Set7/9 Impacts COL2A1 Expression Through Binding and Repression of SirT1 Histone Deacetylation



Type II collagen is a key cartilaginous extracellular protein required for normal endochondral development and cartilage homeostasis. COL2A1 gene expression is positively regulated by the NAD-dependent protein deacetylase Sirtuin 1 (SirT1), through its ability to bind chromatin regions of the COL2A1 promoter and enhancer. Although SirT1/Sox9 binding on the enhancer site of COL2A1 was previously demonstrated, little is known about its functional role on the gene promoter site. Here, we examined the mechanism by which promoter-associated SirT1 governs COL2A1 expression. Human chondrocytes were encapsulated in three-dimensional (3D) alginate beads where they exhibited upregulated COL2A1 mRNA expression and increased levels of SirT1 occupancy on the promoter and enhancer regions, when compared to monolayer controls. Chromatin immunoprecipitation (ChIP) analyses of 3D cultures showed augmented levels of the DNA-binding transcription factor SP1, and the histone methyltransferase Set7/9, on the COL2A1 promoter site. ChIP reChIP assays revealed that SirT1 and Set7/9 form a protein complex on the COL2A1 promoter region of 3D-cultured chondrocytes, which also demonstrated elevated trimethylated lysine 4 on histone 3 (3MeH3K4), a hallmark of Set7/9 methyltransferase activity. Advanced passaging of chondrocytes yielded a decrease in 3MeH3K4 and Set7/9 levels on the COL2A1 promoter and reduced COL2A1 expression, suggesting that the SirT1/Set7/9 complex is preferentially formed on the COL2A1 promoter and required for gene activation. Interestingly, despite SirT1 occupancy, its deacetylation targets (ie, H3K9/14 and H4K16) were found acetylated on the COL2A1 promoter of 3D-cultured chondrocytes. A possible explanation for this phenotype is the enrichment of the histone acetyltransferases P300 and GCN5 on the COL2A1 promoter of3 D-cultured chondrocytes. Our study indicates that Set7/9 prevents the histone deacetylase activity of SirT1, potentiating euchromatin formation on the promoter site of COL2A1 and resulting in morphology-dependent COL2A1 gene transactivation. © 2014 American Society for Bone and Mineral Research.


Type II collagen is upregulated during normal cartilage development and alterations in its mRNA expression pattern may result in skeletal abnormalities and degenerative cartilage diseases.[1, 2] Furthermore, type II collagen serves as a template for skeletal formation via endochondral ossification, and enhances the capacity of articular joints to withstand tensile forces and maintain joint mobility.[1-3] Therefore, elucidating the regulatory mechanisms governing COL2A1 gene expression may lead to an improved understanding of pathologies related to cartilage development, endochondral growth, and cartilage degeneration.

Although Sox9 (SRY-related high mobility group-Box gene 9) has been shown to regulate the expression of the COL2A1 gene by binding the enhancer within the first intron,[4] numerous additional coactivators (eg, LSox5, Sox6, PGC1α, Notch) are involved in activating COL2A1 through enhancer association.[5-7] Because chromatin-modifying enzymes do not directly bind DNA regulatory motifs, they impact gene regulation by interacting with transcription factors and controlling the degree of histone tail modifications and chromatin compaction.[8-12] Accordingly, the histone acetyl-transferase CBP/P300 and the Class III NAD-dependent histone deacetylase Sirtuin 1 (SirT1) have been shown to co-regulate COL2A1 mRNA expression.[13, 14] Interestingly, SirT1 also promotes chondrocyte survival under various stress conditions,[15-20] further supporting its role in maintaining cartilage homoeostasis.

In a previous study, human chondrocyte cell lines stably overexpressing SirT1 displayed increased expression of COL2A1 mRNA and demonstrate enriched occupancy of SirT1, Sox9, and PGC1α on the gene enhancer region.[14] Similarly, TNFα-stimulated human chondrocytes, which possessed reduced COL2A1 mRNA expression, displayed an inactive SirT1 cleaved variant (75SirT1) that yielded reduced occupancy of PGC1α and Sox9 on the COL2A1 enhancer region.[17] These observations are consistent with SirT1s capacity to complex with Sox9[14] and the need for Sox9 to associate with PGC1α on the enhancer region.[6] Furthermore, these findings indicate that a three-way protein complex (ie, Sox9/PGC1α/SirT1) exists on the COL2A1 enhancer region and is necessary for COL2A1 transactivation and expression.

Previous reports also established that the DNA-binding transcription factor SP1 binds the promoter start site of human COL2A1 to induce its transactivation.[21, 22] Despite the identification of several factors involved in COL2A1 promoter binding, little is known about the chromatin-modifying enzymes involved in promoter-driven COL2A1 expression. Of particular interest is the histone methyl transferase Set7/9, which preferentially trimethylates lysine 4 on histone 3 (3MeH3K4) and has been found to be enriched in several promoter sites of activated genes.[23, 24] Interestingly, human chondrocytes overexpressing SirT1 show elevated marks of 3MeH3K4 on the COL2A1 promoter when COL2A1 expression is enhanced.[14] Liu and colleagues[25] reported that Set7/9 and SirT1 form a soluble complex, which prevents SirT1 deacetylation of p53 in 293 cell lines. In light of these observations, the aim of this study is to establish whether Set7/9 complexes with SirT1 to prevent its deacetylase activity and promote local euchromatin formation on the promoter site of COL2A1, leading to its subsequent transactivation. To explore the chromatin regulation of the COL2A1 promoter, we encapsulated human chondrocytes in alginate microbeads (ie, 3D) which elicits a native-like spherical morphology of the cells and activates COL2A1 expression. Several findings demonstrate that three-dimensional (3D) cultures of human chondrocytes induces COL2A1 expression as opposed to plated human chondrocytes, which display reduced COL2A1 levels and spontaneously dedifferentiate with advanced passage.[26, 27] Furthermore, enhanced expression of COL2A1 in 3D cultured human chondrocytes suggests that morphological attributes of the cells cause epigenetic changes in chromatin architecture and folding, which leads to changes in cellular phenotype and gene expression patterns.

Materials and Methods

Human cell culture and transfections

All procedures were performed with Hadassah Medical Center Institutional Review Board approval and in accordance with the Helsinki Declaration of ethical principles for medical research involving human subjects. A formal written informed consent was obtained from 60 osteoarthritic (OA) donors undergoing total knee arthroplasty (mean age 72 years, mean body mass index 31.5 kg/m2), prior to obtaining articular cartilage samples from their knee joints. Normal cartilage tissues were supplied by National Disease Research Interchange (NDRI; Philadelphia, PA, USA).

Human chondrocytes were isolated and plated as described by Derfoul and colleagues.[28] Cells were plated in 14-cm2 tissue culture dishes at a concentration of 1.5 × 106 cells/dish and were grown to confluence (passage 0 [P0]) in DMEM media (Sigma-Aldrich, St Louis, MO, USA) containing 10% fetal calf serum (FCS), 1% penicillin-streptomycin (Biological Industries Beit-Haemek Ltd., Kibbutz Beit-Haemek, Israel). Cells were cultured in standard incubation conditions (37°C, 5% CO2) until confluence. For passage experiments, confluent cells were trypsinized and replated to reach passage 5 (P5) and compared to P0. The retroviral expression plasmids, pHanPuro and pHanPuro-SirT1, were gifts from Dr. Vittorio Sartorelli (NIAMS, NIH, Bethesda, MD, USA). Retrovirus was generated, and infections were carried out as described.[29] Cells resistant to Puromycin (1 µg/mL; Sigma-Aldrich) were pooled and used for chromatin immunoprecipitation (ChIP) analysis.

HEK293 cells were transfected with a Flag-tag human SirT1 expression plasmid using ProFection Mammalian Transfection System (Promega, Fitchburg, WI, USA). Flag-tag human SirT1 expression plasmid was a kind gift of Prof. Danny Reinberg (NYU). Transfected cells were treated with TNFα (50 ng/mL; PeproTech Asia, Rehovot, Israel) for 48 hours and proteins were extracted and processed for immunoblot analysis as indicated in the next section.


The following antibodies were used for immunoblotting (IB), immunoprecipitation (IP), ChIP, and immunofluorescent staining (IF): β-actin (IB-1:2000; Santa Cruz Biotechnology, Santa Cruz, CA, USA; cat#47778), NAMPT (IB-1:1000; Abcam, Cambridge, MA, USA; Ab24149), SirT1 (IB-1:3000, ChIP-1:400, IP-1:150; Epitomics, Burlingame, CA, USA; cat#1054-1), SirT1 (IB-1:3000, IP-1:400; Millipore, Billerica, MA, USA; cat#07-131), SirT1 (IB-1:5000; Bethyl Laboratories, Montgomery, TX, USA; cat# A300-687A), Set7/9 (IB-1:2000, ChIP-1:400, IP-1:150, Millipore; cat#04-805), Sp1 (1B-1:500, ChIP-1:200; Santa Cruz Biotechnology; cat# 17824), DBC1 (IB-1:5000; Abcam; cat#70242), Ac- H4K5 (ChIP-1:400; Millipore; cat# 07-327), Ac-H4K16 (ChIP-1:400; Millipore; cat# 06-762), Ac-H3K9/14 (ChIP-1:400; Millipore; cat# A-4021-050), GCN5 (ChIP-1:400; Santa Cruz Biotechnology; cat#20698), P300 (ChIP-1:400; Santa Cruz Biotechnology; cat#8981), 3MeH3K4 (ChIP-1:400; Millipore; cat# 07-473), 3MeH3K9 (ChIP-1:400; Millipore; cat# 05-1250), 3MeH3K27 (ChIP-1:400; Millipore; cat# 07-449), and type II collagen (Dot blot-1:500, IF-1:50; Santa Cruz Biotechnology; cat# 52658). Secondary antibodies: anti-mouse-alkaline-phosphatase (AP) conjugated (IB-1:2500; Sigma-Aldrich; A3562), anti-rabbit-AP conjugated (IB-1:2500; Sigma-Aldrich; A3687), anti-goat-AP conjugated (IB-1:2500; Sigma-Aldrich; A2168), and Alexa Fluor 488-conjugated antibody (IF-1:100; Invitrogen, Carlsbad, CA, USA; A-11001).

Alginate microbead encapsulation, protein and mRNA extraction

Alginate microbead encapsulation was performed as described by Bonaventure and colleagues.[30] Briefly, following isolation of human chondrocytes, cells were grown to reach passage 1 (P1) and divided into monolayer (passage 2 [P2] or 2D) and 3D alginate cultures. To obtain 3D cultures, P1 human chondrocytes were centrifuged (6 minutes, 1500 revolutions per minute [rpm] [1200g] at room temperature [RT]) and a 1.25% sodium alginate solution was added to obtain a final concentration of 1 × 106 cells/mL. The cell solution was added drop-wise using a 23-guage needle into a 102 mM CaCl2 solution and set to polymerize with constant stirring for 10 minutes at RT. Microbeads were then washed with 0.9% NaCl during two consecutive 5-minute agitations. The 3D microbeads and control 2D human chondrocytes were maintained for 2 weeks in standard incubation conditions.

For protein or mRNA isolation, alginate microbeads were dissolved by adding depolymerization buffer (55 mM sodium citrate and 0.15 M NaCl, pH 6.05) for 10 minutes in 4°C. The cells were then recovered by centrifugation (6 minutes, 1500 rpm, 4°C). 2D cultured human chondrocytes were trypsinized and centrifuged (6 minutes, 1500 rpm, 4°C). Protein extracts were obtained using the protocol described by Perkins and colleagues.[31] Briefly, the harvested cells were treated with lysis buffer for 10 minutes on ice (10 mM HEPES, 0.1% NP-40, 10 mM KCl, 1.5 mM MgCl2,1 mM dithiothreitol [DTT], pH 7.9) to isolate cytosolic proteins. Next, the nuclei pellets were exposed to extract buffer for 10 minutes on ice (0.5 M NaCl, 20 mM HEPES, 25% glycerol, 1.5 mM MgCl2, 1 mM DTT, pH 7.9) to obtain soluble nuclear proteins. Cytosolic and nuclear fractions were then combined to obtain crude extracts for further analyses. All buffers contained 1 mM phenylmethylsulfonyl fluoride (PMSF; Sigma-Aldrich), protease inhibitor cocktail (Roche, Germany), and 10 µg/mL N-Acetyl-L-leucyl-L-leucyl-L-norleucinal (ALLN) (Sigma-Aldrich) to inhibit protease activity. Buffers were also supplemented with 1 µM Trichostatin A (TSA; Sigma-Aldrich), 7 mM β-Nicotinamide adenine dinucleotide (NAM) (Sigma-Aldrich), and 5 mM sodium butyrate (Millipore) to inhibit protein deacetylase activity.

Immunoblot analysis and immunoprecipitation

Protein harvesting and immunoblot procedures were performed according to Perkins and colleagues.[31] Primary and secondary antibodies used for immunoblotting and immunoprecipitation are specified above in Antibodies. Dot blots were carried out for detection of type II collagen protein, by pipetting the protein extract on a 5-mm-diameter circular template on a polyvinylidene fluoride (PVDF) membrane and subsequent incubation with primary and secondary antibodies prior to detection. Following protein extraction, soluble proteins were immunoprecipitated using the IP50 kit (Sigma-Aldrich) according to the manufacturer's instructions. To avoid antibody masking following immunoprecipitation, ReliaBLOT IP/Western Blot kit (Bethyl Laboratories) was used for immunoblot detection of Set7/9 (50 kDa).

All immunoblots were scanned in high resolution and band intensity was determined using Image J software (NIH, Bethesda, MD, USA). Each band was normalized to the corresponding housekeeping protein (ie, β-actin) appearing on the blot. Protein extracts for Type II collagen dot blots were loaded based on Western blot analyses of housekeeping protein. Relative band intensity was presented in arbitrary units (A.U.) adjacent to the representative immunoblot.

Immunofluorescence and immunohistochemistry

Cells were grown on coverslips and immunostaining was performed according to Albrecht and colleagues.[32] Sections were stained with the antibodies specified above in Antibodies. The slides were visualized using a confocal microscope (LSM; Ziess, Germany) with appropriate fluorescent emission/excitation parameters.

To visualize alginate encapsulated chondrocytes, microbeads were fixed with formaldehyde (3.7% for 40 minutes) and permeabilized with 0.2% Triton X-100 for 15 minutes. Following permeabilization, the samples were incubated with 0.2 mg/mL hyaluronidase (Sigma-Aldrich) for 15 minutes at 37°C. After blocking with 1% bovine serum albumin (BSA) for 20 minutes, the beads were incubated overnight with anti–type II collagen (1:50, 4°C; Santa Cruz Biotechnology), and then a secondary antibody with Alexa Fluor 488 conjugate (1:100; Invitrogen) was applied for 30 minutes. Next, the beads were stained with 4,6-diamidino-2-phenylindole (DAPI) for 15 minutes, followed by two PBS washes (5 minutes at RT). The samples were then embedded in mounting solution (Vectashield; Vector Laboratories, Burlingame, CA, USA) and images were captured using confocal microscope (LSM; Ziess) with appropriate emission/excitation parameters and following standardization with a negative control slide incubated only with the secondary antibody.

Fluorescein diacetate (FDA 5 µg/mL; Sigma-Aldrich) was used to observe viable cells in green fluorescence. Samples were viewed under an inverted fluorescence microscope (Model IX70; Olympus, Hamburg, Germany) equipped with a 490-nm bandpass filter and a 510-nm cutoff filter for fluorescent emission.

For histology, OA and normal articular cartilage was fixed in 4% paraformaldehyde, and dehydrated using a graded series of ethanol washes. Cartilage was then embedded in paraffin and cut to 5-µm sections. Alginate microbeads were fixed in OCT and immediately frozen. Five micrometer (5 µm) sections were obtained using a cryostat. Sections were stained with hematoxylin and eosin and visualized under a Leica DMR Light Microscope (Leica Microsystems, Buffalo Grove, IL, USA).

Quantitative PCR analysis

mRNA was isolated using RNeasy mRNA purification columns (Qiagen, Germany). cDNA was then prepared using the OneStep RT-PCR kit according to the manufacturers guidelines (Invitrogen). Real-time PCR reactions were performed using an ABI quantitative PCR (qPCR) model 7300 or 7900 (Applied Biosystems, Carlsbad, CA, USA) with purified samples containing a SYBR Green mix (Applied Biosystems) in accordance to the manufacturer's guidelines. Primers for qPCR were prepared for the following human genes:

  • A Disintegrin And Metalloproteinase with Thrombospondin Motifs (ADAMTS) 4 F:ACAAAGATCCAGGAAAGGAGGGCT; R:AGGGCTGAGGACCGTTAAAGGAAA

Values were normalized to human beta2 microglobulin (hβ2MG) or glyceraldehyde 3-phosphate dehydrogenase (GAPDH), which were both unaffected by the experimental treatments:

  • NAD/NADH ratio and SirT1 activity assays
  • NAD/NADH quantification kit was used according to the manufacturer's instructions (BioVision, Milpitas, CA, USA; cat# k337-100) and based on Dvir-Ginzberg and colleagues.[14] Data is presented as picomoles of NAD/total micrograms of protein.

To determine SirT1 activity a BIOMOL kit (NY; cat# AK500) was used according to the manufacturer's instructions and based on Dvir-Ginzberg and colleagues.[14] Briefly, 1 µM of TSA was added to the cell cultures 1 hour prior to harvesting to inhibit class I and class II histone deacetylases (HDACs). Crude extracts were then immunoprecipitated for SirT1 in a solution containing 1 µM TSA and protease inhibitor cocktail. The immunoprecipitated extracts were added (15 µL) to an opaque multi-well plate with 1 µM TSA and 1 mM of acetylated Fluor-de-Lys substrate. Then 10 mM NAM was added to several samples as a negative control. After adding 50 µL of a developer solution supplemented with 1 µM TSA and 10 mM NAM (final concentrations), the plate was read using a multi-well fluorimeter (excitation 360 nm, emission 460 nm). A standard curve was generated using a deacetylated substrate Fluor-de-Lys (range, 1–40 µM). Values are presented as picomole conversion of the Fluor-de-Lys substrate per microgram protein per minute. The negative controls (10 mM NAM) were subtracted from each treatment to give the final values.

ChIP and ChIP reChIP analyses

ChIP and ChIP reChIP assay was performed based on Furlan-Magarill and colleagues.[33] Specifically, chromatin was cross-linked using 1% formaldehyde for 10 minutes. Cells were then lysed (lysis buffer: 10 mM EDTA, 50 mM Tris-HCl, pH 8.0, 1% SDS) and sonicated (Vibra-Cell; Sonics & Materials, Newtown, CT, USA), at 35 cycles of 95% amplitude for 30 seconds, followed by a 45-second incubation on ice. Samples were then centrifuged and supernatants were aspirated for further analyses. Following ChIP and reChIP, eluates were treated with RNase (1 mg/mL, 30 minutes, 37°C), Proteinase K (1 mg/mL for 1 hour at 42°C) and then DNA was extracted using a Qiaquick spin kit (Qiagen, Germany).

Real-time PCR (qPCR) reactions for ChIP and ChIP reChIP were carried out using primers flanking the COL2A1 promoter start site (–167) to (–65), see illustration in Figure 6A. Additional analyses were performed on the Sox9-binding enhancer region of COL2A1 (+2259) to (+2430). As a negative control, a distal set of primers flanking the intron 38 (+24148) to (+24368) of the COL2A1 gene was used, based on the protocol by Furlan-Magarill and colleagues.[33] The genomic intron 38 region of COL2A1 also served as negative control for all the ChIP and ChIP reChIP protocols. Quantitative PCR analysis of this region showed negligible detection (less than 6%) compared to the promoter or enhancer regions. For ChIP analyses the following primers were used: COL2A1 (NM_001844.4) Promoter F: CGCTGGGCTGTAACCTGAAC; R: GGAAGCGTGACTCCCAGAGA; COL2A1 Enhancer F: ATCCTCCTTTGTGAGGCTTGTT; R: AGTACGAGAGAACCCACTGGAC; and COL2A1-intron 38 (negative control) F: GCCAGAACCAAGCTGCTGAT; R: CAATGGACCCCTGAGGTTTTC.

Statistical analysis

All experiments were performed on multiple donor samples (n > 4 per experiment), as indicated in each figure legend. Statistical analysis was obtained using nonparametric Mann-Whitney analyses. Values of p of plotted data equaling 0.05 or less were considered statistically significant.

3D/2D data plots show a dashed line that delineates the 2D sample which served as control compared to the 3D sample. Immunoblots, immunofluorescent, and FDA images show a representative sample derived from four separate donors, unless otherwise indicated.


Passaged human chondrocytes express reduced COL2A1 mRNA and type II collagen protein, which correlates with decreased SirT1 activity

To explore the chromatin regulation of the COL2A1 promoter, we chose to examine two experimental cultures (ie, P0/P5 and 2D/3D), that produce distinct variations in COL2A1 gene expression. Chondrocytes tend to dedifferentiation with advanced passage and possess decreased COL2A1 expression.[26, 27, 34] To validate this phenomenon, we examined COL2A1 mRNA and type II collagen protein levels (Fig. 1). Figure 1A displays reduced COL2A1 mRNA expression in P5 passaged chondrocytes when compared to P0 chondrocytes. Consistent with gene expression, type II collagen dot blots and immunofluorescent staining show reduced protein levels (Fig. 1B and C, respectively), suggesting that chondrocytes dedifferentiate with advanced passage. Figure 2A shows lower SirT1 activity, despite no change in SirT1 mRNA and protein levels (Fig. 2B and C, respectively). In Fig. 2C, passaged chondrocytes show a statistically significant 30% decrease in protein levels of deleted in breast cancer 1 (DBC1), a known repressor of SirT1 activity.[35] Whereas SirT1 activity decreases with advanced passage, cellular levels of NAD, a cofactor of SirT1 activity, are increased with passage (Fig. 2D), possibly due to a slight but statistically significant (10%) increase in the salvage pathway rate-limiting enzyme NAMPT (Fig. 2E, F), which generates cellular NAD.[36] These results demonstrate that chondrocyte passaging reduces cellular SirT1 activity, which is likely a result of increased inhibitory DBC1.

Figure 1.

Human chondrocytes express reduced COL2A1 mRNA and type II collagen protein levels with passage. (A) Human chondrocytes were isolated from OA knees and plated in a standard monolayer plate at passage 0 or 5 (denoted as P0 or P5, respectively). qPCR of COL2A1 gene expression in P0 or P5 (n = 5). The graph depicts a 50% decrease in COL2A1 gene expression with passage. (B) Dot blots of P0 and P5 protein extracts display decreased protein levels of type II collagen with advanced passage (n = 4). (C) Immunofluorescent staining for type II collagen (green intensity) protein levels show a decrease with advanced passage (n = 4). Cell nuclei were stained with DAPI and appear in blue fluorescence. Bars beneath the images indicate magnification. The graph error bars indicate the SD and the statistical significance according to Mann-Whitney analyses, assuming a p value of 0.05 or less to be statistically significant. A.U. = arbitrary units of protein intensity on the blot.

Figure 2.

Decreased cellular SirT1 activity with passage of human chondrocytes. Human chondrocytes were isolated from OA knees and plated in a standard monolayer plate (P0 and P5). (A) SirT1 deacetylase enzymatic activity was estimated according to the protocol specified in the Materials and Methods. Results show a 30% decrease in SirT1 enzymatic activity with passage (n = 5). qPCR (B), immunoblot analyses and band intensity quantification of SirT1 and DBC1 (C), (n = 6). (D) Cellular NAD levels per microgram total protein (n = 6), show increased levels with passage. qPCR (E), immunoblot analyses and band intensity quantification (F) of the rate-limiting salvage pathway enzyme NAMPT (n = 6). The graph error bars indicate the SD and the statistical significance according to Mann-Whitney analyses, assuming a p value of 0.05 or less to be statistically significant. A.U. = arbitrary units of protein intensity on the blot.

3D culture of human chondrocytes show enhanced cellular SirT1 enzymatic activity and COL2A1 gene expression

As an additional experimental system, we employed 3D cultures of human chondrocytes, which have been shown to possess enhanced cartilage-specific gene expression.[27, 37] thus possibly rescuing the effect of dedifferentiation that occurs with advanced chondrocyte passage. The fabrication of 3D cultured human chondrocytes allows the control of cell dispersion (1 × 106 cells/mL) and architecture, which appeared to possess similarities to that of normal articular cartilage midzone (Fig. 3A). Human chondrocytes were entrapped in alginate microbeads (3D) and compared to P2 plated human chondrocytes (2D). Figure 3B shows the gross morphology and viability of the cells in 2D and 3D experimental settings. The dotted white line in the upper left panel of FDA-stained alginate microbeads (Fig. 3B) delineates the contour of the alginate microbead wherein the cells are encapsulated and maintain a spherical morphology. Immunofluorescent (Fig. 3C) and dot blot (Fig. 3D) analyses confirmed that 3D cultures possess enhanced type II collagen protein levels as compared to monolayer cultures.

Figure 3.

Characterizing 3D and 2D human chondrocyte culture morphology and type II collagen protein levels. Human chondrocytes were passaged to reach P2 and divided to monolayer (2D) cultures and alginate encapsulated cultures (3D). (A) Hematoxylin and eosin staining of 3D chondrocytes and normal articular cartilage (midzone), (n = 4). (B) 3D cultures were stained with FDA to asses their morphology and viability under each culture condition (n = 4). The dashed white line delineates the microbead contour and its encapsulated cells. Micrographs in the lowest panel are of phase depictions of 3D/2D cultures. (C) Immunofluorescent staining for type II collagen protein levels (green intensity) shows augmented protein levels in 3D cultures (n = 4). Cell nuclei were stained with DAPI and appear in blue fluorescence. (D) Immunoblot analyses and band intensity quantification for 3D/2D human chondrocytes protein extracts (n = 4). Bars beneath images in B and C indicate magnification. A.U. = arbitrary units of protein intensity on the blot; FDA = fluorescein diacetate.

Consistent with elevated type II collagen protein levels, COL2A1 mRNA expression is augmented in 3D cultures (Fig. 4A). Figure 4B demonstrates that cellular SirT1 enzymatic activity is significantly increased (approximately threefold) in 3D cultured chondrocytes when compared to 2D conditions. Furthermore, SirT1 mRNA expression (Fig. 4C) and protein levels (Fig. 4D) are significantly increased in 3D cultures, whereas DBC1 protein levels remain unchanged (Fig. 4D). Elevated NAMPT protein levels (Fig. 4E) and mRNA expression (Fig. 4F) corresponded with a 13-fold enhancement of cellular NAD levels (Fig. 4G). Given the known connection between NAD levels and SirT1 enzymatic activity, these data suggest that increased levels and bioavailability of NAD in 3D cultured cells yields enhanced cellular SirT1 enzymatic activity.

Figure 4.

3D human chondrocytes present increased cellular SirT1 enzymatic activity and COL2A1 expression. (A) COL2A1 mRNA levels in 3D and 2D culture conditions (n = 6). (B) SirT1 enzymatic activity (n = 6) and (C) SirT1 mRNA (n = 6) are significantly enhanced in 3D cultures. Quantified immunoblots of (D) SirT1 and DBC1 (n = 6) show a dramatic increase in SirT1 protein levels. (E) NAMPT immunoblots and protein quantification of 3D and 2D cultures (n = 4). (F) NAMPT mRNA levels (n = 5). (G) Cellular NAD bioavailability in 3D versus 2D cultures. Plots in A, C, and F present 2D culture conditions as a reference (ie, 1) dashed black line. Graph error bars indicate the SD and the statistical significance according to Mann-Whitney analyses, assuming a p value of 0.05 or less to be statistically significant. A.U. = arbitrary units of protein intensity on the blot.

3D cultures also exhibited augmented mRNA and protein levels of the transcription factors SP1 and Sox9 (Fig. 5A. B), which are known to regulate COL2A1 expression by binding the DNA-motifs of the promoter and enhancer regions, respectively.[4, 38, 39] The increase in Sox9 protein levels in 3D cultures is consistent with previous reports by Tew and colleagues,[39] which showed that disrupted actin organization, as facilitated by 3D culture of chondrocytes, stabilizes Sox9 mRNA, which may also be the case for SP1.

Figure 5.

SP1 and Sox9 are upregulated in 3D human chondrocytes cultures. (A) Immunoblots for SP1, Sox9, and Set7/9 are presented for 3D/2D cultures and quantified in the graph (right panel), (n = 4). (B) SP1 and Set7/9 mRNA levels (n = 4) in 3D and 2D cultures. 2D culture conditions are presented as a reference (ie, 1) dashed black line. (C) Fibronectin and Laminin mRNA levels (cartilage nonspecific genes) are evaluated for 3D and 2D cultures (n = 4). Cartilage matrix degrading enzymes (ie, ADAMTS4, ADAMTS5, and MMP-13) are evaluated for 3D and 2D cultures (n = 4). Graph error bars indicate the SD and the statistical significance according to Mann-Whitney analyses, assuming a p value of 0.05 or less to be statistically significant.

In contrast, the histone methyltransferase Set7/9 showed unchanged mRNA expression and protein levels in 3D and 2D culture settings (Fig. 5A and B, respectively). As an additional control, we monitored mRNA levels of tissue nonspecific genes (ie, fibronectin and laminin), which remained unaffected (Fig. 5C, upper panel), whereas cartilage degrading proteases as MMP13, ADAMTS4, and ADAMTS5 were significantly reduced in 3D versus 2D human chondrocytes cultures (Fig. 5C, lower panel).

ChIP analysis reveals a SirT1/Set7/9 complex is formed on the chromatin of the actively transcribed COL2A1 promoter

We observed that 3D chondrocytes possess enhanced SirT1 activity, which correlates with augmented COL2A1 mRNA expression and type II collagen protein levels. To understand the regulation of the active COL2A1 promoter, we performed ChIP assays using specific primers flanking the promoter start site and enhancer site within the first intron (illustrated in Fig. 6A). We found enhanced occupancy of the DNA-binding transcription factor SP1 on the promoter start site of 3D cultured cells, whereas passaged cells (P5) showed a slight but insignificant reduction of SP1 (Fig. 6B). Additionally, passaging significantly decreased Sox9 occupancy on the enhancer site (Fig. 6B), whereas 3D chondrocytes possessed a dramatic increase in Sox9 binding. These data suggest that whereas both SP1 and Sox9 are involved in transactivating COL2A1 in 3D cultures, only decreased Sox9 contributes to reduced COL2A1 expression in passaged chondrocytes.

Figure 6.

Set7/9 and SirT1 form a complex on the COL2A1 promoter of 3D human chondrocytes. (A) Illustrates the alignment of COL2A1 promoter and enhancer. The black double-headed arrows present the primer flanking sites and their position. The scheme additionally outlines the SP1-binding on the promoter as well as the Sox9-binding site on the enhancer of COL2A1. (B) Graphs depict ChIP assays for COL2A1 promoter (left) and enhancer (right) of P0/P5 cultures (lower plots) and 2D/3D cultures (upper plots), (n = 5). (C) Coimmunoprecipitation for Set7/9 and SirT1 in 2D and 3D cultures (n = 4). (D) Box plot presenting ChIP-reChIP analyses (n = 5), assessing SirT1 and Set7/9 binding on the promoter of COL2A1 of 2D/3D cultures. Box plot in D provides a visual summary of the median (solid line in the box), the mean (dashed lines in the box), the interquartile range (box height), and the quartile skew (the relative size of box halves). Results were analyzed using Mann-Whitney one-tail test and present a statistical significance at p < 0.05.

While SP1 is enriched on the promoters of 3D chondrocytes, chromatin modifying enzymes must act in concert to promote the euchromatic state of the COL2A1 promoter to potentiate its expression. To investigate whether SirT1 and Set7/9 are involved in inducing COL2A1 mRNA expression by altering chromatin, we conducted additional ChIP assays on the COL2A1 promoter and enhancer loci. Enrichment of SirT1 and Set7/9 was observed on the COL2A1 promoter of 3D cultures (Fig. 6B, left graph, upper panel), but not on the enhancer region of P0/P5 chondrocytes (Fig. 6B, left graph, lower panel) or 2D/3D cultures. Taken together, these results further indicate that Set7/9 impacts COL2A1 gene expression via promoter occupancy.

To test whether SirT1 and Set7/9 association contributes to chromatin relaxation, we examined whether the two proteins form a complex in cultured chondrocytes. Coimmunoprecipitation showed that SirT1 and Set7/9 formed equal amounts of soluble protein complexes in 2D and 3D cultures (Fig. 6C). To further evaluate the possible binding between these chromatin-modifying enzymes on the COL2A1 promoter, we performed ChIP reChIP analyses on 3D and 2D cultured human chondrocytes (Fig. 6D). Results showed increased SirT1/Set7/9 protein complexes on the COL2A1 promoter of 3D chondrocytes.

3MeH3K4 and acetylation levels of SirT1-histone targets are enhanced on the COL2A1 promoter of 3D cultured human chondrocytes

In line with Set7/9 enrichment, its histone mark 3MeH3K4 was also found to be enhanced fourfold on the COL2A1 promoter of 3D human chondrocytes (Fig. 7A) and is reduced with advanced passage (Fig. 7B). These data suggest that SirT1 causes Set7/9 association on the COL2A1 promoter, leading to increased levels of 3MeH3K4. Based on these data, we next examined SirT1-overexpressing chondrocyte lines and found enriched levels of the methyltransferase Set7/9 in the promoter region of (Fig. 7C), further supporting Set7/9 involvement in COL2A1 expression. Despite the enhancement of SirT1 on the COL2A1 promoter (Fig. 6B) of 3D chondrocytes, SirT1s' preferential deacetylation targets (ie, H3K9/14 and H4K16),[40, 41] showed increased acetylation (Fig. 7D) in 3D human chondrocytes. To further decipher which histone acetyltransferases could cause increased targeted acetylation,[42-47] we examined histone acetyltransferase P300 and GCN5 levels given their previously reported enrichment on the promoter/enhancer sites of SirT1-overexpressing chondrocytes.[14] Figure 7E shows significant recruitment of P300 and GCN5 to the COL2A1 promoter region of 3D cultured human chondrocytes. The enhanced acetylation of H3K9/14, H4K5, and H4K16 (Fig. 7D) are in line with previous reports indicating that P300 acetylates H3K14/9,[42, 43] whereas GCN5 targets H3K14/9, H4K5, and H4K16.[44-46] These results suggest that histone hyperacetylation on the COL2A1 promoter, together with the increased presence of 3MeH3K4, may cause the recruitment of the transcription machinery and subsequent gene activation of COL2A1.[23, 47]

Figure 7.

Modified histone marks on the COL2A1 promoter of 3D human chondrocytes. (A) ChIP analyses show enhanced 3MeH3K4 marks on the promoter of COL2A1 of 3D chondrocytes (n = 4) and (B) passaged chondrocytes (n = 4). (C) ChIP analyses for Set7/9 occupancy on the COL2A1 promoter and enhancer sites of SirT1 overexpressing human chondrocyte lines as compared to pHan control lines (n = 5). (D) ChIP analyses for AcH3K9/14, AcH4K16, and AcH4K5 in 2D/3D cultures. (E) ChIP analyses for histone acetyltransferases P300 and GCN5 in 2D/3D cultures. (F) ChIP (n = 4) for inhibitory histone marks (ie, 3MeH3K9 and 3MeH3K27) on the COL2A1 promoter of in 2D/3D cultures. (G) Suggested model for COL2A1 transactivation in 3D cultured human chondrocytes. The scheme illustrated shows a local complex of SirT1/Set7/9 on the COL2A1 promoter in conjunction with enhanced association of the promoter binding SP1 transcription factor. Although Set7/9 induced 3MeH3K4, it simultaneously inhibits SirT1 deacetylase activity. Abrogated SirT1 activity, together with enhanced levels of HATs P300 and GCN5, results in elevated marks of AcH4K16, AcH4K5, and AcH3K9/14. (H) Immunoblots of N-terminally flag human SirT1 overexpressed in HEK293 cells induced with TNFα (50 ng/mL, 24 hours). Protein extracts of HEK293 overexpressing Flag-SirT1 were obtained and immunoblotted for Flag (left blot) following treatment. Arrows show the characteristic Flag-positive 75-kDa SirT1 fragment that appears following TNFα stimulation. This fragment does not appear in 2D/3D and P0/P5 chondrocytes, indicating that cleavage of SirT1 does not occur in these cultures (n = 3). Plots in A, D, and E present 2D culture conditions as a reference (ie, 1) dashed black line. Graph error bars indicate the SD and the statistical significance according to Mann-Whitney analyses, assuming a p value of 0.05 or less to be statistically significant. HAT = histone acetyltransferase.

As an additional control, we examined repressive histone methylation marks (ie, 3MeH3K9 and 3MeH3K27; Fig. 7F). We found that these histone methylation targets are reduced on the COL2A1 promoter of 3D versus 2D human chondrocytes. These observations, together with enhanced 3MeH3K4 and histone acetylation of H3K9/14, H4K5, and H4K16, indicate that chromatin relaxation and enhanced COL2A1 expression is impacted by SirT1/Set7/9 binding on the COL2A1 promoter, which is influenced by cell morphology and shape.


3D biomimetic culture techniques present a valuable tool for deciphering epigenetic mechanisms of gene expression during cell differentiation, as previously shown by dynamic changes in chromatin architecture and compaction during differentiation of embryonic stem cells.[48] Epigenetic changes could be rendered through control of cell-surface ligands[49] that may indirectly impact various cytoskeletal and nuclear scaffold components (eg, Lamins, SATB1), driving structural chromatic alterations and leading to aberrant gene expression profiles and cellular phenotype.

Cellular SirT1 enzymatic activity was enhanced in 3D cultured chondrocytes, possibly due to increased SirT1 and NAMPT protein levels. Increased NAMPT protein seemingly also caused the dramatic increase in cellular NAD bioavailability. Despite the enhanced SirT1 enzymatic activity presented in 3D chondrocytes, its histone targets (ie, H3K9/14, H4K16) on the COL2A1 promoter were acetylated, indicating that it possessed reduced deacetylase activity when occupying the COL2A1 promoter region of 3D chondrocytes. Additional enrichment of the histone acetyltransferases P300 and GCN5 enhanced acetylation of the histone targets recognized by SirT1 (ie, H3K9/14, H4K16), as well as H4K5,[42-46] and contributed to a relaxed chromatic state of the COL2A1 promoter site. Interestingly, a previous report showed association of GCN5, P300, and SirT1 in response to high NAD levels,[50] which we also observed on the COL2A1 promoter of 3D human chondrocyte (see illustration in Fig. 7G).

We found that SirT1 and Set7/9 formed a protein complex which deprives chromatin-bound SirT1 of its enzymatic deacetylase activity, confirming results from a previous study.[25] Although 3D versus 2D cultures did not show an overall increase in SirT1/Set7/9 protein association, this complex was significantly enhanced on the COL2A1 promoter of 3D human chondrocytes, as compared to their monolayer equivalent cells. Set7/9 appeared to preferentially target the COL2A1 promoter and not enhancer region, which is consistent with previous reports indicating 3MeH3K4 are abundant in promoter start sites of actively transcribed genes.[23, 24] To confirm these results we attempted knockdown experiments of SirT1 and Set7/9 in 3D chondrocytes; however, due to technical limitations we were unable to successfully downregulate these target proteins within a 3D setting. It is important to note that SirT1/Set7/9 protein association in 3D was also accompanied by enhanced Sox9 DNA occupancy on the enhancer of COL2A1. These two events may synergize in promoting COL2A1 transactivation, possibly by an enhancer-promoter looping mechanism.[14]

SirT1 has been previously reported to associate with the histone methyltransferases SUV39H1 and subsequently promote heterochromatin formation via enhanced 3MeH3K9 marks.[51] Augmented 3MeH3K9 marks result from SirT1's ability to deacetylate SUV39H1 and activate its lysine methyltransferase enzymatic activity.[51] Based on our observations, SirT1's activity is impaired in the presence of Set7/9 and therefore likely to promote inactivation of SUV39H1 by maintaining its acetylated state.

Accumulating evidence demonstrates that SirT1 may undergo posttranslational and posttranscriptional alterations effecting its enzymatic activity.[15, 17, 52-54] We did not observe C-terminal truncated SirT1 (ie, 75SirT1), nor the deleted exon 8 SirT1 form[54] in either 2D (P0, P2, P5) or 3D cultures (Fig. 7H). Nonetheless, modifications impairing SirT1 enzymatic activity may be present on the SirT1 occupying the COL2A1 promoter site. These modifications may also play a role in preferential association of SirT1 in various protein complexes.

In summary, we find that Set7/9 complexes with SirT1 on the COL2A1 promoter and enhances COL2A1 expression in conjunction with P300 and GCN5. Interestingly, Set7/9, is absent in the promoter region of passaged P5 human chondrocytes, further highlighting its role in inducing COL2A1 mRNA expression. Our data indicate that SirT1/Set7/9 complex reduces SirT1 histone deacetylation capacity on the COL2A1 promoter, which supports euchromatin formation, leading to COL2A1 expression in human chondrocytes in a cell morphology–dependent manner.


All authors state that they have no conflicts of interest.


This work was supported by the Marie Curie European IRG reintegration grant (Proposal N° 268214). We thank Prof. Tim Hardingham (Manchester University), Dr. Eran Meshorer (Hebrew University of Jerusalem), Dr. Rachel Lichtenstien (Ben-Gurion University of the Negev), and Prof. Itai Bab (Hebrew University of Jerusalem) for their constructive input. We thank Dr. Ariel Ginzberg for his help with the manuscript figures and graphs.

Authors' roles: M.D-G and H.O. designed research; M.D-G, H.O., H.M., I.S. and A.K. performed research; M.D-G, A.H., L.K.,Y.M., M.L. contributed new reagents/analytic tools and clinical samples; M.D-G., H.O., I.S., A.K. and A.Z. analyzed the data and performed statistical analysis; and M.D-G wrote the paper.