Rheumatoid arthritis (RA) is a systemic autoimmune and inflammatory disease, characterised by recurrent episodes of inflammatory arthritis, which are often symmetrical, involving peripheral joints with the hands, feet and wrists being the most commonly affected. Active disease in RA implicates both innate and adaptive elements of the immune system with persistent inflammation of synovial tissue and extra-articular organ involvement, more frequently seen in patients with severe disease. Regulatory mechanisms normally used by the body to contain self-directed immune responses may be insufficient to regulate the production of self-reactive autoantibodies like rheumatoid factor (RF) and anti–citrullinated protein/peptide antibodies (ACPA) also known as anti-cyclic citrullinated peptide antibodies (anti-CCP) [1, 2]. This inappropriate activation of B cells and dysregulated immune responses may eventually result in marked deformities of bone and cartilage, due to extensive erosive damage, mainly in the smaller joints of the hands and feet, with associated systematic symptoms, such as fatigue, general malaise and musculoskeletal pain. The opportunity to target these pathways has provided patients with therapeutic hope. However, despite the increasing knowledge in treating patients with RA, disappointingly, there are still patients who do not have favourable responses to existing therapies. In this review, we explore the pathogenesis of RA, and discuss the roles of the main signalling pathways, including metabolic reprogramming, that have been implicated in RA disease progression in order to develop a conceptual framework for more precise deployment of existing therapies, and also to provide a rationale for producing molecular inhibitors of these pathways. Improved knowledge of the many intracellular signalling pathways in RA will complement current precision medicine strategies, particularly for patients with difficult-to-treat RA, and especially for those with multidrug resistance disease.

Different databases, including Pubmed, Web of Science, SCOPUS and Google Scholar, were used for the keywords of ‘Rheumatoid arthritis, Intracellular signalling pathways, endoplasmic reticulum (ER) stress, Treatment response, Multidrug resistance, Autoimmune/autoimmunity, Inflammation, Refractory RA, Metabolic reprogramming/metabolites’. Inclusion criteria included all of the following keywords; RA pathogenesis, inflammation, immune tolerance, janus kinase/signal transducer and activator of transcription (JAK/STAT) pathway, stress-activated protein kinases/mitogen-activated protein kinases (SAPK/MAPK) pathway, PI-3K/AKT/mTOR pathway, Interferon signalling pathway, ER stress and the unfolded protein responses (UPR) pathway, vascular endothelial growth factor (VEGF) signalling pathway, spleen tyrosine kinase (SYK) signalling pathway, Wingless/Integrated (Wnt) signalling pathway, Notch signalling pathway, metabolites and pathogenesis, response to treatment and drug resistance. Exclusion criteria were the studies that included other rheumatological diseases apart from RA.

RA is a heterogenous condition and ACPA/anti-CCP autoantibodies, which are a hallmark of RA, are present in approximately 70% of these patients, in most reported series [3]; the specificity of ACPA/anti-CCP antibodies for RA ranges between 87% to 98% which makes their presence an excellent diagnostic indicator/marker of this disease. Shared epitope positive HLA-DR alleles are associated with anti-citrullinated peptides and with genetic susceptibility to developing RA. However, the molecular basis for these two associations is unknown.

The process of citrullination refers to post-translational conversion of arginine to citrulline residues by peptidylarginine deiminase enzymes (PADs) and is a key factor in the pathogenesis of RA [4]. The mechanisms controlling the citrullination process in RA are both dysregulated and upregulated. Under normal circumstances, the citrullinating enzymes are controlled by levels of calcium and redox conditions in order to limit their hyperactivation. Perforins and bacterial toxins induce prominent calcium influx which induces the hyperactivation of PADs and hypercitrullination in the RA joint. In addition, the extracellular environment of synovial joints in RA is in a state of oxidative activation [5], which is known to inactivate PADs: efficient extracellular citrullination in RA is likely to require the constant release of active enzymes from dying cells, which may be accelerated in the presence of PAD-activating autoantibodies.

PADs are inactivated extracellularly, and the presence of PAD activating co-factors, such as anti-PAD activating antibodies, and the continuous release of active PADs enzymes from dying and activated cells are necessary for efficient extracellular citrullination in RA. Therefore, targeting the signalling pathways that lead to PAD enzyme hyperactivation has the potential to prevent the development and progression of the RA citrullinome, with consequent clinical improvement in patients with RA [6].

An early study has suggested that one citrullinated peptide from vimentin, Vim R70Cit, could bind shared epitope positive with greater affinity than shared epitope negative HLA-DR molecules thus presenting citrullinated peptides to T cells that help production of IgG ACPA [4]. However, a thorough investigation of the binding of 180 peptides, either citrullinated or native, encompassing the entire alpha and beta chains of human fibrinogen did not find that citrullinated peptides bound shared epitope positive HLA-DR molecules better than their native counterparts. Besides, half of the patients with RA have IgG autoantibodies to the peptidyl arginyl deiminase 4 (PAD4) enzyme. This suggests the presence, in these patients, of helper T cells recognizing peptides from PAD4, which could provide help to B cells specific for citrullinated epitopes present on proteins being citrullinated by PAD4, by a hapten-carrier mechanism [7].

The main cause(s) of RA is unknown; however, some genetic and environmental factors are definitely involved in initiating disease onset [8]. Studies on patients with RA have revealed genetic factors that underlie the disease, such as the human leukocyte antigen (HLA)–DRB1 locus, where some alleles contain a five amino acid motif sequence (QKRAA), in residues 70–74 in the HLA-DRB1 region, named the shared epitope. Other factors include selection of certain predisposing T-cell repertoires, and antigen presentation; alteration in peptide affinity may also lead to autoreactive immune responses in RA [9]. The prognosis in patients with RA who are seropositive for ACPA and/or RF, is worse than those with ACPA-negative disease [10]. Neuro-immunological factors also influence disease development [8].

The environmental risk factors which are reported in patients with RA include smoking, exposure to silica, periodontitis, and gut microbiota. Another risk factor in RA is gender, as the disease is three times more common in women than in men [9].

The genetic and environmental risk factors as well as epigenetic elements lead to post-transcriptional modifications and alteration in self-protein citrullination. Anti-citrullinated peptides such as $\alpha$-enolase, keratin, fibrinogen, fibronectin, collagen, and vimentin are recognised as foreign antigen in the sera of patients [11]. Infectious agents and the gut microbiome are also associated with progression of the disease [12]. In patients with RA, dysregulation of the gut microbiota can affect the host immune responses which leads to disease development. The gut microbiota also can be used for prediction of the susceptibility of patients to RA and may be considered as have potential targets to control the incidence of RA. The enzymes and products of gut microbiota also affect the toxicity or efficacy of drugs, either directly or indirectly. On the other hand, different medications can modulate the gut microbiota composition; the administration of etanercept, which contains the TNF receptor, may restore the beneficial microbiota [13].

Variations in specific microbiota are proven to be associated with both disease development and response to drugs. In this regard, specifically targeting the gut microbiota in each individual patient may alleviate RA disease activity, increase drug efficacy and improve the outcome [14]. The different gut microbiota composition in patients with RA, compared to healthy controls, shows the importance of gut dysbiosis in the onset of RA. One of the most important environmental factors that modulate the gut composition is nutritional content of the diet [15].

To prevent and manage the degree of inflammation in RA, modulation of the gut microbiota through diet is recommended. For instance, the Mediterranean diet has beneficial effects on the gut microbiota and may improve the outcome in RA [16]. The use of Lactobacillus acidophilus and Lactobacillus casei in probiotic products are also helpful in reducing disease activity while bacillus spp. and other lactobacillus spp. are increased in quantity in the gut of patients with RA. It seems that different lactobacilli have different roles in RA pathogenicity and diseases activity [14]. In addition, faecal microbiota transplantation (FMT) from healthy individuals to patients with RA may also reduce disease activity [17]. Furthermore, the content of gut microbiota could potentially be considered as biomarkers for RA diagnosis, especially in early onset disease. The abundance of beneficial intestinal bacteria, such as Bacteroidetes, is decreased in early RA while Prevotella copri are increased and take the place of other beneficial bacteria [18]. Not only gut microbiota but also the oral microbiota are important in RA development. Dysbiosis of oral microbiota may lead to periodontitis and the inflamed gingiva contains citrullinated proteins and ACPA. Porphyromonas gingivalis, a common microorganism in gingiva, can citrullinate the proteins for ACPA development, and is involved in the onset of RA disease [19].

A normal synovium contains mesenchymal-derived fibroblast-like synovial cells (FLS) and macrophages. In RA the FLS acquire specific characteristics, including increased proliferation, apoptosis resistance, as well as secretion of high-level disease-associated cytokines and chemokines, adhesion molecules, matrix metalloproteinases (MMPs), and tissue inhibitors. FLS are directly involved in cartilage destruction and chronic inflammation [20].

Considerable insights into the pathogenesis of RA may be gained from study of the mechanisms of action of drugs shown to be effective in RA. Methotrexate (MTX) is a conventional synthetic disease modifying antirheumatic drug (csDMARD) that is often used as a first line treatment for RA. MTX is a folic acid antagonist that inhibits the synthesis of both DNA and RNA and exerts anti-inflammatory effects on many pathways involved in RA. Furthermore, MTX works synergistically with different anti-TNF therapies, independently of its property of reducing the titres of anti-drug antibodies [21].

The balance between pro- and anti-inflammatory cytokine levels is disturbed in rheumatoid joints, leading to the process of autoimmunity, chronic inflammation, and associated joint damage. Inflammation of the synovium begins with infiltration, accumulation, and migration of leukocytes into synovial micro-vessels, which induce the expression of adhesion molecules such as integrins, selectins, in addition to members of the immunoglobulin superfamily and chemokines. These adhesion molecules and chemokines facilitate leukocyte rolling, adhesion, and transmigration into the synovial tissue. Within the synovium, leukocytes release proinflammatory cytokines, such as TNF, IL-1$\beta$, and IL-6, which further activate synoviocytes to perpetuate the inflammatory response with production of more cytokines, chemokines, and matrix-degrading enzymes, all leading to joint destruction and cartilage damage. This persistent inflammation and joint damage may ultimately result in chronic joint disease as found in RA [22]. Self-directed immune responses induce autoantibodies to trigger an adaptive response with increased level of T-cells against citrullinated self-proteins [23]. T-cell differentiation towards a TH1/TH17 phenotype is linked to the presence of IL-15, IL-1, IL-6, TGF$\beta$, IL-12, and IL-23 in synovial tissue [24]. B-cells in the synovium also contribute to antigen presentation and cytokine secretion [25]. Both macrophage-derived and dendritic-cell–derived transforming growth factor $\beta$ as well as interleukin-1$\beta$, 6, 21, and 23 trigger Th17 differentiation and prevent the production of regulatory T cells (Treg) [26]. This imbalance between Th17 and Treg induces TNF, which further inhibits Treg activity [27]. Furthermore, the macrophage-derived cytokines including TNF, IL-1, IL-6, IL-15, and IL-18 also stimulate many proinflammatory cytokines in synovial tissue, where myeloid cells, plasmacytoid dendritic cells (pDCs) release cytokines such as IL-12, 15, 18, and 23 and HLA class II molecules, as well as costimulatory molecules which may cause further T-cell activation [23, 28].

Cytokine production is essential to the development of RA. The early stages of disease show cytokine expression patterns of IL-4, 13, and 15 TNF and IL-6 and during inflammation the other cytokines, chemokines and adhesion molecules are also activated. DCs also have an important role in the initiation and perpetuation of RA through presentation the arthritogenic antigens to T-cells. Two major DC subsets, myeloid DCs (mDCs) and pDCs, in RA synovial tissue are reported to have different cytokine profiles [29]. In RA synovial tissues, both pDCs and mDCs release IL-15 and IL-18, while pDCs express IFN $\alpha$/$\beta$ and mDCs secretes IL-12p70 and IL-23p19 [30]. Moreover, the rate of angiogenesis is increased [31], and IL-6 exerts systemic effects, such as the acute phase response, anaemia, cognitive impairment, and dysregulation of lipid metabolism. IL-1 family cytokines (including IL-1$\alpha$, 1$\beta$, 18, 33) promote the activation of leukocytes, endothelial cells, chondrocytes, and osteoclasts in RA [31]. Cytokines are also involved in osteoclast maturation and activation and nuclear factor-$\kappa$B ligand-receptor activator (RANKL) plays a central role in this process, along with TNF, IL-17, and IL-1 [22].

When loss of immune tolerance to self-antigens occurs in diseases such as RA, protective immune mechanisms switch to auto-aggressive immunity in dedifferentiated synoviocytes, with T cell-mediated joint damage and loss of tissue-protective macrophage populations. The Treg subpopulation of helper T-cells prevents the development of autoimmunity and maintains immune tolerance by using a variety of mechanisms, including elimination of self-reactive T cells and production of anti-inflammatory cytokines, such as IL-10 and TGF-$\beta$ [32].

In effect, Treg cells are divided into three categories according to their origin and differentiation: they are produced by immature T lymphocytes during thymus development, and with a phenotype is CD4+CD25+Foxp3+ T cells, which are called natural Treg (nTreg) cells that constitutively express CD25 and express the specific nuclear transcription factor, Foxp3. Upon peripheral antigen stimulation or immunosuppressive factor induction, mature CD4+CD25- T cells are transformed into acquired Treg (iTreg) cells, including Tr1 and Th3 subsets; the former mainly secretes IL-10 and TGF-$\beta$, while the latter mostly produces TGF-$\beta$. In addition to regulatory CD4+ T cells, regulatory CD8+ T cells also exist in the CD8+ T cell lineage.

In the thymus, nTreg cells are derived from hematopoietic progenitor cells and migrate into the peripheral blood to maintain immune tolerance and to prevent autoimmunity [33].

During the development of peripheral Treg cells, naïve CD4+ T cells first migrate into the peripheral blood without any TCR activation. Once stimulated by antigens in the peripheral blood, naïve CD4+ T cells can differentiate into Foxp3+ Treg cells in the presence of both TGF-beta (TGF-$\beta$) and IL-2. Despite their low frequency, peripheral Treg cells can also prevent inflammation developing in barrier tissues. When Treg cells are generated in vitro, they are termed iTreg cells. These cells are also suppressive in nature and help maintain immune homeostasis; it should be emphasised in this context that iTreg cells are less stable than nTreg cells.

There is an additional CD4+ T cell subset that mediates immunosuppression in vitro and which is characterised by secretion of TGF-$\beta$1 and IL-10, without expressing the Foxp3 transcription factor. This subset is termed type 1 regulatory T cells (Tr1 cells). However, no surface biomarkers or transcription factors have so far been identified for this specific subset, although some promising candidates have been reported. The suppressive mechanisms, used by Tr1 cells, are similar to those employed by nTreg cells and include the production of immunosuppressive cytokines, cell–cell contact-mediated suppression, cytotoxicity, and metabolic disruption. IL-10 and other cytokines, including IFN-$\alpha$, IL-6 and IL-27, are required for the generation of Tr1 cells. Importantly, Tr1 Treg cells are reported to be less suppressive than nTreg cells in the early stage of life as they fail to rescue IPEX patients, who have a complete lack of Foxp3. In addition, it has been been reported that Foxp3${}^{+}$ CD8${}^{+}$ T cells, CD4${}^{-}$CD8${}^{-}$ cells and gamma/delta T cells all share some suppressive properties, but no evidence has produced to show that they play a significant role in self-tolerance [34].

The mechanisms of self-tolerance are complicated, however, and dysfunction of Treg cells and dendritic cells (DCs) are among the proposed mechanisms underlying the breakdown of self-tolerance in the pathogenesis of RA. Furthermore, the ratio of Treg/Th17 and Th1/Th2 cells is vitally important and the imbalance between these specialised subpopulations of T cells induces the initiation of RA pathology [35]. The anti-TNF biologic, etanercept, in combination with MTX has been shown to ameliorate disease in RA by normalising the distribution of Th17 and Treg [36] cells. In a number of chronic autoimmune diseases, enhancement of Treg function by the use of tolerogenic dendritic cells or stem cell transplantation are both potentially effective treatments to re-establish immune tolerance [37, 38]. Moreover, tolerogenic vaccine platforms to deliver the self-antigens to specific antigen-presenting cell subtypes are other therapeutic strategies that may be employed for treat RA [22].

In autoimmunity, it has been postulated that citrullination of peptides and self-antigens may expose cryptic peptides by modifying protein structure and proteolytic cleavage, with subsequent changes to antigen processing by the MHC-II, which may, in turn, lead to loss of immune tolerance to self-antigens because of exposure of these cryptic peptides previously hidden from immune system [6].

Despite the wide-ranging targets and armamentarium of therapy available to treat RA, there are still patients with RA who unfortunately never quite achieve remission or even low disease activity states, the so-called difficult-to-treat RA [187]. These include drug resistant RA, recognised as refractory RA [188]. Drug resistance in RA is likely to be multifactorial, and can stem from several causes, such as the development of neutralising antibodies against biologics, alterations in drug metabolism, or shifts in the underlying disease mechanism [189]. There is also the recognition that the age of onset of RA can affect the response to treatment, with more drug resistance with increasing age of onset [190].

Personalised medicine and the ability to stratify and predict patients’ response to therapy may reduce the drug resistance in RA. Recently, it has been shown that an increase in dendritic cell precursors can predict the prognosis of RA [191]; such findings provide opportunities for novel targets for precision RA therapies. With improved understanding of the pathogenesis of RA comes advancement in the various facets of precision medicine in RA. Multidrug resistance is one of the most common reasons for refractory RA [192], but the ability to identify the right drug for a particular patient with a RA is somewhat lagging behind [193]. Real world registry studies may have a place in helping to close the gap in RA precision medicine, as we understand how patients respond to different therapies, particularly as patients with resistant RA are likely to have cycled through a number of different biologics and targeted synthetic DMARDs (tsDMARDS) [194].

Delineation of the intracellular signalling pathways activated in RA and development of novel targeted therapies are essential steps to improving patient outcome in this complex disease.

As befits the heterogeneity of drug resistant RA, there is a complex myriad of potential targets for therapy [195, 196]. Therefore, the knowledge gained in discovering the intracellular signalling pathways in RA will complement precision medicine strategies in providing difficult-to-treat RA, particularly those exhibiting drug resistance and may also offer the hope of achieving disease remission, or, at least, a less active disease state.

MR and MFM conceived the study and retrieved the literature; MPK, JL and ALT retrieved the literature. All the authors were involved in writing the paper, and all authors read and approved the final manuscript. All authors have participated sufficiently in the work and are accountable for all aspects of the work. All authors contributed to editorial changes in the manuscript.

Not applicable.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.