Investigation of the role of endosomal Toll-like receptors in murine collagen-induced arthritis reveals a potential role for TLR7 in disease maintenance

Rheumatoid arthritis (RA) is a chronic autoimmune inflammatory disease that affects 1% of the population. Disease progression is characterized by a destructive inflammation of the joints, which can lead to progressive disability and a reduced life expectancy. The synovial membrane in RA is infiltrated by activated immune cells, most abundantly macrophages and T cells, resulting in the chronic production of pro-inflammatory cytokines and matrix metalloproteinases, leading to inflammation and cartilage and bone degradation [1]. The treatment of RA has been revolutionized by the development of biological therapies specifically targeting immune mediators. These include tumor necrosis factor (TNF) [2], interleukin-1 (IL-1), the IL-6 receptor, B cells, and activated T cells (reviewed in [3]). However, these biologics are not orally available and are expensive to manufacture; their cost severely limits use. Side effects are also common; for example, systemic inhibition of TNF confers an increased risk of infection in patients [4]. Thus, there is a requirement for cheaper and more targeted therapies to treat RA.

To improve the therapies available to patients, it is essential to gain a better understanding of the mechanisms that sustain inflammation in RA. Despite the effectiveness of biological therapies in many patients, disease activity usually resumes once treatment has stopped. This indicates that the upstream mechanisms that generate inflammation are still functional and most likely unaffected by these treatments. Many studies from both murine and human models have suggested a role for a family of innate immune receptors, the Toll-like receptors (TLRs) in RA pathogenesis [5].

TLRs form part of a network of receptors that alert the host to the presence of infection and tissue damage. TLRs can be classified into two distinct groups on the basis of cellular distribution and ligand repertoire. Cell surface-expressed TLRs 1, 2, 4, 5, and 6 recognize ligands of mainly bacterial and fungal origin, whereas TLRs 3, 7, 8, and 9 are expressed intracellularly in endosomes and detect nucleic acids from bacteria and viruses [6]. TLR activation induces a strong inflammatory response, which is characterized by the increased expression of TNF among many other mediators. In addition to pathogen-associated ligands, TLRs can engage a number of endogenous molecules that can be produced during tissue damage and are often found at the sites of chronic inflammation [7].

The concept of endogenous ligand-driven activation of TLRs makes these receptors potential candidates for the induction or maintenance (or both) of chronic inflammatory conditions [8]. A potential role has been emerging for the endosomal TLRs in autoimmune diseases such as RA and systemic lupus erythematosus (SLE), in which it has become apparent that DNA- and RNA-associated autoantigen immune complexes can activate B cells and dendritic cells through activation of TLRs [9‐11]. Clinical data supporting this concept have come from the study of patients deficient in Unc93B1 [12], a protein required for TLR3, 7, 8, and 9 signaling [13]. Unc93b1-deficient patients show increased numbers of naïve autoreactive B cells in the periphery, similar to patients with RA [12], but do not develop autoreactive antibodies or autoimmunity [12].

In a previous study, we identified an off-target effect of the antidepressant mianserin and showed that it inhibited the activation of the endosomal TLRs 3, 7, 8, and 9 and significantly decreased TNF and IL-6 production from human RA synovial membrane cultures [14]. In the present study, we set out to investigate the role of the endosomal TLRs in vivo in an experimental arthritis model using mianserin. Previous work had suggested that TLR8 may be of importance in a human model of RA [14]. However, with no defined ligand, the role of murine TLR8 remains unclear. One study suggests that murine TLR8 may not be functional [15] -leaving murine TLR7 more closely mirroring the behavior of human TLR8. Murine TLR7 and human TLR8 are activated by the same ligands [15‐17] and induce TNF production from macrophages, a key mediator of the disease process in RA [18]. Thus, in this study, we chose to focus on the role of TLR7 in the murine CIA model using TLR7^-/- mice.

In this study, we set out to investigate the role of endosomal TLRs in a murine model of RA by using mianserin, an antidepressant that can inhibit endosomal TLR signaling [14]. Therapeutic administration of mianserin decreased disease progression and appreciably preserved the joint architecture in the CIA model, suggesting a possible role for one or more of these TLRs in the maintenance of disease. This result was consistent with previous data, in which we showed that mianserin could inhibit spontaneous production of TNF and IL-6 from human RA synovial membrane cultures [14]. Conversely, some reports have suggested that 5HT-2A receptor antagonists, including mianserin, can cause adverse drug reactions, including joint problems [28, 29]. However, the discrepancy between these reports and our studies may be explained by the fact that the inhibition of TLRs by mianserin is an off-target effect observed only at doses that are higher than would be safe to administer clinically [30]. Thus, mianserin may produce differing effects via distinct mechanisms when used at low or high concentrations. Accordingly, mianserin does not represent a useful anti-arthritic drug for the clinic. However, these data highlight the potential benefit that might be provided by the development of a more specific set of inhibitors of the endosomal TLRs.

Previous data from a human model of RA had suggested a role for TLR8 in the production of TNF [14]; however, extrapolating this finding across species becomes difficult when considering the role of TLR8. Whereas human TLR7 and 8 can be activated by ssRNA or resiquimod, only TLR7 (and not TLR8) responds to these ligands in rodents [15, 17]. This distinction was recently suggested to be due to a variation between species in leucine-rich repeats 14-15 of TLR8 [31]. The mechanism of murine TLR8 activation and the function of this receptor remain unclear at present. Nonetheless, its functional paralogue murine TLR7 behaves in a manner similar to human TLR8; both receptors are known to be activated by the same natural ligand, ssRNA, and can induce TNF production from macrophages [17]. Thus, this study focused on the role of TLR7, rather than TLR8, in the CIA model.

Here, we show that TLR7 deficiency resulted in a decrease in clinical and histological scores, paw swelling, and disease incidence. Compared with control WT mice in which the clinical score progresses steadily after disease onset, the clinical score in the majority of TLR7^-/- mice failed to increase. Interestingly, disease resolved in about 20% of arthritic mice within 5 days of onset. Initiation of CIA requires CFA, which is a mixture of mineral oil and heat-inactivated MTB; thus, it was possible that TLR7 deficiency may haveprevented the induction of arthritis. Encouragingly, both BMMs and dendritic cells from TLR7-deficient mice responded to heat-inactivated MTB, producing TNF levels comparable to those of cells from WT mice (data not shown). This suggested that the innate immune response to MTB remained intact in TLR7-deficient mice.

Another important mechanism in the initiation of CIA is the generation of anti-collagen antibodies; these antibodies have also been shown to be capable of transferring disease to immunocompromised animals [32, 33]. In a murine model of spontaneous lupus, it was shown that TLR7 deficiency results in decreased serum levels of IgG_2a and IgG3 isotypes [34], which are among the pathogenic isotypes in autoimmune diseases such as SLE [35]. However, it was of interest that we observed that the lack of TLR7 resulted in higher levels of anti-collagen IgG₁ as well as IgG_2a/c in the serum of arthritic mice when compared with WT controls, suggesting that TLR7 signaling has a negative impact on T_H1 responses in the CIA model. Why TLR7 deficiency resulted in opposing effects in two animal models is not understood but may be due to the direct injection of antigen or the presence of adjuvant in the CIA model, perhaps implicating antigen presentation in this discrepancy. Indeed, the increased level of the T_H1-associated IgG_2a/c isotype may have supported the increase in IFNγ production from stimulated DLNCs. Of the T_H1 responses, IFNγ in particular has been associated with protection from disease in experimental models of arthritis (reviewed in [36]). One way in which IFNγ suppresses disease is by inhibiting IL-17 production [37]. Likewise, arthritic TLR7^-/- animals showed a significant decrease in IL-17/T_H17 and an increase in T_reg cells, pointing to a potential role for TLR7 in regulating T-cell plasticity or the balance between T_H17 cell/T_reg cell responses or both. However, on the basis of our findings, it was not possible to establish whether TLR7 deficiency has a direct effect on T cells or acts indirectly via the antigen-presenting cells. T_reg cells have also been implicated in the control of inflammatory arthritis in animal models (reviewed in [36]). TLR7 ligation is reported to suppress foxp3 expression [38] and thus may explain the increase in T_reg cells in TLR7-deficient arthritic mice. However, no change in the level of T_reg cells was observed in DLNs of mianserin-treated mice and this may reflect the difference of specific TLR7 deficiency as opposed to pan-endosomal TLR inhibition.

However, our data do not exclude a role for other TLRs. In the rat pristane model of arthritis, recent studies have shown upregulation of TLR3 in the rat synovium, aggravation of arthritis by the TLR3 ligand poly I:C, and amelioration of disease by TLR3 RNA interference in vivo [39, 40]. TLR9 inhibitors have also been reported to delay disease onset and severity in the rat pristane model [41].

Most studies of murine experimental arthritis have suggested a contribution from TLR4 to the maintenance of inflammation [42‐44]. A role for TLR7 may, in fact, be complementary to these studies, as TLR7-induced type I IFN has been shown to enhance TLR4 activation, most notably in cells taken from patients with RA [45]. Involvement of TLR7 in the maintenance of inflammation in CIA is also in keeping with other studies of inflammatory arthritis models. Low-dose activation of TLR7 with a small synthetic ligand has been shown to induce tolerance of TLR2, 7, and 9 signaling and to suppress disease in a serum transfer model of arthritis [46]. In addition, a recent study of pristane-induced arthritis has shown that splenocytes from an arthritic animal can transfer disease after re-stimulation with heterogeneous nuclear ribonucleoprotein (hnRNP) antigens that activate TLR7 and 9 [47]. Intra-articular lentiviral delivery of TLR7 short-hairpin RNA was recently shown to decrease IL-1 and IL-6 expression in synovial tissue of CIA rats [48]. This is in agreement with our data showing a decrease in IL-1, IL-6, and other cytokines in the paw tissue from the TLR7^-/- CIA mice.

In summary, this study has uncovered an important role for TLR7 in CIA which complements other animal model studies [46, 47] and shows a role comparable to that of its human homologue, TLR8, in RA [14]. Given that the activity of TLR7 and 8 can be modulated by small synthetic agents such as mianserin, these receptors may provide amenable targets for the development of new therapeutics for RA.