Expression of an anti-RNA autoantibody in a mouse model of SLE is sufficient to increase neutrophil and monocyte numbers as well as the amount of IFN-I

Mice

564Igi-B6 mice, 564Igi-B6.Tlr7^−/−, and 564Igi-B6.Rag2^−/− were generated as described previously [20]. C57BL/6 and the Igh^a allotype congenic C57BL/6 control mice were purchased from The Jackson Laboratory. B6. Ifnar1^−/− mice were from Dr. Ann Marshak-Rothstein (University of Massachusetts Medical School) with Dr. Jonathan Sprent’s permission (Garvan Institute of Medical Research, Sydney, Australia), and were bred to 564Igi mice. Mice were maintained under pathogen-free conditions at Tufts University School of Medicine. The mouse protocol was approved by Institutional and Animal Care and Use Committee.

Flow cytometry and cell sorting

Cells were stained for flow cytometry using standard procedures as previously described [21] [20] [22]. Antibodies were purchased from Southern Biotech, Pharmingen, eBiosciences, or BioLegend. Anti-564 Id antibody [20] was purified from the hybridoma clone (B6.256) and conjugated to Alexa 647 (Invitrogen) following manufacturer’s protocol. Anti-mouse FcγRIV monoclonal antibody (9E9, hamster IgG) [23] was labeled with Alexa 647 as described above. For some experiments positively selected neutrophils were isolated by magnetic beads; total BM cells were stained with biotin-labeled Ly-6G antibody (clone 1A8, Biolegend), bound to Dynabeads Biotin Binder (Invitrogen), and then isolated magnetically. On average, 97% of the recovered cells were neutrophils.

Cytospin and HEMA3 staining

BM viable (PI⁻) cells of each cell type were sorted as described above, centrifuged onto glass slides at 800 rpm for 5 min. using a Cytospin 2 centrifuge (Shandon-Elliott), followed by staining with a HEMA3 kit (Fisher Scientific) using the manufacturer’s protocol. Slides were examined at 400X magnification and pictures were analyzed using Image-Pro Plus software (Media Cybemetics).

ELISPOT for IFN-α-producing cells and for 564 antibody-secreting cells

RBC-depleted single-cell suspensions were prepared from BM and spleen [20] . Rabbit polyclonal anti-mouse IFN-αA antibody (PBL) was used as capture and secondary antibody to detect mouse IFN-αA. The same antibody was conjugated with alkaline phosphatase (AP) using a kit (AbD Serotec). Optimal concentrations of capture and secondary antibodies were determined empirically. For detection of 564 antibody-secreting cells, ELISPOT plates (Millipore) were coated with 5 μg/ml of anti-Id antibody (B6.256; mouse IgG1, κ). AP-conjugated secondary antibodies (SouthernBiotech) were added after incubation and washing of the plates. An NBT/BCIP solution (Pierce) was used to develop visible spots.

C1q immune complex (IC) binding ELISAs

Levels of serum immune complexes were determined using a protocol provided by Dr. Ricardo Gazzinelli (UMass Medical School, Worcester, MA). ELISA plates were coated with 5 μg/ml C1q (Complement Tech) overnight at 4°C, blocked with 1% BSA in PBS, and washed 3 times with 1xPBS/ 0.05% Tween 20. Serum samples, or immune complexes, were added and incubated for 1hr at 37°C. Detection of IC with AP-labeled anti-IgG2a and anti-IgG2b (Southern Biotech) was as previously described [20]. Immune complexes were made by incubating C57BL/6 bone marrow cells RNA with 564 hybridoma (IgG2b) and S7.1 hybridoma (564 IgG2a) monoclonal antibodies (RNA-specific) at a 1:10 ratio for 3 hrs at room temperature.

Real-time PCR

All qPCR experiments were as follows: first-strand cDNA synthesis was performed on four-fold serial dilutions of purified RNA. Triplicates were amplified with commercially available mouse Aicda, Tlr7, Mx1, Ifit1, Ifna6, Ifnb1, Oas 1a or Actb (endogenous control)-specific TaqMan primer/probe sets (Applied Biosystems) in an iQ5 real-time PCR system (Bio-Rad). Quantification was determined using standard curves for genes of interest and the Actb control.

Statistical analyses

The two-tailed Students’ t-test was used for all the statistical analyses. Error bars on bar histograms indicate standard errors.

Expansion of myeloid effector cells in 564Igi mice

To detect early 564Igi phenotypes that might provide insights into the mechanisms leading to SLE-like symptoms, we analyzed young mice (3 months) that showed no overt signs of disease. These mice were bred to a C57BL/6 (B6) background (564Igi-B6) to allow comparison to B6 control mice. Cytometric analysis of 564Igi BM and spleen cells showed an expansion of cells exhibiting low forward and high side scatter; features that are characteristic of granulocytes [24] [Supporting Information Fig. 1 (Fig. S1)]. Antibodies specific for surface antigens present on neutrophils, monocytes and pDCs [21] demonstrated a consistent and significant increase in the fraction of neutrophils (CD11b⁺ Gr-1^hi[25] and Supporting Information Fig. S2 and monocytes (CD11b⁺ Ly6C^hi) in the BMs and spleens of 564Igi-B6 mice (Fig. 1 and Supporting Information Fig. S2 and Daley et al., 2008) [25]. The fraction of neutrophils was also increased in blood (Fig. 1A and 1B), although this was not the case for monocytes (Fig. 1C and 1D). In contrast, when we compared the pDC (B220⁺Siglec-H⁺) fraction of 564Igi-B6 and wild-type B6 mice, we found no differences in the BM, blood or spleen (Fig. S3) even though this cell type is generally thought to be a major producer of IFN-I, particularly in response to viral infection [26]. Only about 1% of total viable BM cells were pDCs and we did not find a significant number of pDCs in the blood or spleen (Fig. S3).

Interferon production by neutrophils and monocytes in 564Igi mice

In order to identify the anatomical sites and cell types responsible for IFN-I production in a direct and unbiased manner, we developed an ELISPOT assay for IFN-α and evaluated it by stimulating purified pDCs with type-A CpG for 24 hours on an ELISPOT plate in vitro [19] (Fig. S3A and S3B). For this assay, we stained and sorted B220⁺CD11c⁺ cells as pDCs because the monoclonal antibody (440c) that binds to the murine pDC-specific surface marker (Siglec-H) blocks the secretion of IFN-α [27]. The induced production of IFN-α in the assay was easily observed as an increase in the size and number of spot. Interestingly, ex vivo purified pDCs from 564Igi-B6 appear to produce less IFN-α than B6 pDCs (Fig. S4A and S4B); we have not investigated this further.

The frequency of IFN-α-producing cells in vivo was then determined by ELISPOT. The frequencies of IFN-α-producing cells in the BM of three month-old B6 and 564Igi-B6 mice were about 150–200 and 300–500 per 1×10⁶ cells, respectively (Fig. 2A). Surprisingly, we did not find any IFN-α-producing cells in 2 × 10⁶ total spleen cells from either group of mice (Fig. 2A). These results were reproduced with minor variation between mice. We conclude that IFN-α is constitutively produced in the BM of both normal and autoimmune mice, but no detectable splenic IFN-α-producing cells are found in either group of mice (Fig. 2A). These studies revealed that 564Igi-B6 mice have significantly more IFN-α-producing cells than do B6 mice (Fig. 2A). Because the only difference between 564Igi-B6 and B6 mice is the 564Igi knock-in, our findings indicate that the presence of RNA-specific 564Igi B cells and autoantibody in the 564Igi mice results in an increase in the frequency of IFN-α-producing cells in the BM.

In order to identify the cell types that produce IFN-α in the BM in vivo, we performed ELISPOT assays with sorted populations of neutrophils (CD11b⁺Gr-1^hi), monocytes (CD11b⁺Gr-1^lo/−), and pDCs (B220⁺CD11c⁺) from 564Igi mice (Fig. 2C, and 2D and Fig. S4). Because primary pDCs represent only about 1% of total BM cells (Fig. S3), we pooled purified populations from several mice to obtain a sufficient number of cells for these assays. The morphology of each sorted population was confirmed by microscopic examination of cells spread by Cytospin (Fig. 2B). To our surprise, we found that BM monocytes (CD11b⁺Gr-1^lo/−), not pDCs, are the most effective IFN-α-producing cells in vivo (Fig. 2C). In the 564Igi-B6 BM, about 244 per 10⁶ monocytes produced IFN-α (Fig. 2C). A lower but significant fraction of sorted neutrophils (CD11b⁺Gr-1^hi) was also found to produce IFN-α (about 108 per 10⁶) (Fig. 2C). About 148 in 10⁶ pDCs (B220⁺CD11c⁺) sorted from the 564Igi-B6 BM produced IFN-α (Fig. 2C).

To assess the contribution of each cell type to total BM IFN-α production, we calculated the total number of IFN-α producing cells of each type (Fig. S4C). Because pDCs constitute only about 1% of total BM cells in both B6 and 564Igi-B6 mice, these cells make only a minimal contribution to the total production of IFN-α. Further, we did not find a difference in the number of IFN-α-producing pDCs in B6 and 564Igi-B6 mice (Fig. S4B). Our results show that, in our SLE model, neutrophils and monocytes, not pDCs, are the major IFN-I-producing cells in vivo.

Increased FcγRIV on neutrophils and monocytes in 564Igi-B6 mice

In SLE, myeloid lineage phagocytic effector cells, such as neutrophils and monocytes, play an important role in engulfing immune complexes (ICs) and initiating immune responses via FcγRs [28]. Furthermore, the levels of activating FcγRI are elevated on myeloid cells of SLE patients, and IFN-I can augment this expression [29].

We therefore determined whether the pro-inflammatory milieu in 564Igi-B6 mice facilitates the expression of FcγRs in vivo, thus potentially lowering activation thresholds for myeloid effector cells. Because IgG2a and IgG2b, the most pro-inflammatory IgG subclasses, are found abundantly in 564Igi mice in vivo [20], we chose to examine the expression of FcγRIV, an activating FcγR specific for IgG2a and IgG2b [23].

To analyze FcγRIV expression in neutrophils (CD11b⁺Gr-1^hi) and monocytes (CD11b⁺Ly-6C^hi), we gated these populations from BM, blood, and spleen samples as shown in Fig. 1A and 1C. Using a specific monoclonal antibody, 9E9 [23], we found increased expression of FcγRIV in young, disease-free 564Igi-B6 mice as compared to controls (Fig. 3). The expression of FcγRIV was significantly higher in neutrophils and monocytes from spleens as compared to BM and blood (Fig. 3C and 3D). We did not observe any significant expression of FcγRIV in pDCs in B6 or 564Igi-B6 mice (data not shown).

Neutrophils express both activating and inhibitory FcγRs and the relative expression ratio between the activating and inhibitory FcγRs (FcγRIV/ FcγRIIb ratio) sets the activation threshold [30]. We measured the FcγRIV/ FcγRIIb ratio by real-time PCR. We consistently found that bone marrow, blood and splenic neutrophils from 564Igi-B6 mice have a higher FcγRIV/ FcγRIIb ratio in a comparison with neutrophils from B6 mice (Fig. 3E). These data raise the question as to what induces the higher FcγRIV/ FcγRIIb in 564Igi-B6 mice.

IFN-I signaling in the up-regulation of FcγRIV in neutrophils of 564Igi-B6 mice

We then asked whether IFN-I plays any role in the expression of FcγRIV in neutrophils in 564Ig-B6 mice. To this end, we bred 564Igi-B6 mice to IFN-I-receptor α-chain (Ifnar1)-deficient mice that lack IFN-I signaling [31, 32]. We found that the expression of FcγRIV in 564Igi-B6.Ifnar1^−/− neutrophils was significantly impaired in the absence of IFN-I signaling (Fig. 4A). The most striking difference was found in the spleens of 564Igi-B6.Ifnar1^−/− mice where neutrophils failed to up-regulate FcγRIV expression (Fig. 4A). Interestingly, in 564Igi-B6 mice the overall level of FcγRIV expression in monocytes seems to be lower than in neutrophils (Fig. 3C and 3D); and, despite the absence of IFN-I signaling in the young mice we analyzed, no statistically significant decrease in the FcγRIV expression in monocytes was seen (data not shown). This difference might reflect a differential regulation [21] in the two cell-types and/or the fact that we were analyzing relatively young mice that have not yet manifested a discernible difference. The expression of FcγRIV on pDCs was negligible, or absent, in all B6 and 564Igi-B6 mice (data not shown).

These results indicate that IFN-I-mediated signaling is required for the significant increase of FcγRIV on neutrophils. It is striking that in human SLE and our mouse model, IFN-I induces the up-regulation of an activating FcγR, even though the two species use different receptors [29]

IFN-I signature genes are up-regulated in 564Igi-B6 neutrophils

Several studies have shown that PBMC and BM cells from SLE patients exhibit increased expression of IFN-signature genes [3–5, 33]. We therefore asked whether IFN-I acts on neutrophils in the BM of 564Igi-B6 mice. To test this we measured the expression of two “IFN-signature genes” the Mixovirus resistance 1 (Mx1), and interferon-induced protein with tetratricopeptide repeats 1 (Ifit1) [34]. The results, shown in Fig. 4B, demonstrate that they were induced in 564Igi-B6 neutrophils. These results suggest that 564Igi neutrophils in the BM are activated by IFN-I.

Immune complexes of IgG2a and IgG2b isotypes found in the sera of 564Igi-B6 mice

A high frequency of SLE patients have anti-RNA antibodies in their sera [35 ]. In addition, it has been known for a long time that immune complexes (IC)s are present in the sera of SLE patients and they are deposited in the kidneys resulting in pathology [36–38]. We have therefore measured by a C1q ELISA the presence of IC in the sera of 564Igi mice (Fig. 4C and 4D). We produced IC composed of the RNA-specific monoclonal antibody 564 (mouse IgG2b) or S7.1 (mouse IgG2a) with RNA derived from BM cells at a 1:10 ratio (w/w) of RNA/antibody. As previously reported [37], C1q can efficiently bind RNA/564 IgG2b or RNA/S7.1 IgG2a IC, but neither the purified 564 nor S7.1 monoclonal antibody nor the RNA alone (Fig. 4C, 4D and Fig. S5). Cross-reactive binding of anti-IgG2a antibodies to RNA/564IgG2b ICs is observed (Fig. 4C and Fig. S5A). However, no cross reactivity was observed with anti-IgG2b antibodies to RNA/IgG2a ICs (Fig. 4D and Fig. S5B). We found significant binding to C1q of components of the sera from 564Igi mice as detected by either anti-IgG2b or anti-IgG2a, indicating that IC formed by self-RNA/ 564 antibodies of IgG2a and IgG2b isotypes are present. These results indicate that ICs formed in 564Igi-B6 mice may be responsible for the activation of neutrophils via the FcγRIV.

Increased frequency of class-switched developing B cells in the BM of 564Igi mice

We have previously shown that mature B cells from the spleen of 564Igi mice are anergic and do not secrete much IgM anti-RNA antibody. Nevertheless they have in their sera IgG2a and IgG2b anti-RNA antibodies [20]. In addition, we also found that activation-induced cytidine deaminase (AID) is expressed in developing immature B cells [39]. Taken together, these findings suggested that immature self-reactive B cells might be responsible for the IgG autoantibody production.

Cell surface staining of 564Igi immature B cells (AA4.1⁺) revealed that over 10% of them are class-switched to IgG2a and nearly all of them are Id⁺, suggesting that self-reactive BCR underwent CSR to IgG2a (Fig. 5D, 5E). The frequency of IgG2a⁺ in developing B cells of 564Igi mice was almost twenty fold higher than in the background strain, the allotype-matched congenic C57BL/6 (B6.Igh^a) wild-type mice (0.87 ±0.4 % in wild-type control versus 15.5 ±5.5 % in 564Igi mice) (Fig. 5D, 5E). The increased 564 Id⁺IgG2a⁺ population in 564Igi-B6 BM was a true double positive population, not an artifact by a compromised fluorochrome compensation (Fig. 5A, 5B, 5C).

These results support the notion that developing B cells harboring self-reactive BCRs undergo CSR in the BM via BCR/TLR dual-engagement, as proposed previously [39].

IFN-I signaling for activation of autoreactive developing B cells and for formation of autoantibody-secreting cells

Type I interferon plays a role in enhancing humoral immune responses and CSR [40]. We therefore explored whether IFN-I signaling is required for the production of IgG autoantibody. To accomplish this we used an ELISPOT assay with anti-564 Id antibody as the capture reagent. In the absence of IFN-I signaling, the differentiation of B cells from the BM into 564 antibody-secreting cells (ASCs) was compromised in the BM but not the spleen (Fig. 6A). The difference in the total number of 564 ASCs between 564Igi-B6.Ifnar1^−/− and 564Igi-B6 (Ifnar1^+/+) mice was statistically significant in the BM but not in the spleen. These results indicate that IFN-I signaling is not absolutely required for the regulation of CSR of IgG genes coding for autoreactive autoantibody, but suggest it enhanced IgG autoantibody formation.

Tlr7 and Aicda over-expression in autoreactive RNA-specific developing B cells is dependent on IFN-I receptor

We have shown previously that 564Igi anti-RNA production was partially TLR7 dependent [20]. In order to directly determine the role of IFN-I in the expression of Tlr7, we analyzed developing immature B-lymphocytes from B6 and 564Igi Id⁺ immature B cells from 564Igi-B6.Tlr7^−/−, 564Igi-B6.Ifnar1^−/−, and 564Igi-B6 mice. It has been reported that IFN-I can specifically up-regulate the expression of TLR7 and MyD88 in human naïve splenic mature B cells in vitro [14]. Consistent with this, Tlr7 expression was significantly diminished in developing Id⁺ BM B cells in the absence of IFN-I signaling (Fig. 6B)

We also previously showed that Aicda is expressed during early B cell development, resulting in functional CSR and SHM in vivo [39]. These studies also suggested that Aicda expression was, in part, a consequence of the dual engagement of BCR and TLR during B cell development. Because we found increased Tlr7 expression in immature B cells of 564Igi mice, we determined whether RNA-specific autoreactive 564Igi B cells have increased Aicda expression, perhaps by 564Igi BCR/TLR7 dual-engagement. Real-time PCR results clearly demonstrate that the activation state of RNA-specific 564Igi B cells, as shown by relative Aicda expression (Fig. 6C), correlated with the extent of Tlr7 expression (Fig. 6B and 6C), suggesting that the expression level of Tlr7 dictates the frequency of activated B cells in the 564Igi system. These results suggest that among many other functions, one role of IFN-I during B cell development is a fine-tuning of the level of TLR7 expression.

Increase of IFN-signature genes in autoreactive RNA-specific developing B cells

We then asked whether IFN-I acts on immature B cells in the BM of 564Igi-B6 mice. To test this we measured the expression of a set of “IFN-signature genes” 2′–5′ oligoadenylate synthetase 1A (Oas1a), Mx1 and Ifit1 [34] [49]. In order to assure that only 564Igi Id⁺ B cells were analyzed we obtained cells from 564Igi mice with a Rag2-deficient background. All three IFN-I signature genes were expressed higher in immature BM B cells from 564Igi-B6.Rag2^−/− mice than in B6 (Fig. 6D). These results indicate that immature 564Igi B cells are directly activated by IFN-I in the BM.

Our data clearly suggest that IFN-I works on B cells, neutrophils and monocytes. It not only enhances the production of autoantibody of the most pro-inflammatory IgG subclasses (IgG2a and IgG2b) [41] but also up-regulates the most pro-inflammatory FcγR [23].