|Home | About | Journals | Submit | Contact Us | Français|
It was initially envisaged that upon completion of the Human Genome Project, our understanding of human disease would lead to the development of abundant new targeted therapies with which to modify disease progression or prevent disease onset. Furthermore, advances in next‐generation sequencing (NGS) have allowed for large‐scale DNA and RNA transcriptome analysis of multiple tissue and cell types encompassing a myriad of disease states, leading to an increased understanding of the composition and regulation of the human genome. However, as a result of such studies, it has become increasingly apparent that many chronic diseases are likely to be the result of the interaction between intrinsic genes and the external environment, which can result in modification of the DNA sequence and have an impact on gene expression.
This understanding has led to the emergence of the research field known as epigenetics, which includes the analysis of DNA modifications (including methylation and histone acetylation) as well as the transcription of noncoding RNAs (ncRNAs). Indeed, the protein‐coding elements of the human genome are now known to be restricted to only 2% of the total genetic material present. Initially, the remaining DNA was thought to be mostly “junk,” but much of this so‐called “genomic dark matter” is now known to transcribe multiple families of ncRNAs 1, many of which have been shown to modulate gene expression (Figure (Figure11).
One such family of ncRNAs is the microRNA (miRNA) family 2, which binds to and prevents the translation of target messenger RNAs (mRNAs) (Figure (Figure2A).2A). Importantly, there is now overwhelming evidence that miRNAs are regulators of the inflammatory response 3, and several have been associated with multiple inflammatory diseases including osteoarthritis (OA) 3 and rheumatoid arthritis (RA) 4.
However, recent NGS studies have now identified a new family of ncRNAs known as long noncoding RNAs (lncRNAs) 5, which are defined as transcripts >200 nucleotides in length (Figure (Figure1A).1A). More than 14,000 lncRNAs have thus far been identified in humans 6, and they are currently subclassified, based on their position relative to protein‐coding genes (Figure (Figure1B),1B), as antisense lncRNAs, pseudogene lncRNAs, enhancer RNAs, intronic lncRNAs, and long intergenic noncoding RNAs (lincRNAs). These multiple variants may in part explain the functional diversity that has thus far been assigned to lncRNAs 7, including inhibiting transcriptional machinery, functioning as miRNA sponges, affecting miRNA stability, and epigenetically modifying DNA (Figures (Figures2B–D).2B–D). Many of these functions have been attributed to the theory that lncRNAs form flexible scaffolds that can bind proteins and RNAs.
Notably, increasing evidence suggests that lncRNAs are important regulators of pathologic and physiologic processes, and evidence is now emerging to suggest that lncRNAs are central regulators of the inflammatory response. Therefore, we examined evidence for the role of lncRNAs as regulators of inflammatory joint disease, by highlighting studies that have demonstrated differential expression of lncRNAs in the inflamed joint tissue of patients with RA and those with OA, as well as studies in which lncRNAs have been implicated as regulators of known inflammatory pathways relevant to joint pathology (Tables 1 and 2). LncRNAs are poorly conserved across species, which has been a limitation of studies performed to determine their function. This poor conservation places great importance on the requirement to validate the function of lncRNAs in disease‐relevant human cells and tissues 8. For this reason, our review focuses on lncRNA data generated from analysis of human cells and tissues.
Inflammation is a central hallmark of RA joint disease, and evidence for inflammatory pathway dysregulation and cell dysfunction has been observed in almost all innate and adaptive immune cells. In addition, despite OA historically being seen as a “wear and tear” disease of cartilage, there is now substantial evidence that inflammation is also a key contributor to OA joint pathology 9. OA synovial inflammation (synovitis) has been shown histologically as well as by ultrasonography and magnetic resonance imaging 10, 11, and there is evidence for increased cellular infiltration of activated B cells and T lymphocytes and elevated levels of proinflammatory cytokines in synovial fluid 12. Furthermore, ex vivo stimulation of cartilage tissue with proinflammatory cytokines mimics several pathologic features of the OA joint 13, and inhibitors of inflammatory pathways have modified OA disease in preclinical models 14.
As described below, we examined current evidence for the role of lncRNAs in regulating canonical inflammatory pathways that are relevant to inflammatory joint pathology. Of the major signaling pathways that are known to govern joint inflammation, 4 pathways were highlighted as having lncRNAs associated with their regulation, namely NF‐κB signaling, the p38 MAPK pathway, the arachidonic acid pathway, and Toll‐like receptor (TLR) signaling.
NF‐κB is considered to be a master regulator of inflammation, and its activity has been shown to regulate the production of several proinflammatory cytokines implicated in OA and RA joint pathology, including interleukin‐1β (IL‐1β), IL‐6, IL‐17, and tumor necrosis factor (TNF) 12, 15. Therefore, inhibition of NF‐κB signaling has long been considered to be an attractive pathway for developing an effective therapy for chronic inflammatory disorders 15. However, therapeutic targeting of NF‐κB presents a challenge, because NF‐κB may also have an antiinflammatory role and be important for the resolution of inflammation 16. Therefore, it is important to identify new approaches to appropriately target the proinflammatory activity of NF‐κB within the joint 17, so that systemic effects are avoided and patients are not immunocompromised.
Thus far, several lncRNAs have been identified and implicated in targeting different components of the NF‐κB signaling pathway (Figure (Figure3A).3A). Of most immediate interest with regard to inflammatory joint disease is lincRNA‐p21, whose ability to inhibit NF‐κB signaling by sequestering RelA (the p65 subunit of NF‐κB) in T cells has been demonstrated following methotrexate treatment in patients with RA. The expression of LincRNA‐p21 was shown to be up‐regulated by methotrexate via a DNA‐dependent protein kinase catalytic subunit–dependent mechanism 18.
Another lncRNA that has been implicated as regulating NF‐κB activity is the pseudogene Lethe, which has been shown to be expressed in fibroblasts. Functional studies have demonstrated that Lethe interacts with and blocks the DNA binding of RelA (p65). It is believed that Lethe acts as part of a negative feedback loop that acts to regulate the inflammatory response, whereby following NF‐κB activation by proinflammatory cytokines, up‐regulation of Lethe inhibits NF‐κB signaling. Of interest, the tissue expression of Lethe has been shown to be down‐regulated in aged mice 19. If the same is true in human aging, the reduction in the expression of this potential brake on NF‐κB activity could contribute to age‐associated inflammatory joint disorders such as RA and OA. Finally, studies conducted in human inflammation‐stimulated breast epithelial cells have shown that the lncRNA NKILA (NF‐κB–interacting lncRNA) binds to and blocks phosphorylation sites on IκB, thereby inhibiting IKK‐induced IκB phosphorylation and NF‐κB activation 20. Thus far, there are no studies that demonstrated the functional role of lincRNA‐p21, Lethe, or NKILA in RA or OA joint tissues or cells.
Arachidonic acid metabolites (prostaglandins and leukotrienes) are known mediators of the inflammatory response 21, and several have been shown to be differentially expressed in the joints of patients with either OA or RA 22. Therefore, targeted inhibition of the arachidonic acid pathway has long been considered to be an important area of research for developing a therapeutic agent that combats joint inflammation and pain. Indeed, inhibition of the prostaglandin synthase (cyclooxygenase [COX]) enzymes is the basis for the mode of action of nonsteroidal antiinflammatory drugs (NSAIDs) such as ibuprofen 23 and selective PTGS‐2 (prostaglandin‐endoperoxide synthase 2)/COX‐2 inhibitors such as valdecoxib 24, which have shown efficacy in reducing joint inflammation in several clinical trials.
Of importance, therefore, was the recent finding that the expression of PTGS‐2 was positively regulated by the lncRNA p50‐associated COX‐2–extragenic RNA (PACER) 25 (Figure (Figure3B).3B). PACER was shown to bind to and remove the repressive action of the p50 homodimer (of NF‐κB) at the COX‐2 promoter, leading to activation of COX‐2 gene expression. Furthermore, we recently observed that the expression of PACER was associated with the IL‐1β–mediated inflammatory response in primary human OA chondrocytes and was differentially expressed in human OA cartilage compared with non‐OA cartilage 26.
Unfortunately, long‐term administration of nonselective NSAIDs is associated with gastrointestinal toxicity 27, while selective COX‐2 inhibitors increase the risk of cardiovascular adverse events 28, attributed to a resulting imbalance between thromboxane T and prostacyclin levels. Therefore, identifying new regulators of prostaglandin production such as PACER is important, because it may lead to new targeted approaches that are capable of selectively inhibiting the production of inflammation‐associated arachidonic acid metabolites without any resulting toxicity.
Several studies have provided evidence that the p38 MAPK pathway is central to mediating both the production of proinflammatory cytokines and their signal transduction in the joint 13. Furthermore, p38 small molecule inhibitors have shown efficacy in both reducing cartilage degeneration in OA preclinical models 29, 30 and reducing inflammatory pain in RA models 14, 31, although clinical trials of p38 inhibitors in patients with RA have thus far failed to meet clinical end points.
With regard to lncRNA regulation of the p38 pathway, recent studies suggest the involvement of MALAT1 (metastasis‐associated lung adenocarcinoma transcript 1) 32 (Figure (Figure3C).3C). MALAT1 is one of the most abundantly expressed lncRNAs, with high levels of conservation across multiple tissue types. It has previously been well‐characterized in cancer, in which it is a transcription regulator implicated in cell cycle control and metastasis. However, Liu and colleagues have now highlighted MALAT1 cross‐talk with p38 MAPK as being pathogenic in diabetes mellitus. MALAT1 is believed to activate p38 MAPK, because small interfering RNA (siRNA)–mediated knockdown of MALAT1 reduced p38 phosphorylation 32. Furthermore, MALAT1 expression has been shown to be implicated in inflammation, again in diabetes mellitus under hyperglycemic conditions; in this setting, it was shown to contribute to increased IL‐6 expression and reactive oxygen species generation 33.
Innate immune cells such as macrophages and neutrophils are present in the inflamed joints of patients with RA and those with OA, and it is known that TLR signaling plays a key role in the recruitment and activation of these cells. Therefore, understanding the regulation of cytokine and chemokine signaling by these cells will provide greater clues regarding the onset and maintenance of chronic inflammation within the joint.
THRIL (TNF and heterogeneous nuclear RNP L [hnRNPL]–related immunoregulatory lincRNA) was recently identified in macrophages as an lncRNA that could form a complex with hnRNPL within the nucleoplasm (Figure (Figure3D).3D). HnRNPs are known to be important for the processing, function, and stabilization of mRNAs. Critically, it was demonstrated that the THRIL–hnRNPL complex binds to the promotor of TNF and induces its expression following TLR‐2 activation 34.
Carpenter and colleagues 35 identified an lncRNA (lincRNA–COX‐2) that is located in close proximity to the PTGS‐2 gene but, unlike the aforementioned PACER, this lncRNA had no functional interaction with PTGS‐2. Instead, lncRNA–COX‐2 was shown to be up‐regulated following activation of the TLR‐2 pathway (Figure (Figure3D).3D). Furthermore, following knockdown of lincRNA–COX2 and under the challenge of a synthetic TLR‐2 antagonist, expression of IL‐6 became undetectable, while CCL5 expression increased. Similarly to THRIL, lincRNA–COX‐2 forms complexes with hnRNPs, specifically hnRNP A/B and hnRNP A2/B1, and the formation of such complexes has been shown to modulate target gene expression.
Activation of TLR‐3 has been shown to induce an lncRNA referred to as NEAT1 (nuclear‐enriched abundant transcript 1). NEAT1 was shown to bind to a repressor of IL‐8 transcription known as SFPQ (splicing factor proline/glutamine‐rich) and relocate it to paraspeckle bodies (Figure (Figure3D),3D), resulting in IL‐8 transcriptional activation 36. Ilott and colleagues 37 demonstrated that TLR‐4 activation in monocytes up‐regulated expression of an lncRNA known as IL‐1β–RBT46 (IL‐1β regions of bidirectional transcription 46). IL‐1β–RBT46 shares a promoter with the IL‐1β–coding gene, and following pathway activation, expression of both the lincRNA and the IL‐1β gene is induced (Figure (Figure3D).3D). Locked nucleic acid (LNA)–mediated knockdown of IL‐1β–RBT46 led to a decrease in IL‐1β mRNA expression, suggesting that the lincRNA promotes expression of the gene. A further consequence of IL‐1β–RBT46 knockdown was a reduction in the expression of IL‐8, a well‐defined neutrophil chemoattractant.
The lncRNAs described above represent a pool of lncRNAs that have not yet been associated with inflammatory joint disease. However, given their prevalence in the regulation of central pathways of inflammation, studies determining their expression in joint tissue, and their relationship to OA and/or RA joint pathology would appear to be important.
The significance of genetic variations in the sequences of lncRNAs is currently not well understood and to date has been little studied. There is evidence that large alterations such as chromosomal rearrangements can affect lncRNA expression 38. However, as illustrated by the functional studies highlighted in this review, lncRNA function may be determined largely by its secondary structure, through binding to proteins and RNA. Therefore, smaller mutations and genetic variations such as single‐nucleotide polymorphisms (SNPs) may have little effect on lncRNA function. It is conceivable that SNPs could alter lncRNA function by altering the lncRNA secondary structure or causing alternative splicing of the lncRNA. Indeed, there is evidence that SNPs in the chromosomal region that harbor the lncRNA ANRIL are associated with susceptibility to cardiovascular disorders 39. However, it is not currently known whether these SNPs have an impact on the functional role of the ANRIL lncRNA or simply affect its expression. Predominantly, therefore, studies have thus far linked lncRNA to disease by expression analysis.
Transcriptional analyses by both NGS and microarray have identified several lncRNAs that are differentially expressed in inflamed joint tissue. Thus far, studies using OA patient samples have focused on cartilage tissue, while in RA disease other perijoint tissues such as synovium and subchondral bone have been interrogated.
One lncRNA that has been shown to be differentially expressed in both OA and RA joint tissue is HOTAIR (homeobox antisense intergenic RNA). Located on chromosome 12, HOTAIR was the first lncRNA to be characterized and has been shown to interact with polycomb repressive complex 2, which regulates the chromatin state, and the histone demethylase, lysine‐specific demethylase 1. As such, it regulates epigenetic changes within its target transcripts 40.
In OA cartilage, compared with non‐OA cartilage, HOTAIR was shown to be up‐regulated 41, although as yet its function in cartilage tissue or in isolated chondrocytes has not been reported. Song et al 42 have also identified differential expression of HOTAIR in RA perijoint tissues, namely synovial fibroblasts, osteoclasts, and bone mononuclear cells. Of note, Song et al found that the function of HOTAIR differed depending on the RA cell studied. For example, in RA synovial fibroblasts and osteoclasts, HOTAIR expression was lower than in controls, and its overexpression decreased the expression of matrix metalloproteinase 2 (MMP‐2) and MMP‐13. In contrast, HOTAIR was up‐regulated in RA bone mononuclear cells, where it functioned to facilitate macrophage activation and migration toward the joint.
Another lncRNA for which differential expression has been reported in inflamed joint tissue is H19. H19 lncRNA has been demonstrated to be highly expressed in RA synovial tissue 43 and in OA cartilage 44. Previous studies have demonstrated that H19 is a developmental reservoir of the 2 miRNA‐675 (miR‐675) family members, miR‐675‐p3 and miR‐675‐p5, which have been shown to regulate type II collagen expression and mediate cellular development in multiple tissues 45, 46, 47. Of interest, numerous studies have demonstrated that H19 controls the expression of several genes that are part of the imprinted gene network, a series of parentally inherited genes the expression of which is epigenetically determined within the sperm or oocyte. Such “imprinting disorders” have been linked to a number of human pathologies. Currently, inflammatory disease is not known to be affected by imprinting disorders, but given the inheritability of RA and other inflammatory diseases, studies that investigate the role of H19 in determining epigenetic susceptibility to disease could be of interest.
As previously mentioned, lncRNA analysis in OA has focused primarily on cartilage tissue, and the lncRNA GAS‐5 (growth arrest–specific 5) 48 and maternally expressed gene 3 (49) have been reported to be down‐regulated in OA‐diseased cartilage. Fu and colleagues 50 performed microarray analysis and identified 4,714 lncRNAs that were differentially expressed in OA cartilage compared with non‐OA cartilage. Although in the current study we did not perform any modulation of lncRNA expression to determine lncRNA function, expression correlation patterns were used to predict candidate lncRNA‐regulated genes. This analysis highlighted lncRNA uc.343, which was up‐regulated in OA cartilage, as regulating in cis the expression of homeobox C8 (50), and that many of the differentially expressed lncRNAs worked in concert with the transcription factor Sp1 to modulate expression of trans target genes.
In a separate study, microarray analysis was performed to identify a lncRNA with a specific role in the degradation of the cartilage matrix, named cartilage injury–related lncRNA (lncRNA‐CIR). LncRNA‐CIR was shown to be up‐regulated in OA cartilage compared with non‐OA cartilage, and siRNA‐mediated knockdown of its expression resulted in increased expression of collagen and aggrecan accompanied by reduced expression of MMP‐13 and ADAMTS‐5 (51).
Furthermore, we recently identified lncRNAs associated with the IL‐1β–mediated inflammatory response in primary human OA chondrocytes that were also differentially expressed in diseased OA cartilage. Of particular note were human immunodeficiency virus type I enhancer binding protein 2S (Hive2pAS) and IL‐7AS, 2 intergenic lncRNAs proximal to the Hivep2 and IL‐7 protein‐coding genes, respectively 26. Both Hive2pAS and IL‐7 AS were observed to be up‐regulated in both knee OA and hip OA cartilage compared with non‐OA control cartilage 26. Furthermore, locked nucleic acid (LNA)–mediated knockdown of Hive2pAS and IL‐7AS gene expression in human chondrocytes resulted in an increase in the production of proinflammatory cytokines, including IL‐6 (26), suggesting that these lncRNAs may function to control aberrant joint inflammation.
LncRNA research in inflammatory joint disease is very much a nascent field. However, several lncRNAs have now been identified as being either differentially expressed in diseased joint tissue or as candidate central regulators of inflammatory pathways relevant to joint pathology. Therefore, future studies to determine the functional role and mode of action of these disease‐associated lncRNAs will be insightful, as will joint tissue expression profiling of lncRNAs for which functional roles within key inflammatory pathways have been determined. To this end, it will be important to conduct further transcriptomic analysis of joint tissues (including subchondral bone, cartilage, and synovium) in which diseased joint tissues are compared with age‐matched noninflammatory control joint tissue and to conduct functional mode‐of‐action studies in disease‐relevant cells.
It should be noted that, currently, expression analysis of lncRNAs in the joint tissue of patients with other inflammatory joint diseases including gout, psoriatic arthritis, and juvenile idiopathic arthritis has not yet been reported. Clearly, such studies could also prove to be informative in building a broader pool of inflammatory joint disease–associated lncRNAs. Indeed, because inflammatory joint diseases share several common pathways, such studies are likely to identify lncRNAs that are dysregulated across multiple inflammatory joint disorders. Such findings will add to the recent paradigm shift in our understanding of the dysregulation in chronic disease, whereby the disease is no longer classified solely by its gross pathology but by the involvement of a particular dysregulated pathway.
It is currently too early to determine whether disease‐modifying therapeutic agents could be developed that are designed to directly target and modulate lncRNA function. Recent progress has been achieved using siRNA‐mediated RNA interference–based therapeutics, whereby targeted knockdown of TNF 52 or NF‐ĸB 53 has modified disease progression in animal models of RA and OA. Therefore, it is conceivable that lncRNAs could provide a new class of targets for RNA interference–based therapeutics. Regardless, ultimately determining the mode of action of lncRNAs that regulate inflammatory pathways and determining their relationship to joint pathology will provide a better understanding of how inflammation is epigenetically regulated within the joint. This improved understanding will likely be important in aiding the identification of new targets for therapeutic intervention as well as in identifying at‐risk patient populations.
Drs. Pearson and Jones were involved in drafting the article or revising it critically for important intellectual content, and both authors approved the final version to be published.