B cell-activating factor (BAFF) from dendritic cells, monocytes and neutrophils is required for B cell maturation and autoantibody production in SLE-like autoimmune disease

Purpose and methods B cell-activating factor (BAFF) contributes to the pathogenesis of autoimmune diseases including systemic lupus erythematosus (SLE). Although several anti-BAFF Abs and derivatives have been developed for the treatment of SLE, the specific sources of BAFF that sustain autoantibody (auto-Ab) producing cells have not been definitively identified. Using BAFF-RFP reporter mice, we identified major changes in BAFF-producing cells in two mouse spontaneous lupus models (Tlr7 Tg mice and Sle1), and in a pristane-induced lupus (PIL) model. Results First, we confirmed that similar to their wildtype Tlr7 Tg and Sle1 mice counterparts, BAFF-RFP Tlr7 Tg mice and BAFF-RFP Sle1 mice had increased BAFF serum levels, which correlated with increases in plasma cells and auto-Ab production. Next, using the RFP reporter, we defined which cells had dysregulated BAFF production. BAFF-producing neutrophils (Nphs), monocytes (MOs), cDCs, T cells and B cells were all expanded in the spleens of BAFF-RFP Tlr7 Tg mice and BAFF-RFP Sle1 mice compared to controls. Furthermore, Ly6Chi inflammatory MOs and T cells had significantly increased BAFF expression per cell in both spontaneous lupus models, while CD8- DCs up-regulated BAFF expression only in the Tlr7 Tg mice. Similarly, pristane injection of BAFF-RFP mice induced increases in serum BAFF levels, auto-Abs, and the expansion of BAFF-producing Nphs, MOs, and DCs in both the spleen and peritoneal cavity. BAFF expression in MOs and DCs, in contrast to BAFF from Nphs, was required to maintain homeostatic and pristane-induced systemic BAFF levels and to sustain mature B cell pools in spleens and BMs. Although acting through different mechanisms, Nph, MO and DC sources of BAFF were each required for the development of auto-Abs in PIL mice. Conclusions Our findings underscore the importance of considering the relative roles of specific myeloid BAFF sources and B cell niches when developing treatments for SLE and other BAFF-associated autoimmune diseases.


Introduction
B-cell-activating factor (BAFF) (also known as BLyS or Tnfsf13b) is a member of the TNF family of ligands that plays an essential role for mature B cells to develop, survive and respond to Ag (1)(2)(3)(4). BAFF binds to three different TNF receptor family members: BAFFR, TACI and BCMA; signaling through these receptors functions as a key survival factor for maintaining mature B cell homeostasis (5)(6)(7). Baff -/and Baffr -/mice develop few or no mature B cells due to a block in B cell development from the newly formed transitional 1 (T1) to the T2 B cell stage onwards (5)(6)(7). We and others have shown that BAFF plays a key protective role against infectious diseases, including West Nile virus, HIV and Salmonella typhimurium (8)(9)(10)(11)(12).
On the other hand, BAFF overexpression contributes to the pathogenesis of a number of autoimmune diseases, including IgA nephropathy, type I diabetes, systemic sclerosis, Sjögren's syndrome, rheumatoid arthritis, and systemic lupus erythematosus (SLE) (13)(14)(15)(16)(17)(18)(19). Mice overexpressing BAFF develop an SLE-like disease, and blocking BAFF receptors in these animals inhibits the development of kidney disease and prolongs survival (20,21). Furthermore, mouse strains that develop spontaneous SLE have increased serum BAFF levels (14,15,20). Elevated serum BAFF levels and increased BAFF gene expression in peripheral blood cells are associated with disease activity in SLE patients (22)(23)(24). More recently, a genotype carrying a variant of the gene encoding BAFF was associated with increased disease severity in SLE (25).
The anti-BAFF Ab, belimumab, was the first BAFF-blocking drug approved for the treatment of SLE (15, 26-30). Other BAFFtargeting agents and belimumab in combination with other biologics are being assessed as treatments for lupus patients (29). The heterogeneity in clinical responses to belimumab and other BAFF-blocking drugs underscores the need to understand more fully the complexity of the BAFF system (3,27,28). While the downstream effects of blocking BAFF receptors have been examined extensively, the upstream regulation of BAFF production and how it affects the development of autoimmunity in vivo has not been investigated in detail.
In this study, we used BAFF-red fluorescent protein (RFP) reporter (BAFF-RFP) mice and BAFF conditional knockout (cKO) lines we previously developed (10) to identify BAFF sources implicated in the development of auto-Abs in lupus. We crossed BAFF-RFP reporter mice with two spontaneous mouse lupus models: Tlr7 Tg mice (31)(32)(33) and Sle1 mice (34, 35). Mice overexpressing TLR7 develop pathogenic auto-Abs and an SLElike disease (33). Tlr7 Tg mice also exhibit peripheral myeloid expansion, dependent on Type I IFN (IFN-I) signaling, a major BAFF inducer (10, 36,37). Sle1 (B6.NZMc1) mice contain an SLE-susceptibility locus that mediates the loss of tolerance to nuclear Ags and generates a robust, spontaneous anti-chromatin auto-Ab response that peaks at 7-9 mo of age (35,38). In addition, we used a pristane-induced lupus (PIL) model (39) as an inducible lupus system readily suitable for studies using our BAFF cKO lines. The hydrocarbon oil pristane induces an SLE-like renal disease, ectopic lymphoid tissues and auto-Abs by 6-8 mo p.i (39). In the PIL model, similarly to Tlr7 Tg mice, the SLE-like autoimmune phenotype is dependent on robust TLR7-mediated IFN-I production by inflammatory MOs (40,41). In all three models, BAFF-RFP Tlr7 Tg mice, BAFF-RFP Sle1 mice, and pristane-treated BAFF-RFP mice, we observed increased BAFF serum levels, which correlated with increases in plasma cells (PCs) and auto-Ab production. BAFF-producing DCs, MOs and Nphs each expanded upon induction of autoimmunity, and each of these cellular sources of BAFF was required for the optimal induction of auto-Abs. However, DCs, MOs and Nphs regulated B cells and auto-Ab production through different mechanisms. DCs and MOs, but not Nphs, were required for the systemic production of BAFF and the maintenance of mature B cells in pristane-induce lupus. Our findings highlight how distinct BAFF sources may play different roles in the development of autoreactive B cells in lupus. The identification of specific BAFF-producing cells and niches implicated in SLE autoimmunity may be useful for designing new effective B cell targeted therapies.

Materials and methods
Mice 57BL/6J (WT) mice were purchased from Jackson Labs (Bar Harbor, ME). BAFF-RFP mice and Baff fl/fl mice (on the B6 background) were described previously (10). Briefly, BAFF-RFP reporter mice were generated by replacing a Tnfsf13b/Baff allele with a targeting construct expressing IRES-RFP. The Tnfsf13b/Baff is functionally knocked-out where the endogenous reporter expresses the RFP protein translated under the control of an IRES site. Thus, the BAFF-RFP signal is a measurement of BAFF expression and correlates with both Baff mRNA and BAFF protein expression (10). These heterozygous BAFF-RFP +/-(BAFF-RFP) mice express the RFP protein on one Baff allele and the WT Baff on the other. Tlr7.1 Tg mice and Sle1 mice (31-36) (provided by KBE). zDC Cre , Cx3cr1 Cre and Mrp8 Cre mice were purchased from Jackson Labs (Bar Harbor, ME, USA). BAFF cKO mice where Baff is selectively deleted in either Nphs (Baff fl/fl Mrp8 Cre ), cDCs (Baff fl/fl zDC Cre ) or MOs (Baff fl/fl Cx3cr1 Cre ) were described previously (10). Briefly, a~50% BAFF mRNA reduction was restricted to cDCs in Baff fl/fl zDC Cre , and an~85% reduction of BAFF mRNA was restricted to CX3CR1 hi MOs in Baff fl/fl Cx3cr1 Cre . In Baff fl/fl Mrp8 Cre BAFF mRNA was selectively, but not specifically reduced in Nphs by 85-90%, as some reduction was observed in MOs too. However, our previous studies do not show any overlap in the phenotype of Baff fl/fl Mrp8 Cre vs. Baff fl/fl Cx3cr1 Cre mice (10, 11). All mice were age-and sex-matched for experiments and used at 8-12 wks of age. Mice were housed in a specific pathogen free environment; all procedures were approved by the University of Washington Institutional Animal Care and Use Committee.

Pristane induced lupus
To induce a lupus-like disease, 0.5 ml or 0.4 ml of pristane (2,6,10,14-tetramethylpentadecane, TMPD) (Sigma-Aldrich, St. Louis, MO, USA) was administered by i.p. injection into 2-3 mo old mice (40). Controls mice were left untreated (naïve). Blood samples were obtained before pristane treatment and after treatment at day 7, day 14, day 28 and monthly thereafter. For short-term experiments with BAFF-RFP mice, spleens and peritoneal exudate cells (PECs) were harvested 2 wks after treatment. For long-term experiments with BAFF cKO mice, spleens and BMs were harvested 6 mo after pristane injections; untreated (naïve) and pristane treated mice were 8-9 mo old at the time of tissue harvest. Spleens and BMs from 2-3 mo old BAFF cKO mice and Baff fl/fl mice were also harvested as controls at the time of injection.

Tissue harvest and cell isolation
Spleens, BM cells and PECs were harvested and cell suspension obtained as described previously (9,10,42). Briefly, spleens were removed and dissociated by enzymatic digestion at 37°CC with liberase TL and Dnase I (Roche, Indianapolis, IN, USA), followed by mincing the tissue between the ends of two frosted microscope glass slides to obtain a single cell suspension. BM cells were isolated by cutting one end of the femur and flushing BM cells out of the bone by centrifugation. After erythrocytes were lysed, splenocytes, BM cells and PECs from peritoneal lavage were filtered and processed for staining for flow cytometry.

Auto-Ab and BAFF ELISAs
Sera were isolated from blood, collected via the retro-orbital route. Serum levels of BAFF were determined using the Mouse BAFF/ BLyS/TNFSF13B DuoSet ELISA kit (R&D systems, Inc., Minneapolis, MN) according to the manufacturer's instructions. For specific auto-Ab ELISAs (44), 96 well Immuno plates (Nunc) were coated with: Sm/RNP (5mg/ml; Arotec Diagnostic Limited ATR01-10); yeast RNA (5mg/ml; Sigma R6750); calf thymus dsDNA (25mg/ml; Sigma-Aldrich D4522) or calf thymus histone (10mg/ml; Sigma-Aldrich H9250) diluted in 1X PBS and incubated overnight at 4°C. Prior to coating with yeast RNA and calf thymus histone, the plates were coated with Poly L Lysine (25mg/ml; Sigma P6516) for 2 h at RT or overnight at 4°C. Plates were then blocked for 2 h at RT with 1% BSA/ PBS prior to addition of diluted serum for 2 h at RT. For ssRNA ELISA plates were blocked with 5% Goat serum/PBS for 2 h at RT and diluted serum were incubated overnight at 4°C. Specific Ab were detected using goat anti-mouse IgG Ab coupled with HRP (Southern Biotech, Birmingham, AL) and allowed to incubate for 2 h at RT. HRP activity was visualized using tetramethylbenzidine peroxidase (Bio-Rad Laboratories, Hercules, CA). Absorbance for all ELISA assays were measured at 450 nm with a Synergy plate reader (BioTek). For the standard curve, goat anti-mouse IgG (Southern Biotech, Birmingham, AL) was used for coating, and serial dilutions of recombinant IgG (Southern Biotech, Birmingham, AL) were used to calculate the absolute quantities of anti-SmRNP IgG and anti-ssRNA IgG. Serial dilutions of sera were used to determine the relative amount of anti-dsDNA and anti-Histone in different groups, measured by the OD at 450nm.

Statistical analyses
Timeline data were analyzed using a 2-way ANOVA with Tukey's multiple comparison test. Analyses between more than two groups were performed using 1-way ANOVA with Holm-Sidak method for multiple comparisons. Analyses between two groups were performed using unpaired Student t test or Mann Whitney test. GraphPad Prism was used for statistical analyses. Differences of p<0.05 were considered significant.

Induction of BAFF, autoantibodies and plasma cells in BAFF-RFP Tlr7.1 Tg mice and BAFF-RFP Sle1 mice
To investigate BAFF regulation and BAFF-producing cells in lupus prone mice, we crossed our BAFF-RFP reporter mice with Tlr7.1 Tg mice and Sle1 mice. Heterozygous BAFF-RFP +/-(BAFF-RFP) mice express lower BAFF levels, as one Baff allele expresses the RFP protein as a measurement of BAFF expression and the other allele expresses wild-type (WT) Baff (10, 11). However, BAFF-RFP still retain a reduced population of mature B cells, and thus can be used for functional studies of B cells and Ab responses (10, 11). We first assessed how either the Tlr7.1 Tg or the Sle1 background affected serum BAFF levels. Both 6-8 mo old Tlr7.1 Tg mice and the 7-9 mo old Sle1 mice had increased serum BAFF levels compared to agematched WT control B6 mice, although the Tlr7.1 Tg mice had more pronounced increases in serum BAFF compared to the Sle1 mice ( Figure 1A, upper panel). Similarly, BAFF levels in the sera of BAFF-RFP Tlr7.1 Tg mice and BAFF-RFP Sle1 mice were significantly upregulated compared to BAFF-RFP control mice ( Figure 1A, lower panel). As expected, since the heterozygous BAFF-RFP lines express half the normal levels of BAFF protein, they had about half the BAFF serum levels of the parental lines ( Figure 1A).
Next, we tested whether the Tlr7.1 Tg or Sle1 background was sufficient to induce an autoimmune phenotype in the BAFF-RFP mice, even in the context of reduced mature B cells. As expected, Tlr7.1 Tg mice and Sle1 control mice had elevated titers of anti-SmRNP, -ssRNA and -dsDNA IgG auto-Abs compared to WT mice ( Figure 1B, upper panels). Similarly, BAFF-RFP Tlr7.1 Tg mice and BAFF-RFP Sle1 mice had substantially increased serum auto-Ab levels compared to BAFF-RFP controls ( Figure 1B, lower panels). Surprisingly, the mean levels of serum IgG auto-Abs in BAFF-RFP Sle1 mice were comparable to those in Sle1 mice. Furthermore, the fold increase in anti -SmRNP and -dsDNA IgG auto-Abs in BAFF-RFP Sle1 mice compared to BAFF-RFP mice was greater than in Sle1 mice compared to WT controls (SmRNP: 226 -fold in RFP Sle1/ RFP in vs 37 -fold in Sle1/WT; dsDNA: 13.4-fold in RFP Sle1/RFP vs 2.3 -fold in Sle1/WT). Anti-histone IgG Abs were also upregulated to a similar extent in BAFF-RFP Tlr7.1 Tg mice and Tlr7.1 Tg mice ( Figure 1B).
Consistent with the increased auto-Ab levels, spleens of BAFF-RFP Tlr7.1 Tg mice had increased PC numbers compared to BAFF-RFP mice and were comparable to those of Tlr7.1 Tg mice ( Figure 1C and Supplementary Figure S1A). PCs were not upregulated in the BM of either Tlr7.1 Tg mice or BAFF-RFP Tlr7.1 Tg mice (data not shown). Sle1 mice had expanded levels of PCs in the BM ( Figure 1D and Supplementary Figure S1B), and to a lesser extent, in the spleen (Supplementary Figure S1C). We previously showed that PCs are also BAFF producers (10). BAFF-RFP PCs were also substantially expanded in the spleens of BAFF-RFP Tlr7.1 Tg mice ( Figure 1C, right panel) and in the BM of BAFF-RFP Sle1 mice ( Figure 1D, right panel). In summary, BAFF-RFP mice with either the Tlr7.1 Tg or Sle1 lupus genotype developed an autoimmune phenotype comparable to their parental line including higher systemic BAFF levels, expanded PCs in spleens and BMs, and increased anti-chromatin auto-Abs.
Regulation of BAFF-RFP expression in splenic and BM myeloid and lymphoid cells from BAFF-RFP Tlr7.1 and Sle1 mice Next, we investigated how a Tlr7 or Sle1 background affected BAFF expression in myeloid and lymphoid subsets. Compared to controls, BAFF-RFP Tlr7.1 mice had increased BAFF expression in all splenic MO subsets and in CD8 -DCs (Figures 2A, B) as well as in BM preDCs ( Figure 2C). Interestingly, TLR7 overexpression did not affect BAFF expression in either splenic or BM Nphs, which constitutively express high BAFF levels (Figures 2A-C). Surprisingly, TLR7 overexpression also upregulated BAFF in splenic and BM T cells ( Figure 2E). TLR7 overexpression, in addition to upregulating BAFF expression on a per cell basis, induced the expansion of several BAFF-producing myeloid and lymphoid splenic cell populations ( Figure 2F).
Myeloid and lymphoid cells were not as profoundly expanded in the Sle1 mice as in the Tlr7.1 Tg mice (45). However, similar to the BAFF-RFP Tlr7.1 mice, BAFF was up-regulated in splenic inflammatory Ly6C hi MOs and to a lower extent in T cells from BAFF-RFP Sle1 mice ( Figures 3A, B). In contrast to the TLR7 Tg mice, splenic Nphs from BAFF-RFP Sle1 mice had lower BAFF expression per cell than Nphs from BAFF-RFP controls ( Figures 3A, B), possibly due to Nph activation. Indeed, we and others have shown that upon activation, Nphs not only release BAFF from the intracellular stores but also down-regulate BAFF expression (11). cDC-and B cell-BAFF expression per cell did not change in the spleens of Sle1 autoimmune mice ( Figure 3B). However, similar to TLR7 overexpression, Sle1 induced the expansion of several BAFF-producing lymphoid and myeloid cell populations, including Nphs ( Figure 3C). Thus, in both Tlr7.1 Tg and Sle1 spontaneous SLE models, increased systemic BAFF levels and the development of autoimmunity were associated with a significant expansion of all the major BAFF-producing myeloid cell populations: Nphs, MOs, and cDCs.

Increased serum BAFF and expansion of BAFF-RFP + myeloid cells in the spleen and peritoneal cavity of pristane treated lupus-prone mice
In order to explore the effects of depletion of BAFF in specific immune cell types, we used a pristane-induced mouse model of lupus (10, 11). As with Tlr7Tg mice, the SLE-like autoimmune phenotype of PIL mice is dependent on IFN-I production and expansion of inflammatory myeloid cells (39). Pristane treatment of WT B6 mice induced an increase in serum BAFF levels that peaked at 30 days post-injection ( Figure 4A, left panel). As with WT mice, BAFF-RFP mice up-regulated serum BAFF that peaked and doubled 30 days after pristane injection ( Figure 4A, right panel). Since the BAFF-RFP mice express only one wild-type BAFF allele, constitutive and induced serum BAFF levels were lower in these mice than in WT mice ( Figure 4A). As expected from previous studies in B6 mice (46), WT mice had a modest up-regulation of anti-SmRNP by 3-4 mo post pristane injection ( Figure 4B). In pristane-treated BAFF-RFP mice, auto-Ab responses were delayed and lower than in WT mice, but still detectable and significantly different from BAFF-RFP untreated mice ( Figure 4B). As previously reported (38), pristane did not induce anti-DNA or anti-histone auto-Abs in the B6 strain (data not shown). Reeves et al. showed that pristane injection induces recruitment of Nphs and inflammatory MOs to the peritoneal cavity and spleen, peaking two weeks post-injection (41). At this time point, BAFF-RFP expression was up-regulated in splenic MO subsets, but decreased in Nphs from BAFF-RFP mice, similar to what we detected in BAFF-RFP Sle1 mice ( Figures 4C, D). Also, as in the spontaneous lupus models, BAFF + Nphs and BAFF + inflammatory MOs/DCs were expanded in the spleens of pristane-treated BAFF-RFP mice ( Figure 4E). As expected, two weeks after pristane treatment Ly6C hi inflammatory MO and mature CD11c + Mph/DCs were also expanded in the peritoneal cavity of BAFF-RFP mice, as well as in WT mice ( Figure 5A). These inflammatory MOs and Mph/DCs also up-regulated BAFF-RFP expression on a per cell basis ( Figures 5B, C), and together with BAFF-RFP+ Nphs, were substantially increased in numbers after pristane injection ( Figure 5D). Together with the significant expansion of myeloid populations, peritoneal B2 and B1 B cells are substantially reduced by pristane treatment (41). Interestingly, although B1 B cells were reduced in the peritoneal cavity of either WT or BAFF-RFP mice, PC numbers and BAFF-RFP + PCs were expanded after pristane injection ( Figure 5E). Furthermore, the remaining B1a and B1b B cells, and to a lesser extent B2 B cells, normally expressing low BAFF levels, upregulated BAFF-RFP expression upon pristane injection ( Figure 5F). Thus, our data suggest that in PIL, as well as in the spontaneous lupus models (Figures 1C, D; 2D-F; 3C), autocrine and paracrine BAFF from PCs and other B cell subsets could also contribute to sustain B cells.
In summary, similar to the spontaneous lupus models, PIL induced the expansion of splenic and peritoneal BAFF producing -Nphs and -inflammatory MOs/DCs and an increase in systemic BAFF levels, together with the expansion of PCs and induction of auto-Abs.  Figure S2A). Furthermore, while pristane treatment of Baff fl/fl control mice induced a sustained increase in serum BAFF levels, it did not up-regulate systemic BAFF levels in BAFF cDC cKO mice for up to 6 mo post-injection ( Figure 6A and Supplementary Figure S2B). In mice lacking BAFF in MOs, serum BAFF slightly increased in the first few months postpristane injection but was significantly lower than in Baff fl/fl control mice over the course of the experiment ( Figure 6B and Supplementary Figure S2C). In contrast, BAFF Nph cKO mice initially up-regulated systemic BAFF levels to a similar extent as Baff fl/fl control mice, but at 3 and 6 mo post-pristane injection BAFF was significantly lower in these mice than in the controls ( Figure 6C and Supplementary Figure S2C). Serum BAFF levels in zDC Cre+ , Cx3cr1 Cre+ and Mrp8 Cre+ control mice were similar to Baff fl/fl mice (Supplementary Figure S2D). Therefore, cDC-BAFF is required to sustain homeostatic levels of BAFF in the serum, while BAFF from either MOs or cDCs is required for pristane-induced systemic BAFF levels. Although Nphs are major BAFF producers (10), Nph-BAFF is not required to sustain homeostatic BAFF levels in the serum, but does contribute to the upregulation of serum BAFF at later stages of pristane-induced autoimmunity.

BAFF from DCs and MOs skews MO differentiation toward a more inflammatory Ly6C hi MO during long-term pristane treatment
We next analyzed whether selective depletion of BAFF from DCs, MOs or Nphs altered splenic myeloid and B cell populations. Unexpectedly, we found that BAFF depletion from cDCs or MOs, but not Nphs, affected the phenotype of inflammatory Ly6C hi MO in the spleen (47-49). Specifically, splenic Ly6C int/hi MOs from either 8-9 mo-old untreated or pristane-treated BAFF cDC cKO mice had lower Ly6C MFI levels compared to those from Baff fl/fl mice (Figures 7A-D). The same phenotype was observed in mice lacking BAFF in MOs, but only after pristane injection (Figures 7A-D). In contrast to older mice, Ly6C int/hi MOs from 2-3 mo old mice were not significantly different in BAFF cDC cKO mice, BAFF MO cKO mice and Baff fl/fl mice (data not shown). No changes in Ly6C MFI occurred in Ly6C int/hi DCs (CD11c + ) from BAFF cDC cKO and BAFF MO cKO mice ( Figures 7B, D), suggesting that lack of BAFF production by cDCs or MOs, skewed MO differentiation toward a Ly6C int MO/Mph subset rather than MOderived DCs. Thus, in an autoimmune setting, DC-BAFF and MO-BAFF were required to maintain high expression of Ly6C in inflammatory MOs.

BAFF from MOs and cDCs is required to sustain mature B cell pools in spleen and BM in pristane-treated mice
We then investigated whether the absence of BAFF-producing MOs, DCs or Nphs affected splenic B cell levels in pristane-induced lupus-prone mice. Before pristane injection, B cell subsets in spleens  Figure S3A and data not shown). Thus, the lower BAFF serum levels observed in BAFF cDC cKO mice and BAFF MO cKO mice before pristane injection ( Figure 6) did not affect mature splenic B cells homeostasis, confirming previous findings using BM chimeras (50). In the long-term (6 mo) pristane experiments we left some mice untreated (naïve) as controls. In contrast to the younger mice, these 8-9 mo old naïve mice lacking BAFF in cDCs had lower numbers of mature B cells and PCs than untreated Baff fl/fl mice ( Figure 8A). Specifically, when BAFF was missing in cDCs, spleens had lower numbers of FO, MZ and MZP B cells, while newly formed T2 and T1 B cells were unchanged ( Figure 8A, lower panels). Like older naive mice, pristane-treated BAFF cDC cKO mice had lower mature B cell and PC numbers compared to Baff fl/fl mice ( Figure 8A and Supplementary Figure S3B). However, in addition to mature B cells, pristane-treated BAFF cDC cKO mice had fewer T2 B cells and CD11c + B cells than Baff fl/fl mice, indicating that in pristane-treated mice the lack of BAFF derived from cDCs more broadly affected B cell numbers. In contrast to mice lacking BAFF in cDCs, naïve 8-9 mo old BAFF MO cKO mice had similar numbers of splenic B cells as naïve Baff fl/fl mice ( Figure 8B). However, BAFF from MOs was required to maintain normal mature B cells numbers after pristane injection ( Figure 8B and Supplementary Figure S3B). Lack of BAFF from Nphs did not significantly affect FO or MZ B cells in either 8-9 mo naïve mice or pristane-treated mice ( Figure 8C and Supplementary Figure S3B). Interestingly, pristane-treated BAFF Nph cKO mice had increased numbers of GC B cells compared to Baff fl/fl mice, suggesting that Nph-derived BAFF may affect the generation of GC B cells, induced by pristane ( Figure 8C).
Since cDC-derived BAFF contributed to systemic BAFF levels in naive mice independently of age, and it was required to maintain splenic mature B cells only in older mice, we examined whether BAFF produced by cDCs played a role in mature B cell homeostasis in the BM. BMs from either younger or older, as well as pristanetreated, BAFF cDC KO mice had lower numbers of mature B cells, but similar numbers of B cell precursors, compared to Baff fl/fl mice ( Figure 8D and Supplementary Figure S4). MO-derived BAFF also contributed to mature B cell homeostasis in the BM of 2-3 mo old mice (Supplementary Figure S4). In long-term experiments, the absence of BAFF-producing MOs affected BM mature B cells only after pristane treatment, similarly to what we observed in the spleen ( Figure 8E). BMs from either younger naïve or pristane-treated BAFF MO cKO mice had lower numbers of Pro-B cells, suggesting a possible role for MO-BAFF, but not cDC-BAFF, in earlier stages of B cell development ( Figure 8E and Supplementary Figure S4). In contrast, Nph-derived BAFF did not contribute to mature B cell homeostasis in the BM of pristane treated mice (data not shown).
Taken together our data from

Nphs, MOs and cDCs are all BAFF sources required for the induction of anti-SmRNP IgG autoantibodies in pristane-induced lupus mice
Finally, we tested whether the decrease in systemic BAFF levels in BAFF cDC cKO mice and BAFF MO cKO mice affected the auto-Ab responses induced by pristane. Anti-SmRNP IgG auto-Abs were induced in Baff fl/fl mice and peaked 4-6 mo after pristane injection; however, when BAFF was absent from either cDCs or MOs, the auto-Ab responses were substantially reduced ( Figures 9A, B). Surprisingly, even though systemic levels of BAFF, mature B cells and PCs were normal in mice lacking Nph-BAFF, pristane-induced anti-SmRNP IgG was significantly lower in these mice compared to Baff fl/fl control mice ( Figure 9C). Serum levels of auto-Abs were not reduced in all the corresponding control Cre mice (data not shown). These data suggest that although BAFF-producing Nphs may have a more localized effect on B cells, they still play a significant role in promoting auto-Ab production in this model. In conclusion, BAFF produced from three myeloid subsets-cDCs, MOs. and Nphs-is required for the induction of auto-Abs in a lupus-like autoimmune model.

Discussion
BAFF plays an essential role in the pathogenesis of SLE and other autoimmune diseases. Since the anti-BAFF Ab, Belimumab, was approved for the treatment of SLE, several new drugs targeting different forms of BAFF or the BAFF/APRIL axis are being developed (26, 29, 30). However, still little is known about how, when, and where BAFF regulates the development of autoreactive B cells (15, [27][28][29]51). In the current study, using novel BAFF-RFP reporter mice, BAFF cKO mice and both spontaneous as well an inducible mouse lupus-models, we defined cDC-, MO-and Nph-BAFF as major BAFF sources regulating specific B cell niches. Each of these cellular sources was required for the full development of SLE auto-Ab responses in an inducible lupus-model. We and others have identified BAFF produced by cDCs and MO-derived DCs as essential for B cell and Ab responses in secondary lymphoid organs (10, 11,42,52,53).
Previous studies described peripheral myeloid cells expansion in Tlr7 Tg mice but did not address BAFF expression (36,45). Utilizing heterozygous BAFF-RFP reporter mice, which when crossed with Tlr7.1 Tg or Sle1 mice spontaneously developed a lupus-like phenotype, we found that both mouse models had expanded splenic BAFF + -Nph, -MOs and -cDCs. Furthermore, BAFF expression was up-regulated on a per cells basis in inflammatory MOs in both Tlr7 Tg and Sle1 mice, while CD8 -DCs increased BAFF expression only in the Tlr7 Tg model. In addition to myeloid cells, T cells upregulated BAFF expression and BAFF + T cells were expanded in both models. Although BAFF expression in T cells has been suggested in CD4 T cells in human SLE (54)(55)(56), our results are the first clear demonstration that BAFF expression in T cells may be as high as in myeloid cells in mouse lupus. BAFF + B cells were expanded in the TLR7 Tg mice, and BAFF expression per cell increased in mature B cell subsets. BAFF + PCs and other BAFF + B cells in the peritoneal cavity were expanded in PIL These findings agree with other studies linking the elevated expression of BAFF in B cells in mouse lupus-models and SLE patients and the production of auto-Abs (57,58). The ability of B cells to produce autocrine BAFF and sustain their own survival could play an important role in Type-I IFN and TLR7-driven autoimmunity. The exact contribution of BAFF + T or B lymphocyte populations in the development or exacerbation of autoimmunity needs to be further explored.
Very few studies have investigated the production of tissue or serum BAFF in the PIL model (59-62). Here, we determined that serum BAFF is up-regulated and peaks at one month post-pristane injection. In PIL BAFF was elevated in multiple myeloid cell populations similar to the two spontaneous models. The reduced transcription of BAFF in splenic Nphs from pristane-treated BAFF-RFP mice, as well as BAFF-RFP Sle1 mice may be due to Nph activation as in a Salmonella infection model (11). Nphs constitutively store high levels of BAFF that is released upon activation and drives PC responses (63, 64). Pristane-induced expansion of BAFF-expressing Nphs, MOs and DCs, similar to the spontaneous lupus models, was associated with increased BAFF serum levels, PC's expansion, and induction of auto-Abs. MO recruitment, IFN-I induction and auto-Ab production in PIL all require TLR7/Myd88 signaling (40,65). The inhibition of TLR7mediated activation and BAFF production by DCs protects from PIL (60). Thus, in a lupus model where the TLR7/IFN-I axis is implicated in the development of autoimmunity, BAFF-producing DCs, MOs and Nphs each contribute to the induction of auto-Abs.
All pristane-treated mice in this study developed lipogranulomas (not shown), indicating disease development (65). However, we could not conclusively determine quantitative differences between the groups on lipogranuloma formation. The PIL model on the B6 background has some limitations (39); pristane-treated B6 mice do not develop robust anti-chromatin auto-Ab responses and develop very mild glomerulonephritis (60, 62). Even though renal function was not examined in this study, previous reports have suggested that all three myeloid BAFF sources studied here may be involved in the regulation of IgG deposits in the kidney (60, 62,66). While BAFF overexpression can efficiently amplify underlying predisposition to clinical disease, it may not be sufficient to drive the development of severe glomerular pathology in mice (1,67,68). Therefore, it will be of future interest to test whether the deletion of any of these specific myeloid BAFF sources could improve renal disease in other more robust lupus-prone mice models.
While all three myeloid BAFF sources were required for the induction of auto-Abs, each subset affected systemic BAFF levels and mature B cells pools differently. For example, while either MO-BAFF or cDC-BAFF was required to maintain systemic BAFF levels and mature B cell homeostasis in the BM of younger mice, only BAFF produced by cDCs was required for mature B cell homeostasis in older mice. DC-BAFF was required to sustain mature splenic B cells only in older untreated or pristane treated mice, while MO-BAFF affected the B cell pool only after the induction of autoimmunity. Further studies are needed to assess the involvement of myeloid-derived BAFF in regulating B cell homeostasis during aging, which could also play a role in the development of autoimmunity (69). Nph-BAFF, in contrast to cDC-BAFF and MO-BAFF, did not play a major role in determining systemic BAFF levels or FO and MZ B cell homeostasis. Using long-term neutrophil-depletion, Coquery et al. reported that Nphs contribute to excess serum BAFF levels and B cell responses in a congenic lupus-prone mouse model (70). In contrast, in PIL mice we found that Nph-BAFF was required for auto-Ab production, but only partially supported normal BAFF serum levels. This difference could be due to differences in the two models used or could reflect the possibility that Nph-depletion independently affects other BAFFsources. Surprisingly, conditional BAFF depletion from Nphs increased GC B cell numbers in PIL, implying that Nph-BAFF may play a role in regulating GC formation. Consistent with our finding, Nph depletion early in lupus resulted in a striking acceleration in the onset of lupus renal disease and GC formation in secondary lymphoid organs (71). Given the complex role of Nphs in lupus (66,72), the effect of Nph-BAFF on GC, serum BAFF and auto-Abs in PIL mice can be interpreted in the context of different disease kinetics and the localized effects of Nph-BAFF in specific B cell niches. For example, although Nphs play a major role in extrafollicular (EF) humoral responses, they can also localize in T-cell zones and interact with GC B cells in lupus-prone mice (70, 73). Both EF and GC-dependent B cell activation pathways contribute to pathogenic auto-Ab production in SLE, and clinical data suggest that these two pathways maybe responsible for some heterogeneity in human SLE patients (28). Thus, a possible explanation for our findings is that Nph-derived BAFF may skew toward EF B cell responses, away from GC formation. Our findings are consistent with a growing number of studies suggesting that a BAFF-producing Ly6G lo Nph subset supports Ag-specific humoral responses in lupus as well as after microbial infections (74-76). Thus, our data suggest that Nphs and Nph-BAFF may play a more pathogenic role in the development of autoreactive B cells in lupus and highlight the importance of defining the specific context where Nphs interact with B cells.
Reeves et al. found that in PIL mice, Ly6C hi inflammatory MOs are the primary source of IFN-I (41,77). IFN-I sustains the chronic inflammatory response that leads to autoimmunity, by promoting Ly6C hi inflammatory MOs continuous recruitment to the peritoneum and lymphoid tissues, and blocking their ability to differentiate into Ly6C lo/int MO/Mphs (41,77). Ly6C lo/int MOderived Mphs are mature stages of MO differentiation, more phagocytic and less inflammatory than Ly6C hi MO, they play a major role in the resolution of inflammation, tissue repair and homeostasis (47-49). Here we found that DC-BAFF and MO-BAFF, but not Nph-BAFF, were required to maintain high expression levels of the inflammatory marker Ly6C on MOs in PIL mice. Consistent with our results, BAFF receptors on human MOs are associated with the release of inflammatory mediators, while suppressing phagocytosis and migration in vitro (78)(79)(80). Furthermore, lupus and Sjogren's syndrome patients have abnormally elevated expression of BAFF receptors on MOs (54,81). Thus, an interesting possibility is that DC-BAFF and MO-BAFF contribute to the amplification of the inflammatory cascade by promoting inflammatory Ly6C hi MOs and restraining their differentiation into more mature MOs (47-49). Future research could test whether BAFF directly induces production of IFN-I or other cytokines in MOs creating a positive feedback circuit, and whether BAFF regulation of MOs may play a role in the exacerbation of lupus autoimmunity induced by chronic inflammatory responses (77,82).
As in the PIL model where IFN-I drives the autoimmune response, the high serum BAFF levels we observed in Tlr7.1 Tg mice was likely due to the substantial increase in IFN-I induced by TLR7 overexpression (31,32,36). A similar correlation between upregulation of TLR7 and high BAFF serum levels occurs in immune thrombocytopenia, another autoimmune disorder (83). Both TLR7 and IFN-I signaling, particularly in myeloid populations, are strong drivers of lupus-like pathology and auto-Ab production in Sle mice (84, 85). However, previous studies only found higher BAFF levels in the serum when crossing Sle1 mice with an autoimmune accelerator that overexpresses TLR7, resulting in severe systemic autoimmunity and kidney disease (38,86). Thus, the increase in systemic BAFF in Sle1 mice, which has been previously overlooked, further supports a role for BAFF in the induction of autoreactive B cells and auto-Abs in spontaneous lupus-like disease models (14,15). Increases in systemic BAFF levels correlate with disease severity and auto-Ab responses in other lupus-prone mice like MRL lpr/lpr mice and NZBWF1 (20, 70, 87), as well as in SLE patients (22)(23)(24)(88)(89)(90).
Humoral autoimmune responses in our parental lines were consistent with those described by us and others in Tlr7.1 Tg mice (31)(32)(33)45), and Sle1 mice (35,38). Notably, the induction of auto-Abs and the increase in PC numbers in BAFF-RFP Tlr7.1 Tg mice and BAFF-RFP Sle1 mice, which had only one functional BAFF allele and reduced mature B cells, was comparable to that of their parental lines. Similarly, high PC numbers and some auto-Abs were induced in BAFF-RFP mice also after pristane-induced autoimmunity. In autoimmune settings transitional/newly formed (NF) B cells may become more responsive to BAFF and play a role in the development of autoimmunity including SLE (28,33,44,(91)(92)(93)(94). Since, NF B cells can differentiate in PCs and produce IgG Abs (9,28,33,91,92), it is possible that in the BAFF-RFP mice NF B cells contributed to auto-Abs responses in the lupus-prone models.
Despite the redundancy in the BAFF system, depletion of BAFF in anyone of the three myeloid cell types, DCs, MOs or Nphs was sufficient to ameliorate humoral autoimmunity. However, each BAFF source has a different role in regulating BAFF serum levels and seems to sustain mature B cells acting through different mechanisms. Consistent with our results, a few studies have pointed out a possible involvement of MO-and DC-BAFF in human SLE (95,96). IFNa and SLE-immune complexes (IC) induce BAFF production and mobilization by blood MOs and DCs in SLE-patients with active disease, but not in healthy individuals or SLE patients with non-active disease (95). Serum from SLE patients via IFN-I instructs MOs to differentiate into DCs, which in turn mediate IgG-plasmablast differentiation via BAFF and IL-10 (96). Also, recent studies in mice and humans have highlighted how inflammatory MO-derived DCs and the plasticity of MOs play a critical role in mediating SLE autoimmunity and pathogenesis (77,82,97,98). In addition, specific monocyte/DC signatures related to clinical disease, type I IFN signatures and responsiveness to treatments have been identified in Lupus patients (99,100). Thus, it would be interesting to see whether BAFF expression correlates with different myeloid cell-signatures in pathological conditions and the ability to respond to BAFFneutralizing agents. It is possible that some of the BAFF heterogeneity in SLE patients may reflect variable responses to inflammatory signals that induce BAFF in different subsets/ locations (27,28,101). Our data, suggest that specific BAFF sources and B cell niches, implicated in the development of autoreactive B cells, may be important to consider when designing new lupus treatments (3,19,28,51). In a recent study, BAFF gene silencing has effectively improved disease outcome in experimental SLE (102). DC-and MO-targeting and new technologies are starting to show promising results in autoimmune diseases (103, 104). Thus, targeting upstream signals of BAFF production by myeloid cell populations, or targeting BAFF in specific myeloid cells or lymphoid compartments, could be helpful in future lupus therapies (105).

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

Ethics statement
The animal study was reviewed and approved by University of Washington Institutional Animal Care and Use Committee.

Funding
This work was supported by a grant from the National Institutes of Health (R01 AI44257, EC/NG).