Regulation of inflammatory responses in tumor necrosis factor–activated and rheumatoid arthritis synovial macrophages by JAK inhibitors

Authors


Abstract

Objective

JAK inhibitors have been developed as antiinflammatory and immunosuppressive agents and are currently undergoing testing in clinical trials. The JAK inhibitors CP-690,550 (tofacitinib) and INCB018424 (ruxolitinib) have demonstrated clinical efficacy in rheumatoid arthritis (RA). However, the mechanisms that mediate the beneficial actions of these compounds are not known. The purpose of this study was to examine the effects of both JAK inhibitors on inflammatory and tumor necrosis factor (TNF) responses in human macrophages.

Methods

In vitro studies were performed using peripheral blood macrophages derived from healthy donors and treated with TNF and using synovial fluid macrophages derived from patients with RA. Levels of activated STAT proteins and other transcription factors were detected by Western blotting, and gene expression was measured by real-time polymerase chain reaction analysis. The in vivo effects of JAK inhibitors were evaluated in the K/BxN serum–transfer model of arthritis.

Results

JAK inhibitors suppressed the activation and expression of STAT-1 and downstream inflammatory target genes in TNF-stimulated and RA synovial macrophages. In addition, JAK inhibitors decreased nuclear localization of NF-κB subunits in TNF-stimulated and RA synovial macrophages. CP-690,550 significantly decreased the expression of interleukin-6 in synovial macrophages. JAK inhibitors augmented nuclear levels of NF-ATc1 and cJun, followed by increased formation of osteoclast-like cells. CP-690,550 strongly suppressed K/BxN serum–transfer arthritis, which is dependent on macrophages, but not lymphocytes.

Conclusion

Our findings demonstrate that JAK inhibitors suppress macrophage activation and attenuate TNF responses and further suggest that suppression of cytokine/chemokine production and innate immunity contribute to the therapeutic efficacy of JAK inhibitors.

Rheumatoid arthritis (RA) is a chronic inflammatory disease that preferentially targets synovial tissue, cartilage, and bone. Multiple cytokines produced by innate and adaptive immune cells are implicated in pathogenesis of RA (1). Imbalance between proinflammatory and antiinflammatory cytokines leads to autoimmunity, chronic inflammation, and tissue destruction. Several biologic agents against specific cytokines and their receptors have been developed, with tumor necrosis factor (TNF) inhibitors leading the pack, and have demonstrated clinical efficacy in chronic inflammatory diseases, including RA (2, 3). However, resistance to therapy in subpopulations of patients, increased infection rates, high treatment costs, difficulty in titrating the dosage, and injection-related complications have prompted the search for orally active small-molecule compounds that can selectively interfere with molecular mediators of cytokine signaling. Recently, the JAK family of nonreceptor tyrosine kinases, which plays a critical role in mediating inflammatory and immune responses, has gained significant interest as a therapeutic target (4–6).

The JAK family is composed of 4 enzymes (JAK-1, JAK-2, JAK-3, and Tyk-2) that control signaling by numerous cytokines that are important for acquired and innate immunity and hematopoiesis (4, 7). In resting cells, JAKs associate with the intracellular domains of type I and II cytokine receptors. Upon ligation of cytokine receptors, JAKs transactivate each other and phosphorylate tyrosine residues on the receptor cytoplasmic domain, leading to the recruitment and phosphorylation of STATs, which culminates in STAT dimerization, translocation to the nucleus, and activation of gene transcription (7–9). Studies in mice and humans with deleted or mutated JAKs revealed their specific role in the regulation of cytokine signaling. JAK-1, JAK-2, and Tyk-2 regulate signaling triggered by the activation of both type I and II cytokine receptors, whereas JAK-3 is specifically associated with interleukin-2 (IL-2) receptor common γ-chain (γc) shared by the receptors for cytokines important for the development and function of T cells, B cells, and natural killer cells (10–14).

Several small-molecule JAK inhibitors are currently in clinical development for the treatment of transplant rejection, hematopoietic disorders, and autoimmune and inflammatory diseases, including RA (4, 6). Among them, CP-690,550 (tofacitinib) demonstrated significant efficacy in RA (5, 15). CP-690,550 was initially developed as a selective JAK-3 inhibitor; however, recent studies demonstrated that in cell culture, it suppressed cytokine signaling mediated by JAK-1/3, JAK-1/2, and JAK-1/Tyk-2, with much less activity against JAK-2 homodimers important for signaling by hematopoietic factors (16, 17).

INCB18424 (ruxolitinib) has higher specificity against JAK-1, JAK-2, and Tyk-2, and has also demonstrated clinical efficacy in RA clinical trials (6, 18). Despite the successful results of clinical trials and efficacy in animal models of arthritis, the precise mechanism of action by CP-690,550 and INCB018424 that suppresses disease activity in RA is not clear. Consistent with effective inhibition of γc cytokines required for lymphocyte proliferation and function, several in vivo and in vitro studies of CP-690,550 demonstrate suppression of lymphocyte activation and proliferation in various animal models (15, 19–21). Also, CP-690,550 interferes with Th1 and Th2 differentiation and impairs the production of inflammatory Th17 cells (17). It was recently suggested that CP-690,550 can also target innate immunity in vivo (17); the underlying mechanisms are completely unknown, as JAKs do not play a direct role in signaling by many receptors that are important for innate immune responses, such as TNF, IL-1, or Toll-like receptors (TLRs).

Macrophages are innate immune cells that play an important role in synovial inflammation and tissue destruction in RA (22). Macrophages contribute to RA pathogenesis in part by producing key inflammatory cytokines, such as TNF, IL-1, and IL-6, and chemokines, such as IL-8 and CXCL10 (interferon-γ–inducible protein 10). RA synovial macrophages express high levels of STAT-1 and an “interferon (IFN)/STAT-1 signature” reflecting activation by synovium-expressed cytokines that signal via JAK-1, JAK-2, and Tyk-2, likely including IFNβ, IFNγ, and IL-6 (1, 8, 12, 13, 23, 24). Importantly, JAK/STAT signaling in macrophages can be indirectly activated by innate immune receptors, such as TLRs and TNF receptors, via induction of an autocrine loop mediated by cytokines such as IFNβ and IL-6 (1, 25, 26). JAK/STAT signaling in macrophages augments the production of multiple inflammatory cytokines and chemokines (27), and the importance of a TNF/IFNβ/JAK/STAT-1 autocrine loop in cell activation and inflammatory gene expression has been recently established (25, 28). This suggests that JAK inhibitors may also target macrophages to suppress inflammatory cytokine and chemokine production.

We therefore examined the effects of JAK inhibition on inflammatory responses in human blood–derived and RA synovial macrophages, with a focus on the key pathogenic cytokine, TNF, which activates JAK/STAT signaling indirectly and with delayed kinetics. JAK inhibitors abrogated the expression of STAT-dependent cytokines such as CXCL10. Unexpectedly, JAK inhibitors also decreased the nuclear localization of NF-κB subunits, and CP-690,550 significantly decreased the expression of IL-6 in synovial fluid macrophages. Both JAK inhibitors augmented nuclear levels of NF-ATc1 and cJun, followed by increased formation of osteoclast-like cells. Last, CP-690,550 effectively suppressed K/BxN serum–transfer arthritis, a model that is solely dependent upon innate immune mechanisms. Our data demonstrate that JAK inhibitors suppress the inflammatory functions of macrophages, in part by altering cell responses to the key pathogenic cytokine TNF. These findings suggest that suppression of macrophages and innate immunity may contribute to the therapeutic efficacy of JAK inhibitors in RA.

MATERIALS AND METHODS

Cell culture and media.

Synovial fluids from RA patients were obtained by their physicians as a part of standard medical care using a protocol approved by the Institutional Review Board of the Hospital for Special Surgery. De-identified specimens that would otherwise have been discarded were used in this study. Peripheral blood mononuclear cells were isolated from blood leukocyte preparations (NYC Blood Center) or from synovial fluids by density-gradient centrifugation, and CD14+ cells were purified using anti-CD14 magnetic beads (Miltenyi Biotec). Human monocytes were cultured overnight in α-minimum essential medium (α-MEM; Invitrogen Life Technologies) supplemented with 10% fetal bovine serum (FBS; HyClone), 100 units/ml of penicillin/streptomycin (Invitrogen Life Technologies), 2 mM L-glutamine (Invitrogen Life Technologies), and 20 ng/ml of human macrophage colony-stimulating factor (M-CSF; PeproTech). The following reagents were added to cell cultures as indicated: recombinant human TNF, 40 ng/ml (PeproTech), recombinant universal type IFNα A/D, 5,000 units/ml (PBL Interferon Source), recombinant human IFNγ, 100 units/ml (Roche Applied Science), CP-690,550 0.1–10 μM, and INCB18424 0.1–1 μM (Active Biochemicals).

Multinuclear cell/osteoclast differentiation.

Human CD14+ cells (0.25 × 106 cells/ml) were incubated in α-MEM supplemented with 10% FBS, 20 ng/ml of M-CSF, and 40 ng/ml of human TNF for various times in the presence or absence of JAK inhibitors. Cytokines were replenished every 3 days. At the end of culture period, cells were stained for tartrate-resistant acid phosphatase (TRAP) activity, according to the manufacturer's instructions (Sigma). Multinucleated (>3 nuclei), TRAP-positive cells were counted in triplicate wells of 96-well plates. For bone resorption assays, cells were cultured for 25 days as described above on Corning Osteo Assay Surface 96-well plates. Cells were removed by incubation for 10 minutes with 10% bleach, and the area of resorption was quantified using IPLab imaging software (BD Biosciences).

Real-time quantitative PCR.

Total RNA was extracted using an RNeasy Mini kit (Qiagen) with DNase treatment, and 0.5 μg of total RNA was reverse transcribed using a First-Strand cDNA Synthesis kit (Fermentas). Quantitative PCR was performed using Fast SYBR green Master Mix and a 7500 Fast Real-time PCR System (Applied Biosystems). Expression of the tested genes was normalized relative to the levels of GAPDH.

Immunoblotting.

Cytoplasmic and nuclear cell extracts were obtained, and equal amounts of total protein were fractionated on 7.5% polyacrylamide gels using sodium dodecyl sulfate–polyacrylamide gel electrophoresis, transferred to PVDF membranes (Millipore), incubated with specific antibodies recognizing NF-ATc1, STAT-2 (BD Biosciences), RelB, NF-κB p100/p52, phospho–NF-κB p65 (Ser536), c-Jun, Akt, and phospho–STAT-1 (Tyr701) (Cell Signaling Technology), phospho–STAT-2 (Tyr689) (Millipore), lamin B1 (Abcam), and p38 (Santa Cruz Biotechnology), and horseradish peroxidase–conjugated secondary antibodies were used for detection with enhanced chemiluminescence (Amersham). The signal intensities of bands specific for transcription factors were quantified using IPLab imaging software (BD Biosciences) and normalized relative to the intensity of loading control lamin B1.

Mouse arthritis model.

We used C57BL/6 mice (The Jackson Laboratory) in our study. Animals were maintained in the Animal Facility of the Hospital for Special Surgery, and protocols were approved by the Institutional Animal Care and Use Committee. K/BxN serum pools were prepared as described elsewhere (29). Arthritis was induced by intraperitoneal injection of 100 μl of K/BxN serum (experimental group; n = 10 mice) or PBS (control group; n = 10 mice) on days 0 and 2. Control and arthritic animals were divided into 2 additional groups and administered vehicle (0.5% methylcellulose/0.025% Tween 20; Sigma-Aldrich) or 50 mg/kg of CP-690,550 resuspended in 0.5% methylcellulose/0.025% Tween 20 twice daily by oral gavage beginning on day 1. The severity of arthritis was monitored by measuring the thickness of both the wrist and ankle joints using dial-type calipers (Bel-Art Products). For each animal, the joint thickness was calculated as the sum of the measurements of both wrists and both ankles. The joint thickness was represented as the average for every treatment group. For histopathologic assessment, mice were euthanized on day 9 after the first serum injection and the fore paws and hind paws were harvested and fixed in 10% neutral buffered formalin for 24 hours.

Histopathology.

Fixed paws were decalcified for 7 days with 10% neutral buffered EDTA (Sigma-Aldrich) and embedded in paraffin. To assess inflammation, sections measuring 5 μm thick were stained with hematoxylin (Sigma-Aldrich), fast green (Acros Organics), and Safranin O (Sigma-Aldrich) using standard techniques.

Statistical analysis.

Statistical analysis was performed using GraphPad Prism analytical software version 5.03 for Windows. Statistical tests included nonparametric Wilcoxon matched-pairs signed-ranks test, two- and one-way analysis of variance with post hoc tests for multiple comparisons. P values less than 0.05 were considered significant.

RESULTS

JAK inhibitors CP-690,550 and INCB018424 inhibit IFN-induced STAT activation.

The efficacy of JAK inhibitors can vary according to cytokine receptor, associated JAKs, and cell type, and we wanted to test and compare the effects of CP-690,550 and INCB018424 on signaling by cytokines that activate STAT-1 in human macrophages and can contribute to the “STAT-1 signature” observed in RA (30, 31). We stimulated primary human macrophages for 15 minutes with IFNα (Figure 1A), which activates JAK-1/Tyk-2 or IFNγ (Figure 1B), which signals through JAK-1/2 (23, 32). IFNα activates STAT-1 and STAT-2, whereas IFNγ activates primarily STAT-1. We prepared nuclear extracts and analyzed STAT activation by measuring nuclear translocation and tyrosine phosphorylation.

Figure 1.

Dose-dependent effect of JAK inhibitors on STAT signaling in human macrophages. Human macrophages were left untreated or were stimulated with interferon-α (IFNα) (A) or IFNγ (B) in the presence of increasing concentrations of JAK inhibitors CP-690,550 (CP) or INCB018424 (INCB) for 15 minutes. Nuclear extracts were analyzed by immunoblotting for the expression and tyrosine phosphorylation of STAT proteins. Results are representative of 2 independent experiments.

CP-690,550 and INCB018424 blocked IFNα- and IFNγ-induced STAT-1 and STAT-2 nuclear translocation and tyrosine phosphorylation in a dose-dependent manner, and strong inhibition was observed at nanomolar concentrations of JAK inhibitors (Figures 1A and B). IFNγ signaling was inhibited more effectively than IFNα signaling, which is most likely explained by the lower efficacy of Tyk-2 inhibition by these compounds (5, 6, 33, 34). Overall, INCB018424 inhibited IFN signaling at lower concentrations than did CP-690,550, which is consistent with the differing potency of these compounds in suppressing JAK-1 and JAK-2 (6, 33). These findings demonstrate that CP-690,550 and INCB018424 can inhibit IFN/JAK/STAT signaling in primary human macrophages at concentrations similar to those reported for other cell types (5, 6, 16, 17, 28) and show modestly different dose-dependent effects of these JAK inhibitors.

JAK inhibitors decrease TNF-dependent STAT-1 activation, STAT-1 expression, and induction of IFN-dependent genes.

We wished to test the effects of JAK inhibitors on macrophage responses to the key pathogenic cytokine TNF. We recently demonstrated that in macrophages, TNF-induced expression of key T cell chemokines, such as CXCL10 and CXCL11, is dependent on synergy between canonical TNF signaling and a TNF-induced IFNβ/JAK/STAT-mediated autocrine loop, which also activates the delayed expression of classic IFN response genes, such as interferon-induced protein with tetratricopeptide repeats 1 (IFIT-1) and interferon regulatory factor 7 (IRF-7) (25). We found that both CP-690,550 and INCB018424 inhibited TNF-induced expression of CXCL10 and IFIT-1 in a dose-dependent manner (Figure 2A).

Figure 2.

Down-regulation of tumor necrosis factor (TNF)–induced STAT-1 activation and expression of STAT-1–dependent genes by JAK inhibitors. A, Human macrophages were left untreated or were stimulated with TNF in the presence of increasing concentrations of CP-690,550 (CP) or INCB018424 (INCB) for 6 hours, and the expression of STAT-1–dependent genes was analyzed by quantitative polymerase chain reaction (qPCR). Results are representative of 2 independent experiments. B and C, Macrophages were stimulated with TNF for the indicated times in the presence or absence of JAK inhibitors, and mRNA expression of STAT-1–dependent chemokines (B) and interferon response genes (C) was measured by qPCR. Target gene expression is presented relative to expression in untreated cells 1 hour after TNF simulation (results set at 1). Values are the mean ± SD of 6 (for CP-690,550) and 5 (for INCB018424) independent experiments. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001 versus TNF plus each inhibitor, by repeated-measures two-way analysis of variance with Bonferroni post hoc test for multiple comparisons. IRF-7 = interferon-regulatory factor 7; IFIT-1 = interferon-induced protein with tetratricopeptide repeats 1. D, Human macrophages were left untreated or were stimulated with TNF in the presence or absence of CP-690,550 or INCB018424 for the indicated times. Tyrosine phosphorylated and total STAT-1 in nuclear (top) and cytoplasmic (bottom) extracts were analyzed by immunoblotting. Results are representative of 5 independent experiments.

TNF-induced IFNβ expression can be detected within 2 hours after stimulation, reaches a maximum at 6 hours, returns to baseline after 24 hours of culture, and leads to sustained STAT-1 activation and related gene expression for several days (25, 35). We therefore performed a time-course analysis of the effects of JAK inhibition on the expression of chemokines and IFN response genes (Figures 2B and C; note logarithmic scale on ordinate) from 3 to 48 hours after TNF stimulation. CP-690,550 and INCB018424 strongly suppressed the TNF-mediated induction of the CXCL10 and CXCL11 chemokine genes and the IFIT-1 and IRF-7 IFN response genes over the entire time course (Figures 2B and C). Multiple TNF-induced intermediate response genes (regulated similarly to CXCL10 and CXCL11) and classic IFN response genes were inhibited by CP-690,550 and INCB018424 without a significant effect on cell viability; TNF-induced IFNβ expression was not affected by JAK inhibitors (data not shown). Thus, inhibition of JAKs not only resulted in the expected suppression of IFN response genes, but also strongly suppressed inflammatory chemokine genes. This suggests that canonical NF-κB signaling is not sufficient to fully induce the expression of these chemokine genes and that JAK inhibitors broadly suppress TNF responses.

Next, we examined TNF-activated STAT-1 signaling and found that CP-690,550 and INCB018424 abrogated the tyrosine phosphorylation that regulates the transcription activity of STAT-1 and suppressed the nuclear translocation of STAT-1 (Figure 2D, upper panel; compare lanes 4 versus 5 and 6; lanes 10 versus 11 and 12; and lanes 16 versus 17 and 18). JAK inhibitors suppressed TNF-induced STAT-1 activation at both early and late time points (Figure 2D), and this inhibition correlated with suppression of TNF-induced gene expression (Figures 2A–C). STAT-1 itself is a target of JAK/STAT signaling and is highly expressed in RA synovium (30). Inhibition of JAKs decreased total STAT-1 protein and mRNA expression in TNF-treated macrophages at 24 and 48 hours (Figure 2D, lanes 16–18 and 22–24; data for mRNA expression not shown). Taken together, our results demonstrate that JAK inhibitors abrogate TNF-activated IFN/STAT-1 signaling and suppress STAT-1 expression in human macrophages, which in turn, leads to decreased expression of proinflammatory chemokines and suppression of IFN-regulated genes.

JAK inhibitors increase TNF-induced NF-ATc1 activation and formation of osteoclast-like cells.

We recently found that prolonged exposure (days) of human macrophages to TNF activates an NF-ATc1–mediated gene program important for cell fusion and osteoclastogenesis (35). Activation of NF-AT transcription factors requires dephosphorylation, which allows nuclear translocation and transcription of target genes (36). We examined TNF-induced NF-ATc1 activation in the presence of JAK inhibitors (Figure 3A) and found that CP-690,550 and INCB018424 strongly increased nuclear expression of NF-ATc1 starting at 24 hours of culture (Figure 3A, lanes 4–6 and lanes 10–12). This finding with TNF is consistent with previous reports showing that IFN/STAT signaling can also inhibit RANKL-induced NF-ATc1 activation and osteoclastogenesis (37, 38).

Figure 3.

JAK inhibitors augment tumor necrosis factor (TNF)–induced expression of NF-ATc1 and cJun and increase the formation of osteoclast-like cells. Human macrophages were stimulated with TNF or were left untreated in the presence or absence of CP-690,550 (CP) or INCB018424 (INCB). A and B, Nuclear extracts were analyzed for the expression of NF-ATc1 (A) and cJun (B, left). Results are representative of 5 independent experiments. The intensity of the cJun bands at 24 hours was quantified by densitometry (B, right). Values are the mean ± SEM of 5 independent experiments. C, Quantification of TNF-induced tartrate-resistant acid phosphatase (TRAP)–positive multinuclear cells (MNCs). Cells were fixed and stained for TRAP activity after 8–12 days of culture. Symbols represent individual data points; horizontal lines show the mean. ∗∗ = P < 0.01 by one-way analysis of variance with Dunn's post hoc test. D, Quantification of TNF-induced resorption. Values are the mean ± SEM of 4 independent experiments. ∗ = P < 0.05 by one-way analysis of variance with Dunn's post hoc test.

In human macrophages, the cJun member of the activator protein 1 (AP-1) family is important for TNF-mediated activation of NF-ATc1 (35). CP-690,550 and INCB018424 treatment increased the nuclear expression of cJun at 24 hours after TNF stimulation (Figure 3B, lanes 4–6; densitometric quantification of cJun induction in 5 independent experiments shown at the right), which correlated with up-regulation of NF-ATc1 nuclear levels (Figure 3A, lanes 4–6). Next, we examined the effects of JAK inhibition on TNF-induced osteoclastogenesis and found that CP-690,550 treatment significantly increased the formation of TRAP+ multinuclear cells in 90% of experiments (Figure 3C) and strongly enhanced the resorptive activity of osteoclasts (Figure 3D). INCB018424 treatment had variable effects on osteoclastogenesis, with increased cell fusion in 70% of experiments (Figure 3C and data not shown) without increasing the resorptive activity (Figure 3D). Moreover, cell fusion was observed more rapidly in the presence of JAK inhibitors (data not shown). Overall, the results show that JAK inhibitors can enhance aspects of TNF-induced cell fusion and osteoclast differentiation.

JAK inhibitors attenuate the late phase of TNF-induced NF-κB activation and affect expression of inflammatory cytokine genes.

CP-690,550 and INCB018424 can decrease plasma levels of inflammatory cytokines (6, 16, 17). However, the cellular basis of this phenomenon is not known. Cytokine induction in response to inflammatory stimuli, such as lipopolysaccharide and TNF, occurs rapidly and declines after several hours. On the other hand, late expression of inflammatory cytokines in response to TNF has not been explored. We therefore analyzed the expression of IL-1β, TNF, and IL-6 in human macrophages stimulated with TNF for 1–48 hours in the presence or absence of JAK inhibitors (Figures 4A and B).

Figure 4.

Effect of JAK inhibition on the expression of inflammatory cytokines and NF-κB signaling. Human macrophages were left untreated or were stimulated with tumor necrosis factor (TNF) in the presence or absence of CP-690,550 (CP) or INCB018424 (INCB) for the indicated times. A and B, Levels of mRNA for TNF (top) and interleukin-6 (IL-6) (bottom) (A) and for interleukin-1β (B, top) were determined by quantitative polymerase chain reaction. Gene expression is presented relative to expression in untreated cells 1 hour after TNF simulation (set at 1). Results are representative of 6 (for CP-690,550) and 5 (for INCB018424) independent experiments. Values are the mean. The effects of CP-690,550 and INCB018424 on IL-1β mRNA expression 48 hours after TNF stimulation was also determined (B, bottom). Each symbol represents an independent experiment (n = 5 or 6). ∗ = P < 0.05 by Wilcoxon's matched pairs signed rank test. C, Nuclear extracts were analyzed by immunoblotting for the expression of NF-κB subunits (top). Results are representative of 3 (for phosphorylated-p65) and 5 (for RelB and p52) independent experiments. The intensity of the RelB and p52 bands was quantified by densitometry (bottom). Values are the mean ± SEM of 5 independent experiments. ∗ = P < 0.05 by one-way analysis of variance with Dunn's post hoc test.

The expression of TNF and IL-6 followed the expected transient expression pattern described above (Figure 4A, top and bottom). Surprisingly, IL-1β expression demonstrated a second wave of increase, with a second peak at 24 hours post-TNF stimulation (Figure 4B, top). CP-690,550 and INCB018424 did not affect the early expression of proinflammatory cytokines (Figures 4A and B, top), but in contrast, suppressed the late wave of IL-1β induction, with significant inhibition by CP-690,550 (Figure 4B, lower panel; note the logarithmic scale on the ordinate).

To explain the suppression of the late expression of IL-1β, we analyzed the effects of JAK inhibitors on the late phase of TNF-induced signaling. We previously demonstrated that TNF induces nuclear accumulation of components of both canonical and noncanonical NF-κB pathways, with biphasic kinetics characterized by sustained nuclear expression of phospho–(Ser536)-p65 (RelA), p52, and RelB at 24–72 hours after TNF stimulation (35). JAK inhibition affected TNF-induced nuclear accumulation of NF-κB subunits at 24 and 48 hours after TNF stimulation, with the most prominent inhibitory effect for RelB and p52 at the 48-hour time point (Figure 4C, lanes 4–6 and lanes 10–12; densitometric quantification of the results from 5 independent experiments shown at the bottom). However, RNA interference–mediated knockdown of RelB or upstream kinase IKKα had minimal effects on the late phase of IL-1β expression (data not shown), suggesting that an alternative JAK/STAT-dependent mechanism contributes to the late phase of IL-1β regulation.

JAK inhibitors affect RA synovial macrophages.

Next, we investigated the direct pathophysiologic importance of our findings by testing the effects of JAK inhibitors on the inflammatory phenotype of RA synovial macrophages. RA synovium and synovial macrophages demonstrate an “IFN signature,” as evidenced by increased expression of IFN-regulated genes, including STAT-1 and the chemokine and IFN response genes analyzed in this study. However, the function of the “IFN/STAT-1 signature” in synovial macrophages is not well understood (39). We used JAK inhibitors to test the role of JAK/STAT signaling in RA synovial macrophages.

As shown on Figure 5A, CP-690,550 and INCB018424 strongly and significantly suppressed the expression of CXC chemokines, IFN response genes, and STAT-1 in RA synovial macrophages. Interestingly, CP-690,550 also significantly decreased IL-6 expression (Figure 5A, top), whereas INCB018424 displayed variable effects on IL-6 expression in synovial macrophages (Figure 5A, bottom). Consistent with these results, CP-690,550 and INCB018424 decreased nuclear expression of tyrosine-phosphorylated STAT-1, total STAT-1, RelA, and RelB in RA synovial macrophages (Figure 5B). We previously demonstrated that NF-ATc1 is expressed in synovial macrophages from patients with inflammatory arthritis (35). JAK inhibitors further increased nuclear expression of NF-ATc1 in RA synovial macrophages (Figure 5B). These results show that JAK inhibitors suppress the inflammatory phenotype of RA synovial macrophages while augmenting the expression of NF-ATc1.

Figure 5.

Effects of JAK inhibitors on synovial macrophages isolated from patients with rheumatoid arthritis (RA). CD14+ cells were isolated from the synovial fluid of patients with RA and were cultured for 24 hours with or without CP-690,550 (CP) or INCB018424 (INCB). A, Expression of mRNA for interleukin-1β (IL-1β), IL-6, tumor necrosis factor (TNF), CCL5, CXCL9, CXCL10, CXCL11, STAT-1, interferon-regulatory factor 7 (IRF-7), and interferon-induced protein with tetratricopeptide repeats 1 (IFIT-1) in JAK inhibitor–treated cells is shown as the percentage of expression in untreated cells (set at 100% [broken line]). Results are representative of 7 (CP-690,550) and 6 (INCB018424) independent experiments. Values are the mean ± SEM. ∗ = P < 0.05 for no stimulation versus each inhibitor, by Wilcoxon's matched-pairs signed rank test. B, Synovial macrophages isolated from patients with RA (lanes 3–7) were treated as in A, and nuclear extracts were analyzed by immunoblotting for the expression of the indicated transcription factors. Human donor (HD) macrophages cultured with TNF for 24 hours (lane 2) or for 72 hours (lane 8) served as a positive control, whereas cells cultured without TNF for same durations (lanes 1 and 9) represent a negative control.

CP-690,550 ameliorates joint inflammation in the K/BxN serum–induced mouse model of arthritis.

We evaluated the effect of JAK inhibition in the K/BxN serum–transfer model of arthritis, which is driven by innate immunity and inflammatory cytokines, including TNF and IL-1β (29, 40). K/BxN arthritis is mediated by innate immune cells, including macrophages, and does not require T cells and B cells that express common γ-chain receptors that are sensitive to JAK-3 inhibition (40, 41). Arthritis was induced by intraperitoneal injection of pooled K/BxN serum on days 0 and 2, and CP-690,550 or vehicle control treatment was started from day 1. As expected, arthritis developed rapidly in mice injected with K/BxN serum and vehicle control (Figure 6A). CP-690,550 treatment significantly and nearly completely suppressed the development of arthritis, as assessed by measurements of joint thickness (Figure 6A) and by histologic examination of the ankle joints (Figure 6B). Histologic analysis revealed that CP-690,550 suppressed synovial hyperplasia, with decreased numbers of synovial lining cell layers and decreased synovial thickness (Figure 6B). Thus, inhibition of JAKs effectively suppressed the effector phase of arthritis that depends solely on innate immune mechanisms.

Figure 6.

CP-690,550 attenuates K/BxN serum–induced arthritis. Arthritis was induced in C57BL/6 mice as described in Materials and Methods. A, The time course of arthritis development in the presence and absence of CP-690,550 (CP) treatment. Values are the mean ± SEM of 5 mice per group. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗∗ = P < 0.0001 versus K/BxN serum plus CP, by repeated-measures two-way analysis of variance with Bonferroni correction for multiple comparisons. B, Histologic features of the ankle joints of mice examined on day 9 after the first serum injection. Hind legs were fixed and decalcified, and ankle joint sections were prepared and stained with hematoxylin, fast green, and Safranin O. Original magnification × 40. Large arrows indicate synovium; asterisks indicate bone; small arrows indicate cartilage.

DISCUSSION

Several small-molecule JAK inhibitors are currently in development for the treatment of RA, with CP-690,550 being in advanced stage of clinical trials. Results of multiple studies suggest that beneficial, as well as adverse, effects of JAK inhibitors are related to inhibition of multiple JAKs in different cell types. However, the inhibition of JAK signaling in T cells has been the main focus of research, and little is known about the effects of JAK inhibitors on cells of the innate immune system. In this study, we demonstrated that JAK inhibitors CP-690,550 and INCB018424 could effectively suppress the activation of blood-derived and RA synovial macrophages, including a subset of inflammatory responses induced by the pathogenic cytokine TNF.

In addition to interrupting an IFN-mediated autocrine loop and STAT-1, which promote inflammatory chemokine production, JAK inhibitors unexpectedly suppressed late phases of NF-κB activation and of inflammatory cytokine production, while augmenting the TNF-mediated induction of c-Jun and NF-ATc1. CP-690,550 effectively suppressed K/BxN serum–transfer arthritis, which is entirely dependent on innate immune cells. Overall, our findings demonstrate that JAK inhibitors such as CP-690,550 and INCB018424 effectively inhibit human macrophages, thus identifying another cellular target for JAK inhibition therapy. The results also suggest that inhibition of JAK/STAT signaling in innate immune cells and attenuation of TNF responses contribute to the efficacy of JAK inhibitors in the treatment of RA.

A key question is inhibition of which cell types and which cytokines is responsible for the therapeutic effectiveness of JAK inhibitors. Previous reports have suggested a role for inhibition of T cells and fibroblasts (16, 17, 28), and we have now added macrophages to this list. It is possible that inhibition of other innate immune cell types, such as neutrophils and mast cells, may contribute to the efficacy of CP-690,550 in K/BxN arthritis, although these cell types are not prominently regulated by JAK/STAT signaling cytokines. In terms of explaining efficacy based on which cytokine is being targeted, it is likely that inhibition of T cell γc cytokine/JAK-3 signaling contributes to the efficacy of CP-690,550, although perhaps less so with INCB018424, which is more selective for JAK-1 and JAK-2.

Many cytokines expressed in RA synovium that act on macrophages and innate immune cells are implicated in the pathogenesis of RA, including IL-6, IL-15, granulocyte–macrophage colony-stimulating factor, type I IFNs (IFNα/β), and IFNγ. Of these, IL-6 is an attractive candidate target for explaining the efficacy of JAK inhibitors, since IL-6 blockade is an effective therapy for RA. However, inhibition of K/BxN arthritis, which is independent of IL-6 (41), by CP-690,550 indicates that inhibition of signaling by other cytokines contributes to the clinical efficacy of JAK inhibitors on the effector phase of arthritis. Our results raise the possibility that inhibition of TNF and IFN signaling helps to explain the therapeutic efficacy of JAK inhibitors.

IFN/STAT-1 signaling, as evidenced by high levels of expression of STAT-1 and IFN-target genes known as an “IFN signature,” occurs in RA synovial cells (30, 31, 42). This IFN signature is induced in RA synovial macrophages at least partly by TNF (25, 43) and may contribute to pathogenesis. One mechanism by which an IFN signature can contribute to synovitis is by expression of IFN-inducible genes that promote inflammation, such as the chemokines CXCL10 and CXCL11, which were shown to be sensitive to JAK inhibitors in this study. In addition, IFN-stimulated cells and cells that express high levels of STAT-1 respond more strongly to inflammatory stimuli, such as TLRs and inflammatory cytokines, and increased cytokine production associated with such enhanced responses likely contributes to disease pathogenesis (23, 30, 44). On the other hand, type I IFN plays a protective role in animal models of arthritis, possibly related to inhibition of stromal and endothelial cells (39, 45–47). In most arthritis models, IFNγ also can be protective, depending on timing and context (1, 27).

Thus, inhibiting IFN signaling using JAK inhibitors can have both beneficial and harmful effects that are relevant to the pathogenesis of RA. The balance between these effects, and thus the functional outcome, will be determined by the timing, context, and cell type in which JAKs are inhibited. To date, it appears that JAK inhibition is overall strongly beneficial for suppressing disease activity. Interestingly, our findings showed that JAK inhibitors also partially suppressed macrophage responses to TNF, a cytokine that is clearly pathogenic in RA (1–3). This raises the question of how JAK inhibitors block cellular responses to TNF, which does not signal directly by the JAK/STAT pathway.

JAK inhibitors work in part by suppressing a TNF/IFNβ/JAK/STAT-1 autocrine loop, which we previously described (25) and likely is operative in RA synovial macrophages (43). Among TNF-induced STAT-1 target genes suppressed by JAK inhibitors, the CXCL9, CXCL10, and CXCL11 group of chemokines that interacts with CXCR3 receptors on T cells has been related to the pathogenesis of arthritis (31, 48). Moreover, the genes encoding these chemokines were among the genes most strongly suppressed by JAK inhibitors in RA synovial macrophages. In addition, JAK inhibitors had unexpected inhibitory effects on TNF responses, namely, suppression of the late phase of NF-κB signaling and, in parallel, suppression of the production of inflammatory cytokines, including IL-1 and IL-6. The suppression of IL-6 expression was especially notable in RA synovial macrophages. Thus, the efficacy of JAK inhibitors in RA may be partially explained by inhibition of innate immune cytokine production by synovial macrophages. The most likely mechanism is inhibition of JAK-dependent priming effects that elevate STAT-1 and augment inflammatory cytokine production in response to various macrophage-activating factors (44).

Our findings also revealed that inhibition of JAKs resulted in an increased TNF-mediated induction of c-Jun and NF-ATc1, and a parallel increase in osteoclastogenesis. These results are consistent with reports that JAK/STAT signaling can inhibit osteoclastogenesis (37, 38). These results raise a cautionary note that JAK inhibition may lead to increased bone resorption in certain settings. Evidence against this possibility are the results of clinical studies and animal experiments showing a protective role of CP-690,550 against joint destruction (5, 15). This is most likely because JAK inhibitors so effectively suppress inflammation (as also observed in our experiments with the K/BxN arthritis model) that inflammation-induced factors that drive synovial osteoclastogenesis, such as RANKL, are suppressed. Alternatively, JAK inhibitors can suppress bone erosion by suppressing osteoclastogenic Th17 cells (17) and may also promote osteoblast function (49). The increase in NF-ATc1 observed in synovial macrophages treated with JAK inhibitors, however, suggests that increased osteoclast formation might be a potential problem and that careful monitoring of bone resorption is probably warranted for patients receiving JAK inhibitor therapy.

In conclusion, taken together with previous findings in T cells and synovial fibroblasts, our results indicate that JAK inhibition can affect multiple steps in the pathogenesis of RA by targeting cytokine and chemokine production and by affecting the function of innate and acquired immune cells. Suppression of the expression of STAT-1 and STAT-1–dependent chemokines, of inflammatory cytokine production by synovial macrophages, and of TNF responses likely contributes to the beneficial effects of JAK inhibitors in RA.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Ivashkiv had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Yarilina, Ivashkiv.

Acquisition of data. Yarilina, Xu, Chan.

Analysis and interpretation of data. Yarilina, Xu, Chan, Ivashkiv.

Acknowledgements

We thank the patients and physicians of the Hospital for Special Surgery for providing synovial fluid samples, as well as K.-H. Park-Min and B. Zhao for critical review of the manuscript.

Ancillary