A plasma cell differentiation quality control ablates B cell clones with biallelic Ig rearrangements and truncated Ig production

We previously showed in normal B lineage cells that upon skipping of some noncoding V exons, so called excluded alleles often yield shortened κ mRNAs which would translate into ΔV-κLCs (Chemin et al., 2010). Observing such shortened mRNAs is surprising because the production of truncated Ig is usually considered a specific feature of some lymphoproliferative diseases (Cogné et al., 1992). To extend this study, we further analyzed the conditions of exon skipping, checked whether shortened Igκ transcripts were translated into ΔV-κLCs, and addressed whether the latter affects B cell differentiation. Out-of-frame Vκ-to-Jκ rearrangements result in nonsense codons affecting either the 3′-end of the V exon (V^PTC) or the C exon (C^PTC; Delpy et al., 2004a; Fig. 1 A). We compared ΔV-κLC mRNA levels in cell lines transfected with constructs mimicking nonproductive VκJκ5 rearrangements from both the V^PTC and the C^PTC class (Fig. 1, B and C). This revealed high ΔV-κLC mRNAs in V^PTC-expressing cells, whereas such transcripts were barely detectable in C^PTC-expressing cells and absent in cell lines expressing productive Igκ genes (Fig. 1 D). ΔV-κLC mRNA levels were quantified using capillary electrophoresis after normalization to the amount of productive κLC mRNAs, as previously described (Chemin et al., 2010). Compared with C^PTC-expressing cells, ΔV-κLC mRNAs were ∼3.5- and 6.7-fold higher in V^PTC-expressing B (A20) and PC (S194) lines, respectively (Fig. 1 E). Hence, skipping of the V exon occurs mostly when this exon itself carries the PTC, and with an efficiency approximately two times higher in PCs than in B cells.

Next, we checked whether ΔV-κLC mRNAs were indeed translated into of V domain-less truncated Ig, thus expected to mostly show up in V^PTC-expressing PCs. Consistent with data in Fig. 1, complete κLCs (∼25 kD) and truncated ΔV-κLCs (∼12 kD) were detected in cells transfected with productive and V^PTC constructs, respectively (Fig. 2 A). In contrast, no ΔV-κLCs (or a background level similar to nonsecretory Sp2/0 cells) was observed in untreated C^PTC-expressing cells (Fig. 2 A). We further explored whether ΔV-κLCs were routed toward proteasomal degradation. In V^PTC-expressing cells, treatment with MG132 induced an approximately two- to four-fold increase of ΔV-κLCs in cell pellets and culture supernatants, respectively (Fig. 2, A and B). In contrast in C^PTC-expressing cells, very low amounts of ΔV-κLCs were found in cell pellets, and only upon treatment with MG132 (Fig. 2 A). For controls, no major changes were observed for complete κLCs (Fig. 2, A and B).

The survival of PCs depends on their ability to alleviate the ER stress response associated with massive Ig synthesis (Cenci et al., 2011). PC survival might thus be impacted by the production of proteins with abnormal structures and folding. To address this issue, we analyzed the apoptotic index of S194 PCs with regard to the abundance of ΔV-κLCs. Interestingly, V^PTC-expressing cells exhibit higher apoptosis than nontransfected and C^PTC-expressing cells, upon treatment with MG132 (Fig. 2 C). In addition, mRNA levels of C/EBP homologous protein (Chop), a transcription factor involved in ER-stress induced apoptosis (Obeng et al., 2006), were approximately two-fold higher in V^PTC-expressing cells than in nontransfected cells upon treatment with MG132 (Fig. 2 D). As expected, Chop expression in C^PTC-expressing cells was intermediate, in between nontransfected and V^PTC-expressing cells (Fig. 2 D). Therefore, the presence of ΔV-κLCs likely provokes excessive ER-stress that sensitizes plasma cells to CHOP-dependent apoptosis upon proteasome inhibition.

Differential sensitivity to apoptosis can hardly be followed in vivo due to immediate phagocytosis of apoptotic cells. However, such differences would be expected to result in a bias in the selection of terminally differentiated cells, thus providing indirect arguments for a TIE checkpoint. To address this issue, we analyzed the Igκ repertoire at the DNA level in B cells and PCs from WT animals. We determined CDR3 lengths and distinguished in-frame (F or VJ⁺) V^PTC and C^PTC VκJκ junctions (Fig. 1 A). In agreement with previous works (Chen et al., 1993; Ehlich et al., 1993; Novobrantseva et al., 1999; Bertocci et al., 2003; Delpy et al., 2004a), VκJκ5 junctions were readily detectable in precursor B cells that did not express a BCR (pro–B/pre–B cells: B220^lo/IgL^neg). As expected, numerous (∼66.6%) out-of-frame junctions were retrieved in the absence of BCR-mediated selection, with an equivalent repartition of V^PTC and C^PTC subclasses (Table 1). In contrast, the percentage of F junctions was much higher in GC B cells (B220^hi/PNA^hi) having been selected for functional BCR expression, with a reciprocal fall of V^PTC and C^PTC subclasses (Table 1). Unexpectedly, the frequency of V^PTC junctions still more drastically decreased in PCs (B220^neg/CD138^pos; approximately nine-fold lower than GC B cells; P < 0.01), whereas that of C^PTC junctions was unchanged (Table 1). Next, we addressed whether the disappearance of PCs harboring V^PTC-rearranged Igκ alleles prone to exon-skipping correlated with a decrease in ΔV-κLC mRNAs. ΔV-κLC mRNA levels were analyzed by RT-PCR in sorted B cell populations and PCs (Fig. 3 A). Although, a short band corresponding to ΔV-κLC mRNAs was readily detectable in B cell populations, these alternative transcripts were not retrieved in PCs (Fig. 3, B and C). The complete loss of ΔV-κLC mRNAs in PCs was consistent with the drastic disappearance of V^PTC junctions in their repertoire. Altogether, the selective elimination of PCs with biallelic Igκ rearrangements in VJ⁺/V^PTC configuration, likely reflects toxic effects of ΔV-κLCs. This suggests that, independently of B cell functionality and specificity for antigen, a late TIE checkpoint counterselects B lineage cells producing aberrant Ig and prevents their survival as terminally differentiated PCs.

According to previous observations showing that ∼40–50% of B lymphocytes exhibit biallelic V(D)J rearrangements of Ig genes (Mostoslavsky et al., 2004; Daly et al., 2007), approximately half of these B cell clones (∼20–25%), i.e., those harboring V^PTC-rearranged Igκ alleles, are candidates for an Ag-independent TIE checkpoint, eventually leading to their elimination at the step of PC differentiation. However, the rarity of V^PTC junctions in PCs at the DNA level is only an indirect proof that cells go through such a checkpoint. To directly address the impact of ΔV-κLCs on late B cell development and on humoral responses, we created Igκ knock-in mice allowing inducible expression of ΔV-κLCs in B lineage cells (Fig. 4 A). In these mutants, later referred as inducible TIE (iTIE) mice, the replacement of endogenous Jκ segments by a leader exon (L_1-33) under the control of a pV_H promoter led to enforced expression of ΔV-κLC mRNAs and proteins. Conditional Cre-mediated expression of ΔV-κLC mRNAs was achieved by inserting a loxP-flanked neomycin-resistance (NeoR) cassette that blocks transcription and splicing of the L_1-33 exon (Fig. 4 A). In heterozygous (iTIE/+) animals, the WT allele (+) supports normal Igκ expression during B cell development, whereas the knock-in (iTIE) allele allows Cre-dependent expression of ΔV-κLCs. After mating iTIE/+ mice with a CMV-Cre–expressing strain, high ΔV-κLC mRNA levels and significant truncated Ig synthesis were observed in Cre-positive (Cre^pos) B lineage cells, with no major leakiness in Cre-negative (Cre^neg) controls (Fig. 4 B). In agreement with the aforementioned data from transfected cells (Fig. 2), ΔV-κLC amounts increased after treatment with the proteasome inhibitor bortezomib (Bz; Fig. 4 B). Interestingly, these knock-in mice mimic the outcome of the physiological TIE checkpoint, as the production of ΔV-κLCs severely impairs PC differentiation without perturbing B cell development and GC formation (Fig. 4, C and D). In controls, the sole expression of Cre-recombinase affected neither total B220^pos and GC B cells nor PCs (unpublished data). Consistent with a global defect in PC differentiation, Cre^pos iTIE/+ mice exhibited significantly lower T cell–independent and T cell–dependent antibody responses than Cre^neg iTIE/+ controls (Fig. 4, E and F). These findings were further confirmed in competitive experiments mimicking the occurrence of the TIE checkpoint during late B cell differentiation and in only a fraction of terminally differentiated PCs. For that, we analyzed B cell and PC contents after mating iTIE mice with AID-Cre-EYFP mice, allowing conditional Cre-recombinase expression in GC B cells upon treatment with tamoxifen (Dogan et al., 2009). We found significantly lower amounts of EYFP-positive PCs in AID-Cre-EYFP iTIE/+ mice, compared with AID-Cre-EYFP +/+ littermates (Fig. 5, A and B). In contrast, similar frequency and absolute numbers of EYFP-positive B cells were retrieved in these animals (Fig. 5, C and D). Thus, the PC defect observed in Cre^pos iTIE mice confirms that the production of ΔV-κLCs induces deleterious effects in terminally differentiated B lineage cells.

Next, we sought to elucidate how these truncated Ig control plasma cell maturation using the iTIE mouse model with enforced ΔV-κLC expression. Experiments were performed in homozygous iTIE/iTIE animals that have no functional Igκ alleles and exhibit only Igλ-expressing B lineage cells. Like heterozygous mutant mice (Cre^pos iTIE/+), but to a higher extent, homozygous animals (Cre^pos iTIE/iTIE) exhibited very low amounts of B220^negCD138^pos PCs (Fig. 6 A) together with a strong reduction in serum Ig levels (Fig. 6 B), compared with Cre^neg controls. ΔV-κLC mRNA amounts were determined by qPCR in B cells and PCs isolated from bone marrow of Cre^pos iTIE/iTIE mice. We found that the remaining PCs exhibited approximately five-fold higher ΔV-κLC mRNA levels compared with sorted B cells (Fig. 6 C). To evaluate whether ΔV-κLC production in PCs affected Ig synthesis, we analyzed the expression of Igλ in Cre^pos and Cre^neg iTIE/iTIE B220^negCD138^pos PCs. Interestingly, we found an approximately two-fold lower mean fluorescence intensity (MFI) for intracellular Igλ in Cre^pos iTIE/iTIE PCs, compared with Cre^neg iTIE/iTIE counterparts (Fig. 6 D). Hence, PCs expressing ΔV-κLCs synthesized low levels of Ig. Next, we evaluated whether the truncated Ig impaired the renewal or survival of PCs in Cre^pos and Cre^neg iTIE/iTIE mice injected with BrdU. Similar amounts of BrdU^neg and BrdU^pos B cells were retrieved for both strains (Fig. 6 E). In contrast, the frequency of BrdU^pos (cycling) B220^negCD138^pos cells was significantly increased in Cre^pos iTIE mice compared with Cre^neg controls, indicating higher proportion of recently divided plasmablasts (PBs) upon production of ΔV-κLCs. Inversely, we found lower amounts of noncycling BrdU^neg PCs in Cre^pos iTIE mice. Next, we sought to elucidate how the production of truncated Ig limited the lifespan of PCs. We analyzed the expression of ER stress markers, together with unfolded protein response (UPR) components in sorted PCs (B220^negCD138^pos). In agreement with an exacerbated ER stress response upon production of ΔV-κLCs, Chop, homocysteine-induced ER protein (Herp), X-box binding protein 1 spliced (Xbp1s), and BiP mRNA levels were strongly increased in PCs isolated from Cre^pos iTIE mice (Fig. 6 F). Thus, the rise of ER stress as a result of the production of ΔV-κLCs is transiently tolerated in early differentiating PBs but clearly compromises their survival as noncycling long-lived PCs secreting high Ig amounts (Nutt et al., 2015).

ΔV-κLCs can induce multiple deleterious effects. On one hand, their presence could impede the normal assembly of IgH and IgL chains and relate to the classical quality control of BCR assembly (Melchers et al., 2000). On the other hand, the TIE checkpoint could involve an intrinsic toxicity of truncated Ig independent of BCR expression. To distinguish between these two hypotheses, we determined whether the elimination of PCs containing V^PTC-rearranged Igκ alleles also occurred in the absence of IgH chain expression. Igκ repertoires were thus compared between B cells (B220^pos) and PCs (B220^neg/CD138^pos) isolated from spleens of DH-LMP2A mice (Lechouane et al., 2013). In this model, the Epstein-Barr virus LMP2A protein mimics the BCR tonic signal and B lymphocytes develop without any IgH chain (Casola et al., 2004). As expected upon such a by-pass of BCR-driven positive selection, approximately two thirds of VκJκ5 junctions were out-of-frame in DH-LMP2A B cells, with an equivalent repartition of V^PTC and C^PTC subclasses (Table 2). Regarding nonproductive junctions, we found again that the V^PTC class was drastically decreased in PCs, compared with B cells (Table 2). In addition, we confirmed that the sole presence of ΔV-κLCs impaired PC differentiation by breeding iTIE mice on a DH-LMP2A background (Fig. 7 A). Thus, the elimination of PCs containing V^PTC-rearranged Igκ alleles occurred independent of IgH expression and BCR assembly. According to an exacerbated ER stress response and activation of the UPR, CHOP, HERP, BiP, Xbp1s, and IRE1α expression were strongly increased upon production of ΔV-κLCs (Fig. 7). Likewise, high ΔV-κLC mRNA levels were correlated with a strong expression of ER stress markers, in PBs sorted 4 d after LPS stimulation (Fig. 7, D and E). Altogether, these findings indicate that the TIE checkpoint is unrelated to a disturbed IgH/IgL assembly, but rather to a toxic ER stress provoked by the sole presence of ΔV-κLCs.

The allelic exclusion of Ig genes is stringently established before or during early B cell maturation and ensures the monospecificity of B lymphocytes through stepwise V(D)J recombination of Ig alleles. In cells harboring biallelic V(D)J rearrangements, the noise coming from transcription of nonproductive alleles has long been neglected because these transcripts are either degraded by NMD or do not encode functional Ig chains (Jäck et al., 1989; Li and Wilkinson, 1998; Delpy et al., 2004a; Chemin et al., 2010; Tinguely et al., 2012). However, we report here that alternative splicing of these passenger transcripts leads to the production of aberrant Ig chains that seriously impede PC differentiation. These findings identify a novel TIE checkpoint that eliminates ∼20–25% of terminally differentiated B lineage cells, i.e., those transcribing nonproductive Igκ alleles prone to exon skipping. Consistent with an Ag-independent checkpoint, ΔV-κLCs exert intrinsic toxic effects in PCs, likely mediated by ER stress–induced apoptosis. In addition, PCs expressing ΔV-κLCs synthesize low amounts of Ig and are mostly found among short-lived PBs. We propose that the TIE-checkpoint favors the selection of long-lived PCs with limited basal ER stress supporting high levels of Ig secretion.

Regarding nonproductively rearranged Igκ alleles, V^PTC and C^PTC junctions are highly similar and often exhibit a single nucleotide difference within the CDR3 sequences. Despite minor changes at the DNA level, we found that exon skipping was only enhanced for V^PTC class transcripts, with higher magnitude in PCs than in mature B cells. Although the molecular mechanisms underlying NAS are not completely understood, this response is triggered by two distinct pathways referred as class I and class II NAS (Maquat, 2002; Wang et al., 2002; Chang et al., 2007). Class I NAS occurs upon disruption of splicing motifs, such as exonic/intronic splicing enhancers/silencers, by either nonsense or missense mutations, whereas class II NAS is a reading frame–dependent process strictly induced by nonsense mutations (Chang et al., 2007). Thus, our data strongly suggest that a class II NAS response occurs for nonproductive Igκ transcripts that further relies on PTC recognition within the skipped V exon, but not within the downstream Cκ exon. We also observed an approximately two-fold increase in ΔV-κLC mRNA levels in S194 plasma cells, compared with A20 B cells. In PCs, Ig genes are localized in transcription factories near the nuclear pore, and these specialized areas authorize cooperation of enhancers and facilitate Ig gene transcription (Park et al., 2014). In addition, a rise of exon skipping has been observed for highly transcribed genes, leading to the assumption that alternative splicing is closely correlated to the rate of RNA polymerase II elongation (Nogués et al., 2003; Shukla and Oberdoerffer, 2012). Altogether, we suggest that the boost of Ig gene transcription accompanying PC differentiation detrimentally promotes exon skipping during splicing of nonproductive Igκ premRNAs.

Biallelic Igκ rearrangements with a VJ⁺/V^PTC configuration represent around 20–25% of mature B cells (Mostoslavsky et al., 2004) and hereby yield the population affected by the TIE checkpoint in terminally differentiated PCs. Sensitivity to prolonged ER stress likely explains the elimination of PCs producing ΔV-κLCs. Accordingly, ER stress markers and UPR components such as CHOP, BIP, HERP, and XBP1/IRE1α are up-regulated upon production of ΔV-κLCs in PCs. In addition, we found that Chop mRNA levels were already elevated in B cells expressing truncated Ig, suggesting that the early expression of this proapoptotic factor could influence the fate of differentiating PCs. Although a previous study showed that CHOP is dispensable for the development of LPS-induced plasmablasts (Masciarelli et al., 2010), our data are in agreement with a recent work highlighting its role in differentiating PCs (Gaudette et al., 2014) and also demonstrating that BCL-XL expression protects PCs from CHOP-associated apoptosis. The iTIE mouse model described herein should be useful to decipher the complex interplay between pro- and anti-apoptotic factors during PC differentiation.

With regard to PC dyscrasias, truncated Ig have been observed in some cases of myeloma (Cogné et al., 1988, 1992; Cogné and Guglielmi, 1993). However, it is a rare feature of primary PC dyscrasias and of malignant PC lines, suggesting that the TIE checkpoint also shapes the human Ig repertoire and that PCs producing aberrant Ig molecules are rarely rescued by oncogenic events (Decourt et al., 2004). Accordingly, we found that ΔV-κLCs were harmful for survival of normal PCs but exerted no obvious effect when expressed in tumor PCs in the absence of PI treatment. The survival of normal and malignant PCs is influenced by ER stress and the balance between load versus capacity of the proteasome (Cenci and Sitia, 2007). Being able to modulate ER stress and proteasome activity is of considerable interest for multiple myeloma (MM) patients (Meister et al., 2007) and new therapeutic strategies have emerged combining inhibitors of chaperones (Hsp90: Heat shock protein 90) or factors involved in unfolded protein response (IRE1α) with PI (Ishii et al., 2012; Mimura et al., 2012). Our findings are consistent with these combined approaches, increasing the amount of aberrant proteins while decreasing their proteasomal degradation, and suggest that splicing modulators or NMD inhibitors could also be useful tools to reinforce the production of truncated proteins including ΔV-κLCs.

Interestingly, the late TIE checkpoint stands as a quality control of individual and unassembled ΔV-κLCs and is not a result of a disturbance of BCR or secreted Ig assembly (which could then be viewed as a failed Ag response). Thus, the late occurrence of the TIE checkpoint, at the terminal differentiation stage, can be considered an intrinsic waste and weakness of the immune system. It allows uncensored B cells to accumulate as mature B cells carrying V^PTC-rearranged Igκ alleles, to potentially respond to Ag, and undergo affinity maturation in GCs, while being unable to yield long-lived PCs secreting high amounts of Ig (Nutt et al., 2015). The TIE checkpoint then cuts off the diversity of PCs arising from Ag-stimulated B cells for reasons unrelated to Ag recognition and affinity.

Whereas NMD basically protects cells from truncated protein synthesis, we show that activation of NAS herein exerts an opposite effect, finally ending with loss of ΔV-κLC–expressing PCs. This study thus shows that transcription and splicing of nonproductive Ig alleles, although often neglected or considered to be under control, are in fact often inappropriately handled by RNA surveillance pathways, allows production of truncated Ig and ends with a PC wastage.

Homologous recombination at the Igκ light chain locus was performed as described previously (Sirac et al., 2006). In brief, a 12.7-kb BamHI genomic fragment corresponding to the germline mouse JκCκ cluster was used to generate the iTIE targeting construct (Van Ness et al., 1982). A 2.2-kb BsmI–SacII fragment spanning all the Jκ segments was replaced with a cassette containing a V_H promoter (pV_H), a human leader exon (L_1-33), and a loxP-flanked neomycin-resistance gene driven by the Herpes simplex thymidine kinase promoter (hsvTk-neoR). The hsvTk-neoR cassette was inserted in opposite orientation to block the transcription and/or splicing of the L_1-33 exon (Fig. 4 A). Mouse embryonic stem cells (E14) were transfected with the linearized vector by electroporation and selected using 300 µg/ml Geneticin and 2 µM ganciclovir.

2–3-mo-old mice were used in all experiments and maintained in our animal facilities, at 21–23°C with a 12-h light/dark cycle. Experiments were performed according to the guidelines of the ethics committee in Animal Experimentation of Limousin (registered by the National Committee under the number C2EA-33) and were approved as part of the protocol registered under the number CREEAL 6–07-2012. Heterozygous mutant mice (iTIE/+) were backcrossed to C57BL/6 (B6) for at least three generations, and then mated with Cre-expressing mice to induce deletion of the hsvTk-neoR cassette. B6 CMV-Cre mice were obtained from the Mouse Clinical Institute (Illkirch, France). B6 AID-Cre-EYFP mice, harboring a tamoxifen-inducible Cre recombinase enzyme controlled by the Aicda (activation-induced cytidine deaminase) promoter and a loxP-flanked EYFP (enhanced yellow fluorescent protein) reporter gene, have been described elsewhere (Dogan et al., 2009). B6 AID-Cre-EYFP mice were obtained from the Imagine Institute (Paris, France). In DH-LMP2A mice (with mixed B6/BALB/c backgrounds), the Epstein-Barr virus LMP2A protein drives B cell development and plasma cell differentiation, as previously described (Casola et al., 2004; Lechouane et al., 2013). DH-LMP2A mice were obtained from the Institute of Molecular Oncology Foundation (Milano, Italy).

A20, S194, and Sp2/0 murine cell lines were cultured (10⁶ cells/ml) in RPMI supplemented with 10% fetal calf serum (Invitrogen), sodium pyruvate, nonessential amino acids, β-mercaptoethanol, 100 U/ml penicillin, and 100 µg/ml streptomycin (Gibco). Cells (2 × 10⁶) were stably transfected by electroporation according to the manufacturer’s instructions (Amaxa). Transfections were performed using in-frame (F), V^PTC, and C^PTC linear plasmid constructs separately, or using equimolar ratios of F and nonproductive constructs as previously described (Chemin et al., 2010). VκJκ5 junctions were built by artificially joining the Vκ_1-135 and Jκ5 segments. To create a reading frameshift, 5 and 10 additional nt were introduced at the VκJκ junction in V^PTC and C^PTC constructs, respectively (Fig. 1 B). Cells (10⁶ cells/ml) were treated or not with MG132 (1 µg/ml) for 5 or 8 h, as indicated, and control cells were treated with DMSO alone (dilution factor 1/1,000).

The frequency of apoptotic cells was determined after staining with Annexin V (BD) and propidium iodide using a LSRII Fortessa (BD). Data were analyzed with FACSDiva software (BD). Bone marrow B cell precursors and PCs were sorted on a FACSVantage (BD) after staining with anti–mouse B220 (RA3-6B2; BioLegend), anti–mouse Igκ (187–1; Beckman Coulter), anti–mouse Igλ (JC5-1; Beckman Coulter), and anti–mouse CD138 (281–2; BD) mAbs. The gates used for cell sorting are indicated in Fig. 3 A. GC B cells were isolated from Peyer’s patches after staining with anti–mouse B220 and PNA (peanut agglutinin; Sigma-Aldrich) as previously described (Delpy et al., 2004b). In all experiments, purity of sorted cells was >90%. B cells were also isolated from spleens by negative selection using anti-CD43 microbeads (Miltenyi Biotec) and stimulated (0.5 × 10⁶ cells/ml) with 1 µg/ml of LPS (LPS-EB Ultrapure; InvivoGen) for 4 d.

To determine the frequency of GC B cells and PCs, erythrocyte-depleted spleen cells were stained with anti-mouse GL7 (GL7; BD), anti-B220, and anti-CD138 mAbs, 7 d after intraperitoneal (IP) injection of sheep red blood cells (SRBCs; bioMérieux). When indicated, mice received additional subcutaneous bortezomib (Sillag) injections at day 5 and 6 (0.5 mg/kg). To induce nuclear translocation of Cre-recombinase, AID-Cre-EYFP mice were treated with Tamoxifen (1 µg/mouse, IP; Sigma-Aldrich) at day 2, 4, and 6 after SRBC immunization.

To analyze the renewal of B lineage cells in bone marrow, mice received i.p. BrdU (Sigma-Aldrich) injections (1 mg/mouse at day 0, then 0.5 mg every 48 h) for 10 d. Cells were then stained according to the BrdU Flow kit (BD) protocol. Cells were treated with DNase I before staining with anti-BrdU mAb. Intracellular Igλ staining was performed in bone marrow cells isolated from homozygous iTIE/iTIE mice using the Cytofix/Cytoperm kit (BD).

For Western blot analysis, a 4–20% Mini-PROTEAN TGX polyacrylamide gel (Bio-Rad Laboratories) was used. Each sample was then denatured at 94°C for 3 min before being loaded. Gels were blotted onto TransBlot Turbo polyvinylidene fluoride membranes (Bio-Rad Laboratories), and blocked in PBST buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, and 0.1% Tween 20, pH 7.42) containing 5% nonfat dry milk. The signal was measured by chemoluminescence (ECL plus; GE Healthcare). Western blots were performed using rabbit anti-mouse HERP (Santa Cruz Biotechnology, Inc.), CHOP, BIP, IRE1α (Cell Signaling Technology), goat anti–mouse Igκ (Beckman Coulter), and using goat anti–mouse GAPDH (R&D Systems) Abs for normalization.

Ig titers were determined in culture supernatants and mouse sera using unlabeled and alkaline phosphatase–conjugated goat anti–mouse Abs (IgM, IgG1, IgG3, IgG2b, Igκ, Igλ, and total IgG; Southern Biotech) as described (Pinaud et al., 2001; Sirac et al., 2006). Total Ig amounts were revealed by the addition of p-nitrophenyl phosphate (Sigma-Aldrich), and plates were read at 405 nm.

To analyze T cell–dependent humoral responses, mice received two IP injections with 10 µg hen egg OVA (Sigma-Aldrich) emulsified in 50% CFA (day 0) and IFA (day 14). For T cell–independent responses, mice received two IP injections (days 0 and 14) with 50 µg of 2,4, Dinitrophenyl-Amino-Ethyl-Carboxy-Methyl-Ficoll (DNP-Ficoll; Biosearch Technologies) emulsified in 50% IFA. Blood samples were collected (day 7 and 24) and antigen-specific antibody titers were determined in polycarbonate 96 multiwell plates (Maxisorp; Nunc) coated overnight at 4°C with OVA or albumin-DNP (Sigma-Aldrich) solution (10 µg/ml), in carbonate buffer.

Genomic DNA and total RNA were prepared using standard proteinase K (Eurogentec) and Tri-reagent (Invitrogen) procedures, respectively. RT-PCR was performed on DNase I-treated (Invitrogen) RNA and was negative in the absence of reverse transcription, ruling out contamination by genomic DNA. Reverse transcription was performed using Superscript II (Invitrogen) on 1 to 5 µg of total RNA. Priming for reverse transcription was done with random hexamers.

Amplifications and capillary electrophoresis were performed as previously described (Chemin et al., 2010). Quantification of PCR products was also performed using an Agilent 2100 Bioanalyzer (Agilent Technologies) according to the Agilent High Sensitivity DNA kit instructions. Sequences of VκJκ5 junctions were analyzed after cloning of PCR products into pCR2.1-TOPO vector (Invitrogen), using V-QUEST software (IMGT, the international ImMunoGeneTics information system). Real-time PCR were performed on a ABI PRISM 7000 Sequence Detection System (Life Technologies). Transcripts were quantified according to the standard 2^−ΔΔCt method after normalization to Gapdh. Sequences of primers and probes are available upon request.

Results are expressed as mean ± SEM and overall differences between variables were evaluated by a two-tailed unpaired Student’s t test using Prism GraphPad software. A χ² test was done to analyze the distribution of F, V^PTC, and C^PTC VκJκ junctions.

We thank the staff of our animal facility, as well as C. Carrion and C. Ouk-Martin for technical assistance with microscopy and cell cytometry. We also thank K. Rajewsky (The Max Delbrück Center for Molecular Medicine, Berlin, Germany) and S. Casola (Institute of Molecular Oncology Foundation, Milano, Italy) for providing DH-LMP2A mice. AID-Cre-EYFP mice were kindly provided by CA Reynaud and JC Weill (Imagine Institute, Paris, France). We are grateful to J. Cook-Moreau (Centre National de la Recherche Scientifique UMR 7276, Limoges, France) for proofreading of the manuscript.

This work was supported by grants from Fondation ARC (#PGA120150202338 and SFI20121205821), Ligue Contre le Cancer (comité Haute-Vienne), Comité d'Organization de la Recherche sur le Cancer du Limousin (CORC), and Fondation pour la Recherche Médicale. N. Srour was funded by Conseil Régional du Limousin. A. Tinguely and G. Chemin were funded by French government fellowships.

The authors declare no competing financial interests.