Introduction
The matrix metalloproteinases (MMPs) are a family of zinc-dependent endopeptidases responsible for the degradation of extracellular matrix (ECM) components. While low levels of these enzymes are required for the homeostatic ECM turnover seen in wound healing, angiogenesis, and development, high levels have been implicated in the pathology of atherosclerosis, tumor metastasis, and the arthritides. In the case of osteoarthritis (OA) and rheumatoid arthritis (RA), members of the collagenase subgroup of the MMPs, specifically
MMP-1 and
MMP-13, are particularly important in the progression of joint disease [
1,
2]. The ability to cleave the collagen triple helix is unique to the collagenases, and the overexpression of
MMP-1 and
MMP-13 in chondrocytes in response to proinflammatory cytokines such as interleukin-1-beta (IL-1β) and tumor necrosis factor-alpha is critical in the pathogenesis of OA and RA [
1].
Many efforts to design small-molecule inhibitors of MMP activity (MMPIs) have succeeded in creating potent compounds; however, due to the highly conserved nature of the catalytic domain among family members, these compounds demonstrate significant inhibitory efficacy against multiple MMPs [
3]. This lack of specificity has been identified as the likely cause of the debilitating side effects observed in clinical trials with these compounds which presented as a chronic musculoskeletal syndrome (MSS) that was characterized by reduced mobility with joint pain and edema due to tendonitis and inflammation [
4‐
6]. The root cause of the MSS is thought to be the disruption of normal connective tissue turnover, secondary to the inhibition of multiple MMPs [
7]. Unfortunately, the MSS has continued to hinder many newly developed MMPIs, resulting in the discontinuation of multiple clinical trials [
8], although promising results with MMP-specific compounds are emerging [
9]. Specific inhibition of MMP gene synthesis is an alternative strategy for counteracting the overexpression of MMPs involved in particular diseases. Although many MMP promoters share similarities, particular variations in MMP promoter structure and in the signaling pathways required for their expression may make it possible to target certain family members with specific ligands.
Peroxisome proliferator-activated receptor-gamma (PPARγ) is a nuclear hormone receptor (NHR) initially recognized as a regulator of genes active in adipogenesis and insulin sensitivity [
10]. PPARγ forms an obligate heterodimer with the retinoid X receptor (RXR:PPARγ) that binds to direct repeat-1 (DR-1) motifs, known as PPARγ response elements (PPREs), in the promoter DNA of regulated genes [
11]. NHRs are typically thought to exert their transcriptional regulatory effects through interaction with coregulatory complexes, which modify the local chromatin environment via multiple mechanisms, including the enzymatic activity of histone deacetylases (HDACs) and histone acetyltransferases (HATs) [
12]. HDAC activity results in a decrease in histone acetylation and a subsequent decrease in transcriptional activity, whereas HAT activity leads to an increase in histone acetylation and a subsequent increase in transcriptional activity [
13].
Recent work has identified an anti-inflammatory role for PPARγ in chondrocytes when the receptor is activated by ligands such as the thiazolidinedione compound rosiglitazone and the prostaglandin 15-Deoxy-Δ12,14-prostaglandin J2 [
2,
14‐
16]. Notably, this anti-inflammatory effect of PPARγ ligands extends to the inhibition of IL-1β-induced expression of
MMP-1 and
MMP-13 in rabbit chondrocytes [
2,
17,
18], and administration of these compounds blunts the development of joint disease in animal models of arthritis [
18,
19]. François and colleagues [
17] have proposed a mechanism to explain rosiglitazone-mediated inhibition of IL-1β-induced expression of rabbit
MMP-1 that involves binding of the RXR:PPARγ heterodimer to a degenerate DR-1 site in the proximal (approximately -72 base pairs) region of the rabbit
MMP-1 promoter. This DR-1 site overlaps a binding site for the transcription factor activator protein-1 (AP-1), which is largely responsible for the proinflammatory cytokine-induced upregulation of
MMP-1 [
20]. In this competitive binding model, binding of the RXR:PPARγ heterodimer to the DR-1 element precludes binding of AP-1 proteins to its site and thereby antagonizes the expression of
MMP-1. François and colleagues [
17] also identified a similar degenerate DR-1/AP-1 site in the promoters of human
MMP-1, MMP-9, and
MMP-13, although the function of this site has not been experimentally verified in the human genes.
Previous work by our laboratory has shown that LG100268 (LG268), a ligand specific for RXR, inhibits IL-1β-induced
MMP-1 and
MMP-13 transcription in the SW-1353 human chondrosarcoma cell line and is associated with a decrease in histone acetylation proximal to the transcription start site in the
MMP-1 and
MMP-13 promoters [
21]. While RXR is an obligate dimer partner for a number of other NHRs, including retinoic acid receptors, thyroid receptor, vitamin D receptor, PPARs, liver X receptors (LXRs), and farnesoid X receptor (FXR) [
22], the ligand LG268 activates only a subset of the RXR catalog of partners, including RXR:FXR, RXR:LXR, RXR:PPARα, and RXR:PPARγ heterodimers, as well as RXR homodimers [
23‐
25]. Of these dimers, only RXR:PPARα, RXR:PPARγ, and RXR homodimers bind to the DR-1 element [
11], suggesting that all or any of these three dimers may be responsible for mediating the inhibitory effect of LG268 on
MMP-1 and
MMP-13. However, since PPARγ-specific, but not PPARα, ligands block
MMP-1 and
MMP-13 gene expression, RXR:PPARγ heterodimers as well as RXR homodimers may be mediating this suppression [
2,
18,
26].
Recent investigations into the mechanisms by which PPARγ inhibits the expression of genes involved in inflammation have identified a molecular pathway of ligand-dependent conjugation of the small ubiquitin-like modifier (SUMO) to lysines in the PPARγ receptor [
14,
27]. This SUMO-conjugated form of PPARγ then binds to corepressor complexes containing HDAC activity and to other promoter-bound proteins. This anchors the corepressors and prevents their release upon proinflammatory stimulation, thereby blocking recruitment of coactivator complexes with HAT activity. The presence of multiple functional SUMOylation sites (SUMO consensus sequence = ψ KXE/D, where ψ is a hydrophobic amino acid, X is any amino acid, and K is the specific SUMOylation target) within PPARγ has been confirmed [
14,
27], and Floyd and colleagues [
27] describe multiply SUMOylated forms of PPARγ. Pascual and colleagues [
14] demonstrate that SUMOylation at different sites confers different modifications of receptor activity and identify K365 as the SUMOylation site required for transrepression of inflammatory genes by PPARγ. SUMOylation of RXR has also been reported [
28].
We hypothesized that, because LG268 and PPARγ ligands target the same NHR complex and have similar inhibitory effects on MMP production, both ligands may be activating similar mechanisms to inhibit MMP gene expression. The competitive binding model implicates competition for binding to the degenerate DR-1 site between RXR:PPARγ and AP-1 proteins as a possible mechanism for rosiglitazone-mediated inhibition of
MMP-1 [
17]. In addition, we hypothesized that LG268, as a ligand for RXR, may also induce increased binding of the heterodimer to the DR-1 site and that combination treatment with both ligands would further increase binding to the DR-1 site since both NHRs would be liganded. As a result, combined treatment should lead to greater inhibition of
MMP-1 and
MMP-13 gene expression compared with either compound alone. In this paper, we demonstrate that combined treatment with the RXR ligand LG268 and the PPARγ ligand rosiglitazone suppresses
MMP-1 and MMP13 gene expression more effectively than either compound alone. In addition, we document that this inhibition is transcriptionally mediated and involves genetic and epigenetic mechanisms but does not appear to involve competitive binding between RXR:PPARγ and AP-1 at the DR-1/AP-1 element.
Materials and methods
Cell culture
SW-1353 human chondrosarcoma cells were obtained from the American Type Culture Collection (Manassas, VA, USA). These cells were propagated at 37°C with 5% CO in Dulbecco's modified Eagle's medium (DMEM) (Mediatech, Inc., Manassas, VA, USA) containing 10% fetal bovine serum (FBS) (HyClone, Logan, UT, USA), 100 U/mL penicillin, 100 μL/mL streptomycin, and 2 mM glutamine. Cells were washed three times with Hanks' balanced salt solution (HBSS) and passaged 1:10 using 0.25% trypsin (Mediatech, Inc.). Experiments were performed with cells from passages 10 to 30, and subsequent cultures were refreshed from frozen stocks.
Cell treatments
The synthetic rexinoid LG268 was kindly provided by Ligand Pharmaceuticals (San Diego, CA, USA). LG268 and the PPARγ ligands rosiglitazone and GW-9662 were solubilized in dimethylsulfoxide, stored in 10 μM aliquots at -20°C, and added to culture media at varying concentrations. Recombinant human IL-1β (Promega Corporation, Madison, WI, USA) was solubilized in sterile H2O, stored in 10 μg/mL aliquots at -80°C, and added to media at 1 ng/mL. For most experiments, SW-1353 cells were grown to approximately 90% confluence in six-well plates and washed twice with HBSS to remove trace serum and waste metabolites. Two milliliters of serum-free DMEM supplemented with 0.2% lactalbumin hydrosylate (DMEM/LH) and appropriate concentrations of LG268 and/or rosiglitazone were added for 1 to 24 hours. IL-1β was then added to the media for an additional 1 to 24 hours followed by cell harvest.
Quantitative real-time reverse transcription-polymerase chain reaction
After experimental treatment, the cells were washed twice with cold 1× phosphate-buffered saline (PBS), scraped off the plate, and homogenized using QIAshredder spin columns (Qiagen Inc., Valencia, CA, USA). Total cellular RNA was isolated using the RNeasy Mini Kit (Qiagen Inc.) in accordance with the manufacturer's instructions, including DNA contamination removal by on-column treatment with the RNase-Free DNase Kit (Qiagen Inc.). The reverse transcription (RT) reaction was performed on 4 μg of purified total RNA using Moloney murine leukemia virus reverse transcriptase (Invitrogen Corporation, Carlsbad, CA, USA) with oligo(dT) or random hexamer primers (Applied Biosystems, Foster City, CA, USA) for mRNA and heterogeneous nuclear RNA (hnRNA) studies, respectively. The RT reactions were performed in a PTC-100 thermal cycler (MJ Research, now part of Bio-Rad Laboratories, Inc., Hercules, CA, USA). Real-time polymerase chain reaction (PCR) was performed using the SYBR Green PCR Master Mix kit (Applied Biosystems) in accordance with the manufacturer's instructions. PCRs were run with experimental triplicates and machine (on-plate) duplicates or triplicates for each sample. To enable quantitative between-plate comparisons, standard curves were generated with each mRNA assay. Both experimental and standard reactions were run using 125 ng each of the appropriate forward and reverse primers for the MMPs analyzed (sequences described previously in [
21]). Target gene expression was normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA expression and reported as mean copies ± standard deviation of target gene mRNA per copy of GAPDH mRNA. Several real-time RT-PCR experiments in which standard curve plasmids were not available were performed. In these cases, the relative mRNA levels of the experimental gene under different treatment conditions were normalized to GAPDH mRNA levels using the 2
-ΔΔCt statistical method [
29].
Western blotting
Trichloracetic acid (TCA) protein precipitation and Western blotting were performed as described previously [
21]. Briefly, SW-1353 cells were grown to confluency in six-well plates in DMEM with 10% FBS. The media were aspirated, the cells were washed with HBSS, and 2 mL of DMEM/LH was added to each well. Cells were pretreated for 24 hours with rosiglitazone, LG268, or both, and IL-1β was added for an additional 24 hours. Protein was TCA-precipitated from 1 mL of media from each well and resuspended in 40 mL of Laemmli buffer. Samples were resolved using Tris-HEPES-SDS precast 10% polyacrylamide gels (catalog number 25201; Pierce, Rockford, IL, USA) and transferred to an Immobilon-P polyvinylidene difluoride membrane (Millipore Corporation, Billerica, MA, USA). The membranes were probed for
MMP-1 using a polyclonal rabbit anti-human
MMP-1 antibody (AB8105; Chemicon International, Temecula, CA, USA) or for
MMP-13 with a polyclonal
MMP-13 antibody generously provided by Peter Mitchell (Pfizer Inc, New York, NY, USA). Protein bands were visualized by incubation with a goat anti-rabbit secondary antibody conjugated to horseradish peroxidase (Cell Signaling Technology, Inc., Danvers, MA, USA) and enhanced chemiluminescence analysis with the Western Lightning reagent (PerkinElmer Life and Analytical Sciences, Inc., Waltham, MA, USA).
Collagen degradation assay
This assay was performed as previously described [
21,
30]. Briefly, fibrillar collagen preparations were made from Vitrogen 100 bovine type I collagen (Cohesion Technologies, Inc., Palo Alto, CA, USA) in accordance with the manufacturer's instructions. The collagen solution was diluted to 2 mg/mL and the pH was adjusted to 7.3 ± 0.2. Once neutralized, an equivalent volume of DMEM/LH containing SW-1353 cells was added, resulting in a final collagen concentration of 1 mg/mL and 2.5 × 10
5 cells per well of a six-well plate. Rosiglitazone, LG268, or both were added to the collagen/cell suspension of specific experimental wells. Following incubation at 37°C for 60 minutes, the collagen gelled and 1 mL of DMEM/LH was added on top of the cell-containing collagen plug. After 24 hours of incubation in DMEM/LH, IL-1β was added to the media to induce MMP production and subsequent collagen degradation. Approximately 24 hours after the addition of IL-1β, the media were removed from each well and weighed to quantify the extent of collagen degradation.
Luciferase reporter assays
Luciferase reporter plasmids incorporating four copies of the putative overlapping DR-1/AP-1 site of the
MMP-1 (MMP1-ENDOG-Luc) or
MMP-13 (MMP13-ENDOG-Luc) promoters were constructed using the pGL3-basic plasmid (Promega Corporation). Control reporters were constructed in a similar fashion, with four scrambled copies of the DR-1/AP-1 element of
MMP-1 (MMP1-SCRAM-Luc) or
MMP-13 (MMP13-SCRAM-Luc). SW-1353 cells were plated in six-well plates at a density of 1.5 × 10
5 cells per well. The next day, cells were transiently transfected in six-well plates with 2 μg/well of the PPRE-tk-luciferase plasmid [
31], or the custom DR-1/AP-1-luciferase plasmids described above, using 5 μL/well of Lipofectamine 2000 (Invitrogen Corporation) in accordance with the manufacturer's instructions. Four to six hours after transfection, cells were washed twice with HBSS followed by the addition of 2 mL of DMEM/LH media containing the indicated NHR ligand. After 24 hours of ligand treatment, IL-1β was added to the media for an additional 24 hours. The cells were then washed three times with cold 1× PBS, and lysates were harvested using 1× Passive Lysis Buffer (Promega Corporation). Protein concentration was determined using Bio-Rad Protein Assay reagent (Bio-Rad Laboratories, Inc.), and equal amounts of total protein were loaded for each sample. Luciferase activity was measured in relative light units using an Lmax II luminometer (Molecular Devices Corporation, Sunnyvale, CA, USA).
Chromatin immunoprecipitation
The chromatin immunoprecipitation (ChIP) protocol was adapted from the 'fast ChIP method' [
32]. SW-1353 cells were grown to confluence in 150-mm plates (approximately 10
7 cells). Crosslinking was performed by adding 40 μL of 37% formaldehyde per milliliter of cell culture media directly to the culture media, and the plates were rocked gently at room temperature for 10 minutes. Crosslinking was quenched by adding 141 μL of 1 M glycine per milliliter media and gently rocking for 5 minutes at room temperature. Cells were washed twice with ice-cold 1 × PBS, scraped, and collected in 15-mL conical tubes on ice. Cells were pelleted by centrifugation at 2,000
g for 5 minutes at 4°C, resuspended in 1 mL of ChIP buffer (150 mM NaCl, 50 mM Tris HCl pH 7.5, 5 mM EDTA [ethylenediaminetetraacetic acid], 0.5% NP40, 1% Triton X-100) with protease inhibitors (complete mini tabs; Roche, Nutley NJ, USA), and lysed on ice for 10 minutes. Nuclei were collected by centrifugation at 12,000
g for 1 minute at 4°C and then washed twice by aspirating the supernatant and resuspending with 1 mL of ChIP buffer. Chromatin was sonicated on ice with 15 × 15 second pulses at power setting #40 on a Sonics Vibro-Cell VC 130PB-1 ultrasonoic processor (Newtown, CT. USA). Debris was cleared by centrifugation at 12,000
g for 10 minutes at 4°C, and the supernatant was split into 200-μL aliquots in 1.5-mL microcentrifuge tubes for immunoprecipitation (IP). Two micrograms of specific antibodies to the HA epitope tag (Abcam, Cambridge, UK), acetylated histone H4 (Upstate, now part of Millipore Corporation), PPARγ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), or normal IgG was added to each tube, and tubes were rotated overnight at 4°C. Twenty microliters of protein A/G agarose (Santa Cruz Biotechnology, Inc.) per IP was washed three times, resuspended 1:1 with ChIP buffer, and distributed (40 μL per IP) to 1.5-mL microcentrifuge tubes. IP reactions were centrifuged at 12,000
g for 10 minutes at 4°C, and then 180 μL of supernatant was transferred to the protein A/G agarose tubes and rotated for 45 minutes at 4°C. Beads were collected by centrifugation at 2,000
g for 30 seconds at 4°C and then washed five times by removing the supernatant and resuspending in ice-cold ChIP buffer. After washing, the pellet was resuspended in 100 μL of 10% Chelex-100 (Fisher Scientific Co., Pittsburgh, PA, USA), boiled for 10 minutes, and then cooled on ice. One microliter of proteinase K (20 μg/μL) was added to the cooled solution, vortexed, incubated at 55°C for 30 minutes, boiled for 10 minutes, and centrifuged at 12,000
g for 1 minute. Eighty microliters of supernatant was transferred to a new microcentrifuge tube, and 120 μL of water was added back to the original tube, vortexed, and centrifuged as before, and 120 μL of supernatant was transferred to the previous 80 μL. Samples were stored at -20°C or immediately quantified using real-time PCR with primers flanking the DR-1/AP-1 site or with negative-control primers flanking an upstream control region (-3 kb for
MMP-1 and -1 kb for
MMP-13), normalized to IgG-precipitated DNA, and expressed as a fold-change over untreated cells.
Immunoprecipitation
For the SUMO IP experiments, cellular proteins were immunoprecipitated following the ExactaCruz system instructions from Santa Cruz Biotechnology, Inc. Briefly, cells were grown to confluency in 150-mm dishes and treated for 1 hour with LG268 (50 nM) and/or rosiglitazone (50 nM) in serum-free DMEM/LH media. The cells were then treated with 1 ng/mL IL-1β for 1 hour and harvested using cold radioimmunoprecipitation assay buffer. Cell lysates were homogenized using QIAshredder columns. IP reactions were performed using 4 μg of anti-SUMO-1 (Santa Cruz Biotechnology, Inc.). IP fractions were resolved using PAGE as described above. The presence of RXR and PPARγ in the IP fractions was detected using 4 μg of RXR (ΔN197) or PPARγ (H-100) antibodies from Santa Cruz Biotechnology, Inc.
Discussion
Previous work has demonstrated that the PPARγ ligand rosiglitazone [
17,
18,
38] and the RXR ligand LG268 [
21] each inhibit proinflammatory cytokine induction of
MMP-1 and
MMP-13 gene expression. In this study, we address the inhibitory effects of adding both ligands on
MMP-1 and
MMP-13 expression in IL-1β-stimulated SW-1353 cells and investigate the mechanisms responsible for this inhibition. We show that rosiglitazone treatment selectively reduces
MMP-1 and
MMP-13 mRNA and note that the pattern and magnitude of MMP reduction parallel those seen with LG268 [
21]. Additionally, combined treatment results in greater reduction of
MMP-1 and
MMP-13 mRNA and hnRNA than those seen with a single ligand. We show that the effect of combined treatment on
MMP-1 and
MMP-13 is observed at the protein level, where the addition of both ligands leads to decreased collagen destruction by IL-1β-activated SW-1353 chondrosarcoma cells over single-ligand treatment. Our investigations into the molecular mechanisms of this inhibition centered on the possibility of a shared mechanism since both ligands bind to the RXR:PPARγ heterodimer. We explored the role of RXR:PPARγ binding to the DR-1/AP-1 site in suppressing
MMP-1 and
MMP-13 production and addressed possible mechanisms of inhibition, including competitive binding between RXR:PPARγ and AP-1 proteins, altered histone acetylation at the DR-1/AP-1 promoter element, and changes in SUMOylation of PPARγ and RXR.
The competitive binding model identified binding of PPARγ to a degenerate DR-1 site as central to the inhibition of rabbit
MMP-1 by rosiglitazone, and initial evidence suggested that the inhibition was due to competition between PPARγ and AP-1 for binding at the degenerate DR-1 site [
17]. In keeping with this model, we show that, in SW-1353 cells, rosiglitazone and LG268 result in similar levels of consensus DR-1-driven reporter activation, suggesting that treatment with each compound may increase binding to the DR-1 element. However, we did not observe similar results with luciferase reporter constructs driven by the endogenous DR-1/AP-1 elements from the human
MMP-1 and
MMP-13 promoters. These data contrast with previous studies in which IL-1β-driven luciferase reporter activity was inhibited in a dose-dependent manner by rosiglitazone; however, those experiments were conducted in rabbit cells transiently transfected with a luciferase construct driven by the rabbit
MMP-1 promoter [
17], whereas the present investigation uses human constructs in human cells. Previously, we noted a discrepancy in the behavior of transiently transfected
MMP-1 promoter constructs in rabbit [
39] versus human [
20,
21,
40] cells in response to treatment with IL-1β. We attribute these differences to a more complex regulation of the human gene and emphasize the importance of measuring expression of the endogenous gene. Since reporter constructs driven by elements of the human
MMP-1 and
MMP-13 promoters often do not mirror expression of the endogenous genes, we shifted to a more direct,
in vivo approach, using ChIP assays to detect changes in PPARγ binding at the DR-1/AP-1 site of the endogenous
MMP-1 and
MMP-13 promoters in genomic DNA.
A key aspect of the competitive binding model is mutually exclusive binding of PPARγ and AP-1 transcription factors at the DR-1/AP-1 element, and when considering the competitive binding model, one would expect a decrease in PPARγ binding at the site in cells treated with IL-1β as compared with rosiglitazone- or LG268-treated cells. On the contrary, ChIP analysis using either an HA-PPARγ expression construct (Figure
7) or endogenous PPARγ (Figure
6) detected an increase in PPARγ at the DR-1/AP-1 sites in IL-1β-treated cells, but not rosiglitazone- or LG268-treated cells. These data are inconsistent with the competitive binding model. The large increase in PPARγ at the DR-1/AP-1 site in IL-1β-treated cells was unexpected and may indicate a novel function of PPARγ in IL-1β signaling at the
MMP-1 and
MMP-13 promoters. Other studies have implicated IL-1β in regulating expression of PPARγ in chondrocytes, with evidence for both decreasing and increasing PPARγ expression [
41,
42]. IL-1β inhibits PPARγ mRNA expression in SW-1353 cells, as measured by real-time RT-PCR (data not shown). However, this repression does not affect expression of the PPRE luciferase reporter construct (Figure
5). Therefore, we speculate that PPARγ may be interacting with AP-1 transcription factors at the proximal promoter regions of
MMP-1 and
MMP-13. Incorporation of PPARγ in the AP-1 complex may place the nuclear receptor in a position to more efficiently regulate AP-1-driven transcription of
MMP-1 and
MMP-13. PPARγ directly interacts with at least one member of the AP-1 transcription factor family [
43], and there are examples of other NHRs directly interacting with the AP-1 transcription factors [
44‐
46].
As the PPARγ ChIP did not appear to support the competitive binding model, we next used ChIP to examine the
MMP-1 and
MMP-13 promoters for evidence of nuclear receptor-associated coactivator and corepressor activity by detecting changes in histone acetylation. We have previously shown that IL-1β leads to an increase in histone acetylation at these promoters in SW-1353 cells and that this increase in acetylation is blocked when the cells are pretreated with LG268 [
21]. Our results show that, as seen previously with LG268, treatment with rosiglitazone prevented histone acetylation at the DR-1/AP-1 site in the proximal promoters of
MMP-1 and
MMP-13. Most importantly, dual treatment with rosiglitazone and LG268 resulted in the additive reduction of histone acetylation at both promoters. Given that HDAC activity is typically associated with transcriptional repression [
47], this result is consistent with the decrease in
MMP-1 and
MMP-13 gene expression in cells treated with a combination of both ligands, suggesting that the compounds may be inhibiting expression of these genes through a common histone acetylation-associated mechanism.
To further investigate this mechanism, we returned to the PPARγ literature. The model describing inhibition of proinflammatory genes by a SUMOylated form of PPARγ proposed by Pascual and colleagues [
14] is attractive for several reasons. First, we are investigating the inhibition of MMPs induced by IL-1β, a prototypical inflammatory cytokine. Second, the model requires that the SUMOylated NHR act to anchor the corepressors before they are released by proinflammatory stimuli, possibly providing an explanation for the required pretreatment [
21]. Lastly, the mechanism involves regulation via post-translational modification of histones, which suggests that native chromatin conformation is important for the regulation of
MMP-1 and
MMP-13 and may help to explain the difficulties seen with transiently transfected luciferase reporter constructs containing endogenous promoter sequences [
20,
21].
Our data show that treatment with either rosiglitazone or LG268 induces a multiply SUMOylated form of PPARγ (Figure
9a), suggesting that both compounds may work through PPARγ to inhibit proinflammatory genes. This result was rather surprising because, in addition to confirming that a PPARγ ligand can cause SUMOylation of PPARγ in SW-1353 cells, it indicates that treatment with an RXR ligand causes SUMOylation of PPARγ. In addition, we show that LG268 leads to SUMOylation of RXR. This suggests that, similar to rosiglitazone and PPARγ, LG268 inhibits proinflammatory genes by inducing a SUMOylated form of its target receptor, RXR. Interestingly, rosiglitazone also induces SUMOylation of RXR, a result that complements our observation demonstrating LG268-induced SUMOylation of PPARγ. These data may be explained by the fact that, when RXR and PPARγ heterodimerize, they are in close proximity to one another. When the SUMOylation machinery is recruited to the heterodimer through the liganding of one partner, attachment of SUMO may be a somewhat leaky process, thereby leading to SUMOylation of the unliganded dimer partner.
Our data also suggest that, while treatment with LG268 or rosiglitazone induces SUMOylation of RXR, the PPARγ ligand preferentially leads to a higher molecular weight form that is consistent with a doubly SUMOylated version of the receptor. To address this point, we analyzed the protein sequence of RXRα and have identified three putative consensus SUMOylation sites (K201, K245, and K364). The presence of multiple SUMOylation sites on RXR may help to explain the different molecular weight forms of SUMOylated RXR by suggesting that rosiglitazone induces two SUMO molecules to be added to RXR whereas LG268 induces conjugation of only a single SUMO molecule. Because both LG268 and rosiglitazone inhibit IL-1β-induced MMP production and only a singly SUMOylated form of RXR persists when IL-1β is added, perhaps only one site needs to be SUMOylated to cause the inhibitory effect. Interestingly, the K364 site in RXR is very similar in location to the K365 site in PPARγ that Pascual and colleagues [
14] demonstrated was required for rosiglitazone-mediated inhibition. Because of the conserved nature of NHR domain structures, K364 may be the required RXR SUMOylation site for inhibition of
MMP-1 and
MMP-13 through this mechanism.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PSB and ACS were responsible for study design, acquisition of data, analysis and interpretation of data, manuscript preparation, and statistical analysis and contributed equally to this work. YR was responsible for acquisition of data, analysis and interpretation of data, and manuscript preparation. MBS was responsible for study design and manuscript preparation. CEB was responsible for study design, analysis and interpretation of data, and manuscript preparation. All authors read and approved the final manuscript.