The mTORC1/4E-BP/eIF4E Axis Promotes Antibody Class Switching in B Lymphocytes

During a response to infection, B cells become activated to produce different isotypes of Abs with varying effector functions (1). Early in the immune response, some activated B cells differentiate into plasmablast cells that mainly secrete low-affinity IgM Abs. Others become germinal center (GC) B cells and through T cell–dependent interactions undergo class switch recombination (CSR) to produce other classes of Abs, including IgG, IgA, and IgE. During the GC reaction, they can also undergo somatic hypermutation to diversify and produce higher affinity Abs. The resulting B cells that survive selection will then either become plasma cells that secrete these Abs to fight off the infection or become long-lived memory B cells that initiate a faster response during a second infection.

Class switching is initiated when the BCR recognizes Ag, and B cells are further stimulated through CD40 and cytokine receptors. All of these signals activate the mammalian (also known as mechanistic) target of rapamycin (mTOR). This ubiquitously expressed serine/threonine kinase integrates receptor and nutrient signals to promote many cellular processes, including mRNA translation, lipid biogenesis, and nucleotide synthesis. The mTOR kinase forms two complexes, mTOR complex 1 (mTORC1) defined by the Raptor subunit and mTORC2 defined by the Rictor subunit. A major function of mTORC2 is to phosphorylate AKT, a kinase that promotes survival and metabolic reprogramming. mTORC1 is activated downstream of PI3K and AKT and phosphorylates many substrates to promote biosynthetic pathways that support cell growth. Key mTORC1 substrates include S6 kinases and a family of mRNA translation inhibitors known as eIF4E-binding proteins (4E-BPs). Phosphorylation of 4E-BPs by mTORC1 leads to the formation of the eIF4F translational initiation protein complex composed of the cap-binding protein eIF4E, eIF4A helicase, and eIF4G scaffold that promote cap-dependent mRNA translation.

The immunosuppressive drug rapamycin (Rap), which allosterically binds and inhibits mTORC1 formation, has long been known to be a potent inhibitor of B cell Ab production (2). Because of the profound antiproliferative effect of Rap on both B and T cells, it has been difficult to uncouple the roles of mTORC1 in B cell proliferation and differentiation. However, several studies have now presented conclusive evidence that mTORC1 activity is dynamically regulated in GC B cells (3, 4) and that mTORC1 has a B cell–intrinsic role to promote Ab class switching from IgM to IgG and other isotypes (5–8). Although these findings highlight the important role of mTORC1 in B cell differentiation, the mechanism by which mTORC1 promotes class switching has not been addressed. Specifically, a key question is which mTORC1 downstream effectors control B cell commitment to isotype switching. Mechanistic investigation of eIF4E activity using both genetic and pharmacological tools shows that inhibiting eIF4E decreases Ab class switching and activation-induced cytidine deaminase (AID) protein. Our findings suggest that cap-dependent translation plays a role in class-switched Ab production and is a novel mechanism of regulating AID and B cell differentiation.

C57BL/6J (B6) mice were bred at the University of California, Irvine and used at between 6 and 12 wk of age. All animals were studied in compliance with protocols approved by the Institutional Animal Care and Use Committee of the University of California, Irvine. Mice carrying an AID-GFP reporter on a B6 background were obtained from The Jackson Laboratory (stock number 018421). Mice harboring a transgenic allele encoding a constitutively active form of 4E-BP (4E-BP1^M) under a tetracycline-responsive element were described previously (9, 10). These mice were crossed to a strain harboring an optimized form of rtTA (rtTA-M2) inserted downstream of the Rosa26 promoter, which was purchased from The Jackson Laboratory (stock number 006965). MLN0128 and MK-2206 were purchased from Active Biochem. Rap was purchased from Cell Signaling Technology. S6K1 inhibitor LY2584702 was purchased from Tocris Bioscience. SBI-756 was synthesized as described (11). Inhibitors were included throughout the indicated cell treatment periods.

Splenic B cells were purified by negative selection (eBioscience MagniSort Mouse B cell Enrichment Kit). B cell purity measured by FACS analysis (FACSCalibur and CellQuest software; BD Biosciences) using anti-B220 Ab (BioLegend) yielded >95% B220⁺. Purified B cells were seeded at a final concentration of 1.5 × 10⁶ cells/ml. For class switching experiments, B cells were stimulated with 5 μg/ml anti-CD40 (HM40-3) agonistic Ab (eBioscience) or 5 μg/ml LPS (Sigma-Aldrich), together with 5 ng/ml mouse IL-4 (R&D Systems) for 96 h to induce switching to IgG1. Treatment with 5 μg/ml LPS (Sigma-Aldrich) induced switching to IgG3. Treatment with 5 μg/ml LPS (Sigma-Aldrich) with 5 ng/ml mIL-4 (R&D Systems), 5 ng/ml mIL-5 (Tonbo Biosciences), 5 nM retinoic acid (Sigma-Aldrich), and 5 ng/ml TGF-β (Thermo Fisher Scientific) induced switching to IgA. All cells were cultured with RPMI 1640 supplemented with 10% (v/v) heat-inactivated FCS, 5 mM Hepes, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μM 2-ME.

Phospho-AKT S473 (Cell Signaling Technology), phospho-S6 S244/240 (Cell Signaling Technology), phospho-4E-BP-1 T37/46 (Cell Signaling Technology), and AID (eBioscience) protein on western blots were analyzed using ImageJ to quantitate signal of each band. Signal was normalized with actin measurements, and fold change was calculated using the stimulated/no drug control.

Mice were immunized i.p. with 100 μl of a final count of 5 × 10⁸ washed SRBCs (Innovative Research) and sacrificed 8 d after immunization. Mice were immunized with 100 μl of NP(18)-LPS (9) (Biosearch Technologies) mixed at a 1:1 ratio with Imject alum (Pierce Biotechnology) at a final concentration of 0.05 mg/ml and sacrificed on day 10. Mice were injected i.p. with 100 μl PBS vehicle, 100 μl at 1 mg/kg, and 0.3 mg/kg or 0.075 mg/kg Rap. Blood was collected for serum, and spleen was harvested for measuring GC and T follicular helper cells.

Serum SRBC Ab production was measured as described in (12).

For ELISA to measure NP-IgM, Nunc MaxiSorp plates (Thermo Fisher Scientific) were coated with anti-NP at 10 μg/ml in 50 μl of total sample in PBS and allowed to incubate overnight at 4°C. Serum was harvested from mice and diluted 1:100 in 2% (w/v) BSA in PBS and incubated on coated plates for 1 h at 37°C. HRP-conjugated rabbit anti-mouse IgM secondary Ab (Zymed Laboratories) was used.

Cells were incubated with TruStain FcX in FACS buffer (0.5% BSA plus 0.02% NaN₃ in 1× HBSS) for 10 min on ice. Staining with Abs was performed with FACS buffer and on ice for 20 min. Flow cytometry Abs and other reagents used are as follows: B220 (RA3-6B2; eBioscience), IgG1 (A85-1; BD Biosciences), IgG3 (R40-82; BD Biosciences), IgA (Cl0-3; BD Biosciences), IgD (11-26c.2a; BioLegend), and 7-aminoactinomycin D (Invitrogen). CFSE or eFluor 670–labeling of B cells was performed by resuspending cells to a concentration of 10 × 10⁶ cells/ml with a concentration of 2.5 μM. Flow cytometric data were analyzed using FlowJo software (Tree Star).

eIF4E–eIF4G interactions were detected in situ using the Duolink detection kit (Sigma-Aldrich) according to manufacturer instructions. Mouse anti-eIF4E (BD Biosciences) and rabbit anti-eIF4G (Cell Signaling Technology) primary mAbs were used. Confocal images were acquired on a Leica SP8 fluorescence microscope using the UV laser for DAPI and an excitation laser of 561. Proximity ligation assay (PLA) and DAPI signal areas were measured by ImageJ. Fold change relative to vehicle was quantified with PLA signal divided by DAPI signal.

Cap-dependent translation was measured using the luciferase reporter construct pRSTF-CVB3 containing the 5′ noncoding region of the Coxsackie B3 virus cloned between the firefly and Renilla luciferase (13). The construct was electroporated in serum-free medium and cells recovered in complete RPMI 1640 medium for 2 h followed by inhibitor treatment for 8 h. Following treatment, cells were lysed, and Renilla and firefly luciferase expression were measured using the Dual luciferase assay kit (Promega) using a luminometer. Renilla luciferase expression was normalized to firefly luciferase expression, and results were expressed relative to vehicle treated.

RNA was extracted using TRIzol reagent (Invitrogen), followed by purification with Quick-RNA MiniPrep according to the manufacturer’s protocol (Zymo Research). For Aicda, purified B cells were stimulated with αCD40 plus IL-4, LPS plus IL-4, or LPS for 72 h. For germline transcripts Iγ-Cγ, Iγ3-Cγ3, and Iα-Cα, B cells were stimulated for either 48 h or 72 h before harvesting. cDNA synthesis was carried out on 300 ng to 1 μg of total RNA using iScript Reverse Transcription Supermix (Bio-Rad Laboratories) for reverse transcription–quantitative PCR (qPCR). Gene expression of Aicda or germline transcripts was performed using iTaq Universal SYBR Green Supermix (Bio-Rad Laboratories), appropriate primers, and StepOnePlus Real-Time PCR System by Applied Biosystems. Mouse L32 was used as the housekeeping gene. Ratio of gene expression change was determined by the ΔΔCt method.

The following primers were used for qPCR: Aicda forward primer 5′-TGCTACGTGGTGAAGAGGAG-3′ and reverse primer 5′-TCCCAGTCTGAGATGTAGCG-3′; mL32 forward primer 5′-AAGCGAAACTGGCGGAAAC-3′ and reverse primer 5′-TAACCGATGTTGGGCATCAG-3′; Iγ-Cγ forward primer 5′-TCGAGAAGCCTGAGGAATGTG 3′and reverse primer 5′- GGATCCAGAGTTCCAGGTCACT-3′; Iγ3-Cγ3 forward primer 5′- GAGGTGGCCAGAGGAGCAAGAT-3′ and reverse primer 5′- AGCCAGGGACCAAGGGATAGAC-3′; and Iα-Cα forward primer 5′-CAAGAAGGAGAAGGTGATTCAG-3′ and reverse primer 5′-GAGCTGGTGGGAGTGTCAGTG-3′.

Mouse splenic B cells were stimulated with LPS plus IL-4 for 48 h and treated with inhibitors for 4 h before harvesting. Cells were then fixed in 4% PFA and stained with anti-B220 PE and rabbit anti-FOXO (Cell Signaling Technology) primary and anti-rabbit secondary. Cells were then stained with Hoechst before collection. Samples were analyzed using Amnis ImageStream Mark II Imaging Flow Cytometer (MilliporeSigma). FOXO nuclear localization was analyzed using the IDEAS software onboard Nuclear Localization Wizard algorithm. The ratio of nuclear to cytoplasmic localization was determined and expressed as fold change to vehicle.

Ab class switching occurs after activated B cells commit to proliferation (14–17). Because Rap inhibits B cell proliferation, further studies were required to establish a separable role for mTORC1 in switching. Experiments in which mice were treated with low doses of Rap after viral infection provided evidence that partial mTORC1 inhibition selectively affects class switching while preserving or enhancing IgM responses that also require B cell clonal expansion (6, 8). In accord, we found that low-dose Rap suppressed IgG1 production and GC B cell percentage without reducing IgM responses in mice immunized with the T cell–dependent Ag SRBCs (Fig. 1A–C) or the T cell–independent Ag NP-LPS (Fig. 1D–F). Another study found that deletion of Raptor in B cells postactivation (using Cγ1-Cre) suppresses switching to IgG1 while preserving IgM responses in vivo (7). Together, these findings support the conclusion that mTORC1 promotes class switching in a manner independent from proliferation. This is an important distinction because it may be possible to develop inhibitors of mTORC1 downstream effectors to selectively modulate B cell differentiation without the broad immunosuppressive effects of Rap.

We sought to apply in vitro methods to study the B cell–intrinsic effects of different genetic and pharmacological perturbations on Ab class switching. Stimulation of purified B cells in vitro with αCD40 plus IL-4 or LPS plus IL-4 mimics T cell–dependent or T cell–independent signals (18, 19), respectively, to induce switching to IgG1. By labeling B cells with the cell division tracker CFSE prior to activation, it is possible to control for differences in proliferation by gating on divided cells (CFSE-lo). In B cells stimulated with αCD40 plus IL-4, Rap caused a dose-dependent decrease in switching to IgG1 among divided cells (Fig. 2A). We used the cyclin-dependent kinase inhibitor palbociclib to suppress proliferation (Fig. 2A) without affecting mTORC1 signaling (Fig. 2B,). Notably, palbociclib did not reduce the percentage of IgG1⁺ among the residual dividing cells (Fig. 2A). These data confirm that the ability of Rap to suppress switching is not due simply to slower proliferation.

To identify conditions in which complete mTORC1 inhibition does not block proliferation, we tested different timepoints at which we added 20 nM Rap or 50 nM MLN0128 (a dual mTORC1/mTORC2 inhibitor). We stimulated splenic B cells with αCD40 plus IL-4 or LPS plus IL-4 and added the inhibitors at the same time as activation (0-h addition), 24 h after activation (24-h addition), or 48 h after activation (48-h addition). The 0-h addition completely blocked division as expected, whereas 24-h addition partly reduced division, and 48-h addition had no effect (Supplemental Fig. 1A). Western blot confirmed that 48 h addition Rap reduced p-S6 and partially inhibited 4E-BP2 phosphorylation, whereas the mTOR kinase inhibitor MLN0128 completely reduced S6, 4E-BP1, 4E-BP2, and AKT phosphorylation (Supplemental Fig. 1B). Nevertheless, the B cells were still able to divide and seemed to have committed to cell cycle progression as seen by cyclin D2 expression; 0-h Rap addition decreased cyclin D2 measured 4 h after activation, whereas Rap added at 48 h did not affect ongoing cyclin D2 expression (Supplemental Fig. 1C). The 48-h addition of Rap significantly decreased the percentage of IgG1⁺ in divided cells compared with 0.1% DMSO control (Fig. 2C, Supplemental Fig. 1D); similar effects were observed over a range of LPS and IL-4 concentrations (data not shown). The 48-h addition of MLN0128 did not affect the percentage of IgG1⁺ (Supplemental Fig. 1D), probably because of opposing effects of mTORC1 and mTORC2 inhibition on CSR (5). Over multiple experiments, measurement of proliferation index showed that 48-h addition of Rap had a significant but minor effect on the proliferation of B cells stimulated with αCD40 plus IL-4 or LPS plus IL-4 (Supplemental Fig. 2A). Moreover, 48-h addition of Rap reduced the percentage of IgG1⁺ in each division in cells stimulated with αCD40 plus IL-4 (Supplemental Fig. 2B). Focusing on cells in the third division showed significantly reduced percentage of IgG1⁺ in the Rap-treated condition (Supplemental Fig. 2B).

We also tested other stimuli that promote switching to IgG3 (LPS alone) or IgA (LPS plus IL-4 plus IL-5 plus TGF-β plus Retinoic Acid; henceforth referred to as IgA stimuli). The 48-h Rap addition reduced the percentage of IgG3⁺ in B cells activated with LPS but did not significantly decrease switching to IgA (Fig. 2C). To control for changes in proliferation efficiency, we also measured IgG1⁺ among cells that had divided three times, as determined by CFSE dilution (Supplemental Fig. 2B). This analysis revealed similar trends with 48-h addition of Rap reducing switching under IgG1 and IgG3 conditions. Using this analysis approach, switching efficiency to IgA was slightly reduced (Supplemental Fig. 2C). Overall, this suggests that the role of mTORC1 in switching is confined to a subset of isotypes but not selective to only IgG1.

To exclude the possibility that the lower percentage of IgG⁺ cells was due to reduced translation of Ig mRNAs, we measured the surface expression of IgG1 (determined by mean fluorescence intensity [MFI]) in cells that did switch following Rap treatment. Notably, the MFI was not lower than the vehicle-treated cells (121.1 ± 40.6 versus 109 ± 34.1) (Fig. 2C). We also assessed the percentage of nonswitched B cells (IgD⁺) and found that 0-h and 48-h addition of Rap increased the percentage of IgD⁺ cells following αCD40 plus IL-4, LPS plus IL-4, or LPS but not IgA stimulation (Fig. 2D). This suggests that Rap reduces switching overall rather than rerouting the switch to other Ab isotypes.

Because mTORC1 is a sensor of environmental inputs, including nutrients and growth factors, we wanted to test the effect of physiological mTORC1 inhibition by nutrient restriction. Glutamine is used as an energy source by activated lymphocytes (20) and is also required for import of essential amino acids that activate mTORC1 (21). B cells require glutamine to proliferate and undergo Ig synthesis (22); however, the effect of glutamine restriction on Ab class switching has not been tested. Typical lymphocyte cell culture medium contains 4 mM of glutamine. We found that 1 and 2 mM glutamine still preserved proliferation and class switching (percentage of IgG1⁺) (Fig. 3A, 3B). Complete glutamine withdrawal strongly reduced proliferation and percentage of IgG1. As glutamine concentration increased to 0.4 mM, proliferation was mostly restored, whereas switching to IgG1 was still reduced, similarly to low concentrations of Rap. B cells stimulated in 0.25 mM glutamine had reduced mTORC1 activity as measured by pS6 and cell size (Fig. 3C, 3D). Gating on cells in the third division confirmed that decreasing glutamine impairs switching to IgG1 among cells that have divided similarly (Fig. 3E). Overall, these results demonstrated that chemical (Rap) or physiological (glutamine deprivation) inhibition of mTORC1 suppresses class switching independently from proliferation.

Two major mTORC1 downstream effectors are known to regulate gene expression: S6Ks and the 4E-BP/eIF4E axis. We assessed the role of these pathways in Ig class switching. To test the effects of S6K inhibition, we treated B cells with an S6K1-selective inhibitor (S6K1i) and confirmed reduced phosphorylation of the direct S6K substrate, ribosomal protein S6 (Supplemental Fig. 3A). S6K1 inhibition also caused increased p-AKT that frequently occurs because of loss of feedback (Fig. 3C). S6K1 inhibition significantly reduced switching to IgG1 following αCD40 plus IL-4 or LPS plus IL-4 stimulation but had no effect in LPS (IgG3) or IgA-stimulating conditions when measured in total divided cells or in division 3 (Supplemental Fig. 3B, 3C). These results suggest that S6K1 activity contributes to mTORC1-driven switching under some conditions. Interestingly, S6K1 inhibition partially blocked proliferation with α-CD40 plus IL-4 but not in cells stimulated with LPS plus IL-4 (Supplemental Fig. 3D).

Next, we investigated the role of the 4E-BPs, a family of mTORC1 substrates that in nonphosphorylated forms compete with eIF4G for the same binding domain of eIF4E and inhibit eIF4F complex formation. We previously showed that Rap inhibits eIF4E function in lymphocytes more than other cell types, in part because of expression of the 4E-BP2 isoform that is more Rap sensitive than 4E-BP1 [(9), Supplemental Fig. 1B]. Thus, to test the role of eIF4E in class switching, we used transgenic mice in which a mutant form of 4E-BP1 (4E-BP1M) can be inducibly expressed by doxycycline (DOX) treatment. The 4E-BP1M protein has five S/T sites changed to alanine, such that the protein can no longer be phosphorylated by mTORC1. 4E-BP1M binds more avidly to eIF4E and inhibits initiation of cap-dependent translation regardless of mTORC1 activity (Fig. 4A). When 4E-BP1M is induced prior to B cell activation, proliferation is strongly suppressed (9). Therefore, we used a late addition of DOX (48 h) and confirmed 4E-BP1M protein expression by Western blotting (Fig. 4B). In B cells stimulated with αCD40 plus IL-4 or LPS plus IL-4 and treated at 48-h addition of increasing DOX concentrations, 4E-BP1M reduced the percentage of IgG1⁺ without reducing proliferation (Fig. 4C–F). 4E-BP1M expression also reduced the percentage of IgG3⁺ but not the percentage of IgA⁺ (Fig. 4D, 4E, Table I).

To complement this genetic model, we tested the compound SBI-756, which binds the scaffolding protein eIF4G and prevents its interaction with cap-binding protein eIF4E in melanoma cells (11). In B cells stimulated with LPS plus IL-4, 150 nM SBI-756 partially reduced switching to IgG1, although proliferation was mainly preserved (Fig. 5A). As SBI-756 concentration was increased, switching and proliferation were reduced further (Fig. 5A). We also tested 48-h addition of SBI-756 (Fig. 5B). Again, switching to IgG1 or IgG3 but not IgA was significantly reduced (Fig. 5C, 5D, Table I). SBI-756 did slightly reduce cell accumulation under LPS plus IL-4 and IgA-stimulating conditions; however, the number of cell divisions was similar under all conditions (Fig. 5E). Hence, direct inhibition of eIF4F function does not affect ongoing B cell proliferation yet specifically inhibits Ig class switching. This selective effect is consistent with previous studies showing that 4E-BP/eIF4E axis affects the translation of subsets of specific mRNAs in a cell context-specific manner (23–25).

To confirm that SBI-756 disrupts the cap-binding complex and to assess the effect of Rap, we used an in situ PLA (26). Analysis by confocal microscopy allows visualization and quantitation of the interactions between eIF4E and eIF4G (Fig. 6A) in activated B cells. We stimulated B cells with LPS plus IL-4 and used the mTOR kinase inhibitor, MLN0128, as a control because this compound strongly disrupts eIF4E interaction with eIF4G (9). MLN0128 (100 nM), Rap (2 or 20 nM), and SBI-756 (300 nM) all significantly reduced eIF4E:eIF4G association compared with DMSO vehicle control (Fig. 6A). In activated B cells treated with Rap or SBI-756 at 48 h and tested 4 h later, we observed trends of reduced PLA signal that were not statistically significant (data not shown). To increase sensitivity, we performed a dual luciferase reporter assay using a bicistronic dual Renilla–firefly luciferase reporter construct. This construct measures cap-dependent translation by synthesis of the Renilla luciferase as well as cap-independent, polio IRES-mediated translation by synthesis of firefly luciferase as an internal control. In B cells activated with LPS plus IL-4, 48 h addition of Rap or SBI-756 significantly reduced cap-dependent translation by 30–40% compared with vehicle (Fig. 6B).

Overall, these results suggest that cap-dependent translation is necessary for CSR. To identify at which step of the recombination event the mTORC1/eIF4E axis is involved, we first investigated the production of germline transcripts. When a B cell is activated, secondary signals such as IL-4, TGF-β, and IFN-γ can direct the class of Ab to switch by promoting production of noncoding sterile RNAs (germline transcripts) encoded by the DNA of the accepter switch region. During transcription elongation of the region, the chromatin becomes more accessible for the CSR machinery to facilitate recombination to the targeted switch region (27). Using RT-PCR, we measured germline transcripts Iγ-Cγ, Iγ3-Cγ3, and Iα-Cα, which are produced from IgG1, IgG3, and IgA switch regions, respectively. As expected, Iγ-Cγ was detected under αCD40 plus IL-4– or LPS plus IL-4–stimulating conditions but not under IgG3 (LPS only) or IgA conditions (Supplemental Fig. 4A). Similarly, Iγ3-Cγ3 and Iα-Cα were detected only under IgG3– or IgA–switching conditions, respectively. In all conditions, 0-h addition of Rap strongly inhibited germline transcript production (Supplemental Fig. 4A), consistent with a study of Raptor-deficient B cells (7). S6K1 inhibition reduced Iγ-Cγ transcripts in B cells stimulated with LPS plus IL-4 (Supplemental Fig. 3E), suggesting a novel role for S6K1 in mTORC1-driven germline transcription under these conditions. However, the S6K1 inhibitor did not reduce germline transcripts under αCD40 plus IL-4–, IgG3-, or IgA-switching conditions (Supplemental Fig. 3E). We also tested 48-h addition of Rap or SBI-756, which only reduced germline transcript production in LPS plus IL-4 and IgA-stimulating conditions but not with αCD40 plus IL-4 or LPS (Supplemental Fig. 4B). These results did not follow the same trends as observed in our flow-based class switching experiments (Table I). This lack of correlation suggests that germline transcript production is not the mechanism by which mTORC1 activity and cap-dependent translation regulates class switching at the 48-h timepoint.

Class switching requires AID, a DNA mutator protein that initiates CSR by introducing mutations into the targeted switch regions of the IgH locus (28, 29). Western blotting (Fig. 7A) showed that Rap added at 0-h or 48-h timepoints after LPS plus IL-4 stimulation significantly reduced AID protein amounts. We used a GFP knock-in reporter of AID (AID-GFP mice) (30) to measure AID protein amounts using flow cytometry. Under all stimulating conditions, we found that 0-h addition of a low concentration of Rap (0.2 nM) reduced AID expression per cell and that 48-h addition of Rap or SBI-756 also significantly reduced expression (Fig. 7B, 7C, Table I). The AKT inhibitor MK-2206 was used as a control to increase AID expression (5).

Reduced AID expression under IgA conditions was surprising given that 48 h addition of Rap or SBI-756 did not affect switching to IgA. We tested the effect of adding Rap at earlier timepoints (24, 40, and 44 h) and found that switching to IgA was minimally affected even when Rap reduced proliferation (data not shown). However, Rap did reduce the percentage of IgG3⁺ in the same experiment, suggesting that the reduction in AID mainly affects switching to other isotypes under the IgA-stimulating conditions (data not shown).

To determine if AID expression is regulated posttranscriptionally in activated B cells, we compared changes to Aicda mRNA and AID protein. Under LPS plus IL-4–stimulating conditions, 0-h addition of Rap strongly reduced Aicda mRNA (Fig. 7D), consistent with studies of B cells lacking Raptor (7). However, 48-h Rap addition had a lesser effect and did not reduce Aicda mRNA under αCD40 plus IL-4– or LPS-stimulating conditions. Notably, 48-h addition of SBI-756 did not reduce Aicda mRNA under any switching conditions. Considering that Foxo proteins promote Aicda transcription (31) and Rap treatment can lead to increased AKT activity, we used ImageStream analysis to determine whether Rap reduced nuclear localization of Foxo1 in activated B cells (Supplemental Fig. 4C). Instead, we found that 48-h addition of either Rap or SBI-756 had the opposite effect, increasing the ratio of nuclear/cytoplasmic Foxo1 (Supplemental Fig. 4D). Together, these data suggest that the reduction in AID protein is not due to a decrease in Aicda transcription.

Following the initial discovery of Rap as a potent immunosuppressant over 40 y ago (32), this natural product was found to inhibit proliferation and impair differentiation of both T and B lymphocytes (2, 33, 34). High doses of Rap profoundly impair Ab production (35), yet low doses have a selective effect on class switched Ab responses (6–8). mTORC1 is dynamically regulated in GCs (3) and responsive to environmental conditions such as hypoxia (4); furthermore, genetic deletion of Raptor has established that the mTORC1 complex has a B cell–intrinsic role in Ab class switching (7). However, the mechanisms by which mTORC1 regulates class switching remain unknown. In this study, we show that Rap reduces eIF4E and eIF4G association to inhibit cap-dependent translation in activated B cells and that directly inhibiting eIF4E activity can also reduce Ab class switching. We also show that the mechanism involves posttranscriptional inhibition of AID protein expression, which is a novel mechanism of regulating AID and B cell differentiation.

We reported previously that the 4E-BP/eIF4E axis is Rap sensitive in lymphocytes and required for both growth and proliferation following Ag receptor stimulation (9). Supporting the role of eIF4E in B cell proliferation, we found that 0-h addition of SBI-756 suppresses B cell proliferation and eIF4E:eIF4G association. However, mTORC1/eIF4E activity is not necessary for proliferation 48 h after activation. The finding that ongoing proliferation is mTORC1-independent is consistent with early studies showing lesser effects of Rap when added at late timepoints after B or T cell activation (2, 36). More recent work has confirmed this in vivo by showing that mTORC1 activity is not necessary to drive B cell proliferation during the dark zone stages of the GC (3). Because mTORC1 is active in light zone B cells (3) and required for IgG1 switching in vivo (7), it will be interesting to see how eIF4E signaling regulates GC differentiation. We tested SBI-756 in the SRBC immunization model and did not observe any differences in IgG1 secretion or GC B cell percentages (data not shown). This might be due to the pharmacokinetics of SBI-756 in vivo, which might not facilitate sustained inhibition of the eIF4E:eIF4G interaction. Further studies with genetic models in which mutations disrupt eIF4G and eIF4E binding will be useful to address this question and to study B cell–intrinsic effects.

We found that 48-h addition of Rap or SBI-756 reduced switching to IgG3 and IgG1 but did not reduce IgA switching although AID protein was reduced. One possible explanation is that the cells switching to IgA are enriched in marginal zone and B-1 cells (37) and that mTORC1 might not regulate AID and switching in these subsets. Marginal zone cells make up ∼10% of B cells purified from the spleen and perhaps the overall AID being produced is outweighed by the reduction of AID in the follicular B cells.

B cell activation is accompanied by rapid and sustained activation of S6 kinases (38); however, little is known about the role of S6Ks in B cell function. Consistent with our previous work (9), we show in this study that S6K1 inhibition has minimal effect on proliferation of B cells stimulated with αIgM plus IL-4. However, S6K1 inhibition did partially reduce proliferation induced by αCD40 plus IL-4 and diminished switching to IgG1. Although S6K1 inhibition may be partially responsible for the effects of early Rap addition on Ab class switching, additional studies are required to better understand how S6K1 responds under the different stimuli to promote switching or proliferation.

Our findings demonstrate that the mTORC1/eIF4E axis plays a novel role in B cell differentiation. We and others have previously published that PI3K/AKT signals are involved in B cell differentiation and transcription of AID (5, 39, 40). However, the predominant role of PI3K/AKT in activated B cells is to oppose CSR as determined by treating cells with PI3K/AKT inhibitors or by deleting PTEN (5, 39, 40). The uncoupling of PI3K/AKT and mTORC1 function in B cell differentiation may be due in part to mTORC1 receiving signals from other inputs (38), along with the independent and opposing role of mTORC2/AKT/FOXO in CSR (5, 31). The finding that AID expression can be regulated posttranscriptionally by mTORC1 illustrates the need for better understanding and investigation of the translatome during B cell differentiation and how different signals control translation efficiency of different mRNAs. The 4E-BP/eIF4E axis downstream of mTORC1 has been studied extensively in cancer and shown to preferentially promote translation of mRNAs involved in metastasis and invasion, and growing evidence supports a key role for regulated translation in CD4 and CD8 T cell differentiation (41, 42). However, which mRNAs in activated B cells are regulated by translation efficiency remains an important question in the field. Additional experiments are required to establish whether AID and other proteins involved in CSR are regulated at the level of translation. Cap-dependent translation is an attractive target for cancer therapy and efforts are underway to develop small-molecule inhibitors of eIF4F components (eIF4E, eIF4G, eIF4A) for oncology (11, 26, 43–45). Further study of this pathway may provide mechanistic insight into how the immune system will be affected by eIF4F-targeted therapies. Indeed, the finding that the eIF4G antagonist SBI-756 reduces AID protein suggests that it may be useful as a combination therapy with idelalisib, a PI3Kδ inhibitor, which increases AID expression and mutational activity in activated B cells and in human patients with chronic lymphocytic leukemia (46). Further studies are needed to determine whether eIF4F inhibitors suppress AID expression in human B leukemia and lymphoma cells.

Our findings emphasize that 4E-BP/eIF4E can regulate expression of important proteins necessary for B cell differentiation. Study of this pathway may provide mechanistic insight into Ab-mediated autoimmune diseases such as lupus and arthritis as well as the formation of protective Ab responses following vaccination.

We thank Bert Semler for the pRSTF-CVB3 luciferase reporter plasmid, Jennifer Atwood for training on Amnis ImageStream, Egest Pone for helpful discussions, and Ivan Topisirovic for insightful comments on the manuscript.

D.A.F. received research support from Revolution Medicines. D.R. is a shareholder of eFFECTOR Therapeutics, and an uncompensated member of the Scientific Advisory Board. The other authors have no financial conflicts of interest.