Approximately 40% of people older than 70 years are affected by osteoarthritis (OA), making the disease the most common form of arthritis worldwide. It is a complex degenerative disease of the whole joint but is epitomized by focal degradation of articular cartilage. Cartilage is composed of an extracellular matrix (ECM) rich in type II collagen and the proteoglycan aggrecan, the synthesis and degradation of which are tightly regulated in healthy individuals by the resident cell type, the chondrocyte (1). In patients with OA, this homeostatic balance is shifted toward ECM degradation, which is mediated by proteolytic enzymes, including the collagenases and aggrecanases. Direct inhibition of the activity of these metalloproteinase enzymes, although effective in vitro and in vivo in animal studies, has failed in clinical trials due to adverse side effects, partly because of a lack of selectivity (2). For this reason, other strategies to reduce the expression of these catabolic enzymes, such as modulation of inflammation, have been considered.
Many metalloproteinase enzymes are highly induced by inflammatory cues, including proinflammatory cytokines, and although still a subject of some controversy, inflammation does appear to play a role in OA, as evidenced by the presence of synovial inflammation (synovitis) in the disease (3). The major signaling pathway through which inflammatory mediators impart their action is via NF-κB, a transcription factor dimer, the composite subunits of which provide the specificity to its transcriptional response. In this issue of Arthritis & Rheumatism, de Andrés and colleagues demonstrate a potentially novel mechanism for inflammatory gene regulation in OA chondrocytes, the demethylation of an enhancer that allows the binding of NF-κB and thus facilitates transcription of the regulated gene, inducible nitric oxide synthase (iNOS/NOS2) (4).
“Epigenetic”' refers to a nongenetic mechanism of gene regulation that could be passed on to daughter cells. Such mechanisms are thought to underlie developmental programs and the maintenance of a stable differentiated phenotype. Epigenetic regulation can be evoked by three basic mechanisms: DNA methylation, histone modifications, and noncoding RNAs. There is increasing evidence that these three mechanisms are tightly linked and intertwined.
DNA methylation represents the most stable epigenetic modification and is the conversion of a cytosine at CpG dinucleotides to 5-methylcytosine (5mC) by the addition of a methyl group by a DNA methyltransferase enzyme family member. Interestingly, a food supplement taken by some OA patients, but with limited evidence of efficacy, S-adenosyl methionine, is the methyl donor used by these enzymes (5).
CpG sequences are often clustered in “islands,” areas of high CpG density, near the promoters of about half of all genes. Methylation of these CpG islands correlates with repression of the expression of the corresponding gene. This mechanism is believed to be especially important in the silencing of transposons (6).
In differentiated cells, such as chondrocytes, promoter-associated CpG islands are generally unmethylated, meaning that genes are accessible for transcription as and when required. In fact, most tissue-specific DNA methylation seems to occur at regions termed “shores,” which are regions of lower CpG density that reside close to CpG islands (7). CpG island and shore–containing genes (or genomic regions) are likely to be regulated by general 5mC DNA binding proteins, such as methyl-CpG–binding protein 2, and subsequent chromatin modifications to generally silence gene expression.
Of course, the remaining 50% of genes lack CpG islands and shores, as is the case for iNOS. It is probable that the mechanism of regulation of island/shore-containing genes and those with only sparse CpG sites differ. In a simplistic model, 5mC within sparse CpG-containing genes regulates the binding of transcription factors or repressors, leading to an altered transcriptional state of the cell. The findings of de Andrés and colleagues indicate that such a mechanism occurs at the iNOS enhancer in OA, where demethylation of the enhancer allows the binding of NF-κB, and thus, induces iNOS expression (4). Certainly, it has been known for a number of years that NF-κB DNA binding is methylation sensitive (8). Others have reported that the sparse CpG promoter of matrix metalloproteinase 13 (MMP-13), the metalloproteinase almost certainly responsible for type II collagen cleavage in cartilage, was differentially methylated in OA. As in the study by de Andrés et al, this allowed the binding of a transcription factor, which in the case of MMP-13 was CREB and its partner, CREB binding protein, rather than NF-κB, and the subsequent induction of the gene (9).
Inducible NOS is the proinflammatory cytokine–inducible form of nitric oxide synthase. NO is a short-lived molecule most noted for its vasodilatory properties. In skeletal muscle, small increases in NO production lead to beneficial effects, such as the regulation of glucose uptake and perhaps blood flow. The role of NO produced by cartilage chondrocytes is less clear, but it is possible that excess NO leads to apoptosis. High concentrations of NO can interact with reactive oxygen species such as superoxide (O2·−) to form the reactive nitrogen species peroxynitrite, causing damage to cells and cellular components through the formation of nitrotyrosine, the presence of which accumulates in aging tissue and OA cartilage (10).
De Andrés and colleagues very clearly demonstrate an up-regulation of both the iNOS gene and protein in OA cartilage compared to normal cartilage (obtained from patients with femoral neck fracture). However, in a similar study, microarray analysis showed a down-regulation of the gene in late-stage OA (11). An early study also failed to identify iNOS in OA chondrocytes unless they were stimulated by interleukin-1 (IL-1), a stimulus often used to induce iNOS in OA chondrocytes (12). Further studies are therefore needed to examine the expression of iNOS not only in late-stage OA, but also, importantly, during the progression of human OA, especially because selective inhibition of iNOS has been shown to reduce the progression of experimental OA in vivo (13).
Several other studies have investigated the role DNA methylation could play in OA in general, focusing on 5mC status and gene expression at the promoters of candidate genes known to be important for cartilage maintenance. These have included the promoters of a number of metalloproteinases, IL-1β, superoxide dismutase 2, bone morphogenetic protein 7 (also known as osteogenic protein 1), growth differentiation factor 5 (GDF-5), leptin, SOX9, and runt-related transcription factor 2 (also known as core-binding factor α1) (5), several of which were examined by the same team of scientists reporting iNOS enhancer demethylation in this issue of Arthritis & Rheumatism (4).
Candidate gene studies are likely to continue to identify differentially methylated regions that may well be important in OA. In general, published studies have shown a correlation between promoter methylation and the expression of the associated gene. However, although the genes are up-regulated in OA, the promoters of aggrecan or type II collagen (COL2A1) are hypomethylated, irrespective of the disease status of the cartilage from which the chondrocytes are isolated (14, 15). This could suggest that DNA methylation may not be important in the regulation of these genes or perhaps that the correct genomic region has yet to be analyzed. Importantly, the study by de Andrés and colleagues (4) demonstrates that the proximal promoter may not be the major site of regulation by DNA methylation and that enhancer methylation should be examined. Indeed, both aggrecan and COL2A1 have several well-characterized enhancers that should be analyzed for differential methylation.
Enhancers are segments of genomic regulatory sequences that act either regionally or frequently over a large distance from their targets to provide much of the instructional information that controls both spatial and temporal gene expression, especially during development. Thus, it is very likely that these elements will be platforms for epigenetic regulation. In fact, recent data indicate that many enhancers are transcribed into long noncoding RNAs, which positively regulate the expression of nearby protein coding genes, emphasizing the complex interplay of the various epigenetic mechanisms (16).
The studies of DNA methylation in OA have been small scale and low throughput, and as a result, as described, focused on the analysis of genes already thought to be important in the disease. Recent technical advances, through the use of genome tiling microarrays (or similar techniques) or high-throughput DNA sequencing, have revolutionized the study of epigenetics in other research fields, allowing almost complete genome-wide CpG methylation analysis. Although no current technique for global CpG analysis is entirely comprehensive, the observations from the OA candidate gene studies indicate that such a screen in OA is warranted. It will of course be important to choose the correct tissue for these studies, with analysis of other joint tissue along with cartilage being important.
At the moment, it is not possible to determine if the methylation changes observed in OA are the cause or, in fact, a consequence of the gene expression changes observed or even if these changes actually genuinely affect gene expression in vivo; of course, these are major challenges for all epigenetic research. In OA this is in part because almost all of the work has focused on tissue with end-stage disease, which is accessible only after joint replacement surgery. Analysis of DNA methylation changes in in vitro models of chondrogenesis or, ideally, in OA progression in vivo may be able to resolve some of these concerns.
The causes of primary OA are currently unclear, but are likely a consequence of the interaction of local and systemic risk factors, with genetics being a major risk factor for the development of the disease. Although heritability studies suggest that genetic background accounts for ∼50% of the risk of developing OA, only a few loci have been associated with the disease at genome-wide significance levels despite extensive efforts, and even these signals have a limited effect size (17).
As a result of the lack of associated genetic loci, it has been suggested that epigenetic modifications could in part provide this missing heritability for OA. Although there is increasing evidence of transgenerational epigenetic inheritance in mammals (see, for example, ref.18), this is likely to be limited because germ cells, which give rise to the gametes, undergo a resetting of their genomic methylation pattern, which is essential in their reprogramming to a totipotent state (19). However, although methylation patterns may be wiped for each generation, it is logical to assume that the genotype is important for determination of these patterns and therefore their reinstigation. For example, DNA methylation patterns correlate more between monozygotic than dizygotic twins and more between twins than between unrelated pairs of individuals (20). These data indicate a degree of DNA methylation heritability, even if this is due to underlying genome heritability. Therefore, a rational first step would be to understand how epigenetic modifications or factors affect the penetrance of the currently identified OA susceptibility alleles (17), integrating genetic and epigenetic data and, thus, adding credence to the findings of recent genome-wide association studies (21). For example, a single-nucleotide polymorphism in GDF-5, which is consistently and robustly associated with OA, can affect how DNA methylation influences the expression of the gene (22).
Whether epigenetic mechanisms can explain the missing heritability of OA or other common diseases remains to be determined. Epigenetic changes also accumulate over years due to a number of factors, for example, diet or environment. How differences in DNA methylation for the iNOS enhancer between OA and normal cartilage arose may perhaps be irrelevant. If such an epigenetic mechanism could be proved to be a contributing factor to disease, then a highly targeted therapeutic intervention is possible because DNA methylation is essentially biologically reversible.
Differential iNOS enhancer methylation is on its own unlikely to be the cause of cartilage loss in OA, but these new data do indicate that enhancer methylation may be an important mechanism in the maintenance of a stable chondrocyte phenotype. The novel research by de Andrés and colleagues adds to our knowledge of the pathophysiology of the disease and thus may highlight new pathways for translational research and potential therapeutic applications.