A Broad Blockade of Signaling from the IL-20 Family of Cytokines Potently Attenuates Collagen-Induced Arthritis

Rheumatoid arthritis (RA) is a chronic and systemic inflammatory disease with cartilage destruction, severe synovitis, and leukocyte infiltration (1). The pathogenesis involved in the development of RA has been widely studied, yet the precise molecular mechanism underlying the disease remains to be fully elucidated (2). Previous studies have revealed pathogenic roles of proinflammatory cytokines such as IL-6, IL-17, TNF-α, and the IL-20 subfamily cytokines in RA (3–8). Although biologic drugs targeting TNF-α have been in clinical use in treating RA for >15 y, nearly one-third of the RA patient population become refractory to the drug class (7).

IL-19, IL-20, and IL-24 are members of the IL-20 subfamily of cytokines (8). These three cytokines share a common β receptor subunit IL-20R2 with all three being able to signal through IL-20R1/IL-20R2, whereas IL-20 and IL-24 can also signal through IL-22R1/IL-20R2 heterodimeric receptor complex (9–11). Such significant receptor sharing among these cytokines has raised questions about whether the three cytokines have redundant biological functions in vivo or whether they may use the same receptors for different biological endpoints in a tissue-specific or temporally regulated manner (12). Substantial evidence supports the linkage between the development of autoimmune diseases—such as RA, psoriasis, osteoporosis, and atopic dermatitis—and the abnormal overexpression of IL-19, IL-20, and IL-24 (8, 13, 14), with synovial tissues having been identified as a major source of the three cytokines in patients with RA (8, 13–15). A neutralizing Ab against IL-20 was shown to be able to attenuate disease severity of collagen-induced arthritis (CIA) in rats (16). Similar efficacy was also reported for anti–IL-19 mAb in the same animal model (13), with IL-24–specific blockers yet to be tested. Although in these studies, both anti–IL-20 and anti–IL-19 mAbs exhibited similar efficacy to that of etanercept in reducing disease symptoms in a rat CIA model, combination therapy for simultaneously blocking IL-20 and anti–IL-19 signaling has yet to be studied and would be expected to result in a better outcome. In contrast to complete failure for treating psoriasis in human clinical trials (17), anti–IL-20 mAb has demonstrated some moderate therapeutic effect against RA in a phase 2 clinical trial, although seen as inferior to anti-TNF biologics (18). Given the promiscuous nature of receptor sharing for the IL-20 subfamily of cytokines, it would be clearly advantageous to develop a biologic therapy that could simultaneously block signaling of all three cytokines that signal through IL-20R2.

Previous studies from our laboratory showed that IL-24 had overlapping functions with IL-20 and IL-22 in the epidermis (19), and IL-20R2 was directly involved in ligand binding (9, 19). Partially purified and characterized recombinant soluble IL-20R2-Fc fusion proteins were shown to have variable specificity and potency toward its ligands and failed to bind to IL-20 in cell-based assays (11, 19). In this study, we demonstrate that a highly purified recombinant human soluble IL-20R2-Fc (rhIL-20R2-Fc) fusion protein can directly bind to IL-19, IL-20, and IL-24 with similar high-affinity in vitro and potently inhibits these individual ligands in Ba/F3 cell–based receptor signaling in vitro as well as CIA in DBA/1 mouse model in vivo. Furthermore, we show that the same set of immune infiltrate cells congregating in large numbers in the disease tissues of RA are not only positive for TNFR expression but also are producing ligands that are involved in IL-20R2 signaling.

rhIL-20R2-Fc expression vector (19) was stably transfected into GH-Chinese hamster ovary (CHO) (dhfr^−/−) cell line (GenHunter, Nashville, TN) using Fugene 6 (Roche) grown in IMDM with 10% FBS and 1% penicillin-streptomycin supplemented with HT (Sigma-Aldrich). After stepwise gene amplification with increasing concentrations (0–1 μM) of MTX (Sigma-Aldrich), the clone with highest rhIL-20R2-Fc titer assayed by ELISA with protein A capture (5 μg/ml) and detection by IL-24-alkaline phosphatase (AP) was obtained. The cells were then adapted to SFM-4-CHO (Hyclone) serum-free medium, and rhIL-20R2-Fc was produced in a 14-l Celligen bioreactor (Eppendorf) under Fed-batch process with CellBoost 5 (Hyclone) added every other day from day 3 until harvest. rhIL-20R2-Fc titer was monitored daily with a Bio-Monolith Protein A column (Agilent) and SDS-PAGE. After 13–16 d culturing in bioreactor, rhIL-20R2-Fc was purified to homogeneity from the conditioned medium using MabSelect Sure (GE Health Sciences) and Superdex 200 (GE Healthcare), according to the manufacturer’s instructions. Briefly, Mabselect Sure affinity chromatography was used to capture the soluble receptor-Fc fusion protein. A total volume of 500 ml of culture supernatant was loaded on a 25-ml MabSelect column. After washing off unbound impurity with 10 column volumes of buffer A (20 mM phosphate buffer + 0.15 M NaCl [pH 7.2]), 1 M Tris-HCl (pH 10) was added to adjust the elution sample to pH 7.2. Superdex 200 prep grade size-exclusion chromatography (SEC) (XK16/100 column with 150-ml bed volume) was employed to further purify the protein using buffer C (0.1 M PBS [pH 7.1]) with a flow rate at 1 ml/min. The purity of rhIL-20R2-Fc was determined by SEC-HPLC (Tosoh Bioscience, Shanghai, China). Purified rhIL-20R2-Fc was ultra-filtrated into PBS before used for biological assays.

Human IL-19-AP, IL-20-AP, and IL-24-AP were described previously (19, 20). The soluble human IL-20R2 coding region was obtained as a HindIII and BglII fragment from the plasmid encoding rhIL-20R2-Fc (19) and subcloned into pAP-Tag2D expression vector (GenHunter) to allow in-frame fusion of the soluble receptor to human placental AP. The recombinant plasmid was then transformed into GH2 competent cells (GenHunter), and positive clones were verified by colony PCR using primers: L-AP (5′-GAACCCACTGCTTACTGGC-3′) and R-AP (5′-GCCTCGCGGTTCCAGAAG-3′) from GenHunter. After confirmation by DNA sequencing, the IL-20R2-AP expression vector was then stably transfected into GH-CHO (dhfr^−/−) cell lines (GenHunter). IL-20R2-AP fusion protein from the SFM-4-CHO serum-free conditioned medium was used directly for the ligand binding studies. TNF-α-AP and human soluble TNF-RII-Fc fusion protein were obtained from GenHunter.

One unit of conditioned medium from 293T or CHO cells expressing IL-19-AP, IL-20-AP, IL-24-AP, IL-20R2-AP, and human placental AP were analyzed by Western blot on a 10% SDS-PAGE using anti-AP rabbit polyclonal Ab (GenHunter), followed by goat anti-rabbit IgG-HRP (Southern Biotechnology Associates) to verify each AP-Tagged ligands (19). For in situ ligand-receptor affinity staining, 2 μg of either purified rhIL-20R-Fc (rhIL-20R2-Fc) or TNFRII-Fc (rhTNFR-Fc) fusion protein (GenHunter) were analyzed by SDS-PAGE under nonreducing condition, followed by Western blot transfer to a polyvinylidene difluoride (PVDF) membrane (GenHunter). The blots were then probed with 1 U/ml of either AP alone or each corresponding ligand-AP fusion proteins, followed by visualization with AP Assay Reagent S (GenHunter). Solid-phase ELISAs for quantitative binding of IL-20 family of ligand-AP fusion proteins to rhIL-20R2-Fc were carried out by first coating the wells with 5 μg/ml protein A (Prospec), followed by 100 μl of 2 μg/ml rhIL-20R2-Fc. One hundred microliters of each corresponding AP fusion proteins at various concentrations were then loaded, and ligand-AP/rhIL-20R2-Fc complexes were detected at OD405 with AP assay reagent A (GenHunter), according to manufactures instruction. Estimated K_ds were concentrations of AP-tagged ligand, where half of the receptors were saturated.

For cell proliferation assays, an IL-3–dependent Ba/F3 stable cell line [Ba/F3 (IL-22R1/IL-20R2)] expressing IL-22R1/IL-20R2 receptor complexes (21) was first washed free of IL-3 with RPMI 1640 medium and then seeded at 5000 cells/well in 6-well plates and cultured in RPMI 1640 medium containing 10% FBS and 1% penicillin-streptomycin with either 1 U/ml (1 μg/ml) of IL-19-AP, IL-20-AP, or IL-24-AP in the presence or absence of 100 μg/ml rhIL-20R2-Fc. Mouse IL-3 (Prospec) at 1 ng/ml was used as a control for the target specificity for rhIL-20R2-Fc. Cell densities were determined at different time points using Vi-Cell XR Cell Counter (Beckman Coulter).

For phosphor-Stat analysis, Ba/F3 (IL-22R1/IL-20R2) cells were deprived of IL-3 for 24 h and starved in RPMI 1640 medium containing 0.5% FBS at 37°C for 6 h. Cells were then stimulated with 1 U/ml IL-19-AP, IL-20-AP, or IL-24-AP, respectively, in the presence or absence of rhIL-20R2-Fc (100 μg/ml) for 30 min. Mouse IL-3 (Prospec) at 1 ng/ml in the presence or absence of rhIL-20R2-Fc (100 μg/ml) was used as a control for the target specificity for rhIL-20R2-Fc. Cells (1 × 10⁶) were washed free of RPMI 1640 medium with PBS and lysed in 1% Triton X-100 buffer containing 1% Triton X-100, 50 mM Tris-HCl, 200 mM NaCl, 3 mM Na₃VO₄, 5 mM EDTA, 5 mM NaF, and 1 mM PMSF (Sigma-Aldrich). Total cell lysates were separated by a reducing SDS-PAGE. STAT3 activation was analyzed by Western blot using Ab specific for phosphorylated STAT3 (catalog number 9145; Cell Signaling Technology, Beverly, MA), according to the manufacturer’s instructions. β-Actin was used as a loading control.

Three- to five-week-old male DBA/1 mice were purchased from Beijing HFK Bioscience and kept under standard pathogen-free conditions in the animal care center at Sichuan University and received humane care. All animal experiments were approved by a state-appointed board on animal ethics and were conducted according to international guidelines for animal experimentation. For pharmacokinetics study, three mice were dosed at 30 mg/kg with rhIL-20R2-Fc via i.p injections. Sera samples were collected from mouse tails at 0, 0.17, 0.5, 1.5, 2, 4, 8, 16, and 24 h after drug administration. Serum drug level was determined by ELISA as described above. Ke was evaluated by least square regression of the points describing the terminal log-linear decaying phase (T_1/2 = ln2/Ke). The areas under rhIL-20R2-Fc concentrations versus time curves from 0 to 24 h (AUC_{0–24 h}) were calculated by GraphPad prism 6.0 software.

For pharmacodynamics study, the established mouse model for CIA was used following the protocol with two immunizations and treatment regimens (22). Briefly, 6- to 8-wk-old male DBA/1 mice were immunized intradermally each at the base of tail with 150 μg of bovine type II collagen (Condrex) in CFA (Condrex). The mice were boosted after 3 wk with 100 μg of bovine type II collagen in IFA (Condrex) to initiate the CIA. One week after the second boost injection when animals began manifest mild swollen digits, mice were randomly divided into four treatment groups and one healthy control group (n = 6). Each treatment group received 200 μl of PBS, purified rhIL-20R2-Fc in 200 μl of PBS (12.5 mg/kg) every 16 h or rhTNF-RII-Fc (etanercept; Amgen) in 200 μl of PBS (12.5 mg/kg) every 48 h or rhIL-20R2-Fc plus rhTNF-RII-Fc (12.5 mg/kg of each protein) at the same intervals for each, respectively, all via i.p injections. Mice were monitored daily over a 6-wk period from the initial immunization for signs of arthritis. Standard scoring system for the severity of arthritis in each hind paw was followed with 0, no swelling or redness; 1, slight ankle swelling or redness in the toe; 2, progressive swelling and redness in the joints; 3, severe swelling and redness of the entire foot, including the toes; and 4, severe swelling and redness of soft joint tissue and joint stiffness. Each limb was graded, giving a maximum possible score of 16 per mice. The severity of arthritis in each mouse was determined blindly and independently by four observers, and average of scores was counted. Statistical analysis of daily average arthritis scores of five groups was expressed as mean ± SEM. Experimental data between groups were statistically analyzed by on-one ANOVA, and differences between groups were compared with LSD test, and p < 0.05 was considered significant. Statistics were performed using SPSS software. Infrared thermography of the hind paws of the CIA mice on day 21 was captured with a FLIR T600 Thermal Imaging Camera. Regional temperature was analyzed by XJ-Infrared report analysis software. Images of radiographs of the hind paws from the CIA mice on day 21 were captured with a Kodak Point-of-Care CR140 medical X-ray system. Standard scoring system for the severity of bone destruction in the ankle joint (16) was followed with 0, no swelling or bone damage; 1, slight ankle swelling or joint space narrowing; 2, moderate bone erosion and ankle swelling; and 3, severe bone erosion. The severity of arthritis in each mouse was determined blindly and independently by four observers, and average of scores was calculated.

rhIL-20R2-Fc was biotinylated with NHS-PEG4-biotin kit (Thermo Fisher Scientific) following the instruction of the manufacturer. Unbound NHS-PEG4-biotin was removed via ultra-filtration in PBS. The protein concentration of biotinylated rhIL-20R2-Fc was determined by Pierce BCA Protein Assay Kit (Thermo-Fisher Scientific).

Hind paw samples from the mouse model of CIA were fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned at 5 μm thickness. For in situ ligand-receptor affinity tissue staining, the sections were rehydrated with ethanol of gradient concentration and stained with 1 U/ml of corresponding AP fusion proteins or AP alone as previously described (19, 23), whereas 10 μg/ml of corresponding Fc fusion proteins were included during AP-fusion protein binding in competition experiments. For double staining, the slides were sequentially stained for biotin-labeled rhIL-20R2 and TNF-α-AP following the strategy described previously, except paraffin-embedded instead of cryosections (24). In brief, sections were rehydrated with ethanol of gradient concentration. Nonspecific binding was blocked using 0.01% avidin (Procept) at room temperature for 20 min. The CIA tissue sections were first incubated with biotinylated rhIL-20R2 at 20 μg/ml for 1 h, followed by signal detection using 1:5000 dilutions of HRP-labeled streptavidin (Jackson ImmunoResearch Laboratories) and a standard diaminobenzidine kit (Vector Laboratories). After image capture for rhIL-20R2 binding signals, the same slides were incubated with 1 U/ml TNF-α-AP for 1 h, followed by signal detection using AP Assay Reagent S (GenHunter). All tissue-staining images were captured with an Olympus BX53 upright microscope.

Hind paw samples from the mouse model of CIA were fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned at 5 μm thickness. The sections were deparaffinized, rehydrated with ethanol a gradient, and placed in 0.1 M citrate buffer Ag retrieval solution (pH 6) for 18 min. Endogenous peroxidase activity was quenched with 0.3% hydrogen peroxide, and nonspecific Ab binding was blocked using 2% BSA for 30 min. The slides were first incubated with a macrophage specific rat anti-mouse F4/80 mAb (catalog number NBP1-60140; Novus, Littleton, CO) at 1:200 dilution for overnight, followed by signal detection in brown color with a corresponding HRP-labeled secondary Ab (catalog number ZB-2307; ZSGB-BIO, Beijing, China) and the diaminobenzidine substrate kit (Vector Laboratories) according to manufacturer’s instructions. After image captures for F4/80 binding signals, the same slides were incubated with 1 U/ml TNF-α-AP for 1 h, followed by signal detection using AP Assay Reagent S (GenHunter) in blue color. All tissue-staining images were captured with an Olympus BX53 upright microscope.

On the basis of our earlier findings that IL-20R2 alone, when expressed either on the surface of transfected Cos-1 cells or as soluble receptor-Fc fusion, could directly bind to IL-24 (9, 19), we had predicted that rhIL-20R2-Fc might also bind to IL-19 and IL-20 and thus serve as a better biologic blocker against the IL-20R–mediated signaling than any Abs to individual cytokines. However, previous indirect binding assays based on the neutralization of either ligand-mediated signal transduction (11) or receptor binding (19) resulted in inconsistent findings. It seemed from these studies that rhIL-20R2-Fc could not bind to IL-20, while its ability to block IL-19 and IL-24 differed in potency depending on which assays were used. To better determine and explore the biochemical and therapeutic functions of rhIL-20R2-Fc, we screened for high-titer production clones of transfected CHO cells via MTX-mediated gene amplification. The resulting leading clone was adapted to serum-free medium and fed-batch cell culture from bioreactor led to high-level expression of rhIL-20R2-Fc close to 2 g/l (Fig. 1A, 1B). rhIL-20R2-Fc was purified to near homogeneity via protein A–based affinity chromatography followed by gel filtration (Fig. 1C, 1D).

To determine whether the rhIL-20R2-Fc fusion protein could bind to the three ligands in the IL-20 subfamily, we used AP-tagged human IL-19, IL-20, and IL-24 (hIL-19-AP, hIL-20-AP, and hIL-24-AP) (9, 19) as probes to conduct in situ ligand-receptor affinity binding. After confirming the expression of hIL-19-AP, hIL-20-AP, hIL-24-AP, and AP control as secreted proteins (Fig. 2A), these AP-Tag ligands were each tested for in situ receptor binding after SDS-PAGE separation and transfer of the purified rhIL-20R2-Fc fusion protein to a PVDF membrane. The result indicated that under nonreducing condition, the rhIL-20R2-Fc could bind to all three AP-tagged cytokines, whereas AP alone failed to do so (Fig. 2B). As a control, a recombinant human soluble TNFR2-Fc fusion protein was also included for the in situ binding assay, which confirmed the specificity of the ligand-receptor interactions. To more accurately determine the binding affinity of each AP-Tagged ligand to rhIL-20R2-Fc, we then conducted quantitative ELISA by first capturing the soluble receptor-Fc fusion protein via protein A coating. Then increasing concentration of each AP-Tagged ligand was allowed to saturate the receptor and the results indicated that hIL-19-AP, hIL-20-AP, and hIL-24-AP all bound to rhIL-20R2-Fc with similar saturation kinetics with an estimated K_d of ∼10 nM (Fig. 2C, 2D).

To further demonstrate that the rhIL-20R2-Fc could inhibit the cell signaling from each of the three IL-20 family of cytokines in vitro, we used a Ba/F3 cell line stably expressing IL-22R1/IL-20R2 (21). We showed that rhIL-20R2-Fc at 100 μg/ml (∼100 times excess to ligands) could completely block both IL-20-AP– and IL-24-AP–mediated cell proliferation while having little effect on IL-3–dependent growth (Fig. 3A). Although it has been reported that IL-19 could not signal through IL-22R1/IL-20R2, we observed that IL-19-AP supported a residual cell survival and growth in this cell line, which was also completely abolished by rhIL-20R2-Fc.

To further elucidate the mechanism of rhIL-20R2-Fc in inhibiting the cell signaling from each of the three IL-20 family of cytokines in vitro, STAT3 activation was assessed in ligand-stimulated Ba/F3 (IL-22R1/IL-20R2) cells in the presence or absence of rhIL-20R2-Fc. The result showed that treatment of IL-3–starved cells with each of the three IL-20 family of cytokines led to rapid phosphorylation of STAT3, whereas the activation, but not by IL-3, could be essentially blocked by rhIL-20R2-Fc (Fig. 3B).

Although abnormal activation in signal transduction pathways via TNFR and IL-20R2 has been both clearly linked to the pathogenesis of RA, the nature of the interplay and cross-talk between these two signal transduction pathways remain obscure. Past analysis of tissue expression of these cytokines and their receptors in RA from both human patients and animal models has relied on the immunohistochemistry using Abs that are prone to problems in target specificities. To more accurately evaluate the cell types involved in the production of IL-20 family cytokines and TNFR, paraffin-embedded disease tissues from CIA of DBA/1 mice were sectioned and analyzed by in situ ligand-receptor functional staining using AP-tagged soluble IL-20R2 (IL-20R2-AP) and TNF-α-AP. In comparison with normal tissues from control mice that had not been injected with collagen, disease tissues from the swollen paws of CIA mice exhibited a large number of immune infiltrates that were stained strikingly positive for both IL-20R2-AP and TNF-α-AP (Fig. 4A). To ensure that the signals from the in situ ligand-receptor binding assays were ligand-specific, TNF-α and rhIL-20R2-Fc were tested for their ability to compete for TNF-α-AP or IL-20R2-AP binding, respectively. The results indicated that the positive stainings could be largely extinguished by 10-fold excess of the corresponding competitors, supporting that TNF-α-AP ligand binding was receptor-specific and IL-20R2-AP receptor binding was ligand-specific (Fig. 4B).

Given the striking similarity in the pattern of both sets of immune infiltrates that populated the disease tissues, we next examined whether the expression of TNFR and IL-20 family of cytokines came from the same cell type or neighboring cells. To this end, we conducted a sequential ligand-receptor binding analysis of the CIA tissues by first using a biotin-labeled IL-20R2 (biotin-IL-20R2) followed by detection of streptavidin-labeled HRP to visualize the expression of the IL-20 subfamily of cytokines. This yielded a pattern of positively stained immune infiltrates (light brown color) similar to that of stained by IL-20R2-AP (Fig. 4C). After capturing the images, the same CIA tissue section was then probed with TNF-α-AP followed by visualization of bound AP activity using 5-bromo-4-chloro-3-indolyl phosphate substrate, which gave darker bluish-purple color. The results revealed that IL-20 family cytokines and TNF-α receptor were expressed in the same set of immune infiltrates as the two signals completely overlapped (Fig. 4C). This finding provides direct evidence that TNF and IL-20R signaling pathways seem to converge in the pathogenesis of rheumatoid arthritis. To identify the cell types that are IL-20R2 and TNF positive in the immune infiltrates, we have conducted a sequential costaining for anti-F4/80 (macrophage specific), followed by TNF-α-AP based on our previous work on IL-24 transgenic mice (19). Indeed, we have confirmed that these IL-20R2 and TNF-positive immune infiltrates were macrophages as we suspected because signals from anti-F4/80 completely overlapped with that of TNF-α-AP on the same CIA tissue slide (Fig. 4D).

Having demonstrated that rhIL-20R2-Fc can bind to all three ligands (IL-19, IL-20, and IL-24) from the IL-20 subfamily and that abnormal expression of these cytokines are evident from large numbers of infiltrating leukocytes within disease tissues of RA, we then set out to test whether rhIL-20R2-Fc could be therapeutically efficacious in the treatment of collagen-induced arthritis. To determine the dosing strategy, we first conducted a pharmacokinetic study of rhIL-20R2-Fc and showed that the half-life of rhIL-20R2-Fc in DBA/1 mouse was ∼5 h and AUC_{0–24 h} = 1.033 mg.h/ml (Fig. 5).

Treatment regimens were followed for the CIA model of DBA/1 mice. Treatment of established RA with rhIL-20R2-Fc injections (i.p.) every 16 h led to significant reduction in the severity scores of RA, compared with nontreatment control (p < 0.05) (Fig. 6). As a positive control, the mice were also treated with rhTNFRII-Fc (etanercept) every 48 h as well as a combination therapy with both rhIL-20R2-Fc and rhTNFRII-Fc. The results showed that rhIL-20R2-Fc and rhTNFRII-Fc gave similar efficacy in the treatment of established CIA, each being able to reduce the severity scores of RA by more than half compared with nontreated group (Fig. 6A, 6B). Moreover, combination therapy resulted in mild synergistic efficacy, although the benefit was not additive. The measurements in RA severity scores were corroborated with both infrared and radiography analysis, which supported the pharmacodynamic functions of rhIL-20R2-Fc in reducing the severity of joints damage for CIA mice (Fig. 6C–F).

It is known that the three cytokines IL-19, IL-20, and IL-24 from the IL-20 subfamily all depend on IL-20R2 receptor subunit for signaling as part of two heterodimeric receptors. Although IL-20R2 does not seem to be directly involved in signal transduction because of its short cytoplasmic domain, in contrast to with of IL-20R1 and IL-22R1 (12), it was shown to be directly involved in IL-24 binding (9, 19). To test whether IL-20R2 can also bind to the other two ligands in the family and be used as a broad antagonist for signaling through receptor complexes involving IL-20R2, we developed a CHO cell line and complementary serum free cell culture process from which rhIL-20R2-Fc can be produced at high level. As a result, large quantity of highly purified rhIL-20R2-Fc was obtained and tested for ligand binding. Using AP-tagged ligands, direct bindings of IL-19, IL-20, and IL-24 to rhIL-20R2-Fc were evident even with denatured soluble receptor-Fc fusion protein separated on an SDS-PAGE. ELISA-based assays provided a more quantitative analysis of the affinity for each cytokine to the native rhIL-20R2-Fc and showed that all three ligands bound to the soluble receptor with comparable high affinity in the nanomolar range. To further determine the specificity and efficacy of rhIL-20R2-Fc in blocking each individual cytokine from the IL-20 subfamily, we then used a Ba/F3 cell line with IL-22R1/IL-20R2 (21). This cell line can depend on IL-3 for growth. In the absence of IL-3, it can also proliferate in the presence of IL-20 and IL-24 (21). We showed that rhIL-20R2-Fc could completely block both IL-20– and IL-24–mediated cell proliferation while having little effect on IL-3–dependent growth. Although it has been reported that IL-19 could not signal through IL-22R1/IL-20R2, our data showed that IL-19 supported a residual cell survival and growth in this cell line, which was also completely abolished by rhIL-20R2-Fc. Anti–phospho-STAT3 analysis of these cells also confirmed that rhIL-20R2-Fc could specifically abolish the signaling from IL-19, IL-20, and IL-24 but not from IL-3, confirming the specificity of the targets. We believe that this important bioassay further establishes that rhIL-20R2-Fc can functionally block the receptor signaling via each individual cytokines from the IL-20 subfamily.

Although aberrant signaling from TNF and IL-20 subfamily cytokines have been linked to the pathogenesis of RA and anti-TNF biologics have been in clinical use, the nature of their biological functions and cross-talk in contributing to the disease progression remain obscured. Before testing rhIL-20R2-Fc in RA treatment using the mouse CIA model, we first used AP-tagged ligands and soluble receptors to examine the cell types that were involved in signal transduction from both TNF and the IL-20 subfamily of cytokines using paraffin-embedded disease tissues. Compared with immunohistochemistry using Abs, ligand and soluble receptor-AP fusion proteins in theory would offer a much better specificity in functional binding to their biological targets. When normal and disease tissues (hind paws) from the mouse CIA model were analyzed, one striking finding was that large numbers of disease-specific immune infiltrates were detected by both TNF-α-AP and IL-20R2-AP staining. This important finding suggests that these immune infiltrates are positive in expressing both TNFR and the IL-20 subfamily of cytokines. In contrast, decreased expression of both TNF-α and the receptor for IL-20 subfamily of cytokine were seen from the epidermis region of the affected digits when TNFR2-AP and IL-24-AP were used (data not shown). To find out whether TNF-α-AP and IL-20R2-AP detected in the same cell type in RA tissues, we then conducted sequential staining using biotin-labeled IL-20R2, followed by TNF-α-AP as a probe. The result indicated that the overexpression of TNFR and the IL-20 subfamily of cytokines in fact came from the same sets of immune infiltrates. Further sequential staining with an anti-F4/80 Ab and TNF-α-AP unmistakably identified these IL-20R2- and TNF-positive immune infiltrates as macrophages.

These important findings provide direct evidence that TNF and IL-20R2 signaling pathways converge on the pathogenesis of RA. It is tempting to hypothesize that the two pathways may function in tandem via macrophages as a relay system in amplifying or maintaining the inflammation where TNF-α acts as a signal input and its receptor activation leads to the production of the IL-20 subfamily of cytokines, which in turn may attract more leukocytes to the site of inflammation as shown in IL-24 transgenic mice (18). Consistent with this hypothesis, rhIL-20R2-Fc, which is able to block all three ligands from signaling through IL-20R2, was as efficacious as etanercept in the treatment of CIA in DBA/1 mouse model. Moreover, combinational therapy of etanercept with rhIL-20R2-Fc demonstrated little synergistic effect, supporting that the two pathways operate in tandem, rather than independently in the pathogenesis RA. The therapeutic effects seen are similar to that of individual mAbs targeting IL-19 and IL-20, each of which also exhibited similar efficacy as etanercept (13, 16, 25). Given that multiple cytokines from the IL-20 subfamily appear to be involved in the pathogenesis of RA in rodent CIA models and that anti-human IL-20 mAb previously yielded inferior therapeutic benefits to that of anti-TNF biologics in human clinical trial for RA (18), the development of rhIL-20R2-Fc as a broad inhibitor of the IL-20 subfamily of cytokines may provide a more effective blockade against signaling from IL-20R2, thus offering a potential new treatment option for RA and/or related diseases. Conceivably, similar goal may also be achievable using therapeutic mAbs against the IL-20R2 receptor subunit instead of individual ligands of the IL-20 subfamily of cytokines.

We thank Jamie Walden and Jonathan Meade from GenHunter Corporation for the proofreading of the manuscript.

The authors have no financial conflicts of interest.