Objective Based on previous data that have linked the small ubiquitin-like modifier-1 (SUMO-1) to the pathogenesis of rheumatoid arthritis (RA), we have investigated the expression of the highly homologous SUMO family members SUMO-2/3 in human RA and in the human tumour necrosis factor α transgenic (hTNFtg) mouse model of RA and studied their role in regulating disease specific matrixmetalloproteinases (MMPs).
Methods Synovial tissue was obtained from RA and osteoarthritis (OA) patients and used for histological analyses as well as for the isolation of synovial fibroblasts (SFs). The expression of SUMO-2/3 in RA and OA patients as well as in hTNFtg and wild type mice was studied by PCR, western blot and immunostaining. SUMO-2/3 was knocked down using small interfering RNA in SFs, and TNF-α induced MMP production was determined by ELISA. Activation of nuclear factor-κB (NF-κB) was determined by a luciferase activity assay and a transcription factor assay in the presence of the NF-κB inhibitor BAY 11-7082.
Results Expression of SUMO-2 and to a lesser extent of SUMO-3 was higher in RA tissues and RASFs compared with OA controls. Similarly, there was increased expression of SUMO-2 in the synovium and in SFs of hTNFtg mice compared with wild type animals. In vitro, the expression of SUMO-2 but not of SUMO-3 was induced by TNF-α. The knockdown of SUMO-2/3 significantly increased the TNF-α and interleukin (IL)-1β induced expression of MMP-3 and MMP-13, accompanied by increased NF-κB activity. Induction of MMP-3 and MMP-13 was inhibited by blockade of the NF-κB pathway. TNF-α and IL-1β mediated MMP-1 expression was not regulated by SUMO-2/3.
Conclusions Collectively, we show that despite their high homology, SUMO-2/3 are differentially regulated by TNF-α and selectively control TNF-α mediated MMP expression via the NF-κB pathway. Therefore, we hypothesise that SUMO-2 contributes to the specific activation of RASF.
Statistics from Altmetric.com
Hyperplasia of the synovium, infiltration of different inflammatory cells and invasion of synovial fibroblasts into joint structures are characteristic features of human rheumatoid arthritis (RA).1 ,2 Stable activation of RA synovial fibroblasts (RASFs) plays a major role in the rheumatoid destruction of the cartilage matrix,3 mainly through the expression of matrixmetalloproteinases (MMPs) which are regulated by inflammatory cytokines such as tumour necrosis factor α (TNF-α) and interleukin (IL)-1.4 ,5
Small ubiquitin-related modifiers (SUMOs) belong to a subfamily of ubiquitin-like proteins. Post-translational modification of proteins by SUMO is involved in a variety of cellular processes, including protein localisation,6 ,7 transcriptional regulation,8 ,9 protein stability, cell survival and death.10–12 There are four mammalian SUMO genes: SUMO-1, SUMO-2, SUMO-3 and SUMO-4. SUMO-2 and SUMO-3 share 96% amino acid sequence identity, while SUMO-1 is approximately 50% similar to SUMO-2 and SUMO-3; thus SUMO-2 and SUMO-3 are referred as SUMO-2/3.13 SUMO-2 and SUMO-3 can be modified in vivo by SUMO-1 and SUMO-2/3.14 ,15 Interestingly, SUMO can be targeted by post-translational modifications belonging to the ubiquitin family and by acetylation and phosphorylation. For SUMO-2 it was shown that the lysine residues K11, K32 and K41 can be modified by ubiquitin.16 ,17 A recent study, showed that SUMO-3 can be phosphorylated at serine 2, while SUMO-2 cannot be phosphorylated because it has an alanine at this position.18 Thus, these may be functional differences between SUMO-2 and SUMO-3 which, however, have not been investigated in detail.
In line with their important role in cellular functions, the expression and activity of MMPs are tightly regulated at multiple levels of gene transcription, synthesis and extracellular activity.
Previously, it was shown that the MMP-1 promoter is hyperacetylated in RASF compared with SFs from osteoarthritis (OA) patients (OASFs). Interestingly, overexpression of SUMO specific protease-1, which is downregulated in RASF, leads to normalisation of the acetylation pattern in the MMP-1 promoter and decreases the production of MMP-1.19 Cytokines such as TNF-α and IL-1β stimulate the production of MMPs through the activation of cellular signalling pathways involving mitogen-activated protein kinases and nuclear factor-κB (NF-κB).20
In the current study, we have investigated the expression of SUMO-2/3 in human RASFs and in human TNF-α transgenic (hTNFtg) mice,21 where transgenic overexpression of the hTNF-α leads to a chronic inflammatory and destructive polyarthritis. Here, we demonstrate for the first time a role of SUMO-2/3 in the activation of TNF-α mediated MMP-3 and MMP-13 expression in RASFs. We show that SUMO-2/3 is part of regulatory mechanisms that limit expression of distinct MMPs by inhibiting the activation of NF-κB.
Elevated expression of SUMO-2/3 in the synovium and in synovial fibroblasts from patients with RA
First, we addressed the question of whether SUMO-2/3 is expressed in synovial fibroblasts from the inflamed synovium of human RA patients. As shown in figure 1A, immunohistochemical stainings of human RA synovial tissue revealed high expression of SUMO-2/3. The proteins are predominantly localised in the synovial lining layer as well as in the sublining layer, whereas only marginal expression was found in synovial tissue of patients with OA. For further analysis of the subcellular distribution of SUMO-2/3 in RASFs and OASFs, immunocytochemical stainings were performed. SUMO-2/3 was found predominantly within the nucleus (figure 1B). When compared with OASFs, RA cells exhibited a markedly enhanced nuclear staining for SUMO-2/3. These data were confirmed by western blot analysis at protein level (figure 1C) using a SUMO-2 specific antibody (see online supplementary figure S3) which additionally recognises SUMO-3. Interestingly, analysis by quantitative real-time PCR in figure 1D shows an increased expression of SUMO-2 in fibroblasts isolated from RA synovium compared with the control cells from patients with OA. The mRNA level of SUMO-3 was far less increased in RASFs compared with in OASFs.
TNF-α regulates SUMO-2, but not SUMO-3
The expression data led us to investigate whether SUMO-2/3 is regulated by inflammatory cytokines such as TNF-α. To this end, we stimulated synovial fibroblasts from RA patients (n=5) with recombinant human TNF-α and analysed whether there is a dose-response effect of TNF-α (0,1; 1; 10 and 100 ng/ml) on SUMO-2 and SUMO-3 at transcriptional level and subsequently analysed the expression of SUMO-2/3 at protein level. As shown in figure 1E, TNF-α (100 ng/ml) stimulation of synovial fibroblasts from RA patients for 24 h resulted in an upregulation of SUMO-2/3 at protein level. In addition, the conjugation of target proteins by SUMO-2/3 was increased. Interestingly, TNF-α was only able to induce a dose-dependent increase in SUMO-2 mRNA expression, while SUMO-3 remained unaffected (figure 1F). These data suggest that only SUMO-2 is regulated by TNF-α and that although SUMO-2 and SUMO-3 proteins are closely similar in their amino acid sequence, they perform different functions during inflammatory conditions. Next, we investigated the expression of SUMO-2/3 in hTNFtg mice,21 as a model for inflammatory polyarthritis. For this, we first performed immunohistochemical stainings of hTNFtg mice in tissue sections. As shown in figure 2A, we confirmed the increased expression of SUMO-2/3 in tissue sections from hTNFtg mice as well as in western blot analysis of primary synovial fibroblasts in comparison with wild type (wt) samples (figure 2C). SUMO-2 mRNA level was strongly increased in synovial fibroblasts from hTNFtg mice compared with wt mice, but the SUMO-3 RNA level was not increased to the same extent (figure 2B).
Knockdown of SUMO-2/3 affects MMP-3 and MMP-13 expression in synovial fibroblasts from RA and OA patients
To investigate the role of elevated SUMO-2/3 expression in RA, we analysed the production of MMP-1, MMP-3 and MMP-13 in synovial fibroblasts obtained from RA and OA patients. As expected, the levels of MMP-1, MMP-3 and MMP-13 expression were strongly upregulated in RASF compared with in OASF. Interestingly, we found that knockdown of SUMO-2 and SUMO-3 using specific small interfering RNA (siRNA), leads to a significant upregulation of MMP-3 and MMP-13 after stimulation with TNF-α in comparison with the cells transfected with mock siRNA (figure 3A) without changing TNF-Receptor I expression (see online supplementary figure 1A). In contrast, the production of MMP-1 was not affected by silencing SUMO-2/3 (figure 3A). These observations could be confirmed at the transcriptional level by semiquantitative PCR (figure 3B). We found a significant upregulation of TNF-α induced MMP-3 expression in RASFs, but not of MMP-1 after knockdown of SUMO-2/3. In addition to TNF-α stimulation we used the cytokine IL-1β, where a similar effect on MMP-3 and MMP-13 production after SUMO-2/3 knockdown was observed (see online supplementary figure S2A).
SUMO-2/3 regulates the activity of MMP-3 and MMP-13 regulation via the NF-κB pathway
Based on our findings, we analysed underlying mechanisms of how SUMO-2/3 regulates MMP-3 and MMP-13. As MMP-3 and MMP-13 are regulated predominantly through the NF-κB pathway, we analysed the functional consequences of SUMO-2/3 knockdown on the activity of NF-κB. First, in light of data showing an effect of SUMO-1 on NF-κB activity via the modification of IκB-α,22 we measured the expression of SUMO-1 after SUMO-2/3 knockdown using real time PCR and found that the expression of SUMO-1 was not affected (see online supplementary figure S1B). The direct role of SUMO-2/3 in NF-κB activation was assessed following SUMO-2/3 knockdown using siRNA. As shown in figure 4A and B, the loss of SUMO-2/3 enhanced the activity of NF-κB after stimulation with 100 ng/ml TNF-α for 16 h. Transfection of HeLa cells with siRNA against SUMO-2/3 resulted in a knockdown that was detectable at the protein level (figure 4A). These results indicate that downregulation of SUMO-2/3 positively regulates NF-κB transcriptional activity and hence may increase the expression of MMP-3 and MMP-13. Based on this observation, we used the NF-κB inhibitor BAY 11-7082 to determine whether upregulation of MMP-3 and MMP-13 via SUMO-2/3 was indeed dependent on the NF-κB pathway. Figure 5A and B show that inhibition of TNF-α induced phosphorylation of IκB-α and NF-κB activation using the BAY 11-7082 inhibitor totally blocked SUMO-2/3 induced MMP-3 and MMP-13 production. The MMP-13 levels after treatment with BAY 11-7082 were undetectable.
Materials and methods
Isolation of synovial fibroblasts and cell culture
All studies were approved by the ethics committees of the Medical University of Vienna and the University Hospital Muenster. Samples of synovial tissues from patients with RA or OA were obtained from joint replacement surgery and provided by the Department of Orthopaedic Surgery, St. Joseph Hospital, Sendenhorst, Germany, the Department of Orthopaedic Surgery of the University of Magdeburg, School of Medicine, Magdeburg, Germany, and the Department of Orthopaedic Surgery, KMG-Kliniken Kyritz, Germany. Murine synovial fibroblasts were isolated from tarsus of hind paws of wt and hTNFtg mice. Synovial fibroblasts were isolated by enzymatic digestion and were cultured in Dulbecco's Modified Eagle Medium (DMEM) (Invitrogen)+10% fetal calf serum (FCS) (PAA)+antibiotics/antimycotics (PAA). All cells were cultured in DMEM with 10% FCS at 37°C and 5% CO2 from passage 3 to 5 for all experiments.
Immunohistochemical analysis of human and mouse synovial tissue
For immunohistochemical analysis, deparaffinised, ethanol-dehydrated tissue sections of human synovial tissue were pretreated with trypsin, blocked with 10% horse serum, stained with antibodies against SUMO-2/3 (Zymed Laboratories), and counterstained with methyl green (Sigma-Aldrich).
siRNAs and transfection
A single siRNA was used to suppress SUMO-2 and SUMO-3 expression (SUMO-2+3 siRNA). Negative control siRNA (target sequence AATTCTCCGAACGTGTCACGT) were synthesised by QIAGEN. Transfection of siRNAs was performed by using Lipofectamine2000 (Invitrogen) according to the manufacturer's instructions. Synovial fibroblasts were transfected 48 h before further analysis.
Semiquantitative and quantitative RT-PCR
Total RNAs from RASF/OASF were prepared by RNeasy miniprep kit (QIAGEN) according to the manufacturer's protocol. cDNAs were synthesised by reverse-transcriptase. Quantitative real-time PCR was performed using the Bio-Rad iQ2 system.
The following primers were used: MMP-1 sense 5′-CTGAAGGTGATGAAGCAGCC-3′, antisense 5′-AGTCCAAGAGAATGGCCGAG-3′; MMP-3 sense 5′-CTCACAGACCTGACTCGGTT-3′, antisense 5′-CACGCCTGAAGGAAGAGATG-3′; SUMO-1 sense 5′-GACCAGGAGGCAAAACCTTCAACTG-3, antisense 5-TCTCACTGCTATCCTGTCCAATGACT-3′; SUMO-2 sense 5′-CACACCTGCACAGTTGGAAATGG-3′, antisense 5′-ACCTCCCGTCTGCTGTTGGAA-3′; SUMO-3 sense 5′-GACACTCCAGCACAGCTGGAGATG-3′, antisense 5′-AAACTGTGCCCTGCCAGGCT-3′; GAPDH sense 5′-GGTGAAGGTCGGAGTCAACGGATT-3′, antisense 5′-TGGTGACCAGGCGCCCAATACGA-3′; β-Actin sense 5′-CCACACCCGCCACCAGTTCG-3′, antisense 5′-TGCTCTGGGCCTCGTCACCC-3′.
Protein extraction and western blot analysis
Total protein from human and murine synovial fibroblasts were extracted in radioimmunoprecipitation assay (RIPA) lysis buffer (50 mM Tris/HCl, 150 mM NaCl, 1 mM EDTA, 0.1% sodium dodecyl sulfate (SDS), 0.5% deoxycholate, and 1% Triton X-100) containing complete protease inhibitor cocktail (Roche Applied Science) and 10 mM N-ethylmaleimide. For western blotting, 30 µg of total cellular protein was separated by gradient SDS-polyacrylamide gel electrophoresis (PAGE) (4–15%), transferred electrophoretically onto a polyvinylidene fluoride (PVDF) membrane (Millipore) and blocked for 1 h with phosphate buffered saline (PBS) containing 5% non-fat milk, then incubated with primary antibody overnight at 4°C, followed by three PBS washes. Endogenous SUMO proteins were visualised using a SUMO-2/3 antibody recognising SUMO-2 and SUMO-3 isoforms (Zymed Laboratories). Detection was performed using horseradish peroxidase-conjugated secondary antibodies and chemiluminescent substrates (ECL Western Blotting Detection Reagents; Amersham).
MMP expression in human RASF/OASF
MMP-1, MMP-3 and MMP-13 production by synovial fibroblasts from RA and OA patients after stimulation with recombinant human TNF-α or IL-1β for 24 h were measured by ELISA (R&D systems) measurement of the supernatants. Before stimulation with TNF-α the cells were treated with the NF-κB inhibitor BAY 11-7082, 5 µM (Adipogen) or with dimethyl sulfoxide (DMSO) for 30 min.
NF-κB p65 transcription factor assay
HeLa cells were transfected with specific siRNA against SUMO-2/3 for 48 h.
Activation of NF-κB p65 transcription factor was achieved by the stimulation of HeLa cells with 100 ng/ml TNF-α for 16 h. The nuclear extracts were prepared and NF-κB p65 transcription factor assay was performed according to the manufacturer's protocol (Active Motive).
Luciferase reporter assay
Hek 293T cells were co-transfected with siRNA SUMO-2/3 or mock siRNA, in addition to the reporter plasmid, pNF-κB-Luc (Stratagene) and Renilla luciferase vector (pRL-TK) control reporter vector (Promega) using Lipofectamine2000 (Invitrogen). The pNF-κB-Luc plasmid contains a firefly luciferase reporter gene that was derived from a basic promoter element joined to five tandem repeats of an NF-κB consensus-binding element. The pRL-TK plasmid (Promega), which expresses Renilla luciferase, was used for the normalisation of transfection efficiency. After 12 h of stimulation with 100 ng/ml TNF-α, the cells were harvested and their luciferase activities were measured using the dual luciferase reporter assay system (Promega). The data represent the means±SEM, and each transfection was performed in triplicate.
Data are shown as arithmetic means±SEM. Statistical analysis were performed using GraphPad Prism Software, 5.0c (Graph Pad Software Inc., San Diego, California, USA). According to data distribution and number of groups, a parametric (t test) or non-parametric (Mann-Whitney) test was performed. Values from ELISA data were compared by paired Student's t test. *p≤0.05; **p≤0.01; ***p≤0.001 was considered as statistically significant.
In this study, we have shown for the first time that SUMO-2/3 is involved in the activation of synovial fibroblasts in RA and, thus, the disregulation of sumoylation is likely to contribute to the pathogenesis of the disease. SUMO-2 is upregulated in RASF and regulates TNF-α and IL-1β induced expression of MMP-3 and MMP-13.
Following previous data that have demonstrated the increased expression of SUMO-1 in synovial fibroblasts from patients with RA,23 we investigated the expression of SUMO-2/3 in RASF as well as in the hTNFtg mouse model. In our study, we have demonstrated an increased expression of SUMO-2/3 in synovial tissue sections from RA patients compared with OA patients as well as in synovial fibroblasts from these two patient groups. Based on this data, we wanted to investigate the effect of the proinflammatory cytokine TNF-α, which is one of the major players contributing to RA. Interestingly, we found a dose-dependent increase in SUMO-2 expression at mRNA level, while, expression of SUMO-3 was unchanged. Additionally, protein levels of free SUMO-2/3 and of the conjugated proteins was increased. Therefore, we conclude that TNF-α selectively regulates the expression of SUMO-2, but not SUMO-3. SUMO-3, but not SUMO-2, can be phosphorylated at serine 2.18 This published functional difference between SUMO-2 and SUMO-3 may be the reason for the different contributions of TNF-α in the expression of SUMO-2 and SUMO-3. Furthermore, we have also confirmed increased expression of SUMO-2/3 in synovial tissue sections from hTNFtg mice and in synovial fibroblasts obtained from these mice compared with wt mice, suggesting that chronic exposure to TNF-α increases the expression of SUMO-2/3.
To find out which functional role SUMO-2/3 has in contributing to the pathogenesis of RA, we knocked down SUMO-2/3 and measured TNF-α and IL-1β induced MMP expression. Surprisingly, knockdown of SUMO-2/3 enhances TNF-α and IL-1β induced expression of MMP-3 and MMP-13, but not of MMP-1. TNF-α is one of the key regulators in the pathogenesis of RA and is overexpressed in synovial fibroblasts from RA patients. These data suggest that the increased expression of SUMO-2 may be part of a protective rather than a disease-promoting mechanism in synovial fibroblasts from RA patients to counteract and limit the increased and unbalanced expression of MMP-3 and MMP-13. These data raise the question of which transcription factors are involved in increasing MMP-3 and MMP-13 expression following TNF-α stimulation of SUMO-2/3 silenced cells. NF-κB plays an important role in RA pathogenesis24 and cytokine activity for example, TNF-α induces the activation of NF-κB in RA.25 Inactive NF-κB is composed of a heterodimer of p50 and p65 subunits in complex with an inhibitory IκB subunit. Activation of this factor requires phosphorylation, ubiquitination and proteasomal degradation of IκB. Using site-directed mutagenesis, it was shown that lysine K21 in IκB-α is the primary site on the target protein for sumoylation.22 SUMO-1 was shown to regulate NF-κB activity by stabilising the IκBα-NF-κB complex.22 Induction of MMPs by cytokines is cell type and MMP-specific. In articular chondrocytes and in SW1353 cells, inhibition of the transcription factor NF-κB suppresses TNF-α induced MMP-13 expression.25 NF-κB activity is also essential for upregulation of MMP-1 and MMP-3 in rabbit and human vascular smooth muscle cells.26 Our results show that silencing of SUMO-2/3 by specific siRNA upregulates the transcriptional activity of NF-κB in response to TNF-α. Our observed data is in line with a recent study, which has shown that mouse SUMO-2 modifies IκB-α and inhibits the translocation of NF-κB into the nucleus in dendritic cells.27
Our observations suggest that after knockdown of SUMO-2/3, NF-κB translocates into the nucleus and promotes increased transcription and secretion of MMP-3 and MMP-13.
With respect to the differential regulation of individual members of the MMP family, it has been well established that in addition to NF-κB, MMPs are regulated by other pathways and transcription factors, such as AP-1.2 Therefore, our results on the one hand may reflect the different levels to which MMPs are regulated specifically by NF-κB. On the other hand, sumoylation pathways have been demonstrated to regulate MMPs also through complex epigenetic pathways, which also may explain the differences.19 Additionally we have confirmed these results by pretreatment of RA synovial fibroblasts with the highly specific inhibitor of IκB-α phosphorylation BAY 11-7082. Addition of BAY 11-7082 inhibitor totally blocked SUMO-2/3 induced MMP-3 and MMP-13 production, indicating that SUMO-2/3 regulates MMP-3 and MMP-13 expression through NF-κB pathway. Furthermore, there is a growing body of evidence showing that hypoxia and SUMO activity are closely linked through regulation of hypoxia inducible factor (HIF)1-α. Different studies demonstrate this controversial hypothesis concerning the regulation of HIF1-α expression and subsequent regulation of MMP-1 and MMP-3.28–30 In our study, although we demonstrate a role of SUMO-2/3 in regulation of TNF-α mediated MMP expression via the NF-κB signalling pathway, we cannot exclude the effect of other integrating signalling pathways upon the overall regulation of MMPs in RA.
In conclusion, our results have demonstrated that downregulation of SUMO-2/3 enhances TNF-α and IL-1β induced expression of MMP-3 and MMP-13 through upregulated transcriptional activity of NF-κB. Overall, our results suggest that the fine tuning and balance of sumoylation pathways is important in the pathogenesis of RA. Furthermore, because many proteins which are regulated by SUMO modification are also targets of drugs and therapies against RA, it is important to invest in a deeper understanding of this mechanism in order to provide us with novel targets for drug design against RA.
We would like to thank Borna Truckenbrod, Rene Gronewold and Vera Eckervogt for excellent technical assistance. Furthermore, we thank George Kollias for providing us the hTNFtg mice. This work was supported by DFG (Deutsche Forschungsgemeinschaft) and by Institute for Arthritis Research (IAR) Epalinges.
Handling editor Tore K Kvien
Contributors SF: Performed research, analysed data, wrote the manuscript. MAP: Designed research, analysed data. CW: Discussion of results. SS: Discussion of results. CK-W: Discussion of results. MH: Discussion of results. AH: Discussion of results. JB: Discussion of results. JS: Discussion of results. CS: Provided tissue samples. SG: Designed research. TP: Designed research, wrote the manuscript.
Competing interests None.
Funding DFG Pa689/9-1 and IAR, Epalinges; IMI-BTCure.
Ethics approval Ethics committee Muenster, approval number 2009-049-f-s.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data were obtained by the first author.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.