To investigate the ability of systemic lupus erythematosus (SLE) autoantigen– and Sjögren's syndrome (SS) autoantigen–associated U1 small nuclear RNA (U1 snRNA) and hY1RNA to induce interferon-α (IFNα) production.
To investigate the ability of systemic lupus erythematosus (SLE) autoantigen– and Sjögren's syndrome (SS) autoantigen–associated U1 small nuclear RNA (U1 snRNA) and hY1RNA to induce interferon-α (IFNα) production.
In vitro–transcribed U1 snRNA or hY1RNA and lipofectin were added to peripheral blood mononuclear cell (PBMC) cultures. Purified U1 snRNP particles and IgG from SLE patients (SLE-IgG) were added to cultures of PBMCs, enriched monocytes, or natural interferon–producing cells (NIPCs); the latter are also known as plasmacytoid dendritic cells (pDC). Cells were double-stained for IFNα and either blood dendritic cell antigen 2 (NIPCs/pDC) or CD14 (monocytes) and then analyzed by flow cytometry. In some experiments, RNase or inhibitors of Fcγ receptor IIa (FcγRIIa) (specific antibodies), endocytosis (chloroquine, bafilomycin A), or Toll-like receptors (TLRs; oligodeoxynucleotide 2088) were used. The produced IFNα was measured by immunoassay.
Lipofected U1 snRNA and hY1RNA both induced IFNα production in monocytes, but not in NIPC/pDC. In contrast, U1 snRNP combined with SLE-IgG induced IFNα production only in NIPCs/pDC, and this response was decreased by RNase treatment or inhibition of the FcγRIIa, the endocytosis pathways, or the TLRs.
Our finding that U1 snRNA and hY1RNA have IFNα-inducing capacity indicates that immune complexes containing such RNA, for example U1 snRNP particles, can be at least partly responsible for the ongoing IFNα production seen in SLE and SS. These results may help to explain the molecular mechanisms behind the pathogenesis of these and other autoimmune diseases in which autoantibodies to RNA- binding proteins occur.
Systemic lupus erythematosus (SLE) is a multisystem autoimmune disease that is characterized by immune complexes (ICs) in the circulation and IC deposits in tissues. The formation of immune complexes is attributable to production of autoantibodies to a wide variety of autoantigens, many of which are complexes between proteins and nucleic acids. Examples of such autoantigens are chromatin and nucleosomes, SSA/Ro and SSB/La proteins that associate with hYRNA, as well as small nuclear RNP (snRNP) such as U1 snRNP 70K, A, and C, and the group of Sm proteins that form complexes with the U series of snRNA (1–3).
It has been proposed that the type I interferon (IFN) system plays a pivotal etiopathogenic role in SLE (4–6). The evidence for this includes findings of ongoing production of type I IFN in SLE patients, which correlates with disease activity and severity (7–11). Many patients have detectable IFNα levels in blood, or more commonly, increased expression of type I IFN–inducible genes at the tissue level, termed the IFN signature. Furthermore, IFNα therapy for malignancies and viral infections is frequently associated with the appearance of various manifestations of autoimmunity, including production of anti–double-stranded DNA (anti-dsDNA) antibodies and SLE-like disease (12–15). Recently, single-nucleotide polymorphisms in 2 genes with major function in type I IFN production and action, interferon regulatory factor 5 and tyrosine kinase 2, were found to be strongly associated with human SLE (16). Finally, an important role for type I IFN in murine NZB, NZB/NZW, and C57BL/6-lpr/lpr experimental models of SLE has also been reported (6, 17).
A role for type I IFN, especially IFNα, in SLE is most likely related to the immunomodulatory effects of these cytokines, which include promotion of survival, differentiation, and function of key cells (including myeloid and plasmacytoid dendritic cells (pDC), and cytotoxic, Th1, and B lymphocytes) in immune responses (6, 18). Moreover, type I IFNs have many other effects on cells, including increased expression of class I and class II major histocompatibility complex molecules, regulation of cell migration via induced expression of chemokines (6, 18), and increased expression of SLE-related autoantigens (19). Consequently, the ongoing production of IFNα in SLE can stimulate production of autoimmune effector T cells and antibodies, and ultimately affect autoimmune disease in many ways.
Central issues related to activation of the type I IFN system in SLE are the identity of IFN-producing cells (IPCs) and inducers of the expression of type I IFN genes. We previously reported that SLE patients frequently had an IFNα-inducing factor in blood, which consisted of small ICs containing DNA and antibodies and was mimicked by combining plasmid DNA containing unmethylated CpG dinucleotides and anti-dsDNA autoantibodies (20, 21). Interferogenic ICs selectively trigger IFNα synthesis in the highly specialized natural IPCs (NIPCs), also known as pDC (20, 21). This requires dual engagement of Fcγ receptor IIa (FcγRIIa) and Toll-like receptor 9 (TLR-9), both of which are expressed by IPCs (22, 23). The frequency of pDC in the blood of SLE patients is severely reduced (24), which may be due to recruitment of these cells into inflamed tissues, where IFNα-producing cells have been shown to occur in higher numbers than in blood from healthy individuals (25, 26).
The DNA present in the interferogenic ICs in SLE is most likely from apoptotic cells. Thus, interferogenic ICs can be generated by mixing apoptotic or necrotic cells, or material released by them, with purified IgG from SLE patients (SLE-IgG), but not normal IgG (27, 28). However, the interferogenic activity of both the necrotic and apoptotic cell material was found to be sensitive to RNase treatment (27), and the interferogenic activity of the IgG correlated with the presence of antibodies to RNA-binding proteins, especially anti-RNP antibodies (27, 28). Furthermore, together with material from apoptotic or necrotic cells, IgG from patients with Sjögren's syndrome (SS-IgG), an autoimmune disease affecting mainly salivary and lacrimal glands, could also induce IFNα production (29). Patients with SS usually do not have anti-DNA antibodies, and IFNα induction by the combination of SS-IgG and apoptotic/necrotic cell material was sensitive to RNase treatment. It also correlated with the presence of antibodies to RNA-binding proteins, especially the major autoantigens anti-SSA and anti-SSB.
Interestingly, SS patients have both infiltration of IFNα-producing cells (29) and an IFN signature (30) in their salivary glands, suggesting that endogenous RNA-containing ICs are relevant type I IFN inducers in both SS and SLE. Because the formation of such interferogenic ICs correlated with the presence of anti-RNP and anti-SSA/SSB autoantibodies, the associated U1 snRNA and hYRNA are obvious candidate inducers. These RNA have secondary structures containing stem-loops and single-stranded stretches (2, 3) that may be capable of activating IFNα production. It is known that guanosine and uridine (GU)–rich single-stranded RNA (ssRNA) can induce IFNα via interaction with TLR-7/8, while dsRNA act via TLR-3 or intracellular sensors, such as retinoic acid–inducible gene I (RIG-I) and melanoma differentiation–associated antigen 5 (Mda5) (31, 32).
In the present investigation we analyzed the interferogenic ability of U1 snRNA and hY1RNA and found that these RNA, when mixed with lipofectin, induced IFNα production in monocytes, but not in NIPCs/pDC. Using native purified U1 snRNP particles, we found that they selectively activated IFNα production in NIPCs/pDC, but only in the presence of autoantibody-containing SLE-IgG. This activity required endocytosis as well as interaction with FcγRIIa and TLRs, possibly TLR-7/8.
U1 snRNP were purified from nuclear extracts of HeLa cells as previously described, using affinity chromatography with 2,2,7-trimethylguanosine monoclonal antibodies (mAb), MonoQ ion-exchange chromatography, and sucrose gradient centrifugation (33). The final preparations contain U1 snRNP at a purity of at least 90%, determined by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and sequential Coomassie blue and silver staining as described previously (34). The particles were used at a final concentration of 2.5 μg/ml in the cell cultures, unless otherwise stated.
In vitro–transcribed full-length U1 snRNA and hY1RNA were produced using the MEGAshortscript T7 Kit (Ambion, Austin, TX) with a T7 promoter–flanked polymerase chain reaction (PCR) product as template. The PCR product was produced from a plasmid containing the complete U1 snRNA or hY1RNA sequence (GenBank accession no. K00788.1 and V00584.1, respectively). The quality and the size of the produced RNA were controlled on an agarose gel. Endotoxin-free plasmid pcDNA3 was used as DNA, and poly(I-C) (Pharmacia P-L Biochemicals, Milwaukee, WI) was used as dsRNA. The plasmid pcDNA3 (10 μg/ml), poly(I-C) (10 μg/ml), hY1RNA (100 ng/ml, unless otherwise stated), or U1 snRNA (100 ng/ml, unless otherwise stated) was mixed with lipofectin (10 μg/ml; Invitrogen, San Diego, CA) and incubated for 15 minutes at room temperature. All concentrations represent the final concentrations in the cell culture. Before addition to the IFNα inducers, the lipofectin (40 μg/ml) was preincubated for 45 minutes at room temperature in Macrophage-SFM medium (Invitrogen) complemented with HEPES (20 mM), penicillin (60 μg/ml), and streptomycin (100 μg/ml).
Herpes simplex virus type 1 (HSV-1) was propagated in human amnion WISH cells, inactivated by ultraviolet light, and used at a final concentration corresponding to 2 × 106 plaque-forming units per milliliter, as previously described (27). Oligodeoxynucleotide (ODN) 2216 (3 μg/ml; Cybergene AB, Huddinge, Sweden) and resiquimod R848 (100 ng/ml; GLSynthesis, Worcester, MA) were used as TLR-9 and TLR-7/8 agonists, respectively (35, 36).
Necrotic cell material was prepared from the human monocyte line U937 as described previously (27). Briefly, U937 cells (5 × 107 /ml) in Macrophage-SFM medium supplemented as described above were freeze-thawed 4 times. Supernatants were collected after centrifugation (400g for 5 minutes), stored at −80°C, and used at a final concentration of 10% in the cell cultures.
Plasma (obtained by plasmapheresis) from 2 SLE patients, or serum from 1 SLE patient, and serum from 3 SS patients were used for preparation of IgG. One SLE patient had autoantibodies to DNA, histones, U1 snRNP 70K, U1 snRNP A, U1 snRNP C, SmB, and SmD. Another SLE patient had antibodies to histones, ribosomal P antigen, U1 snRNP A, U1 snRNP C, SmB, SmD, and low levels of antibodies to DNA. The third SLE patient had antibodies to DNA, U1 snRNP 70K, U1 snRNP A, U1 snRNP C, SmB, SmD, Ro 52, and Ro 60. One SS patient had antibodies to Ro 52, Ro 60, and La. The second SS patient had antibodies to Ro 52 and threshold levels of antibodies to Ro 60 and La. The third patient with SS had antibodies to Ro 52 and Ro 60. The analysis for autoantibodies was carried out using an in-house assay for antibodies to dsDNA (27) and the Inno-Lia ANA Update line blot assay (Innogenetics, Ghent, Belgium). Plasma from 3 autoantibody-negative healthy blood donors was used as control. The IgG was prepared by converting plasma to serum, followed by treatment with 2,000 units/ml DNase I (Roche Applied Sciences, Basel, Switzerland), purification of IgG on protein G columns (Amersham Biosciences, Uppsala, Sweden), and dialysis against culture medium (27). The final concentration of IgG used in the cell cultures, unless otherwise stated, was 1 mg/ml.
Peripheral blood mononuclear cells (PBMCs) were prepared by Ficoll-Hypaque (Amersham Biosciences) density-gradient centrifugation of buffy coats from healthy blood donors and cultured at a density of 2 × 106/ml in 0.1-ml volumes in 96-well flat-bottomed plates (Nunclone; Nunc, Roskilde, Denmark). Macrophage-SFM medium was used, supplemented as described above and, in addition, with 500 units/ml IFNα2b (Introna; Schering-Plough, Bloomfield, NJ) and 2 ng/ml granulocyte–macrophage colony-stimulating factor (GM-CSF) (Leukine; Berlex, Montville, NJ). The PBMCs used for inhibition of endocytic and TLR pathways were cultured in RPMI 1640 (Cambrex Bioscience, Baltimore, MD) supplemented with L-glutamine (2 mM), HEPES (20 mM), penicillin (60 μg/ml), streptomycin (100 μg/ml), and 5% human AB serum, as well as IFNα2b and GM-CSF.
Enrichment and depletion for blood dendritic cell antigen 4 (BDCA-4) (NIPC/pDC)– or CD14 (monocyte)–positive cells were performed using the MACS BDCA-4 and MACS CD14 Cell Isolation Kits (Miltenyi Biotec, Bergisch Gladbach, Germany). These cells, including the original PBMCs, were cultured at a concentration of 1 × 106/ml.
RNase-free DNase I (Qiagen, Hilden, Germany) was used at a final concentration of 20 Kunitz units/ml, and DNase-free RNase A (AB-Gene, Surrey, UK) was used at a final concentration of 8 μg/ml. The inducers were incubated with the enzymes for at least 3 hours at 37°C before addition of lipofectin or IgG.
The anti-FcγRIIa mAb IV.3 (no. HB-217; American Type Culture Collection, Rockville, MD) was purified from hybridoma supernatants on a protein G column, according to the instructions of the manufacturer (Amersham Biosciences), and dialyzed against phosphate buffered saline. The antibody was added to PBMC cultures at the indicated concentrations, together with the IFNα inducers, using the isotype-matched mAb MPC-11 (BD Biosciences, San Jose, CA) as a control.
Endocytosis was inhibited using chloroquine and bafilomycin A1 (Sigma-Aldrich, St. Louis, MO). The inhibitory phosphorothioate ODN 2088 (5′-TCCTGGCGGGGAAGT-3′) (37) and the inactive phosphorothioate ODN 20958 (5′- TCCTAACAAAAAAAT-3′) (both from Coley Pharmaceutical Group, Wellesley, MA) were used to explore the role of TLR-7, -8, and -9 in IFNα induction.
Flow cytometry. The expression of BDCA-2 or CD14 on IFNα-producing cells was detected by flow cytometry, as previously described (38). Briefly, 7 hours after induction, brefeldin A (10 μg/ml) was added to the cell cultures, and after an additional 2 hours, cells were harvested and stained using fluorescein isothiocyanate (FITC)–labeled anti–BDCA-2 IgG1 mAb (Miltenyi Biotec) or FITC-labeled anti-CD14 IgG1 mAb (Serotec, Oxford, UK). An FITC-labeled IgG1 mAb (BD Pharmingen, San Diego, CA) was used as control. The cells were fixed overnight in paraformaldehyde before staining for intracellular IFNα with a biotin-labeled anti-IFNα mAb (LT27:295), followed by streptavidin–phycoerythrin (Jackson ImmunoResearch, West Grove, PA). A biotin-labeled IgG1 mAb (Dako, Glostrup, Denmark) was used as control. Staining for BDCA-2– or CD14-positive cells was also performed to determine the efficiency of NIPC/pDC and monocyte enrichment and depletion. Cells were analyzed using a FACScan flow cytometer and CellQuest software (BD Biosciences). During BDCA-2/CD14 and IFNα staining, dead cells were excluded by gating in the forward and side scatter channels, leaving primarily live cells for the analysis.
The IFNα concentrations in 24-hour culture supernatants were determined by dissociation-enhanced lanthanide fluoroimmunoassay, as previously described (24). Briefly, microtiter plates were coated with the anti-IFNα mAb LT27:293, which recognizes most IFNα subtypes but not IFNα2b, and bound IFNα was detected using the europium-labeled anti-IFNα mAb LT27:297. The IFNα standard was calibrated against the National Institutes of Health reference leukocyte IFNα GA-23-02-530. This assay has a detection limit of 2 units/ml.
We initially investigated whether U1 snRNA and hY1RNA were capable of inducing production of IFNα in human PBMCs. The RNA was added, together with the transfection agent lipofectin, to aid uptake into cells. We found that both hY1RNA and U1 snRNA mixed with lipofectin induced strong IFNα production that was dependent on the RNA concentration (Figure 1). Furthermore, RNase A treatment markedly decreased the IFNα production induced by lipofected RNA. In contrast, RNA or lipofectin alone lacked IFNα-inducing ability.
In order to determine whether the IFNα production occurred in NIPCs/pDC or monocytes, PBMCs were cultured with U1 snRNA or hY1RNA and then double-stained for intracellular IFNα and either the NIPC/pDC marker BDCA-2 or the monocyte marker CD14. Cells producing IFNα were detected, both in cultures stimulated with U1 snRNA and lipofectin (Figures 2A–C) and in cultures stimulated with hY1RNA and lipofectin (Figures 2D–F). The results clearly show that IFNα production in response to either U1 snRNA or hY1RNA occurred mainly in CD14-positive cells (Figures 2B and E). However, CD14-negative cells were also IFNα positive. In contrast, essentially no IFNα-positive cells were BDCA-2 positive (Figures 2A and D) compared with the matched FITC-IgG1 control antibody (Figures 2C and F), while stimulation with HSV-1 caused strong production of IFNα in BDCA-2–positive cells (Figure 2G) but not in CD14-positive cells (Figure 2H). Lipofectin alone resulted in no IFNα-positive cells (Figure 2I), and staining with an isotype-matched control antibody to the anti-IFNα mAb gave no positive results (data not shown). Results show that U1 snRNA and hY1RNA activated IFNα production in a small fraction of monocytes and possibly other cell types, but not in NIPCs/pDC.
The fact that lipofected U1 snRNA and hY1RNA did not trigger IFNα production in NIPCs/pDC raised the possibility that the RNA must be in a native modified form, complexed with proteins, or even present in ICs in order to be taken up and activate NIPCs/pDC. We therefore examined the IFNα-inducing capacity of purified native U1 snRNP particles, alone or in combination with SLE-IgG that contained anti-RNP and anti-Sm autoantibodies.
We found that U1 snRNP particles combined with SLE-IgG induced production of high levels of IFNα in cultures of PBMCs from healthy blood donors, in a dose-dependent manner (Figure 3A). U1 snRNP particles, alone or combined with normal IgG from healthy individuals (Figure 3A), induced no IFNα production, nor did SLE-IgG alone (Figure 3B). There was variation in the IFNα-producing capacity of PBMCs from different blood donors with U1 snRNP and SLE-IgG stimulation, but a clear induction of IFNα synthesis was seen in all donors tested (Figure 3B). The IFNα levels observed with U1 snRNP and SLE-IgG stimulation were comparable with those seen when the combination of material from freeze-thawed cells and SLE-IgG was used as an inducer (Figure 3B). In contrast, SS patient IgG containing anti-Ro/La but not anti-RNP/Sm antibodies induced no IFNα when combined with U1 snRNP particles, but did induce high levels of IFNα when combined with material from freeze-thawed cells (Figure 3B).
In order to identify the cell type(s) responsible for the IFNα production induced by U1 snRNP and SLE-IgG, PBMCs were stained for intracellular IFNα and either the NIPC/pDC marker BDCA-2 or the monocyte marker CD14. All IFNα-producing cells were found to be BDCA-2 positive and clearly negative for CD14 (Figures 4A and B) compared with matched isotype controls (Figures 4C and F). The same pattern of induction was seen with cells stimulated with HSV-1 (Figures 4D and E). This indicates that NIPCs/pDC were the IFNα producers.
However, the forward and side scatter pattern of cells stimulated with U1 snRNP and SLE-IgG (Figure 4G) differed considerably from that of unstimulated cells and cells stimulated with HSV-1 (Figure 4H). Thus, cells stimulated with U1 snRNP and SLE-IgG contained a large population of cells with high granularity in side scatter and small size in forward scatter, resembling dead cells. There was also a marked decrease in the number of both BDCA-2–positive (Figure 4A) and CD14-positive cells (Figure 4B) that was not seen with unstimulated cells or cells stimulated with HSV-1 (Figures 4D and E). These effects were, however, not seen in cells that were stained without paraformaldehyde fixation and permeabilizing (results not shown) and may therefore be an artifact introduced during the staining procedure. Furthermore, direct analysis of the cells with low forward or side scatter in the U1 snRNP– and SLE-IgG–stimulated cultures revealed that these cells were negative not only for IFNα, but also for CD14 and BDCA-2 (results not shown). To exclude the possibility that these cells had produced IFNα or expressed either of the surface markers, samples were stimulated with U1 snRNP and SLE-IgG in cultures enriched or depleted for BDCA-4–positive pDC or CD14-positive monocytes. The cells enriched for pDC contained 28% BDCA-2–expressing cells, compared with 0.6% in the original PBMCs and 0.3% in the depleted cells. The monocyte-enriched cultures contained 86% CD14-positive cells, compared with 13% in the original PBMCs and 0.6% in the depleted cells.
The pDC-enriched cells produced much higher levels of IFNα than did the original PBMCs in response to SLE-IgG combined with either U1 snRNP or material from freeze-thawed cells (Figure 5A). Similar results were found with the control inducers HSV-1 and Sendai virus. The pDC-depleted cultures produced lower levels than the original PBMCs in response to all inducers (Figure 5A). The monocyte-enriched PBMCs essentially failed to produce IFNα in response to U1 snRNP combined with SLE-IgG (Figure 5B). They displayed reduced production in response to material from freeze-thawed cells combined with SLE-IgG and to HSV-1, but increased production of IFNα when stimulated with Sendai virus (Figure 5B). The monocyte-depleted cells produced as much IFNα as the original PBMCs, and in some experiments even more, when exposed to HSV-1 or to SLE-IgG combined with U1 snRNP or material from freeze-thawed cells but produced lower levels of IFNα when exposed to Sendai virus (Figure 5B).
The importance of snRNA in the induction of IFNα production by U1 snRNP was examined by RNase and DNase treatment. Pretreatment with RNase A strongly reduced IFNα induction by U1 snRNP and SLE-IgG, while DNase I treatment had no effect (Figure 5C). These enzymes had similar effects on the interferogenic capacity of material from freeze-thawed cells (Figure 5C). The specificity of the enzymes was verified by demonstrating that, when mixed with lipofectin, the IFNα-inducing capacities of the plasmid pcDNA3 or the dsRNA molecule poly(I-C) were inhibited by pretreatment with DNase I and RNase A, respectively, as expected (Figure 5C). In contrast, the enzymes did not affect the IFNα-inducing capacity of HSV-1.
The importance of FcγRIIa, which is known to be expressed by NIPCs/pDC (22), in the induction of IFNα by U1 snRNP combined with SLE-IgG was determined using the specific blocking mAb IV.3. When added to cultures of PBMCs, this mAb strongly inhibited the IFNα production induced by U1 snRNP combined with SLE-IgG, by ∼80% at the highest mAb concentration tested (Figure 6A). Monoclonal antibody IV.3 also inhibited the IFNα production induced by material from freeze-thawed cells combined with SLE-IgG (data not shown), but not that induced by HSV-1 or ODN 2216 (Figure 6A). The isotype-matched control mAb had no significant effects on the IFNα production caused by any of the inducers (Figure 6B).
PBMCs stimulated with the combination of U1snRNP and SLE-IgG were treated with inhibitors of the endocytic pathway and the TLR pathways. Both chloroquine and bafilomycin A1, which inhibit endosomal acidification and maturation (39), inhibited the induction of IFNα by U1 snRNP or material from freeze-thawed cells combined with SLE-IgG. Chloroquine and bafilomycin A1 also inhibited induction of IFNα by the control inducers HSV-1, TLR-7/8 agonist R848, and TLR-9 agonist ODN 2216 (Figures 6C and D). ODN 2088, which blocks TLR-7, -8, and -9, inhibited the IFNα production caused by all inducers, including the combination of U1 snRNP and SLE-IgG, whereas control ODN 20958 had no effect (Figures 6C, E, and F).
One major finding in the current study was that the 2 RNA molecules U1 snRNA and hY1RNA were able to trigger IFNα production in normal PBMCs, provided that the RNA was mixed with lipofectin. Such RNA species are ubiquitous in eukaryotic cells and are released from cells dying by necrosis or apoptosis (40). The absolute requirement for lipofectin in the induction of IFNα suggests that the RNA must reach either an intracellular cytoplasmic or an endosomal compartment. Simple lipofectin-mediated protection of RNA against degradation by RNases appears less likely in view of the complete lack of IFNα-inducing ability of U1 snRNA and hY1RNA, even at high concentrations. The common view is that endocytosis is the principal route of uptake for cationic liposomes, and the pathway through which materials such as DNA or RNA finally enter the cytoplasm and cell nuclei (41, 42). In this way, lipofected nucleic acid can interact with endosomal TLR-3, -7/8, and -9 (which recognize dsRNA, ssRNA, and CpG-rich DNA, respectively), as well as with intracellular receptors such as RIG-I and Mda5 (which recognize dsRNA) (31, 32, 43).
U1 snRNA and hY1RNA have secondary structures containing stem-loops and single-stranded stretches (2, 3) that may be capable of activating IFNα production via dsRNA- and ssRNA-dependent pathways. However, our finding that IFNα production occurred mainly in monocytes and not at all in the NIPCs/pDC was unexpected. NIPCs/pDC do have endosomal TLR-7 and TLR-9 but little or no expression of other TLRs (43), and are readily activated by lipofected GU-rich ssRNA or CpG-containing DNA (44,45). In addition, NIPCs/pDC express the intracellular sensors RNA-regulated protein kinase and RIG-I (43) and can be activated by intracellular delivery of lipofected short dsRNA or viral RNA (46). The lack of induction of IFNα in NIPCs/pDC by U1 snRNA or hY1RNA combined with lipofectin could be attributable to a lack of uptake of such liposomes into these cells. Another possibility is that U1 snRNA and hY1RNA activate IFNα production in monocytes via trigger mechanisms that are not available in NIPCs/pDC. These may involve TLR-3 or TLR-8, which are present in the endosomes of monocyte/macrophages but not in NIPCs/pDC (47), or monocyte-specific intracellular sensors that remain to be defined.
Induction of interleukin 6 (IL-6) and IL-8 by U1 snRNA via TLR-3 in human endometrial cell lines was recently reported (48). However, in that case the cytokine induction did not require lipofectin or other transfection agents, suggesting that another mechanism may be involved in the induction of IFNα, or that TLR-3 is not restricted to the endosomal compartment in these endometrial cell lines.
RNA molecules such as hY1RNA and U1 snRNA undergo extensive nucleoside modifications, and the native forms are also associated with proteins in cells (2, 49). The U1 snRNA, 3 unique U1 snRNP proteins (A, C, and 70K), and 7 Sm proteins form the U1 snRNP particles, with the proteins also being major autoantigens in SLE. Interestingly, we found that although purified U1 snRNP particles alone completely lacked IFNα-inducing ability, they became potent IFNα inducers when mixed with IgG preparations that contained anti-RNP/Sm autoantibodies from SLE patients. The IFNα induction was abolished by RNase and was therefore dependent on intact U1 snRNA. Furthermore, NIPCs/pDC, and not monocytes, were the actual IFNα producers. Thus, these results differ from those obtained with lipofected U1 snRNA, and indicate that native U1 snRNA, containing modified nucleosides and associated with proteins, is indeed interferogenic in NIPCs/pDC.
It was recently demonstrated that naturally occurring modifications of RNA molecules, including methylations and pseudouridine formation, suppress the ability of lipofected RNA to activate production of IL-12, tumor necrosis factor α, and IFNα in DCs, including NIPCs/pDC (50). Because such modifications are more common in eukaryotic than in prokaryotic RNA, they could be a way by which eukaryotic RNA released by apoptotic or necrotic cells is rendered less immunostimulatory. However, our results show that this does not apply to the IFNα-inducing ability of U1 snRNA present in ICs.
Furthermore, we found that the induction of IFNα by U1 snRNP combined with SLE-IgG required interactions with FcγRIIa and endocytosis, because induction was inhibited by antibodies blocking FcγRIIa and by the endocytosis inhibitors bafilomycin A and chloroquine. The engagement of FcγRIIa may not only be necessary for internalization of the ICs, but also could be directly involved in the intracellular signaling preceding IFNα induction in NIPCs/pDC. Thus, the lack of signaling via FcγRIIa could be another reason for the inability of U1 snRNA combined with lipofectin to induce IFNα in these cells. There were also indications that TLRs were involved, because the IFNα induction was inhibited by ODN 2088, which blocks both TLR-7 and TLR-9 (51). TLR-7 is probably the most relevant, because it can be activated by both ssRNA and short dsRNA (52) and has recently been shown to mediate triggering of IFNα production in NIPCs/pDC by ICs consisting of anti-RNP antibodies and material from apoptotic cells (53).
The fact that monocytes failed to produce IFNα in response to U1 snRNP ICs may be due to the inability of the TLR-7/8 pathways to trigger IFNα production in these cells or in monocyte-derived DCs (54). Another possibility is that ICs have a negative impact on monocytes, but a lesser impact on NIPCs/pDC. This may, for instance, be mediated by FcγRIIb, which is present on monocytes, but not on NIPCs/pDC. The FcγRIIb contains immunoreceptor tyrosine-based inhibitory motif, and when activated by ICs this receptor strongly inhibits the tyrosine phosporylation required for FcγRIIa-mediated phagocytosis and production of cytokines (55).
We have previously demonstrated that RNA-containing material released by dying cells (promptly at necrosis and with a delay at apoptosis) can trigger IFNα production specifically in NIPCs/pDC when combined with autoantibodies (27, 29). The findings in the present study confirm our hypothesis that RNA associated with major protein autoantigens, exemplified by the U1 snRNP particles, is interferogenic. Cells dying by apoptosis rapidly accumulate such RNA–protein complexes in, and on the surface of, apoptotic blebs and bodies (40). The increased apoptosis and decreased scavenging of apoptotic cells reported in SLE (40) can enhance formation of interferogenic ICs. Because the 2 examined RNA species, U1 snRNA and hY1RNA, could both induce IFNα production, it is quite possible that several other RNA species could also prove to be interferogenic when present in ICs.
We conclude that autoantigens associated with either DNA or RNA can form interferogenic ICs, and may be the reason for the sustained IFNα production seen in patients with SLE, SS, and several other autoimmune disorders in which such autoantibodies occur. The results of our current investigation could explain the observed association between the type I IFN signature and the presence of autoantibodies to RNA-binding proteins (56), and support a role of RNA-containing autoantigens in the development of these diseases. Thus, not only the autoantibodies but also the autoantigens themselves are important in the etiopathogenesis of autoimmune disease. These autoantigens in ICs would contribute to continuous NIPC/pDC activation and IFNα production, resulting in a type I IFN–driven autoimmune process, as previously described (27, 29, 57). Finally, inhibition of the IC-induced IFNα production by TLR antagonists suggests that this may constitute a new therapeutic strategy for these diseases.
After submission of this manuscript, a report by Vollmer et al was published, describing a study demonstrating that U1 snRNP–containing ICs induce IFNα production in NIPCs/pDC, and that highly conserved RNA sequences within the snRNP can stimulate TLR-7 and TLR-8 (58). Furthermore, Lau et al recently showed that RNA-containing ICs also can activate proliferation of murine autoreactive B cells (59). These data, together with the results of the present investigation, reveal that at least some endogenous RNA may have a pivotal role in promoting the autoimmune process in SLE. The results also suggest that inhibition of the action of such RNA could be the basis of new strategies for the treatment of this disease.
We thank Anne Trönnberg for outstanding technical assistance and Drs. Arthur M. Krieg, Marion Jurk, and Jörg Vollmer for kindly providing the inhibitory ODN 2088 and its control ODN 20958.