The demonstration in 1985 that low-dose, intermittent methotrexate is a potent and effective therapy for rheumatoid arthritis (RA) 1 led to a dramatic change in the way that patients with RA are treated. Indeed, methotrexate is no less efficacious than specific anti-tumor-necrosis-factor (anti-TNF) therapy for the relief of symptomatic joint inflammation in early RA, and the difference between methotrexate and etanercept with respect to protection from structural injury in RA is probably not biologically significant 2. Thus, methotrexate remains the cornerstone of therapy for RA, and understanding the mechanism(s) responsible for the therapeutic efficacy of this agent may lead to the development of new therapies.

Methotrexate was first developed in the 1940s as a specific antagonist of folic acid. This drug inhibits the proliferation of malignant cells, primarily by inhibiting the de novo synthesis of purines and pyrimidines. Because administration of high doses of reduced folic acid (folinic acid) or even folic acid itself can reverse the antiproliferative effects of methotrexate, it is clear that methotrexate does act as an antifolate agent. Interestingly, although not originally designed as such, methotrexate appears to be a 'pro-drug', i.e. a compound that is converted to the active agent after uptake. Methotrexate is taken up by cells via the reduced folate carrier and then is converted within the cells to polyglutamates 3. Methotrexate polyglutamates are long-lived metabolites that retain some of the antifolate activities of the parent compound, although the potency for inhibition of various folate-dependent enzymes is shifted 3456.

Low-dose methotrexate was introduced for the treatment of RA because of its presumed antiproliferative properties, although it was unclear how inhibiting proliferation of the lymphocytes thought to be responsible for synovial inflammation in RA for one day a week might lead to effective suppression of disease activity. However, it soon became clear that inhibition of folic acid metabolism could not be completely responsible for the anti-inflammatory effect of methotrexate. During the past 15 years, it has become clear that administration of folic acid in doses of 1–5 mg per day helps to prevent much of the toxicity of methotrexate without interfering with the anti-inflammatory efficacy of the drug, whereas very high doses of folinic acid also prevent methotrexate toxicity but may interfere with its efficacy 7891011121314151617181920. There are two potential explanations for the capacity of high doses of folinic acid to reverse the therapeutic effects: first, folinic acid may bypass the effects of methotrexate on reduction of folic acid and thereby bypass the therapeutic effects of the drug; alternatively, folinic acid but not folic acid may compete with methotrexate for a single transport site into the cell (Fig. 1) and may thus interfere with cellular uptake of methotrexate 21. Moreover, the expected inhibition of cellular proliferation is manifested as bone marrow suppression, and oral and gastrointestinal ulcers, and may require lowering the dose of the drug and, usually, the efficacy of the therapy, suggesting that inhibition of cellular proliferation alone is not responsible for the anti-inflammatory effects of methotrexate. Thus, folate antagonism appears to play, at most, a minimal role in the anti-inflammatory mechanism of methotrexate.

Another potential mechanism by which methotrexate may diminish inflammation in the joint is by diminishing cytokine production. Numerous studies have demonstrated diminished levels of inflammatory cytokines in the serum of patients. The adenosine A_2A receptor agonist CGS-21680 is a potent inhibitor of neutrophil leukotriene synthesis in vitro, and, similarly, methotrexate therapy leads to diminished production of leukotriene B₄ by neutrophils stimulated ex vivo2223. The mechanism by which methotrexate diminishes these cytokine levels remains unexplained and it is difficult to determine from these studies whether the effects of methotrexate therapy on production of inflammatory mediators results in diminished inflammation or is secondary to other anti-inflammatory events.

Similarly, methotrexate-mediated effects on T-cell function, either in vivo or in vitro, have been demonstrated. Indeed, Genestier and colleagues have reported that methotrexate diminishes antigen-stimulated T-cell proliferation both in vitro and in T cells taken from patients taking methotrexate 24. That the effects of methotrexate on T-cell function are completely reversed by folic acid and that the effects of therapy on T cells studied ex vivo are present for only 48 hours a week would strongly suggest that this cannot be responsible for the bulk of the anti-inflammatory effects of the drug.

A third proposed mechanism of action is based upon the observation that polyamines accumulate in the synovium of patients with RA and that metabolism of these polyamines by macrophages leads to the production of toxic oxygen products that diminish stimulated T-cell function 252627. Indeed, methotrexate therapy does diminish polyamine levels in the joints of patients with RA 282930, but this effect, like that of methotrexate on T-cell proliferation, is reversed by folic acid. Moreover, there are more than enough toxic oxygen metabolites being generated in the rheumatoid synovium to mediate the tissue damage present in this disease; another source of toxic agents would add relatively little.

Our laboratory originally proposed the hypothesis that the beneficial effects of methotrexate result from the intracellular accumulation of intermediates in purine biosynthesis that, by a mechanism that has not been completely worked out, leads to increased concentrations of adenosine in the extracellular space 31. This hypothesis sprang from the prior demonstration that intracellular accumulation of specific intermediates in the de novo synthesis of purines leads to adenosine release 32 and from our interest in the anti-inflammatory effects of adenosine, which are mediated by specific receptors on inflammatory cells. Prior work had demonstrated that methotrexate polyglutamates inhibit the enzyme aminoimidazolecarboxamidoadenosineribonucleotide (AICAR) transformylase more potently than the other enzymes involved in purine biosynthesis 4533. This inhibition occurred at pharmacologically relevant concentrations of methotrexate and might be expected to occur more readily with infrequent loading with methotrexate, since methotrexate polyglutamates are long-lived metabolites (persisting for weeks). The presence of increased concentrations of AICAR metabolites in the urine of RA patients treated with methotrexate supports these findings 3435. The accumulation of AICAR and its metabolites has a direct inhibitory effect on at least two key enzymes, adenosine deaminase and AMP deaminase, with the end result of increased concentrations of adenosine and adenine nucleotides intracellularly 4. Methotrexate in doses similar to that used in the treatment of RA has been known to cause the accumulation of AICAR in animal models of RA, and this accumulation is associated with an elevation in adenosine concentration in the extracellular space 3236. The exact mechanisms by which the elevation of extracellular adenosine arises are not fully understood, but dephosphorylation of adenine nucleotides is likely to be a major contributor, partly because of the ubiquitous nature of ATP in tissues and partly because of the widespread existence of ecto-5'-nucleotidase, an enzyme that catalyzes the dephosphorylation of AMP to adenosine 37.

All this evidence points to adenosine as a key mediator in the anti-inflammatory actions of methotrexate. In vivo experiments support this contention. The nonselective adenosine receptor antagonist 8-phenyl theophylline potentiated inflammatory responses in a hamster-cheek-pouch model 38. Infusion of adenosine directly into the knee in rats inhibited the development of adjuvant-induced arthritis, and an adenosine receptor antagonist effectively reduced the severity of joint inflammation in a collagen-induced arthritis model in mice 3940. We have previously shown that the anti-inflammatory effects of methotrexate in carrageenan induced mouse air pouch inflammation is reversed by an antagonist to the adenosine A_2A receptor, or by the addition of adenosine deaminase, an adenosine-metabolizing enzyme, suggesting that adenosine is indeed responsible for the anti-inflammatory effects of methotrexate in vivo36. An interesting study by Silke et al. showed that ingestion of caffeine, a nonselective antagonist of adenosine receptors, in coffee correlates with poor clinical response to methotrexate, and patients with a high caffeine intake are more likely to discontinue methotrexate than those with a low caffeine intake 41.

To better appreciate how adenosine influences biological responses in the network of events taking place in an inflammatory milieu, something must be said about this autocoid and the cellular receptors with which it interacts to produce these physiological responses. Adenosine receptors, or P1 receptors, fall into four known subclasses: A₁, A_2A, A_2B, and A₃. These are members of the large, seven-transmembrane-receptor family of receptors that influence cell signaling mechanisms by coupling to G proteins. The receptor sequences have been characterized and, with the exception of the A₃ receptor, they are highly conserved during evolution. Adenosine receptors modulate a vast array of physiological functions, from heart rate to the state of wakefulness. Adenosine, acting on P1 receptors, exerts a number of actions on a variety of cell types relevant to the anti-inflammatory effect of methotrexate.

Rheumatoid factor, or autoantibodies directed against the Fc portion of IgG, is a hallmark of RA, although its exact role in the pathogenesis of the disease has been debated. The effect of methotrexate on the levels of circulating IgM rheumatoid factors has also been controversial. While some workers have reported no suppression of serum rheumatoid factor levels with methotrexate treatment, Alarcon et al. observed significant drops in the levels of both IgM and IgA rheumatoid factors in methotrexate-treated patients, and particularly of the concentration of IgM rheumatoid factor in those who showed clinical improvement 75. These findings were confirmed by other groups in studies done both in vivo and ex vivo7677787980, although it is unclear whether this is a primary or secondary effect of adenosine.

T lymphocytes have received much attention in relation to the pathogenesis of RA and opinions differ in their contribution to the causation of the disease. The presence of these cells in the affected synovium and the strong ethnicity-dependent HLA-DR associations implicate T lymphocytes as key players in the disease process. One possible explanation of the beneficial actions of methotrexate in RA is the diminution of both the size and reactivity of the T-lymphocyte population. There are suggestions that this may be accomplished by the induction of apoptosis in activated T cells 24. This suggestion is consistent with the observations of reductions in peripheral blood T and B lymphocyte populations after short-term methotrexate treatment 81, and methotrexate induction of apoptosis in inflammatory cells may be relevant to its antirheumatic actions in vivo82. In contrast, significant increases in the CD3- and CD4-positive peripheral blood cells and enhancement of stimulated lymphocyte proliferation have been observed after long-term treatment with methotrexate 83, and adenosine, acting through and A_2A A_2B receptors, may play a role in T-cell deactivation 8485. Nonetheless, the role of these shifts in T-cell function and trafficking in the pathogenesis of RA is unclear.

Cytokines are messengers with major roles in inflammatory and immune responses and have been targets of interest in recent therapeutic developments in chronic arthritis, with TNF-α and IL-1 as the focus of interest 86. In animal models of chronic arthritis, methotrexate was thought to be useful in reducing the production of IL-1 8788. In support of these findings, clinical studies of RA patients receiving methotrexate treatment have documented reductions in monocytic IL-1 production but not serum concentrations of IL-1 89. Others have disputed this view and suggested that alterations in IL-1 responses were related to diminutions in the ability of cells to respond to IL-1 rather than to direct inhibition of its production, perhaps through dose-dependent ligand binding 909192.

Methotrexate is also known to suppress TNF activity by suppressing TNF-induced nuclear factor-κB activation in vitro, in part related to a reduction in the degradation and inactivation of an inhibitor of this factor, IκBα, and probably related to the release of adenosine 93. The generation of TNF-α by peripheral blood mononuclear cells is suppressed by an adenosine kinase inhibitor, by virtue of its ability to limit adenosine uptake and metabolism and thereby enhance extracellular adenosine concentration 94. TNF-α synthesis in T cells and macrophages is suppressed 95. In the murine collagen-induced arthritis model, in vivo intraperitoneal methotrexate treatment reduced TNF serum levels and diminished TNF production by splenic T cells and macrophages 96. Methotrexate suppresses the production of both TNF and IFN-γ by T-cell-receptor-primed T lymphocytes from both healthy human donors and RA patients 97. In early RA, in which the disease duration is less than 6 months, methotrexate treatment is associated with a significant decrease of TNF-α-positive CD4⁺ T cells, while the number of T cells expressing the anti-inflammatory cytokine IL-10 increased 98. Methotrexate is also known to suppress the IL-6-induced generation of reactive oxygen species in the synoviocytes of RA patients 99. Serum IL-6 levels have also declined after methotrexate treatment in RA patients in some studies 100. Constantin et al. reported that ex vivo treatment of peripheral blood monocytes with methotrexate increased expression of IL-4 and IL-10 while IL-2 and interferon-γ expression were decreased, suggesting that the immunoregulatory role of methotrexate is also targeted at adjusting the balance between Th1 proinflammatory and Th2 anti-inflammatory cytokines 101. Again, the molecular mechanism of these changes is unclear.

Our search for mechanisms governing the inflammatory response has uncovered many facets relevant to the pathogenesis of arthitic diseases. The success of methotrexate as an antirheumatic agent rests on its many actions that affect a wide variety of pathogenic mechanisms, many of which are mediated by the release of adenosine. The molecular mechanism for many of these phenomena is related to the enhanced release of adenosine into the extracellular space, where it can activate its receptors on relevant cell types. In this respect, methotrexate is an excellent example of how knowledge and continuing research in molecular biology and pharmacology can be employed in the refinement of existing medications originally used on an observational basis. Such understanding will form the basis for the development of new and more effective therapy for the treatment of rheumatic diseases.

AICAR = aminoimidazolecarboxamidoribonucleotide; Fc = crystallizable fragment (of antibody); IFN = interferon; IL = interleukin; RA = rheumatoid arthritis; Th = T helper (cells); TNF = tumor necrosis factor.