Activated PI3Kδ signals compromise plasma cell survival via limiting autophagy and increasing ER stress

Antibodies produced by plasma B cells (PCs) are a hallmark of humoral immunity and provide protection against infection and reinfection. Depending upon the context of antigen encounter, activated B cells proliferate and differentiate into antibody-secreting cells (ASCs) or memory B cells (MBCs). Plasmablasts are short-lived ASCs that are generated early during the immune response (Nutt et al., 2015). In contrast, long-lived PCs are predominately derived from germinal centers (GCs) and home to, and preferentially reside in, the bone marrow (BM) to sustain high titers of protective antibodies for decades. In parallel, GCs also produce MBCs that are retained in lymphoid tissues and poised to rapidly differentiate into short- and long-lived PCs following reexposure to antigen (Kurosaki et al., 2015). Thus, understanding key factors that control the differentiation, function, and survival of PCs is critical for development of new therapeutic approaches to either enhance humoral immunity to vaccination or to attenuate pathogenic humoral responses in autoimmune or malignant diseases.

Differentiation of B cells into PCs requires a coordinated genetic reprograming via modulation of key transcription factors, including silencing of Pax5, Bach2, and Bcl6 and activation of IRF4 and Blimp-1 (Nutt et al., 2015). As terminally differentiated B cells, PCs exhibit an exceptional capacity to produce and secrete high levels of antibodies. To accommodate these sustained functional activities, the PC differentiation program entails unique mechanisms to permit enhanced protein accumulation and ER stress while maintaining survival. First, PCs display a distinctive morphology with an expanded cytoplasm and a tightly arranged cisternae of rough ER. Second, PCs continuously activate the unfolded protein response (UPR), a homeostatic, adaptive signaling program that promotes the recovery of ER proteostasis and thereby permits cell survival in the setting of these major metabolic challenges (Bettigole and Glimcher, 2015). Further, autophagy counterbalances ER stress by eliminating excess ER while supplying metabolic substrates to maintain PC viability (Pengo et al., 2013).

Phosphatidylinositide 3-kinase (PI3K) signaling plays a critical role in the activation and function of nearly all cell lineages. Among PI3K family proteins, p110δ, the catalytic subunit of PI3Kδ, is expressed predominantly in lymphocytes and is essential for humoral immune responses (Lucas et al., 2016). p110δ is required for antigen receptor–driven activation and differentiation of B and T cells (Preite et al., 2019). Both loss- and gain-of-function mutations in the gene encoding p110δ (PIK3CD) lead to human immune dysregulation (Fruman et al., 2017; Lucas et al., 2016), albeit through different mechanisms, events that have been partially elucidated in mouse models. Mice expressing catalytically inactive (or lacking) p110δ in T cells exhibit impaired humoral responses due to defective T follicular helper cell formation (Okkenhaug et al., 2002; Rolf et al., 2010). Conversely, B cell–specific activation of PI3K signaling using Foxo1 or Pten knockout mice results in loss of B cell class-switch recombination (CSR) and affinity maturation in response to T cell–dependent (TD) immunizations (Suzuki et al., 2003; Omori et al., 2006; Dengler et al., 2008; Sander et al., 2015; Dominguez-Sola et al., 2015). Further, mice expressing activating mutations in PIK3CD, referred to hereafter as activated (a)PIK3CD mice, display a broad range of alterations in immune function, including decreased antigen-specific class-switched antibodies (Wray-Dutra et al., 2018; Preite et al., 2018; Avery et al., 2018; Stark et al., 2018). aPIK3CD mice mimic the most common heterozygous mutation present in the human immunodeficiency disease, activated PI3Kδ syndrome (APDS; Angulo et al., 2013; Crank et al., 2014; Lucas et al., 2014; Coulter et al., 2017). Both global and B cell–intrinsic aPIK3CD mice (Wray-Dutra et al., 2018) exhibit a reduction in class-switched antibodies, mirroring the reduced IgG levels and proportion of MBCs in APDS subjects. However, while previous work has suggested these deficits primarily reflect defective CSR (Preite et al., 2018; Avery et al., 2018), the specific role for aPIK3CD signaling in differentiation and function of activated, mature B cells has not been determined.

Herein, we investigated how B cell–intrinsic aPIK3CD signals specifically impact MBC and PC biology. Surprisingly, we found aPIK3CD increased the generation of antigen-specific MBCs but altered the antibody response kinetics following secondary antigen challenge. Strikingly, our detailed investigation showed that aPIK3CD markedly limited PC generation and survival. Elevated PI3Kδ activity led to increased ER stress and impaired autophagy in PCs, thereby compromising PC survival. Together, our findings demonstrate that modulation of PI3Kδ signaling is essential for PC development via regulating the UPR and autophagy via the mTORC1 pathway to mitigate ER stress and thereby ensure PC survival.

To investigate the impact of activated PI3Kδ signaling in MBC function, we first crossed mice harboring an endogenous loxP-flanked allele encoding the murine equivalent of the most common gain-of-function mutation of PIK3CD in APDS patients (E1021K [E1020K in mice]) with CD21-Cre mice (Cd21^Cre/+ Pik3cd^E1020K/+) to express the variant in mature B cells (hereafter CD21-aPIK3CD; Wray-Dutra et al., 2018). To assess the TD memory response, we immunized CD21-aPIK3CD and CD21-WT littermate controls with virus-like particles (VLPs) containing a TLR ligand (VLP-ssRNA; Hou et al., 2011). 90 d after immunization, VLP-specific MBCs were isolated using magnetic bead–based enrichment as previously described (Liao et al., 2017; Krishnamurty et al., 2016; Taylor et al., 2012). VLP⁺ B cells were gated within B220⁺ and B220^lo CD138⁺, and VLP-specific MBCs were identified as CD138⁻GL7⁻CD38⁺ (Fig. S1 A). Based on this gating strategy, CD21-aPIK3CD mice had increased numbers of VLP-specific IgM⁺ MBCs and similar numbers of switched (IgM⁻IgD⁻) MBCs compared with control mice (Fig. 1, A–C; and Fig. S1 B). To further define cells with a memory phenotype, we assessed VLP⁺CD73⁺CD80⁺ MBCs (Fig. S1 C). CD21-aPIK3CD mice exhibited significantly increased numbers of both IgM⁺ and switched VLP⁺CD73⁺CD80⁺ MBCs compared with controls (Fig. 1 C).

To assess the effect of activated PI3Kδ on MBC development in the same inflammatory environment, we used chimeras containing cells with both genotypes. To minimize confounding effects of PI3K on development, we induced aPIK3CD in GC B cells by crossing Pik3cd^E1020K/+ with activation-induced cytidine deaminase (AID)–Cre (hereafter AID-aPIK3CD). Mixed BM chimeras were generated by transferring a 1:1 mixture of CD45.2 AID-aPIK3CD and CD45.1/2 AID-WT cell or CD45.2 AID-WT and CD45.1/2 AID-WT cells, respectively, into lethally irradiated CD45.1 WT hosts. Reconstituted mice were immunized with VLP-ssRNA and analyzed 90 d after immunization. Consistent with the proportional input populations, all chimeras exhibited a 1:1 ratio of CD45.2 versus CD45.1/2 cells within the IgD⁺ naive/MBC compartment. In contrast, AID-aPIK3CD:AID-WT recipients exhibited a ∼4:1 skewing for switched MBCs compared with 1:1 ratio for AID-WT:AID-WT controls (Fig. 1 D). Together, these findings demonstrate that activated PI3Kδ signaling promotes the generation of both IgM⁺ and switched MBCs and raised the question of how CD21-aPIK3CD mice would respond to a secondary challenge.

To understand whether aPIK3CD could impact the kinetics of the humoral response following secondary challenge, we immunized CD21-aPIK3CD and CD21-WT mice with VLP-ssRNA. On day 90 after immunization, the mice were left unchallenged or rechallenged with RNA-free VLP (Fig. 1 E), an immunogen that induces class-switched antibodies only in the setting of primary immunization with VLP-ssRNA. As predicted (Wray-Dutra et al., 2018), IgG2c VLP-specific antibodies were reduced in CD21-aPIK3CD primary-only recipients compared with controls (Fig. 1 F). Surprisingly, within 3 d of rechallenge, VLP-specific IgG2c levels in CD21-aPIK3CD mice increased over primary-only levels, while titers in WT mice remained unchanged (Fig. 1 F, far left). By day 5, CD21-aPIK3CD mice exhibited IgG2c levels that exceeded those in WT mice (Fig. 1 F) and remained above WT counterparts at 21 d after rechallenge (Fig. 1 F, middle). In contrast, at later time points, CD21-aPIK3CD mice exhibited a decline in IgG2c, reaching levels equivalent to controls by day 90 (Fig. 1 F, right).

To gain insight into these unexpected kinetics, we performed VLP-specific ELISPOT assays to quantify ASC number and productivity. Consistent with serum antibody levels, splenic populations in CD21-aPIK3CD mice contained a higher number of spots at day 5 after rechallenge (Fig. 1 G, middle). In contrast, the spot size of these splenic ASCs was significantly smaller than in controls (Fig. 1 G, left and right). Notably, consistent with the rapid induction of splenic ASC after rechallenge (as well as the reduced initial primary antibody response), ASC numbers in the BM of CD21-aPIK3CD were lower than in controls at day 5. Further, at 90 d after rechallenge (Fig. 1, G and H), a time point where long-lived PCs (LLPCs) are predicted to populate the BM, spot numbers remained lower in CD21-aPIK3CD mice, findings consistent with a more rapid decay of antibody levels. We also evaluated VLP-specific IgM response upon rechallenge. VLP-specific IgM was not detectable before rechallenge in either genotype. However, by day 5, CD21-aPIK3CD mice exhibited a trend for increased IgM (Fig. S1 D). Despite this increase, and in contrast to control mice, VLP-specific IgM ASCs remained undetectable in the spleen or BM (Fig. S1 E). Together, these data show that CD21-aPIK3CD mice exhibit an increased ASC response upon antigen rechallenge but are not able to sustain antibody production over time, suggesting that inappropriate PI3Kδ signaling disrupts PC longevity and/or function but does not alter MBC formation.

We next interrogated the effect of activated PI3Kδ on PC formation during a primary immune response in vivo. CD21-aPIK3CD and CD21-WT mice were immunized with VLP-ssRNA. At day 14 after immunization, VLP-specific CD138⁺ PCs were markedly reduced in CD21-aPIK3CD mice compared with controls (Fig. 2 A). Moreover, CD21-aPIK3CD exhibited significantly reduced VLP-specific IgG and IgG2c PCs in spleen and BM, as measured by intracellular flow cytometry (Fig. 2 B) and ELISPOT (Fig. 2 D). In contrast, IgM PCs were present in similar numbers in CD21-aPIK3CD and controls (Fig. 2, C and D). The decrease in PC numbers CD21-aPIK3CD mice was not secondary to a reduced GC response. Total GC B cells were increased (Fig. S2 A), and the absolute number of VLP-specific and Ig2C⁺ VLP-specific GC B cells in CD21-aPIK3CD mice was similar to controls (Fig. S2, B and C).

Next, to evaluate the potential impact of aPIK3CD on PC generation in a competitive environment, we established BM chimera mice as in Fig. 1 D, immunized these recipients with VLP-ssRNA, and analyzed them at 14 d after immunization (Fig. S3 A). VLP-specific aPIK3CD GC B cells outcompeted WT GC cells, resulting in ∼2.5:1 skewing as compared controls (Fig. S3, B and C). Similar to what we observed previously (Wray-Dutra et al., 2018), AID-aPIK3CD GC B cells were enriched for cells with a light zone (LZ) phenotype (Fig. S3 D). Strikingly, despite the preferential expansion of aPIK3CD cells within GC, the VLP-specific PC-to-GC B cell ratio was significantly lower AID-aPIK3CD chimera mice (Fig. 2 E and Fig. S3 B), suggesting counterselection of antigen-specific PCs. These findings demonstrate that while aPIK3CD increased the expansion and selection of antigen-specific GC B cells, aPIK3CD impairs the generation or survival of PCs.

To begin to determine how aPIK3CD modulates PC biology, we next used an in vitro culture system capable of modeling differentiation of mouse naive B cells into PCs (Nojima et al., 2011). In this system, CD21-aPIK3CD or CD21-WT B cells were cocultured with CD40LB stromal cells coexpressing membrane-bound CD40L and the secreted prosurvival cytokine B cell activating factor. To generate plasmablasts, cultures were costimulated with R848 and IL-21. Although CD21-aPIK3CD B cells cultures generated slightly but significantly more CD138⁺ plasmablasts at day 4, CD21-aPIK3CD B cell cultures generated significantly fewer CD138⁺ plasmablasts at days 6 and 7 (Fig. 2 F and Fig. S3 E). Similar data demonstrating a reduction in plasmablasts in CD21-aPIK3CD B cell cultures were observed with other relevant activating stimuli, including IL-4 + IL-21, LPS + IL-4, and LPS + IL-21 (Fig. S3 F). Together, these data demonstrate that hyperactive PI3Kδ signaling inhibits PC formation.

Previous studies attributed the impaired humoral response caused by hyperactive PI3Kδ to a CSR defect (Avery et al., 2018; Preite et al., 2018), mainly driven by reduced AID expression (Avery et al., 2018). Our mouse model represents a robust tool to examine the effect of aPIK3CD at specific B cell developmental stages. We previously showed that mice expressing aPIK3CD under the control of Aicda locus, in which B cells become overactive following AID expression, still exhibit diminished production of VLP-specific class-switched antibodies despite normal GC reactions (Wray-Dutra et al., 2018). To evaluate whether aPIK3CD signaling impacts affinity maturation, we immunized AID-aPIK3CD and controls with NP-CGG in alum and measured the titers of total NP30-binding and high-affinity NP4-binding antibodies at day 14 by ELISA. We found that both NP4- and NP30-specific IgG1 levels in the serum were significantly lower (Fig. S3 G) in AID-aPIK3CD mice. However, there was no difference in the ratio of NP4 to NP30 IgG1, which suggested that aPIK3CD does not alter affinity maturation, consistent with previously published studies (Preite et al., 2018; Avery et al., 2018). Next, we assessed the humoral response using the Cγ1-Cre strain, where aPIK3CD is expressed subsequent to CSR, leading to IgG1 expression. Cγ1-aPIK3CD mice exhibited a significant decrease in NP4- and NP30-specific serum IgG1 levels at 14 d after immunization with NP-CGG, again without differences in the ratio of NP4 to NP30 IgG1 (Fig. S3 H). Together, these data demonstrate that aPIK3CD expression leads to a defect in the magnitude of class-switched, antigen-specific antibody responses via a mechanism independent of AID expression and CSR.

Based on the above findings, we next performed studies to directly determine whether aPIK3CD exerts a PC-intrinsic impact on PC survival or function. We crossed Pik3cd^E1020K/+ mice with Blimp1-Cre strain (hereafter, Blimp1-aPIK3CD). To track Blimp1-expressing cells in this model, we also crossed Blimp1-Cre with a loxP-flanked stop cassette EYFP strain controlled by the constitutively expressed Rosa26 promoter. Following immunization with VLP-ssRNA, 80–85% of PCs expressed YFP, while less than 8% of naive B cells, T cells, GC B cells, and non-B/T cells expressed YFP (Fig. S4 A). Thus, the Blimp1-aPIK3CD mice largely represent a PC-intrinsic model of activated PI3Kδ.

To evaluate the humoral response in these mice, Blimp1-aPIK3CD and control mice were immunized with VLP-ssRNA and analyzed 14 d after immunization. Consistent with near absence of aPIK3CD expression before PC commitment, we observed no difference in GC events, including the number and frequency of total and VLP-specific GC B cells, switching to IgG2c, or dark zone (DZ)/LZ polarization (Fig. S4, B–E). However, the proportion and number of VLP-specific PCs in Blimp1-aPIK3CD mice was significantly reduced compared with controls (Fig. 3 A). Moreover, the number of VLP-specific IgM⁺, IgG⁺, and IgG2c⁺ PCs were significantly decreased in Blimp1-aPIK3CD compared with controls, as measured by flow cytometry and ELISPOT (Fig. 3, B–D). Finally, serum levels of VLP-specific IgM, IgG, and IgG2c antibodies were significantly diminished in Blimp1-aPIK3CD (Fig. 3 E). Together, these data demonstrate that hyperactive PI3Kδ signaling prevents maturation of PCs.

To explore the cellular mechanisms underlying the humoral defect observed in aPIK3CD mice, we performed RNA sequencing (RNA-seq) of flow-sorted PCs, GC (LZ and DZ populations) and naive B cells from immunized aPIK3CD and WT littermate controls. To eliminate any effects of aPIK3CD on early B cell development and to permit use of naive B cells as a negative control, we used AID-aPIK3CD and AID-WT mice for RNA-seq studies. Principal-component analysis showed that the GCs, naive B cells, and PCs clustered into transcriptionally distinct groups (Fig. S4 F). We did not identify differences between genotypes in naive B cells or either GC subset (adjusted P < 0.05; Fig. S4, G–I). However, analysis of transcriptome of PCs identified 472 differentially expressed genes (Fig. 4 A). As expected, gene set enrichment analysis (GSEA) revealed that conventional targets of PI3K–AKT–mTORC1 signaling and mTORC1 signaling were increased in AID-aPIK3CD PCs, validating our model (Fig. 4 B). We also found that genes associated with glycolysis, protein secretion, and the UPR were increased in AID-aPIK3CD PCs. UPR genes specifically up-regulated in AID-aPIK3CD PCs included the chaperones BiP (Hspa5), Grp75 (Hspa90), and DjaJ/Hsp40 (Dnaja3, Dnajb1, and Dnajb11); members of the ER-associated protein degradation pathway (Edem1 and Herpud1); protein disulfide isomerases (Pdia4 and Pdia3); protein trafficking effectors (Gosr2 and Srprb); and metabolic regulators (Chac1, Asns,and Slc7a5). Also, gene ontology enrichment analysis revealed that aPIK3CD PCs were enriched for ER-genesis and activation of the secretory apparatus, including protein trafficking, ER–Golgi transport, and Golgi–vesicle transport (Fig. 4 C). Altogether, these data suggest that aPIK3CD is associated with increased transcription of PC genes associated with protein secretion.

To directly assess the potential role for aPIK3CD in regulating transcription of ER genes in PCs, we quantified the amount of two ER-resident proteins, PDI and BiP. Consistent with the RNA-seq data, in vitro–generated CD21-aPIK3CD PCs had an increase in mRNA encoding PDI and BiP as measured by qPCR (Fig. 5 A). We also stained PCs with an ER-Tracker fluorescent dye to measure ER content. Flow cytometry analysis revealed that the intensity of ER-Tracker staining was significantly increased in CD21-aPIK3CD PCs compared with control PCs (Fig. 5 B), indicative of a higher abundance of ER proteins.

As an independent qualitative analysis of PC morphology, we used electron microscopy to determine if aPIK3CD altered ER structure or size and/or other ultrastructural features. CD21-aPIK3CD PCs displayed multiple abnormalities as compared with controls (Fig. 5 C). The cytoplasm of most aPIK3CD PCs was densely occupied with ribosomes and mitochondria, and uniformly spaced ER structures were absent or altered in comparison to control cells (Fig. 5 C). Although some CD21-aPIK3CD ASCs exhibited an organized ER structure, in contrast to controls, these cells also contained multiple large vacuole-like structures (Fig. 5 C).

Consistent with the deregulation of ER by aPIK3CD, we found enhanced ER stress in CD21-aPIK3CD PCs, as evidenced by increased spliced XBP1, total XBP1, and Blimp1, and there was no difference in ATF4 or ATF6 (Fig. 5 D). Supporting our in vitro data, ex vivo CD21-aPIK3CD VLP-specific and polyclonal PCs, but not naive or GC B cells, displayed a significantly increased ER-Tracker staining compared with controls at days 7 and 14 after immunization with VLP-ssRNA (Fig. 5 E). Together, these data demonstrate a direct control of ER homeostasis mediated by PI3Kδ-dependent signaling.

Next, we investigated whether the autophagy pathway, critical for the breakdown of excess or damaged ER (Pengo et al., 2013), was altered in aPIK3CD PCs. Because macroautophagy (hereafter autophagy) is mainly controlled by mTORC1 (Jung et al., 2010), a key downstream target of the PI3Kδ pathway, we first assessed PI3Kδ and mTORC1 activity in control and CD21-aPIK3CD B cell subsets by measuring phosphorylated AKT (T308) and ribosomal protein S6 (S235/236), respectively. CD21-aPIK3CD B cells and PCs displayed significant enhanced levels of pAKT and pS6 (Fig. S5 A). Since mTORC1 negatively regulates autophagy, we hypothesized that aPIK3CD suppresses autophagy in differentiating PCs. To measure autophagy, we quantified the processing of LC3 to its autophagosome membrane-bound lipid-conjugated form (LC3-II). Immunoblot analysis of primary splenic B cells differentiated into PCs in vitro showed that CD21-aPIK3CD cells have less accumulation of LC3-II compared with controls (Fig. 5 F). Treatment with chloroquine, an inhibitor that prevents the lysosomal breakdown of autophagosomes, increased LC3-II accumulation in both CD21-aPIK3CD and CD21-WT cells but did fully not rescue the deficit in CD21-aPIK3CD cells (Fig. 5 F). These results were further validated by quantifying the formation of LC3 puncta using fluorescence microscopy. Again, aPIK3CD cells exhibited significantly less LC3⁺ autophagosome accumulation (Fig. 5 G and Fig. S5 B). Together, these data demonstrate that autophagy is inhibited in PCs with hyperactive PI3Kδ signaling.

PCs require UPR and autophagy to control ER stress, and dysregulation in these pathways impairs PC survival (Auner et al., 2010; Pengo et al., 2013). Our data showed that hyperactive PI3Kδ signaling leads to a decrease in splenic and BM PC numbers, suggesting that altered PC survival might be the primary driver of these abnormalities. We evaluated PC apoptosis using caspase and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays. In vitro–cultured aPIK3CD activated B cells exhibited slightly reduced caspase and TUNEL signals at day 6 after stimulation with R848 and IL-21, consistent with slightly increased survival (Fig. S5, C and D). In contrast, CD21-aPIK3CD ASCs exhibited significant increases in both caspase and TUNEL signal relative to controls (Fig. 6, A and B). Consistent with this, MCL-1 expression, an essential anti-apoptotic BCL-2 family member for PC survival (Peperzak et al., 2013), was decreased in CD21-aPIK3CD ASCs compared with controls (Fig. 6 C). Next, we assessed survival of PCs directly ex vivo after VLP-ssRNA immunization. Our data revealed that CD21-aPIK3CD VLP-specific and polyclonal PCs exhibited significant increases in caspase compared with controls, while naive and GC B cells did not (Fig. 6 D). Together, these data show that aPIK3CD promotes PC apoptosis.

We next tested whether inhibition of PI3Kδ can promote improved survival of aPIK3CD PCs. To explore this, in vitro–cultured CD21-aPIK3CD and CD21-WT B cells were allowed to expand for 4 d, and then idelalisib, a selective PI3Kδ inhibitor, was added to cultures for 48 h before analysis. Idelalisib decreased the intensity of ER-Tracker staining in aPIK3CD PCs to a level comparable to untreated WT (Fig. 6 E). Importantly, the defect in PC numbers derived from in vitro–differentiated CD21-aPIK3CD B cells was also fully restored by inhibiting PI3Kδ (Fig. 6 F). Next, to assess whether mTORC1 regulates PC survival in the context of hyperactive PI3Kδ, we performed similar in vitro experiments and treated cells with rapamycin, a potent mTORC1 inhibitor. Indeed, rapamycin reduced the intensity of ER-Tracker staining and nearly restored PC formation in in vitro–differentiated B cells isolated from CD21-aPIK3CD mice (Fig. 6, E and F). Because mTORC1 is a key regulator of multiple biological processes, we next sought to increase autophagy, independently of direct pharmacological targeting of mTORC1 using starvation (Jung et al., 2010). To assess whether induction of autophagy can contribute to PC generation, we did similar in vitro experiments, and then cultures were serum starved for 48 h before analysis. We found that starvation decreased the intensity of ER-Tracker staining and significantly increased the proportion of PCs differentiated from CD21-aPIK3CD B cells (Fig. 6 G). Together, these results demonstrate that aPIK3CD perturbs ER homeostasis and PC generation via an mTORC1-dependent signaling program.

PIK3CD signaling exerts a critical role in immune regulation and humoral immunity (Lucas et al., 2016). Notably, the activity of PI3Kδ must be precisely regulated to achieve optimal immune responses, a concept highlighted by the distinct immunodeficiency syndromes observed in human subjects with inactivating versus activating mutations in p110δ (Fruman et al., 2017). Loss of p110δ function leads to a severe deficit in T-independent (TI) IgM and TD IgM and IgG antibody responses (Lucas et al., 2016). In contrast, activating mutations in PIK3CD cause a human primary immune disorder characterized, in part, by enhanced TI antibody responses and conversely, reduced IgG levels and a deficit in switched and unswitched MBCs (Angulo et al., 2013; Crank et al., 2014; Lucas et al., 2014; Coulter et al., 2017). The complex and diverse phenotype observed in APDS patients, which extends beyond features described here (Coulter et al., 2017), underscores the pleiotropic effects of PI3Kδ in distinct immune cell lineages. To better define the impact of activated PI3Kδ signaling in specific cell types, we previously generated a conditional aPIK3CD mouse model that permits precise lineage- and developmental stage–specific modeling of aPIK3CD in an immune competent setting (Wray-Dutra et al., 2018). In this study, we investigated the specific role of elevated PI3Kδ activity in the differentiation and function of MBCs and PCs. B cell–intrinsic expression of aPIK3CD led to increased formation of antigen-specific IgM and switched MBCs. Consistent with the elevated numbers of MBCs, aPIK3CD mice generated increased numbers of ASCs and more robust antibody responses following rechallenge. However, despite this initial amplified response, antibody levels declined rapidly correlating with reduced numbers of splenic and BM PCs. Our studies provide a detailed mechanistic explanation for these important and surprising observations. We demonstrate that aPIK3CD uniquely impairs PC survival by limiting autophagy and increasing ER stress, perturbations likely to be driven by the mTORC1 pathway. Our combined findings show that PI3Kδ must be precisely modulated to control ER homeostasis and permit PC generation and survival.

APDS patients exhibit a proportional and, over time, numerical reduction in unswitched and class-switched MBCs (Coulter et al., 2017; Angulo et al., 2013; Crank et al., 2014; Avery et al., 2018). However, the impact of aPIK3CD on antigen-specific MBC populations has not been explored. To address the cell-intrinsic impact of aPIK3CD in B cell memory, we assessed antigen-specific MBCs in aPIK3CD mice immunized with TD antigens. Both IgM and switched antigen-specific MBC compartments were expanded in CD21-aPIK3CD mice. These findings correlated with the impact of aPIK3CD within the GC, leading to generation of larger numbers of antigen-specific MBCs, observations corroborated by the competitive advantage of aPIK3CD versus WT antigen-specific GC B cells in BM chimeras. Further, the increase of IgM MBCs in CD21-Cre, but to a lesser extent in AID-Cre, aPIK3CD mice implies that many IgM MBCs are likely generated extrafollicularly, independent of GCs (Viant et al., 2021; Harms Pritchard et al., 2019,Preprint). These IgM MBCs may preferentially derive from activation of B cells within the expanded marginal zone compartment present in aPIK3CD mice; findings also consistent with the proportional increase of IgM versus switched MBCs in APDS subjects. Importantly, while our data showing increased MBCs in aPIK3CD mice appear distinct from the reduced number of switched MBCs in APDS patients, our current and previous observations suggest that the MBC deficit in the setting of global aPIK3CD likely reflects the combined impacts of reduced generation of naive B cells (secondary to reduced BM B lymphopoiesis), alterations in the BCR repertoire with MZ compartment skewing and a partial CSR defect. Finally, despite our new data and previous findings implying that the humoral deficits in APDS are predominantly B cell intrinsic, we cannot exclude additional B cell–extrinsic impacts caused by aPIK3CD on MBC development, particularly with respect to the functional activities of T follicular helper cells (Bier et al., 2019; Preite et al., 2018).

The expanded MBC compartment in B cell–intrinsic aPIK3CD mice correlated with greater numbers of splenic ASCs and a more robust initial antibody response following antigen rechallenge. However, splenic ASCs in aPIK3CD mice were smaller in size, and antigen-specific antibody levels rapidly declined. Further, despite the increase in splenic ASCs, BM ASC numbers did not increase, suggesting that ASC survival was reduced in aPIK3CD mice. Consistent with this concept, despite preferential expansion of antigen-specific aPIK3CD GCs and MBCs, we observed a bottleneck in the generation of antigen-specific PCs derived from aPIK3CD-expressing B cells in BM chimeras. In further support of the concept that aberrant PIK3CD signaling compromises PC formation, primary immunization in both B cell– and PC-intrinsic aPIK3CD mice resulted in reduced generation of antigen-specific PCs. Using in vitro experiments, we confirmed that differentiated aPIK3CD ASCs manifest increased apoptosis and resulted in fewer ASCs following extended culture. Of note, a previous study in APDS subjects identified both an increase in circulating plasmablasts and increased apoptosis following in vitro stimulation of primary B cells (Wentink et al., 2017). In agreement with these observations, our data showed that aPIK3CD promoted increases in ASC formation yet decreases in PC survival.

Previous work demonstrated that aPIK3CD impairs the humoral response through defective CSR (Avery et al., 2018; Preite et al., 2018), correlating with reduced AID expression (Avery et al., 2018). Further, in vitro–stimulated transitional B cells from APDS patients failed to generate ASCs and acquire a normal PC differentiation program despite the expression of key early PC genes, including PRDM1, XBP1, and IRF4 (Avery et al., 2018). Our new observations expand upon these collective findings, showing that expression of aPIK3CD following AID up-regulation or class-switching to IgG1 (using AID-Cre and Cγ1-Cre strains, respectively) limits class-switched antibody production. Additionally, our findings imply that these events occur independently of intact B cell expansion and selection within GCs (Wray-Dutra et al., 2018). Consistent with our findings, aPIK3CD or alternative models of enhanced PI3K signaling also induced elevated Blimp-1 expression and promoted premature ASC differentiation (Setz et al., 2018; Preite et al., 2018). Thus, while aPIK3CD does impact CSR, our combined data demonstrate that PC survival limits sustained antibody production in aPIK3CD mice and represents an additional driver for reduced IgG levels, implying that a similar mechanism is operative APDS subjects.

PCs continually produce and secrete vast amounts of antibodies (Helmreich et al., 1961). To adapt to this enormous metabolic challenge, the UPR is induced as activated B cells differentiate into PCs (Gaudette et al., 2020; Gass et al., 2002). Simultaneously, autophagy promotes degradation of unfolded proteins and damaged ER, a process that optimizes antibody secretion and sustains PC viability in mice and humans (Arnold et al., 2016; Halliley et al., 2015; Pengo et al., 2013). Based upon transcriptional, protein blotting, confocal microscopy, electron microscopy, and flow cytometry studies, we identified marked changes in aPIK3CD PCs, including reduction in autophagosome assembly, altered PC ultrastructure and ER contents, and up-regulation of the UPR. The observed defect in autophagy is one possible reason we find increased ribosome and ER quantity, and it may also lead to the accumulation of misfolded protein and accentuate the UPR program. Additionally, the PI3Kδ–mTORC1 pathway regulates several aspects of cell growth and metabolic programs in activated B cells such as mitochondrial function and ROS production, which play important roles in PC differentiation, function, and survival (Jang et al., 2015; Jellusova and Rickert, 2016). Thus, it is possible that such metabolic changes in aPIK3CD B cells may contribute to ER stress and/or PC death, independently of autophagy. Collectively, these events lead to increased ER stress likely contributing to reduced PC survival.

We determined whether blocking PI3Kδ-dependent signals might reverse the observed defects in aPIK3CD PCs. Treatment with the PI3Kδ-specific inhibitor idelalisib, as well as with the mTORC1-specific inhibitor rapamycin, reversed the elevated ER content and promoted PC survival. The impact of rapamycin alone supports the interpretation that elevated aPIK3CD-driven mTORC1 signaling, which inhibits autophagy (Jung et al., 2010), promotes ER stress and PC death in aPIK3CD B cells. These results may seem confounding given that mTORC1 initiates the UPR before Blimp-1 (Benhamron et al., 2015; Gaudette et al., 2020) and that mTORC1 is required for PC generation (Jones et al., 2016), presumably via its role in priming B cells for antibody secretion by up-regulating UPR components (Benhamron et al., 2015). We propose that during PC development, mTORC1 is required early to promote UPR but that shortly after Blimp1 expression, mTORC1 is down-regulated to enable clearance of damaged ER via autophagy. This idea is consistent with transcriptional profiling data showing that mTORC1 is down-regulated in terminally differentiated human PCs (Garimalla et al., 2019). Collectively, these findings show that PCs exploit PIK3CD–mTORC1 to prepare for antibody secretion but that precise and dynamic tuning of this pathway is required for the survival of newly generated PCs to elicit optimal humoral immunity; however, the impact of aPIK3CD on mature PCs after LLPC generation remains to be addressed.

Inhibitors of PI3Kδ (Rao et al., 2017; Chiu et al., 2017) and mTORC1 (Rae et al., 2016; Maccari et al., 2018; Lucas et al., 2014) partially reverse the immune dysregulation present in APDS patients, indicating that enhanced PI3Kδ activity leads to broad alterations of immune function through skewing toward the mTORC1 pathway. Ongoing clinical trials for PI3Kδ inhibitors in APDS will assess their long-term safety and efficacy. Our data support a potential role for PI3K inhibitors to boost humoral immune responses to acute vaccination even in the setting of short-term inhibitor therapy. Consistent with this idea, published work in both mouse and human cells has shown mTORC1 is required for PC generation but dispensable for survival of LLPCs (Nguyen et al., 2018; Jones et al., 2016). Additionally, maintenance of the UPR is essential for plasmablast adaptation to ER stress to prolong survival and sustain PC persistence and function (Tellier et al., 2016; Gass et al., 2002; Auner et al., 2010). Thus, our work supports utilization of PI3Kδ/mTORC1 inhibitors in APDS and suggests that targeted immunization during such therapy may provide an efficient strategy to establish sustained pathogen-specific, PC mediated IgG production.

In summary, this study provides insight into the critical, balancing role for PI3Kδ signals in PC biology, thereby expanding our understanding of PI3Kδ signaling in humoral immunity. Most notably, we identify a previously unrecognized requirement for modulating PI3Kδ activity in order to balance autophagy and the UPR, events essential to modulate ER stress and ensure PC survival. Our data also suggest that vaccination of APDS subjects in association with PI3Kδ–mTORC1 inhibition may permit generation and survival of pathogen-specific LLPCs, improving immunity to infection. Conversely, enhancing PI3Kδ activity may provide a means to trigger early PC death and dampen autoantibody responses and/or potentially limit survival of PC-derived malignancies.

CD21-Cre (Cr2-Cre, 006368), AID-Cre (Aicda-Cre, 007770), Cγ1-Cre (Ighg1-Cre, 010612), Blimp1-Cre (Prdm1-Cre, 008827), R26R-EYFP (006148), B6 (C57BL/6J, 000664), and B6 CD45.1 (002014) mice were originally purchased from Jackson Laboratory and bred/maintained in specific pathogen–free facility at Seattle Children’s Research Institute. Pik3cd^E1020K/+ mice were generated in our laboratory as previously described (Wray-Dutra et al., 2018). All mice were maintained on C57BL/6 background. Male and female 10- to 14-wk-old mice were used in the experiments. Animal experiments were conducted according to the protocols approved by Institutional Animal Care and Use Committee at Seattle Children’s Research Institute.

Mice were injected intraperitoneally with 2 µg VLP-ssRNA, or RNA-free VLP for secondary challenge experiments, in PBS. For NP-CGG immunization, mice were intraperitoneally immunized with 100 µg NP-CGG (Biosearch Technologies) in alum (Imject Alum Adjuvant; Thermo Fisher Scientific) at 1:1 ratio. Analysis was performed after immunization at time points described in the text. All VLP reagents were provided by B. Hou (Chinese Academy of Sciences, Beijing, China).

Spleens and BM were harvested into ice-cold FACS buffer (PBS with 2% FBS). Cells were resuspended in RBC lysis buffer (ACK), incubated for 2–3 min, and then spun down and resuspended in FACS buffer. Samples were then filtered through 70-µm cell strainers (Corning). Single-cell suspensions were stained with fluorescence-labeled antibodies for 20–30 min at 4°C. Intracellular staining was performed using Fixation/Permeabilization (Cytofix/Cytoperm; BD Biosciences) following cell-surface staining according to the manufacturer’s instructions. For Mcl-1 staining, intracellular staining was performed using the Foxp3 Staining Buffer Set (eBioscience) following cell-surface staining according to the manufacturer’s instructions. For pS6 and pAKT, cells were directly fixed in 4% paraformaldehyde for 20–30 min at 4°C, followed by permeabilization cells at using cold Phosflow Perm Buffer III (BD Biosciences) for at least 30 min, then cells stained with surface and phospho-site antibodies for 35 min at 4°C.

The initial gating strategy performed included forward scatter (FSC)-A/side scatter (SSC)-A, exclusion of doublets (through SSC-H/SSC-W and FSC-H/FSC-W), and live cells. Cell counts were determined using Spherotech Accucount Fluorescent Beads (Thermo Fisher Scientific). After total cell calculation, the number of specific populations was determined by the frequency of the population determined by flow cytometry. Samples were acquired on a LSRII (BD Biosciences) flow cytometer, and data were analyzed using FlowJo software (Tree Star). For cell sorting, cells were prepared and stained as described above and sorted on FACSAria II (BD Biosciences) flow cytometer.

The following reagents were used according to the manufacturer’s instructions: violet live caspase probe (BD Biosciences), in situ cell death detection kit, fluorescein (TUNEL; Sigma), and ER-Tracker Red (BODIPY TR Glibenclamide; 200 nM; Invitrogen). Anti-murine antibodies used in the study include the following: B220 (RA3-6B2), CD19 (1D3), CD138 (281–2), GL7 (GL7), CD38 (90), IgM (II/41), and IgD (11–26) from BioLegend; IgG2c (polyclonal) from Southern Biotech; CXCR4 (2B11), CD86 (GL1), CD73 (TY/11.8), CD80 (16-10A1), F4/80 (BM8), CD3 (145-2C11), CD45.1 (A20), CD45.2 (104), Mcl1 (D2W9E), H2kd (SF1-1.1), SA-APC-Cy7, and SA-FITC from BD Biosciences; SA-PE-AF647 from Thermo Fisher Scientific; and pS6 (Ser235/236; D57.2.2E) and pAKT (Thr308; D25E6) from Cell Signaling.

Splenic cell suspensions were prepared and resuspended in 200 µl FACS buffer and first incubated with 1 μg PE-AF647 (Decoy) at room temperature (RT) for 10 min. 1 μg VLP-AF647 was then added, with a final VLP-AF647 concentration ∼2 nM, and incubated for 30 min at 4°C. Cells were washed, incubated with anti-Cy5/AF647 MicroBeads for 30 min at 4°C, and isolated with magnetic LS columns (both from Miltenyi Biotec). This enrichment method was adapted from previously described studies (Krishnamurty et al., 2016; Taylor et al., 2012; Liao et al., 2017).

BM cells were isolated from leg bones (femurs and tibias) from congenically marked donors of the following genotypes: AID-Cre Pik3cd^E1020K/+ (CD45.2), AID-Cre WT (CD45.2), and AID-Cre WT (CD45.1/2). Bones were rinsed in PBS and then ground using a pestle. Extracted cells were washed in PBS, resuspended in RBC lysis buffer (ACK), and filtered through 70-µm strainers. Single-cell suspensions were resuspended in PBS and mixed at 1:1 ratio, and the ratio of cells mixed was confirmed prior injection by flow cytometry. WT (CD45.1) recipient mice were lethally irradiated (950 rads) and retro-orbitally injected with total 5 × 10⁶ cells of the indicted genotypes into each recipient ∼4–6 h after irradiation. Mice were placed on Baytril-medicated water (5 ml/liter) for 2 wk after irradiation. Resulting BM chimeras were bled every 4 wk after transplant to check degree of chimerism, and mice were held for 10–12 wk to allow for hematopoietic reconstitution before performing experiments.

96-well Nunc-Immuno MaxiSorp plates (Thermo Fisher Scientific) were coated overnight at 4°C with 2 µg/ml VLP, 35 µg/ml NP4-BSA, or 35 µg/ml NP30-BSA (Biosearch Technologies) in PBS. Plates were blocked with 2% BSA before sample incubation, and then plates were incubated with serially diluted serum at RT for 2 h Subsequently, plates were incubated with HRP-conjugated goat anti-mouse IgM, IgG, IgG1, or IgG2c (1:2,000 dilution; Southern Biotech). Peroxidase reactions were developed using OptEIA TMB substrate reagent set (BD Biosciences) and stopped with sulfuric acid. Absorbance was measured at 450 nm using a SpectraMax i3X microplate reader (Molecular Devices).

MultiScreen 96-well ELISPOT plates (Millipore) were coated overnight at 4°C with 2 μg/ml VLPs. Plates were blocked with 10% FBS in complete DMEM (Thermo Fisher Scientific) for 1 h at RT. Serially diluted cells were added to the plate and then incubated overnight at 37°C in 5% CO₂. Cells were washed off, and plates were incubated with HRP-conjugated goat anti-mouse IgM, IgG, or IgG2c (1:2,000 dilution; Southern Biotech). Nonspecific (background) spots were determined in wells containing no cells. Spots were developed using the AEC substrate kit (Vector Laboratories) and scanned and counted using the CTL ELISPOT reader and Immunospot analysis software (Cellular Technology).

The in vitro differentiation system using 40LB stromal cells was performed as described previously (Haniuda and Kitamura, 2019; Nojima et al., 2011). In brief, 40LB cells (in complete DMEM supplemented with 10% FBS; Thermo Fisher Scientific) were inactivated by irradiation. Splenic B cells, enriched using EasyStep mouse B cell isolation kit (Stem Cell Technologies), were plated onto the 40LB cells cultured in IMDM medium (supplemented with 10% FBS and 2-mercaptoethanol; Thermo Fisher Scientific) and stimulated with recombinant IL-4 (1 ng/ml), IL-21 (10 ng/ml), R848 (1 µg/ml), and/or LPS (1 µg/ml) for 4 d. At day 4, stimulated B cells were transferred onto fresh 40LB cells and cultured in new IMDM media with indicated stimulations. For inhibitor experiments, vehicle (DMOS), idelalisib (100 nM; Selleck Chemicals), or rapamycin (0.5 nM; Selleck Chemicals) was added to the cultures 48 h before analysis. For serum starvation experiments, cultures were either received full serum (10%) or serum deprived (0%) media. Cells were analyzed by flow cytometry at indicated time points.

500 cells (from four biological replicates for each genotype, where each replicate represents one mouse) were sorted directly into reaction buffer from the SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (Takara), and reverse transcription was performed, followed by PCR amplification to generate full-length amplified cDNA. Sequencing libraries were constructed using the NexteraXT DNA sample preparation kit (Illumina) to generate Illumina-compatible barcoded libraries. Libraries were pooled and quantified using a Qubit Fluorometer (Life Technologies). Dual-index, single-read sequencing of pooled libraries was performed on a HiSeq2500 sequencer (Illumina) with 58-base reads using HiSeq v4 Cluster and SBS kits (Illumina) with a target depth of 5 million reads per sample. Base calls were processed to FASTQs on BaseSpace (Illumina), and a base call quality-trimming step was applied to remove low-confidence base calls from the ends of reads. The FASTQs were aligned to the mouse reference genome using STAR v.2.4.2a and gene counts were generated using htseq-count. Quality control and metrics analysis was performed using the Picard family of tools (v1.134).

Differential expression was analyzed for each cell type by DESeq2. Genes with an adjusted P value of < 0.05 were called as significantly differentially expressed. GSEA was performed using the GSEA software from the Broad Institute. Genes were ranked based on −log(padj)*log2FoldChange from the DESeq2 package and compared with GSEA-defined gene sets. RNA-seq datasets were deposited in the Gene Expression Omnibus (accession no. GSE171795)

CD138⁺ PCs were sorted, gently pelleted, and resuspended in 1/2 strength Karnovsky’s EM fixative (2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4) overnight at 4°C. Cells were rinsed by pelleting and resuspending with several changes of 0.1 M cacodylate buffer and postfixed with 1% osmium tetroxide (EMS Diasum) for 90 min at 4°C. The samples were rinsed several times with cacodylate buffer and dehydrated through a graded series of ethanols. Cell pellets were incubated in a 1:1 epon 12 resin (Ted Pella) to ethanol mixture overnight on a rotator and exchanged with two fresh resin changes the following day. A final resin change was made and allowed to polymerize overnight in a 60°C oven. 70-nm ultrathin sections were obtained with a Leica UC7 ultramicrotome, put onto 150-mesh carbon-coated grids, and stained with uranyl acetate and lead citrate. Sections were examined with an FEI TF20 transmission electron microscopy (TEM) operated at 200 kV and imaged with a high-angle annular dark-field scanning TEM (HAADF STEM) detector. Images were inverted to bright field using ImageJ.

Total RNA was isolated from FACS-sorted PCs using RNeasy Mini Kit (QIAGEN) and converted into cDNA by reverse transcription (Maxima Reverse transcription; Thermo Fisher Scientific) according to the manufacturer’s instructions. Real-time PCR was performed using a CFX96 Real-Time PCR Detection System with iTaq Universal SYBR Green Supermix (Bio-Rad). Primers used were as follows: ATF4, forward 5′-GCC​TGA​CTC​TGC​TGC​TTA​CAT​TAC-3′, reverse 5′-CCC​TTG​CCT​TAC​GGA​CCT​CTT​C-3′; ATF6, forward 5′-CAG​AGG​CAG​CAC​ACG​CAT​TC-3′, reverse 5′-GGT​CAG​CAG​GAG​CAG​AGA​AGG-3′; Blimp-1, forward 5′-GTG​TGG​TAT​TGT​CGG​GAC​TTT​G-3′, reverse 5′-GGA​CAC​TCT​TTG​GGT​AGA​GTT​C-3′; tXBP1, forward 5′-AAG​AAC​ACG​CTT​GGG​AAT​GG-3′, reverse 5′-ACT​CCC​CTT​GGC​CTC​CAC-3′; sXBP1, forward 5′-GAG​TCC​GCA​GCA​GGT​G-3′, reverse 5′-GTG​TCA​GAG​TCC​ATG​GGA-3′; PDI, forward 5′-TAG​CAA​AGG​TGG​ATG​CCA​CA-3′, reverse 5′-CAC​CAT​ACT​GCT​GAG​CCA​GGT-3′; BiP, forward 5′-TAT​TGG​AGG​TGG​GCA​AAC​CAA​G-3′, reverse 5′-CGC​TGG​GCA​TCA​TTG​AAG​TAA​G-3′; histone H3, forward 5′-GTG​AAG​AAA​CCT​CAT​CGT​TAC​AGG​CCT​GGT-3′, reverse 5′-CTG​CAA​AGC​ACC​AAT​AGC​TGC​ACT​CTG​GAA-3′. PCR was performed with the following conditions: denaturation at 95°C for 5 min; amplification, 40 cycles at 95°C for 10 s, 60°C for 30 s, and 72°C for 30 s. All reactions were standardized to the housekeeping gene histone H3.

FACS-sorted B cells were seeded onto poly-L-lysine–coated glass coverslips (BD BioCoat) in complete RMPI 1640 and allowed to attach for 1 h at 37°C. After plating, cells were fixed with 4% paraformaldehyde for 20 min at RT and permeabilized with saponin (0.2% saponin in PBS, 0.03 M sucrose, and 1% BSA) for 10 min at RT. Cells were blocked to prevent nonspecific binding by incubating for 1 h at RT with blocking buffer (2% goat serum, 1% BSA, 0.1% cold fish skin gelatin, 0.1% saponin, and 0.05% Tween-20 in 0.01 M PBS, pH 7.2). Cells were incubated overnight at 4°C with primary antibody (LC3, 1:500 dilution) in dilution buffer (PBS, 0.05% Tween-20, 1% BSA, and 0.1% saponin). Cells were washed three times in dilution buffer (5 min each time), incubated in secondary antibodies (1:500 dilution) in dilution buffer for 1 h at RT, and washed three times in dilution buffer. Cell nuclei were stained with Hoechst 33342 and washed twice with PBS, and the coverslips were mounted with ProLong Glass Antifade (Invitrogen). Cells were imaged on a ×63 oil objectives (aperture 1.4), and images were captured on a Leica TCS SP5 confocal microscope. Postacquisition analysis, such as contrast adjustment and z-slice stacking, were performed using FIJI software, and quantification of LC3 puncta was performed using 3D Object Counter plugin from FIJI. LC3 puncta were quantified by defining LC3-staining patterns as distributed around the nucleus and as punctate. The volume of LC3 punctate was obtained by setting 85–95 intensity threshold and 3–60 voxels unit size, and the percentage of LC3 punctate–positive cells was obtained by counting total nucleus and nucleus surrounded with LC3 puncta.

Sorted cells were lysed for 30 min on ice in radioimmunoprecipitation assay buffer (Thermo Fisher Scientific) supplemented with phosphatase and protease inhibitor cocktail and dithiothreitol. Lysates were centrifuged for 10 min at 4°C at 14,000 g and supernatant was collected. Proteins were separated by electrophoresis using NuPage-Bis-Tris gels (Invitrogen) and blotted onto polyvinylidene fluoride membranes. Nonspecific binding was blocked by 5% BSA in TBS-Tween followed by incubation with LC3 primary antibody (1:2,000 dilution) overnight at 4°C and secondary antibody HRP-conjugated antibody (1:5,000 dilution) for 1 h at RT. Membranes were thoroughly washed with TBS-Tween after antibody incubations and developed using ECL reagents (Millipore). For reprobing, blots were stripped for 10 min at RT with Restore PLUS stripping buffer (Thermo Fisher Scientific). Blots were then incubated with β-actin primary antibody (1:5,000 dilution) and secondary antibody HRP-conjugated antibody (1:5,000 dilution) for 1 h at RT, washed with TBS-Tween, and developed using ECL reagents (Cytiva Amersham). ImageJ software was used for band densitometry.

Statistical significance was determined with Prism 9 (GraphPad) software Version 8.0/9.0 using two-tailed unpaired Student’s t test or ANOVA multiple comparison as indicated in each figure. The P values were considered significant when P < 0.05 (*), P < 0.01 (**), P < 0.001 (***), and P < 0.0001 (****). No statistical methods were used to predetermine sample size.

Fig. S1 shows the gating strategy and characterization of VLP-specific MBCs in CD21-aPIK3CD and control mice. Fig. S2 shows the GC response in CD21-aPIK3CD and control mice 14 d after VLP-ssRNA immunization. Fig. S3 demonstrates the GC response in BM chimeras after VLP-ssRNA immunization. The figure also shows reduced PC generation in vitro from CD21-aPIK3CD B cells following stimulation with IL-4 + IL-21, LPS + IL-4, and LPS + IL-21. The figure also demonstrates reduced levels of NP-specific IgG1 antibodies in AID-aPIK3CD and Cγ1-aPIK3CD mice after NP-CGG immunization. Fig. S4 shows a similar GC response in Blimp1-aPIK3CD and control mice. It also shows the gene expression analysis of naive and GC B cells sorted from AID-WT and AID-aPIK3CD mice after immunization. Fig. S5 depicts elevated levels of pAKT and pS6 in CD21-aPIK3CD B cells and PCs. The figure also shows no impact of aPIK3CD on survival of activated B cells.

We thank V. Gersuk and the Benaroya Research Institute Genomics Core for help with RNA-seq; S. MacFarlane and P. Guo at the Fred Hutchinson Cancer Research Center Cellular Imaging Shared Resource and E. Lavoie at the University of Washington Molecular Analysis Facility for assistance with TEM and image analysis; D. Kitamura at Tokyo University of Science for kindly providing CD40L stromal cells and sharing protocol and R. Pitner for initiating and maintaining 40LB cultures in our laboratory; and S. Khim and A. Zielinska-Kwiatkowska for mouse colony maintenance.

This work was supported by the King Abdulaziz City for Science and Technology (Saudi Arabia) Graduate Scholarship Program, and the Department of Immunology, University of Washington (F. Al Qureshah) and National Institutes of Health awards DP3-DK111802 (D.J. Rawlings) and R01CA201135 (R.G. James). Additional support was provided by the Children's Guild Association Endowed Chair in Pediatric Immunology (D.J. Rawlings), the Tom Hansen Investigator in Pediatric Innovation Endowment (D.J. Rawlings), the Benaroya Family Gift Fund (D.J. Rawlings), and the Seattle Children’s Research Institute, Program for Cell and Gene Therapy. EM studies were supported by the Cellular Imaging Shared Resource of the Fred Hutchinson Cancer Research Center/University of Washington Cancer Consortium (P30 CA015704).

Author contributions: F. Al Qureshah designed and performed experiments, analyzed data, generated figures, and wrote the manuscript. S. Sagadiev, S. Liu, and M. Acharya performed and analyzed immunohistochemistry and confocal microscopy experiments. C.D. Thouvenel analyzed RNA-seq data. Z.H. and B. Hou provided VLP reagents and intellectual input into experimental design and interpretation. R.G. James and D.J. Rawlings designed and supervised the study, interpreted data, and wrote the manuscript.