NZM2410 (Taconic and Jackson Laboratories), NZB-ZW/F1J, BALB/c, and C57BL/6J (Jackson Laboratories) mice were bred and maintained in accordance with the guidelines of the IACUC and all the experimental procedures were approved by the IACUC of the Children's Hospital of Philadelphia, an AAALAC accredited facility.

To study DCs in the spleen ex vivo, we used a modified protocol from Vremec and colleagues 16. In brief, we injected excised spleens with a solution of 0.8 mg/ml collagenase and 0.1 mg/ml DNAse, then teased the tissue into small pieces, and incubated it for 30 minutes at 37°C. From this point onward, all procedures were performed on ice to minimize the spontaneous activation of DCs in vitro. We pushed spleen fragments through a cell strainer (100 μm) and inhibited collagenase with 5 mM EDTA; then we lysed the red blood cells with Hybrimax (Sigma, St Louis, MO, USA) and counted the cells. We prestained the total population of splenocytes for 10 minutes with rat anti-mouse CD16/CD32 (clone 2.4G2) antibodies to block FcγR, and then stained for 30 minutes with the following monoclonal antibodies from BD Bioscience Pharmingen (San Diego, CA, USA): Allophycocyanin -conjugated hamster anti-mouse CD11c (to recognize DCs) and PerCP-Cy5.5-rat anti-mouse CD19 (to gate out CD11c^low B cells). In some experiments, biotinylated anti-F4/80 and NK1.1 monoclonal antibodies, followed by Streptavidin PerCP-Cy5.5 were used to gate out CD11c^low macrophages and natural killer cells, respectively. We found that these cells do not contribute significantly to the CD11c⁺ population.

We studied DC costimulatory molecules with phycoerythrin-conjugated rat anti-mouse CD80, CD86, and CD54 and with fluorescein isothiocyanate-conjugated hamster anti-mouse CD40. To study CD40 expression in DC subsets, we used phycoerythrin-conjugated rat anti-mouse CD8α, CD11b, and B220. In parallel tubes we stained cells with isotype control antibodies to determine non-specific staining. After staining, cells were washed in phosphate-buffered saline, fixed in 1% formaldehyde, and analyzed on a FACSCalibur flow cytometer (BD Biosciences Pharmingen, San Diego, CA, USA). We collected a large number of cells per sample (5 × 10⁵ cells per tube), allowing us to study small populations among the splenocytes, such as the CD11c⁺ DCs (about 2 to 4% of the splenocytes) and the smaller subsets: the myeloid DCs (recognized as CD11c⁺ CD11b⁺ cells), the pDCs and the CD8α⁺ DCs (recognized as CD11c⁺ B220⁺ cells and CD11c⁺ CD8α⁺ cells, respectively). The large number of cells yielded a sufficiently large number of events (more than 2,000 for subsets, and up to 18,000 for the all DC populations) to guarantee reliable results.

We generated bone marrow-derived DCs (BM-DCs) as described previously 17. In brief, we eliminated T cell and B cell contaminants from bone marrow with anti-Thy 1.2 and anti-B220 magnetic beads and MACS columns (Miltenyi Biotech, Bergisch Gladbach, Germany); then seeded bone marrow precursors in 24-well plates at 10⁶ cells/ml in 10% fetal bovine serum in complete IMDM (Iscove's modified Dulbecco's medium) containing glutamine, 2-mercaptoethanol, and antibiotics, and enriched with 3 ng/ml granulocyte/macrophage colony-stimulating factor (GM-CSF) and 2.5 ng/ml interleukin-4 (BD Bioscience Pharmingen, San Diego, CA, USA). We used BM-DCs at day 6 to 7 of culture.

We isolated splenic myeloid DCs as described previously 18. In brief, we freed the DCs from the extracellular matrix, as described above. We plated the splenocytes on polystyrene Petri dishes to let DCs and macrophages adhere, and washed out the non-adherent cells after 90 minutes. We then incubated the adherent cells overnight in complete medium enriched with 10 ng/ml GM-CSF. On the following day, we collected the floating DCs while leaving the strongly adherent macrophages on the plastic. After harvesting, we stained the isolated splenic DCs as described above.

We analyzed the results with the non-parametric Kruskal-Wallis test and the non-parametric post hoc Mann-Whitney test to evaluate the differences between experimental groups. We considered p < 0.05 to be significant.

We studied DCs in spleens of lupus-prone NZM2410 mice in comparison with non-autoimmune age-matched BALB/c and B6 mice. NZM2410 mice, derived by selective inbreeding of the progeny of NZB-W/F1 mice, resemble, both serologically and clinically, the manifestations of human systemic lupus erythematosus (SLE) 19. Because the parental NZB and NZW mice display some autoimmune features, we chose as controls two non-autoimmune strains of mice, B6 and BALB/c, which in the literature are the most commonly used for comparison with NZM2410 2021 and NZB-W/F1 mice 1222, respectively. Differences shared by these two non-autoimmune mice in comparison with the NZM2410 mice are likely to be lupus specific. We first studied mice with lupus at 6 to 9 months of age (adult mice). The prevalence of the autoimmune disease is about 90% at this age. For our experiments, we selected mice that showed high titers of anti-DNA antibodies in their sera, detected with a specific ELISA (not shown). Additionally, we assessed kidney damage by detecting elevated proteinuria (more than 300 mg/dl) and blood urea nitrogen (more than 26 mg/dl), and an increase in body weight (45 to 50 g) due to fluid retention from kidney failure (not shown). Age-matched BALB/c and B6 mice did not have anti-DNA antibodies and displayed levels of proteinuria in the normal range for mice (30 to 100 mg/dl), normal blood urea nitrogen (less than 26 mg/dl), and stable adult body weight (25 to 30 g) (not shown).

We also investigated DCs from spleens of mice 6 to 8 weeks old (young). At this age, no sign of autoimmunity (anti-DNA antibodies or kidney damage) is evident in the NZM2410 or in the NZB-W/F1 model of lupus.

To study DCs ex vivo we performed no cell selection, to avoid any loss of DC subsets. We therefore stained the entire population of splenocytes and analyzed a large number of cells by flow cytometry (see details in Methods). Figure 1a shows the gate used to analyze DCs. We found that, after the onset of disease, splenic DCs accumulate with age (Figure 1b,c). Indeed, we found significant differences in percentages and absolute numbers of DCs between young and adult NZM2410 mice. Such accumulation with age is not present in either BALB/c or B6 mice: in these strains, DC percentages and absolute numbers were not significantly different in young and adult mice (Figure 1b,c). These data are consistent with the significant difference in CD11c⁺ DC numbers between lupus-prone and BALB/c adult mice reported in NZW-BXSB/F1 9 and in NZB-W/F1 mice (LC, JD, DKS, LF, RC, SG, unpublished data). In addition, we propose the new concept that the distribution of DCs is a variable among different strains (compare B6 with BALB/c) and that DC accumulation needs to be measured by comparing different ages in a single strain. These observations suggest that DC accumulation is common to at least three murine models of lupus 8911 and therefore that it is likely to be important in lupus development.

The analysis of the costimulatory molecules expressed by DCs ex vivo (Figure 2a,b) showed that NZM2410 mice 6 to 9 months old (adult) had a lower percentage of DCs expressing CD80, whereas the positivity for CD86 was similar to that of the non-autoimmune mice (Figure 2c–e). Moreover, the comparable levels of CD80 and CD86 in BALB/c and B6 DCs confirmed the feasibility of using these two strains of mice for comparison with the lupus-prone strain. As was expected, we found a similar profile (normal CD86 and low CD80) in age-matched NZB-W/F1 mice (not shown). Furthermore, in both NZM2410 and NZB-W/F1 mice, we found a decreased expression of the adhesion molecule CD54 on lupus DCs in comparison with non-autoimmune age-matched BALB/c DCs (Figure 2b and not shown).

A defect in the expression of the costimulatory molecule CD80 and the adhesion molecule CD54, in the presence of normal levels of the costimulatory molecule CD86, suggests a pathologic state of activation of DCs arising during the disease.

We also found that CD40 expression was highly significantly upregulated in DCs from spleens of diseased NZM2410 mice in comparison with age-matched non-autoimmune mice (Figure 2b,e). The increase in the percentage of CD40-positive DCs was accompanied by an increase in CD40 mean fluorescence intensity (MFI) (Figure 2f). The levels of CD40 positivity in DCs of the two non-autoimmune strains were significantly different, but they were both much lower than in lupus DCs (Figure 2e,f). We found the same constitutive overexpression of CD40 in NZB-W/F1 DCs (Figure 2g). The increase in CD40 MFI was not accompanied by an increase in the maximal levels of CD40 expression (Figure 2b). These data suggest an increased number of activated DCs rather than an increased capacity of lupus DCs to express CD40 once activated.

We also investigated CD40 expression in the same group of young mice. We found that young NZM2410 mice have a higher percentage of CD40-positive DCs in their spleens than either BALB/c or B6 mice. The two control strains had very similar levels of CD40-positive DCs (Figure 3c), both significantly lower than in NZM2410 mice. The percentages of CD40-positive DCs decreased with age in the three strains, with a more pronounced effect in B6 mice (compare Figure 2e with Figure 3c). Although the functional significance of these age-related differences is worth further investigation, our results show that DCs from lupus-prone mice express abnormal and long-lasting levels of this fundamental marker.

In summary, we found that DCs from lupus-prone mice show a pathologic state of activation in comparison with age-matched control mice. Our analysis suggests that the decreased CD80/CD86 ratio may be linked to the full development of the autoimmune disease, whereas the overexpression of CD40 pre-dates the occurrence of overt autoimmunity and may therefore be involved in its pathogenesis.

To determine whether the overexpression of CD40 was a DC-specific phenomenon, we analyzed the expression of CD40 in B cells, the other important population of immune cells that normally express this marker. As expected, a large percentage of B cells constitutively expressed high levels of CD40. We found that NZM2410 mice and age-matched BALB/c mice had similar percentages of CD40-positive B cells, both in adult animals (Figure 4a) and in young animals (Figure 4b), whereas B6 B cells showed overall lower levels of expression, especially in adult mice.

Because the percentage of CD40-positive B cells in NZM2410 mice was significantly higher only in comparison with B6 but not with BALB/c mice, and this single difference was present only after onset of the disease, we propose that the overexpression of CD40 in lupus-prone mice is DC specific.

DCs can be subdivided into subsets, characterized by specific surface markers and functions 23. We therefore asked which DC subset is responsible for the increase in CD40-positive DCs in lupus-prone mice. In young NZM2410 mice, we observed overexpression of CD40 largely associated with myeloid DCs. This subset continued to overexpress CD40 after the onset of the disease (Figure 5a). CD8α⁺ DCs constitutively expressed high levels of CD40 in all three strains (Figure 5b). Nevertheless, we observed a slightly but significantly higher percentage of CD40-positive cells in CD8α⁺ DCs from lupus-prone mice than in non-autoimmune mice at a young age. Such overexpression disappeared after lupus onset (Figure 5b). The pDCs, which normally present a more immature phenotype than other subsets, expressed low levels of CD40 in all three strains, with normal percentages of CD40 positivity in young lupus-prone mice (Figure 5c) and increased percentages after lupus onset, although the increase was not statistically significant (Figure 5c). Because we found greater variability in the CD40 expression of pDCs from old mice with lupus than from the other groups of mice, we take into account that technical reasons may have hampered the statistical significance in this comparison (Kruskal-Wallis p = 0.018, Mann-Whitney NZM versus BALB/c = 0.064, NZM versus B6 = 0.023). We therefore do not completely exclude the involvement of pDCs in the increased DC expression of CD40 after the onset of the disease.

This thorough analysis suggests that the increase in CD40-positive DCs is due mainly to the myeloid DCs and to a smaller extent, before the onset of the disease, to the CD8α⁺ DCs, and after disease onset, to pDCs.

The increase in the percentage of CD40-positive DCs that we observed in NZM2410 and in NZB-W/F1 mice could be due to a pro-inflammatory environment present in lupus-prone mice even months before the onset of the disease, or to a primary alteration in the state of activation of DCs.

To discriminate intrinsic DC abnormalities from those induced by environmental factors, we analyzed CD40 expression in NZM2410, BALB/c, and B6 BM-DCs, grown in culture with a protocol that generates resting CD11c⁺ CD11b⁺ myeloid-like DCs 17. In this artificial environment, DCs are influenced only by themselves and by the cytokines that are provided to them in culture (Figure 6a). We found that lupus BM-DCs expressed levels of CD40 comparable to those of BALB/c and B6 BM-DCs in terms of both percentage of positivity (Figure 6b,c) and MFI (Figure 6b). We found the same results in BM-DCs grown from both old (Figure 6a–c) and young mice (data not shown).

These results suggest that lupus DCs do not have a primary alteration in their expression of CD40, but that the overexpression observed ex vivo may be the result of a chronic exposure or altered response to activators in vivo.

To determine whether lupus DCs have a normal capacity to upregulate CD40 expression after activation, we used the traditional protocol by Inaba and colleagues to isolate splenic DCs 18. This protocol gives rise to a pure (more than 90%; data not shown) population of cells double positive for CD11b and CD11c and negative for CD8α or B220, therefore resembling myeloid DCs. Moreover, DCs are highly activated by the procedure, with more than 80% of DCs expressing high levels of costimulatory molecules (not shown). Although the stimulus that induces DC activation is unclear, this protocol allowed us to analyze the intrinsic capacity of DCs to upregulate CD40 1824. We found that lupus DCs from both NZM2410 mice (Figure 6d) and NZB-W/F1 mice (Figure 6e) upregulated CD40 to the same extent as the non-autoimmune DCs, therefore suggesting that the regulation of CD40 expression is not intrinsically altered.