G_αq-containing G proteins regulate B cell selection and survival and are required to prevent B cell–dependent autoimmunity

Our laboratory previously showed that G_αq regulates the in vitro chemotaxis of BM-derived neutrophils and DCs (Shi et al., 2007). Based on these data, we wanted to test whether G_αq was also required for B cell migration. However, mice lacking G_αq (Gnaq^−/− mice) are difficult to study as they are born runted and exhibit motor defects (Offermanns et al., 1997). Therefore, to analyze the role of G_αq specifically in immune cells, we generated BM chimeric mice by reconstituting lethally irradiated C57BL/6J recipient mice with either C57BL/6 BM (WT chimeras) or Gnaq^−/− BM (Gnaq^−/− chimeras). To determine whether G_αq expressed by hematopoietic cells was required for the positioning of B and T cells within lymphoid tissues, we examined the splenic architecture of Gnaq^−/− and WT chimeras. Using antibodies (Abs) to B220, CD21, and CD90.2 (Thy1.2) we identified B220⁺ B follicles containing CD21⁺ follicular DC networks and CD90.2⁺ T cell zones in the spleens of WT and Gnaq^−/− chimeras (Fig. 1 A). However, the numbers of CD21^hi B cells within the splenic marginal zone (MZ) appeared to be increased in the Gnaq^−/− chimeras (Fig. 1 A). To address whether this alteration was a result of changes in the migratory potential of the Gnaq^−/− B cells, we performed chemotaxis assays using spleen cells from WT and Gnaq^−/− chimeras. We found that the Gnaq^−/− B cells migrated normally to CXCL13, CXCL12, CCL19, and S1P (Fig. 1 B). Furthermore, upon transfer of purified WT and Gnaq^−/− B cells to intact hosts, we found that the Gnaq^−/− B cells migrated to both the B cell follicle and MZ of the spleen (Fig. 1 C). Therefore, the expansion of the MZ in the Gnaq^−/− chimeras did not appear to be a result of overt changes in the migratory capacity of the Gnaq^−/− B cells.

Previous experiments indicated that G_αi-containing G proteins regulate BCR signaling, B cell activation, and partitioning of B cells between the follicular and MZB cell subsets (Dalwadi et al., 2003). We therefore tested whether G_αq, like G_αi, regulates B cell development or maturation in the BM of chimeric mice (Fig. S1 A). We found that the total numbers of B220⁺ B cells in the BM of both groups of chimeras were equivalent (Fig. 1 D). Likewise, the absolute numbers of pro–B, pre–B (large and small), and immature B cells were indistinguishable between WT and Gnaq^−/− chimeras (Fig. 1 D).

Next, we examined peripheral B cell development using CD93, IgM, and CD23 to identify the T1, T2, and T3 subsets in the spleens of Gnaq^−/− and WT chimeras (Fig. S1 B). We found no differences between the numbers of T2 and T3 cells in the spleens of Gnaq^−/− and WT chimeras (Fig. 1 E). However, we reproducibly observed an increase in the percentage (Fig. S1 B) and number (Fig. 1 E) of T1 cells in the spleens of Gnaq^−/− chimeras. Interestingly, the percentages (Fig. S1 C) and absolute numbers (Fig. 1 E) of CD19⁺ splenic B cells, CD19⁺CD21^hiCD23^lo MZ B cells (MZBs), and CD19⁺CD21^loCD23^hi follicular B cells (FOBs) were higher in the Gnaq^−/− chimeras compared with WT chimeras. Likewise, the percentage of CD21^hi B cells expressing the MZB cell marker CD1d^hi was also significantly elevated in the spleens of Gnaq^−/− chimeras (Fig. S1 D). However, inconsistent with the phenotype of conventional MZB cells, ∼50% of the CD21^hiCD23^lo MZ-like B cells found in the spleens of the Gnaq^−/− chimeras also expressed high levels of IgD (Fig. S1 E).

To determine whether the expansion of MZ-like B cells in the Gnaq^−/− chimera spleens was the result of changes in the number of MZB precursors, we first examined the B220⁺CD93⁻CD23^hiCD21^hi MZB cell precursors that are thought to arise from the T2 or FOII population (Srivastava et al., 2005; Cariappa et al., 2007). We found no difference between the percentages (Fig. S1 F) or numbers (Fig. 1 F) of these cells in spleens of Gnaq^−/− and WT chimeras. We next examined the MZB precursors (B220⁺CD93⁺CD23⁻IgM^hiCD21^hi) found in the T1 population (Carey et al., 2008). Interestingly, both the frequency (Fig. S1 G) and total number (Fig. 1 F) of the T1-derived MZB precursors were significantly elevated in the spleens of Gnaq^−/− chimeras. Collectively, the data suggest that G_αq is not required for B cell development in the BM. However, G_αq does modulate peripheral B cell development and appears to control the numbers of T1 cells, T1-derived MZB cell precursors, and mature MZB and FOB cells.

Given the alterations in splenic B cell homeostasis in Gnaq^−/− chimeras, we hypothesized that G_αq may regulate B cell expansion or survival. To test this hypothesis, we cultured splenic WT and Gnaq^−/− B cells in media alone or with increasing amounts of anti-IgM F(ab′)₂ for 48 h, pulsed the cells with ³H-thymidine for 5 h, and measured thymidine incorporation. No major differences in DNA replication were observed between unstimulated and anti-IgM–stimulated Gnaq^−/− and WT B cells (Fig. 2 A). Likewise, we did not observe any differences in DNA replication between Gnaq^−/− and WT B cells stimulated with agonistic anti-CD40 Abs (Fig. 2 B). However, Gnaq^−/− B cells did proliferate more strongly (∼30% increase) in response to LPS than their WT counterparts (Fig. 2 C), suggesting that either a higher proportion of Gnaq^−/− B cells are responsive to LPS or that Gnaq^−/− B cells are hyperresponsive to TLR4 ligands.

Because the loss of G_αq only modestly affected the in vitro proliferative responses of splenic B cells, we next examined whether G_αq regulates B cell survival. We cultured purified WT or Gnaq^−/− B cells in media alone or with anti-IgM F(ab′)₂ and monitored viability. Surprisingly, a larger fraction of Gnaq^−/− B cells remained viable in both unstimulated and anti-IgM–stimulated cultures (Fig. 2 D), suggesting that G_αq plays a role in regulating the in vitro survival of mature B cells. To address whether G_αq also modulates immature B cell survival, we sort purified T1 B cells and T2/T3 B cells from the spleens of WT and Gnaq^−/− chimeras, cultured the cells in media alone or with anti-IgM F(ab′)₂ for 18 h, and measured B cell viability. As expected (Su and Rawlings, 2002), WT T1 B cells died rapidly after anti-Ig stimulation (Fig. 2 E), whereas WT T2/T3 B cells were largely resistant to anti-IgM–induced cell death (Fig. 2 F). Interestingly, significantly more of the Gnaq^−/− T1 (Fig. 2 E) and T2/T3 B cells (Fig. 2 F) survived the culture period, regardless of whether the B cells were stimulated with anti-IgM or cultured in media alone. Thus, Gnaq^−/− B cells survive in vitro in higher numbers than WT B cells at all stages of transitional and mature B cell development.

GTP-bound G_αq is reported to suppress activation of PI3K and Akt in nonhematopoietic cells (Bommakanti et al., 2000; Ballou et al., 2003). Given the enhanced in vitro survival of Gnaq^−/− B cells, we next tested whether the loss of G_αq leads to enhanced Akt activation in B cells. We purified splenic B cells from WT and Gnaq^−/− chimeras, stimulated the cells with anti-IgM F(ab′)₂ for 0–30 min, prepared total protein lysates, and performed Western blot analysis. We found that the basal levels of phospho-Akt (p-Ser473) were higher in Gnaq^−/− B cells than in WT B cells (Fig. 3 A). Furthermore, upon anti-IgM stimulation phospho-Akt levels increased more rapidly in Gnaq^−/− B cells (Fig. 3 A), and the peak amount of phospho-Akt was higher in the anti-IgM–stimulated Gnaq^−/− B cells than in the anti-Ig–stimulated WT B cells (Fig. 3 B). We also found that the levels of phosphorylated PLCγ2 (Y759) and Erk were significantly increased in the anti-IgM–stimulated Gnaq^−/− B cells relative to the WT B cells (Fig. 3, A and C). Consistent with these results, we noted that several protein species were hyperphosphorylated on either tyrosine (Fig. S2 A) or serine (Fig. S2 B) in the unstimulated Gnaq^−/− B cells relative to the WT B cells and were further phosphorylated as early as 1 min after anti-IgM stimulation. Thus, several signaling substrates, including Akt, are hyperphosphorylated in Gnaq^−/− B cells both before and after BCR ligation.

To address whether the increased Akt phosphorylation observed in the anti-IgM–stimulated splenic Gnaq^−/− B cells was a result of the presence of more MZB cells in the Gnaq^−/− spleens, we assessed phospho-Akt levels in CD43^neg naive B cells and enriched populations of FOB and MZB cells. As expected, phospho-Akt levels were constitutively elevated in naive CD43^neg Gnaq^−/− B cells and more rapidly reached peak levels after anti-IgM stimulation (Fig. 3 D). In contrast, the phospho-Akt levels in the enriched FOB cells were equivalent between the WT and Gnaq^−/− B cells both before and after anti-IgM stimulation (Fig. 3 D). However, phospho-Akt levels were constitutively elevated in the Gnaq^−/− MZB cells relative to the WT MZB cells (Fig. 3 D). Furthermore, phospho-Akt levels increased to peak levels more rapidly and were sustained for extended periods of time after BCR stimulation in the Gnaq^−/− MZB cells (Fig. 3 D). Together, these data suggest that Akt activation after BCR engagement is specifically enhanced in Gnaq^−/− MZB cells.

Collectively, our data suggested that Gnaq^−/− B cells may survive better than WT B cells in vivo. To test this, we transferred equivalent numbers of splenic B cells from WT or Gnaq^−/− chimeras (CD45.2⁺) into intact congenic CD45.1⁺ WT hosts (Fig. 4 A). At various time points after transfer, we enumerated donor CD45.2⁺ B cells in the spleen by FACS. As shown in Fig. 4 (B and C), the percentages and numbers of donor Gnaq^−/− B cells recovered at 1 wk after transfer were significantly higher compared with those found in recipients receiving WT B cells. At days 14 and 21 after transfer, very few of the transferred WT B cells could be found in the congenic hosts (Fig. 4 B). However, the transferred Gnaq^−/− B cells were easily detectable and present at 10-fold higher frequencies and numbers than the WT B cells (Fig. 4, B and C). Thus, the data suggested that Gnaq^−/− B cells survive longer in vivo than WT B cells.

Next, to address whether the Gnaq^−/− B cells are more fit to survive than WT B cells, we prepared mixed BM chimeras by transferring equivalent numbers of WT (CD45.1⁺) and Gnaq^−/− (CD45.2⁺) BM cells into lethally irradiated WT CD45.1⁺ recipient mice (Fig. 4 D) and determined the relative composition of the splenic B cells after reconstitution. We found that the ratio of Gnaq^−/− B cells to WT B cells was already significantly increased by 1 mo after reconstitution (Fig. 4 E). Furthermore, by 8 wk after reconstitution the ratio of Gnaq^−/− to WT B cells was 4:1, and within 6 mo the ratio was 19:1 in favor of the Gnaq^−/− B cells (Fig. 4 E). We found that the FOB and MZB compartments were both skewed in favor of the Gnaq^−/− B cells at 8 wk after reconstitution; however, the MZB cell compartment was more highly skewed than the FOB compartment toward the Gnaq^−/− B cells (Fig. 4 F). Likewise, all of the transitional B cells in the spleen were already highly skewed toward the Gnaq^−/− B cells by 8 wk after reconstitution (Fig. 4 F). Interestingly, the T1 MZB precursor population was almost entirely derived from the Gnaq^−/− B cells within 8 wk after reconstitution (ratio of 20:1; Fig. 4 F). The skewing of the peripheral B cell compartments toward Gnaq^−/− B cells was not a result of changes in the output of B cell progenitors from the BM, as the ratio of Gnaq^−/− to WT immature B cells in the BM (fraction E) remained very close to 1:1 at 8 wk after reconstitution (Fig. 4 F). However, the ratio of Gnaq^−/− to WT B220⁺CD93⁺IgM⁺CD23⁺ transitional B cells in the BM (Lindsley et al., 2007) increased slightly (∼2.0) and the recirculating mature B cells in the BM were highly skewed toward the Gnaq^−/− genotype (10:1 ratio; Fig. 4 F).

Finally, to evaluate whether the increase in Gnaq^−/− B cells relative to WT B cells in the spleen of the mixed chimeric mice was the result of increased proliferation of the Gnaq^−/− B cells, we injected the 50:50 BM chimeras with 5-ethynyl-2′-deoxyuridine (EdU) at 8 wk after reconstitution and then determined the proportion of WT and Gnaq^−/− B cells that proliferated in 18 h. As shown in Fig. 4 G, the percentage of replicating B cells was not significantly different between Gnaq^−/− and WT B chimeras. Together, these data argue that the Gnaq^−/− B cells are not overtly more proliferative than WT B cells but are able to efficiently outcompete WT B cells for access to or retention within peripheral survival niches.

Because Gnaq^−/− B cells quickly outcompete WT B cells in the spleen, we next tested whether the Gnaq^−/− B cells are less reliant on survival factors such as BAFF. We generated 50:50 mixed BM chimeras by transferring equal amounts of CD45.1⁺ WT BM and CD45.2⁺ Gnaq^−/− BM to irradiated CD45.1 WT hosts. We treated the recipients with a neutralizing anti-BAFF or isotype control Ab 8 wk after reconstitution and then monitored WT and Gnaq^−/− B cells in the peripheral blood and spleen over the next 3 wk (Fig. 5 A). We determined the percentage of B cells in peripheral blood that were either G_αq deficient or WT at each of the time points (Fig. S3 A) and then calculated the ratio of Gnaq^−/− B cells to WT B cells. The ratio of peripheral blood Gnaq^−/− B cells to WT B cells was ∼1.5 at 8 wk after reconstitution (Fig. 5 B), indicating that Gnaq^−/− B cells were already overrepresented in the blood at this time. The ratio of Gnaq^−/− B cells to WT B cells in peripheral blood modestly increased over the next 3 wk in the chimeras that were treated with the isotype control Ab (Fig. 5 B), which is consistent with the gradual takeover by Gnaq^−/− B cells. However, the ratio of Gnaq^−/− B cells to WT B cells increased even more rapidly in the anti-BAFF–treated chimeras (Fig. 5 B).

Next, we determined the percentage (Fig. S3 B) and number (Fig. 5 C) of mature WT (CD45.1⁺) and Gnaq^−/− (CD45.2⁺) B cells present in the spleens of the mixed BM chimeras on day 21 after treatment. As expected, approximately three times more mature Gnaq^−/− B cells were recovered relative to WT B cells in the control Ab–treated chimeras (Fig. 5 C). After BAFF blockade, we observed a significant reduction in the number of total mature B cells, regardless of genotype (Fig. 5 C). However, although the number of WT B cells decreased by 23-fold after anti-BAFF treatment, the number of Gnaq^−/− B cells dropped by only 13.5-fold (Fig. 5 C). Furthermore, the ratio of splenic Gnaq^−/− B cells to WT B cells increased from 3.1 in the control Ab–treated chimeras to 5.0 in the anti-BAFF–treated chimeras (Fig. 5 D). This change in the ratio of splenic Gnaq^−/− B cells to WT B cells was not the result of an obvious difference in BAFFR expression levels between WT and Gnaq^−/− B cells (Fig. S3, C and D). Thus, Gnaq^−/− B cells are significantly more resistant to BAFF depletion than their WT counterparts.

BAFF signaling in B cells is reported to stimulate PI3K and induce activation of the Akt signaling pathway (Patke et al., 2006). Given the increase in constitutively phosphorylated Akt observed in the Gnaq^−/− B cells and the increased resistance of these B cells to BAFF depletion, we evaluated whether Gnaq^−/− B cells are more responsive to BAFF signals than their WT counterparts. We isolated splenic CD43^neg naive B cells, FOB cells, and MZB cells from WT and Gnaq^−/− chimeras. We cultured the cells overnight in the absence of BAFF, purified the live cells, stimulated the BAFF-starved B cells with recombinant BAFF protein, and then monitored activation of the Akt signaling pathway by measuring phosphorylation of Akt and one of its downstream signaling targets, S6 ribosomal protein. As shown in Fig. 5 E, phospho-Akt was observed at higher levels in BAFF-starved Gnaq^−/− CD43^neg B cells than in WT CD43^neg B cells. After BAFF stimulation, phospho-Akt levels increased more rapidly in the Gnaq^−/− B cells. This finding was recapitulated when either FOB or MZB cells were analyzed (Fig. 5 E) but was most striking in the Gnaq^−/− FOB cell subset (Fig. 5 E). Interestingly, although G_αq deficiency amplified BAFF-dependent Akt activation in both FOB and MZB cell subsets, phosphorylation of S6 ribosomal protein was equivalent between BAFF-stimulated WT and Gnaq^−/− FOB cells (Fig. 5 E). However, phospho-S6 levels were constitutively higher in Gnaq^−/− MZB cells than in WT B cells and were induced to higher levels in the Gnaq^−/− MZB cells after BAFF stimulation (Fig. 5 E). Together, these data indicate that the Akt prosurvival pathway is constitutively more active in BAFF-deprived Gnaq^−/− B cells and that Gnaq^−/− B cells engage this pathway more efficiently than WT B cells after BAFF stimulation.

Because Gnaq^−/− B cells are more resistant to BAFF deprivation, we postulated that Gnaq^−/− B cells may survive peripheral tolerance checkpoints that normally eliminate B cells with self-reactivity. If this were true, then Gnaq^−/− B cells might be activated and responding to self-antigens. To test this hypothesis, we made 50:50 mixed BM chimeras by reconstituting CD45.1⁺ WT recipients with an equal mixture of CD45.1⁺ WT BM and CD45.2⁺ Gnaq^−/− BM. We isolated cells from spleen, BM, and lung at 3 mo after reconstitution and used FACS to phenotypically characterize the B cells. Gnaq^−/− B cells from the mixed BM chimeras expressed statistically higher levels of CD80, CD40, and MHC class II than the WT B cells isolated from the same animals (Fig. S4 A). Next, using peanut agglutinin (PNA) and FAS to identify B cells with a germinal center (GC) phenotype (PNA^hiFas⁺), we found only a very small proportion of WT B cells with this phenotype in any tissue (Fig. 6, A and B). However, significantly higher proportions of the Gnaq^−/− B cells from the same animal expressed PNA^hiFas⁺ in spleen, lung, and BM (Fig. 6, A and B), suggesting that Gnaq^−/− B cells make up most of the GCs present in these mice. In support of this conclusion, we identified well defined B cell follicles in the lungs of G_αq-deficient chimeras, and some of the B cells within these follicles expressed the GC marker GL7 (Fig. S4 B). As these structures are not found in the lungs of naive normal mice (Moyron-Quiroz et al., 2004), the data suggest that Gnaq^−/− B cells are being activated by endogenous self-ligands.

To address whether Gnaq^−/− B cells are self-reactive, we evaluated whether serum Abs from WT and Gnaq^−/− chimeras were reactive in a standard anti-nuclear Ab (ANA) assay. As expected, none of the serum samples from WT chimeras exhibited ANA reactivity at any time point after reconstitution (Fig. 6 C). In contrast, ∼20% of the serum samples from Gnaq^−/− chimeras were ANA reactive within 2 mo after reconstitution. By 5 mo after reconstitution, 100% of the serum samples from Gnaq^−/− chimeras were ANA⁺ (Fig. 6 C). Interestingly, we observed that some of the serum samples strongly stained the entire nucleus, whereas others exhibited a more punctate pattern of staining (Fig. S4 C), suggesting that multiple auto-Ags can be recognized by Gnaq^−/− B cells. Indeed, we also identified significantly higher levels of anti–double-stranded (ds) DNA reactivity in some Gnaq^−/− chimera serum samples (Fig. 6 D).

To determine whether the development of autoreactive B cells in Gnaq^−/− chimeras is the result of an intrinsic defect in the Gnaq^−/− B cells, we lethally irradiated WT Igh^a allotype congenic mice and reconstituted them with 50% WT Igh^a BM and 50% Igh^b Gnaq^−/− BM (Gnaq^−/−:WT chimera) or with 50% WT Igh^a BM + 50% WT Igh^b BM (WT:WT chimera; Fig. 6 E). At various time points after reconstitution, we collected serum and peripheral blood cells. As expected, peripheral blood B cells from the WT:WT chimeras were evenly distributed, with ∼50% of the B cells expressing an Igh^a BCR and the other half expressing Igh^b BCRs (Fig. 6 F). In contrast, in the Gnaq^−/−:WT chimeras, the Igh^b-expressing Gnaq^−/− B cells outcompeted the Igh^a-expressing WT B cells and eventually became the dominant B cell population present in the peripheral blood (Fig. 6 G). We next evaluated the serum samples for ANA reactivity using allotypic-specific Abs to detect the ANAs. As expected, neither Igh^a nor Igh^b Abs from the serum of WT:WT chimeras exhibited ANA reactivity (Fig. 6 H). However, analysis of serum from the Gnaq^−/−:WT chimeras, revealed that the Igh^b Abs (derived from the Gnaq^−/− B cells) efficiently stained the nucleus of HepG2 cells (Fig. 6 H). In contrast, the Igh^a Abs (derived from the WT B cells present in the same host) did not stain HepG2 nuclei (Fig. 6 H). Indeed, ANA reactivity of the WT Igh^a Abs from the Gnaq^−/−:WT chimeras did not rise above background and was equivalent to that observed for the Igh^a Abs from WT:WT chimeras (Fig. 6 I). This was not the result of a lack of Igh^a Ab in the Gnaq^−/−:WT serum samples, as nonautoreactive Igh^a Abs were present at low, but still detectable, levels in the Gnaq^−/−:WT chimera serum samples (Fig. S4 D). Together, the data indicate that the development of autoreactive B cells is dependent on the loss of G_αq expression in B cells. Furthermore, the loss of G_αq in other hematopoietic cells is not sufficient to drive normal B cells to become autoreactive.

Given the propensity of Gnaq^−/− B cells to secrete autoreactive Abs, we predicted that Gnaq^−/− chimeras would develop autoimmune disease. To test this, we determined whether immune complexes were present in the kidneys of WT and Gnaq^−/− chimeras. No Ab-containing immune complexes were detected in the WT kidney sections (Fig. S5, A–C). In contrast, IgG2a- and IgG2c-containing immune complexes were reproducibly observed in kidney sections from Gnaq^−/− chimeras (Fig. S5, D–F). However, despite immune complex deposition in the kidney, kidney function, as assessed by measuring protein concentration in the urine, was normal in the Gnaq^−/− chimeras (unpublished data). Histopathologic analysis of the kidneys from Gnaq^−/− chimeras showed mild to moderate mesangiolysis (thrombotic microangiopathy) with splitting of the basement membranes in the glomeruli and increased mesangial matrix within the glomeruli (Fig. S5, G–J). Although this pathology is atypical for the most common forms of immune complex–mediated disease, thrombotic microangiopathy can be a manifestation of disease in Lupus patients with anti-coagulant, anti-phospholipid, or anti-endothelial Abs (Tektonidou, 2009).

In addition to changes in the kidney, the chimeras exhibited other symptoms consistent with autoimmune disease. For example, we noted that the BM from the older Gnaq^−/− chimeras appeared to have reduced numbers of RBCs. To test whether the Gnaq^−/− chimeras develop anemia, we performed complete blood counts on arterial blood from WT and Gnaq^−/− chimeras. We found that the number of RBCs and the concentration of hemoglobin were significantly reduced in the Gnaq^−/− chimeras (Fig. 7, A and B). These symptoms are consistent with a diagnosis of autoimmune hemolytic anemia as a result of the presence of anti-RBC Abs, or they could also be a consequence of the autoimmune kidney thrombotic microangiopathy (Tektonidou, 2009).

We also monitored joint swelling in WT and Gnaq^−/− chimeras, as we observed that many of the older Gnaq^−/− chimeras were less mobile than their WT counterparts. Significant differences in the size of the joints of Gnaq^−/− chimeras was observed within 3.5 mo of reconstitution and the penetrance of disease in the Gnaq^−/− chimeras was 100% (unpublished data). By 5 mo after reconstitution, significant swelling was observed in both ankle and wrist joints of Gnaq^−/− chimeras compared with the WT chimeras (Fig. 7 C). Histological analysis of knees, ankles, and portions of feet from Gnaq^−/− chimeras revealed a variety of arthritic changes including chronic/active areas of inflammation within the joints, evidence of synovitis and bone resorption, exostotic bone development, and osteolytic activity (Fig. 7, D–K).

Collectively, our data suggest that Gnaq^−/− chimeras develop systemic autoimmune disease with multiorgan involvement. Consistent with this, we noted a significant difference between the survival of Gnaq^−/− and WT chimeras with up to 50% of the Gnaq^−/− chimeras dying within 6 mo of reconstitution (Fig. 7 L). To test whether the loss of G_αq in B cells is responsible for the increased mortality observed in the Gnaq^−/− chimeras, we depleted B cells in WT and Gnaq^−/− chimeras at 8 wk after reconstitution using anti–mouse CD20 (Hamel et al., 2008; Yu et al., 2008). We then monitored survival in the anti-CD20 and control Ab–treated chimeras over the next several months. As expected, mortality rates were very low in WT chimeras treated with a single dose of control or anti-CD20 Ab (Fig. 7 M). Gnaq^−/− chimeras treated with the control Ab had high rates of mortality, with 60% of the chimeras dying within 10 mo of reconstitution (Fig. 7 M). In striking contrast, only 10% of the anti-CD20–treated Gnaq^−/− chimeras died during the study (Fig. 7 M). Thus, G_αq expression by B cells is necessary for normal peripheral B cell development and tolerance induction and also plays a role in preventing early onset mortality in mice.

It is estimated that >800 seven-transmembrane GPCRs are encoded in the mammalian genome. B lymphocytes, like other hematopoietic cells, express GPCRs, some of which play important roles in regulating B lymphocyte adhesion and trafficking (Wettschureck et al., 2004). In this paper, we show that G_αq regulates B cell repertoire selection by controlling B cell survival during transitional cell differentiation. Aberrant regulation at this point allows inappropriate survival of autoreactive B cells, ultimately resulting in the development of autoimmune disease.

G_αq is expressed in B lineage cells throughout development (Wilkie et al., 1991). However, G_αq appears to have little to no effect on the survival or expansion of BM B cell progenitors. Instead, G_αq exerts its effects at the earliest T1 stage of transitional B cell development and extends its influence to all transitional and mature B cells subsets. Interestingly, the loss of G_αq particularly favors the development of MZB-like cells. At least three precursors for the mature MZB cells have been identified: the T2 MZB precursor (Srivastava et al., 2005), the T1-MZB precursor (Carey et al., 2008), and the FOII precursor (Cariappa et al., 2007). Interestingly, only the T1-MZB precursor population was expanded in the Gnaq^−/− chimeras and this population was the most highly skewed in favor of Gnaq^−/− B cells in the mixed BM chimeras. Therefore, we think it likely that the Gnaq^−/− MZB-like cells develop directly from the T1 MZB precursors.

The MZB compartment contains B cells that are positively selected from a subpopulation of proliferating T2 B cells (Meyer-Bahlburg et al., 2008) as well as potentially autoreactive B cells that escape deletion at the T1 to T2 checkpoint (Thien et al., 2004). T1 B cells with low affinity for self-antigens normally up-regulate BAFFR, colonize the B cell niches of the spleen, and enter the mature B cell pool (Thien et al., 2004). In contrast, B cells of intermediate affinity for self-antigen are excluded from the splenic niches and rapidly die (Cyster et al., 1994). However, when BAFF is readily available, the intermediate affinity autoreactive B cells are rescued and positively selected into the MZ compartment (Thien et al., 2004). These results indicate that B cells with a higher affinity for self-antigen compete less effectively than the lower affinity B cells for a limiting pool of BAFF. It is clear from our data that the Gnaq^−/− B cells can outcompete WT B cells under homeostatic conditions and that Gnaq^−/− B cells are less dependent on BAFF for survival than WT B cells. This would permit relaxed stringency of negative selection for the Gnaq^−/− B cells and allow for the entry of potentially autoreactive cells into the mature MZB compartment. However, it is important to note that the T1 compartment, which is initially BAFF unresponsive, is also substantially expanded in the Gnaq^−/− chimeras. Thus, either the loss of G_αq affects more than just BAFF responsiveness or a proportion of the T1 cells can become BAFF responsive as they transit this developmental window. Regardless, our combined data strongly imply that loss of G_αq facilitates the inappropriate survival of B cells in the periphery.

The data presented in this paper suggests that G_αq is an important regulator of B cell homeostasis and survival. Although the neuronal defects in Gnaq^−/− mice make it impractical for us to monitor the long-term development of clinical autoimmunity without generating BM chimeras, B cell homeostasis was identical in the Gnaq^−/− mice and the Gnaq^−/− chimeras. Therefore, our results are not an aberrant manifestation of irradiation and reconstitution but are a result of the loss of G_αq specifically in hematopoietic cells. In addition, our data strongly argue that the Gnaq^−/− B cells are intrinsically defective and are more resistant than the WT B cells to cell death–inducing signals such as BAFF withdrawal or strong BCR signals. Although we do not yet know all the mechanistic details surrounding G_αq-mediated suppression of B cell survival, we do know that unstimulated and BCR- or BAFF-activated Gnaq^−/− B cells have higher levels of active Akt as measured by phosphorylation of Ser473, a phosphorylation site which is under PI3K control. Prior work studying fibroblast cell lines (Bommakanti et al., 2000; Ballou et al., 2003) and cardiomyocytes (Howes et al., 2003) showed that G_αq normally represses Akt activation. Furthermore, overexpression of G_αq in cardiomyocytes leads to cardiac hypertrophy as well as cardiomyocyte apoptosis (Adams et al., 1998). Based on these data, it is speculated that low-level G_αq activity may lead to PI3K-dependent Akt activation and cell survival, whereas high or sustained G_αq activity results in PI3K/Akt repression and cell death (Hubbard and Hepler, 2006).

Although the active GTP-bound form of G_αq interacts with different signaling partners, including PLC-β and Btk (Hubbard and Hepler, 2006), published data indicate that G_αq mediates its pro-apoptotic effect in myocytes by binding to p110α and suppressing the production of PIP3 by PI3K (Bommakanti et al., 2000; Ballou et al., 2003). This G_αq-dependent inhibition of PI3K activity leads to reductions in Akt activation and suppression of the prosurvival mTOR pathway (Manning and Cantley, 2007). Interestingly, expression of activated p110α in B lymphocytes is sufficient to ensure survival of the mature B cell population, even when the BCR has been deleted (Srinivasan et al., 2009). Furthermore, deletion of PTEN, an inhibitor of PI3K, also promotes mature B cell survival (Srinivasan et al., 2009) and can rescue autoreactive B cells from anergy (Browne et al., 2009). Interestingly, the autoreactive prone MZB cell compartment is also expanded in mice expressing activated p110α or lacking PTEN (Anzelon et al., 2003; Srinivasan et al., 2009). Finally, B cells isolated from multiple models of autoimmunity are reported to express elevated levels of phospho-Akt (Wu and Mohan, 2009), and perturbations in the PI3K/Akt axis can lead to the development of autoimmunity (Wu and Mohan, 2009). Based on these data, we speculate that the loss of G_αq in B cells leads to increased active PI3K, resulting in the augmented phosphorylation of Akt, activation of the mTOR pathway, and an increase in the MZB compartment. Our experiments indicate that the increased Akt activity observed after BCR engagement is most pronounced in the MZ-like B cells found in Gnaq^−/− chimeras. This selective enhanced engagement of the prosurvival Akt pathway may explain why these MZB cells are present in larger numbers in the Gnaq^−/− chimeras. Interestingly, activation of Akt was very similar in the anti-IgM–stimulated FOB cells from Gnaq^−/− and WT chimeras. This result was initially unexpected given that both MZB and FOB cell subsets are expanded in the Gnaq^−/− mice. However, Akt is also constitutively more activated in BAFF-deprived Gnaq^−/− B cells and is increased more strongly after BAFF stimulation. This enhanced response to BAFF was observed in both the FOB and MZB cells from Gnaq^−/− chimeras. Furthermore, BAFF-induced phosphorylation of S6, a downstream target of the Akt–mTOR pathway, was evident in both the MZB and FOB subsets, although it was most affected in the Gnaq^−/− MZB cells. Collectively, these data suggest that G_αq deficiency confers both FOB and MZB cells with an enhanced ability to activate the PI3K–Akt pathway.

Although we have made progress in identifying the prosurvival signaling pathway that is normally suppressed by G_αq, we do not yet know which GPCR or ligand mediates this dampening signal in peripheral B cells. Our preliminary transcriptome analysis indicated that >100 different GPCRs, including chemokine receptors, are expressed by mature B cells. Our data suggest that B cell chemotaxis to CXCR4, CXCR5, Edg1, and CCR7 ligands is not abrogated by the loss of G_αq. Although these data do not exclude the possibility that chemokine receptors play a separate G_αq-dependent role in regulating B cell survival, it is equally likely that one of the many non-chemokine receptor GPCRs expressed by B cells regulates B cell survival. Unfortunately, surprisingly little is known regarding the role for these other GPCRs in B cells and more work will be needed to evaluate their potential roles in B cell development and selection.

The autoimmune disease that develops in Gnaq^−/− chimeras is similar, but not identical, to the disease that develops in BAFF Tg mice (Mackay et al., 1999). For example, both strains of mice have increased numbers of MZB cells, develop auto-Abs, and exhibit signs of lupus-like disease. However, kidney failure did not occur in Gnaq^−/− chimeras, despite evidence of immune complex deposition and mesangiolysis. Instead, the Gnaq^−/− chimeras develop arthritis and become anemic. Furthermore, the Gnaq^−/− chimeras develop inflammatory lesions and ectopic lymphoid tissues in multiple organs, including the lung. Given the speed with which autoimmune disease manifests in the Gnaq^−/− chimeras and the number of affected tissues, we believe that the animals suffer from a widespread autoimmune syndrome that affects multiple organ systems and likely involves autoreactive B cells of different specificities. Importantly, the increased mortality observed in Gnaq^−/− chimeras is dependent on B cells, as a single round of B cell depletion significantly delayed the onset of morbidity and mortality. Although we do not yet know whether G_αq deficiency in B cells alone is sufficient to induce autoimmune disease, we do know that the loss of G_αq specifically within B lineage cells is required for the development and survival of the autoreactive B cells. Together, our results indicate that the loss of G_αq in other hematopoietic lineage cells, including T cells, is not sufficient to drive the selection of normal B cells into the autoimmune repertoire.

Although Gnaq^−/− T cells are not sufficient to induce B cell autoreactivity, we cannot exclude the possibility that Gnaq^−/− T cells contribute to the disease. Indeed, the presence of GCs containing Gnaq^−/− B cells in the spleen, lung, LNs, and BM suggests that autoantigen-specific T cells are present and able to drive the GC reaction. Interestingly, a recent publication reported that G_αq regulates TCR signaling (Ngai et al., 2008). Preliminary unpublished experiments from our group indicate that Gnaq^−/− T cells outcompete normal T cells in the thymus of mixed BM chimeras. The competitive advantage of Gnaq^−/− T cells is observed at the double-negative stage of thymocyte development and persists throughout thymic and peripheral T cell development. However, the Gnaq^−/− T cells found in the LNs and spleen of the mixed chimeras are not constitutively activated as measured by expression levels of CD44, CD62L, or CD25. Thus, although G_αq plays a role in T cell development, it is still not clear whether the loss of G_αq expression on T cells will be necessary for the initiation of the autoimmune disorders observed in the Gnaq^−/− chimeras.

In conclusion, our work illustrates a critical intrinsic role for G_αq in the maintenance of peripheral B cell immunological tolerance and shows that deficiencies in G_αq expression contribute to the development of autoimmune disease. These data also provide the first demonstration that the survival and selection of B cells into the mature B cell repertoire is not only dependent on NF-κB activation in response to BCR and BAFFR engagement but is also controlled by G protein signals.

All experimental procedures involving mice were approved by the Trudeau Institute Institutional Animal Care and Use Committee (IACUC) and the University of Rochester Committee on Animal Resources. C57BL/6J (B6), Gnaq^−/− (n > 5 backcross to C57BL/6J), CD45.1⁺ congenic B6, and B6.Cg-Igha Thy1a Ptrpca Pep3b/Boy (Igh^a) mice were bred in the University of Rochester or Trudeau Institute animal facilities. BM chimeras were generated by irradiating recipient mice with a split dose of 950 Rads from a ¹³⁷Cs irradiator and then reconstituting the recipients with 1–2.5 × 10⁶ BM cells from B6 or Gnaq^−/− donors. In some experiments, CD45.1⁺ recipients were reconstituted with a mixture of 50% WT and 50% Gnaq^−/− BM (50:50 chimeras). In other experiments, BAFF was depleted by injecting 50:50 BM chimeras i.p. with 250 µg of anti–mouse BAFF/BLyS (clone 10F4; Human Genome Sciences) or control hamster IgG (BioExpress) in PBS 1 time at 8 wk after reconstitution. CD20⁺ B cells were depleted by injecting Gnaq^−/− and WT chimeras i.p. with 200 µg of mouse anti–mouse CD20 (clone 18B12 IgG2a; Hamel et al., 2008) or with an isotype control Ab (clone 2B8) one time at 9 wk after reconstitution. B cell–depleted and nondepleted chimeras were monitored for health and survival. Mice that appeared hunched with ruffled fur and/or had difficulty moving were euthanized in accordance with IACUC guidelines.

B220 APC-AF750 and CD23-APC were obtained from Invitrogen. SA–Pacific orange, SA-APC-AF750, and λ5-AF488 were purchased from Invitrogen. CD19 Pacific blue, CD93 biotin, anti-BAFFR-AF647, and CD45.2-APC-AF750 were produced by eBioscience, and IA/IE Pacific Blue and CD80-AF647 were obtained from BioLegend. All other Abs were purchased from BD. To analyze in vivo cell proliferation, chimeras were injected i.p. twice with the nucleic acid homologue EdU (Invitrogen) 18 h before harvest. Splenocytes were then stained with Abs to CD19 and CD45.2, fixed with 1% neutral buffered formalin, permeabilized with saponin-based buffer, and incubated with the manufacturer-provided Pacific blue–conjugated EdU detection agent. Flow cytometry data were collected using FACSCalibur, Canto II, or LSR II instruments (BD) from the University of Rochester Flow Cytometry Core and were analyzed using FlowJo (Tree Star, Inc.).

Spleen cells from 50:50 mixed BM chimeras were isolated and stained with Abs to B220 and CD45.2 to identify WT (CD45.2^neg) and Gnaq^−/− (CD45.2⁺) B cells. 3 × 10⁵ total spleen cells were added to the top well of a transwell with a 5-µm pore size polycarbonate filter (Costar), and the indicated amount of chemokine was added to the bottom well. Cells were recovered from the bottom well 2 h later, fixed, and then analyzed and counted on a flow cytometer to determine the number of B cells of each genotype that migrated to the bottom of the transwell. Data are reported as the chemotaxis index, which represents the fold change in the number of cells that migrated in response to the chemokine versus the number of cells that spontaneously migrated in the absence of chemokine. S1P was purchased from Avanti Polar Lipids and CXCL12, CXCL13, and CCL19 were purchased from R&D Systems.

B cells were purified from the spleens of CD45.2⁺ Gnaq^−/− and CD45.1⁺ WT chimeras by negative selection (>90% purity) using Abs to CD3, CD11b, and GR1 (MACS; Miltenyi Biotec). WT B cells were incubated at room temperature for 15 min in a solution of 80 µg/ml biotin in PBS. Gnaq^−/− B cells were incubated at room temperature in PBS only. Cells were washed three times in PBS, resuspended in FACs buffer, and counted. The cells were mixed at a 1:1 ratio and the volume was adjusted to 5 × 10⁷ cells/ml. CD45.1⁺ recipient B6 mice (n = 4 mice) were then injected i.v. with the cell mixture (10⁷ cells/recipient). Spleens from the recipient mice were harvested 18 h later and frozen sections were prepared. Sections were stained with PE-labeled anti-CD45.2 (to detect transferred Gnaq^−/− B cells) and a rat anti–mouse Ab to MOMA-1 (to identify the MZ). Sections were washed and then stained with SA-488 to detect transferred biotinylated WT B cells and an Alexa Fluor 647–labeled anti–rat Ab to detect the MOMA-1⁺ cells. Slides were viewed on a fluorescent microscope (Axio Observer.Z1; Carl Zeiss, Inc.) using a 10× objective (for 100× total magnification). Images were collected using AxioVision 4.5 software and converted into TIFFs.

Total B cells were purified by either positive (CD19⁺) or negative (CD3, CD4, CD8, CD43, CD11b, and CD11c negative) selection by MACS (Miltenyi Biotec). T1 and T2/T3 B cells were sort purified using a FACSAria sorter in the University of Rochester Flow Cytometry Core. Purity of the B cells was >90% in all experiments. To measure proliferation, purified B cells (5 × 10⁵/ml) were cultured in complete media for 2 d with 0–25 µg/ml of polyclonal goat anti–mouse anti-IgM F(ab′)₂ (SouthernBiotech), 0–40 µg/ml of monoclonal rat anti–mouse CD40 (clone 1C10), or 0–10 µg/ml LPS (Escherichia coli 055:B5) followed by the addition of 1 µCi 3H-thymidine for 5 h. Thymidine incorporation was measured using a scintillation counter. For in vitro survival experiments, 10⁶ B cells/ml were cultured in media alone or were stimulated with 10–15 µg/ml of anti-IgM F(ab′)₂. Cells from the cultures were collected at 18–24 h and stained with propidium iodide to determine the percentage of live cells.

Splenic B cells from WT and Gnaq^−/− chimeras (both CD45.2⁺) were purified by positive selection using MACS. 2.5 × 10⁷ donor B cells were transferred i.v. into individual CD45.1⁺ hosts. Spleens from host mice were isolated at 7, 14, and 21 d after transfer and the percentage and total number of transferred donor B cells were determined using flow cytometry.

Total naive B cells were purified (negative selection with biotinylated CD4, CD8, CD11c, and F4/80 Abs and SA microbeads or anti-CD43 beads; Miltenyi Biotec). CD43^neg B cells were further separated using biotinylated mouse anti-CD21/35 (eBioscience) and SA microbeads (Miltenyi Biotec). The cells not retained on the column were highly enriched for FOB (CD23^hiCD21^lo), whereas the B cells retained on the column were enriched for MZB as defined by CD21/CD23 and CD1d staining. The enrichment of FOB and MZB cells was equivalent between the Gnaq^−/− and WT groups, allowing for direct comparison between the groups.

After isolation, B cells were rested for 1 h in RPMI at 37°C before stimulation with 20 µg/ml anti-IgM F(ab′)₂ (8 × 10⁶ B cells per time point). Alternatively, for BAFF stimulations, B cells were cultured overnight in the absence of BAFF in RPMI supplemented with 10% FBS, 2 mM L-glutamine, 50 µM 2-ME, and antibiotics. Viable cells were then isolated by centrifugation using 1-step Polymorph gradients (Accurate Chemical & Scientific Corporation). Cells were washed with unsupplemented RPMI and rested for 1 h at 37°C before stimulation with 25 ng/ml BAFF (R&D Systems) in RPMI for the indicated times. B cells were lysed after stimulation in 1% NP-40 lysis buffer containing protease inhibitors. Cell lysates were recovered and frozen at −70°C until used. For Western blotting assays, 4 × 10⁶ cell equivalents per lane were resolved on 8% SDS-PAGE gels and transferred to PVDF membranes. Membranes were blocked and then probed with Abs specific for total phospho-tyrosine (clone 4G10; Millipore), total phospho-Serine PKC substrates, phospho-Akt, total Akt, and phospho-S6 ribosomal protein (Cell Signaling Technology), phospho-ERK, ERK, and PLCγ2 (Santa Cruz Biotechnology, Inc.) or phospho-Y759 of mouse PLCγ2 (Humphries et al., 2004). Membranes were washed and then incubated with anti–rabbit IgG IRDye800 (Rockland Immunochemicals) or anti–mouse IgG Alexa Fluor 680 (Invitrogen). Proteins were detected and quantified by Infrared Imaging System (LI-COR Biosciences). Gel images were captured using the Gel-Doc It Imaging System (UVP) and the GelCam310 charge-coupled device camera and Visionworks LS software. Gamma settings, background, and contrast were not adjusted.

Serum was diluted 1:100 in PBS/0.2% BSA and incubated on slides with acetone-fixed HepG2 cells. Slides were washed and stained with anti–mouse IgG FITC (SouthernBiotech). In some experiments, biotinylated allotype-specific Abs (recognizing Igh^a or Igh^b Abs; BD) were used to detect the ANA⁺ Abs in the serum samples. The allotype-specific detection Abs were then visualized using SA-AF488. Slides were mounted with Aqua PolyMount and analyzed on a fluorescent microscope (Axio Observer.Z1) using a Plan-Apochromat 20×/0.8 M27 objective. Images were acquired using a camera (1394 ORCA-ERA; Hamamatsu Photonics) with exposure times ranging from 150 to 400 µs. Data were collected using AxioVision Rel 4.7 software and then converted to TIFF files.

To detect circulating total Igh^a or Igh^b Ab in serum from 50:50 mixed BM chimeras, serum was diluted 1:2,700 in PBS and used to coat ELISA plates. Plates were then probed with biotinylated allotype-specific Abs that could distinguish Igh^a and Igh^b Abs from one another (BD). The allotype-specific detection Abs were visualized using SA-HRP (SouthernBiotech). To detect anti-dsDNA Abs, 96-well Immuno Plates (Thermo Fisher Scientific) were coated with 0.01% poly-l-lysine solution in PBS (Sigma-Aldrich) followed by 100 µg/ml deoxyribonucleic acid (Sigma-Aldrich). After blocking, serum samples, diluted 1:200 in 0.05% Tween 20/0.5% BSA/PBS, were added. Plates were washed and then developed with goat anti–mouse IgG HRP (SBA). The OD readings from three wells were averaged for each serum sample.

Whole blood from the descending aorta of freshly euthanized mice was collected using a tuberculin syringe. Blood was added to EDTA coated Vacutainer tubes (BD) and was counted on a HESKA Veterinary Hematology System, according to manufacturer’s instructions.

Wrists and ankles of Gnaq^−/− and WT chimeras were measured using digital calipers every 2 wk beginning at 4 wk after reconstitution. To assess pathological changes to the joints, chimeras were euthanized and dissected at 5 mo after reconstitution. The carcasses were fixed in 10% neutral buffered formalin (Sigma-Aldrich). Skeletal tissues were decalcified in a formic acid solution (Cal-Rite; Thermo Fisher Scientific), embedded in paraffin, and cut on a microtome (6-µm sections). Sections were stained with H&E and evaluated by a board-certified veterinary pathologist who was blinded to the experimental design. Images were collected using a microscope (Eclipse 80i; Nikon) with Plan Fluro 10×/0.30, Plan Fluro 20×/0.50, and Plan Fluro 60×/0.85 objectives. Images were recorded with a Digital Sight D5-5M camera using NIS Elements Acquisition Software (Nikon). Postimage processing was performed with NIS Elements for total picture contrast, balance, and color. All changes were made across the entire image and no spot changes were performed.

Spleens from WT or Gnaq^−/− chimeras were harvested in OCT buffer (Tissue-Tek), snap frozen in LN₂, sectioned (5 µm) on a cryostat, and then stained with anti-B220 Alexa Fluor 488, anti-CD21 Alexa Fluor 594, and anti-CD90.2 biotin followed by SA–Alexa Fluor 350. Slides were mounted in PolyMount fluorescent mounting media and viewed on an Axio Observer.Z1 fluorescent microscope at 100× (10× objective lens) using AxioVision 4.5 software. Images were converted to TIFFs and postprocessing imaging was performed in Canvas. Minor adjustments to brightness were performed equally across the whole image.

Fig. S1 shows FACS analysis of WT and G_αq-deficient B cell development in BM and spleen. Fig. S2 shows that BCR ligation of G_αq-deficient B cells leads to elevated tyrosine and serine phosphorylation of proteins. Fig. S3 shows that increased resistance of G_αq-deficient B cells to BAFF blockade is not a result of altered BAFFR expression. Fig. S4 shows that G_αq-deficient B cells are activated and display autoreactive specificities. Fig. S5 shows that G_αq-deficient chimeras develop immune complexes in the kidney and exhibit signs of kidney damage.

The authors would like to thank Lizz LaMere and Kyle Martin for management of animal breeding and genotyping and would like to acknowledge the University of Rochester Flow Facility Core for help with sorting and analysis of samples. The authors would also like to thank Randall Rossi and the University of Rochester Wilmot Cancer Center for use of the HESKA Veterinary Hematology System and Dr. Alex Rosenberg for analysis of GPCR mRNA expression levels in naive B cells.

This work was supported by the University of Rochester and National Institutes of Health grants AI068056 to F.E. Lund, F32AI080104 to R.S. Misra, HL69409 to T.D. Randall, HD037091 to D.J. Rawlings, and DK076126 to C.E. Alpers. M. Simon is supported by a Senior Scholar award from the Ellison Medical Foundation and G. Shi is supported by Natural Science Foundation of China grants 30830094 and 30972678.

The authors declare no financial conflicts of interest related to this work.