Rheumatoid arthritis is a multisystem autoimmune disorder [¹]. Inflammatory processes in rheumatoid arthritis lead to accelerated atherosclerosis and afflicted patients have an elevated risk for cardiovascular events leading to a higher mortality [¹]. In particular, elevated levels of immune complexes and cytokines (tumor necrosis factors, interleukins, and interferons (IFN)) contribute to the pathogenesis of both the articular [²] and cardiovascular manifestations of rheumatoid arthritis [³]. As such, the cardiovascular actions of antiarthritic medications are an important determinant for multiple aspects of patient care [³].

Although traditional nonsteroidal anti-inflammatory drugs (NSAIDs) [⁴], cyclooxygenase (COX)-2 inhibitors [⁵], methotrexate [⁶], and other disease-modifying antirheumatic drugs (DMARDs) are effective treatments for pain and inflammation, each of these drugs is linked with particular adverse effects which necessitate consideration. While NSAIDs are associated with gastrointestinal complications [⁷], some COX-2 inhibitors are associated with an elevated risk of cardiovascular events [⁵]. The COX-2 inhibitor cardiovascular risk has resulted in a market recall of several of these medications [⁸]. Although still associated with some adverse effects, methotrexate has proven to be one of the safest antirheumatic drugs [⁹].

In contrast to COX-2 inhibitors, methotrexate has demonstrated atheroprotective properties [¹⁰]. Recent studies have begun unravelling the molecular interactions underpinning the cardiovascular influence of COX-2 inhibition [¹¹, ¹²] and methotrexate [¹³]. While these drugs have multiple cellular effects, cardiovascular modulation in part relies upon the regulation of reverse cholesterol transport [¹³, ¹⁴]. This review aims to provide an overview of the mechanisms by which methotrexate interacts with cholesterol homeostasis to modulate atherogenesis in inflammatory disease.

Methotrexate is an anti-inflammatory medication commonly used in the treatment of rheumatoid arthritis and other inflammatory diseases such as psoriasis and inflammatory bowel disease [⁹]. The primary anti-inflammatory actions of methotrexate are attributable to adenosine release, triggered by methotrexate's polyglutamate metabolites [⁶].

Adenosine, a nucleoside produced by many cells and tissues in response to physical or metabolic stresses, is an endogenous anti-inflammatory mediator [¹⁵]. The physiological effects of adenosine are mediated by G-protein coupled 7-transmembrane receptors [¹⁶] that exist on almost all human cell types [¹⁵]. Adenosine receptors are divided into four classes, A₁, A_2A, A_2B, and A₃, based on the differential selectivity of adenosine analogues and molecular structure [¹⁵].

It has been noted that some of the anti-inflammatory effects of methotrexate may be reversed by inhibition of the A_2A receptor [¹⁷]. Further research suggests that stimulation of the A_2A receptor interferes with inflammatory processes such as the biosynthesis and release of proinflammatory cytokines, inhibits oxidative activity, prevents platelet aggregation, and reduces adhesion and degranulation of neutrophils [¹⁸]. Together these results demonstrate that the A_2A receptor plays a significant role in the mediation of inflammation [¹⁷, ¹⁸].

Although adenosine has anti-inflammatory properties principally through its A_2A [¹⁷] and A₃ [¹⁹] receptors, it can also act in contradictory fashions via alternate receptors [²⁰]. Ligation of adenosine to the A₁ receptor initiates inflammatory processes [²⁰]. The adenosine receptors influence inflammatory processes by modulating cAMP signalling cascades [²⁰]. In particular, the opposing effects of A_2A and A₁ activation are accounted for, respectively, by an elevation or reduction in intracellular cAMP [²⁰]. Consequently, the overall effect of adenosine on inflammatory processes depends upon the relative temporal and spatial distribution of the various adenosine receptors, as well as the influence of other effectors in the inflammatory milieu [²⁰].

Accumulating evidence indicates that the systemic inflammatory load in lupus disrupts cholesterol dynamics, increasing vulnerability to cholesterol accumulation in cells of the artery wall, including macrophages and endothelium [⁴³]. Patients with lupus [⁴⁴] and rheumatoid arthritis [¹] are known to have an increased risk for atheromatous cardiovascular disease. The immune-complex dysregulation seen in these conditions may play a role in the atherogenic process, a notion supported by evidence from cholesterol loading and efflux [⁴⁴, ⁴⁵].

In some inflammatory disease, antibodies to oxidized LDL are generated [⁴⁵]. The resulting formation of oxLDL-antibody complexes enhances uptake of oxLDL into macrophages and promotes foam cell formation [⁴⁴, ⁴⁵]. Furthermore, stimulation of cultured human monocytes and THP-1 human monocytoid cells with the immune reactants IFN-γ or immune complexes impede cellular cholesterol efflux by markedly decreasing 27-hydroxylase and ABCA1 [⁴⁶]. Collectively, these immune and inflammatory mediators promote atherosclerosis by disabling mechanisms that prevent the cells of the artery wall from being overloaded with cholesterol, leading to the formation of lipid laden foam cells [⁴⁶, ⁴⁷].

Research suggests that IFN-γ inhibits RCT by modifying signal transducer and activator of transcription (STAT) protein activity via two alternate mechanisms [⁴⁸, ⁴⁹]. In the first process, IFN-γ binds to the IFN receptor, initiating the dimerization of the two IFN-γ receptor sub-units which phosphorylates the bound janus kinases [⁴⁸, ⁵⁰]. The activated kinase phosphorylates the IFN-γ receptor, recruiting STAT proteins [⁵⁰]. These STAT proteins are tyrosine phosphorylated, dimerize and translocate to the nucleus, where they downregulate LXRα, which in turn decreases ABCA1 expression [⁴⁸].

In the second process, IFN-γ induces a calcium flux which activates the calcium/calmodulin-dependent protein kinase II (CaMKII) [⁴⁹]. CaMKII directs the phosphorylation of the STAT residue, Ser⁷²⁷ which maximizes STAT activity [⁵¹]. The ability of both immune complexes and IFN-γ to influence RCT and macrophage to foam cell conversion emphasizes the importance of cardiovascular considerations in the treatment of inflammatory autoimmune disorders [⁴⁷].

Across a multiplicity of reviewed studies, COX-2 inhibition elevated cardiovascular risk [⁵] while methotrexate significantly reduces cardiovascular disease mortality amongst rheumatoid arthritis patients [¹⁰], despite some disagreement for patients with prior atherosclerotic development [⁵²]. Given the fundamental role of ABCA1 and 27-hydroxylase in regulating foam cell formation, we postulated a potential role for RCT in the atherosclerotic effects following treatment with COX-2 inhibitors and methotrexate [¹³]. The application of COX-2 inhibitors to THP-1 monocytes cultures resulted in significant decreases in the gene expression of 27-hydroxylase and ABCA1 [¹⁴]. The resultant reduction in ABCA1 and 27-hydroxylase protein creates an environment where cholesterol efflux is compromised, promoting foam cell formation [¹⁴].

Significantly, the effects of COX-2 inhibition on cholesterol metabolism and efflux may be reversed by the addition of adenosine A_2A receptor agonists [¹¹]. Given the previously described ability of methotrexate to induce adenosine release [⁶], it is plausible that methotrexate is capable of upregulating RCT as its mechanism for atheroprotection.

Using RT-PCR and immunoblotting, the levels of expression of ABCA1 and 27-hydroxylase were evaluated in cells in the presence of COX-2 inhibitors or IFN-γ with or without methotrexate [¹³]. Methotrexate reversed foam cell formation and the down regulation of ABCA1 and 27-hydroxylase induced by COX-2 inhibition or IFN-γ [¹³]. However, this methotrexate-induced upregulation was prevented by an inhibition of adenosine A_2A receptors [¹³]. Thus, the capacity of methotrexate to reduce the onset of atherosclerosis may be partially attributed to the adenosine-driven upregulation of ABCA1 and 27-hydroxylase which expedites cholesterol efflux [¹³].

The specific mechanism by which adenosine A_2A receptor ligation modifies 27-hydroxylase and ABCA1 expression has not yet been elucidated [¹²]. However, the inhibition of PKA prevents adenosine A_2A stimulation of RCT proteins, demonstrating a dependency upon a cAMP-PKA-mediated process [⁴⁶]. Generally, A_2A receptor ligation activates adenylate cyclase, leading to the accumulation of cAMP [⁵³]. cAMP activates PKA which subsequently activates CREB proteins [⁴⁶]. These proteins then translocate into the nucleus, where it interacts with a consensus element in the gene promoter, modulating the expression of the target gene [⁴⁶] (Figure 1).

As previously discussed, 27-hydroxycholesterol upregulates ABCA1 expression [⁴⁰, ⁴¹]. A reasonable hypothesis is that ABCA1 upregulation is dependent upon the increased expression of 27-hydoxylase. However, inhibition of the 27-hydroxylase enzyme does not hinder the ability of adenosine A_2A ligation to upregulate ABCA1, suggesting that the methotrexate-induced increase in ABCA1 expression is not contingent on this feedback loop [⁴⁶].

In addition to stimulating increased ABCA1 expression, A_2A ligation augments ABCA1 activity [⁵⁴]. The generation of cAMP by A_2A stimulation activates protein kinases and Epac [⁵⁴], a guanine nucleotide exchange factor [⁵⁵]. Protein kinase generates phospho-ABCA1 and Epac activates ABCA1 function [⁵⁴]. Thus, methotrexate promotes cholesterol efflux and hinders atherogenesis by both upregulating and activating RCT proteins.

Beyond the A_2A receptor, adenosine may counteract the effects of IFN-γ via the A₃ receptor [⁴⁹]. The addition of adenosine to IFN-γ stimulated cells decreases ser⁷²⁷ phosphorylation, thereby reducing STAT activity [⁴⁹]. This decrease in STAT activity decreases the expression of genes involved in inflammation and lipid uptake [⁴⁸]. Inhibition of the A₃ receptor reverses the suppressive effects of adenosine on STAT ser⁷²⁷ phosphorylation [⁴⁹]. Given the role of STAT in suppressing RCT [⁴⁸], this activity may additionally contribute to the antiatherosclerotic actions of methotrexate [⁴⁹]. Although the A₃-mediated pathway has yet to be uncovered, it has been demonstrated that A₃ receptor activation prevents the accumulation of Ca²⁺ [⁵⁶]. Thus, it is possible that A₃ activity prevents the activation of CaMKII, which subsequently reduces the phosphorylation of ser⁷²⁷ [⁴⁹] (Figure 2).

While methotrexate improves cardiovascular risk in a number of inflammatory diseases [¹⁰], COX-2 inhibition promotes atherosclerosis [⁵]. These opposing cardiovascular influences of methotrexate and COX-2 inhibitors may arise from differing effects on the expression of ABCA1 and 27-hydroxylase, key proteins in RCT [²⁵]. While methotrexate upregulates their expression [¹³], COX-2 inhibition causes a downregulation [¹⁴]. Methotrexate increases the expression of these proteins via adenosine acting upon the A_2A [¹³] and A₃ receptors [⁴⁹]. Activation of A_2A increases expression of the proteins and activity of ABCA1 by a cAMP-mediated process [⁴⁶] while the A₃ receptor counteracts the IFN-γ induced downregulation by inhibiting STAT [⁴⁹]. Furthermore, as proposed in the Cardiovascular Inflammation Reduction Trial, methotrexate administration may provide additional antiatherosclerotic benefits in patients already receiving traditional cholesterol-lowering therapy, regardless of inflammatory disease status [⁵⁷]. Thus, the understanding of the mechanisms by which these medications act to induce cardiovascular effects provides a platform for the design of novel and safer anti-inflammatory and antiatherosclerotic drugs in the future, such as selective adenosine A_2A receptor agonists.

E.S.L. Chan declares that he holds a patent pertinent to King Pharmaceuticals on the use of adenosine A_2A receptor antagonists to inhibit fibrosis.

This work was supported by Grants from the U.S. National Institutes of Health (AR057544), the Scleroderma Foundation, the Arthritis Foundation, and the NYU-HHC Clinical and Translational Science Institute (UL1RR029893) (E.S.L.C.). Additionally, this work was supported in part by an Innovative Research Grant from the Arthritis Foundation National Center (A.B.R.). The authors would also like to thank Farheen Manji, at McMaster University in Hamilton, ON, for her assistance with the figures.