A Novel Type I IFN-Producing Cell Subset in Murine Lupus¹

Systemic lupus erythematosus (SLE)³ is a chronic autoimmune disorder affecting multiple organs including the skin, joints, kidneys, lungs, heart, and the nervous system (1). Antinuclear Abs against small nuclear ribonucleoprotein (RNP) and dsDNA are pathognomonic of the disease (1). Recent evidence suggests that type I IFNs (IFN-I), a family of antiviral cytokines, are integral to the pathogenesis of SLE. More than two-thirds of SLE patients display up-regulation of a group of type I IFN-stimulated genes (ISGs) (2, 3, 4). This “IFN signature” is clinically relevant as it correlates with active disease, presence of certain autoantibodies, and an increased incidence of renal involvement (4, 5). Supporting a causal role of IFN-I, therapeutic use of recombinant IFN-α is linked to a wide array of autoimmune manifestations and there are reports of SLE following treatment (6). Although an association between elevated IFN-I levels and SLE is well established, its origin is unclear.

Experimental lupus induced by 2,6,10,14-tetramethylpentadecane (TMPD; Sigma-Aldrich) displays key immunologic and clinical features of human SLE, including the production of autoantibodies against dsDNA and small nuclear RNPs, and the development of immune complex-mediated glomerulonephritis and arthritis (7). We recently reported that TMPD-induced lupus is associated with excess IFN-I production and up-regulation of ISGs (8). This murine model is at present the only one reported to have the “IFN signature”. IFN-I signaling is critically required in this model as shown by the absence of lupus autoantibodies and kidney disease in IFN-I receptor α-chain-deficient (IFNAR^−/−) mice) (9). The source of the excess IFN-I, however, has not been examined. In this study, we aimed to identify and characterize the cell population responsible for increased IFN-I production in TMPD-induced lupus.

Wild-type BALB/cJ, and B6.FVB-Tg^{.Itgax-DTR/EGFP.57}Lan/J backcrossed to a BALB/c background, referred to as CD11c-DTR (diphtheria toxin receptor) mice (10, 11), were purchased from The Jackson Laboratory and housed in a conventional facility. CD11c-DTR mice were maintained as heterozygote crosses and littermates not expressing the transgene were used as controls. 129Sv/Ev IFNAR^−/− mice and wild-type controls (129Sv/Ev) were purchased from B&K Universal. Eight-week-old mice received a single i.p. injection of 0.5 ml of TMPD, mineral oil (Harris Teeter), 4% thioglycolate (BD Biosciences), squalene (Sigma-Aldrich), n-hexadecane (Sigma-Aldrich), or PBS. Peritoneal cell isolation and cecal ligation and puncture were performed as described (8, 11). These studies were approved by the Institutional Animal Care and Use Committee.

Real-time PCR and conventional PCR were performed as described (8). Briefly, total RNA was extracted from 10⁶ peritoneal cells using TRIzol reagent (Invitrogen Life Technologies) and cDNA was synthesized using Superscript II First-Strand Synthesis kit (Invitrogen Life Technologies). Real-time PCR was performed using the SYBR Green Core Reagent kit (Applied Biosystems) with an Opticon II thermocycler (MJ Research). The following amplification conditions were used: 95°C for 10 min, followed by 45 cycles of 94°C for 15 s, 60°C for 25 s, and 72°C for 25 s. After the final extension (72°C for 10 min), a melting-curve analysis was performed to ensure specificity of the products. IFN-I genes were amplified by conventional PCR using a PTC-100 programmable thermal controller (MJ Research). The amplification conditions were 95°C for 5 min, followed by 40 cycles of 94°C for 30 s, 60°C for 1 min, and 72°C for 1 min. After a final extension (72°C for 10 min), PCR products were analyzed by agarose gel electrophoresis. Primers used in this study include the following: myxoma resistance protein-1 Mx-1 (forward) GATCCGACTTCACTTCCAGATGG, (reverse) CATCTCAGTGGTAGTCAACCC; MCP-1 (forward) AGGTCCCTGTCATGCTTCTG, (reverse) GGATCATCTTGCTGGTGAAT; IP-10 (forward) CCTGCAGGATGATGGTCAAG, (reverse) GAATTCTTGCTTCGGCAGTT; consensus IFN-α (forward) ATGGCTAGRCTCTGTGCTTTCCT, (reverse) AGGGCTCTCCAGAYTTCTGCTCTG; IFN-α5 (forward) tgtgaccttcttcagactc, (reverse) CTCCTCCTTGCTCAATC; IFN-β (forward) AGCTCCAAGAAAGGACGAACAT, (reverse) ATTCTTGCTTCGGCAGTTAC; TNF-α (forward) GGCAGGTCTACTTTGGAGTCATTGC, (reverse) ACATTCGAGGCTGCTCCAGTGAATTCGG; inducible NO synthase (iNOS) (forward) ATCGACCCGTCCACAGTATG, (reverse) GATGGACCCCAAGCAAGACT; IL-12p40 (forward) GAGTGGACTGGACTCCCGA, (reverse) CAAGTTCTTGGGCGGGTCTG; β-actin (forward) CCCACACTGTGCCCATCTAC, (reverse) CGCTCGGTCAGGATCTTCAT); and 18 S RNA (forward) CGGCTACCACATCCAAGGAA, (reverse) GCTGGAATTACCGCGGCT.

All Abs were purchased from BD Biosciences with the exceptions of anti-Ly6C-FITC, anti-Ly6C-biotin, avidin-allophycocyanin (eBioscience), anti-F4/80-FITC, anti-Moma2-FITC (Serotec), and anti-CD11b-Pacific blue (Caltag Laboratories). Cell staining was performed as described (8). Propidium iodide (Invitrogen Life Technologies) staining was performed following the manufacturer’s protocol. Fifty thousand events per sample were acquired by a Cyan ADP flow cytometer (DakoCytomation) and analyzed with FCS Express 3 (De Novo Software).

Peritoneal cells from TMPD-treated mice (10⁷) were stained with anti-Ly6G-PE, washed, and incubated with magnetic bead-conjugated anti-PE (Miltenyi Biotec). Granulocytes (>99% purity) were positively selected using MS columns, whereas the negative fraction was stained with anti-CD11b-allophycocyanin, washed, and incubated with magnetic bead-conjugated anti-allophycocyanin (Miltenyi Biotec). Positive selection using MS columns yielded >85% Ly6C^high monocytes. The negative fraction consisted of lymphocytes and dendritic cells (DCs). Cell sorting using a FACSDiva flow cytometer (BD Biosciences) yielded similar results with a higher purity of Ly6C^high monocytes (>95%). For morphological analysis, 3 × 10⁴ sorted cells were cytospun onto glass slides and stained using the Hema3 kit (modified Wright stain; Fisher Scientific).

Clodronate-containing liposomes (clo-liposomes), a gift from Roche Diagnostics, and DiD (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate) liposomes were produced as described (12, 13). To label Ly6C⁻ monocytes, 150 μl of DiD-liposomes were injected i.v. into TMPD-treated mice. Clo-liposomes (200 μl) were delivered i.v. 24 h before the injection of DiD-liposomes for the labeling of Ly6C^high monocytes. Analysis was performed 24 h after labeling. To deplete peritoneal monocytes, 200 μl of clo-liposomes were injected i.p. into wild-type BALB/c mice treated with TMPD 2 wk earlier.

DC ablation was performed in TMPD-treated CD11c-DTR mice by i.p. injecting 4 ng/g of body weight of diphtheria toxin (Sigma-Aldrich). Mice were sacrificed for analysis 48 h after clo-liposomes or diphtheria toxin administration.

For quantitative variables, differences between groups were analyzed by the Student’s unpaired t test. Bivariate correlations were assessed using Spearman’s correlation coefficient. Data were presented as mean ± SD. All tests were two-sided, and a value for p < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism 4.0 (GraphPad Software).

Although the link between SLE and IFN-I is firmly established, it remains unclear whether the IFN signature precedes disease manifestations. In the TMPD model of SLE, production of lupus autoantibodies (anti-Smith Ag (-Sm), antinuclear RNP (-nRNP), and anti-dsDNA) and development of immune complex-mediated glomerulonephritis occur ∼4–6 mo after peritoneal exposure to the hydrocarbon oil (7). Up-regulation of IFN-α and IFN-β, conversely, was observed in peritoneal exudate cells within 2 wk of TMPD treatment (Fig. 1,A). Although consensus PCR primers can amplify most IFN-α subtypes, quantification of ISG expression allows assessment of total IFN-I proteins because all IFN-I subtypes bind to single receptor complex to activate ISG expression. Accordingly, a panel of ISGs including Mx-1, MCP-1/CCL2, and IP-10/CXCL10 were also highly up-regulated (Fig. 1 B) 2 wk after TMPD treatment. In contrast, ISG expression was only modestly increased in mice treated with mineral oil, a control hydrocarbon oil that triggers chronic inflammation without features of lupus. Supporting the gene expression data, elevated levels of IFN-β and MCP-1 protein along with the inflammatory cytokines TNF-α and IL-12 were demonstrated in peritoneal lavage fluid of TMPD-treated mice (data not shown). Thus, the IFN signature induced by TMPD was established within 2 wk of treatment, long before the appearance of lupus autoantibodies and kidney pathology.

To identify the source of IFN-I production, we analyzed different cell populations in the peritoneal exudate 2 wk following TMPD treatment. More than 80% of the cells in the inflammatory exudate were CD11b⁺ (data not shown). Because CD11b is expressed by monocytes/macrophages, granulocytes, and peritoneal B1 cells, we used the marker Ly6C to distinguish these populations (14, 15). In untreated or PBS-treated animals, resting peritoneal cells predominantly consisted of macrophages (CD11b^high Ly6C⁻) and B1 cells (CD11b^mid Ly6C⁻) (Fig. 1,C). As previously described (14), peritoneal macrophages expressed the F4/80 Ag, whereas B1 cells were positive for the B cell markers B220 and CD5 (data not shown). Interestingly, both resting cell types were depleted 2 wk following TMPD treatment, whereas two distinct populations emerged: CD11b⁺ Ly6C^high cells (R1), consistent with the phenotype of immature monocytes (15), and CD11b⁺ Ly6C^mid cells (R2) characteristic of granulocytes (Fig. 1, C and D). The latter cell population (R2) also expressed the neutrophil marker Ly6G (data not shown). These populations each represented ∼30% of cells in the peritoneal inflammatory infiltrate. Morphology of the monocyte and granulocyte populations was consistent with their surface marker profile (Fig. 1 E). These findings were consistent in other wild-type strains (C57BL/6 and BALB/c), except that these strains showed a greater influx of granulocytes (data not shown).

Treatment with mineral oil, which does not induce the IFN signature, resulted in similar accumulation of granulocytes, but few Ly6C^high monocytes were present (Fig. 1, C and D). Instead, mature monocytes resembling resting peritoneal macrophages were found (R3) (Fig. 1, C and D). Compared with the Ly6C^high monocytes, these cells were larger, more vacuolated, and displayed rounded nuclei, suggestive of a more differentiated phenotype (Fig. 1,E). The accumulation of Ly6C^high monocytes was specific to TMPD treatment as this pattern was not seen in sterile peritonitis induced by thioglycolate or septic peritonitis induced by cecal ligation and puncture (Fig. 1 C).

Flow cytometric analysis of Ly6C^high monocytes from TMPD-treated animals demonstrated intense expression of the myeloid markers Mac-3 and Moma-2 (Fig. 1,F). Supporting their immature phenotype, only a small fraction of Ly6C^high monocytes expressed the macrophage marker F4/80, MHC class II (I-A), and the costimulatory molecules CD80 and CD86 (Fig. 1,F). This result was in sharp contrast to mature peritoneal macrophages, which expressed these markers at high levels (data not shown). Ly6C^high monocytes also lacked markers of DCs (CD11c and CD205), B cells (B220 and CD19), T cells (CD3), granulocytes (Ly6G), or NK cells (Pan-NK) (Fig. 1,F and data not shown). The absence of B220 and CD11c expression distinguishes these cells from the previously described natural IFN-producing cells, which also express Ly6C, albeit at much lower levels compared with Ly6C^high monocytes (16, 17). Curiously, these monocytes exhibited strong expression of Sca-1 (Ly6A/E) (Fig. 1 F), a marker normally found on hematopoietic stem cells and certain T cell subsets (18).

Although the accumulation of Ly6C^high monocytes was evident after 2 wk, these cells appeared in the peritoneal cavity as early as 1 day after TMPD treatment (Fig. 1,G). However, this observation was not limited to TMPD as a similar pattern of acute peritoneal inflammation was elicited by mineral oil (Fig. 1,G). But the appearance of Ly6C^high monocytes in response to mineral oil was transient compared with TMPD. Only a small population of Ly6C^high monocytes was found in the peritoneal cavity 2 wk after mineral oil treatment, whereas the response was maintained for several months with TMPD treatment (Fig. 1 C and data not shown).

To investigate the source of the accumulating Ly6C^high monocytes, we first performed a cell cycle analysis to determine whether these cells were proliferating in the peritoneal cavity. Propidium iodide staining revealed that virtually all Ly6C^high monocytes in the peritoneal exudate elicited by TMPD were in G₁ phase (Fig. 2 A), indicating that extramedullary myelopoiesis at the site of inflammation is not a likely explanation for the monocyte accumulation.

Recent studies have demonstrated that Ly6C^high monocytes egress from the bone marrow and that expression of Ly6C diminishes as they mature in the circulation (12). We therefore examined whether TMPD treatment alters the maturation profile of peripheral blood monocytes. In untreated animals, mature circulating monocytes (Ly6C⁻) outnumbered their immature Ly6C^high counterparts by ∼2:1 (Fig. 2,B). Two weeks after TMPD treatment, the frequency of Ly6C^high monocytes in the peripheral blood doubled, whereas the Ly6C⁻ subset remained constant (Fig. 2 B).

To demonstrate the migration of Ly6C^high monocytes into the peritoneal cavity following TMPD injection, we selectively labeled monocyte subsets in vivo. As previously described (12), i.v. injection of liposomes containing the fluorescent dye DiD labeled strictly Ly6C⁻ monocytes in the circulation (Fig. 2,C, left). Consistent with the paucity of Ly6C⁻ monocytes in the response to TMPD, DiD-positive cells were not recovered from the peritoneal cavity. Using clo-liposomes first to deplete mature monocytes (12), circulating Ly6C^high monocytes were specifically labeled by the subsequent administration of DiD-liposomes (Fig. 2 C, right). After 24 h, more than one-third of Ly6C^high monocytes in the peritoneal cavity were DiD-positive, indicating that TMPD treatment induces rapid and specific recruitment of this monocyte subset. Taken together, these data suggest that TMPD treatment results in the export of Ly6C^high monocytes from the bone marrow into the circulation and followed by their specific recruitment and accumulation in the peritoneal cavity.

Using magnetic bead sorting, we separated peritoneal exudate cells from TMPD-treated mice into Ly6C^high monocytes, Ly6G⁺ granulocytes, and a negative fraction consisting of lymphocytes and DCs. Although the Ly6C^high monocyte fraction contained a small percentage of contaminating Ly6G⁺ granulocytes, DCs were found only in the negative fraction (data not shown). PCR analysis revealed that Ly6C^high monocytes were the predominant source of IFN-α and IFN-β expression (Fig. 3,A). IFN-I transcripts also were detected in other populations, albeit at significantly lower levels (Fig. 3,A). Ly6C^high monocytes expressed high levels of the IFN-stimulated chemokine MCP-1 and moderate levels of TNF-α and IL-12 (Fig. 3,B) as well. Granulocytes accounted for the remaining TNF-α transcripts and the majority of iNOS expression. Not surprisingly, IL-12 transcripts were found predominantly in cells of the negative fraction because this cytokine was most likely derived from DCs (Fig. 3 B). Similar results were obtained when this experiment was repeated using flow cytometric cell sorting to increase the purity of sorted populations (data not shown).

To confirm that Ly6C^high monocytes were major producers of IFN-I, we asked whether their depletion abolishes the IFN signature. Treatment with clo-liposomes effectively eliminates monocytes in vivo (13). Indeed, a single dose of clo-liposomes i.p. was sufficient to eliminate ∼80% of peritoneal Ly6C^high monocytes in mice pretreated with TMPD (Fig. 3 C). There were also slight reductions in the number of DCs and lymphocytes (data not shown), whereas the number of CD11b⁺Ly6C^midLy6G⁺ granulocytes was unaffected by clo-liposome treatment.

Concomitant with the depletion of Ly6C^high monocytes, the expression of IFN-α, IFN-β, and ISGs was drastically reduced (Fig. 3, D and E). Similarly, the expression of TNF-α, which was highly expressed by Ly6C^high monocytes, also diminished upon their depletion. The expression of IL-12, which was expressed mainly by the negative cell fraction comprised of lymphocytes and DCs, did not change significantly after clo-liposome treatment (Fig. 3 E). The effect of clo-liposomes was transient as the number of Ly6C^high monocytes and the expression of ISGs returned to pretreatment levels after 4 days (data not shown).

Plasmacytoid DCs (PDCs) are capable of secreting large amounts of IFN-I during viral infection and are thought to be primary IFN producers in SLE (17, 19). In the peritoneal cavity of TMPD-treated animals, CD11c⁺I-A⁺ DCs comprised ∼2% of the infiltrating inflammatory cells. Most peritoneal DCs expressed CD11b but not B220 (Fig. 4 A), consistent with the phenotype of myeloid DCs (MDCs). However, PDCs may home to other secondary lymphoid tissue following activation (16, 17). To elucidate the extent to which DCs contribute to IFN-I production in the TMPD model, we used transgenic mice carrying the simian DTR under the control of the CD11c promoter (CD11c-DTR) (10). Injection of diphtheria toxin rapidly ablates both PDCs and MDCs systemically in CD11c-DTR mice, whereas wild-type mice are unaffected by the toxin (10).

Two days following diphtheria toxin injection, TMPD-treated CD11c-DTR mice showed >85% depletion of CD11c⁺ I-A⁺ DCs in the peritoneal exudate compared with wild-type controls (Fig. 4, B and C). In line with previous reports (11, 20), DC depletion was systemic as splenic MDCs and PDCs were also depleted by 70–80% (Fig. 4, B and C). In contrast, there was no significant difference in the peritoneal accumulation of Ly6C^high monocytes, granulocytes, and lymphocytes (Fig. 4,C and data not shown). Both CD11c^highCD11b⁺I-A⁺ MDCs and CD11c⁺B220⁺PDCA-1⁺ PDCs were depleted to a similar degree in the spleen (Fig. 4,D) and lymph nodes (data not shown). Systemic depletion of DCs did not affect TMPD-induced IFN-I production as the expression of IFN-I and ISGs were unaffected in CD11c-DTR animals (Fig. 4, E and F). The expression of TNF-α and iNOS was also unchanged (data not shown). In contrast, IL-12 expression was drastically reduced in the absence of DCs (Fig. 4,F), consistent with the cell sorting experiment (Fig. 3 E). Taken together, these data indicate that DCs were the primary source of IL-12 but not IFN-I.

We also tried to deplete PDCs using the recently described PDC-specific Ab 120G8 (21). Treatment with 120G8 i.p. resulted in ∼70% depletion of splenic PDCs after 24 h, comparable to the levels seen in CD11c-DTR mice. However, peritoneal Ly6C^high monocytes and T lymphocytes were also reduced by >50% (data not shown). Although the Ag bound by 120G8 and PDC Ag-1 is normally expressed on PDCs, its expression can be induced by IFN-I in other cell types (21, 22). Indeed, elevated IFN-I production in TMPD mice is associated with the expression of this PDC Ag on Ly6C^high monocytes, granulocytes and T lymphocytes (data not shown), making selective Ab-mediated depletion of PDCs unfeasible in this model.

Expression of ISGs is associated with autoantibodies against Sm/RNP in SLE patients (4, 5). In the TMPD model, 60–70% of treated BALB/c mice exhibit anti-Sm/RNP Abs after 4–6 mo vs 0% of IFNAR^−/− mice (9). These autoantibodies either appear less frequently or are completely absent in mice treated with other adjuvant oils such as n-hexadecane, squalene, or mineral oil (23). Interestingly, treatment with n-hexadecane (which induces anti-Sm/RNP in ∼25% of treated animals) elicited the accumulation of Ly6C^high monocytes, albeit the response was milder than with TMPD, whereas squalene (which induces a weak anti-Sm/RNP response in <10% of treated animals) recruited mostly mature monocytes/macrophages and few Ly6C^high monocytes, resembling the pattern seen with mineral oil (Fig. 5,A). Because Ly6C^high monocytes were the major source of IFN-I in TMPD-treated mice, we examined whether the frequency of anti-Sm/RNP autoantibodies correlates with the number of these cells. Indeed, the ability of adjuvant oils to elicit anti-Sm/RNP Abs was highly correlated (R² = 0.98) with the accumulation of Ly6C^high monocytes (Fig. 5,B). Numbers of Ly6C^high monocytes also correlated with ISG expression (Fig. 5 C), supporting our finding that these cells are major IFN-I producers. In contrast, the recruitment of granulocytes was similar among all treatment groups and more DCs were present in the peritoneal exudate cells following mineral oil or squalene treatment than TMPD (data not shown).

Because Ly6C is an IFN-inducible gene (24), it is possible that IFN-I production triggered by TMPD treatment contributes to the Ly6C^high monocyte phenotype. To address this issue, we analyzed monocyte subsets in IFNAR^−/− mice. Compared with wild-type controls, untreated IFNAR^−/− mice showed similar levels of Ly6C^high and Ly6C⁻ monocytes in the circulation, accounting for ∼2% and 5% of PBMC, respectively (data not shown). Two weeks following TMPD treatment, IFNAR^−/− mice also displayed an increased number of circulating Ly6C^high monocytes, albeit the response was milder than that of wild-type controls (Fig. 6). Importantly, Ly6C^high and Ly6C⁻ monocytes, as well as the Ly6C^mid granulocyte populations in the peripheral blood were clearly discernible even in the absence of IFN-I signaling. The number of bone marrow precursors of Ly6C^high monocytes was also comparable between wild-type and IFNAR^−/− mice (Fig. 6). These data suggest that although IFN-I has been shown to induce Ly6C expression (24), the Ly6C^high phenotype of the immature monocyte population in naive or TMPD-treated mice was not dependent on IFN-I signaling.

Elevated serum IFN-I was first associated with SLE over two decades ago (25). Recent studies using microarrays and real-time PCR further defined a panel of ISGs overexpressed in the peripheral blood of SLE patients (2, 3, 4). Although this IFN signature has been linked to disease activity, kidney involvement, and autoantibody levels, the source of IFN-I responsible for the IFN signature remains a matter of speculation. This issue has been difficult to address using animal models because most of them do not exhibit up-regulation of IFN-I.

We recently reported that murine lupus induced by TMPD is associated with elevated IFN-I production and ISG expression (8). Disruption of IFN-I signaling completely abrogates the development of kidney disease and the onset of autoantibody production (9). In this study, we show that Ly6C^high monocytes are a major source of IFN-I in the TMPD model of lupus. Up-regulation of IFN-I and ISGs occurred long before the clinical manifestations of lupus and coincided with an accumulation of Ly6C^high monocytes. These immature monocytes expressed large amounts of IFN-I and the IFN signature was rapidly abolished upon depletion of these cells by clo-liposomes. Moreover, the abundance of Ly6C^high monocytes was highly associated with the ability of various adjuvant oils to induce anti-Sm/RNP Abs. In contrast, systemic depletion of DCs did not alter IFN-I production triggered by TMPD.

Although Ly6C^high monocytes were found in the inflammatory infiltrate 1 day after thioglycolate (12, 13) or mineral oil administration, the response was transient as the number of these cells was drastically reduced after 72 h. The recruitment and accumulation of Ly6C^high monocytes seen in TMPD-treated animals, in contrast, persisted for as long as 4 mo after treatment. A recent study shows that the egression of Ly6C^high monocytes from the bone marrow in response to Listeria monocytogenes infection is dependent on the interaction between the monocyte attractant MCP-1/CCL2 and its receptor CCR2 (26). Interestingly, as confirmed in this study, MCP-1 is an ISG that is induced by TMPD treatment, raising the possibility that the production of IFN-inducible chemokines may drive the recruitment of Ly6C^high monocytes. This mechanism is relevant to human SLE as elevated serum levels of MCP-1 have been associated with the IFN signature (3). How TMPD treatment maintains the infiltrating monocytes in an immature state is unknown, but a defect intrinsic to the cells is unlikely as Ly6C^high monocytes spontaneously acquired a mature macrophage-like phenotype (Ly6C⁻ F4/80⁺ I-A⁺) in vitro (data not shown). The immaturity of Ly6C^high monocytes may be important to their ability to produce IFN-I as their mature counterparts elicited by mineral oil displayed significantly lower levels. Analogously, PDCs also are better equipped to secrete large amounts of IFN-I when immature (27).

Prolonged elevation of IFN-I may promote DC maturation (28), T cell survival (29), isotype class-switching (30), and B cell maturation into plasma cells (31), culminating in the loss of tolerance and autoantibody production. It is noteworthy that although IFN-I is essential to TMPD-induced lupus (9), other factors such as IL-12 and IFN-γ play important roles because autoantibody production and kidney disease development are reduced in deficient mice (32, 33).

Our findings challenge the conventional view that PDCs are solely responsible for the elevated IFN-I in SLE. In TMPD-treated mice, the vast majority of the increased IFN-I production is derived from the 20–30% of peritoneal cells that are Ly6C^high CD11b⁺B220⁻CD11c⁻ monocytes. Although, PDCs can synthesize up to 1000-fold more IFN-α than most other cell types (34), the number of circulating PDCs is drastically reduced in SLE patients (4, 35). Although it is plausible that PDCs home to tissues following activation, a view supported by the presence of these cells in lupus skin lesions (36), there is little direct evidence connecting tissue PDCs to the excess serum IFN-I seen in SLE patients. Indeed, as tissue (e.g., spleen) PDCs were greatly reduced by diphtheria toxin treatment of CD11c-DTR mice, despite little effect on ISG expression, it is unlikely that PDCs were the main source of IFN-I in TMPD-induced lupus. Also, the elevation of IFN-I in this model occurs within 2 wk of treatment, long before the appearance of autoantibodies against dsDNA and small nuclear RNP and formation of immune complexes. The proposed mechanism of IFN-I induction in PDCs by endogenous nucleic acids present in immune complexes, therefore, is not a likely explanation of our findings (19). Nevertheless, PDCs may function to amplify IFN-I production once autoantibodies and immune complexes develop. MDCs also play a role in TMPD-induced lupus because they are the major source of IL-12, a cytokine critical for the development of kidney disease in this model (32). The role of DCs in human SLE is more difficult to assess because targeted DC therapy is not yet available.

It is noteworthy that Ly6C^high monocytes also have been recently reported to play a role in atherosclerosis (37). Therefore, these monocyte-like IFN-producing cells could play a role in the premature atherosclerosis seen in SLE patients. A CD14^highCD16⁻ monocyte subset (also called “classical monocytes”) is the equivalent of murine Ly6C^high monocytes in terms of migratory properties (15). Whether this subset or other monocyte subsets produce IFN-I in human SLE warrants detailed investigation.

We thank Neal Benson and the University of Florida Flow Cytometry Core (Gainesville, FL) for assistance with cell sorting.

The authors have no financial conflict of interest.