This study was approved by the Institutional Review Board of Kangnam St. Mary's Hospital (Seoul, Korea) and all participants provided informed consent. Forty-three consecutive patients (two males and 41 females; age 35 ± 8.4 years) who presented at the rheumatology clinic and fulfilled the revised classification criteria for SLE 21 were enrolled in the study. Twenty-six volunteers (one male and 25 females; age 38.4 ± 4.3 years) were recruited to serve as healthy control individuals. Among the 43 patients, two patients were off prescription medication (in remission) and three were receiving prednisolone alone (mean dosage 12.5 mg/day; range 5 to 20 mg/day). The remaining 38 patients were receiving both prednisolone (mean dosage 13.3 mg/day; range 2.5 to 125 mg/day) and adjunctive therapies, such as hydroxychloroquine (34 patients; 200 or 400 mg/day), azathioprine (four patients), mizoribine (four patients), mycophenolate mofetil (three patients), and cyclosporine (one patient). Disease activity was scored using the SLE Disease Activity Index (SLEDAI) 22. Descriptive statistics and clinical data for the SLE patients are described in Table 1.

Blood samples obtained from patients and healthy control individuals were collected in heparinized tubes (BD Biosciences, San Jose, CA, USA), and PBMCs were prepared using Ficoll-Hypaque (Amersham Bioscience, Pascataway, NJ, USA) density gradient centrifugation. Cells were washed and suspended in RPMI 1640 medium (GibcoBRL, Carlsbad, CA, USA) containing penicillin (100 U/ml), streptomycin (100 μg/ml), and 10% fetal bovine serum (GibcoBRL) that had been inactivated by heating to 56°C for 1 hour. Healthy pDCs were isolated from PBMCs using a Diamond Plasmacytoid Dendritic Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany); pDC purity was greater than 95%. The purified pDCs were cultured in RPMI 1640 medium containing 10% fetal bovine serum, granulocyte-macrophage colony stimulating factor (10 ng/ml), and IL-3 (10 ng/ml).

PBMCs were incubated with human IgG to block the Fc receptor and then incubated with anti-CD123-PE-Cy5 monoclonal antibody (Mouse IgG₁; BD Pharmingen™, San Jose, CA, USA), anti-BDCA-2-fluorescein isothiocyanate, and monoclonal antibody (Mouse IgG₁; Miltenyi Biotec) for 30 minutes at 4°C; isotype control experiments were conducted in parallel. After two washes, the cells were re-suspended in phosphate-buffered saline and analyzed by flow cytometry. pDCs were identified by dual staining for both CD123 and blood dendritic cell antigen (BDCA)-2.

PBMCs (1 × 10⁶ cells) were stimulated using 1 μmol/l CpG ODN2216 (InvivoGen, San Diego, CA, USA) or 30% serum from SLE patients. Duplicate cultures were performed in 48-well plates (NUNK, Roskilde Denmark) at a final volume of 500 μl/well. After 24 hours IFN-α was measured from the supernatant.

Supernatants collected from sera and cell cultures were stored at -70°C until further use. The amounts of IFN-α in the sera and supernatants were then measured using a sensitive sandwich ELISA kit (Bender MedSystems, Vienna, Austria). A representative standard curve for the IFN-α ELISA is shown in Additional file 1 (Supplementary Figure 1). All measurements were performed in duplicate and averaged values were used in the data analysis.

Purified pDCs (2 × 10⁴ cells) were incubated with or without the TLR9 ligand CpG ODN2216 (1 μmol/l; InvivoGen) or 30% serum from SLE patients in 96-well plates (NUNK) at a final volume of 200 μl/well. After 24 hours the pDCs were carefully washed with serum-free RPMI and re-treated with or without CpG ODN2216 (1 μmol/l) or 30% serum from SLE patients. The supernatants were harvested after an additional 24 hours and IFN-α production was measured using ELISA. In the recovery assay, pDCs were treated with 1 μmol/l CpG ODN2216 for 24 hours, washed with serum-free medium, cultured with serum-containing medium for 0, 24, or 48 hours, and then treated again with CpG ODN2216 (1 μmol/l). After 24 hours IFN-α production was measured from the supernatant.

Relative cell viability was measured by Quick Cell Proliferation Assay kit (BioVision, Mountain View, CA, USA). Briefly, 1/10 volume of WST-1/electrocoupling solution were added into the culture media, incubated 4 hours in 5% carbon dioxide incubator, and measured the absorbance of the treated and untreated samples with water-soluble tetrazolium salt (WST)-1/electrocoupling solution using a microtiter plate reader at 450 nm. Each sample was duplicated and averages of the absorbance were used in comparisons. The differences in absorbance between treated and untreated samples was shown as relative cell viability.

Total RNA was extracted from isolated PBMCs or cultured cells using RNAzol B (Biotex Laboratories, Houston, TX, USA), in accordance with the manufacturer's instructions. Reverse transcription using 2 μg total RNA was carried out at 42°C using the Superscript reverse transcription system (Takara, Shiga, Japan). PCR amplification was performed in a reaction mixture containing 2.5 mmol/l dNTPs, 2.5 U Taq DNA polymerase (Takara), 0.25 μmol/l sense and antisense primers, and PCR buffer (1.5 mmol/l MgCl₂, 50 mmol/l KCl, 10 mmol/l Tris-HCl [pH 8.3]). Reactions were processed in a DNA thermal cycler (Perkin-Elmer Cetus, Wellesley, MA, USA). PCR products were separated on a 2.5% agarose gel and stained with ethidium bromide. The latest cycle number during which PCR products were not yet saturated was selected and used to compare treatments using Quantity One software (version 4.5.2; BioRad, Hercules, CA, USA). Results are expressed as the ratio of target PCR products to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) or β-actin product. The used primer pairs are described in Table 2.

To measure the serum levels of immune complexes in SLE patients, we used the CIC-C1Q Circulating Immune Complexes ELISA kit (Bühlmann Laboratories AG, Schonenbuch, Switzerland). Circulating immune complexes from sera from patients with chronic SLE and control individuals were incubated with human C1q, which was adsorbed onto microtiter wells. After a washing step, we added alkaline phosphatase-conjugated protein A, which binds to the Fc region of human IgG. After an additional washing step, the enzyme substrate (paranitrophenyl-phosphate) was added, followed by a stop solution. The absorption at 405 nm of each sample was measured. We used 14 SLE sera, two of which were from patients with SLEDAI values of 12 and 16, and the remainder were from patients with SLEDAI values of less than 4. The two active SLE sera exhibited 1.8 U and 6.3 U of immune complexes.

Differences between groups were analyzed using the Mann-Whitney U-test or Student's t-test. Correlation analyses were performed using Spearman's rank correlation test. All analyses were performed using SPSS (version 10.0; SPSS Inc., Chicago, IL, USA). Data are expressed as the mean ± standard deviation.

We used ELISA to compare serum concentrations of IFN-α between SLE patients and healthy control individuals (Figure 1a). Patients with active SLE (SLEDAI > 4) exhibited significantly higher levels of IFN-α than did patients with inactive SLE and healthy control individuals (P = 0.002 and P = 0.007, respectively). However, the levels of IFN-α in total SLE patients (active and inactive) versus healthy control individuals were not significantly different (P = 0.280). No correlation was observed between serum IFN-α levels and the clinical characteristics of SLE, such as disease duration, or medications such as steroids and hydroxychloroquine (data not shown).

Because the number of circulating pDCs is very low among total PMBCs, making it difficult to isolate this cell type from a blood sample, IFN-α production was measured in total PBMCs using a pDC-specific TLR9 ligand 20. PBMCs from patients with SLE and healthy controls werestimulated in vitro for 24 hours using the TLR9 ligand CpG ODN2216. IFN-α production was then measured using ELISA. CpG DNA induced IFN-α production was markedly reduced in PBMCs from SLE patients as compared with PBMCs from healthy control individuals (342.46 ± 636.82 pg/ml in SLE patients versus 1610.35 ± 759.56 pg/ml in healthy control individuals; P < 0.001; Figure 1b). In addition, IFN-α production was slightly higher in patients with active SLE than in those with inactive SLE, but the difference was not significant (P = 0133). No correlation was observed between serum levels of IFN-α and CpG-induced IFN-α production in vitro in patients with active SLE. Interestingly, IFN-α production was completely abolished in PBMCs from one-third of SLE patients. These data are almost identical to those reported previously 18. Although CpG ODN2216 was not used in that previous work, CPG oligonucleotides, herpesviruses, and DNA-containing immune complexes are all TLR9 ligands. Our findings are in accordance with thosse of previous studies showing that PBMCs from SLE patients have reduced capacity to produce IFN-α in response to TLR9 stimulation 1617.

Because pDCs are major producers of IFN-α, decreased IFN-α levels in SLE patients may be the result of a drop in pDC count 161723. To test this possibility, we stained PBMCs with anti-BDCA-2 and anti-CD123, which recognize specific surface markers for human pDCs 23. During fluoresence-activated cell sorting analysis, PBMCs were selected in gate R1 and BDCA-2⁺/CD123⁺ cells from gate R1 were deemed pDCs, as shown in gate R2 (Figure 2a). The percentage of pDCs among PBMCs was slightly reduced in SLE patients compared with healthy control individuals (0.23 ± 0.11% in SLE patients versus 0.30 ± 0.14% in controls; P > 0.10; Figure 2b). Although this difference was not statistically significant, a decrease in pDC count may partially contribute to reduced IFN-α production. We observed similar results for the absolute number of circulating pDCs (data not shown). Because pDCs are a major source of IFN-α in human PBMCs 101112, we calculated the relative IFN-α producing capacity of pDCs in vitro as follows. The amount of IFN-α produced upon CpG ODN2216 stimulation in vitro was divided by the absolute number of pDCs. As expected from results presented in Figures 1b and 2b, the relative IFN-α producing capacity of pDCs was significantly lower in SLE patients than in control individuals (0.2 ± 0.18 in SLE patients versus 1 ± 0.5 in control individuals; P < 0.01; Figure 2c). These data, and the fact that PBMCs from one-third of SLE patients did not produce IFN-α, suggest that the observed decrease in IFN-α production is caused by aberrant function in SLE pDCs or depletion of some proportion of pDCs, and not by a decrease in total pDC count. However, there is no way to determine the proportion of pDCs in the blood.

Downregulation of TLR9 may also contribute to decreases in IFN-α production in SLE patients. We used semiquantitative reverse transcription PCR to determine the expression of TLR9 in PBMCs. Interestingly, TLR9 expression was higher in SLE PBMCs than in cells from healthy control individuals (Figure 2d). These data demonstrate that the decrease in IFN-α production was not caused by downregulation of TLR9 expression.

Differences in the numbers of circulating pDCs do not fully explain the difference in IFN-α production capability between the SLE and healthy PBMCs. Furthermore, it has been reported that TLR9 expression levels in the pDCs of SLE patients and healthy control individuals are similar 24. We suspected that the presence of DNA-containing immune complexes, which function as TLR9 ligands, affect pDC function. In SLE patients, these immune complexes stimulate pDCs to produce IFN-α via TLR9 and FcγRIIa (CD32) 1415. We incubated healthy PBMCs with serum from SLE patients and healthy control individuals, and then measured the production of IFN-α. As expected, serum from SLE patients induced IFN-α production to a much greater degree than did serum from healthy control individuals (24.1 ± 27.5 pg/ml versus 4.1 ± 3.14 pg/ml, respectively; P < 0.005; Figure 3a). Paradoxically, SLE patients whose serum induces low levels of IFN-α production in PBMCs from healthy individuals possess PBMCs that exhibit higher IFN-α production when stimulated by the TLR9 ligand CpG ODN2216 (Figures 1b and 3b). Vice versa, those patients with SLE whose serum induces high levels of IFN-α production in PBMCs from healthy individuals possess PBMCs that exhibit lower IFN-α production in response to CpG ODN2216. In other words, the degree of CpG-induced IFN-α production in SLE PBMCs correlated inversely with SLE serum induced IFN-α production in healthy PBMCs. Also, SLE serum induced IFN-α production in healthy PBMCs correlated with the amount of immune complexes in SLE serum (Figure 3c), and the degree of CpG-induced IFN-α production in SLE PBMCs was inversely correlated with the amounts of immune complexes in matched SLE patients (Figure 3d).

Because there was an inverse correlation between IFN-α producing capacity in SLE PBMCs and SLE serum induced IFN-α production in control PBMCs, we hypothesized that the persistent presence of DNA-containing immune complexes may desensitize pDCs to TLR9 stimulation. TLR tolerance, in which cells exhibit no response to subsequent challenges with the same TLR stimulus, has been reported in epithelial cells, neutrophils, macrophages, and conventional dendritic cells 25262728. To test this hypothesis, we induced TLR9 tolerance by repeatedly stimulating pDCs isolated from healthy individuals using CpG ODN2216 or 30% serum derived from SLE patients. Both IL-3 and granulocyte-monocyte colony stimulating factor are essential cytokines for pDC survival and were thus included in the culture medium 29. Repeated stimulation using CpG ODN2216 or 30% serum from SLE patients resulted in 65% and 90% decreases in IFN-α production, respectively (Figure 4a); cell viability was not affected (data not shown). These data indicate that repeated stimulation induces TLR9 tolerance in pDCs and are similar to previous data indicating that pDCs produce large amounts of IFN-α within the first 12 hours of activation and become refractory to subsequent stimulation 30.

TLR9 tolerance may be the result of negative feedback in the TLR signaling pathway 31. Therefore, we examined whether TLR9 tolerance is reversible. pDC PBMCs from healthy individuals were incubated with CpG ODN2216 or 30% SLE serum for 1 day, washed, and then re-stimulated with CpG ODN2216 or 30% SLE serum after a given interval (0, 24, or 48 hours), as indicated in Figure 4b. IFN-α production was measured 24 hours after re-stimulation. IFN-α production decreased upon repeated stimulation at the 0 hours interval and recovered within 48 hours of the first stimulation in repeatedly stimulated pDCs (Figure 4c,d). The cell viability did not differ significantly between stimulation and re-stimulation groups (lower panels of Figure 4c,d). Although we could not test the recovery of IFN-α production capability by pDCs purified from SLE patients, the PBMCs from SLE patients recovered IFN-α production capability over time in vitro, without further stimulation, as shown in Additional file 1 (Supplementary Figure 2). These data indicate that TLR9 tolerance is reversible over time and that continuous TLR9 stimulation is required to induce persistent nonresponsiveness in pDCs.

Because the overall amounts of IFN-α in SLE sera were not significantly different from those in sera from healthy control individuals (Figure 1a), it was unclear whether circulating pDCs had already become TLR9 tolerant. Therefore, we used fluorescence-activated cell sorting to analyze the expression of CD80, CD86, and HLA-DR, which are markers of pDC maturation and activation. However, the number of circulating pDCs was too low to provide conclusive evidence regarding the activation of PDCs. Instead, we examined the expression of IFN-α signature genes in SLE PBMCs 32, including IFIT1 (IFN-induced protein with tetratricopeptide repeats 1), IFI44 (IFN-induced protein 44), and PRKR (IFN-inducible double-stranded RNA-dependent protein kinase). The expression of all three genes was elevated in SLE PBMCs compared with healthy PBMCs (Figure 5a–c). These data imply that SLE PBMCs had been exposed to IFN-α to a greater degree, even though IFN-α levels did not differ between the sera of SLE patients and those of healthy control individuals.