ARTHRITIS & RHEUMATOLOGY Vol. 67, No. 1, January 2015, pp 225–237 DOI 10.1002/art.38882 © 2015, American College of Rheumatology

Mechanisms of Tumor Necrosis Factor ␣ Antagonist–Induced Lupus in a Murine Model Yuan Xu, Haoyang Zhuang, Shuhong Han, Chao Liu, Hai Wang, Clayton E. Mathews, John Massini, Lijun Yang, and Westley H. Reeves Objective. Tumor necrosis factor ␣ (TNF␣) antagonists are effective for treating rheumatoid arthritis and other inflammatory diseases, but their use can be complicated by the development of lupus-like phenomena. This study was undertaken to investigate the role of TNF␣ in a murine model of lupus. Methods. Toll-like receptor 7 (TLR-7) ligand– driven lupus was induced by injection of pristane into C57BL/6 (B6) mice deficient in TNF␣ (TNF␣ⴚ/ⴚ) or TNF␣-intact B6 mice as wild-type controls. Autoantibody and type I interferon (IFN) production was measured in each group of mice, and the effects of the presence or absence of TNF␣ on type I IFN–producing plasmacytoid dendritic cells (PDCs), Ly-6Chigh monocytes, and TNF␣-producing neutrophils were determined. Results. TNF␣ⴚ/ⴚ mice did not spontaneously develop autoantibodies or clinical manifestations of lupus, suggesting that TNF␣ deficiency alone is insufficient to cause lupus. Although the levels of type I IFN were comparable between untreated TNF␣ⴚ/ⴚ and wildtype control mice, untreated TNF␣ⴚ/ⴚ mice had increased circulating levels of PDCs and PDC-like cells, which enhanced the potential for production of type I IFN. When treated with pristane, TNF␣ⴚ/ⴚ mice developed more severe lupus compared to pristane-treated

controls, characterized by increased levels of anti-Sm/ RNP autoantibodies, type I IFN, PDCs, and peritoneal inflammatory (Ly-6Chigh) monocytes. In pristanetreated TNF␣ⴚ/ⴚ mice, the numbers of neutrophils, a cell type that promotes resolution of inflammation, were decreased considerably, whereas the responses of inflammatory monocytes and PDCs and the production of type I IFN were increased and prolonged. Conclusion. Low levels of TNF␣ will increase the number of circulating PDCs in mice, thereby enhancing the potential to produce type I IFN. However, this does not necessarily lead to type I IFN production or autoimmunity unless there is concomitant exposure to endogenous TLR-7 ligands, which are released from dead cells following pristane treatment. In humans, the rate of clearance of dead cells, along with the levels of TNF␣, may influence who will develop lupus following treatment with TNF␣ inhibitors. Type I interferon (IFN) is strongly implicated in the pathogenesis of systemic lupus erythematosus (SLE). Peripheral blood mononuclear cells (PBMCs) from ⬃60% of adult patients with SLE and from nearly all pediatric patients with SLE are known to overexpress type I IFN–regulated genes (i.e., the IFN signature) (1). Moreover, individuals with 3 copies of the type I IFN gene cluster develop lupus-like disease (2), and IFN␣ exacerbates lupus in NZB/NZW mice (3). In contrast, tumor necrosis factor ␣ (TNF␣) is critical to the pathogenesis of rheumatoid arthritis, inflammatory bowel disease, and psoriasis (4), diseases that are treated with TNF␣ inhibitors. Up to 60% of patients treated with TNF inhibitors develop antinuclear antibodies (ANAs) and 10– 20% develop anti–double-stranded DNA (anti-dsDNA) autoantibodies, but ⬍1% of TNFi-treated patients develop clinical lupus (5–7). The pathogenesis of TNFiinduced lupus remains uncertain and the risk factors are

Supported by research grants from the Lupus Research Institute and the NIH (National Institute of Arthritis and Musculoskeletal and Skin Diseases grant R01-AR-44731 and National Institute of Allergy and Infectious Diseases grant P01-AI-42288). Dr. Zhuang’s work was supported by the NIH (training grant T32-AR-007603). Yuan Xu, BS, Haoyang Zhuang, PhD, Shuhong Han, PhD, Chao Liu, BS, Hai Wang, MD, PhD, Clayton E. Mathews, PhD, John Massini, MD, Lijun Yang, MD, Westley H. Reeves, MD: University of Florida, Gainesville. Dr. Xu and Dr. Zhuang contributed equally to this work. Address correspondence to Westley H. Reeves, MD, Division of Rheumatology and Clinical Immunology, University of Florida, PO Box 100221, Gainesville, FL 32610-0221. E-mail: [email protected]. Submitted for publication April 1, 2014; accepted in revised form September 11, 2014. 225



incompletely understood. Patients with Sjögren’s syndrome treated with a TNFi have increased levels of type I IFN (8), suggesting that TNF␣ negatively regulates type I IFN production. Consistent with that possibility, TNF␣ inhibits the generation of plasmacytoid dendritic cells (PDCs) and type I IFN secretion by PDCs upon viral triggering. Furthermore, in vitro culture of PBMCs with a TNFi increases IFN␣ expression (9). In this study, we examined whether TNF␣deficient (TNF␣⫺/⫺) mice develop lupus. Although TNF␣ deficiency was associated with increased generation of PDCs, it did not induce autoantibodies or lupus-like manifestations spontaneously. However, when lupus was induced by the inflammatory hydrocarbon pristane, the levels of type I IFN and autoantibodies were 5-fold higher in TNF␣⫺/⫺ mice compared to TNF␣-intact wild-type control mice. These observations may have implications in the pathogenesis of TNFiinduced lupus in humans. MATERIALS AND METHODS Mice. Mice were bred and maintained under specific pathogen–free conditions at the University of Florida Animal Facility. CL57BL/6 (B6) and B6,129S-Tnftm1Gkl/J (TNF␣⫺/⫺) mice were from The Jackson Laboratory. For the induction of lupus, 0.5 ml pristane (Sigma-Aldrich) was injected intraperitoneally (IP) into each strain of mice (10,11); some mice were left untreated as a referent control. Serum from the tail vein of mice (at 0–24 weeks after pristane injection) was analyzed for the presence of autoantibodies (10). These studies were approved by the Institutional Animal Care and Use Committee. Enzyme-linked immunosorbent assay (ELISA). Total Ig levels were measured by ELISA, as described previously (11,12). For the anti-Sm/RNP ELISA, mouse serum (1:400 diluted) was incubated with goat anti-mouse IgG secondary antibodies (1:1,000 diluted) (10). Anti-dsDNA antibodies were quantified by ELISA in mouse serum samples (1:400 diluted) using S1 nuclease–treated calf-thymus DNA as antigen (13). Anti–U1-A (anti-RNP subset) antibodies were quantified by ELISA in mouse serum samples (1:400 diluted) incubated with recombinant U1-A as antigen and alkaline phosphatase– conjugated goat anti-mouse isotype antibodies (1:1,000 diluted) (10). To quantify anti–U1-A autoantibodies in wild-type control mice compared to TNF␣⫺/⫺ mice, equal amounts of serum from mice in each group (n ⫽ 12 mice per group) were pooled and serially diluted (fold dilutions from 1:400 to 1:2,624,400). The optical density (OD) values (determined by ELISA) in the samples from the 2 groups were plotted as a function of the serum dilution. To estimate the relative levels of anti–U1-A antibodies, the OD of the pooled samples was plotted as a function of the log dilution at an OD of 0.7, and X-intercepts were determined. Fluorescent ANA assay. Serum ANAs were detected using the Crithidia luciliae kinetoplast staining assay (Inova), following the manufacturer’s protocol. The serum dilution was

1:100, and the secondary antibody was goat anti-mouse IgG (20 ␮g/ml). Quantitative polymerase chain reaction (qPCR). Quantitative PCR analyses were performed as described previously (12,14), using RNA extracted from 106 peritoneal cells in TRIzol (Invitrogen). Peripheral blood RNA was isolated with a QIAamp RNA Blood Mini kit (Qiagen). Complementary DNA was synthesized using a Superscript II First-Strand Synthesis kit (Invitrogen) according to the manufacturer’s protocol. SYBR Green qPCR analysis was performed using an Opticon II thermocycler (Bio-Rad). The primer sequences in the qPCR analyses were as follows: for Isg15 (IFN-stimulated gene 15), forward 5⬘GAGCTAGAGCCTGCAGCAAT-3⬘ and reverse 5⬘-TAAGACCGTCCTGGAGCACT-3⬘; for Irf7 (IFN regulatory factor 7), forward 5⬘-ACAGCACAGGGCGTTTTATC-3⬘ and reverse 5⬘-GAGCCCAGCATTTTCTCTTG-3⬘; for Mx1 (myxovirus resistance protein 1), forward 5⬘-GATCCGACTTCAC TTCCAGATGG-3⬘ and reverse 5⬘-CATCTCAGTGGTAGTCCAACCC-3⬘; for CXCL5, forward 5⬘-CCCCTTCCT CAGTCATAGCC-3⬘ and reverse 5⬘-TGGATTCCGCTTA GCTTTCT-3⬘; for Tcf4 (T cell transcription factor 4 or E2-2), forward 5⬘-GTGGACATTTCACTGGCTCA-3⬘ and reverse 5⬘-CCCTGCTAGTCATGTGGTCA-3⬘; for SpiB, forward 5⬘AACCACCATGCTTGCTCTG-3⬘ and reverse 5⬘-CTGGGTAACTGAAGGGCTTG-3⬘; for CXCL2, forward 5⬘AAGTTTGCCTTGACCCTGAA-3⬘ and reverse 5⬘-CGAGGCACATCAGGTACGAT-3⬘; for CXCL3, forward 5⬘CCACTCTCAAGGATGGTCAA-3⬘ and reverse 5⬘GGATGGATCGCTTTTCTCTG-3⬘; for Dock2 (dedicator of cytokinesis 2), forward 5⬘-CTTCTTCCAAGTCTCAG ATGG-3⬘ and reverse 5⬘-TTCCCACAGTGCTCGGCTCA-3⬘; for Il1a (interleukin-1␣), forward 5⬘-TGTGACTGCCCA AGATGAAG-3⬘ and reverse 5⬘-CTTAGTGCCGTGAG TTTCCC-3⬘; for Il1b, forward 5⬘-TGAAGCAGCTATG GCAACTG-3⬘ and reverse 5⬘-AGGTCAAAGGTTTGG AAG-3⬘; and for Tnfsf13b (BAFF), forward 5⬘-AGGGACCA GAGGAAACAGAA-3⬘ and reverse 5⬘-AAAGCTGAGAAG CCATGGAA-3⬘. Flow cytometry. Flow cytometry was performed as described previously (10,15). Cells were incubated with antimouse CD16/CD32 (Fc Block; BD Biosciences) before being stained with the primary antibody or with isotype control antibodies (for 10 minutes at 22°C). For each sample, 10,000– 50,000 events were acquired using a CYAN ADP flow cytometer (Beckman-Coulter), with results analyzed using the FCS Express 3 program (De Novo Software). The following antibodies were used: phycoerythrin (PE)–conjugated anti–Sca-1, allophycocyanin-conjugated anti-TNF␣, PE-conjugated anti– Ly-6G, allophycocyanin–Cy7–conjugated anti-B220, allophycocyanin-conjugated anti-CD11c, fluorescein isothiocyanate (FITC)–conjugated anti-B220, FITC-conjugated anti– Ly-6C, PE-conjugated anti-CCR7 (BD Biosciences), PE–Cy7– conjugated avidin, Brilliant Violet–conjugated anti-CD11b, FITC-conjugated anti-CCR9, and biotin-conjugated anti– PDCA-1 (BioLegend). Intracellular TNF␣ staining was performed as described previously (16). Monocyte depletion. Clodronate-containing liposomes (anionic) were purchased from Formumax and administered as described previously (17). To deplete peritoneal monocytes in TNF␣⫺/⫺ mice, 150 ␮l of clodronate-containing liposomes was


injected IP into mice after the mice had been treated with pristane 2 weeks earlier. Peritoneal cells were collected 2 days later, and the percentages of Ly-6Chigh and Ly-6Clow monocytes and PDCs were determined. Expression levels of the IFN-regulated markers Sca-1 (measured as the mean fluorescence intensity on flow cytometry) and Irf7 (determined by qPCR) were quantified. Neutrophil transfer. For cell transfer experiments, neutrophil donors (wild-type B6 mice) were treated with 1.5 ml thioglycollate IP, and 5 hours later, peritoneal cells were harvested from the mice. Clodronate-containing liposomes (0.1 ml IP) were injected 30 minutes after the injection of thioglycollate. The donor peritoneal cells were pooled after harvesting. Splenocytes from the same donor mice served as a control and were treated similarly with clodronate-containing liposomes. Red blood cells (RBCs) were lysed with 10 ml RBC lysis buffer (Qiagene). Cells were pelleted at 4°C and immediately resuspended at 5 ⫻ 106/ml in medium. Three TNF␣⫺/⫺ mice pretreated with pristane 2 weeks earlier received either 5 ⫻ 106 monocyte-depleted peritoneal cells (mainly neutrophils) or 5 ⫻ 106 splenocytes (mainly lymphocytes), injected IP. The composition of the donor cell populations was verified by flow cytometry. As a positive control, TNF␣⫺/⫺ mice were treated with pristane for 2 weeks but did not receive any donor cells; untreated TNF␣⫺/⫺ mice served as negative controls. Cytokine ELISAs. ELISAs for CCL19 and CXCL5 (R&D Systems) were performed in accordance with the manufacturer’s instructions. The OD value was converted to a concentration using standard curves, based on the expression of recombinant chemokines. Results were analyzed using the Softmax Pro 4.3 program (Molecular Devices). Assessment of glomerulonephritis. Glomerular cellularity was evaluated by counting the nuclei present per glomerular tissue cross-section (18). Renal immune complex deposition was evaluated by direct immunofluorescence of 4-␮m frozen sections of renal tissue stained with FITC-conjugated rat anti-mouse C3 monoclonal antibodies (Cedarlane) (18). Statistical analysis. Statistical analyses were performed using Prism software version 4.0 (GraphPad). Differences between groups were analyzed using Student’s unpaired t-test (for normally distributed data) or Mann-Whitney U test (for non-normally distributed data). Data are expressed as the mean ⫾ SD for normally distributed data sets. Normality was determined by the D’agostino and Person omnibus test. All tests were 2-sided. P values less than 0.05 were considered significant.

RESULTS Pristane-treated B6 mice are known to develop a Toll-like receptor 7 (TLR-7)–driven lupus syndrome associated with production of type I IFN and TNF␣ and accumulation of dead cells in the mouse bone marrow (BM) (16,19). It is likely that uncleared dead cells are the source of the endogenous TLR-7 ligands driving the inflammatory response to pristane. We therefore examined the effect of TNF␣ deficiency on the pathogenesis


of lupus in pristane-treated mice as a means of modeling TNFi-induced lupus. Increased autoantibody production in TNF␣ⴚ/ⴚ mice. In both TNF␣⫺/⫺ mice and TNF␣-intact wild-type control mice, anti–U1-A autoantibodies were first detected at 6–12 weeks after pristane treatment (Figure 1A, left). Although the time course of onset of these autoantibodies was similar between the groups, the levels of anti–U1-A were increased in TNF␣⫺/⫺ mice compared to wild-type control mice at 24 weeks after pristane treatment (P ⫽ 0.0006, by Student’s t-test). Moreover, at 6 months after pristane treatment, 10 of 12 TNF␣⫺/⫺ mice had high levels of anti–U1-A autoantibodies, compared to 2 of 12 controls. Anti–U1-A autoantibody titers (at 24 weeks) were substantially higher in the pooled serum from TNF␣⫺/⫺ mice compared to wild-type controls (Figure 1A, right). A significant difference was still observed at a serum dilution of 1:874,800. When the data were analyzed using a log plot, the mean anti–U1-A autoantibody level in TNF␣⫺/⫺ mice was 5.37-fold higher than that in wild-type control mice (results not shown). TNF␣⫺/⫺ mice also had higher levels of anti-Sm/ RNP antibodies at 24 weeks after pristane treatment (Figure 1B). However, consistent with previous observations (19), pristane-treated wild-type mice did not produce anti-dsDNA or anti–single-stranded DNA antibodies (Figure 1B). Similarly, anti-DNA autoantibodies were absent in pristane-treated TNF␣⫺/⫺ mice, suggesting that the absence of TNF␣ did not break tolerance to DNA. Levels of total IgG2b/IgG2c, the predominant isotypes of anti-Sm/RNP, were similar between TNF␣⫺/⫺ mice and wild-type control mice (Figure 1C). In many patients, TNFi therapy induces the production of ANAs. Untreated TNF␣-deficient mice were negative for ANAs up to 12 months of age (n ⫽ 6 mice tested) (Figure 1D, left, top). In contrast, at 28 weeks after pristane treatment, the serum from 3 of 6 TNF␣⫺/⫺ mice was strongly ANA positive (Figure 1D, left, bottom). Glomerulonephritis is generally mild in pristanetreated B6 mice (19). Although the intensity of C3 staining at 12 weeks after pristane treatment was somewhat increased in TNF␣⫺/⫺ mice compared to wild-type control mice (Figure 1D, center), the extent of glomerular cellularity was similar between the groups (Figure 1D, right). Increased type I IFN production in pristanetreated TNF␣ⴚ/ⴚ mice. Since autoantibody production in pristane-treated mice is dependent on type I IFN (20), we examined the expression of the IFN-inducible pro-



Figure 1. Increased autoantibody production in tumor necrosis factor ␣–deficient (TNF␣⫺/⫺) mice. A, Left, TNF␣⫺/⫺ mice and TNF␣-intact wild-type C57BL/6 (B6) control mice were treated with pristane. Serum was collected at the indicated time points after treatment (n ⫽ 4–6 mice per group at 3 days and 2, 6, and 12 weeks; n ⫽ 8–12 mice per group at 3 weeks and 24 weeks) and anti–U1-A levels were measured; serum from untreated B6 mice was used as a control (broken horizontal line). Right, Anti–U1-A antibody titration curves were determined in the pooled serum of each group at 24 weeks after pristane treatment (mean of 12 mice per group). B, IgG anti-Sm/RNP and IgG anti–double-stranded DNA (anti-dsDNA) and anti–single-stranded DNA (anti-ssDNA) autoantibody levels were determined by enzyme-linked immunosorbent assay at 24 weeks after pristane treatment (12 mice per group). Broken horizontal line indicates the cutoff value for the positive control (pos ctr). C, Serum levels of total IgG1, IgG2b, IgG2c, and IgG3 were determined in pristane-treated TNF␣⫺/⫺ mice (n ⫽ 11) and wild-type control mice (n ⫽ 12). Bars show the mean ⫾ SD. D, Left, Serum from an untreated and a pristane-treated (28 weeks posttreatment) TNF␣⫺/⫺ mouse was assessed by fluorescent staining for antinuclear antibodies. Center, Renal tissue (4-␮m sections) from a wild-type control mouse and a TNF␣⫺/⫺ mouse at 3 months after pristane treatment was assessed for immune complex deposition by staining with fluorescein isothiocyanate–conjugated anti-mouse C3 antibodies and hematoxylin and eosin (H&E). Results are representative of a total of 6 mice per group. Right, Glomerular cellularity was evaluated by counting the number of nuclei per glomerulus. Similar counts were obtained by 2 independent observers. Symbols represent individual mice; horizontal bars denote the mean. ⴱ ⫽ P ⬍ 0.05; ⴱⴱⴱ ⫽ P ⬍ 0.001 versus controls, by Student’s unpaired t-test (A) or Mann-Whitney U test (B). NS ⫽ not significant.

tein Sca-1 on circulating B cells. The mean fluorescence intensity of Sca-1 protein expression on peripheral B cells was increased in TNF␣⫺/⫺ mice as early as 1 week after pristane treatment (Figure 2A). Similarly, qPCR

showed that the expression of messenger RNA (mRNA) for the IFN-stimulated genes Irf7, Mx1, and Tnfsf13b (BAFF) was increased in peritoneal cells from TNF␣⫺/⫺ mice 1 week after pristane treatment (Figure 2B).



Figure 2. Increased type I interferon (IFN) production in pristane-treated TNF␣⫺/⫺ mice. A, TNF␣⫺/⫺ and wild-type B6 control mice were injected with pristane. Peripheral blood samples were collected at different time points after treatment. Erythrocytes were lysed and leukocytes were gated on the CD19⫹ B cell population and stained for the expression of Sca-1 protein, assessed as the mean fluorescence intensity (MFI). B, Expression of the IFN-stimulated genes Irf7, Mx1, and BAFF (Tnfsf13b) was examined by quantitative polymerase chain reaction in mouse peripheral blood cells at 1 week and 6 weeks after pristane treatment, assessed as fold expression relative to untreated mice. Results in A and B are the mean ⫾ SD in 3–5 mice per group. C, Total peritoneal exudate cells from wild-type control mice, obtained at several time points after pristane treatment, were surface-stained with anti–Ly-6G and stained intracellularly for TNF␣ (TNF␣⫹Ly-6G⫹ cells as a percentage of total peritoneal exudate cells). ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01 versus controls. See Figure 1 for other definitions.

Expression of Irf7 and Mx1 mRNA was further increased in TNF␣⫺/⫺ mice at 6 weeks, but the levels were not significantly different from those in wild-type control mice. Levels of BAFF declined in TNF␣⫺/⫺ mice, but did not differ between the strains, at 6 weeks after pristane treatment. CD11b⫹Ly-6G⫹ peritoneal neutrophils from wild-type control mice exhibited increased intracellular levels of TNF␣ by 2 weeks after pristane treatment. Moreover, the intensity of intracellular staining for TNF␣ continued to increase up to 6 weeks after treatment, before returning to baseline levels (Figure 2C). Production of TNF␣ previously has been shown to require the presence of TLR-7 (16). Effect of TNF␣ on Ly-6Chigh (inflammatory) monocytes. The total number of peritoneal cells was determined at different time points after pristane treatment in TNF␣⫺/⫺ and wild-type control mice (Figure 3A, right). Ly-6Chigh monocytes are a major source of

type I IFN in pristane-treated mice (21). In the peritoneum, IFN␣ receptor signaling blocks the maturation of Ly-6Chigh monocytes to Ly-6Clow monocytes/macrophages. The percentage of peritoneal Ly-6Chigh monocytes (CD11b⫹Ly-6G⫺Ly-6Chigh cells) was increased in TNF␣⫺/⫺ mice up to 12 weeks after pristane treatment (Figure 3A, left). Consistent with the increased type I IFN activity observed at 1 week (Figures 2A and B), Ly-6Chigh monocyte counts also were increased in TNF␣⫺/⫺ mice at 1 week after pristane treatment (Figure 3B, top). In contrast, at 2 weeks and 6 weeks, Ly-6Chigh monocyte counts were comparable in TNF␣⫺/⫺ and wild-type control mice (Figure 3B, top) and type I IFN activity was less divergent (Figure 2B). However, resolution of inflammation was delayed in TNF␣⫺/⫺ mice, as indicated by a higher percentage of Ly-6Chigh monocytes at 32 weeks post–pristane treatment (Figure 3B, bottom). Moreover, the intensity of cell surface staining for Ly-6C



Figure 3. Effects of TNF␣ on Ly-6Chigh (inflammatory) monocytes. A, Peritoneal cells from TNF␣⫺/⫺ and wild-type B6 control mice were analyzed by flow cytometry at 0–12 weeks after pristane injection to determine the percentages of CD11b⫹Ly-6Chigh monocytes (boxed areas) among total peritoneal exudate cells (PECs) (left) and numbers of total peritoneal cells (right). B, Top, Numbers of CD11b⫹Ly-6Chigh monocytes were determined at different time points after pristane injection. Bottom, Percentages of Ly-6Chigh monocytes (CD11b⫹Ly-6ChighLy-6G⫺) among total PECs were assessed at 32 weeks after pristane injection (left), and Ly-6C expression (mean fluorescence intensity [MFI]) on peritoneal Ly-6Chigh monocytes was assessed at 6 weeks after treatment (right). Results in A and B are the mean ⫾ SD in 3–5 mice per group. C, Top, Expression of Ly-6C was assessed by flow cytometry in peritoneal cells from pristane-treated TNF␣⫺/⫺ mice 2 days following treatment with clodronate-containing liposomes (Clo-Lip) (or phosphate buffered saline [PBS] as control). Boxed areas indicate the population of Ly-6Chigh monocytes. Bottom, Percentages of peritoneal Ly-6Chigh monocytes and plasmacytoid dendritic cells among total PECs were determined in PBS- or Clo-Lip–treated TNF␣⫺/⫺ mice. D, Expression of Sca-1 (MFI, determined by flow cytometry) on peritoneal B cells and Irf7 mRNA (fold expression relative to untreated mice, determined by quantitative polymerase chain reaction) on PECs of TNF␣⫺/⫺ mice was assessed after PBS or Clo-Lip treatment. Results in C and D are the mean ⫾ SD in 4 mice per group. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01; ⴱⴱⴱ ⫽ P ⬍ 0.001 versus controls, by Student’s unpaired t-test. See Figure 1 for other definitions.

(an IFN-inducible gene product) was increased in the Ly-6Chigh monocytes from TNF␣⫺/⫺ mice compared to wild-type controls (Figure 3B, bottom). Ly-6Chigh monocytes are a major source of type I IFN production in pristane-treated B6 mice. To evaluate the source of type I IFN production in TNF␣⫺/⫺ mice, we depleted monocytes using an injection of clodronatecontaining liposomes 2 weeks after pristane treatment (Figure 3C). At 48 hours after pristane treatment, the

percentage of CD11b⫹Ly-6Chigh cells was reduced by ⬃90% following IP injection of clodronate-containing liposomes, whereas there was little effect on CD11b⫺CD11c intermediate B220⫹PDCA-1⫹ PDCs, which are poorly phagocytic. Blood samples obtained from the TNF␣⫺/⫺ mice 24 hours before the injection of clodronate-containing liposomes (or phosphate buffered saline [PBS] as controls) were examined by flow cytometry to verify that the


2 treatment groups had similar expression of the IFNinducible marker Sca-1 on peripheral blood B220⫹ cells (results not shown). Forty-eight hours after the injection of clodronate-containing liposomes, the expression of Sca1 in the peritoneal exudate cells (PECs) (as determined by flow cytometry) was reduced substantially compared to that in PBS-treated controls, and there was a trend toward significance in the reduction of Irf7 mRNA expression in mice injected with clodronatecontaining liposomes (as determined by qPCR) (Figure 3D). Taken together, these findings suggest that, as has been shown previously in wild-type B6 mice (20), Ly6Chigh monocytes are responsible for much of the type I IFN produced in the peritoneum of pristane-treated TNF␣⫺/⫺ mice. Effect of defective neutrophil recruitment on inflammation resolution in TNF␣ⴚ/ⴚ mice. Since neutrophils are the main source of TNF␣ in the BM and peritoneum of pristane-treated B6 mice (16) and their apoptosis promotes resolution of inflammation (16), we examined CD11b⫹Ly-6G⫹ peritoneal neutrophils (Figure 4A). TNF␣⫺/⫺ mice had markedly fewer peritoneal neutrophils than did controls, both as a percentage of the total cell population and as an absolute count, especially near the peak of the acute inflammatory response (6 weeks) (Figure 4A, left and center). Similarly, the percentage of peripheral blood neutrophils was decreased in TNF␣⫺/⫺ mice (Figure 4A, right). There was no significant change in the percentage of neutrophils in the mouse BM at 6 weeks after pristane treatment (results not shown), suggesting that TNF␣⫺/⫺ mice do not have an intrinsic deficiency in neutrophil maturation. Levels of Il1a and Il1b transcripts in the peritoneum, as well as expression of the neutrophil chemokines CXCL2, CXCL3, and CXCL5, were all decreased in TNF␣⫺/⫺ mice compared to wild-type controls (Figure 4B). Moreover, decreased production of CXCL5 in TNF␣⫺/⫺ mice was confirmed by ELISA (results not shown). Apoptotic neutrophils are thought to play a critical role in the resolution phase of inflammation (22). To explore the possibility that a lack of peritoneal neutrophils contributes to the prolonged inflammatory response to pristane in TNF␣⫺/⫺ mice, we performed neutrophil transfer experiments and assessed apoptosis/ necrosis. TNF␣⫺/⫺ mice and wild-type control mice were injected with pristane, and 1 week later, PECs were stained for apoptotic and necrotic cells using annexin V (AnxV) and 7-aminoactinomycin D (7-AAD) (Figure 4C). Although there was no significant difference in the


percentage of early apoptotic cells (AnxV⫹, 7-AAD⫺ cells), TNF␣⫺/⫺ mice had more late apoptotic/necrotic cells (AnxV⫹, 7-AAD⫹ cells) than did wild-type control mice. Since necrotic cells tend to be proinflammatory, this may contribute to the increased inflammatory response in TNF␣⫺/⫺ mice. To examine the functional role of neutrophils, we transferred enriched peritoneal neutrophils (elicited by thioglycollate injection) or splenic lymphocytes into pristane-treated TNF␣⫺/⫺ mice (Figure 4D). Prior to treatment, Sca1 expression on peripheral blood B cells (B220⫹ cells) was similar in all recipient mice (Figure 4D, left). The recipient mice received either 5 ⫻ 106 clodronate-containing liposome–treated PECs (enriched in neutrophils) or 5 ⫻ 106 clodronate-containing liposome–treated splenocytes (enriched in lymphocytes) as a control (Figure 4D, pretreatment, right panel). Two days after the cell transfer, Sca1 expression (determined by flow cytometry) on recipient peripheral blood B cells was similar between the 2 treatment groups (Figure 4D, after treatment, left panel) and comparable to that in TNF␣⫺/⫺ mice that had not received cell transfer. However, 3 days after transfer, TNF␣⫺/⫺ mice that were treated IP with peritoneal neutrophils exhibited significantly lower Sca1 expression on peritoneal B cells than did recipients of an equal number of splenocytes (Figure 4D, after treatment, right panel). Sca1 expression in the recipients of splenocytes was comparable to that in positive control (no cell transfer) mice. Thus, increasing the number of peritoneal neutrophils led to suppression of peritoneal IFN levels (measured as expression of the IFN-inducible cell surface marker Sca1) in pristane-treated mice, consistent with the proposal that neutrophils are involved in the resolution of inflammation. Taken together, the findings from these experiments suggest that there are 2 potential mechanisms by which TNF␣ deficiency might promote chronic inflammation: 1) by increasing the number of necrotic cells, and 2) by decreasing the number of peritoneal neutrophils, a cell type that promotes resolution of inflammation. Increased numbers of PDCs in TNF␣ⴚ/ⴚ mice. Although PDCs represent ⬍3% of all peritoneal cells, they produce up to 1,000-fold more type I IFN compared to myeloid dendritic cells (23). Because TNF␣ inhibits the differentiation of PDCs from CD34⫹ hematopoietic progenitor cells (9), we sought to determine whether the numbers of PDCs (CD11c intermediate PDCA-1 high B220⫹Ly-6C⫹ cells) were increased in TNF␣⫺/⫺ mice. Indeed, the PDC count was increased in the PECs of TNF␣⫺/⫺ mice at 6 weeks after pristane treatment



Figure 4. Defective neutrophil recruitment in TNF␣⫺/⫺ mice. Peritoneal cells from TNF␣⫺/⫺ and wild-type B6 control mice were analyzed by flow cytometry at 0–12 weeks after pristane treatment. A, Percentages of CD11b⫹Ly-6G⫹ neutrophils (boxed areas) at 0–6 weeks after pristane treatment (left), absolute numbers of CD11b⫹Ly-6G⫹ neutrophils at 0, 1, 6, and 12 weeks (calculated as total peritoneal cell count ⫻ fraction of neutrophils) (center), and percentages of CD11b⫹Ly-6G⫹ neutrophils at 0–6 weeks (right) were determined in the peripheral blood from each group of mice. B, Expression of Il1a and Il1b and the neutrophil-attractive chemokines CXCL2, CXCL3, and CXCL5 was assessed by quantitative polymerase chain reaction in mouse peritoneal exudate cells (PECs) at 6 weeks after pristane treatment. C, Percentages of apoptotic (annexin V–positive [AnxV⫹], 7-aminoactinomycin D–negative [7-AAD⫺]) and necrotic (AnxV⫹, 7-AAD⫹) cells were determined in the PECs of TNF␣⫺/⫺ and wild-type control mice 2 weeks after pristane treatment. D, Left, Recipient TNF␣⫺/⫺ mice (recipients of peritoneal neutrophils [RPEC] or splenocytes [RSPL]), at 2 weeks after pristane treatment, were compared for levels of Sca-1 expression (mean fluorescence intensity [MFI]) in peripheral blood mononuclear cells (PBMCs) before donor cell transfer. Composition of the B6 mouse donor cells (PECs or splenocytes) was assessed following treatment of the mice with clodronate-containing liposomes (Clo-Lip) (n ⫽ 3 mice per group). Right, Sca-1 expression (MFI, by flow cytometry) was assessed in B220⫹ cells from PBMCs and PECs at 2 days and 3 days, respectively, after donor cell transfer into recipient TNF␣⫺/⫺ mice. Results are the mean ⫾ SD in 3–5 mice per group. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01; ⴱⴱⴱ ⫽ P ⬍ 0.001 versus controls, by Student’s unpaired t-test. See Figure 1 for other definitions.

(Figures 5A and B and Table 1). PDC levels increased as early as 1 week after pristane treatment in both strains, but the levels were dramatically elevated in TNF␣⫺/⫺ mice (Figure 5B). In wild-type control mice, the PDC levels peaked at ⬃6 weeks, and declined thereafter. In

contrast, in TNF␣⫺/⫺ mice, the PDC levels remained elevated through the 12-week time point (Figure 5B). Interestingly, TNF␣⫺/⫺ mice exhibited a second population of cells (designated P2 in Figure 5A) with surface markers similar to those of PDCs, except for a



Figure 5. Increased levels of plasmacytoid dendritic cells (PDCs) in TNF␣⫺/⫺ mice. Peritoneal cells were collected from TNF␣⫺/⫺ and wild-type B6 control mice at 6 weeks after pristane treatment. A, Left, After gating on live cells, the CD11b⫺CD11c⫹ cell population was analyzed for B220 and PDCA-1 staining, and PDCA-1⫹ cells were analyzed for Ly-6C staining, with boxed areas showing PDCs and the P1 and P2 cell populations. Right, PDCs (CD11b⫺CD11c⫹PDCA-1⫹B220⫹) from peripheral blood mononuclear cells (PBMCs) were assessed for Ly-6C expression by flow cytometry (n ⫽ 3–6 mice per group) at 6 weeks after pristane injection. Boxed area shows the P2 cell population in TNF␣⫺/⫺ mice. B, The numbers of PDCs in peritoneal exudate cells (PECs) at 0–12 weeks after pristane treatment (top) and the percentages of P1/P2 cells (bottom, left) and PDCs (bottom, right) among total PECs were determined in each group. C, Left, Histograms show surface staining of the P1/P2 cell populations, PDCs, and Ly-6Chigh monocytes (mono) for CD11b, Ly-6C, CD11c, B220, and PDCA-1; staining with isotype control is shown as a gray-filled histogram. Values to the right of the histograms are the mean fluorescence intensity (MFI) of each surface stain in wild-type control mice (black) and TNF␣⫺/⫺ mice (red) (see also Table 1). Right, Levels of CCL19 in peritoneal lavage fluid at 6 weeks after pristane treatment (determined by enzyme-linked immunosorbent assay [ELISA]) and staining of PDCs for CCR7 and CCR9 (MFI, by flow cytometry) were determined in each group of mice (top), and the percentage of PDCs in PBMCs from each group over time was determined (bottom). Results in B and C are the mean ⫾ SD in 3–6 mice per group. D, Left, The percentage of PDCs (CD11b⫺CD11c⫹PDCA-1⫹B220⫹) was determined in the bone marrow (BM) of untreated (No Tx) and 6-week pristane–treated mice from each group. Right, Expression of PDC differentiation factors (E2-2, SpiB) and the PDC regulatory factor Dock2 was evaluated by quantitative polymerase chain reaction in the BM of untreated mice and the PBMCs (at 1 week) and PECs (at 6 weeks) from pristane-treated mice. Bars show the mean ⫾ SD in 3–4 mice per group. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01; ⴱⴱⴱ ⫽ P ⬍ 0.001 versus controls, by Student’s unpaired t-test. See Figure 1 for other definitions.

lower expression of B220. This P2 population of cells expressed Ly-6C and PDCA-1, both of which are important PDC markers, with expression levels comparable to

those in regular PDCs. The P2 cell population was nearly absent in wild-type control mice, whereas a population of cells that exhibited higher expression of



Table 1. Expression of cell surface markers in TNF␣⫺/⫺ mice compared to wild-type control mice* Population 1/2

Ly-6Chigh monocytes



Wild-type control, P1 cells

TNF␣⫺/⫺, P2 cells

Wild-type control


Wild-type control

CD11b Ly-6C CD11c B220 PDCA-1

4.9 ⫾ 0.3 8.4 ⫾ 1.0 126.6 ⫾ 20.5 10.1 ⫾ 1.1 125.9 ⫾ 12.2

3.3 ⫾ 0.2 165.3 ⫾ 47.8 56.8 ⫾ 3.2 11.9 ⫾ 0.3 221.8 ⫾ 12.1

2.6 ⫾ 0.3 270.2 ⫾ 142.7 56.5 ⫾ 6.1 140.6 ⫾ 22.2 515.9 ⫾ 58.8

2.3 ⫾ 0.1 322.9 ⫾ 47.8 52.8 ⫾ 4.6 116.3 ⫾ 7.7 503.1 ⫾ 43.0

120.6 ⫾ 47.0 862.9 ⫾ 101.7 6.8 ⫾ 1.5 6.0 ⫾ 3.4 251.2 ⫾ 63.3

TNF␣⫺/⫺ 138.0 ⫾ 26.9 1,327.7 ⫾ 80 5.9 ⫾ 0.4 1.8 ⫾ 0.2 215.7 ⫾ 27.2

* Results, presented as the mean ⫾ SD, are the mean fluorescence intensity in tumor necrosis factor ␣ (TNF␣)–intact C57BL/6 (B6) wild-type control mice and B6 mice deficient in TNF␣ (TNF␣⫺/⫺). PDCs ⫽ plasmacytoid dendritic cells.

CD11c and CD11b, lower expression of PDCA-1, and no expression of Ly-6C (designated P1 in Figures 5A and C) was seen instead. P1 cells may represent conventional dendritic cells (Table 1). P2 cells also were seen in the peripheral blood of TNF␣⫺/⫺ mice at 6 weeks after pristane treatment (Figure 5A). The P1 and P2 cell populations were readily distinguishable from Ly-6Chigh monocytes (CD11c⫺CD11b⫹PDCA-1intermediateB220⫺Ly6Chigh cells) (Table 1 and Figure 5C). It is likely that the increased numbers of PDCs and P2 cells were not related to altered chemokine production, since the PDC-attractive chemokine CCL19 was present at similar levels in TNF␣⫺/⫺ and wild-type control mice (Figure 5C, right, top). We also analyzed the expression of CCR7 and CCR9 on the surface of peritoneal PDCs (CD11b⫺CD11c intermediate B220⫹PDCA-1⫹ cells) from TNF␣⫺/⫺ and wild-type control mice 6 weeks after pristane treatment and found no significant difference in the expression of CCR7 (Figure 5C, right, top). However, the expression of CCR9⫹ PDCs was lower in TNF␣⫺/⫺ mice. Unexpectedly, the mean fluorescence intensity of CCR9 staining was lower on PDCs from TNF␣⫺/⫺ mice than on PDCs from wild-type control mice (Figure 5C, right, top), suggesting that decreased chemokine receptor expression is not responsible for the increased numbers of PDCs in TNF␣⫺/⫺ mice. Even before treatment, TNF␣⫺/⫺ mice had a higher percentage of peripheral blood PDCs compared to controls. This persisted up to 6 weeks after pristane treatment (Figure 5C, right, bottom). In contrast, fewer PDCs were seen in the BM from untreated TNF␣⫺/⫺ mice and the BM from TNF␣⫺/⫺ mice at 6 weeks after pristane treatment, compared to the BM from wild-type controls (Figure 5D). Thus, PDCs may mature more rapidly and/or migrate more rapidly to sites of inflammation in TNF␣⫺/⫺ mice. Consistent with that possibility, levels of mRNA for E2-2 and SpiB, key transcription

factors mediating PDC maturation (24), and levels of mRNA for Dock2, a hematopoietic cell–specific Rac GTPase activator essential for PDC and lymphocyte migration in response to CCL19 and other chemokines (25–27), were increased in TNF␣⫺/⫺ mice (Figure 5D). DISCUSSION ANAs and anti-dsDNA antibodies are common in patients treated with TNF inhibitors, but ⬍1% of patients treated with a TNFi develop clinical SLE (5– 7,28). Increased levels of type I IFN may promote lupus in these patients, although the factors determining who will develop disease remain unclear. The view of lupus as strictly type I IFN–mediated may be overly simplistic, and the presence of multiple TLR-7–stimulated cytokines, including type I IFN and TNF␣, may help explain the diversity of clinical manifestations (16). Analogous to the polar forms of leprosy, which depend on the balance of IL-4 and IFN␥ (29), the presence of autoantibodies in nephritis and hematologic manifestations in arthritis may represent opposite ends of a clinical spectrum of SLE that might reflect the balance of type I IFN and TNF␣, respectively (16). We found that TNF␣⫺/⫺ mice did not spontaneously develop serologic or clinical manifestations of lupus or the IFN signature (Figures 1 and 2), despite the increased numbers of circulating PDCs (Figure 5). However, when chronic TLR-7–driven inflammation was induced with pristane, the levels of autoantibodies, type I IFN, and Ly-6Chigh monocytes were all higher in TNF␣⫺/⫺ mice compared to wild-type control mice. In contrast, pristane-treated TNF␣⫺/⫺ mice do not develop hematologic abnormalities (16). Thus, although TNF␣ deficiency alone is insufficient to cause lupus, the increased numbers of PDCs may help polarize TLR-7– stimulated cytokine production toward high type I IFN levels, thereby promoting autoantibody production.


Anti-Sm/RNP autoantibodies have been detected in patients with TNFi-induced lupus (6,7), but antidsDNA antibodies are more common (although only present in ⬃10–20% of patients). In our experiments, we found that, compared to wild-type controls, TNF␣⫺/⫺ mice had markedly higher levels of anti-Sm/RNP autoantibodies (a more than 5-fold increase in the levels of anti–U1-A) after pristane treatment (Figure 1A), suggesting that the absence of TNF␣ promotes either the breaking of tolerance or the maturation of autoantibody-producing plasma cells. Pristane-treated TNF␣⫺/⫺ mice, similar to wild-type controls (19), did not produce anti-DNA autoantibodies (Figure 1B), and anti-Sm/RNP autoantibody production was not accelerated in TNF␣⫺/⫺ mice (Figure 1A), suggesting that TNF␣ deficiency does not fundamentally alter B cell tolerance in pristane-induced lupus. Somewhat unexpectedly, pristane-treated TNF␣⫺/⫺ mice did not exhibit major differences in splenic B cell subsets when compared to wild-type controls (Xu Y: unpublished observations). TNF␣⫺/⫺ mice lack normal B cell follicles, organized follicular dendritic cell networks, and germinal centers, since TNF␣ is required during the development of secondary lymphoid tissue to create a milieu permissive of germinal center formation (30,31). In contrast, germinal centers form normally after immunization of TNFi-treated adult mice. TNFi-treated mice have impaired T cell–dependent antibody responses, but TNF␣⫺/⫺ mice produce antigen-specific IgG after immunization with TNP–keyhole limpet hemocyanin, although at lower levels than that in wild-type mice, whereas IgG responses to the TI-2 antigen TNP-Ficoll are enhanced (30). Although TNF␣⫺/⫺ mice lack normal germinal centers, early T cell responses are intact in the mice following immunization (31). The enhanced IgG anti-Sm/RNP response in TNF␣⫺/⫺ mice (Figure 1) is suggestive of a TI-2 response, but pristane does not induce anti-Sm/RNP in T cell–deficient mice (32). Thus, pristane may stimulate the extrafollicular differentiation of anti-Sm/RNP autoantibody–producing cells, as has also been reported for rheumatoid factor (33,34). It will be of interest to see whether IL-21–producing extrafollicular T cells (35) drive this anti-Sm/RNP response. Further studies are needed to definitively address the question of how TNF␣ deficiency enhances serum autoantibody levels. Autoantibody production and renal disease in pristane-induced lupus is highly dependent on type I IFN production (19). In wild-type mice, much of this type I IFN is produced by Ly-6Chigh monocytes (20).


However, PDCs produce up to 1,000-fold more type I IFN than has been observed in other cell types, and PDCs have been implicated as key players in the pathogenesis of lupus (36). In our study, similar to wild-type mice, Ly-6Chigh monocytes were a significant, but not the only, source of type I IFN in TNF␣⫺/⫺ mice (Figure 3). Untreated TNF␣⫺/⫺ mice had decreased numbers of BM PDCs and increased numbers of circulating PDCs (Figure 5), the latter of which may be poised to overproduce type I IFN. However, type I IFN levels were comparable in untreated TNF␣⫺/⫺ mice and wild-type control mice (Figure 2), suggesting that these circulating PDCs are not detrimental unless the mice are exposed to TLR-7 ligands (37), probably derived from dead cells (16). Interestingly, the number of late apoptotic/necrotic cells was higher in pristane-treated TNF␣⫺/⫺ mice compared to controls (Figure 4C), and this could serve as a source of endogenous TLR-7 ligands driving inflammatory cytokine production and possibly B cell differentiation. We also identified a novel PDC-like cell population in TNF␣⫺/⫺ mice (designated P2 cells [Figure 5]), the nature of which remains to be determined. We speculate that the P2 cells identified in the blood and PECs of TNF␣⫺/⫺ mice are PDC precursors that prematurely exit the BM. The E2-2–inducible transcription factor SpiB regulates Ly-49Q expression on naive mouse peripheral PDCs and mature mouse BM PDCs and is necessary for optimal IFN␣ production (38,39). The absence of SpiB causes premature export of immature (Ly-49Q⫺) PDCs from the BM (40). Additional studies are needed to evaluate SpiB expression in BM PDCs and circulating P2 cells and to determine whether the P2 cells are Ly-49Q⫺. Neutrophils are recruited to the peritoneum of pristane-treated mice in response to IL-1␣ and the neutrophil-attractive chemokine CXCL5 (21). In our study, this response was attenuated in TNF␣⫺/⫺ mice (Figure 4). Although the levels of other neutrophil chemokines (CXCL2, CXCL3) were also reduced, CXCL5 production is TNF␣-driven (41,42), and we found that CXCL5 mRNA and protein levels were decreased in TNF␣⫺/⫺ mice (Figure 4B), suggesting that decreased CXCL5 expression was at least partly responsible for the attenuated peritoneal neutrophil influx. Neutrophil extracellular traps that are released by dying neutrophils are proinflammatory (43). However, neutrophil apoptosis also promotes resolution of inflammation by inducing the development of antiinflammatory macrophages and by generating annexin A1 and other mediators, blocking further neutrophil re-



cruitment (22). Thus, insufficient neutrophil recruitment into the peritoneum could delay the resolution of inflammation. Consistent with that possibility, transfer of neutrophils into the peritoneum of pristane-treated TNF␣⫺/⫺ mice inhibited local type I IFN production while having little effect on type I IFN levels in the peripheral blood (Figure 4D). Finally, since type I IFN prevents Ly-6Chigh inflammatory monocytes from developing into noninflammatory peritoneal macrophages (17), dysregulated production of type I IFN in TNF␣⫺/⫺ mice may favor chronic inflammation over resolution. The mechanisms of TNFi-induced lupus are uncertain. One hypothesis is that TNF inhibitors cause apoptosis of inflammatory cells, thereby leading to the release of endogenous immunostimulatory molecules (44,45). Alternatively, TNF␣ antagonists may promote autoimmunity by increasing susceptibility to infections (46) or deviating cytokine production away from Th2 toward a Th1 profile (44,47). Finally, TNFi-induced lupus may be precipitated by removing the negative regulatory effects of TNF␣ on PDCs or other type I IFN–producing cells (9). Our findings are consistent with the latter hypothesis, but suggest that removal of TNF␣ alone is insufficient to induce autoimmunity. A hereditary or acquired inability to rapidly clear autologous TLR-7 ligands may synergize with TNF␣ blockade in TNFi-treated patients to drive type I IFN production by an expanded PDC population. Consistent with that idea, the risk of developing lupus is higher in individuals who are ANA/anti-dsDNA autoantibody positive before treatment with a TNFi (6,48,49). It may be useful to evaluate the clinical utility of pretreatment screening for autoantibodies and/or delayed clearance of apoptotic cells as a means of assessing the risk of TNFi-induced lupus. 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. Reeves 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. Xu, Zhuang, Han, Liu, Wang, Mathews, Massini, Yang, Reeves. Acquisition of data. Xu, Zhuang, Han, Liu, Wang, Mathews, Massini, Yang. Analysis and interpretation of data. Xu, Zhuang, Han, Liu, Wang, Mathews, Massini, Yang, Reeves.








9. 10. 11. 12. 13.




17. 18.



1. Baechler EC, Batliwalla FM, Karypis G, Gaffney PM, Ortmann WA, Espe KJ, et al. Interferon-inducible gene expression signa-


ture in peripheral blood cells of patients with severe lupus. Proc Natl Acad Sci U S A 2003;100:2610–5. Zhuang H, Kosboth M, Lee P, Rice A, Driscoll DJ, Zori R, et al. Lupus-like disease and high interferon levels corresponding to trisomy of the type I interferon cluster on chromosome 9p. Arthritis Rheum 2006;54:1573–9. Mathian A, Weinberg A, Gallegos M, Banchereau J, Koutouzov S. IFN-␣ induces early lethal lupus in preautoimmune (New Zealand Black ⫻ New Zealand White) F1 but not in BALB/c mice. J Immunol 2005;174:2499–506. Feldmann M, Maini RN. Anti-TNF␣ therapy of rheumatoid arthritis: what have we learned? Annu Rev Immunol 2001;19: 163–96. Williams VL, Cohen PR. TNF␣ antagonist-induced lupus-like syndrome: report and review of the literature with implications for treatment with alternative TNF␣ antagonists. Int J Dermatol 2011;50:619–25. Ramos-Casals M, Brito-Zeron P, Munoz S, Soria N, Galiana D, Bertolaccini L, et al. Autoimmune diseases induced by TNFtargeted therapies: analysis of 233 cases. Medicine (Baltimore) 2007;86:242–51. Costa MF, Said NR, Zimmermann B. Drug-induced lupus due to anti-tumor necrosis factor ␣ agents. Semin Arthritis Rheum 2008;37:381–7. Mavragani CP, Niewold TB, Moutsopoulos NM, Pillemer SR, Wahl SM, Crow MK. Augmented interferon-␣ pathway activation in patients with Sjögren’s syndrome treated with etanercept. Arthritis Rheum 2007;56:3995–4004. Palucka AK, Blanck JP, Bennett L, Pascual V, Banchereau J. Cross-regulation of TNF and IFN-␣ in autoimmune diseases. Proc Natl Acad Sci U S A 2005;102:3372–7. Xu Y, Zeumer L, Reeves WH, Morel L. Induced murine models of systemic lupus erythematosus. Methods Mol Biol 2014;1134: 103–30. Perry D, Sang A, Yin Y, Zheng YY, Morel L. Murine models of systemic lupus erythematosus. J Biomed Biotechnol 2011;2011: 271694. Xu Y, Lee PY, Li Y, Liu C, Zhuang H, Han S, et al. Pleiotropic IFN-dependent and -independent effects of IRF5 on the pathogenesis of experimental lupus. J Immunol 2012;188:4113–21. Richards HB, Satoh M, Shaw M, Libert C, Poli V, Reeves WH. Interleukin 6 dependence of anti-DNA antibody production: evidence for two pathways of autoantibody formation in pristaneinduced lupus. J Exp Med 1998;188:985–90. Li Y, Lee PY, Kellner ES, Paulus M, Switanek J, Xu Y, et al. Monocyte surface expression of Fc␥ receptor RI (CD64), a biomarker reflecting type-I interferon levels in systemic lupus erythematosus. Arthritis Res Ther 2010;12:R90. Weinstein JS, Delano MJ, Xu Y, Kelly-Scumpia KM, Nacionales DC, Li Y, et al. Maintenance of anti-Sm/RNP autoantibody production by plasma cells residing in ectopic lymphoid tissue and bone marrow memory B cells. J Immunol 2013;190:3916–27. Zhuang H, Han S, Xu Y, Li Y, Wang H, Yang LJ, et al. Toll-like receptor 7–stimulated tumor necrosis factor ␣ causes bone marrow damage in systemic lupus erythematosus. Arthritis Rheumatol 2014;66:140–51. Lee PY, Li Y, Kumagai Y, Xu Y, Weinstein JS, Kellner ES, et al. Type I interferon modulates monocyte recruitment and maturation in chronic inflammation. Am J Pathol 2009;175:2023–33. Nacionales DC, Kelly-Scumpia KM, Lee PY, Weinstein JS, Lyons R, Sobel E, et al. Deficiency of the type I interferon receptor protects mice from experimental lupus. Arthritis Rheum 2007;56: 3770–83. Reeves WH, Lee PY, Weinstein JS, Satoh M, Lu L. Induction of autoimmunity by pristane and other naturally occurring hydrocarbons. Trends Immunol 2009;30:455–64. Lee PY, Weinstein JS, Nacionales DC, Scumpia PO, Li Y,


21. 22. 23. 24. 25. 26. 27.

28. 29. 30.



33. 34. 35.

Butfiloski E, et al. A novel type I IFN-producing cell subset in murine lupus. J Immunol 2008;180:5101–8. Lee PY, Kumagai Y, Xu Y, Li Y, Barker T, Liu C, et al. IL-1␣ modulates neutrophil recruitment in chronic inflammation induced by hydrocarbon oil. J Immunol 2011;186:1747–54. Ortega-Gomez A, Perretti M, Soehnlein O. Resolution of inflammation: an integrated view. EMBO Mol Med 2013;5:661–74. Liu YJ. IPC: professional type 1 interferon-producing cells and plasmacytoid dendritic cell precursors. Annu Rev Immunol 2005; 23:275–306. Ghosh HS, Cisse B, Bunin A, Lewis KL, Reizis B. Continuous expression of the transcription factor E2-2 maintains the cell fate of mature plasmacytoid dendritic cells. Immunity 2010;33:905–16. Fukui Y, Hashimoto O, Sanui T, Oono T, Koga H, Abe M, et al. Haematopoietic cell-specific CDM family protein DOCK2 is essential for lymphocyte migration. Nature 2001;412:826–31. Reif K, Cyster J. The CDM protein DOCK2 in lymphocyte migration. Trends Cell Biol 2002;12:368–73. Gotoh K, Tanaka Y, Nishikimi A, Nakamura R, Yamada H, Maeda N, et al. Selective control of type I IFN induction by the Rac activator DOCK2 during TLR-mediated plasmacytoid dendritic cell activation. J Exp Med 2010;207:721–30. Aringer M, Smolen JS. Efficacy and safety of TNF-blocker therapy in systemic lupus erythematosus. Expert Opin Drug Saf 2008;7: 411–9. Modlin RL. The innate immune response in leprosy. Curr Opin Immunol 2010;22:48–54. Pasparakis M, Alexopoulou L, Episkopou V, Kollias G. Immune and inflammatory responses in TNF␣-deficient mice: a critical requirement for TNF␣ in the formation of primary B cell follicles, follicular dendritic cell networks and germinal centers, and in the maturation of the humoral immune response. J Exp Med 1996; 184:1397–411. Cook MC, Korner H, Riminton DS, Lemckert FA, Hasbold J, Amesbury M, et al. Generation of splenic follicular structure and B cell movement in tumor necrosis factor-deficient mice. J Exp Med 1998;188:1503–10. Nacionales DC, Weinstein JS, Yan XJ, Albesiano E, Lee PY, Kelly-Scumpia KM, et al. B cell proliferation, somatic hypermutation, class switch recombination, and autoantibody production in ectopic lymphoid tissue in murine lupus. J Immunol 2009;182: 4226–36. William J, Euler C, Christensen S, Shlomchik MJ. Evolution of autoantibody responses via somatic hypermutation outside of germinal centers. Science 2002;297:2066–70. Shlomchik MJ. Sites and stages of autoreactive B cell activation and regulation. Immunity 2008;28:18–28. Odegard JM, Marks BR, DiPlacido LD, Poholek AC, Kono DH, Dong C, et al. ICOS-dependent extrafollicular helper T cells elicit IgG production via IL-21 in systemic autoimmunity. J Exp Med 2008;205:2873–86.


36. Gilliet M, Cao W, Liu YJ. Plasmacytoid dendritic cells: sensing nucleic acids in viral infection and autoimmune diseases. Nat Rev Immunol 2008;8:594–606. 37. Lee PY, Kumagai Y, Li Y, Takeuchi O, Yoshida H, Weinstein J, et al. TLR7-dependent and Fc␥R-independent production of type I interferon in experimental mouse lupus. J Exp Med 2008;205: 2995–3006. 38. Tai LH, Goulet ML, Belanger S, Toyama-Sorimachi N, FodilCornu N, Vidal SM, et al. Positive regulation of plasmacytoid dendritic cell function via Ly49Q recognition of class I MHC. J Exp Med 2008;205:3187–99. 39. Omatsu Y, Iyoda T, Kimura Y, Maki A, Ishimori M, ToyamaSorimachi N, et al. Development of murine plasmacytoid dendritic cells defined by increased expression of an inhibitory NK receptor, Ly49Q. J Immunol 2005;174:6657–62. 40. Sasaki I, Hoshino K, Sugiyama T, Yamazaki C, Yano T, Iizuka A, et al. Spi-B is critical for plasmacytoid dendritic cell function and development. Blood 2012;120:4733–43. 41. Liu Y, Mei J, Gonzales L, Yang G, Dai N, Wang P, et al. IL-17A and TNF-␣ exert synergistic effects on expression of CXCL5 by alveolar type II cells in vivo and in vitro. J Immunol 2011;186: 3197–205. 42. Griffin GK, Newton G, Tarrio ML, Bu DX, Maganto-Garcia E, Azcutia V, et al. IL-17 and TNF-␣ sustain neutrophil recruitment during inflammation through synergistic effects on endothelial activation. J Immunol 2012;188:6287–99. 43. Knight JS, Kaplan MJ. Lupus neutrophils: ‘NET’ gain in understanding lupus pathogenesis. Curr Opin Rheumatol 2012;24: 441–50. 44. Bout-Tabaku S, Rivas-Chacon R, Restrepo R. Systemic lupus erythematosus in a patient treated with etanercept for polyarticular juvenile rheumatoid arthritis. J Rheumatol 2007;34:2503–4. 45. D’Auria F, Rovere-Querini P, Giazzon M, Ajello P, Baldissera E, Manfredi AA, et al. Accumulation of plasma nucleosomes upon treatment with anti-tumour necrosis factor-␣ antibodies. J Intern Med 2004;255:409–18. 46. Ferraccioli G, Mecchia F, Di Poi E, Fabris M. Anticardiolipin antibodies in rheumatoid patients treated with etanercept or conventional combination therapy: direct and indirect evidence for a possible association with infections. Ann Rheum Dis 2002;61: 358–61. 47. Steiner G, Smolen J. Autoantibodies in rheumatoid arthritis and their clinical significance. Arthritis Res 2002;4 Suppl 2:S1–5. 48. De Bandt M, Sibilia J, Le Loet X, Prouzeau S, Fautrel B, Marcelli C, et al. Systemic lupus erythematosus induced by anti-tumour necrosis factor ␣ therapy: a French national survey. Arthritis Res Ther 2005;7:R545–51. 49. Wetter DA, Davis MD. Lupus-like syndrome attributable to anti-tumor necrosis factor ␣ therapy in 14 patients during an 8-year period at Mayo Clinic. Mayo Clin Proc 2009;84:979–84.

Copyright of Arthritis & Rheumatology is the property of John Wiley & Sons, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Mechanisms of tumor necrosis factor α antagonist-induced lupus in a murine model.

Tumor necrosis factor α (TNFα) antagonists are effective for treating rheumatoid arthritis and other inflammatory diseases, but their use can be compl...
1MB Sizes 2 Downloads 3 Views