Reduced B Lymphoid Kinase (Blk) Expression Enhances Proinflammatory Cytokine Production and Induces Nephrosis in C57BL/6-lpr/lpr Mice

BLK, which encodes B lymphoid kinase, was recently identified in genome wide association studies as a susceptibility gene for systemic lupus erythematosus (SLE), and risk alleles mapping to the BLK locus result in reduced gene expression. To determine whether BLK is indeed a bona fide susceptibility gene, we developed an experimental mouse model, namely the Blk+/−.lpr/lpr (Blk+/−.lpr) mouse, in which Blk expression levels are reduced to levels comparable to those in individuals carrying a risk allele. Here, we report that Blk is expressed not only in B cells, but also in IL-17-producing γδ and DN αβ T cells and in plasmacytoid dendritic cells (pDCs). Moreover, we found that solely reducing Blk expression in C57BL/6-lpr/lpr mice enhanced proinflammatory cytokine production and accelerated the onset of lymphoproliferation, proteinuria, and kidney disease. Together, these findings suggest that BLK risk alleles confer susceptibility to SLE through the dysregulation of a proinflammatory cytokine network.


Introduction
Systemic lupus erythematosus (SLE) is a chronic multisystem autoimmune disorder that afflicts more than 1.5 million Americans. There is strong evidence for a genetic basis to this disease, and many candidate genes, which predispose an individual to SLE, have been identified from studies in patients with SLE and in mouse models of lupus [1][2][3]. With recent advances, however, such as the completion of the Human Genome Project and the International HapMap Project, it is now possible to perform genome-wide association studies to identify additional susceptibility genes in humans. Indeed, several groups, using this experimental approach, have identified and confirmed over 25 new susceptibility genes in SLE patients of different ethnicity and race [4][5][6][7][8][9][10]. Notably, many of these new susceptibility genes are not among those known to be associated with autoimmune disease; therefore, follow-up studies are necessary to determine the mechanisms by which they promote development of SLE.
One of the newly identified susceptibility genes is BLK, which encodes B lymphoid kinase (Blk). Multiple single nucleotide polymorphisms (SNPs) in the BLK locus, mapping primarily to the promoter and first intron, are associated with disease risk [4][5][6][7][8][9][10]. A handful of these SNPs have been studied in more depth to determine how the specific nucleotide change affects BLK expression. All studies to date report a 25 to 70% reduction in BLK expression depending on whether individuals are heterozygous or homozygous for the risk allele [5,[11][12][13]. These findings suggest that the genetic variants in the BLK locus predispose an individual to SLE by reducing Blk expression.
Blk was first described over two decades ago as a B cell-specific member of the Src family of tyrosine kinases (SFKs) [14]. Even though early reports demonstrated functional redundancy among Blk, Lyn, and Fyn in B cell development and activation [15,16], a recent report has revealed a requirement for wild-type levels of Blk in the development and function of marginal zone (MZ) B cells [17], a mature splenic B cell subset involved in both microbial immunity and autoimmunity [18]. In both Blk +/2 and Blk 2/2 mice, MZ B cells are fewer in number but exhibit augmented in vitro and in vivo responses to BCR stimulation in comparison to Blk +/+ mice [17]. With age, this B cell hyperactivity leads to autoimmunity, as evidenced by the display of multiple autoimmune phenotypes in 6-month-old Blk +/2 mice, including increased numbers of MZ and B1 B cells, detection of B cells with an activated surface phenotype, and production of a low but significant level of serum anti-nuclear antibodies (ANAs) [17]. Given the well-documented role of B cells in autoimmune disease development and pathogenesis [19], these data suggest that BLK risk alleles promote development of SLE by altering BCR signaling responses and, by extension, B cell development, function, and tolerance.
It is important to note that Blk is also expressed outside of the B cell lineage, in both immune and non-immune cells. In humans, Blk is expressed in unfractionated thymocytes, cd T cells, plasmacytoid dendritic cells (pDCs), and pancreatic b cells [13,[20][21][22], while in mice, it is expressed in bone marrow progenitor cells, immature CD4 2 CD8 2 (double negative; DN) thymocytes, cd thymocytes, IL-17-producing cd T (cd-17) cells, and pancreatic b cells [22,23]. More importantly, analysis of Blkdeficient mice has revealed a requirement for Blk not only in early T cell development but also in the development and function of cd-17 cells [23]. Therefore, because Blk is expressed in a diverse array of immune cells, it is conceivable that reducing its expression could have wide-ranging effects on immune cell development, activation, and effector function.
To determine whether and how reduced Blk expression levels contribute to autoimmune disease development and pathogenesis, we established an experimental mouse model in which Blk transcript and Blk protein levels are reduced by approximately 45% [17], which is within the range reported for individuals carrying a BLK risk allele [5,[11][12][13]. In addition, the mouse model carries the lpr mutation in Fas, which not only results in severely reduced Fas expression but also accelerates the development of disease when introgressed on an autoimmune susceptible background [24,25]. We present here that, in addition to B cells and cd-17 cells, Blk is expressed in murine pDCs and in IL-17producing DN ab T (DN-17) cells. Furthermore, we found that solely reducing Blk expression in B6.lpr mice enhanced proinflammatory cytokine production by both Blk-positive andnegative immune cells and accelerated the onset of lymphoproliferation, proteinuria, and kidney disease. These findings demonstrate that BLK is indeed a bona fide susceptibility gene and suggest that BLK risk alleles promote autoimmune disease development through the dysregulation of a proinflammatory cytokine network.

Ethics statement
All research involving animals has been conducted according to the relevant national and international guidelines with respect to husbandry, experimentation and welfare. Mouse protocols were approved by the State University of New York (SUNY) Upstate Medical University Committee on the Humane Use of Animals (CHUA protocol numbers 262 and 368).

Mice
C57BL/6J (B6) and B6.MRL-Fas lpr /J (B6.lpr) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). B6.Blk tm1 (Blk 2/2 ) mice [15] were provided by A. Tarakhovsky (Rockefeller University) and V H 3H9 site directedtransgenic (3H9 Tg) mice [26] on the B6 background were provided by M. G. Weigert (University of Chicago). All mice used in this study were bred and maintained in a barrier facility in the Department of Laboratory Animal Resources at SUNY Upstate Medical University in accordance with the specifications of the Association for Assessment and Accreditation of Laboratory Animal Care.

Flow cytometric analysis
Flow cytometric analysis for surface antigen expression was performed by pre-incubating cells with the anti-CD16/CD32 antibody for at least 10 minutes to block non-specific binding of immunoglobulins to Fc receptors, followed by staining with fluorochrome-conjugated mAbs against various surface antigens. Intracellular staining for Blk, RORct, T-bet, and Foxp3 was performed using the Foxp3/Transcription factor staining buffer set (eBioscience, San Diego, CA, USA) according to the manufacturer's instructions. Intracellular staining for IL-6, TNFa, IL-17A, IFNc, and IL-10 was performed by first fixing cells in a final concentration of 1.5% formaldehyde for 10 minutes at 37uC. Fixed cells were stained for surface antigens, permeabilized with Perm/Wash Buffer (BD Pharmingen, San Jose, CA, USA) for 20 minutes at 4uC, and then stained with antibodies against the appropriate cytokines. For all experiments, 0.5 to 1 6 10 6 cells were acquired on a BD LSRFortessa using FACSDiva software (BD Immunocytometry Systems, San Jose, CA, USA). Data analysis was performed using FlowJo software (Tree Star, Inc., Ashland, OR, USA). Dead cells were excluded from analysis based on forward and side scatter profiles.

ELISAs
Cytokine, IgM, IgG, and autoantibody levels in supernatants or sera were determined with ELISA kits purchased from eBioscience, BioLegend, R and D systems (Minneapolis, MN, USA), or Alpha Diagnostics (San Antonio, TX, USA).

In vitro stimulation of immune cell subsets
To assess pDC function, pDCs from 2-month-old B6.lpr or Blk +/2 .lpr spleens were purified by negative selection using the Mouse Plasmacytoid Dendritic Cell Isolation Kit II (Miltenyi, Auburn, CA, USA) and then stimulated with Type C CpG ODN 2395 (1 mg/ml; Invivogen, San Diego, CA, USA) for 24 hours at 37uC. Supernatants were collected and assayed for IFNa by ELISA.
To polarize naïve CD4 + T cells towards the Th17 and Tfh effector lineages, naïve CD4 + T cells were purified from the spleens and peripheral lymph nodes (pLNs) of B6 mice by negative selection using magnetic bead separation as previously described [23]. Cells were stimulated with 5 mg/ml of immobilized anti-CD3 antibody and 1 mg/ml of soluble anti-CD28 antibody in the presence of 1 ng/ml anti-IFNc antibody, 1 ng/ml anti-IL-4 antibody, and either IL-6 (100 ng/ml; Miltenyi) for Tfh polarization or IL-6 (100 ng/ml) plus TGFb (1 ng/ml; Miltenyi) for Th17 polarization. After 3 days at 37uC, cells were harvested and their Blk expression levels were analyzed by flow cytometry.
For all other in vitro stimulation assays, 3 6 10 6 pLN or spleen cells from 3-month-old B6, Blk +/2 , B6.lpr, or Blk +/2 .lpr mice were cultured for 4 hours at 37uC in the presence of brefeldin Acontaining Leukocyte Activation Cocktail (BD Biosciences) to assess cytokine production by various immune cell subsets or in the presence of Cell Stimulation Cocktail (eBioscience) to measure cytokines in the supernatant.

Histopathology
Kidneys, lungs, and livers were harvested, fixed in 10% neutralbuffered formalin, and then embedded in paraffin. For light microscopy, tissue samples were sectioned at 4 mm and then stained with H&E or periodic acid-Schiff (PAS). For electron microscopy, specimens were chosen from paraffin-embedded kidneys, using H&E stained sections to locate glomeruli. Speci-mens were reprocessed as follows: dewaxing in xylene, rehydrating in progressively diluted ethanol, post-fixing in osmium tetraoxide, dehydrating in progressively concentrated ethanol, infiltrating with propylene oxide, embedding in Epon, and then sectioning with an ultramicrotome. Glomerular damage was assessed by a renal pathologist (A.H.T.), who was blinded to the genotype of the kidney samples.
Quantitative real-time RT-PCR analysis and RT 2 Profiler PCR array RNA was isolated from homogenized kidneys and splenic B cells using Qiagen's RNeasy kit, and cDNA was synthesized using Invitrogen's SuperScript First-Strand Synthesis System. The expression of genes associated with the IL-23/IL-17 axis was assessed using Qiagen's ''Th17 for Autoimmunity and Inflammation'' RT 2 Profiler PCR array. For quantitative real-time PCR analysis, all primer sets, as well as the SYBR Green PCR Master Mix, were purchased from Qiagen (Valencia, CA, USA).

Listeria Infection
Infection of mice with Listeria monocytogenes and subsequent analysis of T cell effector function were performed as previously described [27], except that a cocktail of IL-23 (50 ng/ml; BioLegend), IL-1 (10 ng/ml; BioLegend), and Pam 3 Cys (1 mg/ ml; Invivogen) was used to elicit IL-17 production from both cd-17 and DN-17 cells.

Statistical analysis
Data are presented as either values for individual mice or the mean 6 SEM for a group of mice. All statistical analyses were performed using GraphPad Prism software (La Jolla, CA, USA). The Student's t test was used to analyze parametric data, the Mann-Whitney U test was used to analyze nonparametric data, and the Spearman's rank order correlation was used to measure the strength of the relationship between TNFa serum levels and urine protein concentrations and between IgG serum levels and cytokine serum levels.

Results
Blk is expressed in murine B cells, cd and DN ab T cells with the potential to produce IL-17, and pDCs The first step in determining the contribution of BLK risk alleles to SLE development and pathogenesis is to identify which cells express this SFK. We have previously shown, using an intracellular flow cytometric assay that we developed to measure Blk expression levels, that Blk is expressed in T lineage cells, specifically immature DN thymocytes, cd thymocytes and cd-17 cells [23]. Here, we have extended these previous findings to include DN-17 cells ( Figure 1A), which are defined by their expression of RORct, a transcription factor required for IL-17 expression [28]. Blk, however, is not a universal marker for IL-17producing cells, as other cell types that produce IL-17, such as T helper 17 (Th17) cells, CD4 + NKT cells and NK cells, did not express Blk ( Figure 1B). Moreover, Blk was not detected in Th1  [29,30]. Last, we did not observe Blk expression in CD4 + and CD8 + T cells from B6.lpr mice ( Figure 1C), indicating that neither the expression of the Fas lpr mutation in T cells nor the microenvironment induced by the expression of the Fas lpr mutation resulted in abnormal expression of Blk in conventional ab T cells. These data indicate that Blk expression in mature T lineage cells is limited to unconventional IL-17-producing T cells.
Blk, along with other components of the BCR-coupled signaling pathway, are expressed by human pDCs [21]. Notably, we found that murine pDCs also express Blk, albeit at levels lower than those observed in B cells ( Figure 1D). No Blk expression, however, was detected in murine macrophages and conventional dendritic cells (cDCs) ( Figure 1D). These data demonstrate that Blk is expressed in B cells, cd-17 cells, DN-17 cells, and pDCs. Although these are functionally disparate cells, it is important to note that all of these cell types play a role in autoimmune disease development and/or pathogenesis [31][32][33][34].
Development of proteinuria and nephrosis in 5-monthold Blk +/2 .lpr mice B6.lpr mice do not manifest severe autoimmune disease, specifically glomerulonephritis, until 9 months of age [35]. Consequently, we reasoned that if BLK were a bona fide susceptibility gene, then reducing its expression would either accelerate the onset of glomerulonephritis or increase its incidence and severity in B6.lpr mice. To test this, we monitored the concentration of protein in the urine, as a measure of glomerular barrier function, starting at 3 months of age. At 5 months of age, 60% of the Blk +/2 .lpr mice, but none of the B6.lpr mice, displayed proteinuria (defined as $100 mg/dl) ( Figure 2A). Surprisingly, by light microscopy, none of the Blk +/2 .lpr mice showed mesangial proliferation or any other signs of immune complex (IC)-induced inflammation ( Figure 2B). In agreement with this, we found no difference in the serum levels of ANAs between 5-month-old B6.lpr and Blk +/2 .lpr mice ( Figure S1A). We did note, however, narrowing of the capillary lumens and PAS-positive hyaline deposits in the glomeruli of Blk +/2 .lpr mice but not in those of B6.lpr mice ( Figure 2B). To explain the development of proteinuria in the absence of IC-induced inflammation, we examined the ultrastructure of the Blk +/2 .lpr glomerulus by electron microscopy. In addition to the narrowed capillary lumens and hyaline deposition, we observed damage to the podocyte, a component of the glomerular filtration barrier, as evidenced by the considerable focal shortening (i.e., effacement) of their foot processes in the Blk +/2 .lpr glomerulus ( Figure 2C, Figure S2). Taken together, these data indicate that 5-month-old Blk +/2 .lpr mice suffer from nephrosis, a kidney disease that is frequently observed in SLE patients with renal involvement [36,37]. In fact, a recent classification, termed lupus podocytopathy, has been developed to describe SLE patients who present with nephroticrange proteinuria and podocyte foot process effacement but no ICinduced inflammation [36,38].
One possible explanation for the development of proteinuria in Blk +/2 .lpr mice is that Blk is expressed by resident kidney cells, and that reducing its expression leads to impaired glomerular barrier function. To test this, we used quantitative RT-PCR analysis to determine whether Blk, along with other SFK genes, are expressed in wild-type B6 kidney. Notably, we found that cells within the kidney expressed Src, Lyn, and Fyn transcripts but not Blk transcripts ( Figure 2D), which is consistent with recent reports showing that neither the Blk gene nor the Blk protein is expressed in total kidney, kidney cell lines and primary podocytes [39,40]. Nonetheless, to rule out an intrinsic role for Blk in maintaining glomerular barrier function, we determined whether Blk-deficient mice develop proteinuria with age. As shown in Figure 2E, only trace amounts of protein are detected in the urine of 3-to 4month-old as well as of 7-to 9-month-old Blk 2/2 mice. Taken together, these data suggest that the development of proteinuria in 5-month-old Blk +/2 .lpr mice is through an extrinsic mechanism.
In both humans and mice, there is a strong correlation between serum levels of TNFa and the severity of proteinuria [41,42]. Accordingly, another possible explanation for the development of proteinuria in Blk +/2 .lpr mice is that reducing Blk expression in B6.lpr mice promotes TNFa production. To test this, we compared TNFa serum levels in 5-month-old B6.lpr and Blk +/2 . lpr mice and found that they were indeed significantly higher in Blk +/2 .lpr mice than in B6.lpr mice ( Figure 2F). Moreover, TNFa levels correlated with severity of proteinuria (r 2 = 0.594; p = 0.02), suggesting that elevated TNFa levels contribute to the development of proteinuria, and possibly nephrosis, in Blk +/2 .lpr mice.
Notably, other phenotypes that are associated with autoimmune disease were also observed in 5-month-old Blk +/2 .lpr mice. These included splenomegaly and multifocal lymphocytic infiltration of the lung and liver ( Figure 3). Collectively, these data indicate that solely reducing Blk expression levels in B6.lpr mice leads to early onset lymphoproliferation, proteinuria and nephrosis. To gain insight into the mechanisms by which Blk expression levels regulate autoimmune disease development, we performed a phenotypic and functional analysis of Blk +/2 .lpr mice, prior to the onset of proteinuria and nephrosis, in order to exclude any diseaserelated effects on immune cell function. We settled on 3 months of age because, at this age, Blk +/2 .lpr mice exhibit lymphoproliferation but have minimal levels of serum autoantibodies and no serum TNFa nor proteinuria (Table 1, Figure S1B and C, data not shown).

Blk is required to regulate B cell functional responses not B cell tolerance
We have previously shown that MZ B cell numbers are significantly decreased in 2-month-old Blk +/2 mice but, with age, MZ B cell numbers increase and, at 6 months of age, even surpass those observed in age-matched B6 mice [17]. At 3 months of age, in both Blk +/2 and Blk +/2 .lpr mice, we found that MZ B cell numbers were equivalent to those in B6 and B6.lpr mice, respectively, although percentages of MZ B cells were significantly reduced in comparison to B6 and B6.lpr mice (Table 1, Figure S3). Except for an increase in the number of splenic B1 B cells in Blk +/ 2 mice compared to B6 mice, no differences in the numbers of mature B cell subsets were observed between B6 and Blk +/2 mice and between B6.lpr and Blk +/2 .lpr mice at 3 months of age (Table 1).
One way B cells contribute to autoimmunity is through the production of autoantibodies [31]. We found that the serum levels of ANAs and anti-cardiolipin antibodies in 3-month-old B6.lpr and Blk +/2 .lpr mice were low and were comparable to those in agematched B6 and Blk +/2 mice (Figure S1C, data not shown). Nonetheless, we did observe significant increases in the numbers of plasma (CD138 + ) cells, both short-lived (B220 + ) and long-lived (B220 2 ), and in the total serum levels of IgM and IgG in Blk +/2 .lpr mice compared to B6.lpr mice ( Figure 4A-C). Notably, we also observed more long-lived plasma cells and higher IgM serum levels in Blk +/2 mice than in B6 mice ( Figure 4B-C). Together, these results suggested that reducing Blk expression leads to generalized B cell hyperactivity and not to a loss of B cell tolerance. To investigate this further, we generated and analyzed Blk +/2 mice carrying the well-described 3H9 IgH transgene, which forms an anti-DNA antibody when paired with most endogenous light chains [26]. Phenotypic analysis of 6-month-old Blk +/+ 3H9 Tg and Blk +/2 3H9 Tg mice revealed no differences in the percentage and number of anti-DNA (Igl1 + IgM a+ ) B cells, and that the vast majority of anti-DNA B cells in both genotypes had similar surface phenotypes and were arrested at the transitional (CD93 + ) B cell stage ( Figure 5A-D). More importantly, equivalent levels of anti-dsDNA IgG were detected in the sera of 6month-old Blk +/2 3H9 Tg and Blk +/+ 3H9 Tg mice ( Figure 5E).  These data indicate that Blk does not act intrinsically to regulate B cell tolerance. In addition to autoantibody production, B cells contribute to autoimmunity by secreting effector cytokines and functioning as antigen presenting cells (APCs) [31,[43][44][45]. Accordingly, we assessed the ability of B cells from 3-month-old B6, Blk +/2 , B6.lpr, and Blk +/2 .lpr mice to secrete proinflammatory and antiinflammatory cytokines and to express surface antigens associated with APC function. Following a short-term PMA/ionomycinstimulation, we found that B cells from all four genotypes produced IL-6, IFNc, and IL-10, but not TNFa ( Figure 4D, data not shown). When we quantified the number of cytokineproducing B cells, we noted ,2-fold more IL-6 + B cells in Blk +/2 . lpr mice than in B6.lpr mice ( Figure 4D and E). Importantly, this same difference in IL-6 + B cell numbers was also noted in Blk +/2 mice in relation to B6 mice ( Figure 4D and E). However, no differences in the numbers of IFNc + and IL-10 + B cells were observed among B6, Blk +/2 , B6.lpr, and Blk +/2 .lpr mice (data not shown). Interestingly, the IL-6 + B cells in all four genotypes were CD21 lo/2 ( Figure 4D), which is the same phenotype as the B cells that are the major source of IL-6 in mice with experimental autoimmune encephalomyelitis [45].
Regarding the ability of B cell subsets to function as APCs, we found that CD86 and IA b surface levels were equivalent between B6.lpr and Blk +/2 .lpr B cells (data not shown), but ICOSL surface levels on follicular (FO) B cells from Blk +/2 .lpr mice were reduced compared to those on FO B cells from B6.lpr mice ( Figure 6A). This phenomenon cannot be explained by an increase in the size of Blk +/2 .lpr FO B cells, as their cell size was comparable to those of B6 and B6.lpr FO B cells ( Figure 6B). Since BAFF-R signaling regulates ICOSL expression on B cells [46,47], we next determined whether a decrease in BAFF serum levels, BAFF-R expression levels, or both can explain the lower ICOSL expression on Blk +/2 .lpr FO B cells. Notably, while BAFF serum levels were higher, BAFF-R expression levels on B cells were lower, in both Fas lpr genotypes compared to both Fas wt genotypes ( Figure 6C and D). Nonetheless, as no difference was observed between B6.lpr and Blk +/2 .lpr mice in either their BAFF serum levels or their BAFF-R expression levels, it is unlikely that reduced BAFF-R signaling accounts for the lower ICOSL expression on Blk +/2 .lpr FO B cells. It is possible, however, that the reduced ICOSL expression is a consequence of prior contact with ICOS + T cells, as ICOSL expression on B cells is downregulated following engagement with ICOS on T cells [46,48,49]. To test this, we examined splenic T cell subsets for the expression of ICOS, a costimulatory molecule that is only expressed on activated T cells [50]. ICOS surface levels were higher on all T cell subsets from B6.lpr and Blk +/2 .lpr mice than on those from B6 and Blk +/2 mice; however, the CD4 + and CD8 + T cell subsets from Blk +/2 .lpr mice expressed significantly higher levels of ICOS than their B6.lpr counterparts ( Figure 6E). These data suggest that the reduced ICOSL expression on Blk +/2 .lpr FO B cells is due to augmented ICOS-ICOSL interactions in Blk +/2 .lpr mice. Thus, B cells in 3-monthold Blk +/2 .lpr mice have the capacity to play an autoantibodyindependent role in the early stages of disease development, as evidenced by their enhanced ability to secrete IL-6 and to function as APCs.

Blk is required to regulate T cell-mediated proinflammatory cytokine production
In light of the hyperactive B cell phenotype in 3-month-old Blk +/2 .lpr mice, it was important to assess its effect on T cell numbers, phenotype and function. Interestingly, total T cell numbers were significantly higher in the Blk +/2 .lpr spleen than in the B6.lpr spleen, with increased numbers of CD4 + and DN ab T cells accounting for the higher T cell numbers (Table 1, Figure  S4A and B). We also noted increased numbers of CD4 + , CD8 + , DN ab, and cd T cells in the Blk +/2 spleen relative to the B6 spleen (Table 1). However, despite the increase in total splenic T cell numbers in Blk +/2 .lpr mice, the relative percentages of CD69 + cells within the ab and cd T cell subsets, as well as the relative percentages of CD4 + T cells with a memory cell phenotype (CD62L l uCD44 hi ), were comparable between the two Fas lpr genotypes ( Figure S4C and D). Last, there was no difference in regulatory T cell numbers between B6 and Blk +/2 mice and between B6.lpr and Blk +/2 .lpr mice (Table 1, Figure S4A).
Given that ICOS-ICOSL interactions appear to be augmented in Blk +/2 .lpr mice and that ICOS-ICOSL signaling is associated with Th1, Tfh and Th17 effector fates [51][52][53][54], we next enumerated effector subsets and/or cytokine-producing cells in B6.lpr and Blk +/2 .lpr mice. Significant increases in the numbers of Tfh cells as well as in the numbers of IFNc-producing effector subsets were observed in Blk +/2 .lpr mice relative to B6.lpr mice ( Figure 7A, D, E). Furthermore, using both IL-17 production and phenotype (RORct + CCR6 + ) to identify cd-17 cells, we detected significantly more cd-17 cells in Blk +/2 .lpr mice than in B6.lpr mice ( Figure 7B, C, F). The numbers of IL-17 + CD4 + and DN ab T cells, on the other hand, were equivalent between the two Fas lpr genotypes ( Figure 7B, C, F). Lastly, consistent with increased local production of these proinflammatory cytokines, we found that the serum levels of IFNc, IL-17A and IL-21 were significantly higher in Blk +/2 .lpr mice than in B6.lpr mice ( Figure 8). Together, these data demonstrate that reducing Blk expression in B6.lpr mice enhances proinflammatory cytokine production by both Blkpositive and -negative T cells.
We have previously shown that Blk-deficiency results in a selective loss of cd-17 cells [23]. For this reason, we were intrigued to find that there were 50% more cd-17 cells in Blk +/2 mice than in B6 mice ( Figures 7F and 9A). As this number is quite different than the predicted one, based on a gene dose effect, of 50% fewer cd-17 cells than B6 mice, we conclude that Blk-haploinsufficiency and Blk-deficiency differentially affect the development and/or homeostasis of cd-17 cells. To determine whether Blk-haploinsufficiency and Blk-deficiency also have different effects on cd-17 cell function, we compared the in vivo cd-17 response in B6, Blk +/2 and Blk 2/2 mice following infection with Listeria monocytogenes, a model pathogen known to elicit a robust cd-17 response [55]. On day 5 post infection, we found that the percentage of IL-17 + cd T cells is significantly higher in infected Blk +/2 mice than in infected B6 mice, while virtually no IL-17 + cd T cells are detected in infected Blk 2/2 mice ( Figure 9B). Taken together, these data demonstrate that Blk-deficiency and Blk-haploinsufficiency have opposing effects on the generation and differentiation of cd-17 cells.
Blk is required to regulate macrophage-mediated IL-6 production Notably, macrophages, in addition to B cells, produced elevated amounts of IL-6 in Blk +/2 and Blk +/2 .lpr mice. Specifically, we observed a 3-fold increase in the number of IL-6 + macrophages in Blk +/2 mice compared to B6 mice and in Blk +/2 .lpr mice compared to B6.lpr mice ( Figure 10A). By contrast, nominal IL-6 production was detected in cDCs from all four genotypes (data not shown). Moreover, when we assayed for other proinflammatory cytokines, such as IL-1b, IL-12, IL-18, and IL-23, in the sera and the supernatants of short-term PMA/ionomycin-stimulated splenocytes from B6.lpr and Blk +/2 .lpr mice, we were only able to detect IL-18, and it was produced in equivalent quantities in both Fas lpr genotypes (data not shown). These results demonstrate that reducing Blk expression in B6.lpr mice increases IL-6 production from macrophages, an immune cell that does not express Blk.
Blk regulates pDC-mediated IFNa production pDCs play an important role in autoimmunity through their production of type I interferon [32]. To determine whether their function is altered in Blk +/2 .lpr mice, we measured serum levels of IFNa in B6.lpr and Blk +/2 .lpr mice. Surprisingly, we observed significantly less IFNa in the sera of Blk +/2 .lpr mice than in the sera of B6.lpr mice ( Figure 10B). The decrease in serum IFNa levels is not the result of a defect in pDC development, as increased numbers of pDCs were detected in Blk +/2 .lpr mice compared to B6.lpr mice (Table 1). However, the decrease in serum IFNa levels may be due to TLR tolerance, as a result of chronic TLR stimulation [56], since purified pDCs from younger Blk +/2 .lpr mice were able to secrete IFNa, and at higher levels than pDCs from age-matched B6.lpr mice, following in vitro CpG stimulation ( Figure 10C). Alternatively, the decreased serum levels of IFNa may be due to high local production of TNFa, which is known to inhibit INFa production from pDCs [57]. Although there were equivalent percentages of TNFa-producing T cells, DCs and macrophages following short-term PMA/ionomycin stimulation ( Figure 10D), considerably more TNFa + splenic T cells are present in Blk +/2 .lpr mice than in B6.lpr mice, as a result of there being 40% more T cells in the Blk +/2 .lpr spleen than in the B6.lpr spleen (Table 1). Collectively, these data suggest that the proinflammatory microenvironment established in the lymphoid tissue of Blk +/2 .lpr mice negatively regulates IFNa secretion by pDCs.

Discussion
With the completion of both the Human Genome Project and the International HapMap Project, researchers are able to scan markers across the human genome to identify genes that contribute to complex human diseases, such as SLE [62]. One of the newly identified SLE susceptibility genes is BLK [4][5][6][7][8][9][10], and SNPs that map to this locus result in a reduction in BLK expression [5,[11][12][13]. In this report, we follow-up on the identification of BLK as a susceptibility gene by determining whether and how reduced BLK expression contributes to SLE disease development. Using the autoimmune-prone B6.lpr mouse model, in which we solely reduced Blk expression levels, we noted enhanced proinflammatory cytokine production by both Blk-positive and -negative immune cells in addition to early onset lymphoproliferation, proteinuria, and kidney disease. Together, these findings not only confirm BLK as a bona fide susceptibility gene for SLE but also reveal new functions for Blk in immune cell activation and regulation.
Systemic autoimmune disease is the result of synergistic actions of multiple susceptibility genes, with each susceptibility gene making a small contribution to disease development [63]. By analyzing both Blk +/2 and Blk +/2 .lpr mice, we are unraveling how reduced Blk expression alters immune cell activation and in turn contributes to the development of autoimmunity. The increased production of IL-6 by B cells and macrophages in 3-month-old  Blk +/2 mice may foster a proinflammatory microenvironment that supports the development and/or maintenance of T cell subsets that have the potential to produce IFNc, IL-17A, IL-21 and TNFa. Indeed, IL-6 has been shown to play a role in the generation of these cytokine-producing T cells [64][65][66]. Not until the Fas lpr mutation is introduced onto the Blk +/2 background, however, do these T cell subsets become activated and secrete their respective cytokines. Enhanced ICOS-ICOSL signaling in the Blk +/2 .lpr mouse model may contribute to their differentiation into cytokine-producing effectors at 3 months of age. In Blk +/2 .lpr mice, the reduced ICOSL expression on FO B cells combined with the increased ICOS expression on CD4 + and CD8 + T cells strongly suggest prior and augmented ICOS-ICOSL signaling [46,48,49]. Interestingly, similar changes in ICOS and ICOSL expression are also observed in SLE patients [49] and in NZB/W F 1 mice, a lupus mouse model in which ICOS/ICOSL interactions are known to contribute to disease development [46]. Importantly, the interactions between B cells and T cells are reciprocal at this stage, as evidenced by increases in plasma cell numbers and serum IgG levels in Blk +/2 .lpr mice compared to B6.lpr mice. Eventually, the increased production of TNFa, along with other proinflammatory cytokines, culminates in accelerated onset of proteinuria in 60% of 5-month-old Blk +/2 .lpr mice. This high degree of penetrance in our experimental model is striking given that a similar incidence of proteinuria is observed in MRL.lpr mice at 3 to 4 months of age [67] and in NZB/W F 1 mice at 6 to 7 months of age [46].
The vast majority of allelic polymorphisms, on which gene expression studies have been performed, result in a change in protein expression or activity, not a null mutation [5,11,12,68]. Adjacent graph compares the number of IL-6 + macrophages between 3-month-old B6 (n = 4) and Blk +/2 (n = 4) mice and between 3-month-old B6.lpr (n = 7) and Blk +/2 .lpr (n = 8) mice. (B) Comparison of serum IFNa levels between 3-month-old B6 and Blk +/2 mice and between 3-month-old B6.lpr and Blk +/2 .lpr mice. Each symbol represents an individual mouse. (C) Purified pDCs from 2-month-old B6.lpr (n = 3) and Blk +/2 .lpr (n = 3) mice were stimulated with CpG ODN 2395 for 24 hours and supernatants were collected and assayed for IFNa by ELISA. (D) Splenocytes from 3-month-old B6.lpr (n = 4) and Blk +/2 .lpr mice (n = 4) were stimulated with Leukocyte Activation Cocktail for 4 hours. Dot plots showing CD3 versus TNFa expression in gated CD3 + T cells (top), B220 versus TNFa expression in gated DCs (center), and F4/80 versus TNFa expression in gated macrophages (MWs) (bottom). Numbers represent percentages of TNFa + cells. Very few (#0.5%) B cells were TNFa + . (E) Expression data for Il17a, Il17f, Il17c and Il23a from RT 2 profiler array. Data were normalized to Gapdh levels and are presented as fold change over 3-month-old B6.lpr kidney (set to 1). Data represent 4 to 5 mice per genotype. *p # 0.05; **p # 0.01; #p # 0.001. doi:10.1371/journal.pone.0092054.g010 Consequently, to study how these allelic polymorphisms increase disease risk, it is critical to use animal models in which gene expression approximates that of the human risk allele(s). This is especially true for BLK, as we have discovered a major phenotypic difference between Blk +/2 and Blk 2/2 mice, which could have an effect on disease progression and pathogenesis. Even though both cd-17 and DN-17 cells express Blk, Blk 2/2 mice exhibit a ,75% decrease in cd-17 cell numbers, but no change in DN-17 cell numbers, compared to B6 mice. Surprisingly, instead of observing a phenotype consistent with a gene dosage effect, Blk +/2 mice have ,50% more cd-17 cells and ,100% more DN-17 cells than B6 mice. Moreover, following infection with Listeria monocytogenes, which elicits both cd-17 and DN-17 cell responses [61], we found that the percentage of IL-17 + cd T cells, but not of IL-17 + DN ab T cells, was significantly greater in infected Blk +/2 mice than in infected B6 mice. By contrast, virtually no IL-17 + cd T cells were detected in the spleen of infected Blk 2/2 mice. Notably, there is a similar bias towards cd-17 cells in Blk +/2 .lpr mice, as there are 3fold more CCR6 + RORct + cd T cells than CCR6 + RORct + DN ab T cells in Blk +/2 .lpr mice yet equivalent numbers of these cells in B6.lpr mice. IL-17A has the attributes of a key effector cytokine in the Blk +/2 .lpr model, not only because its serum levels are significantly higher in Blk +/2 .lpr mice than in B6.lpr mice, but also because it is the only T cell cytokine whose serum levels correlate positively with IgG serum levels in Blk +/2 .lpr mice (r 2 = 0.551; p = 0.04). This correlation suggests that IL-17A mediates immunoglobulin class switch and antibody production in our experimental mouse model, which is in agreement with other reports demonstrating a role for IL-17A in IgG production [58,[69][70][71]. Accordingly, if our study of the role of BLK in the development of SLE had used Blk 2/2 .lpr mice, then there would be no functional cd-17 cells and, most likely, no enhanced IL-17A and IgG production at 3 months of age. Importantly, other phenotypes that are observed in 3-and 5-month-old Blk +/2 .lpr mice, as a direct or indirect consequence of elevated IL-17A production, may be delayed or may not even occur in Blk 2/2 .lpr mice.
There are three key steps in the development of systemic autoimmunity, commencing with loss of tolerance to self antigens, progressing to dysregulation of both the innate and adaptive immune systems, and ending with inflammation and tissue damage [1,72,73]. Previous studies, using different mouse models of lupus, have identified susceptibility genes and gene clusters that act at each of the three steps [1,72,73]. Since B cell tolerance is maintained in Blk +/2 3H9 Tg mice role, we conclude that BLK variants do not confer susceptibility to SLE by breaking B cell tolerance to self-antigens and, therefore, do not act at the first step of this model. Our findings, instead, reveal a role for Blk in the regulation of a proinflammatory cytokine network, involving cells of both the innate and adaptive immune systems. Such a regulatory function would indicate that BLK variants act at the second step in the three-step model. Notably, other genes and gene clusters that are also reported to act at the second step and participate in the dysregulation of innate and adaptive immune cell function include Fas, Lyn, Sle2, Sle3, Tlr7, and Ptpn6 [1,72,73].
In summary, we have demonstrated that solely reducing Blk expression levels in autoimmune-prone B6.lpr mice results in elevated proinflammatory cytokine production prior to the onset of proteinuria and nephrosis. Accordingly, we conclude that SNPs in the BLK locus increase risk to SLE, and to other autoimmune diseases, through the dysregulation of a proinflammatory cytokine network. Determining the hierarchy and interdependence of the cytokines in this network may lead to new tests for early detection as well as to new therapies that can be implemented in at risk individuals prior to the onset of autoimmune disease. different T cell subsets between 3-month-old B6 and Blk +/2 mice and between 3-month-old B6.lpr and Blk +/2 .lpr mice. *p#0.05; **p#0.01. (C) Histograms comparing CD69 expression on gated splenic CD4 + , CD8 + , DN ab, and cd T cell subsets from 3-monthold B6.lpr and Blk +/2 .lpr mice. CD69 expression levels on the corresponding splenic T cell subsets from age-matched B6 mice are also shown (shaded histogram). (D) Dot plots showing CD44 versus CD62L expression on gated CD4 + splenocytes from 3-month-old B6, Blk +/2 , B6.lpr and Blk +/2 .lpr mice. Numbers in plots represent percentages of naive (CD62L hi CD44 lo ), effector (CD62L hi CD44 hi ), and memory (CD62L l uCD44 hi ) CD4 + T cells. (DOCX)