SV129 EV mice and FoxP3DTReGFP mice were originally from B and K Universal, North Humberside, England and Jackson Laboratories, ME, USA, respectively. CD4-Cre^+/−IFNAR flox/flox and CD4-Cre^−/−IFNAR flox/flox were received as a kind gift from Ulrich Kalinke, Twincore, Germany. Mice were further bred in the animal facility of Linkoping University, Sweden. Foxp3DTReGFP mice were bred heterozygously, and their offspring were genotyped for the mutant (Foxp3DTReGFP^+/−) allele or wild-type (WT) allele (Foxp3DTReGFP^−/−) by PCR using the primers 5′-CCCAGGTTACCATGGAGAGA-3′ and 5′-GAACTTCAGGGTCAGCTTGC-3′ for the mutant allele, and 5′-CAAATGTTGCTTGTCTGGTG-3′ and 5′-GTCAGTCGAGTGCACAGTTT-3′ (Invitrogen) for the WT allele according to the protocol for this strain provided by Jackson Laboratories at https://www.jax.org/strain/011003. Littermates without the mutant allele (WT) were used as control mice. All experimental procedures were performed according to the guidelines provided by the Swedish Animal Welfare Act and approved by the Ethical Committee Board, in Linköping (Dnr 12-01) and Stockholm (N271-14).

Arthritis was induced in female mice (8–13 weeks of age) as previously described (18) (Figure 1). Briefly, female mice were immunized subcutaneously with 200 µg methylated Bovine Serum Albumin (mBSA, Sigma-Aldrich) emulsified in Freund’s incomplete adjuvant (Sigma-Aldrich), with or without 1,000 U of IFN-α (IFNαA, PBL Assay Science, Piscataway, NJ, USA) or 15 mg/kg kynurenine (Sigma-Aldrich, St. Louise, MO, USA) also combined in the emulsion. One week later, a booster immunization of 100 µg mBSA emulsified in Freund’s incomplete adjuvant, with or without 1,000 U IFN-α or 15 mg/ml kynurenine, was injected at the base of the tail. On day 21, the mice were rechallenged with an intra-articular injection of 20 µl of mBSA solution (1.5 µg/µl) in the left knee joint. As a control, 20 µl of PBS was injected in the right knee joint. Mice were sacrificed on day 28, and knee joints were separated, fixed with 4% paraformaldehyde, decalcified and cut into sagittal sections (4 µm). Each section was stained with eosin and hematoxylin (Sigma-Aldrich) and the severity of arthritis was assessed with scores (0–3) on the basis of cellular infiltration in the synovial cavity, synovial layer thickening, and cartilage and bone erosion as earlier described (6). The score 0 represents no inflammation whereas score 1–3 represents increasing degrees of inflammation.

Foxp3DTReGFP mice [also called depletion of regulatory T-cells (DEREG)] were utilized to deplete the T_regs in vivo. These mice carry a diphtheria toxin (DT) receptor gene coupled to a green fluorescent protein gene, which is controlled by a Foxp3 promoter (19). Upon administration of DT, the Foxp3⁺ T_reg population can be depleted for a limited period in vivo without affecting other cell populations. The T_regs repopulate to the original amount after 7–10 days (19). Following several optimization experiments, we finalized a single i.p. injection of 250 µg DT in 100 µl PBS that can deplete up to 90% of Foxp3⁺ T_regs 2 days after DT injection (Figure S1 in Supplementary Material). Foxp3⁺ T_regs were depleted in separate experimental groups each receiving a single injection of DT at either day 0, day 5, or day 19 of AIA.

Blood was collected from the tail vein on days 0, 4, 10, 14, 20, 24, and 28 from mice during AIA and mixed with heparin to prevent coagulation. Spleens and a pool of draining lymph nodes (axillary, popliteal, and inguinal) were collected on days 0, 4, 10, 24, and 28 of AIA, from which single cell suspensions were prepared by gently crushing and passing the spleen and lymph nodes through a 70 µm nylon cell strainer and lysing the red blood cells with an RBC lysing solution (Sigma-Aldrich, Dusseldorf, Germany). The single cell suspensions or 100 µl of heparinized blood were surface stained with rat anti-mouse CD4 FITC antibody (Clone GK1.5, BD Biosciences, San Jose, CA, USA) and rat anti-mouse CD25 PE antibody (Clone PC61.5, eBioscience, San Diego, CA, USA). The cells were then fixed and permeabilized with a Foxp3-staining set (eBioscience, USA) according to the manufacturer’s instructions and stained intra-cellularly with rat anti-mouse Foxp3 APC antibody (Clone FJK-16s, eBioscience, USA). Analysis was performed with FACS Gallios (Beckman Coulter, Inc., Brea, CA, United States), and collected data were analyzed with Kaluza^® Flow Analysis Software, Beckman Coulter (version 1.5). The percentages of Foxp3⁺ T_regs among gated CD4⁺ cells were determined by FMO gating as earlier described (20).

A pooled single cell suspension of splenocytes and lymph node cells prepared as described above were re-suspended in Iscove’s Modified Dulbecco’s Media (Sigma-Aldrich) supplemented with 10% heat inactivated fetal bovine serum, 4 mM glutamine, 50 µM β-mercaptoethanol, 100 U/ml penicillin, and 0.1 mg/ml streptomycin (Sigma-Aldrich). 2 × 10⁵ cells of leukocytes were cultured for 72 h and stimulated with 50 µg/ml mBSA or 1 µg/ml anti-CD3 antibodies in the presence or absence of 500 U/ml of IFN-α. After 72 h, cells were harvested and analyzed for CD4, CD25, and Foxp3 expression as described above.

Indoleamine 2,3-dioxygenase was blocked using DL 1-MT (Sigma-Aldrich), which was prepared in drinking water (5 mg/ml) as described previously (8). The 1-MT drinking solution was administered 2 days before the first immunization until day 20 of AIA.

CD4⁺ T cells from single cell suspensions prepared from days 4, 10, 20, and 28 of AIA were sorted using magnetic bead based MACS technology (CD4⁺ T Cell Isolation Kit, Miltenyi Biotec, Cologne, Germany). From the sorted CD4⁺ cells, CD4⁺CD25⁻ cells, and CD4⁺CD25^+high cells were further sorted by FACS Aria using rat anti-mouse CD4 PE (Clone GK1.5, BioLegend, San Diego, CA, USA) and rat anti-mouse CD25 APC antibodies (Clone PC61, BioLegend, USA). For suppression assays, CD⁺CD25⁻ cells and CD4⁺CD25^+high cells were used as responder T cells (T_resp cells) and T_reg, respectively. CFSE-stained T_resp cells from IFN-α-treated or -untreated mice at days 4, 10, 20, and 28 of AIA were prepared as described (21). 50,000 CFSE-stained T_resp cells isolated from IFN-α-treated or -non-treated mice were cocultured in a 96-well round-bottom culture plate with decreasing number of T_regs isolated from IFN-α-treated (with or without 1-MT in drinking water) or -non-treated mice in the T_reg:T_resp ratio of 1:2, 1:4, 1:8, and 0:1 (T_resp cells only). In each well, 100,000 irradiated splenocytes as antigen presenting cells from naïve mice were added. Cultures were stimulated either with anti-CD3 (1 µg/ml) or mBSA (50 µg/ml). Cells were harvested after 72 h and analyzed by FACS Gallios, and the suppressive capacity of T_regs against T_resp cell proliferation was determined as previously described (22). The proliferation of CFSE-stained T cells was calculated based on CFSE MFI of proliferating cells and non-proliferating cells by Kaluza software. The suppression percentage was calculated as [(T_resp proliferation without T_regs) − (T_resp proliferation with T_regs)]/(T_resp proliferation without T_regs) × 100.

Spleens and draining lymph nodes (axillary, popliteal, and inguinal) were collected at day 20 of AIA from IFN-α-treated or -non-treated mice. CD4⁺CD25^+high T_regs were isolated from combined splenocytes and lymph node cell suspensions as described earlier. 50,000 or 250,000 T_regs in a volume of 100 µl PBS were intravenously injected in the tails of mice at day 20 of AIA. Arthritis was induced in these recipient mice by intra-articular injection of mBSA as described earlier on day 21 (1 day after transfer). The mice were sacrificed at day 28, and arthritis severity was evaluated as described earlier.

Differences in arthritis severity between groups with different treatments were measured by Mann–Whitney U test. Student’s t-test was used to compare the suppressive capacity of T_regs from PBS or IFN-α-treated mice (with or without in vivo 1-MT). A p value <0.05 was considered significant. All statistical tests were performed using GraphPad Prism 7.02.

Previously, we have shown that IFN-α protects against AIA (6, 8). Here, we wanted to determine if protection rendered by IFN-α in AIA is dependent on T_regs. To this end we used DEREG mice where the Foxp3⁺ population can be transiently depleted in vivo by DT. DT was administered once at day 0, day 5, or day 19 during AIA resulting in a 90% depletion of Foxp3⁺ in blood 2 days after DT injection (Figure S1 in Supplementary Material). DT injection to wt littermate mice (Foxp3DTReGFP^−/−) did not result in T_reg depletion (data not shown). The effect of T_reg depletion on arthritis in IFN-α-treated and control DEREG mice was evaluated on day 28 of AIA. In DEREG mice where the DT was administered at day 0 or 5 (sensitization phase of AIA), IFN-α could still significantly mitigate arthritis as compared with control DEREG mice receiving DT day 0 or day 5, respectively (Figure 1B). However, in DEREG mice where DT was administered at day 19 (i.e., shortly before arthritis induction by intra-articular injection of mBSA), the protective effect of IFN-α was totally abolished (Figure 1A), i.e., not significantly different compared with control DEREG mice receiving DT day 19 (Figure 1B) and significantly different from IFN-treated DEREG mice without depletion. This indicates that T_regs are critically required for IFN-α-protection in the arthritis phase of AIA (Figures 1B,C).

We analyzed the percentage of T_regs (Foxp3⁺ cells) among CD4⁺ cells in blood, spleens, and draining lymph nodes collected from IFN-α-treated and non-treated mice at different days during AIA. The percentage of Foxp3⁺ T_regs in IFN-α-treated mice did not differ significantly from percentages in non-treated mice in the blood at any day (Figure 2A). Likewise, no significant difference was observed in spleen or lymph nodes, although the total number of Foxp3⁺ T_regs increased in lymph nodes from day 0 to day 28 (Figure 2A). To test whether IFN-α affected the function of T_regs, we performed T_reg suppression assays. The ex vivo suppressive capacity of T_regs (CD4⁺CD25^+high) isolated at days 4, 10, 20, and 28 of AIA from PBS (control) or IFN-α-treated mice was tested against proliferation of T-responder cells (T_resp cells, CD4⁺CD25⁻), isolated at the same days of AIA either from control mice, or from IFN-α-treated mice. For the suppressive assay, 50,000 CFSE-stained T_resp cells were cocultured with APCs and decreasing numbers of T_regs (T_reg:T_resp ratios 1:1, 1:2, 1:4, and 1:8) and stimulated with mBSA or anti-CD3 for 72 h. The suppressive capacity of T_regs was thereafter evaluated by flow cytometry by comparing the proliferation of T_resp cells in cocultures relative to T_resp cells without T_regs. As depicted in Figure 2B, the more T_regs that are present in the cultures, the higher the suppression (see, e.g., top panel in Figure 2B: T_reg:T_resp ratio 1:2 vs T_reg:T_resp ratio 1:4 or 1:8). No significant differences were observed in the suppressive capacity of T_regs isolated at day 4 between PBS and IFN-α-treated mice (Figure 2B, day 4). T_regs isolated at day 10 and day 20 from IFN-α-treated mice during AIA exhibited clearly higher suppression of antigen-specific (mBSA)-induced proliferation than T_regs from PBS-treated mice. From day 20 and onward, an enhanced suppressive capacity of T_regs isolated from IFN-α-treated mice was in fact also observed against polyclonal (anti-CD3) stimulation of T_resp cells (Figure 2B, day 20 and day 28). The increased suppressive capacity of T_regs from IFN-α-treated mice was also observed on proliferation of T_resp cells isolated from IFN-α-treated mice (Figure S2 in Supplementary Material).

Three days after the second immunization (day 10) we observed an increased suppressive capacity of T_regs from IFN-α-treated mice (Figure 2B, day 10). This was observed when T_resp cells were restimulated with the immunizing antigen (mBSA), but not when the same T_resp cells were stimulated with the polyclonal activator anti-CD3 (Figure 2B, day 10). To explore this antigen requirement further, we counted the number of T_regs generated in vitro by IFN-α in the presence or absence of antigen-specific restimulation. Pooled spleen and lymph node cells, either from naïve mice or from mice immunized once or twice with the antigen (mBSA) were restimulated for 72 h with mBSA or anti-CD3, with or without 500 U of IFN-α. No effect of the mBSA antigen or IFN-α on the quantity of T_regs was observed in naïve leukocytes when stimulated with anti-CD3 or mBSA in the presence of IFN-α (Figure 2C). In contrast to naïve cell cultures, FoxP3⁺CD4⁺ T_regs were significantly increased by IFN-α in cultures using cells from mice once or twice immunized with mBSA, when restimulated ex vivo with mBSA + IFN-α (Figure 2C). The IFN-α-mediated increase of T_regs was not observed when cells from mBSA-immunized mice were stimulated with anti-CD3 antibodies plus IFN-α (Figure 2C). Thus, IFN-α increases the number of T_regs in vitro, but only in the presence of antigen in already antigen-sensitized cells.

As T_regs are crucial for protection against arthritis by IFN-α, we investigated whether protection is a direct effect of IFN-α on T helper cells. To this end the protective effect of IFN-α against AIA was tested in CD4-Cre^+/− IFNAR flox/flox mice where the IFN-α receptor (type I IFN receptor) is knocked out in T cells (23). When treated with IFN-α, mice lacking the type I IFN receptor on CD4⁺ T cells were equally protected against AIA by IFN-α compared with their WT counterpart (CD4-Cre^−/− IFNAR flox/flox), as depicted in (Figure 3).

Because IDO is implicated in the generation of T_regs (24, 25) we tested whether IDO contributed to the increased suppressive function conferred by IFN-α on T_regs during AIA. To this end, we performed the same ex vivo suppressive assay with T_regs from IFN-α-treated mice as described in Figure 2B, but in which the IDO inhibitor 1-MT was administered during the sensitization phase of AIA as earlier described in Ref. (8). Presence of 1-MT during Ag sensitization did not impair the suppressive function of T_regs in general, but the enhancing effect of IFN-α on the suppressive capacity of T_regs (Figure 2B) was clearly hampered if mice were treated with the IDO inhibitor 1-MT (Figure 4A). This was apparent both in T_regs isolated in the sensitization phase (day 10 of AIA) as well as in T_regs isolated after induction of arthritis (day 28 of AIA). In line with the abolished increased suppressive capacity of T_regs, IFN-α + 1-MT-treated mice were not protected against arthritis as were mice receiving IFN-α + vehicle (Figure 4B) and as earlier reported (8).

Thus, one way by which IFN confers a better suppressive capacity upon T_regs is by activating the enzymatic activity of IDO, which converts tryptophan to kynurenine. To test whether kynurenine also employs T_regs to prevent arthritis, Kyn was administered at immunization in AIA to DEREG mice with or without depletion of T_regs by DT. As depicted in Figure 4C, Kyn readily protected DEREG mice from arthritis, in line with our earlier observations (8). When mice were treated with DT on day 19, resulting in a potent but transient depletion of Foxp3⁺ T_regs (Figure 4D), the protective effect was abolished (Figure 4C). Thus, as for IFN-α (Figure 1A) T_regs are required for the protective effect of Kyn against AIA.

Knowing that IFN-α treatment in vivo can enhance the ex vivo suppressive activity of T_regs, we tested the ability of such T_regs to protect against AIA. To this end we isolated T_regs from mice immunized with mBSA with or without IFN-α (20 days after the first immunization). 50,000 FACS- sorted T_regs (CD4⁺CD25^+high) were intravenously transferred to recipient mice subjected to AIA (day 20 of AIA). Arthritis was induced in recipient mice one day after transfer (day 21) by intra-articular injection of mBSA, and the severity of arthritis was evaluated at day 28. Transfer of 50,000 T_regs from IFN-α-treated mice clearly prevented development of arthritis whereas the same number of T_regs from PBS-treated mice did not (Figure 5). To verify that T_regs from PBS-treated mice were still functional we transferred a five times higher dose of T_regs to mice subjected to AIA. Adoptive transfer of 250,000 T_regs from mBSA-immunized mice, irrespectively of whether IFN-α was included at immunization or not, to recipient mice subjected to AIA prevented arthritis development (data not shown). Thus, IFN-α confers a suppressive capacity to T_regs, so that an inferior number of T_regs are sufficient to prevent arthritis development as compared with T_regs generated in the absence of IFN-α.