUNX2, labeled ◊). Conversely, we found that RUNX2 stabilized/enhanced SOX9 rather than promoted its degradation (Fig. 2D). The stabilizing effects of RUNX2 on SOX9 are also clearly shown in Fig. 2C (also compare lanes 5 and 6 in the top panel, labeled *). These results suggested that SOX9 inhibited the transactivity of RUNX2 by promoting its degradation, whereas RUNX2 inhibited the transactivity of SOX9 by other mechanisms.

RUNX2 affects the nuclear expression of SOX9 in chondrocytes

The results of whole-cell lysate Western blotting indicated that the total protein of SOX9 was enhanced by RUNX2, but its transactivity was inhibited by RUNX2 in HEK293 cells. To check whether RUNX2 can affect the nuclear expression of SOX9, we transfected EYFP-tagged RUNX2 into human primary articular chondrocytes, followed by IMF for endogenous SOX9. We found that the nuclear expression of SOX9 was decreased in the cells transfected with RUNX2 (compare the expression level indicated by arrows in Fig. 2E), which indicates that RUNX2 may block the nuclear translocation of SOX9 or promote the degradation of SOX9 in the nucleus.

RUNX2 inhibits SOX9 transactivity during chondrogenesis in ATDC5 cells

To determine whether RUNX2 can inhibit the transactivity of endogenous SOX9 during chondrogenesis, we performed the transactivity assay in a prechondrogenic mouse cell line, ATDC5, which expresses endogenous SOX9 during chondrogenesis. We found that the transfection of RUNX2 caused an up to twofold decrease of 4Col2-Luc in a dose-dependent manner (Fig. 2F). The inhibition of RUNX2 on endogenous SOX9 is not as significant as that on the overexpressed SOX9 (compare with Fig. 2B), and the low level of SOX9 may be part of the reason during the initiation of the chondrogenesis of ATDC5.39, 40

SOX9-induced RUNX2 degradation is ubiquitin-proteasome-independent

Since RUNX2 is degraded through the ubiquitin-proteasome pathway,22, 25, 41 we hypothesized that SOX9 promoted RUNX2 degradation through this mechanism. To test this hypothesis, the proteasome inhibitor MG13242 was applied to HEK293 cells in which we overexpressed RUNX2, SOX9, or both. We found that MG132 stabilized RUNX2 when expressed alone (compare lane 6 with lane 2 in Fig. 3A for RUNX2 expression), which is consistent with the previous founding that RUNX2 is degraded through proteasomes. In addition, with the presence of SOX9, MG132 can stabilize RUNX2 (compare lane 8 with lane 4 in Fig. 3A). Importantly, SOX9 can promote the degradation of RUNX2 in both the absence and the presence of MG132 (compare lane 4 with lane 2 and lane 8 with lane 6, middle panel, of Fig. 3A). By comparing the quantified expression level of RUNX2 when expressed alone or with the expression of SOX9, we confirmed that SOX9 promoted the degradation of RUNX2 both in the absence and in the presence of MG132, and therefore, the degradation potency of SOX9 on RUNX2 was not affected by inhibition of proteasomal degradation (Fig. 3C, black arrow). In both cases, RUNX2 expression levels decreased by around 50% with the coexpression of SOX9 (Fig. 3C). This implied that the process of SOX9-induced RUNX2 degradation is proteasome-independent. However, replacement of a phosphorylation site serine with arginine in RUNX2 (S540A-RUNX2)23 partly blocked the effects of SOX9 on RUNX2 degradation (compare lane 2 with lane 4 in Fig. 3B). The expression level of SA-RUNX2 decreased only 17% by SOX9, but that of wild type RUNX2 was reduced 49% by SOX9 (compare the histograms on the left in Fig. 3C and D). MG132 abolished the effect of SOX9 on the degradation of SA-RUNX2 (histogram on the right in Fig. 3D). Phosphorylation assays have shown that strong phosphorylation was detected for wild-type (WT) RUNX2 by the incubation of recombinant proteins with cyclin D1-Cdk4 enzymes, but only a weak phosphorylation was detected for SA-RUNX2.23 These findings implied that the SOX9-induced RUNX2 degradation is a phosphorylation-dependent process and that the proteasome pathway may be involved in the regulation of SOX9 on the SA-RUNX2. Interestingly, SOX9 levels were stabilized/enhanced by both WT-RUNX2 and SA-RUNX2 in the presence and absence of MG132 (Fig. 3A, B, top panels).

thumbnail image

Figure 3. SOX9-induced RUNX2 degradation is ubiquitin-proteasome-independent. (A) HEK293 cells were transfected with SOX9, RUNX2, or both. Then, 42 hours after transfection, cells were treated with MG132 (40 µM) or vehicle (DMSO) for 6 hours, followed by Western blot analysis. (B) HEK293 cells were transfected with SOX9, SA-RUNX2, or both. Then, 42 hours after transfection, cells were treated with MG132 (40 µM) or vehicle for 6 hours, followed by Western blot analysis. (C) Protein levels of RUNX2 were quantified by densitometry (mean ± SD from three independent experiments; the results of one experiment are shown in part A). The data are shown as the fold change to the mean density of lane 4 in the middle panel in part A. (D) Protein levels of SA-RUNX2 were quantified by densitometry (mean ± SD from three independent experiments; the results of one experiment are shown in part B). The data are shown as the fold change to the mean density of lane 4 in the middle panel of part B.

Download figure to PowerPoint

SOX9 decreases the ubiquitination of RUNX2

To confirm the lack of involvement of ubiquitin-proteasome pathway in SOX9-mediated degradation of RUNX2, a 6× histidine-tagged ubiquitin plasmid (Ub-His) was used for in vitro ubiquitination analysis in HEK293 cells. In the presence of MG132, we found that SOX9 induced a dramatic loss of ubiquitinated RUNX2 (Fig. 4A; compare lanes 6 and 8). In the absence of MG132, most of the ubiquitinated RUNX2 was degraded in the proteasome (compare lane 6 in Fig. 4A, B), but it is still noticeable that SOX9 decreased the RUNX2 ubiquitination (Fig. 4B; compare lanes 6 and 8). These results further confirmed that SOX9 did not promote the ubiquitination of RUNX2 and prevents it from being degraded in the proteasome. In contrast, RUNX2 appeared to enhance the ubiquitination of SOX9 (compare lanes 7 and 8, bottom panels, in Fig. 4A, B). Although SOX9 can be polyubiquitinated, we found that most of the ubiquitinated SOX9 was monoubiquitinated (lane 8, bottom panel, in Fig. 4A, B), consistent with the finding reported by Akiyama.26

thumbnail image

Figure 4. Alternation of RUNX2 by ubiquitination in the presence of SOX9. HEK293 cells were transfected with SOX9, RUNX2 or both. Ub-His (6× histidine-tagged ubiquitin) was cotransfected for ubiquitination analysis. Then, 42 hours after transfection, cells were treated with MG132 (A) or vehicle (B) for 6 hours, followed by whole-cell lysate collection and histidine-tagged protein purification, and the results were subjected to Western blotting.

Download figure to PowerPoint

RUNX2 C-terminal domain is necessary for degradation enhancement by SOX9

The C-terminus of RUNX2 has been reported to be important for its stability and interaction with CHIP, a chaperone protein implicated in the regulation of osteogenesis.12 We therefore asked if the C-terminal domain of RUNX2 was important for its stability regulated by SOX9. A C-terminal domain deletion (168 amino acids) construct of RUNX2 (RUNX2-N) was generated, and we demonstrated that it was resistant to SOX9-mediated degradation by Western blot analysis (Fig. 5A, top panel, lanes 1 and 2) in the presence and absence of MG132 (Fig. 5A, top panel, lanes 3 and 4). Then we used the same cell lysate as that in Fig. 5A, lanes 3 and 4, for histidine-tagged protein purification to analyze the ubiquitination of RUNX2-N. We found that even though SOX9 did not affect the stability of RUNX2-N, it still decreased the ubiquitination of RUNX2-N, and the ubiquitination pattern was different from that of the WT RUNX2 (Fig. 5B, lanes 1 and 2). Since SUMO conjugation could compete with ubiquitin modification at a lysine residue,43 we searched for potential ubiquitination sites within the C-terminal deletion region of RUNX244 and found a potential ubiquitination site at residue 434. Removal of this site in RUNX2-N may contribute the different pattern of ubiquitination from that of WT RUNX2 (compare lane 6, top panels, in Figs. 4 and 5B).

thumbnail image

Figure 5. Requirement of the C-terminal domain of RUNX2 for SOX9-enhanced degradation. (A) C-terminal truncated RUNX2 (RUNX2-N) and full-length SOX9, along with Ub-His, were transfected into HEK293 cells as indicated. Then, 42 hours after transfection, cells were treated with MG132 or vehicle for 6 hours, followed by whole-cell lysate collection, and the results were subjected to Western blotting. (B) C-terminal truncated RUNX2 (RUNX2-N) and full-length SOX9, along with Ub-His, were transfected into HEK293 cells as indicated. Then, 42 hours after transfection, cells were treated with MG132 for 6 hours (the same as lanes 3 and 4 in part A), followed by histidine-tagged protein purification, and the results were subjected to Western blotting.

Download figure to PowerPoint

SOX9 enhances the degradation of RUNX2 through the lysosomal pathway

To identify alternative intracellular pathways that might be involved in the degradation of RUNX2, we performed an in vitro ubiquitination assay with the application of MG132 to inhibit the proteasomal pathway and NH4Cl to inhibit the lysosomal pathway.45 We found that RUNX2 was much more stable in the presence of NH4Cl than in the presence of MG132 when expressed alone (Fig. 6A; compare lanes 1 and 2, middle panel) as well as when co-expressed with SOX9 (Fig. 6A; compare lanes 3 and 4, middle panel). These findings suggested that RUNX2 also was degraded through the lysosomal pathway. Surprisingly, the protein level of RUNX2 had not been altered by SOX9 with the presence of NH4Cl (Fig. 6A; compare lanes 2 and 4), which indicated that SOX9 may promote the degradation of RUNX2 by directing the latter to the lysosome. To confirm this property of SOX9, we determined the ability of SOX9 to regulate β-catenin levels, an alternative protein known to be degraded by the proteasomal pathway that also acts downstream of canonical Wnt signaling, a key regulator of osteo/chondrogenesis.46 We confirmed that β-catenin is normally ubiquitinated and degraded through the proteasomal pathway, in which MG132 can stabilize β-catenin (Fig. 6B; compare lanes 1 and 2, top panel), and not through the lysosomal pathway, in which NH4Cl had no stabilizing effect (Fig. 6B; compare lanes 1 and 3, top panel). In contrast, NH4Cl stabilized β-catenin (Fig. 6B; compare lanes 4, 5, and 6, top panel) with the presence of SOX9. These results show that SOX9 can divert the degradation of different proteins from the proteasomal pathway to the lysosomal pathway 7.

thumbnail image

Figure 6. Diversion of degradation of RUNX2 to the lysosomal pathway by SOX9. (A) HEK293 cells were transfected with the plasmids indicated. Then, 42 hours after transfection, cells were treated with MG132 (40 µM) or NH4Cl (50 mM) for 6 hours. The cells were collected for harvesting via whole-cell lysate or His-tagged protein purification. The protein samples then were subjected to Western blotting against RUNX2. (B) HEK293 cells were transfected with EYFP or SOX9-EYFP. Then, 42 hours after transfection, cells were treated with MG132 (40 µM), NH4Cl (50 mM), or vehicle (DMSO and NaCl) for 6 hours. Whole-cell lysate was collected and subjected to Western blotting. The expression of EYFP and SOX9-EYFP was confirmed by fluorescence microscopy. Magnified cells shown (inset).

Download figure to PowerPoint

thumbnail image

Figure 7. Expression pattern of SOX9 and RUNX2 during chondrocyte hypertrophy in vitro. Human primary articular chondrocytes were cultured as spheroid pellets and treated with BIO to induce the hypertrophy. The chondrocyte pellets were cryosectioned and subjected to IMF for endogenous SOX9 and RUNX2. Nuclei were counterstained with DAPI. Scale bar = 100 µm.

Download figure to PowerPoint

Relative expression patterns of SOX9 and RUNX2 in primary human chondrocytes

To explore the expression pattern of SOX9 and RUNX2 in primary human chondrocytes, we induced hypertrophy on chondrocytes spheroid pellets by applying BIO, which can activate the Wnt/β-catenin pathway and result in hypertrophy of chondrocytes.35, 53 This mimics in some way the natural process of chondrocyte hypertrophy and osteogenesis. Seven days after application of BIO, the expression of SOX9 was decreased, and the expression of RUNX2 was prominently detected [although it was not expressed initially in the chondrocytes (day 3)] and compared with the DMSO control.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

During mouse embryogenesis, SOX9 is expressed in a variety of tissues, including the skeleton, testis, intestine, spinal cord, and pancreas. In the skeleton, SOX9 expression precedes RUNX2, and subsequently, SOX9 expression is maintained during the whole process of bone formation, from chondrocytes to osteoblasts.47 The coordination between SOX9 and RUNX2 is highly important for the determination of MSC fate. It has been shown previously that SOX9 is able to exert a dominant inhibitory effect on RUNX2 function32; however, the precise mechanisms involved are unclear. As well as transcriptional and translational control of function, one common regulatory mechanism is through protein stability. The stability of both SOX9 and RUNX2 is regulated by the ubiquitin-proteasome pathways.12, 22, 23, 26 Here we confirmed the suppressor function of SOX9 on RUNX2 transactivity using an alternative osteocalcin promoter reporter to that described previously.32 Furthermore, we found that SOX9 promoted the degradation of RUNX2 dramatically. Interestingly, these processes are phosphorylation-dependent. Previous work has shown that phosphorylation of the same serine residue in RUNX2 is required for its degradation induced by cyclin D1-Cdk4 via the proteasome.23 Phosphorylation of different serine residues (S104 and S541) on RUNX2 also can negatively regulate its activity.48 Collectively, these data demonstrated the importance of phosphorylation events in regulating RUNX2 function through different mechanisms.

We hypothesized that SOX9 might promote the ubiquitination and proteosome degradation of RUNX2, but this was not the case, and indeed, SOX9 decreased ubiquitination of RUNX2 and directed it toward lysosomal degradation. Similar observations have been reported for Ci-155, a transcription factor involved in the hedgehog signaling pathway, which can be degraded through both the lysosomal and proteasomal pathways, and it can be diverted preferentially to the lysosomal pathway by the binding factor Dbr.49 Significantly, SOX9 seems to have a more general function to regulate the ubiquitination status and/or degradation pathway for different signaling proteins. For example, we found that both RUNX2 and β-catenin, a well-known protein degraded through the ubiquitin-proteasome pathway,50, 51 were subject to SOX9-directed lysosomal compartmentalization. β-Catenin is a key component of the canonical Wnt signaling pathway. In the absence of Wnt ligand, β-catenin is continuously turned over and subject to proteosomal degradation. Following a Wnt signal, β-catenin is stabilized, translocates to the nucleus, and initiates TCF-dependent gene transcription.46 SOX9 has been shown recently to enhance β-catenin phosphorylation in the nucleus and promote its degradation.52 Our findings also show that SOX9 levels can determine β-catenin stability by diverting degradation to the lysosome, which is of particular significance given the importance of Wnt-signaling osteochondrogenesis.53 For the degradation of RUNX2 in the lysosome, polyubiquitination appears to be less important because ubiquitinated RUNX2 was decreased with the coexpression of SOX9 but still was degraded in the lysosome. A similar observation was made recently with Notch3, which does not need to be ubiquitinated before being degraded within the lysosome.54 Interestingly, it was reported recently that SOX9 can repress the expression of RUNX2 in chondrocytes indirectly through the vertebrate homologue of Drosophila bagpipe (Bapx1), a transcriptional repressor of RUNX2.55 This indicates that SOX9 regulated RUNX2 through a variety of mechanisms both transcriptionally and posttranscriptionally. In addition, we found that RUNX2 inhibited the transactivity of SOX9, which is consistent with the finding that SOX9 is expressed in mature osteoblasts47 and provides a mechanism whereby the expression levels of SOX9 may be maintained but its transactivity repressed by RUNX2 during the process of osteogenic differentiation in MSCs. We found that SOX9 was stabilized by RUNX2, and therefore, the ability of RUNX2 to inhibit SOX9 transactivity does not appear to be through degradation. It may be that RUNX2 can block SOX9 nuclear translocation or decrease the binding ability of SOX9 with the Collagen2 promoter. Indeed, we found a decreased SOX9 level in the nuclei of primary human chondrocytes overexpressing RUNX2. This may be part of the reason for the transactivity inhibitory effect of SOX9 on RUNX2. By mimicking chondrocyte hypertrophy, which is a key stage for the chondro-osteogenic differentiation of MSCs, we confirmed the reciprocal inhibitory influence SOX9 and RUNX2. It would appear that fine-tuning between these transcription factors and their crosstalk with other pathways, such as Wnt/β-catenin, regulate the process of cartilage and bone development.

Since Smad ubiquitin regulatory factor 1 (Smurf1) was the key factor indentified as an E3 ligase for RUNX2 ubiquitination and degradation,25 it will be interesting to study further the possible mechanisms of SOX9-mediated RUNX2 degradation, including the role of Smurf1 on RUNX2. In conclusion, we have demonstrated that SOX9 and RUNX2 can exert reciprocal inhibitory effects on each other using different mechanisms. The ability of SOX9 to direct the degradation of RUNX2, as well as other relevant signaling molecules, demonstrates the major role for this transcription factor in the control of mesenchymal differentiation.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

All the authors state that they have no conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We would like to thank Baoqiang Guo (Faculty of Life Science, University of Manchester, UK) for helpful discussions with us. We thank Shuji Muramatsu (Asahi Kasei Pharma Corp.) for the kind gift of the plasmid 4Col2-Luc. We thank York Hospital and Harrogate District Hospital for providing tissue samples. We also thank Peter O'Toole (Technology Facility, University of York, UK) for advice on FRET. This work was supported by funding from the European Commission (OsteoCord 018999).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References