Btk Supports Autoreactive B Cell Development and Protects against Apoptosis but Is Expendable for Antigen Presentation Free

This article is featured in Top Reads, p.

Failed B cell tolerance supports autoimmune disease by producing autoantibodies and via Ag presentation to autoreactive T cells. Central tolerance eliminates the bulk of immature autoreactive B cells, and those that escape to the periphery can be suppressed via anergy (1–3). In humans, 2.5% of B cells are estimated to be autoreactive but anergic, failing to proliferate or produce Ab in response to stimulus (4). Strong signals can reverse anergy, causing proliferation and autoantibody production, as occurs in autoimmune arthritis and systemic lupus erythematosus (5, 6). This reversal is not necessary for Ag presentation, however, an important mechanism by which B cells promote autoimmunity, particularly in autoimmune (type 1) diabetes (7–11).

Bruton’s tyrosine kinase (Btk) is a member of the TEC family of nonreceptor tyrosine kinases whose function is best understood in BCR signaling (12–20). Btk-deficiency prevents autoimmune disease, including type 1 diabetes (T1D) in NOD mice and arthritis in the K/BxN model. It also selectively reduces anti-insulin and anti-GPI autoantibodies while preserving total IgG (18, 21–24). Furthermore, Btk deficiency reduces transgenic anergic anti-insulin B cells by 95% (25), contrasting nonautoreactive cells, which are impaired but reach mature subsets and remain able to respond to T-dependent immunizations (12, 26–28). The mechanism underlying this difference has not been understood. One possibility was that altered signaling properties of anergic autoreactive cells made them more dependent on Btk for survival after maturation. Alternatively, Btk-mediated signaling could be necessary for anergic cells to transition through the negative selective tolerance checkpoint that bridges the immature and mature states. This question is important for therapeutic interventions, as the first possibility would allow Btk inhibitors to eliminate existing autoreactive cells, whereas the second would require ongoing treatment to block maturation of newly emerging autoreactive B cells.

In this study, we report effects of Cre-mediated timed deletion of Btk in adult mice to reveal its role in anergic autoreactive B cells. Btk^flox/Cre-ER^T2 mice (29) were crossed with transgenic (125Tg) mice expressing anti-insulin BCRs (30). In Btk^flox/Cre-ER^T2 125Tg offspring, >95% of B cells bind insulin and Btk is efficiently deleted by tamoxifen-induction of Cre recombinase activity. Surprisingly, mature anti-insulin B cells persisted for months after Btk deletion. They could still mobilize activating and inhibitory signaling, but did not upregulate the antiapoptotic protein Bcl-xL. Furthermore, anti-insulin B cell proliferation, already blunted by anergy (30), was further abrogated by loss of Btk. Most importantly, these cells remain able to present Ag and activate cognate T cells.

Btk^flox mice were developed and bred to Cre-ER^T2 mice [B6.Cg-Tg(UBC-cre/ERT2)1Ejb/1J] as previously described (29) then crossed with mice expressing anti-insulin BCR transgenes using site-directed BCR H chain transgene paired with conventional BCR L chain transgene (125Tg) or paired with endogenous BCR L chains (VH125). Btk^null 125Tg and VH125 B6 controls were generated as previously described (12, 18, 25, 30). Mice were bred and maintained under specific pathogen–free conditions. Cre-activation was induced by i.p. injection of 3 mg of tamoxifen-free base (Sigma-Aldrich) in 200 µl of safflower oil on days −2, −1, and 0 (29).

Single-cell suspensions were obtained (18) and stained using fluorochrome or biotin-conjugated Abs against B220 (RA3-6B2), IgM (μ-chain; Life Technologies), IgMa (DS-1), IgD (11-26c.2a), CD19 (1D3), CD11b (M1/70), CD11c (HL3 or N418; eBioscience), F4/80 (BM8; eBioscience), Ly6G (IA8; Tonbo Biosciences), CD21 (7G6), CD23 (B3B4), CD86 (GL1), CD45 (30-F11; eBioscience), CD4 (RM4-5), CD8a (53-6.7), CD69 (H1.2F3; Tonbo Biosciences), and/or IA/IE (MHC class II, M5/114.15.2). B cell specificity for insulin was confirmed using biotin-conjugated human insulin or fluorochrome-conjugated pork insulin detected with streptavidin-conjugated fluorochromes. Viability dye Violet 510 or Red 710 (Tonbo Biosciences) excluded dead cells. For intracellular staining, cells were fixed using 1.6% paraformaldehyde (Electron Microscopy Sciences), permeabilized with a solution of 0.05% Triton-X-100 (SigmaUltra) or eBioscience intracellular fixation and permeabilization buffer set then stained with rabbit anti-mouse Btk (D3H5; Cell Signaling Technology) directly conjugated to fluorochrome, or followed by fluorochrome-conjugated anti-rabbit IgG(F’ab2) secondary. Cells were stained with anticleaved PARP (F21-852) or anti–Bcl-xL (54H6, Cell Signaling) in apoptosis assays or anti-CD22 (pY822) (12a/CD22), anti-ZAP70 (pY319)/Syk (pY352) (17A-P-ZAP70), or anti-PLCγ2 (pY759) (K86-689.37) for phosphoflow cytometry. Unless otherwise stated, Abs were from BD Biosciences. Samples were collected on an LSRII Flow Cytometer (BD Biosciences) and an Aurora Flow Cytometer (Cytek Biosciences), and data analyzed using FlowJo software (Tree Star).

Splenocytes were suspended at 10 × 10⁶ cells per ml in RPMI (Life Technologies), rested for 30 min, then stimulated for 0, 4, 10, or 30 min with 10 µg/ml goat anti-mouse IgM (µ-chain specific, Jackson Immunoresearch). Reaction was stopped with 1.6% paraformaldehyde before staining.

Splenocytes were stained with CellTrace Violet (Life Technologies), cultured at 1 × 10⁶ cells per ml for 3 d in complete RPMI (cRPMI) alone or stimulated with 10 μg/ml goat anti-mouse IgM (μ-chain specific; Jackson ImmunoResearch), then analyzed by flow cytometry.

Splenocytes were cultured at 10 × 10⁶ cells per ml for 0, 10, or 24 h in cRPMI alone or stimulated with 10 μg/ml goat anti-mouse IgM (μ-chain specific, Jackson ImmunoResearch).

Lymphocytes were purified as previously described (11). Anti-insulin B cells and OT-II T cells were stained with CellTrace Violet (Life Technologies). OT-II T cells were incubated in cRPMI alone or with anti-insulin B cells, with or without pork insulin conjugated to OVA_323–329 peptide (KISQAVHC; Sigma-Genosys). Samples were harvested and analyzed on day 3. Mitotic events were calculated as previously described (11, 18, 32, 33).

The p values were calculated using one-way or two-way ANOVAs, or multiple t tests with Welch correction, as appropriate, using GraphPad Prism 7.00 for Windows (GraphPad Software, La Jolla, CA).

Ninety-five percent of transgenic anti-insulin B cells are culled from the repertoire when Btk is genetically absent (Btk^null) (25). To determine whether Btk is required for autoreactive B cell development versus peripheral maintenance, tamoxifen-inducible Cre recombinase was used to delete the Btk gene from adult mice. Anti-insulin transgenic (125Tg) Btk^flox/Cre-ER^T2 mice were injected with tamoxifen to induce Btk deletion, as were 125Tg Btk^flox and 125Tg Btk^null controls. Anti-insulin B cells in bone marrow and spleen were assessed by flow cytometry 5 d and 10 wk later (Fig. 1, Supplemental Table I). Btk deletion was highly efficient with 92.1%, 98.3%, and 94.3% of pro/pre-B, immature B, and splenic B cells, respectively, Btk-negative by day 5 postinjections (Fig. 1A–C, 1F, 1G). As expected, this did not reduce numbers of bone marrow pre/pro-B cells in Btk^flox/Cre-ER^T2 mice even 10 wk postdeletion (1.09 × 10⁴ ± 3.81 × 10³) as compared with Btk^flox controls (9.53 × 10³ ± 8.43 × 10³) (p = 0.98) and in fact resulted in an increase in immature bone marrow B cells (Btk^flox = 1.20 × 10⁵ ± 7.37 × 10⁴, Btk^flox/Cre-ER^T2 = 2.28 × 10⁵ ± 1.17 × 10⁵, p = 0.02), indicating that Btk is not essential for autoreactive B lymphopoiesis in bone marrow. However, mature recirculating anti-insulin B cells, which originate in lymphoid follicles, are nearly absent in Btk^null controls but were not reduced in Btk^flox/Cre-ER^T2 (2.33 × 10⁵ ± 6.81 × 10⁵) compared with Btk^flox controls (3.07 × 10⁵ ± 1.41 × 10⁵) (p = 0.425).

Splenic anti-insulin B cell numbers were strikingly preserved 5 d after Btk deletion, contrasting their near absence in lifelong Btk-deficient Btk^null controls (Fig. 1H). Even at 10 wk, autoreactive B cells were only reduced by 52% in spleens of Btk^flox/Cre-ER^T2 mice (1.23 × 10⁷ ± 3.66 × 10⁶ versus 2.55 × 10⁷ ± 4.56 × 10⁶ in Btk^flox controls). Although this is a significant loss (p < 0.001), it did not recapitulate the near-total absence of mature B cells seen in the Btk^null model. Thus, Btk-mediated signaling is not required for persistence of autoreactive mature anti-insulin B cells.

Splenic B cells that have just emerged from the bone marrow go through an additional tolerance checkpoint at the early transitional (T1) stage, where they are subject to deletion upon Ag encounter. (Fig. 2B shows that T1 B cell numbers decrease significantly as early as 5 d after Btk deletion (1.81 × 10⁵ ± 8.17 × 10⁴) and remain low at 10 wk, (1.18 × 10⁵ ± 3.97 × 10⁴), matching low numbers in Btk^null mice (9.81 × 10⁴ ± 3.19 × 10⁴) (p = 0.52, p = 0.99) (Fig. 2A, 2B, Supplemental Table II), suggesting that Btk supports development of autoreactive B cells that pass through peripheral tolerance mechanisms at the T1 checkpoint.

At the late transitional (T2) stage, there is an increase in anti-insulin B cells 5 d after Btk deletion (6.48 × 10⁶ ± 1.98 × 10⁶ versus 3.77 × 10⁶ ± 1.02 × 10⁶ in Btk^flox controls, p < 0.001), with a commensurate decrease in follicular (FO) B cells (1.83 × 10⁶ ± 1.29 × 10⁶ versus 5.24 × 10⁶ ± 2.21 × 10⁶ in Btk^flox controls, p < 0.001) (Fig. 2B, Supplemental Table II), reproducing a pattern typically seen in nontransgenic repertoires (12). At 10 wk post–Btk loss, numbers of T2 B cells (2.42 × 10⁶ ± 7.55 × 10⁵) are not reduced compared with Btk^flox controls (p = 0.209). However, a block in development from T2 to FO B cells remains, resulting in a drastic reduction of FO B cells (2.01 × 10⁵ ± 1.17 × 10⁵ versus 5.24 × 10⁶ ± 2.21 × 10⁶ in Btk^flox controls, p < 0.001) (Fig. 2B, Supplemental Table II). The remaining few FO B cells either survived or developed from Btk-positive precursors, as the majority (57.1 ± 23.5%) retained Btk (Fig. 2C). The increase in Btk⁺ FO B cells was not due to globally increased Btk, as splenic macrophages and dendritic cells remain Btk-negative (94.8 ± 2.32 and 96.5 ± 2.04%, respectively) 10 wk after tamoxifen treatment (Supplemental Fig. 1). These findings support the idea that Btk promotes maturation of autoreactive FO B cells by supporting their transition through the early T1 checkpoint. However, FO B cells do not appear to need Btk for survival, as they are maintained for at least 5 d after excision of Btk. Over time, Btk-deleted anti-insulin FO B cells likely reach the end of their lifespans and are not replenished by newly emerging cells, resulting in nearly complete loss of this compartment.

Most autoreactive B cells that remain 10 wk after Btk deletion are marginal zone (MZ) B cells, and their numbers do not differ significantly between Btk^flox/Cre-ER^T2 (6.89 × 10⁶ ± 2.68 × 10⁶) and Btk^flox controls (1.13 × 10⁷ ± 3.93 ×10⁶) (p = 0.152) (Fig. 2B, Supplemental Table II). This contrasts near-complete lack of anti-insulin MZ B cells in lifelong Btk deficiency (4.23 × 10⁵ ± 1.87 × 10⁵) (p < 0.001) (25). About 17% of Btk^flox/Cre-ER^T2 MZ B cells are Btk⁺, and a majority (71.5 ± 19.6%) of the immediate precursor premarginal zone (pMZ) B cells are Btk⁺ (Fig. 2C, 2D, Supplemental Table II). This suggests that anti-insulin B cells that retain Btk may have a competitive advantage in the pMZ compartment, and then successfully mature, joining long-lived MZ cells that survived after Btk deletion. Thus, constitutive Btk-mediated signaling is not required for maintenance of autoreactive MZ B cells but supports transition into this subset.

Homeostatic expansion of otherwise impaired B cells can occur in the absence of a normal repertoire. To determine whether mature anti-insulin B cells lacking Btk still survive when competing with a polyclonal repertoire, we took advantage of the VH125 model. In this setting, the same transgenic anti-insulin H chain pairs with dozens of endogenous L chains, of which only a small percentage confer insulin-binding (Fig. 3A). In the bone marrow anti-insulin B cells make up 0.18% of the B cell population in this model (Supplemental Fig. 2A). Btk deletion was highly efficient in VH125 Btk^flox/Cre-ER^T2 anti-insulin bone marrow B cells, where 91.66% were depleted of Btk by day 5 postinjection (Supplemental Fig. 2B, 2C, Supplemental Table III). As with 125Tg Btk^flox/Cre-ER^T2 bone marrow, Btk proved unnecessary for lymphopoiesis, as total anti-insulin B cell numbers did not differ at 5 d or 10 wk after treatment (5 d: 1.47 × 10³ ± 1.78 × 10³ versus 1.98 × 10³ ± 2.13 × 10³ for VH125 Btk^flox controls, p = 0.912; 10 wk: 1.24 × 10³ ± 1.40 × 10³ versus 1.98 × 10³ ± 2.13 × 10³ for VH125 Btk^flox controls, p = 0.8171) (Supplemental Fig. 2D, Supplemental Table III).

In the spleens, 0.54% of B cells in VH125 mice bound insulin (Supplemental Table III). Most enter the FO compartment (87.4%) with a smaller proportion in MZs (5.9%), reflecting previously published data (30). This differs from 125Tg mice in which a higher proportion enter the MZ (38.% MZ, 56.96% FO) (Supplemental Fig. 2E). The reason for this difference is unknown but may be due to slight differences in affinity between endogenous and transgenic L chains. Btk deletion was highly efficient, with 91.11% of the anti-insulin B cells Btk-negative by day 5 postinjection (Fig. 3C). Like 125Tg, splenic anti-insulin B cell populations were maintained 5 d postinjection (Fig. 3C). Anti-insulin populations were reduced 46.7% by 10 wk postinjection, although this loss was not statistically significant (8.23 × 10⁴ ± 3.63 × 10⁴ versus 1.18 × 10⁵ ± 6.06 × 10⁴ in VH125 Btk^flox controls).

Anti-insulin B cells emerging in T1 5 d after Btk deletion did not demonstrate a reduction as they did in the fixed 125Tg repertoire (Figs. 2B, 3E). However, they did move into the T2 stage phenotype in a similar manner (2.09 × 10⁴ ± 1.36 × 10⁴ versus 5.33 × 10³ ± 4.52 × 10³ in VH125 Btk^flox controls, p = 0.0136). FO anti-insulin B cells did not differ at the 5-d time point (8.76 × 10⁴ ± 3.83 × 10⁴ versus 1.38 × 10⁵ ± 1.00 × 10⁵ in VH125 Btk^flox controls, p = 0.3542). Ten weeks after Btk deletion, T2 B cells remained significantly increased (2.47 × 10⁴ ± 1.36 × 10⁴ versus 5.33 × 10³ ± 4.52 × 10³ in VH125 Btk^flox controls, p = 0.0029), with a corresponding significant decrease in FO B cells (2.26 × 10³ ± 3.45 × 10³ versus 1.38 × 10⁵ ± 1.00 × 10⁵ in VH125 Btk^flox controls, p = 0.0141). Comparable with 125Tg, a substantial proportion (23.63%; (Fig. 3F, 3G) of FO B cells expressed Btk. This supports the notion that B cells containing Btk may have a competitive advantage in the FO compartment.

There were no noticeable changes in pMZ and MZ anti-insulin B cells 5 d post–Btk deletion (1.12 × 10³ ± 1.93 × 10³ versus 1.59 × 10³ ± 1.99 × 10³ in VH125 Btk^flox controls, p = 0.9229 and 4.54 × 10³ ± 6.93 × 10³ versus 1.67 × 10³ ± 1.39 × 10³ in VH125 Btk^flox controls, p = 0.3745, respectively), nor 10 wk post–Btk deletion (1.69 × 10³ ± 1.13 × 10³ versus 1.59 × 10³ ± 1.99 × 10³ in VH125 Btk^flox controls, p = 0.9991 and 4.67 × 10³ ± 3.62 × 10³ versus 1.67 × 10³ ± 1.39 × 10³ in VH125 Btk^flox controls, p = 0.3794, respectively) (Fig. 3E, Supplemental Table III). Overall, these findings indicate that T2, MZ and mature FO anti-insulin B cells do not require Btk to survive, even when competing with a polyclonal repertoire.

In nonanergic B cells, BCR engagement activates tyrosine kinases Lyn and Syk, which phosphorylate and activate Btk and adapter protein BLNK (34–36). This recruits PLC-γ2, which is activated by Syk and Btk. Anti-insulin B cells are anergic in response to insulin stimulus but can partially respond to BCR crosslinking by anti-IgM (11, 30, 37). To assess the role of Btk in BCR signaling of anti-insulin B cells, cells were stimulated with anti-IgM and analyzed by phosphoflow cytometry to measure phosphorylation of PLCγ2. To account for possible signaling differences between splenic B cell subsets, MZ and T2/FO B cells were analyzed separately (38). T2/FO could not be identified separately due to use of anti-IgM Ab to stimulate IgM-BCR. We also assessed activation of upstream kinase Syk and inhibitory mediator CD22 to assess overall integrity of BCR signaling in the absence of Btk. As expected, basal and BCR-induced Syk phosphorylation were not reduced in Btk-deleted (Btk^flox/Cre-ER^T2) and Btk-sufficient (Btk^flox) cells. In fact, T2/FO Btk–deleted B cells showed increased Syk phosphorylation after BCR stimulation (Btk⁺ = 2.48 ± 0.55, Btk^– = 3.36 ± 0.39, p < 0.001) (Fig. 4A, Supplemental Table IV), possibly due to increased surface levels of IgM that occur in the absence of Btk. Phosphorylation of CD22 was slightly enhanced at 10 min in Btk-negative T2/FO cells (Btk⁺ = 1.95 ± 0.43 Btk^– = 2.51 ± 0.35, p < 0.001) but otherwise did not differ (Fig. 4C, Supplemental Table IV). Thus, Syk activation and BCR inhibitory signaling by CD22 do not depend on Btk in mature anti-insulin B cells.

Syk phosphorylation occurs upstream of Btk, whereas PLCγ2 is a target of Btk tyrosine phosphorylation. Surprisingly, PLCγ2 was phosphorylated even in Btk-negative Btk^flox/Cre-ER^T2 T2/FO anti-insulin B cells (4 min = 4.41 ± 0.93, 10 min = 4.85 ± 0.76), at a higher level compared with their Btk-positive Btk^flox counterparts (4 min = 3.20 ± 0.88, 10 min = 2.84 ± 0.85) (p < 0.01, p < 0.001) (Fig. 4B, Supplemental Table IV). Btk has been reported to phosphorylate PLCγ2, so this was unexpected (39, 40). However, Syk can also phosphorylate PLCγ2 (40). A Syk inhibitor, R406, was used to test the hypothesis that Syk provides redundant function in PLCγ2 activation after deletion of Btk from anti-insulin B cells. Syk, CD22, and PLCγ2 all showed significantly less phosphorylation, at every stimulated time point, in the presence of R406 (Fig. 4A–C, Supplemental Table IV). In addition, differences in signaling between Btk-positive and Btk-negative T2/FO B cells were abrogated. Because Syk has many substrates, it may mediate PLCγ2, or CD22 phosphorylation indirectly. However, inhibition shows that Syk is at least partially responsible for the ability of Btk-negative anti-insulin B cells to signal.

Btk^null B cells with endogenous BCRs show reduced proliferation to anti-IgM (12), as do mature endogenous Btk^flox/Cre-ER^T2 B cells after Btk deletion (29). The anti-insulin B cells used in this report were previously shown to have reduced proliferative capacity compared with nonautoreactive cells (30, 31). As shown in (Fig. 5, Btk deletion further diminishes that capacity. Despite their ability to activate proximal signaling, Btk-negative anti-insulin Btk^flox/Cre-ER^T2 B cells are almost completely unable to proliferate to anti-IgM. After 3 d of stimulus, dye dilution analysis showed that 79.5 ± 3.53% of live Btk^flox anti-insulin B cells had undergone at least one cycle of cell division, whereas only 9.40 ± 3.98% of Btk^flox/Cre-ER^T2 anti-insulin B cells did so (p < 0.001; (Fig. 5A, 5B, Supplemental Table V). Surviving Btk-negative anti-insulin B cells were also significantly less able to upregulate CD86 in response to anti-IgM. Btk-sufficient anti-insulin B cells increased expression nearly 50-fold (49.5 ± 20.5), whereas the Btk-negative anti-insulin B cells only increased expression ∼10-fold (9.1 ± 5.3) (p < 0.001) (Fig. 5B, Supplemental Table V). Therefore, functional cellular responses to signaling are abnormal in anti-insulin B cells after deletion of Btk, despite intact proximal signaling, indicating that Btk contributes to additional pathways that support these autoreactive cells.

Btk mediates Akt activation, which regulates apoptotic pathways, including Bcl-xL, an antiapoptotic protein important in B cell proliferation and survival (41–45). We assessed anti-insulin B cell survival and Bcl-xL expression in response to anti-IgM, with and without Btk. Immediately ex vivo, Btk-negative Btk^flox/Cre-ER^T2 and Btk-positive Btk^flox anti-insulin B cells showed similarly low levels of cleaved PARP, a marker of apoptosis (Btk^flox = 0.12 ± 0.02, Btk^flox/Cre-ER^T2 = 0.10 ± 0.01, p = 0.20) (Fig. 5C, 5D, Supplemental Table V). However, after 24 h of anti-IgM stimulus, a significantly higher percentage of Btk-negative anti-insulin B cells were positive for cleaved PARP (64.4 ± 2.15) compared with those with intact Btk (46.1 ± 8.97) (p < 0.001). Btk-negative anti-insulin B cells also showed significantly less cell viability (Btk^flox = 43.4 ± 11.3, Btk^flox/Cre-ER^T2 = 23.4 ± 2.54, p = 0.002) (Fig. 5E, Supplemental Table V). Furthermore, intracellular flow cytometry revealed that anti-insulin B cells do upregulate Bcl-xL in response to anti-IgM stimulation (4.19 ± 0.15 at 24 h), whereas Btk-negative anti-insulin B cells do not (1.75 ± 0.27 at 24 h) (p < 0.001) (Fig. 5G). These data suggest that strong BCR crosslinking can help break tolerance in anergic autoreactive B cells by activating Btk-mediated antiapoptotic pathways.

Anergic insulin-specific B cells do not proliferate or produce Ab, but efficiently present Ag and activate cognate T cells (11). We previously showed that anti-insulin B cells can efficiently internalize Ag without Btk but the small numbers of anti-insulin B cells that remain in Btk^null animals used in that study were too few to allow assessment of Ag presentation (25). Therefore, we purified anti-insulin B cells from Btk^flox/Cre-ER^T2 spleens 5 d after tamoxifen treatment. The anti-insulin B cells were then incubated with purified OT-II CD4⁺ T cells. Insulin was conjugated to OVA peptide to create a cognate T-B Ag, allowing anti-insulin B cells to internalize it via the insulin component, then process it and present the OVA peptide component (11). OT-II T cells did not proliferate when incubated with OVA-insulin alone (data not shown) but significantly proliferated when Btk-negative Btk^flox/Cre-ER^T2 anti-insulin B cells were added (52.0 ± 15.2%, p < 0.001; (Fig. 6A, 6B, Supplemental Table VI). This was similar to percent proliferation induced by Btk-positive Btk^flox anti-insulin control B cells (61.0 ± 15.4%, p = 0.708). Both Btk-positive and Btk-negative B cells induced up to five cell divisions by activated T cells, although T cells incubated with Btk^flox/Cre-ER^T2 anti-insulin B cells underwent slightly fewer cell divisions overall. More T cells reached peak 3 and peak 5 when incubated with Btk^flox versus Btk^flox/Cre-ER^T2 anti-insulin B cells, indicating a higher number of mitotic events overall (T cells with Btk^flox = 3.93 × 10⁴ ± 7.82 × 10³, T cells with Btk^flox/Cre-ER^T2 = 2.25 × 10⁴ ± 7.55 × 10³, p < 0.001) (Fig. 6C, Supplemental Table VI). This difference may not have been due to reduced Ag presenting capacity of Btk-negative Btk^flox/Cre-ER^T2 anti-insulin B cells; however, because MHC class II was equally upregulated on Btk^flox (3.04 ± 1.01) and Btk^flox/Cre-ER^T2 (2.88 ± 1.26) B cells (p > 0.999) (Fig. 6E). Furthermore, Btk-positive cells in culture with OT-II T cells underwent proliferation (50.1 ± 7.69 proliferating), whereas Btk-negative Btk^flox/Cre-ER^T2 anti-insulin B cells failed to respond (14.6 ± 1.20 proliferating) (p < 0.001), resulting in fewer Btk-negative B cells remaining to help activate T cells as the days progressed (Fig. 6D, 6E, Supplemental Table VI). These data show for the first time, to our knowledge, that anergic autoreactive B cells specific for a small soluble Ag do not require Btk to present Ag and activate T cells.

BTK-targeting, using either genetic or pharmacologic methods, has been shown to eliminate autoreactive B cells and prevent autoimmune diseases (21, 25, 46, 47). However, both methods have limitations. BTK-inhibitors affect off-target kinases, which may contribute to their efficacy, whereas lifelong Btk deficiency eliminates autoreactive cells from early development, precluding study of Btk contributions to their survival and function at mature stages. Studies in this report use timed deletion to determine the checkpoints at which autoreactive B cells require Btk. The data demonstrate that Btk supports maturation through early peripheral tolerance checkpoints, but is not needed for maintenance of mature autoreactive cells (Figs. 1, 2). Importantly, these cells retain the ability to internalize autoantigen and present it to cognate T cells, an essential function for some organ-specific autoimmune diseases, exemplified by T1D (Fig. 6) (11, 31). However, Btk-depletion from anti-insulin B cells renders them vulnerable to apoptosis upon BCR engagement, as they fail to upregulate the survival factor Bcl-xL, downstream of Akt, despite the fact that proximal signaling via PLCy2 remains intact (Figs. 4, and (5). Thus, autoreactive clonal populations are unable to expand, even with T cell help (Figs. 5, 6). Overall, the data indicate that autoreactive B cells are most susceptible to Btk-targeting during development and upon BCR crosslinking.

In examining developmental stages at which Btk deletion affects autoreactive cells, the data show immediate reduction in cells emerging from the bone marrow in the T1 stage, matching levels found in mice deficient from birth (Fig. 2). This demonstrates that Btk-mediated signaling is required for autoreactive B cells to bypass this peripheral tolerance checkpoint. Normal T1 B cells undergo deletion in response to BCR crosslinking, an event thought to be useful for enforcing tolerance. However, anti-insulin B cells differ from normal B cells. Insulin is a small, soluble protein that binds BCRs with relatively low affinity. Ag-engagement does not necessarily generate the same kind of strong signal induced by BCR crosslinking used in other experimental settings, but rather may just enhance or support constitutive levels of signaling. Their BCRs are consistently insulin-occupied in vivo, which may support this anergic state, and at the same time may support a low level of Btk-mediated signaling necessary for them to bypass this checkpoint. Cells in the T1 compartment are in the process of moving from the bone marrow, where Ag-engagement promotes apoptosis, to more mature peripheral stages at which Ag-engagement initiates cellular activation. It may be that the T1 stage provides a narrow signaling window in which low level Btk-mediated signals are required for positive selection, whereas stronger signals generate negative selection. One caveat to this interpretation is that myeloid cells also express BTK, and our model does not rule out a role for these cells in supporting T1 B cell processes, a possibility that requires further study.

Once anti-insulin B cells reach mature stages, Btk is no longer required, as evidenced by the persistence of FO cells to at least 5 d, and MZ cells even after 10 wk (Figs. 1, 2). However, the loss of emerging Btk-deficient autoreactive cells through transitional stages results in attrition of mature FO cells over time, so that by 10 wk posttreatment, those cells are nearly absent. In contrast, the long-lived MZ compartment is retained at 10 wk. PreMZ precursors at that time are mostly Btk⁺ cells that escaped deletion, as are 17% of MZ cells. Because anti-insulin B cells that are Btk-deficient from birth populate neither the FO nor the MZ compartment, these findings suggest that Btk-mediated signaling supports development of both compartments, but is not required for their persistence in the repertoire beyond transitional stages.

Btk has a well-established role in proximal signaling, particularly in phosphorylation of PLCγ2 (15). Therefore, it was surprising that this pathway proved to be maintained in anti-insulin B cells after Btk was removed. In fact, PLCγ2 phosphorylation levels in T2/FO B cells were even higher in the absence of Btk, although this is most likely due to the fact that IgM surface levels are also higher in these cells, providing more contact points for signaling in response to crosslinking. Increased Syk phosphorylation is consistent with this concept, as it is downstream of the BCR but upstream of both Btk and PLCγ2. Furthermore, inhibition of Syk abrogated PLCγ2 phosphorylation, indicating that Syk provides redundancy for Btk in this setting, either directly or indirectly. The negative regulator CD22 was also shown to rely on Syk, but not Btk. Thus, proximal signaling pathways in anergic anti-insulin B cells rely on Syk, but not Btk.

Despite intact proximal signaling, anti-insulin B cells failed to proliferate in response to BCR crosslinking after Btk deletion (Fig. 5A, 5B). Even when Btk is present, anti-insulin B cells do not proliferate in response to their Ag. However strong BCR crosslinking using anti-IgM can partially break this anergy, inducing some cell division, although less than in nonautoreactive cells (30). Loss of Btk completely abrogates the ability of this strong BCR signal to break anergy. The fact that proximal signaling is still intact indicates that an additional mechanism is needed to support a functional response by anergic autoreactive B cells. These findings prompted the discovery that anergic autoreactive B cells retain the ability to protect themselves from apoptosis by upregulating Bcl-xL in response to BCR crosslinking (Fig. 5F, 5G). This process is Btk-dependent, as its loss abrogates Bcl-xL responses, resulting in increased PARP cleavage and cell death (Fig. 5C–E). Btk activates the Akt pathway, which in turn supports Bcl-xL expression. Thus, Btk has an important role in supporting autoreactive B cells that is independent of proximal signaling via PLCγ2, but is likely mediated by the Akt pathway. In terms of pathophysiology, this pathway also appears to support anergic B cell response to T cell help, as Btk-deficient anti-insulin B cells also fail to proliferate during Ag presentation assays using cognate T cells (Fig. 6D, 6E).

Despite this inability to respond to T cell help, Btk-deficient anti-insulin B cells remained competent to present Ag and activate their cognate partners. This assay uses a conjugated Ag: insulin protein paired with an OVA peptide recognized by OT-II T cells (11). This approach requires internalization of the insulin component via the BCR followed by cellular processing, and presentation of the OVA peptide component to the T cells, thus testing multiple aspects of Ag internalization, processing, and presentation. We previously used this method to show that anergic B cells could process and present a physiologic autoantigen, supporting the concept that this is their primary role in T1D, an organ-specific, T cell–mediated disease in which autoantibodies do not appear to contribute to islet destruction (10, 31, 48). We had also shown that Btk was unnecessary for Ag internalization by the few anti-insulin B cells that remained when Btk was deficient from birth, but there were too few cells for functional studies (25). This model overcame that limitation and shows that Btk is not necessary for Ag processing and presentation in this context. Our findings differ from two other studies showing that Btk-impaired B cells with endogenous BCRs do require Btk for Ag internalization and/or presentation (49, 50). This may be due to differences in functional capabilities of anergic versus nonanergic cells, or in differences between B cell responses to small soluble Ag such as insulin versus strong crosslinking stimulation used in those studies.

Overall, data in this report show that Btk is not required for anergic autoreactive B cells to survive in vivo once they have passed developmental checkpoints. Furthermore, these cells present Ag and activate cognate T cells. However, Btk does support autoreactive B cell development, antiapoptotic mechanisms and proliferative responses, such that drug targeting may still be effective in reducing their pathogenic contributions. These findings may help guide efforts to use BTK inhibitors to treat or prevent autoimmune diseases, suggesting that long-term dosing at regular intervals may be more effective at preventing development of autoreactive cells than intermittent or short-term treatment.

The goal of treatment for autoimmune disease is to target autoreactive cells while leaving normal B cells intact. Human B cells rely heavily on BTK for early maturation, but our work suggests that late targeting has less effect than genetic deficiency. Indeed, early work with BTK inhibitors suggests they do not recapitulate x-linked agammaglobulinemia, the disease that results from BTK deficiency in humans (51–53). In fact, a recent case study reported that the BTK-inhibitor ibrutinib eliminates autoantibodies and reduces the need for insulin in diabetic patients while leaving vaccine responses intact, similar to our findings in mouse models (18, 21, 54). Therefore, the approach of targeting BTK to treat autoimmune disease remains appealing, and may offer a way to act as a “rheostat,” to dial down autoreactive B cells without inducing B cell immunodeficiency.

Flow Cytometry experiments were performed in the Vanderbilt University Medical Center Flow Cytometry Shared Resource. The Vanderbilt University Medical Center Flow Cytometry Shared Resource is supported by the Vanderbilt Ingram Cancer Center (P30 CA68485) and the Vanderbilt Digestive Disease Research Center (DK058404). Btk^flox mice were generated at the transgenic core, Miller School of Medicine, University of Miami, Miami, FL.

The authors have no financial conflicts of interest.