Mechanism of CD79A and CD79B Support for IgM⁺ B Cell Fitness through B Cell Receptor Surface Expression Free

The development and function of B cells depend on signals from the BCR. The Ag-binding part of the BCR consists of two identical membrane-bound IgHs, each linked to identical IgL chains. The Ig molecule pairs with a signaling unit comprised of a transmembrane heterodimer of CD79A (Igα) and CD79B (Igβ). Upon Ag binding, BCR signaling is initiated by phosphorylation of the ITAMs of CD79A and CD79B by the SRC-family kinase LYN (1). Signaling is further relayed by recruitment and phosphorylation of SYK and activation of downstream signaling including phospholipase C (PLC)-γ, BLNK, and BTK. Deletion of any of the BCR components in mice, including IgH, CD79A, or CD79B, results in a block in B cell development in the bone marrow and absence of mature B cells (2–4). Signaling through CD79A or CD79B is needed in this developmental checkpoint, as replacement of the tyrosines in the ITAMs of CD79A and CD79B in mice leads to a block in B cell development (5). In mature mouse B cells, the presence of BCR on the cell surface provides a low-level constitutive Ag-independent survival signal called tonic signaling. This was first shown in conditional knockout (KO) mice where deletion of Igh or Cd79a in mature B cells depleted the mature B cell pool with a half-life of 3–6 d (6, 7). Mature BCR-negative B cells can be saved by constitutive active PI3K signaling or deletion of the phosphatase Pten, a negative regulator of AKT signaling and a key target of PI3K signaling (8). Primary B cells with high BCR expression have increased survival and elevated tonic signaling compared with cells with low BCR expression (9).

The BCR similarly has an essential role for human B cells. Immunodeficient patients with mutation in IGHM, CD79A, or CD79B do not have mature B cells (10–13), confirming the developmental checkpoint seen in mice. B cell malignancies arising from mature B cells usually retain BCR expression on the surface (14) and are addicted to tonic (15), chronic active (16, 17) or Ag-induced BCR signaling (18–20), hence making this pathway an important therapeutic target. CD79A and CD79B were identified as essential in prior CRISPR/Cas9 screens in lymphoma cell lines representing diffuse large B cell lymphoma (DLBCL) (21–23), the most common aggressive form of B cell non-Hodgkin’s lymphoma. The activated B cell–like subtype of DLBCL is characterized by chronic activation of NF-κB, induced by BCR engagement by self-antigens (17) or activating mutations in genes involved in the BCR pathway (16). Younger activated B cell–like DLBCL patients, especially with the MCD genetic subtype harboring mutations in CD79B and/or MYD88, show high response rates to the BTK inhibitor ibrutinib (24). Mantle cell lymphoma (MCL), a rarer form of aggressive B cell non-Hodgkin’s lymphoma that also shows activation of the BCR pathway and constitutive NF-κB signaling, has a high response rate to ibrutinib (25, 26) and next-generation BTK inhibitors acalabrutinib and zanubrutinib (27).

The germinal center (GC) B cell–like (GCB) subtype of DLBCL depends on tonic BCR signaling relayed via PI3K-mediated activation of AKT and inhibition of FOXO1, a positive regulator of apoptosis (15, 28). In cell lines representing GCB DLBCL, Y188F mutation in the ITAM of CD79A or KO of AKT is toxic, and BCR KO cells can be rescued by forced activation of AKT (15). Furthermore, silencing of FOXO1 in B cell lymphomas renders the cells insensitive to BCR signaling inhibitors (28). Burkitt’s lymphoma, which is a MYC-dependent malignancy, is outcompeted after induction of short hairpin RNA (shRNA) targeting CD79A, suggesting a dependency on tonic signaling (29). However, in a MYC-driven lymphoma mouse model, BCR loss did not kill the cells, but the competitiveness of BCR-negative cells was reduced due to rewiring of the metabolism and increased activity of GSK3β (30). Recently, the requirement of the CD79A/CD79B heterodimer for survival of human B cells was questioned, as the Burkitt’s lymphoma cell line Ramos could be rescued upon reintroduction of the CD79B wild-type (WT) in BCR KO cells (31). In this cell line, CD79B homodimers could associate with CD19 and provide an alternative survival signal when CD79A or IgM was lost (31).

Although it is generally accepted that CD79A and CD79B are needed for BCR assembly and expression, few studies have investigated this in detail using human B cells. Overexpression of BCR components in Drosophila S2 cells have shown that CD79A and CD79B are covalently bound by a disulfide bridge in the extracellular domain (32), and this CD79A/CD79B heterodimer is noncovalently linked to the Ig molecule through aromatic proline interactions in the transmembrane domain (33, 34). In a murine cell line, the CD79A/CD79B heterodimer is observed in a 1:1 stoichiometry to plasma-membrane Ig (35). Findings in the Burkitt’s lymphoma cell line Ramos suggest that the entire BCR complex must be assembled in the endoplasmic reticulum (ER) before transport to the cell membrane (36), a mechanism controlled by chaperone proteins of the ER quality control system that retains improperly folded proteins or improperly assembled protein complexes (37). For instance, in the absence of CD79A in a murine B cell line, the Igµ H chain is retained in the ER through interactions of polar amino acids in the transmembrane domain with the ER chaperone calnexin (38). Transport of protein complexes from the ER through Golgi stacks and trafficking to the plasma membrane depends on N-glycosylation in a well-regulated, multistep process (39). The progression of the BCR complex can be measured by the glycosylation pattern, as immature, pre-Golgi proteins are decorated with a high-mannose N-acetyl glucosamine core whereas mature, post-Golgi proteins have more complex N-glycans (40).

In this study, we aimed to systematically investigate the requirement of CD79A and CD79B for IgM membrane expression and B cell fitness. We used human tonsillar B cells to investigate the association between the level of membrane expression of IgM and protection against apoptosis. For mechanistic studies, we focused on human MCL cell lines as a model system, as lymphoma cells from MCL usually overexpress IgM as compared with normal B cells, which likely contributes to the pathogenesis of this lymphoma type (41). CRISPR/Cas9 editing was used to generate mutants lacking CD79A or CD79B, followed by monitoring of KO cells over time in a competition assay. CD79A and CD79B KO clones were generated to study their interdependence for protein maturation and for maturation of IgM.

Tonsils were obtained from patients undergoing tonsillectomy with written informed consent in accordance with the Declaration of Helsinki and the Regional Committees for Medical and Health Research Ethics, Region Eastern Norway (REK no. 2010/1147a). The tonsils were processed to single-cell suspension by mincing and then stored in liquid nitrogen. Before experiments, the cells were thawed and rested in RPMI 1640 with 10% FCS at 37°C for at least 15 min before counting and further processing. The B cell lines Z-138 (ATCC CRL-3001) and Granta-519 (DSMZ ACC 342) were cultured in RPMI 1640/10% FCS + 2 µM l-glutamine. MINO (DSMZ ACC 687) and SUDHL-6 (DMSZ ACC 572) were cultured in RPMI 1640/10% FCS. Cell line authentication was done by PCR single-locus technology for 21 independent PCR systems (Eurofins, Glostrup, Denmark).

Stable Cas9-expressing cell lines were made as previously described (42). Briefly, Z-138, MINO, SUDHL-6, and Granta-519 cells were transduced with retroviral particles expressing humanized WT Cas9 (MSCV_Cas9_puro was a gift from Christopher Vakoc [Addgene plasmid no. 65655]) (43) and then selected by puromycin at concentrations determined by a cytotoxicity assay (Invitrogen). Single-guide RNAs (sgRNAs) targeting CD79A and CD79B were designed using the online tool CHOPCHOP (44). The three sgRNAs with the highest scores were delivered into Cas9-expressing Granta-519 and MINO cells by electroporation using the ECM 830 square wave electroporation system (BTX). To check the genome editing efficiency, surface staining of CD79A and CD79B was performed 72 h after electroporation and the targeted loci were amplified by PCR and subjected to Sanger sequencing. The results were analyzed using the TIDE online software (45). The sgRNAs with the highest editing efficiencies (CD79A, 5′-GCACCACCGCGCAGAACAGG-3′ and CD79B, 5′-GACCTTTGGGATTCCGGTAC-3′) were selected for functional analyses. Single-cell clones were sorted by FACS (Sony SH800S cell sorter) and expanded for several weeks, before repeating surface staining and Sanger sequencing to enable selection of potential KO clones. Stable clones were obtained for Granta-519 (CD79A KO clone II-G6, homozygous 85-bp deletion; CD79B KO clone E4, homozygous 1-bp insertion) and Z-138 cells (CD79B KO clone B3, homozygous 2-bp deletion, and clone B1, complex alteration with three alleles: 1-bp deletion, 14-bp deletion, and 22-bp deletion [only used in Supplemental Fig. 2]). In addition, clones with in-frame 6-bp deletions in CD79B were obtained for MINO (clones B and D with identical alterations; 1-bp insertion and 6-bp deletion) and Z-138 (clone C7 and C9 with identical homozygous 6-bp deletions). Clones with a 6-bp deletion maintained surface IgM expression.

A population of naive and memory B cells was isolated from tonsils by bead depletion of CD3⁺ T cells and CD38⁺ GC B cells as previously described (46). Approximately 1 million naive/memory B cells were pelleted, total mRNA was extracted (RNeasy Plus kit, Qiagen), and cDNA was synthesized (QuantiTect kit, Qiagen). The CD79B coding sequence was amplified using 10 ng of cDNA and the following primers: CD79B forward, 5′-CAC CAT GGC CAG GCT GGC GTT GTC-3′, reverse, 5′-CTC TCT TGG CTC TCT CCT GGC CTG GGT GCT CAC-3′. The PCR conditions were the following 30 cycles (30 s at 94°C, 30 s at 53°C, and 2 min at 68°C) using Pfx polymerase (Thermo Fisher Scientific, Waltham, MA), generating a 690-bp product that contains the natural start and a replacement of the stop codon. In parallel, a GFP fragment was amplified using a 2A-GFP–containing template (CSK-2A-GFP) (47) with the primers 2Af_UNIVERSAL (5′-AGA GCC AAG AGA GGC AGC GGC G-3′) and pWS_GFPXI CTC (5′-GAG TTA CTT GTA CAG CTC GTC CAT GCC GAG-3′) to generate a GFP fragment containing a 2A coding sequence on its 5′ end. The two fragments were purified, mixed, and amplified using the CD79B forward and pWS_GFPXI primers in these conditions: 20 times for 1 min at 94°C, 1 min at 53°C, and 1 min 30 s at 68°C with Pfx. The amplicon was subsequently purified and subcloned into the pENTR/D-TOPO vector (Thermo Fisher Scientific) and sent to sequencing (MWG Eurofins, Erlangen, Germany). Validated sequences were further recombined into the retroviral pMP71-Gateway vector (48) using a Gateway recombinase system (Thermo Fisher Scientific). CD79B mutants were created by a Pfu-Turbo-based site-directed mutagenesis procedure using pENTR-CD79B-2A-GFP as a template and the following primers: G137S, 5′-GTC TAC CAG GGC TGC AGC ACA GAG CTG CGAG-3′ and 5′-CTC GCA GCT CTG TGC TGC AGC CCT GGT AGAC-3′; Y196N, 5′-GAA GAT CAC ACC AAC GAG GGC CTG GAC-3′ and 5′-GTC CAG GCC CTC GTT GGT GTG ATC TTC-3′; Y196C, 5′-GAA GAT CAC ACC TGC GAG GGC CTG GAC-3′ and 5′-GTC CAG GCC CTC GCA GGT GTG ATC TTC-3′; Y196F, 5′-GGA AGA TCA CAC CTT CGA GGG CCT GGAC-3′ and 5′-GTC CAG GCC CTC GAA GGT GTG ATC TTCC-3′. All mutants were sequence verified and further recombined into the pMP71-Gateway vector.

CD79B retroviral particles were prepared in HEK-Phoenix cells as previously described (48). Granta-519 CD79B KO cells were transduced by one cycle of spinoculation. Nontreated culture plates (Thermo Fisher Scientific) were coated overnight with RetroNectin (Takara, Shiga, Japan) at 4°C and blocked with PBS containing FCS at 2% final concentration for 30 min at room temperature. Wells were then washed and 1 ml of cells at 0.5 million/ml was added and incubated for 10 min. Then, 1 ml of retroviral supernatant was mixed, and the plates were spun at 750 × g for 1 h at 32°C. Cells were harvested after 18 h, washed in PBS, and grown in complete medium for another 48 h. Transgene presence was monitored by flow cytometry analysis of the GFP signal.

Tonsillar single-cell suspensions or B lymphoma cell lines were stained with Abs on ice for 30 min before analysis on a FACSCanto II or LSR II (BD Biosciences). The following Abs from BioLegend were used: CD20-Pacific Blue (clone 2H7), CD3-BV570 (clone UCHT1), CD38-PE-Cy7 (clone HIT2), CD79A-PE-Cy7 (clone HM47), and IgM-Pacific Blue (clone MHM-88). The following Abs from BD Biosciences were used: CD27-allophycocyanin (clone M-T271), CD27-BV605 and CD27-FITC (clone L128), CD3-V500 and CD3-Ax700 (clone UCHT1), CD79B-PerCP-Cy5.5 and CD79B-allophycocyanin (clone SN8/3A2-2E7), IgD-PerCP-Cy5.5 (clone IA6-2), and IgG-allophycocyanin-H7 (clone G18-145). In addition, we used CD3-PE (clone UCHT1) and polyclonal rabbit IgD (catalog no. R5112) from Dako, CD79A-PE (clone 706931) from R&D Systems, polyclonal goat IgA-PE (catalog no. 2052-09) from SouthernBiotech, and polyclonal goat IgG-FITC and IgM-FITC (catalog nos. AHI1308 and AHI 1608) from Invitrogen.

To induce BCR signaling, cell suspensions from tonsils were distributed at 10⁷ cells/ml and rested for an additional 15–30 min, then stimulated with 10 µg/ml anti-IgM F(ab′)₂ or IgG F(ab′)₂ (Invitrogen catalog nos. AHI1601 [discontinued] and AHI1301 or Jackson ImmunoResearch Laboratories catalog no. 309-006-043) for 4 min. Cell lines were stimulated with 20 µg/ml anti-IgM for 15 min as previously described (41, 49). Fixation was done by adding paraformaldehyde (PFA) to a 1.6% final concentration for 5 min at room temperature. In experiments where signaling was induced by FITC-labeled anti-IgM F(ab′)₂ (5 min; Invitrogen catalog no. AHI1608 [discontinued] or Jackson ImmunoResearch Laboratories catalog no. 309-096-043) or anti-IgG F(ab′)₂ (5 min; Invitrogen catalog no. AHI1308), the unstimulated cells were stained for 15 min, at room temperature, after fixation. Cells were washed with PBS and then permeabilized with ice-cold methanol (final concentration >90%) and stored at −20 or −80°C. At the day of flow cytometry analysis, cells were rehydrated by washing twice in PBS. For barcoding, cells were resuspended in PBS and incubated with different concentrations of Pacific Blue dye for 30 min in the dark. Ab staining (p-BLNK-Ax647 [clone J117-1278], p-BTK-Ax647 [clone 24a/BTK], p-BTK-Ax488 [clone N-35-88], p-LCK-Ax488 and p-LCK-PE [clone 4/LCK-Y505], p-PLC-γ-Ax647 and p-PLC-γ-Ax488 [clone K86-689.37], p-SFK-Ax647 [clone 4/LCK-Y505], p-SYK-Ax647, p-SYK-PE, and p-SYK-Ax488 [clone 17a/p-ZAP70], and CD20-PerCP-Cy5.5 [clone H1] from BD Biosciences and p-ERK-Ax647 [clone 197G2] from Cell Signaling Technologies) was performed at room temperature for 15–30 min before washing and running the samples on a FACS LSR II or a FACSCanto II (BD Biosciences). Cytobank software (www.cytobank.org) was used for data analysis, and relative phosphorylation changes were calculated using arcsinh transformation of mean fluorescence intensity of the cell population of interest.

Tonsillar cell suspensions were thawed, rested for 30 min, and distributed at 10⁷ cells/ml. Cells were either left unstained or stained for 5 min with anti-IgM-FITC, then left untreated or treated with FASL. As a live/dead cell stain, Pacific Blue dye was added at a final concentration of 0.1 ng/μl for the last 10 min of culture. After 3 h, cells were fixed by adding PFA (1.6%, 5 min, room temperature). The cells that were left unstained prior to culture were now stained with anti-IgM-FITC (15 min, room temperature), followed by permeabilization with ice-cold methanol. Cells were stored at −20°C until the day of flow cytometry analysis. Ab staining (cleaved caspase-3-Ax647, clone C92-605 from BD Biosciences) was performed at room temperature for 15–30 min before washing and running the samples on a FACS LSR II (BD Biosciences).

CD79B⁻IgM⁻ and CD79B⁺IgM⁺ MINO cells were sorted by FACS (BD FACSAria) on day 3 after transfection with CD79A or CD79B sgRNA, then cultured in separate wells for up to 7 d. Twenty-four hours after sorting, cells were incubated with Alexa Fluor 750 N-hydroxysuccinimide (NHS) ester (1 µg/ml) or a LIVE/DEAD fixable near-IR dead cell stain kit (Invitrogen) for 10 min as live/dead cell stain before fixation in PFA, washing, permeabilization, and staining with anti-cleaved caspase-3-PE (clone C92-605) from BD Biosciences and anti–p-GSK3β-Ax647 (clone D85E12) from Cell Signaling Technology.

Protein lysates were generated by lysis with SDS buffer containing protease and phosphatase inhibitors. Endoglycosidase H (Endo H) treatment was carried out by combining 10 µg of glycoprotein with 1 µl of 10× glycoprotein denaturing buffer (100°C, 10 min) before treatment with GlycoBuffer and Endo H for 1 h as previously described (50). The samples were run on Criterion 15% Tris-HCl, 12 or 7.5% TGX gels from Bio-Rad and analyzed on a ChemiDoc MP (Bio-Rad) applied for imaging. The following Abs were used: anti-CD79A (clone D1X5C) and anti-CD79B (clone D7V2F) from Cell Signaling Technologies, anti-CD79B (clone CB3-1) from eBioscience, anti-IgM rabbit F(ab′)₂ (catalog no. 309-006-043) from Jackson ImmunoResearch Laboratories, anti-histone H3 rabbit polyclonal from Millipore (catalog no. 05-928), and anti-GAPDH rabbit polyclonal from GeneTex (catalog no. GTX100118). Image processing was performed in ImageLab (Bio-Rad). Global background-adjusted volume of bands was quantified, and the lane normalization factor was calculated by normalizing all housekeeping control bands to the strongest housekeeping signal on the gel. Each band of interest was then normalized by the lane normalization factor, and intensities are shown relative to expression of total protein by original cells (lane 1).

RNA was isolated using the Qiagen RNeasy Plus kit, and cDNA was synthesized using the Qiagen QuantiTect kit. Quantitative PCR was carried out using TaqMan universal PCR master mix on a 7500 real-time PCR system with probes against CD79A (Hs00998119_m1), CD79B (Hs00236881_m1), and IGHM (Hs00941538_g1), using GAPDH (Hs99999905_m1) and PGK1 (Hs9999906_m1) as endogenous controls (Applied Biosciences). Relative gene expression was calculated by the comparative cycle threshold method.

Statistical analyses were performed in GraphPad Prism, and data were considered statistically significant when p < 0.05. One-way ANOVA, a one-sample t test, or a two-sample t test was applied to determine the level of statistical significance. In Fig. 1B, linear regression and Pearson correlation were performed to investigate the relationship between signaling and surface Ig expression.

To investigate the relationship between BCR expression and BCR signaling strength, we started by mapping the IgH isotype (IgM, IgD, IgG, IgA) and BCR expression level in tonsillar B cell subsets. Naive, GC, and memory B cells were distinguished using CD20, CD38, CD27, and IgD (Supplemental Fig. 1A). All naive B cells expressed IgM and/or IgD, whereas GC and memory B cells had a mix of IgM⁺, IgD⁺, IgG⁺, and IgA⁺ B cells (Supplemental Fig. 1B, 1C). GC B cells had a large fraction of BCR-negative cells as expected (Supplemental Fig. 1B, 1C). The expression level of CD79B and IgM/IgG showed a linear relationship (Supplemental Fig. 1D, 1E). Similarly, IgD levels were also associated with CD79B levels, but a population of IgD^lowIgM^high cells interrupted the linearity of this relationship (Supplemental Fig. 1D–G). The maximum expression level of IgM, IgG, IgD, IgA, and CD79B was similar across the different B cell subsets (Supplemental Fig. 1H, 1I). In separate experiments, using the same 10 tonsils, we measured BCR signaling by phospho-specific flow cytometry (phospho-flow) in the three B cell populations (Fig. 1A). On a population level, anti-IgM/IgG–induced signaling varied between the subsets but exhibited good correlation between percentage surface IgG/IgM expression and anti-IgM/IgG–induced signaling strength (Pearson r = 0.85 for p-BLNK and r = 0.68 for p-SYK; (Fig. 1B). To directly link BCR signaling strength to the surface IgM or IgG expression level, we induced signaling by activating tonsillar cells with FITC-labeled anti-IgM F(ab′)₂ or anti-IgG F(ab′)₂, whereas unstimulated cells were fixed before staining with the FITC-labeled Abs (Fig. 1C). Different levels of IgM or IgG expression were gated, and BCR-induced signaling was then quantified for specific levels of IgM/IgG relative to the same level in unstimulated B cells (Fig. 1C). This approach revealed a strong association between surface IgM or IgG and induced phosphorylation of the downstream signaling proteins p-SYK, p-BLNK, p-LCK, and p-PLC-γ in naive/memory as well as in GC B cells (Fig. 1D, 1E). A potential limitation in these experiments is that the total levels of signaling proteins were not measured. However, due to the short activation time, changes are not expected. Thus, we conclude that the BCR expression level determined the BCR signaling strength in human mature B cells.

To further examine how the surface expression level of BCR affects B cell function, we measured apoptosis in human tonsillar B cells. GC B cells rapidly underwent spontaneous apoptosis when cultured in vitro for 3 h as measured by expression of cleaved caspase-3 (Fig. 2A). To link IgM expression to apoptosis, we used the same approach as for tracking of BCR-induced signaling and briefly stained tonsillar B cells with FITC-labeled anti-IgM F(ab′)₂ at time 0 for tracking IgM^low and IgM^high B cells. The IgM-FITC signal and IgD expression were stable for at least 6 h (Supplemental Fig. 1J). Of importance, the anti-IgM-FITC induced BCR signaling (Fig. 1), but it did not induce apoptosis, as unstimulated cells and IgM-stained cells displayed the same percentage of apoptotic cells at 3 h (Fig. 2A). To further rule out that anti-IgM-FITC binding to IgM molecules at the start of the culture induced apoptosis, we also stained cells at the end of the culture. These two approaches showed that high surface IgM expression protected GC B cells from FASL-induced apoptosis, as cleaved caspase-3 was significantly lower in IgM^high cells compared with IgM^low cells (Fig. 2B, 2C). Spontaneous apoptosis was also significantly lower in IgM^high cells than IgM^low cells when the IgM staining was performed at the end of the culture (Fig. 2D). IgM-negative GC B cells displayed the highest frequency of apoptotic cells (Fig. 2B–(D). Naive and memory B cells had a very low level of spontaneous or FASL-induced apoptosis during this short-term culture, and no significant difference was found related to the difference in IgM expression level (Fig. 2). These experiments demonstrate a link between IgM levels and apoptosis in GC B cells.

Because IgM surface levels correlated with the level of apoptosis in primary GC B cells, we further investigated the role of CD79A and CD79B for IgM expression and survival using CRISPR/Cas9 in cell line models. After CD79A-CRISPR editing in three IgM⁺ MCL cell lines (MINO, Z-138, Granta-519) and a GCB DLBCL line (SUDHL-6), the cells that lost surface CD79A expression also lost surface IgM and CD79B expression (Fig. 3A, Supplemental Fig. 2A, 2B). We did not observe CD79A-negative cells with CD79B expression as reported in previous studies (31). The CD79A-edited cultures were immunophenotyped regularly for up to 24 d, and in most cell lines the triple-negative population was outcompeted by unedited original cells (Fig. 3B, 3C). In MINO and SUDHL-6 cells, the CD79A⁻IgM⁻ population rapidly declined, whereas Z-138 cells displayed a slower reduction of the edited population. CRISPR/Cas9 editing of CD79B also led to a triple-negative population that was rapidly outcompeted in MINO, SUDHL-6, and Z-138 cells (Fig. 3D–F, Supplemental Fig. 2A, 2B). We also observed a CD79B⁻IgM⁺ population that remained stable over time (Fig. 3D, 3E, 3G). We generated single-cell clones from this population, and CD79B⁻IgM⁺ MINO clones (C7 and C9) had a homozygous 6-bp deletion in CD79B (Supplemental Fig. 2C), which did not reduce the total levels of CD79B or IgM protein (Supplemental Fig. 2D–F, 2H). Z-138 clones (B and D) with a 6-bp deletion in one allele and a 1-bp deletion in the other had reduced levels of CD79B and slightly lower IgM (Supplemental Fig. 2G, 2H). As CD79B protein was still present, we did not further investigate these clones and focused on CD79A⁻CD79B⁻IgM⁻ triple-negative cells.

To determine whether the CD79A⁻CD79B⁻IgM⁻ triple-negative cells were outcompeted or died by apoptosis, we sorted CD79B⁻IgM⁻ and CD79B⁺IgM⁺ MINO cells 3 d after CD79A/CD79B-CRISPR editing and cultured the two populations separately. More than 90% of the sorted CD79A sgRNA-edited cells and >80% of the CD79B sgRNA-edited cells maintained a triple-negative phenotype (on day 10 and day 5, respectively; data not shown). The sorting selected for live cells, and directly after the sort the viability was equal in CD79B⁻IgM⁻ and CD79B⁺IgM⁺ MINO cells (Fig. 3H). During culturing, the viability became significantly lower in CD79B⁻IgM⁻ cells compared with CD79B⁺IgM⁺ cells (Fig. 3H). The day after sorting, we detected high cell death and apoptosis in the CD79B⁻IgM⁻ cells (Fig. (3I, 3J). We also observed lower p-GSK3β in CD79B⁻IgM⁻ cells compared with CD79B⁺IgM⁺ and Cas9 control cells (Fig. (3K). Reduced p-GSK3β has previously been associated with reduced fitness in lymphoma cells (30).

The survival and fitness of Granta-519 cells were not affected by loss of CD79A or CD79B and surface IgM, as the frequency of edited cells remained unchanged after 24 d in culture (Fig. 3C–F). Unlike the other cell lines, Granta-519 cells are EBV-positive and may express EBV genes such as LMP2A, which provides a constitutive surrogate for BCR signaling using LYN and SYK kinases (51), or LMP1, which activates TRAF2/3 and the NF-κB pathway, rendering these cells less dependent on tonic BCR signaling. This might also explain the wide range of IgM surface expression in original Granta-519 cells. These cells displayed a large population of IgM⁻ cells prior to CRISPR/Cas9 editing, contrasting the other cells lines displaying mostly surface IgM^+/high cells (Fig. 3A).

Stable CD79A and CD79B KO clones were established from the cultures of edited Granta-519 cells (Supplemental Fig. 2C). The clones maintained a triple-negative phenotype, lacking surface expression of CD79A, CD79B and IgM (Fig. 4A). To confirm that the clones had no functional BCR, we compared anti-IgM–induced signaling in KO cells to the original Granta-519 cells and did not detect any BCR signaling in the KO cells (Fig. 4B, 4C).

In Z-138 cells that had undergone CRISPR/Cas9 editing of CD79A or CD79B, IgM-induced signaling was measured in bulk populations, using Abs against CD79A, CD79B, and IgM to identify original and edited cells (Fig. 4D). In CD79A^low cells, reduced BCR signaling and low/no IgM expression levels were observed, confirming the loss of a functional BCR in the edited cells (Fig. 4D, 4E). The population of CD79B⁻IgM⁻ double-negative cells in CD79B-edited cells was lost at the time point when these experiments were performed, preventing similar analysis of BCR signaling in these cells.

To investigate the mechanisms underlying loss of surface BCR expression upon deletion of CD79A or CD79B, we investigated gene transcription and protein expression in original cells and CRISPR-edited cells (Fig. 5, Supplemental Fig. 3). Protein lysates from Granta-519 CD79A KO cells did not express CD79A protein (Fig. 5A, lane 2, 5B), and CD79A mRNA was not detected in these cells (Fig. 5C). CD79B KO cells did not express CD79B (Fig. 5A, lane 3, 5B), but CD79B mRNA was detected at the same level as in original cells (Fig. 5C). The total protein level of IgM was not altered in the two KO models (Fig. 5A, lane 1–3, 5B), but IGHM mRNA was increased, possibly through a compensatory mechanism (Fig. 5C). In contrast, the total protein level of CD79B was significantly reduced in CD79A KO cells (Fig. 5A, lane 2, 5B), although the CD79B mRNA expression level was not altered (Fig. 5C). This suggests that the protein stability of CD79B depends on CD79A. CD79A mRNA was, similarly to IGHM, increased in CD79B KO cells (Fig. 5C), but total CD79A protein was not altered (Fig. 5A, lane 3, 5B).

The loss of surface BCR in CD79A/CD79B KO cells could be due to a block in protein maturation and transport of proteins from the ER to the surface. To test this in our KO models, we used the enzyme Endo H to cleave off N-linked glycans from immature proteins. The complex N-linked glycans of mature protein are Endo H resistant, whereas immature protein will have the glycans cleaved off upon Endo H treatment. Western blot analysis can then be applied to separate the longer mature protein from the shorter immature protein (40). Whereas original Granta-519 cells expressed both mature and immature fraction of CD79B (Fig. 5A, lane 4, 5D), CD79A KO cells only expressed the immature fraction of CD79B (Fig. 5A, lane 5, 5D). The mature fraction of CD79A is barely detectable in Granta-519 cells (Fig. 5A, lane 4), but the weak band of mature CD79A is missing in CD79B KO cells (Fig. 5A, lane 6, 5D), indicating that CD79A protein cannot mature without CD79B and vice versa. Original Granta-519 cells had variable and low levels of mature IgM protein (Fig. 5A, lane 4, 5D), in agreement with the low fraction of cells with surface expression observed by flow cytometry (Fig. 3A). We did not observe an effect on IgM protein maturation in Granta-519 CD79A KO and CD79B KO cells (Fig. 5A, lane 4–6, 5D).

Due to the low surface expression and mature fraction of IgM in Granta-519 cells, we used Z-138 cells as a second model system. We were able to obtain a stable Z-138 CD79B KO clone whereas CD79A KO cells did not survive in culture. Instead, CD79A-negative cells were sorted from CD79A-edited cultures before cell lysis. CD79A protein was lost in CD79A-negative cells (Fig. 6A, lane 2) and CD79B protein was lost in CD79B KO cells (Fig. 6A, lane 3), whereas total IgM protein was not altered by CD79A/CD79B KO (Fig. 6A, lane 1–3, 6B). As in Granta-519 cells, the total protein level of CD79B was significantly reduced in CD79A-negative cells whereas the total protein level of CD79A was less affected in CD79B KO cells (Fig. 6A, lane 1–3, 6B). Original Z-138 cells expressed both mature and immature fractions of IgM, but the mature fraction of IgM was lost in CD79A-negative cells and in CD79B KO cells (Fig. 6A, lane 4–6, 6C). Similarly, the mature fraction of CD79A was lost in CD79B KO cells and the mature fraction of CD79B was lost in CD79A-negative cells (Fig. 6A, lane 4–6, 6C). These results support a model where retention of immature CD79A/CD79B and IgM in the ER is the main mechanism for reduced surface BCR expression in CD79A/CD79B KO cells.

We next performed a rescue experiment with reintroduction of WT CD79B or overexpression of a validated dominant-negative mutant, G137S (12), in Granta-519 CD79B KO cells. The G137S mutation is next to the cysteine required for the disulfide bond with CD79A (C136) and was detected in a patient with agammaglobulinemia where the mutation resulted in inefficient heterodimerization of the CD79 molecules, low IgM expression, and a strong reduction of mature peripheral blood B cells (12). To assess equivalent production of the rescue proteins, we linked CD79B to GFP using a 2A ribosome skipping peptide (47). Reintroduction of WT CD79B resulted in strong surface expression of CD79A, CD79B, and IgM, contrasting CD79B G137S, which led to low surface expression of the three BCR components (Fig. 7A). Reintroduction of CD79B variants with point mutations of Y196 led to even higher surface expression of CD79A, CD79B, and IgM, both when comparing the total GFP⁺ population (Supplemental Fig. 4A, 4B) and cells with equivalent GFP levels (Supplemental Fig. 4C), consistent with previous observations (16, 50). Importantly, the low surface expression levels observed with CD79B G137S were not due to lower retroviral transduction efficiency because both CD79B constructs showed similar GFP expression levels (Fig. 7A). In addition, CD79B mRNA expression levels were similar (Fig. 7B), whereas the total level of CD79B protein was higher for CD79B WT than CD79B G137S (Fig. 7C, lane 3 and 4, (Fig. 7D), suggesting that the mutant has reduced protein stability. However, the most striking difference was in protein maturation of CD79B (Fig. 7E, Supplemental Fig. 4D). Whereas cells with WT CD79B expressed both mature and immature CD79B at the same ratio as original Granta-519 cells, the cells expressing CD79B G137S only expressed the immature fraction of CD79B (Fig. 7C, lane 5, 7, and 8, 7E). CD79A mRNA expression was not affected by reintroduction of CD79B (Fig. 7B), but the total protein level of CD79A was increased (Fig. 7C, lane 3, 7D) and the mature fraction was strongly increased in cells with reintroduced CD79B WT (Fig. 7C, lane 7, 7E). Similarly, IGHM mRNA expression was not affected (Fig. 7B), but a strong skewing toward more mature IgM protein (Fig. 7C, lane 7, 7E) and a small increase in IgM total protein were observed after reintroduction of CD79B WT (Fig. 7C, lane 3, 7D).

Collectively, these data demonstrate that CD79B protein governs the maturation and surface expression of both CD79A and IgM. The model that emerges from these results is that heterodimerization of CD79A and CD79B is required for proper protein maturation.

The requirement for CD79A and CD79B for BCR expression and B cell fitness is well established in mice, but few studies have investigated this using human B cells. Prior CRISPR/Cas9 screens in human DLBCL cell lines identified CD79A and CD79B as essential but did not investigate consequences for BCR assembly and trafficking to the cell surface (21–23). In human MCL cell lines, we demonstrated in the present study that surface expression and glycan maturation of IgM and CD79 are lacking after CD79A or CD79B KO, indicating retention of BCR complex components in the ER. The loss of surface IgM had important consequences for B cell survival. In three out of four cell lines, CRISPR/Cas9 editing of either CD79 gene reduced the fitness relative to original cells and the edited cells were outcompeted by non-sgRNA-transfected cells over time. Cell sorting of MINO cells that had lost surface expression of IgM after CD79A or CD79B editing demonstrated that these cells rapidly underwent apoptosis. The consequence of IgM surface level for survival was further shown in primary human GC B cells as high surface density of IgM protected against spontaneous- and FasL-induced apoptosis.

Loss of BCR surface expression following CD79 KO was previously shown in a screen using shRNA in DLBCL cell lines where the degree of CD79A or CD79B knockdown correlated with the level of surface BCR (16). Assembly of the BCR complex occurs in the ER (36). Without dimerization, CD79A and CD79B are blocked from transport to the Golgi apparatus and this mechanism depends on conserved motifs in the transmembrane domain that contains charged amino acids (34, 52). The glycosylation pattern of CD79A, CD79B, and IgM in CD79A or CD79B KO cells and in CD79B G137S-reconstituted cells in our study fit with a model where these proteins are retained in the ER when CD79A/CD79B heterodimer formation is not possible (40). Reintroduction of WT as well as CD79B Y196 variants in Granta-519 CD79B KO cells led to high surface expression of CD79A and IgM, suggesting that CD79B is the rate-limiting factor for the level of surface BCR expression in these cells. The same conclusion has previously been drawn for normal mouse B cells (50). The BCR surface expression in Granta-519 cells was not regulated at the transcriptional level because reintroduction of CD79B WT increased the surface BCR expression without increasing CD79A and IGHM mRNA level, and high levels of mutated CD79B G137S mRNA did not affect CD79A or IGHM mRNA levels. Instead, pools of immature CD79A and IgM protein were matured as the level of mature CD79B increased, shown by the skewing in the mature/immature ratio of CD79A and IgM. In primary mouse B cells, which do not have a large pool of immature IgM, overexpression of CD79B WT led to a small increase in surface BCR expression and a large increase in immature CD79B protein (50). The increased CD79A and IgM maturation in Granta-519 cells depended on CD79A/CD79B heterodimerization, as the CD79B G137S variant did not rescue the protein maturation of CD79A and IgM. In addition, the total protein level of CD79A and IgM increased in cells overexpressing CD79B, suggesting that mature CD79B stabilizes the other two proteins and prevents protein degradation, in agreement with earlier reports (37). In contrast, CD79A was suggested to be the rate-liming factor for BCR expression in the Burkitt’s lymphoma cell line Ramos (36). Using the same cell line, He et al. (31) showed that CD79B can be expressed as homodimers on the surface in the absence of CD79A and IgM. Crystal structures of human and murine CD79B suggest that the protein can form homodimers (33). However, this was not seen in the three MCL cell lines and the GCB DLBCL cell line used in our study, as CD79B was not expressed on the surface of CD79A KO cells. The mechanism for how CD79B can exist as homodimers and be expressed at the plasma membrane in the absence of other BCR components in Burkitt’s lymphoma but not in DLBCL and MCL remains unknown.

Glycosylation defects and retention of IgM in the ER is a mechanism for reduced BCR expression in CLL (53), which can be rescued by IL-4 (54, 55). Furthermore, surface IgM in CLL cells can differ in the N-glycosylation pattern in the Igµ C region, with one being similar to normal B cells with mature complex glycans whereas the other form is high in mannose, characteristic of immature Igµ chains in the ER (56). Similar to CLL cells, tolerant B cells express normal levels of IGHM mRNA and surface IgD, but IgM is selectively blocked in the ER after repeated stimulation by self-antigen (40). B cells expressing mutated MYD88 downregulate their IgM similarly to the trafficking block of IgM induced by chronic exposure to self-antigen (50). This can be overcome by CD79B Y196 ITAM mutations causing elevated surface BCR expression (50). Co-occurring mutations in CD79B and MYD88 are characteristic of the genetic subtype MCD/C5 of DLBCL (57, 58), demonstrating that malignant B cells can acquire mutations to escape IgM retention or maturation block that would normally cause low BCR expression.

CD79A or CD79B KO reduced the fitness of MCL cell lines MINO and Z-138. MINO was rapidly outcompeted by unedited cells due to rapid induction of apoptosis upon loss of surface IgM, whereas Z-138 had a slower decline. It has previously been shown that some MCL cell lines, including MINO, are sensitive to BTK inhibitors due to constitutive BCR activation and canonical NF-κB signaling (26). Z-138 cells represent another subset of MCL cell lines that are insensitive to BTK inhibitors and demonstrate activation of the noncanonical NF-κB pathway (26). This difference in NF-κB pathway activation can explain why Z-138 cells are less sensitive to loss of surface BCR expression. Granta-519 cells were resistant to BCR loss, likely due to EBV (51), showing that lymphoma cells can replace the survival signal that should arise from a functional BCR. In Ramos cells, CD79B formed homodimers in the absence of other BCR complex components and constituted an alternative survival pathway (31). In DLBCL cell lines, KO of the IgH reduced the fitness of DLBCL cell lines, but the decline of the cells varied greatly (15). Reduced proliferation was the cause of the decline in most cells, but apoptosis was also seen in some cells. Collectively, this demonstrates that malignant B cells can acquire mutations and/or express viral proteins that provide better fitness, compared with healthy B cells.

In some lymphoma cells, IgM expression has been shown to have other roles than providing survival signals. In Ramos cells, the production of IgM was crucial for the metabolic fitness of B cells (59). Without expression of immunoglobulins, ER homeostasis was not maintained and the cells died (59). In a myc-driven mouse lymphoma model, BCR depletion did not affect lymphoma growth, but BCR⁻ cells with reduced p-GSK3β were outcompeted by their BCR⁺ counterparts (30). GSK3 is a metabolic sensor downstream of BCR signaling. The kinase promotes the survival of naive cells by restricting cell mass accumulation, but it is inhibited by phosphorylation in GC B cells to allow for growth, proliferation, and increased metabolic activity (60). The BCR⁻ cells could be rescued by GSK3β inhibition, showing that BCR-induced inhibition of GSK3β was essential for metabolic fitness when nutrients were scarce. BCR⁻ MINO cells displayed reduced p-GSK3β, but they also had increased cell death.

Emerging evidence demonstrates that also healthy B cells depend on surface BCR for survival and metabolic fitness. We found that high BCR expression protected against apoptosis in GC B cells. Consistent with this, it has been shown in both mice and humans that BCR^low cells have reduced signaling compared with BCR^high cells (9, 61). All GC B cells, independent of BCR expression level, displayed increased apoptosis compared with naive and memory B cells. We observed >50% apoptosis after 3 h in culture, consistent with a study showing that 50% of GC B cells die within 6 h in vivo (62). In the GC reaction, B cells go through rounds of somatic hypermutation in the dark zone to increase the affinity of the BCR. Downregulation of the BCR is important in this step, as the GC B cells need to re-express and test the affinity of their newly mutated BCR. A block in CD79B downregulation reduces the number of GC B cells and affinity maturation in transgenic mice (63). This downregulation of the BCR explains the large percentage of GC B cells lacking surface BCR expression in our study. A functional BCR must be re-expressed on the surface for dark zone B cells to reach the light zone, and dark zone B cells with damaging mutations in the BCR do not reach the light zone (64). Light zone B cells are programmed to die unless rescued by T cells whereas dark zone cells depend on tonic signaling, as apoptotic cells in this zone are highly enriched for Ig genes with deleterious mutations lacking surface BCR (62). However, the role of Ag-induced BCR signaling in GC B cells is not fully understood. It has been suggested that BCR signaling is suppressed in GC B cells and that T cell help is the main selective mechanism in the GC (65). In our approach, detecting maximal BCR-induced signaling capacity, we observed similar BCR signaling capacity in human GC B cells compared with naive and memory B cells. Newer studies suggest that BCR signals do play an important role during the GC reaction and that the Ag-induced BCR signaling is rewired to inactivate FOXO1 through PI3K-AKT in GC B cells (66).

In this study, combining CRISPR/Cas9 deletion of CD79A or CD79B with studies of glycan maturation in MCL cell lines gave mechanistic insight into protein maturation and BCR complex assembly in human B cells. Our study highlights the central role of the survival signal provided by BCR signaling for primary human GC B cells and in EBV-negative MCL cell lines. Activation through the BCR guides decisions about life, death, and B cell function, and our results showed that the IgM surface level contributes to these decisions. The presence of IgM on the surface of these B cells equally depended on CD79A and CD79B, and when one of these proteins was deleted, it blocked glycan maturation of the other CD79 protein as well as IgM, with subsequent blocked transportation of the IgM complex to the plasma membrane. We have confirmed the requirement of the CD79A/CD79B heterodimer for surface expression of IgM in human B cells as previously comprehensively demonstrated in mice.

We are grateful for services provided by the Flow Cytometry Core Facility at Oslo University Hospital.

E.B.S. has received royalties from licensing of several mAbs against CD Ags. He owns stocks in Nordic Nanovector. He is a coinventor on several patents from the Lymphoma and Leukemia Profiling Project led by Louis M. Staudt (National Cancer Institute). The other authors have no financial conflicts of interest.