Igβ ubiquitination activates PI3K signals required for endosomal sorting

Recognition of polyvalent antigens by the B cell receptor (BCR) initiates two simultaneous processes critical for B cell activation. One is the induction of signaling cascades that modulate transcriptional programs determining cell fate (Kurosaki et al., 2010; Clark et al., 2014). The other is the capture and delivery of antigenic complexes along the endocytic pathway to the MHC class II antigen-presenting compartment (MIIC; Qiu et al., 1994; Ferrari et al., 1997; Clark et al., 2011; Blum et al., 2013). Endocytic trafficking is also necessary for activating, by BCR-captured ligands, TLRs 7 and 9, which take residence in the MIIC after BCR ligation (Chaturvedi et al., 2008; O’Neill et al., 2009; Lee and Barton, 2014). Therefore, BCR endocytic trafficking links signals elicited at the cell surface (signal 1) to costimulatory processes that originate in late endosomes (signal 2; Bretscher and Cohn, 1970).

Signals are initiated through the BCR when Igα and Igβ immunoreceptor tyrosine–based activation motifs are phosphorylated to form a recruitment site for the spleen tyrosine kinase (Syk). Downstream phosphorylation of the B cell linker protein (BLNK or SLP-65; Kabak et al., 2002) forms a platform for assembly of downstream effectors, including Btk, PLCγ2, Nck, Vav, and Grb2 (Herzog et al., 2008; Kurosaki et al., 2010). A particularly important signaling effector is phosphatidylinositol 3-kinase (PI3K), which is required for B cell development (Fruman et al., 1999; Clayton et al., 2002; Ramadani et al., 2010; Clark et al., 2014), regulation of receptor editing (Tze et al., 2005), peripheral B cell maintenance (Srinivasan et al., 2009), and germinal center (GC) responses (Wang et al., 2002; Castello et al., 2013; Sander et al., 2015). There are multiple ways in which PI3K can be activated (Engel et al., 1995; Rickert et al., 1995; Wang et al., 2002; Aiba et al., 2008; Castello et al., 2013; Clark et al., 2014). However, how activation occurs in each functional context is incompletely understood.

Igβ is also inducibly ubiquitinated, and in vitro, this is necessary for sorting internalized BCRs into the MIIC (Zhang et al., 2007). Based on a wealth of in vitro data, ubiquitinated receptors are recognized by components of the endosomal complex required for transport (Raiborg and Stenmark, 2009), which are first recruited to endosomes by inositols phosphorylated at the 3 position, especially phosphatidylinositol (3,4,5)-trisphosphate (PIP₃; Schmidt and Teis, 2012). Surprisingly, the in vivo significance of antigen receptor ubiquitination and PIP₃ in receptor trafficking is largely unexplored.

Within the MIIC, many of the same mechanisms that process antigens into peptides are necessary for activating TLRs 7 and 9 (Lee and Barton, 2014). Linked recognition between BCR and TLR7 or TLR9 is required for anti-RNP and anti-DNA humoral autoimmunity, respectively (Leadbetter et al., 2002; Viglianti et al., 2003; Christensen et al., 2006). TLR7 is also required for humoral immunity to RNA viruses (Koyama et al., 2007), including influenza, whereas TLR9 mediates responses to DNA-containing viruses (Hou et al., 2011).

Targeting the MIIC is dependent on BCR-initiated signals that accelerate receptor internalization (Niiro et al., 2004; Gazumyan et al., 2006; Hou et al., 2006) and enable BCR trafficking to late endosomes (Chaturvedi et al., 2008; Clark et al., 2011). Disruption of proximal signaling pathways, such as what occurs in anergy, blocks both BCR and TLR endocytic transit (Chaturvedi et al., 2008; O’Neill et al., 2009). There is only a partial understanding of the signaling pathways mediating BCR endocytic transit (Chaturvedi et al., 2008; Clark et al., 2011). Furthermore, it is not known whether the signaling mechanisms regulating BCR trafficking are similar or different than those regulating other BCR-dependent development and maturation transitions.

Herein, we use mice expressing an Igβ cytosolic tail mutant (Igβ^KΔR) to demonstrate that Igβ ubiquitination enables BCRs to enter the MIIC by activating PI3K and generating PIP₃ on early endocytic vesicles containing internalized BCR complexes (Murphy et al., 2009; Chaturvedi et al., 2011). Other PI3K-dependent B cell functions were normal in Igβ^KΔR mice, indicating that Igβ ubiquitination mediates a highly specific mechanism of PI3K activation required for one BCR-dependent function: endocytic sorting.

Gene targeting was used to derive C57BL/6 homozygous mice in which the codons encoding the three Igβ cytosolic lysines were mutated to encode arginines (Igβ^KΔR; Fig. S1 A). These mice were then crossed to mice expressing FLP1 recombinase to delete the neomycin gene (Neo). Proper gene targeting and Neo deletion were confirmed by PCR of genomic DNA (Fig. S1 B). mRNA was isolated from WT and homozygous Igβ^KΔR (hereafter Igβ^KΔR) splenic B cells, amplified by PCR, and sequenced to confirm targeted mutations (Fig. S1, C and D). Immunoprecipitating the Igα/Igβ heterodimer from WT and Igβ^KΔR splenic B cells, followed by immunoblotting sequentially with ubiquitin- and Igβ-specific antibodies demonstrated that WT Igβ was ubiquitinated in the resting BCR and that this increased the resulting BCR clustering. In contrast, neither the resting nor clustered Igβ^KΔR BCR was detectably ubiquitinated (Fig. S1 E).

To explore the in vivo consequences of Igβ ubiquitination, bone marrow (BM) was harvested from WT and Igβ^KΔR mice, stained with antibodies specific for B220, CD43, and IgM, and then examined by flow cytometry (Fig. 1, A–C). Early B cell development appeared normal with no significant differences in the pro–B cell (B220⁺CD43⁺IgM⁻) and pre–B cell (B220⁺CD43⁻IgM⁻) compartments of WT and Igβ^KΔR mice. However, the numbers of immature (B220^intermedCD43⁻IgM⁺) and mature (B220^highCD43⁻IgM⁺) B cells in Igβ^KΔR mice were reduced approximately twofold. IgM surface densities on immature and mature B cells were also reduced, with only a few cells expressing high levels of IgM. Mean fluorescence intensity of surface IgM staining on WT and Igβ^KΔR immature B cells was 2,497 ± 255 and 1,305 ± 62, respectively (P = 0.0001). As expected, thymic T cell development was normal in Igβ^KΔR mice (Fig. 1 D).

In the periphery, transitional B cell populations (T1, CD93⁺IgM⁺CD23⁻; and T2, CD93⁺IgM⁺CD23⁺; Merrell et al., 2006) were significantly diminished. However, later developmental stages, including T3/AN1 (CD93⁺IgM^lowCD23⁺), splenic immature (B220⁺IgM⁺IgD⁻), and mature (B220⁺IgM⁺IgD⁺) B cell populations, were only mildly decreased (Fig. 1, E and F). There was a proportional decrease in the number of follicular (B220⁺CD21^loCD23^hi) and marginal zone B cells (B220⁺CD21^hiCD23^lo). IgM surface densities on splenic immature Igβ^KΔR B cells were only slightly diminished compared with WT (1.3-fold), whereas IgD surface densities were normal. Therefore, Igβ^KΔR mice do not have a substantial defect in BCR expression in the periphery.

In immature BM B cells, changes in receptor editing are usually associated with changes in the fraction of cells expressing Igλ (Gay et al., 1993; Tiegs et al., 1993). However, as demonstrated in Fig. 1 G, the frequencies of Igλ⁺ cells in Igβ^KΔR, WT immature, and mature B cell populations were similar. This suggests that receptor editing was normal in Igβ^KΔR mice.

To determine whether the observed phenotypes were cell intrinsic and/or influenced by lymphopenia, we performed competitive BM chimera reconstitutions. Sublethally irradiated Rag2^−/−II2rg^−/− mice were reconstituted with WT (CD45.1) BM mixed 50:50 with either Igβ^KΔR (CD45.2) or WT (CD45.2) BM. 8–9 wk after reconstitution, BM and spleens were harvested and analyzed by flow cytometry. As shown in Fig. 1 H, Igβ^KΔR and WT BM were equally efficient at reconstituting the pro– and pre–B cell populations. In contrast, Igβ^KΔR BM was about a fourfold less efficient than WT BM in reconstituting the immature and mature B cell pools. In the periphery, Igβ^KΔR BM was less fit to reconstitute the transitional (T1 and T2) and marginal zone B cell compartments. In contrast, both WT and Igβ^KΔR BM equally reconstituted An1/T3, follicular, immature, and mature B cell compartments (Fig. 1 I). These data indicate that Igβ^KΔR mice have a defect in late B lymphopoiesis that recovers to normal levels in the periphery.

Cell line studies suggested that Igβ ubiquitination was required for targeting the MIIC (Zhang et al., 2007). Therefore, WT and Igβ^KΔR B splenocytes were stimulated in culture through the BCR with FITC-conjugated anti-IgG and -IgM (heavy chain and light chain [H+L]) F(ab)₂ antibodies for up to 30 min. Cell aliquots were then fixed, stained with Lamp-1– and H2-M–specific antibodies, and visualized by confocal microscopy. As demonstrated (Fig. 2, A and B), Igβ^KΔR BCRs were excluded from H2-M⁺Lamp-1⁺ MIIC late endosomes.

Examination of an earlier time point after 2-min stimulation demonstrated that Igβ^KΔR BCRs did not cap normally (Fig. 2, C and D). Though ∼80% of WT cells formed a single cap after BCR clustering, this was observed in only 20% of Igβ^KΔR splenocytes. Rather, multiple patches of Igβ^KΔR BCRs were usually distributed across the cell surface.

In cell lines, Igβ ubiquitination is dispensable for BCR internalization (Zhang et al., 2007). Indeed, Igβ^KΔR BCRs were internalized more rapidly than WT BCRs after receptor cross-linking (Fig. 2 E). These data indicate that Igβ ubiquitination is required for both BCR capping and sorting into the MIIC. Furthermore, it appeared to modulate early BCR internalization.

In Igβ^KΔR mice, pre-BCR–mediated development was normal, whereas BCR-mediated selection into the immature B cell pool was impaired. We therefore examined whether endocytic trafficking of the pre-BCR and BCR was also different. Large and small pre–B cells, immature B cells, and mature B cells were isolated from WT BM by flow cytometry. The pre-BCR was labeled on ice with the biotinylated antibody SL156, whereas BCRs on immature and mature B cells were labeled with FITC-conjugated anti-IgG and -IgM (H+L) F(ab)₂ antibodies. Samples were then warmed to 37°C for 30 min, fixed, and stained (Fig. 2, F and G). As demonstrated by confocal microscopy, internalized pre-BCRs were completely excluded from Lamp-1⁺ late endosomes. In contrast, BCRs on immature and mature cells colocalized with late endosomes to a similar degree as endocytosed BCRs on splenic B cells.

We next examined early BCR endocytic sorting. Internalized Igβ^KΔR BCRs initially targeted EEA1⁺ early endosomes similarly to WT BCRs (Fig. 2 H). However, Igβ^KΔR BCRs left this compartment much more rapidly than WT BCRs. In B lymphocytes, terminal lysosomes do not contain Lamp-1 (Li et al., 2002). Indeed, Igβ^KΔR BCRs rapidly targeted a Lamp-1⁻Cathepsin L⁺ (Fig. 2, I and J) and acidic (Fig. 2 K) compartment. These data indicate that in the absence of Igβ ubiquitination, ligated BCR complexes are rapidly internalized and target terminal lysosomes.

Numerous in vitro studies associate BCR endocytic trafficking to the MIIC with peptide loading of MHC class II and eliciting T cell help (Qiu et al., 1994; Siemasko et al., 1999; Chen and Jensen, 2008; Clark et al., 2011). To examine the in vivo functions of MIIC targeting, WT and Igβ^KΔR mice were immunized with both T-dependent (NP conjugated to either chicken gamma globulin [CGG], Np-CGG; or NP conjugated to sheep RBCs [SRBCs], NP-SRBCs) and T-independent (NP-Ficoll) antigens intraperitoneally at day 0 and again at day 21. Serum was collected at various time points up to 35 d after initial immunization and assayed by ELISA for NP-specific IgM, as well as low- and high-affinity NP-specific IgG.

Remarkably, humoral immune responses to NP were relatively similar in WT and Igβ^KΔR mice (Fig. 3 A). There was a modest decrease in titers of IgM to NP in response to NP-Ficoll at days 7 and 14, NP-CGG at days 28 and 35, and NP-SRBCs at day 7. Low-affinity IgG immune responses were similar at all time points, and high-affinity IgG immunity was similar except for a modest decrease in the NP-SRBC responses at day 14.

To better understand early humoral immunity to T-dependent antigens, Igβ^KΔR mice were crossed to mice expressing a BCR specific for hen egg lysozyme (HEL; MD4). HEL-specific B cells were sorted from WT×MD4 and Igβ^KΔR×MD4 mice, labeled with CFSE, and transferred into CD45.1 (B6.SJL-Ptprc^aPepc^b/BoyJ) mice. Mice were simultaneously challenged with HEL conjugated to SRBCs. After 3 d, spleens were harvested, labeled with CD86, and assayed by flow cytometry. As seen in Fig. 3 B, dilution of CFSE in HEL-binding WT×MD4 and Igβ^KΔR×MD4 mice was similar. In addition, both populations up-regulated CD86 (Fig. 3 B). After 6 d, intracellular staining of HEL-binding splenic B cells revealed that both populations up-regulated IRF4 and BCL6 (Ochiai et al., 2013) to a similar degree (Fig. 3 C). In aggregate, these data indicate MIIC targeting is not required for primary T-dependent and T-independent humoral immunity.

Confocal microscopy of WT splenic B cells demonstrated that ligated BCRs were rapidly delivered to TLR9⁺Lamp-1⁺ late endosomes (Fig. 4, A and B). In contrast, in Igβ^KΔR splenic B cells, endocytosed BCRs colocalized with TLR9 but were excluded from Lamp-1⁺ late endosomes. These results are similar to those observed in anergic B cells (O’Neill et al., 2009) and suggest that endocytosed BCRs traverse through a TLR9⁺ compartment before entering the MIIC.

When a TLR9 ligand (ODN 1826) was targeted to BCRs on WT splenic cells, there was a robust induction of the transcription factor T-box expressed in T cells (T-bet), a target downstream of TLR9 (Fig. 4 C; O’Neill et al., 2009). In contrast, in Igβ^KΔR splenic B cells BCR-targeted ODN 1826 induced T-bet poorly. Both WT and Igβ^KΔR splenic B splenocytes strongly expressed T-bet in response to IFNγ and not to a control oligonucleotide (ODN 2138). These data suggest that Igβ ubiquitination is required for coupling BCR antigen recognition to endocytic TLR activation.

To test the importance of Igβ ubiquitination for TLR activation in vivo, WT, Tlr7^−/−, and Igβ^KΔR mice were immunized and then boosted 21 d later with inactive influenza virus particles (Koyama et al., 2007). Comparison of serum influenza-specific IgG2a responses revealed that Tlr7^−/− mice had low titers compared with WT mice (Fig. 4 D). Titers in Igβ^KΔR mice were intermediate and significantly lower than WT titers. Differences in the expression of T-bet in splenic B cells were more striking with both Tlr7^−/− and Igβ^KΔR splenocytes having low and comparable levels of T-bet expression compared with WT controls (Fig. 4 E).

We then challenged WT, Tlr7^−/−, and Igβ^KΔR mice with influenza A virus. 27 d after infection, peripheral blood was assayed for neutralizing titers of antihemagglutinin antibodies. As demonstrated, antihemagglutinin antibody titers in Igβ^KΔR and Tlr7^−/− mice were similar, with both being significantly less than that observed in WT mice (Fig. 4 F). These data suggest that targeting the BCR to the MIIC is more important for TLR-dependent than T cell–dependent primary humoral immunity.

BCR-dependent signals enable the coordinated delivery of both the BCR and TLR9 to the MIIC (Chaturvedi et al., 2008; O’Neill et al., 2009). Therefore, we next determined whether defects observed in Igβ^KΔR mice were associated with differences in BCR signaling. Stimulation of B splenocytes in vitro followed by immunoblotting of total cell lysates indicated that BCR-induced tyrosine phosphorylation of total cellular proteins was similar in WT and Igβ^KΔR cells (Fig. 5 A). Furthermore, stimulation of both WT and Igβ^KΔR cells led to rapid recruitment of Syk to surface BCR complexes (Fig. 5, B and C; Hou et al., 2006). These data indicate that proximal tyrosine kinase activation through Igβ^KΔR BCRs is intact.

We next examined downstream signaling pathways in Igβ^KΔR splenocytes. BCR-induced phosphorylation of the MAP kinases ERK and p38 was similar in WT and Igβ^KΔR splenocytes (Fig. 5 D). In contrast, inductive phosphorylation of the PI3K subunit p85α was abolished in Igβ^KΔR splenocytes, and AKT phosphorylation was severely attenuated. This was associated with diminished CD19 phosphorylation after BCR ligation in Igβ^KΔR splenocytes (Carter and Myers, 2008).

We next determined whether diminished PI3K activation contributed to defects observed in Igβ^KΔR mice. Therefore, Igβ^KΔR mice were crossed sequentially with mice expressing the Cre recombinase in the Cd19 locus and then with mice containing a floxed allele of Pten. Deletion of Pten in WT mice did not increase numbers of immature and mature BM cells in either WT or Igβ^KΔR mice (Fig. S2 A). Furthermore, deletion of Pten did not enhance BCR surface densities on Igβ^KΔR immature B cells (Fig. S2 B). These data suggest that the observed defect in late B cell development is not caused by defective PI3K activation.

To determine the role of PI3K activation in endocytic trafficking in vivo, WT, Igβ^KΔR, and Igβ^KΔRCD19^crePten^fl/fl B splenocytes were stimulated through the BCR for 30 min and then imaged by multicolor confocal microscopy as described in Materials and methods. As demonstrated in Fig. 5 (E and F), deletion of Pten reconstituted endocytic trafficking of Igβ^KΔR BCRs into Lamp-1⁺TLR9⁺ late endosomes. Conversely, preincubation of WT splenic B cells with the PI3K inhibitor LY294002 blocked entry of ligated BCRs into Lamp-1⁺TLR9⁺ late endosomes (Fig. 5, G and H). These data suggest that Igβ ubiquitination enables BCR and TLR endocytic trafficking by activating PI3K (O’Neill et al., 2009). Interestingly, deletion of Pten did not reconstitute T-bet induction in Igβ^KΔR B splenocytes (not depicted). Therefore, in addition to PI3K activation, Igβ ubiquitination likely mediates other processes required to couple BCR recognition to TLR activation.

Defective activation of PI3K–AKT was associated with diminished CD19 phosphorylation, which is a well-described signaling intermediate of BCR-dependent PI3K activation (Carter and Myers, 2008). Indeed, cross-linking of the BCR and CD19 induced robust AKT phosphorylation in Igβ^KΔR splenocytes (Fig. 5 I). In contrast, CD19 cross-linking did not substantially enhance AKT phosphorylation in WT cells. Furthermore, cross-linking of the BCR and CD19 on Igβ^KΔR splenocytes rescued BCR endocytic trafficking to Lamp-1⁺ late endosomes (Fig. 5, J and K). These data suggest that Igβ ubiquitination is required to couple the BCR to efficient CD19-mediated PI3K activation.

Phosphorylation of phosphatidylinositols by PI3K on endosomal membranes assembles effectors of endocytic trafficking required for vesicular sorting (Schmidt and Teis, 2012). Indeed, clustering of WT BCRs induced a rapid accumulation of PIP₃ with the BCR at the plasma membrane after 2 min (Fig. 6, A and B). In contrast, there was little accumulation of PIP₃ with Igβ^KΔR BCRs. At 15 min after stimulation, both the WT and Igβ^KΔR BCRs target early endosomes. However, accumulation of PIP₃ on BCR-targeted EEA1⁺ early endosomes was much more robust in WT cells (Fig. 6, C and D). In contrast, there was no difference in colocalization of WT and Igβ^KΔR BCRs with PI (4,5)-diphosphate (PIP₂). The difference in PIP₃ and BCR colocalization at early endosomes persisted up to 60 min (Fig. S3 A). The WT BCR colocalized with PIP₃ on Lamp-1⁺ late endosomes at 30 min, and this was greatly diminished in Igβ^KΔR B cells (Fig. 6, E and F). However, the degree of colocalization of WT BCR and PIP₃ on late endosomes was less than that observed in early endosomes and was not detectable in this assay by 60 min (Fig. 6 F and Fig. S3 B).

The aformentioned data suggested that PIP₃ accumulated primarily on those endocytic vesicles directly targeted by the BCR. Indeed, analysis of BCR⁺ and BCR⁻ EEA1⁺ early endosomes 15 min after stimulation revealed that PIP₃ was primarily on BCR⁺ vesicles (Fig. 6 G). To better examine the spatial relationships between the BCR and PIP₃, WT cells were stimulated for 15 min, stained as above and imaged using superresolution confocal microscopy, which provided 20-nm x–y axis resolution. As demonstrated (Fig. 6 H and Fig. S3 C), only those EEA1⁺ vesicles that contained internalized BCR complexes also had detectable levels of PIP₃. Furthermore, PIP₃ was demonstrable only on that portion of EEA1⁺ vesicles containing BCR complexes (Fig. 6 H, right). These data indicate a direct role for the BCR, and Igβ ubiquitination, in PIP₃ accumulation on early endosomes.

The canonical model of antigen receptor signaling is one in which immunoreceptor tyrosine–based activation motif phosphorylation recruits and activates proximal tyrosine kinases, which assemble signalosomes and drive downstream signaling cascades. We now demonstrate that receptor ubiquitination also contributes to signaling by efficiently coupling the BCR to CD19-dependent PI3K activation and the generation of PIP₃ on endosomal membranes critical for proper receptor sorting. These data both reveal a novel mechanism of receptor signaling and demonstrate how the BCR can deliver signals to endosomal compartments critical for MIIC targeting.

Endosomal signaling is now recognized to be common and necessary to couple receptor complexes to specific signaling pathways and functions (Su et al., 2006; Murphy et al., 2009; Chaturvedi et al., 2011). However, the mechanisms of endosomal signaling have remained largely unknown. Our data demonstrate endosomal signaling. Remarkably, Igβ ubiquitin–dependent PIP₃ accumulation was very local, being restricted to those membrane surfaces containing BCR complexes. This likely drives the assembly of endosomal complexes required for transport (Raiborg and Stenmark, 2009; Vanhaesebroeck et al., 2010) and the sorting of the BCR, but not other receptors, along the endocytic pathway. This spatially restricted signaling mechanism is in contrast to the BCR-initiated kinase cascades that propagate through the cytosol to activate nuclear transcriptional programs (Kurosaki et al., 2010).

PI3K activation by the resting BCR regulates receptor editing (Tze et al., 2005; Verkoczy et al., 2007) and maintains peripheral B cell populations (Srinivasan et al., 2009). Both of these functions appeared normal in Igβ^KΔR mice. Furthermore, primary T-dependent immune responses, which occur in GCs and require PI3K (Clayton et al., 2002; Jou et al., 2002; Omori and Rickert, 2007; Ramadani et al., 2010), were intact. For B cell development and peripheral responses, several mechanisms of PI3K activation have been described, many with very specific functions (Engel et al., 1995; Rickert et al., 1995; Wang et al., 2002; Aiba et al., 2008; Castello et al., 2013; Clark et al., 2014). Igβ ubiquitination appears to be yet another example in which the mechanism and context of PI3K activation dictate cellular responses.

Likewise, the effects on CD19-dependent functions were very restricted. CD19 is required for initiating GC responses, whereas CD19-dependent PI3K activation is required for propagating the GC reaction (Carter and Myers, 2008). Both were normal in Igβ^KΔR mice. This could be because BCR-dependent CD19 phosphorylation was only partially attenuated in Igβ^KΔR splenocytes. Alternatively, Igβ ubiquitin–independent mechanisms of CD19 activation might dominate in GC responses (Carter and Myers, 2008).

Efficient MIIC targeting was dispensable for T-dependent humoral immunity. Although this was surprising, it may reflect the underlying biology of MHC class II trafficking and peptide loading (Blum et al., 2013; Roche and Furuta, 2015). From the trans-Golgi network, MHC class II can either target the endocytic pathway directly or indirectly via the plasma membrane. This ensures that most endocytic compartments contain MHC class II (Blum et al., 2013). In vitro studies have revealed that antigens can be productively processed and loaded onto MHC class II in multiple early and late endosomal compartments (Harding et al., 1990, 1992; Castellino and Germain, 1995; Griffin et al., 1997; Pathak and Blum, 2000). In fact, many T cell epitopes are destroyed in late endosomal compartments by excessive proteolysis (Manoury et al., 2002; Delamarre et al., 2005). Our findings indicate that alternative sites of antigen processing, distinct from the MIIC, are functional and important.

Normal GC responses and intact BCR proximal signaling in Igβ^KΔR splenocytes indicate that most BCR-dependent functions remain intact in Igβ^KΔR mice. Furthermore, PI3K complementation experiments indicate that many of the defects observed in Igβ^KΔR mice can be ascribed to a very specific signaling mechanism. Therefore, it is highly unlikely that the conservative mutations introduced in the Igβ cytoplasmic tail had global effects on BCR structure or function.

In contrast to T-dependent immune responses, TLR7- and 9-dependent responses required BCR ubiquitination and MIIC targeting. This might reflect the importance of endosomal localization for TLR function as proteolytic processing of TLR7 and 9 is necessary for receptor activation (Ewald et al., 2008, 2011; Park et al., 2008; Sepulveda et al., 2009). This dependence on proteolysis links BCR antigen capture to endosomal TLR activation, and ensures discrimination between self and nonself (O’Neill et al., 2009; Mouchess et al., 2011).

Both in vitro and in vivo, increasing signaling through the PI3K–AKT pathway restored normal BCR endocytic trafficking in Igβ^KΔR splenocytes. These complementation experiments clearly demonstrate a specific signaling pathway critical for BCR endocytic trafficking. However, deletion of Pten in vivo did not rescue TLR9 activation by BCR-targeted ligands nor late B cell development. These observations indicate that Igβ ubiquitination does more than activate PI3K. Whether these functions reflect additional signaling mechanisms or recruitment of other effector complexes to ubiquitinated Igβ is unclear.

In Igβ^KΔR mice, there was a defect in the immature B cell compartment that persisted into the transitional stage and then largely resolved by the follicular stage. Earlier stages of B cell development were normal. It is at the immature B cell stage that WT BCRs first become competent to enter late endosomes. In contrast, internalized pre-BCR complexes were excluded from the MIIC. Therefore, development was impaired at the stage at which the BCR first becomes competent to enter the MIIC. In pre–B cells, differentiation is directed by the pre-BCR with the single purpose of selecting cells that have productively rearranged the immunoglobulin heavy chain for subsequent light chain recombination (Clark et al., 2014). In contrast, in immature B cells, the TLR-signaling molecules IRAK-4 and MyD88 (Isnardi et al., 2008) shape naive repertoire in humans. These observations suggest that immature B cells might resemble peripheral B cells that depend on the integration of BCR and endosome-restricted signals to determine cell fate.

Much of our knowledge of receptor ubiquitination is derived from in vitro systems (Zhang et al., 2007; Raiborg and Stenmark, 2009). Our in vivo experiments demonstrate that such experiments do not reliably predict in vivo function. In vitro experiments are, by necessity, reductionist and usually rely on soluble antigens available for extended periods of time (Allen et al., 2007). Therefore, it is not surprising that such in vitro experiments do not recapitulate the complexities of in vivo immune responses. Therefore, additional in vivo studies of how ubiquitination contributes to the function of other receptors are needed.

All mice were on a C57BL/6 background and were used at 6–12 wk of age. Mice were housed in the Gordon Center for Integrated Sciences barrier animal facility, and experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of the University of Chicago. To derive Igβ^KΔR mice, a targeting vector (Fig. S1 A) was constructed in a C57BL/6 BAC clone in pSP72 (Promega). Mutations were introduced by overlap extension PCR at amino acid codons 157, 185, and 218. A lox-P and FRT flanked neomycin resistance gene was inserted 682 bp downstream of the most 3′ mutation in exon 6 by Red/ET recombineering. Targeting vector was confirmed by restriction digestion and sequencing at each modification step. The construct was then electroporated into C57BL/6 embryonic stem cells, and successfully recombined embryonic stem cells were injected into C57BL/6 blastocysts. After transfer into foster mothers, identified F1 pups were bred to homozygosity. Primers used for screening were Igβ^KΔR KI-F, 5′-CATTCCCTGCTTCCTGGATA-3′; Igβ^KΔR-R, 5′-ATGAACAGGGACACCCTCAA-3′; and Neo-R, 5′-CTTCCATCCGAGTACGTGCT-3′, yielding a 1.2-kb band for Igβ^KΔR and 640-bp band for WT. Sequencing primers were S1, 5′-TTAGGATTCAGCACGTTGGAC-3′; and S2, 5′-CATTCCTGGCCTGGATGCTCTCCTAC-3′. Homozygous Igβ^KΔR mice were then crossed to the Flpe deleter strain B6.129S4-Gt(ROSA)26Sor^tm1(FLP1)Dym/RainJ. WT C57BL/6, CD19Cre, Pten^flx/flx, B6.129S4-Gt(ROSA)26Sor^tm1(FLP1)Dym/RainJ, MD4, Tlr7^−/−, and B6.SJL-Ptprc^aPepc^b/BoyJ (CD45.1) mice were purchased from the Jackson Laboratory. B10;B6-Rag2^tm1Fwa II2rg^tm1Wjl were purchased from Taconic.

Single cell suspensions (10⁶ cells/100 µl) were stained on ice for 1 h with antibodies specific for IgM (Il/41), B220 (RA3-6B2), CD25 (PC61.5), IgD (11-26c), CD93 (AA4.1), CD45.1 (A20), CD45.2 (104), CD23 (B3B4), CD4 (GK1.5), CD8α (53-6.7; all from eBioscience), CD3ε (145-2C11; BD Biosciences), CD43 (S7; BD Biosciences), or CD21 (7G6; BD Biosciences) directly conjugated to FITC, PE, PE-Cy7, Percp-cy5.5, APC, APC-Cy7, or APC-780. Data were collected and analyzed using FlowJo.

Age-matched 6–12-wk-old Igβ^KΔR and C57BL/6 male mice were injected intraperitoneally with either 50 µg NP49-aminoethylcarboxymethyl-Ficoll (NP-Ficoll), 50 µg alum-precipitated NP29-CGG (Biosearch Technologies) in 300 µl PBS or 2 × 10⁸ SRBC conjugated to NP-OSu. ELISA plates (Immulon 2 HB Flat Bottom MicroTiter; Corning) were coated with 5 µg/ml of NP52-BSA or NP4-BSA (Biosearch Technologies), blocked and incubated with serially diluted serum, washed, and then incubated with horseradish peroxidase–conjugated goat anti–mouse IgM or anti–mouse IgG secondary antibodies (Jackson ImmunoResearch Laboratories). Plates were developed using a peroxidase substrate kit (Bio-Rad Laboratories), and absorbance was measured at 405 nm. For each panel, absorbances at the same dilution were graphed.

For influenza immunizations, age matched 6–12-wk-old Igβ^KΔR, C57BL/6, and Tlr7^−/− mice were injected intramuscularly with 5 µg of inactive A/Switzerland/9715293/2013 virus particles in 50 µl PBS. A boost using the same formulation and route was given 21 d later. Mice were euthanized 21 d after boosting, and their serum and spleens were collected. ELISAs were done using Immulon 2 HB Flat Bottom MicroTiter plates coated with 2 µg/ml of purified hemagglutinin from the A/Switzerland/9715293/2013 virus and horseradish peroxidase anti–mouse IgG2a secondary antibody (Invitrogen). Plates were developed using Super AquaBlue ELISA Substrate (Invitrogen), and absorbance was measured at 405 nm. B cells were purified from the spleens and then RNA was isolated, and quantitative real-time PCR for T-bet was performed as previously described (O’Neill et al., 2009).

Mice were infected by intranasal instillation with 50 PFU of influenza strain PR8. On day 27, postinfection sera were collected and analyzed for neutralizing antibodies against hemagglutinin (Hirst, 1942) as described previously (Geeraedts et al., 2012).

B cells were purified from the spleen by negative selection with biotinylated anti-CD11b (M1/70), anti-CD11c (HL3), anti-NK1.1 (PK136), anti–Ter-119, anti-CD3 (452C), anti-CD4 (RM4-5), anti-CD8α (53-6.7), anti–Ly-6G, and Ly-6C (RB6-8C5; all from BD Bioscience), followed by streptavidin magnetic beads (MACS; Miltenyi Biotec). Purified splenic B cells (1 × 10⁵ per sample) were stained with 10 µg/ml FITC-conjugated goat anti–mouse IgG and IgM (H+L) F(ab)₂ (Jackson ImmunoResearch Laboratories) on ice for 15 min. Cells were incubated at 37°C for the indicated times, and internalization assays were performed as described previously (Hou et al., 2006).

Splenic B cells were purified by negative selection as described above. Cells were stimulated with 20 µg/ml F(ab)₂ goat anti–mouse IgG and IgM (H+L; Jackson ImmunoResearch Laboratories) at 37°C for the indicated times. For Western blotting, cell aliquots were lysed in modified radioimmunoprecipitation assay buffer containing protease inhibitors and PMSF as described previously (Kabak et al., 2002). For ubiquitin immunoblots, cells were lysed as above with the addition of 20 mM N-ethylmaleimide (Sigma-Aldrich), 10 mM 1,10-phenatholine monohydrate (Sigma-Aldrich), and 50 µM PR-619 (LifeSensors). Cellular lysates were precleared with protein G–Sepharose (Pierce), incubated with primary antibodies specific for Igβ (Hm79b; BD Biosciences), and captured with protein G–Sepharose (Pierce). Lysates or immunoprecipitates were resolved on a 4–15% Mini-Protean TGX gel (Bio-Rad) and transferred onto polyvinylidene fluoride membrane (Immun-Blot; Bio-Rad). Antibodies with the indicated specificities were used: ERK1/2 (137F5; Cell Signaling), p38 (9212; Cell Signaling), CD19 (3574; Cell Signaling), actin (C4; Millipore), phosphotyrosine (4G10; Millipore), CD79b (Luisiri et al., 1996), antiubiquitin (P4D1; Santa Cruz), and phospho-T202/Y204 ERK1/2 (197G2), pT180/T182 p38 (9244S), pT458 p85 (4228), pS473 Akt (193H12), and Y531 pCD19 (3571; all from Cell Signaling).

B10;B6-Rag2^tm1Fwa II2rg^tm1Wjl (CD45.2) mice were irradiated (550 rad) the day before BM reconstitution. BM cells were harvested, and lineage-negative, Sca-1⁺, and cKit⁺ cells from CD45.1 B6.SJL-Ptprc^aPepc^b/BoyJ, CD45.2 WT, or CD45.2 Igβ^KΔR mice were isolated. Cells were mixed 1:1, and cells were introduced via retroorbital injection. After 8 wk, BM and spleen cells were harvested and analyzed by flow cytometry.

B cells from WT×MD4 and Igβ^KΔR×MD4 mice were labeled with CFSE, which were transferred into B6.SJL-Ptprc^aPepc^b/BoyJ mice and challenged i.v. with HEL-conjugated SRBCs (Lampire Biological Products). After 3 d, spleens were harvested, labeled with anti-CD86, and assayed by flow cytometry. After 6 d, splenic B cells were stained with antibodies to IRF4 and Bcl6 and assayed by flow cytometry (Ochiai et al., 2013).

Confocal microscopy was performed as previously described (O’Neill et al., 2009). Images were collected using a confocal microscope (100× objective; SP5-II-STED-CW; Leica). Antibodies were used to visualize intracellular Syk (C-20; Santa Cruz), H2M (E25A; BD Biosciences), Cathepsin L (CPLH-3G10; Santa Cruz), TLR9 (26C593.2; Abnova), Lamp-1 (1D4B), and Cbl-b (B-5; Santa Cruz). Pre–B cells were incubated with 2.4G2 (LifeSciences) and then biotinylated SL156 (BD PharMingen). After permeabilization, cells were incubated with rabbit antibiotin antibodies (Bethyl Laboratories) and then Alexa Fluor 488–conjugated donkey anti–rabbit antibodies (Invitrogen). Acidic compartments were visualized by incubating samples with 1 µM LysoSensor green DND-1 dye (Molecular Probes) at 37°C for 1 h. Samples were then chilled on ice and imaged. PI3K was inhibited with 5 or 10 µM LY294002 (Selleckchem) for 30 min at 37°C. The CD19 receptor was cross-linked with biotin anti-CD19 (1D3; BD PharMingen) simultaneously with BCR cross-linking after incubating with 2.4G2 (LifeScience). The cells were then incubated for 30 min at 37°C, fixed, and stained. For each experimental time point, images of 30 cells were scored for colocalization using the JACoP plug-in (Manders’ coefficient) of ImageJ (National Institutes of Health). Trilocalization was scored using the BlobProb plug-in (Manders’ coefficient) of ImageJ. A threshold of 40 was set for each channel. Superresolution confocal microscopy images were acquired on a ground state depletion/total internal reflection fluorescence superresolution microscope (63× objective, Leica). WT B cells were stimulated with Alexa Fluor 488–conjugated anti-IgG and -IgM (H+L) F(ab)₂ antibodies, fixed, and then stained with antibodies to PIP₃ (Echelon) and EEA1 (C45B10; Cell Signaling).

Biotinylated ODN 1826 and biotinylated control ODN 2138 were purchased from Invivogen. Purified splenic B cells from WT and Igβ^KΔR mice were incubated with 20 µg/ml of streptavidin-conjugated F(ab)₂ IgG and IgM (H+L), followed by either the biotinylated ODN 1826 or control ODN 2138 (Invivogen). Cells were then incubated for 6 h at 37°C. RNA was isolated and quantitative real-time PCR performed as previously described (O’Neill et al., 2009).

Unless otherwise noted, all statistics were done using Student’s t test, using Prism 6 software (GraphPad).

Fig. S1 shows the gene-targeting vector used to derive the mutations in the Igβ^KΔR mouse, confirmed by both PCR genotyping and sequencing. Fig. S1 also shows by immunoprecipitation that these mutations ablated the ability of the Igβ to be ubiquitinated. Fig. S2 shows the cell numbers and surface IgM densities from crossing WT and Igβ^KΔR mice to Pten^−/− mice. Fig. S3 shows the colocalization of PIP₃ and BCR in early endosomes at 60 min, along with the absence of colocalization in late endosomes. Fig. S3 also shows the spatial relationship between BCR, PIP₃, and EEA1 at 15 min using superresolution confocal microscopy.

We would like to thank Sarah Powers for careful reading of this manuscript, Kristen Wroblewski for statistical advice, and Vytas Bindokas and Christine Labno for technical assistance with the superresolution confocal microscopy.

This work was funded by grants from the National Institutes of Health (R01 GM088847 and GM101090).

The authors declare no competing financial interests.

Author contributions: M. Veselits performed most experiments. A. Tanaka performed experiments described in Fig. S1 (D and E) and Fig. 2 E. BM chimeras were performed by K. Hamel and M. Mandal. B. Manicassamy and M. Kandasamy provided experiments described in Fig. 4 E. Y. Chen and P. Wilson assisted with the experiments described in Fig. 4 (D and E). Noncompetitive transfer was performed by R. Sciammas. S.K. O’Neill helped derive the Igβ^KΔR mice, and M.R. Clark oversaw the project and prepared the manuscript.