Conditional Upregulation of IFN-α Alone Is Sufficient to Induce Systemic Lupus Erythematosus

Systemic lupus erythematosus (SLE) is a prototypical autoimmune disease presenting with erythematosus skin rash and characterized by the presence of serum autoantibodies, such as anti-dsDNA Ab. Previous studies have shown that IFN-α can be a principal driver of SLE (1–4). IFN-α was upregulated in sera and cerebrospinal fluids of patients with SLE (5–13). IFN-α–regulated gene transcripts, protein levels of some of IFN-α–induced chemokines, and the levels of IFN-α have been shown to be specifically upregulated in accord with disease activities of SLE (5–17). Occurrence of SLE or the rise of anti-dsDNA Ab has been reported in some patients treated with IFN-α and those with sporadic mutations in IFN-inducing genes (18–27). Autoimmunity in lupus-predisposed mice has been shown to be aggravated by upregulating IFN-α (28–32), and, instead, null mutation in the type I IFN receptor (IFNAR) gene of lpr background mice alleviated lymphoproliferation and immune complex (IC) deposition in the kidney (32). In the NZB mice lacking α-chain of IFNAR (IFNAR1), Abs reactive against erythrocytes, DNA, kidney disease, and mortality were significantly reduced (33). In the 129Sv/Ev mice lacking IFNAR1, autoantibodies, glomerular hypercellularity, and proteinuria did not develop after challenge with 2,6,10,14-tetramethyl pentadecane (34). Blockade of IFNAR by Ab reduced the disease activity of male BXSB mice and, transiently, of MRL-Fas^lpr mice (35), the findings indicating that IFN-α exacerbates preexisting autoimmunity. The findings together suggest the causal relationship between IFN-α and SLE.

However, there are also findings that run counter to the prevailing view that IFN-α causes SLE. In the congenic lupus-prone MRL/lpr mice lacking IFNAR1, lymphoproliferation, autoantibody production, and end-organ disease were rather aggravated (36). In the B6.Sle2 congenic and B57BL/6 mice, Ab-mediated blockade of type I IFN augmented serum autoantibody levels and boosted B1a cell numbers, and, instead, they were alleviated by administering type I IFN (37). In humans, therapy with humanized IgG monoclonal anti–IFN-α Ab showed no (38) or modest effects (39), and those with Abs against IFNAR were effective in less than half of the patients, in which those with high serum IFN-α levels mostly responded (40). There is no direct proof that dysregulation of IFN-α has a causal role in SLE. In the current study, to study the role of IFN-α in SLE, in particular, and to what extent and how IFN-α is responsible as a causative agent, we established a transgenic (Tg) mouse in which the IFN-α1 gene was selectively expressed in the lymphoid cells upon cessation of doxycycline (Dox) and studied the pathological details in this mice.

The study was approved by and performed in accordance with the regulations of the institutional review board, Kobe University (approval no. P091207), Kyushu University (approval no. A29-169-0), and the Institute for Rheumatic Diseases (approval no. A2017-001).

The mouse IFN-α1 (mIFN-α1) cDNA (bases 343–962, 620 bp, X01974.1; GenBank) was amplified by RT-PCR and inserted downstream of the TetO promoter (TetOp) of the pTet Splice vector (Invitrogen, Carlsbad, CA) to generate TetOp-mIFN-α1. TetOp-mIFN-α1 was microinjected into fertilized eggs of C57BL/6 mice to obtain TetOp-mIFN-α1 Tg mice. The mice were mated with EμSR-tTA Tg mice of FVB/N background, expressing the tetracycline transactivator (tTA) gene downstream of the IgH enhancer and SRα (EμSRα) promoter (41), resulting in lymphoid cell–specific expression of the EμSR-tTA gene in C57BL6/FVBN mice (B6/FVBN mice). The mice were also mated with the EμSR-tTA Tg mice of C57BL/6 background obtained after backcrossing EμSR-tTA Tg mice more than five generations to C57BL/6 mice (B6 mice). The mating produced double-Tg TetOp-mIFN-α1/EμSR-tTA (IFN-α Tg) mice of either C57BL6/FVBN (B6/FVBN) or C57BL/6 (B6) backgrounds. These mice were fed 50 μg/ml Dox (Sigma-Aldrich, St. Louis, MO) to maintain the expression of IFN-α being suppressed until the initiation of experiments (i.e., 4 wk after birth). The littermate of respective C57BL6/FVBN or C57BL/6 mice was used as a control. Known mutations either unique or common to both mice are shown in Supplemental Table I.

Serum IFN-α was measured using a μ-IFN-α ELISA kit (PBL Biomedical Laboratories, Piscataway, NJ). Autoantibodies in sera were measured by ELISA using plates coated with ssDNA (Sigma-Aldrich), dsDNA (Worthington Biochemical, Lakewood, NJ) digested by S1 nuclease (Promega, Madison, WI), or Sm Ag (Immuno Vision, Springdale, AR) as capture reagents. Serum IC was measured using solid-phase anti-C3 ELISA (42), which captured C3 to plate-coated goat anti-C3 Ab (Bethyl Laboratories, Montgomery, TX), and IgG included in C3-captured IC was measured by using HRP-conjugated anti-mouse IgG Ab (Kirkegaard and Perry Laboratories, Gaithersburg, MD). Pooled sera of 15- to 20-wk-old MRL/lpr female adult mice were used as positive controls (arbitrary units). Splenocytes were stimulated in vitro with plate-coated 2 μg/ml anti-CD3 (145-2C11; BioLegend, San Diego, CA) and 5 μg/ml anti-CD28 (37.51; BioLegend) Ab for 24 h. Cytokines in the culture supernatant or in sera or BAFF in sera were measured by ELISA (BioLegend; R&D systems, Minneapolis, MN; and eBioscience, San Diego, CA).

Proteinuria were measured semiquantitatively using urine dipsticks (Albustix, Berkeley, CA). Frozen kidney and skin sections were stained for C3 and IgG using immunofluorescent Abs. Abs used were anti-CD3 (Abcam, Cambridge, U.K.), anti-CD4 (Bioworld Technology, Minneapolis, MN), anti-CD8 (Epitomics, Burlingame, CA), anti–IFN-γ (PBL), and anti-B220 (BioLegend). Staining was performed as described (43) using Histofine Simple Stain MAX PO (Nichirei, Tokyo, Japan). Lupus kidney pathological condition was evaluated, being classified according to the human classification criteria of lupus glomerulonephritis (44).

FITC-conjugated Abs against CD3 (145-2C11), CD44 (IM7), CD80 (16-10A1), and TCRβ (H57-597); PE-conjugated streptavidin and Abs against CD62L (MEL-14), CD69 (H1.2F3), CD40 (1C10), TCRγδ (UC7-13D5), CD45R (B220; RA3-6B2), IL-17A (TC11-18H10.1), and TNF-α (MP6-XT22); PE-cyanin 5 (PE-Cy5)–conjugated Abs against CD86 (GL-1) and CD45R (RA3-6B2); PerCP complex–cyanin 5.5 (PerCP-Cy5.5)–conjugated Abs against CD4 (RM4-5) and CD8α (53-6.7); allophycocyanin-conjugated Abs against CD4 (RM4-5), CD117 (c-Kit; 2B8), CD279 (PD-1; 29F.1A12), IL-4 (11B11), and TCRβ (H57-597); and biotin-conjugated Ab against mouse IgG (Poly4053) were purchased from BioLegend. Purified anti–pre-TCR α-chain (pTα) (2F5), PE-conjugated anti–IFN-γ (XMG1.2), PerCP-conjugated anti-CD8α (53-6.7), and allophycocyanin-conjugated anti-CD11c (HL3) Abs were from BD Pharmingen (San Diego, CA). PE-conjugated anti-Helios (22F6) Ab was from eBioscience. Allophycocyanin-conjugated and α-galactosylceramide–loaded CD1d tetramers were from ProImmune (Oxford, UK). To detect intracellular cytokines, splenocytes (1 × 10⁶ per ml) were stimulated with 500 ng/ml ionomycin (Sigma-Aldrich) and 50 ng/ml phorbol 12-myristate 13-acetate for 4 h in the presence of 10 μg/ml brefeldin A (45). For flow cytometry, cells were incubated with Abs against cell surface Ags on ice for 30 min under dark conditions. Cells were then fixed with 2% formaldehyde at room temperature for 10 min and permeabilized with 0.5% saponin at room temperature for 20 min, followed by incubation with Abs against intracellular Ags, including cytokines at room temperature for 30 min under dark conditions. Stained cells were analyzed in FACSCalibur (BD Biosciences, San Jose, CA), in which >30,000 events gated on the lymphocyte with forward scatter and side scatter were collected. Data analysis was performed with CellQuest Pro software (BD Biosciences).

The CD3⁺ T, CD4⁺ T, and CD8⁺ T cells were isolated from spleen thru positive selection using MACS beads and autoMACS Pro Separator (Miltenyi Biotec, Germany). To isolate double-negative T (DNT) cells from spleen, the CD3⁺ T cells were collected from CD4⁻CD8⁻ cells by the positive selection using MACS beads. Spleen of IFN-α Tg mice was isolated 30 wk after cessation of Dox administration, and its CD3⁺ T (5 × 10⁶), CD4⁺ T (5 × 10⁶), CD8⁺ T (5 × 10⁶), or DNT (3 × 10⁵) cell subsets were isolated and transferred into naive recipients every week for 2 wk. Renal histopathology and the skin of the mice were studied 21 and 60 d after cell transfer, respectively.

Cytoplasmic and nuclear proteins (20 μg) of splenocytes, obtained using subcellular proteome extraction kit (Calbiochem, La Jolla, CA) and quantified using Bradford protein assay (Bio-Rad, Hercules, CA) and BioPhotometer Plus (Eppendorf, Hamburg, Germany), were loaded and subjected to SDS-PAGE. Samples were transferred to nitrocellulose membranes (Millipore, Bedford, MA), probed with Abs, including rabbit anti-Phospho-STAT1 (Cell Signaling Technology, Beverly, MA), rabbit anti-STAT1 (Cell Signaling Technology), rabbit anti-Phospho-STAT2 (Millipore), rabbit anti-STAT2 (Cell Signaling Technology), rabbit anti-IRF9 (Santa Cruz Biotechnology, Santa Cruz, CA), goat anti-actin (Santa Cruz, Biotechnology), and rabbit anti-histone H2B (Santa Cruz Biotechnology) Abs, and developed with an ECL detection system (GE Healthcare, Pasadena, CA).

Total RNA was isolated from splenocytes using RNeasy Mini Kit (Qiagen Valencia, CA). RNA samples were quantified in a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE), and the quality of the RNA was confirmed using an Experion system (Bio-Rad Laboratories).

Cyanine-3 (Cy3)–labeled cRNA was prepared from 50 ng of total RNA using Low Input Quick Amp Labeling Kit, One-Color (Agilent Technologies, Santa Clara, CA), according to the manufacturer’s instruction. The cRNA was purified using RNeasy Mini Kit (Qiagen), and its yield and labeling efficiency were checked using Experion system (Bio-Rad). The Cy3-labeled cRNA was hybridized with SurePrint G3 Mouse Gene Expression Microarray 8 × 60K v2 (Agilent) containing 27,122 Entrez target genes for 17 h at 65°C. After washing, all hybridized microarray slides were scanned using an Agilent scanner. Relative hybridization intensities and background hybridization values were calculated using Agilent Feature Extraction Software (v. 9.5.1.1).

According to the procedures recommended by Agilent, raw signal intensities and flags for each probe were calculated from hybridization intensities (gProcessedSignal) and spot information (gIsSaturated, etc.). The raw signal intensities of samples were log₂ transformed and normalized using the quantile algorithm in the preprocessCore library package (46) in the Bioconductor software (47). We selected probes that registered P flag in both samples. To identify up- or downregulated genes, we calculated Z-scores (48) and ratios (non–log-scaled fold change) from the normalized signal intensities of each probe and compared them with control (control mice) and experiment samples (IFN-α Tg mice). The heat map was generated by MeV software (49). We used a hierarchical clustering method to sort the genes. The color indicated the distance from the median of each row. We also extracted the genes shown to be upregulated or downregulated in human SLE (50) and compared them with the upregulated and downregulated genes of our IFN-α Tg mice by referring to DAVID Bioinformatics Resources database, in which genes with a Z-score exceeding 2 were judged to be upregulated and those below −2 were judged to be downregulated. The microarray data presented in this paper are available from the Gene Expression Omnibus (GEO) database under accession number GSE123549 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE123549).

Statistical analyses were performed using Student t test, and the data were expressed as the mean ± SD.

By using tetracycline regulatory system, we produced the mice Tg for both EμSR-tTA and TetOp-mIFN-α1 Tg (IFN-α Tg) that conditionally express IFN-α protein in the lymphoid cells upon cessation of orally administered Dox (Fig. 1A). We found that serum IFN-α was increased to substantially high levels upon cessation of orally administered Dox in the IFN-α Tg female mice of either B6/FVBN or B6 background (Fig. 1B). Genetic backgrounds of the mice, especially the mutations found, are shown in Supplemental Table I. Serum anti-dsDNA Ab was increased after 12 wk (Fig. 1C), followed by subsequent increase of serum IC after 24–28 wk (Fig. 1D). Serum anti-ssDNA Ab was increased, whereas serum anti-Sm Ab was not increased (Fig. 1E). Proteinuria became detectable subsequent to or in conjunction with the increase of serum IC in the IFN-α Tg mice (Fig. 2A). Pathological studies showed that glomerulonephritis with IC deposition was generated (Fig. 2B), in which the glomerulonephritis of B6/FVBN mice was classified correspondingly to the human World Health Organization (WHO) renal pathology criteria (44); 37.5% of glomeruli was WHO type IV diffuse glomerulonephritis, and 25% was WHO type V membranous glomerulonephritis by 28 wk after cessation of Dox (Fig. 2C). Advanced glomerulonephritides, WHO types IV and V, were also observed in 43.8 and 17.8%, respectively, in the kidney of IFN-α Tg mice of C57BL/6 background (B6) (Supplemental Table II). Other pathological lesions found included epidermal liquefaction degeneration of the skin and splenic onion-skin lesion, both of which were classical to human SLE (Fig. 2C). There were Russell bodies in lymph nodes, interstitial lung disease, and inflammatory cell infiltrates to salivary glands and bile ducts. Lupus band test in the skin was positive, and alopecia was also observed in either IFN-α Tg B6/FVBN or B6 mice (Fig. 2D).

In Tg mice, IFN-α signal was confirmed to be transmitted into the cells as revealed by cytoplasmic STAT1/2 phosphorylation and nuclear translocation of IRF9 (Fig. 3A), the sequence of events that happened when IFN-α was solely increased in vivo. Analysis of splenic T cells showed increased number of activated, CD69-positive CD4 T cells, CD8 T cells, and CD3⁺CD4⁻CD8^- T (DNT) cells, and the DNT cells were also increased in the kidney of IFN-α Tg mice (Fig. 3B). The CD8 T cells were also IFN-γ positive. The CD4 and CD8 T cells were with CD62L^lowCD44^high effector phenotype (Fig. 3C). The splenic T cells of IFN-α Tg mice produced increased amounts of IL-4, IL-6, IL-17, and IFN-γ, whereas the production of IL-2 was decreased (Fig. 3D). Gene expression profile in the splenocytes of Tg mice showed an increase in IFN-α signature genes, such as IRF7, ISG15, Ifi44, Ifi204, Ifit3, Ly6C, Lamp2, MX1, OAS, and defensin (Defb15), whereas Ifi44L and Apobec3 said to be IFN-α inducible were not increased (Fig. 3E and the GEO database under accession number GSE123549). Among cytokines, IL-1, VEGF, IL-15, IL-7, and IL-18 were upregulated. TLR4, TLR7, TLR13, and LPS binding protein (Lbp) were upregulated. Although TLR9 gene expression was not increased, the TLR9 signal appeared transmitted into the cell as judged by the increase of grancalcin (Gca) gene expression and TLR9 protein (Fig. 3A). Gene expression related to the membrane, membrane-associated microtubular signals, and G protein–associated signals as well as glycogen and glucose metabolism generating NADPH, cAMP/DG, and G protein signals were upregulated. Whereas TCR signaling molecules were downmodulated, expression of Ig genes was markedly upregulated (the GEO database under accession number GSE123549). When compared with the genes shown to be upregulated in SLE (50), 35% of the genes, including ifit1, Oas3, or irf7, were similarly upregulated in Tg mice, whereas 4% of the genes were instead downregulated in the Tg mice (Fig. 3F). The genes such as Otof or Prkd2 were increased in human SLE but were decreased in the Tg mice, whereas the genes such as Zak, Card9, Impa2, and Icam4 were decreased in human SLE but were increased in the Tg mice (Fig. 3F).

We noted that activated CD69-expressing DNT cells were increased in the spleen and kidney of IFN-α Tg mice (Figs. 3B, 4B upper panel), and they infiltrated into glomeruli (Fig. 4A). The DNT cells were negative for TCRγδ and CD1 tetramer (Fig. 4B), and thus they were distinct from those derived from TCRγδ T cells (51) or invariant NKT cells (52). The DNT cells expressed B220, PD-1, and Helios but were negative for pTα or c-Kit (CD117) (Fig. 4B middle panel). They also expressed increased amounts of cytokines, such as IFN-γ, IL-4, IL-17, or TNF-α (Fig. 4B lower panel). Further, by transferring as few as 3 × 10⁵ of these DNT cells, the organ manifestations, such as alopecia (Fig. 4C, 4D) or IC-deposited glomerulonephritis with significantly enlarged glomeruli, were reproduced in the recipient naive mice (Fig. 4D).

Another finding included significant expansion of activated CD86⁺B220⁺ B cells and CD86⁺CD11c⁺ dendritic cells in the Tg mice (Fig. 3G). BAFF and IL-15 were also increased in the sera of Tg mice (Fig. 3H).

As to whether IFN-α causes SLE, we present direct evidence that conditional upregulation of IFN-α solely is sufficient to induce SLE; the lesion classical to human SLE, such as onion-skin lesion in the spleen; epidermal liquefaction and alopecia of the skin; IC-deposited glomerulonephritis; and anti-dsDNA Ab in the mice otherwise not prone to autoimmunity. It was noted that the serum levels of IFN-α and the levels of IFN-α signaling achieved in the current study were substantially high, serum levels comparable to those of human SLE (8). The pathogenic process in SLE must require high IFNAR availability and signaling (35). This would explain the reason why only a minor fraction of IFN-α–treated patients succumbed from SLE (2, 19, 20, 24, 25), why only lupus-prone mice develop SLE upon continuous IFN-α exposure (28–30), and why type I IFN showed somewhat protective effects on SLE (36, 37).

The present results also show that activated DNT cells as expanded by IFN-α are directly responsible for the induction of classical organ manifestations of SLE, such as IC-deposited glomerulonephritis and alopecia. The causal relationship between IFN-α and lupus kidney disease is compatible with the previous results (28–30). The finding was compatible with the previous findings that TCRαβ⁺ DNT cells represented a large component of kidney-infiltrating T cells (38, 53) and that the DNT cell count was increased in SLE patients (38, 54, 55) and animal models of SLE (56–60). The DNT cells found in the current study expressed B220, PD-1 and Helios and produced increased cytokines, such as IFN-γ, IL-4, IL-17, or TNF-α but were negative for pTα, c-Kit (CD117), TCRγδ, or CD1 tetramer, indicating that the DNT cells possibly derive from CD8 T cells (61–66) and appear to be the self-reactive T cells undergoing tolerance (61, 67). Mixter et al. (68) did show that the B220⁺ IL2Rβ⁺ CD44⁺ CD69⁻ DNT cells, which are almost identical to those found in the current study except for CD69 positivity, could be the cells undergoing apoptosis but are resistant to die and recognizing endogenous retroviral Ags.

When the gene expression profiles in the IFN-α Tg mice (Fig. 3F) were compared with those of human SLE (50), upregulation of IFN-α can be confirmed to be essentially important in the induction of SLE, as the genes behaving inversely to human SLE also existed in the IFN-α Tg mice. The anti-Sm Ab that raises routinely in ∼30% of SLE patients and is considered diagnostic of SLE was not raised in the present IFN-α Tg mice and also in humans treated with IFN-α (21, 22). The therapy targeting IFNAR was effective in less than half of the patients and mostly effective for those with increased serum IFN-α (39, 40). Thus, upregulation of IFN-α is sufficient, but may not be required, for generation of SLE. Disease mechanisms due to dysregulation of Ag processing or late endosomal and endolysosomal trafficking that is IFN-α independent have been proposed (45, 69), and even IFN-α–dependent autoimmunity has been shown to involve dysregulation of mitochondria, endolysosome, and autophagy (70).

The present study was unable to provide further direct evidence on how IFN-α acts to cause SLE. Nevertheless, we note that activated forms of CD86⁺B220⁺ B cells and CD86⁺CD11c⁺ dendritic cells were also significantly expanded in the IFN-α Tg mice (Fig. 3G). Serum BAFF and IL-15 were also increased (Fig. 3H). The findings are compatible with the previous findings that IFN-α lowers the threshold of B cell activation (32), accelerates disease (29) via or in cooperation with BAFF (31), and promotes differentiation of B cells into plasma cells to make autoantibodies (29, 71). Although inconclusive, the direct action of IFN-α to B cells may induce autoantibodies other than anti-Sm Ab in human SLE and also IFN-α Tg mice.

In summary, we show that sole upregulation of IFN-α is sufficient to induce SLE in nonautoimmune prone mice and that the unique DNT cells expanded by IFN-α directly caused the tissue injuries of SLE, such as those of kidney or skin.

We thank Dr. Marc Lamphier for useful discussion and Yuko Hirohata and Dr. Shingo Kamoshida for technical assistance.

The authors have no financial conflicts of interest.