A crucial role for Jagunal homolog 1 in humoral immunity and antibody glycosylation in mice and humans

The humoral branch of the immune system is responsible for the production of circulating antibodies to remove pathogens and harmful molecules. Following antigen exposure, naive B lymphocytes recognize cognate antigens with their B cell receptor and upon provision of appropriate costimulatory signals differentiate into memory B cells and plasma cells, which secrete thousands of antibodies per cell per second (Hibi and Dosch, 1986). These exceptionally high rates of Ig production and secretion are enabled by structural reorganization and extensive generation of the ER and Golgi apparatus during B cell differentiation into plasma cells (Wiest et al., 1990). The ER is a cellular organelle responsible for protein folding and assembly, lipid and sterol synthesis, and calcium storage and is positioned adjacent to the Golgi apparatus, which processes proteins for secretion, e.g., by extensive protein glycosylation (Kornfeld and Kornfeld, 1985).

Ig are secreted glycoproteins that recognize antigens via their antigen-binding fragment regions. Additionally, Ig communicate with other components of the immune system via their fragment crystallizable (Fc) region to exert their effector functions (Huber et al., 1976; Woof and Burton, 2004). The Fc domain of Ig can also interact with complement proteins (Miletic and Frank, 1995). Both the antibody isotype determined by the constant, heavy chain (CH) region as well as the pattern of N-glycosylation are important determinants of antibody functionality (Stavnezer et al., 2008; Tao and Morrison, 1989). Antibody glycans are assembled and attached mainly to asparagine 297 (Asn297) of the CH2 domain of the IgG antibody polypeptide chain in the ER and further modified during passage through the Golgi apparatus. The N-linked glycosylation of an IgG antibody has been shown to affect its conformation and thereby its capacity to interact with Fc receptors on cells of the innate immune system (Vidarsson et al., 2014; Arnold et al., 2007). This can affect both an antibody’s half-life as well as modulate its effector functions (Jennewein and Alter, 2017). These glycosylation-modulated interactions enable immune responses including antibody-dependent cell-mediated cytotoxicity (ADCC), antibody-dependent cellular phagocytosis, complement-dependent cytotoxicity (CDC), or neutrophil activation (Nimmerjahn and Ravetch, 2008; Kellner et al., 2017). Insights into the role of glycosylation in the regulation of antibody-mediated immune functions have been applied to glyco-engineer therapeutic monoclonal antibodies to enhance or alter their effector functions (Jefferis, 2009). Moreover, alterations in antibody glycosylation have been linked to autoimmunity (Seeling et al., 2017). However, the in vivo molecular control and regulation of a B cell’s cellular adaptation to high rates of antibody production, glycosylation, and secretion remains poorly understood.

Jagn1 encodes a ubiquitously expressed transmembrane protein that has been demonstrated to be essential for both ER reorganization in Drosophilamelanogaster oocyte development (Lee and Cooley, 2007) and more recently the differentiation and survival of human neutrophils (Boztug et al., 2014). Thus far, described defects in mammals have been limited to neutrophils, where JAGN1 regulates the process of granule formation (Boztug et al., 2014; Wirnsberger et al., 2014). Since differentiation of B cells into antibody-secreting plasma cells involves a significant ER expansion to support high rates of antibody production and secretion, we set out to study a role for JAGN1 in B cell biology. Here we report that loss of Jagn1 in B cells causes defects in Ig production and secretion as well as aberrant IgG glycosylation and subsequent enhanced Fc receptor binding. Our data uncover a novel role for JAGN1 in B cell biology to critically regulate humoral immunity in mice and humans.

To study a role for Jagn1 in B cells, we bred Jagn1^fl/fl-Mb1-Cre mice (hereafter referred to as Jagn1^ΔB) to delete Jagn1 at the early pro–B cell stage (Hobeika et al., 2006). To exclude Cre effects in the Mb1-Cre line, we analyzed B cell development and B cell subset composition as well as serum Ig levels in Mb1-Cre and control littermates. We did not observe any significant differences between those cohorts in all parameters tested (Fig. S1, A–C) and therefore used Jagn1^fl/fl mice as controls throughout the manuscript. Efficient deletion of Jagn1 in the B cell lineage was determined by measuring Jagn1 mRNA levels in B220⁺ splenic B cells and in vitro–differentiated plasmablasts (Fig. S2 A). Protein loss was confirmed by intracellular staining for JAGN1 in plasma cells using flow cytometry (Fig. S2 B). To assess the subcellular localization of JAGN1 in B cells, we next established a plasmacytoma cell line (MPC-11) by knockdown of endogenous Jagn1 and expression of JAGN1 fused to an amino-terminal V5 tag. Immunohistochemical analysis revealed that V5-JAGN1 localized to the ER in B cells (Fig. S2 C).

To assess B cell function, serum Ig levels of Jagn1^ΔB mice and control Jagn1^fl/fl littermates were measured by immunoassays using the Luminex xMAP technology. Jagn1^ΔB mice exhibited an isotype-dependent 4–18-fold reduction in serum Ig levels in unchallenged mice (Fig. 1 A). We next assessed the role of JAGN1 in the generation of antigen-specific humoral immune responses. Jagn1^ΔB mice displayed significantly reduced titers of OVA-specific IgG and IgM antibodies after immunization with OVA/alum when compared with littermate control mice (Fig. 1 B). Consistent with impaired anti-OVA responses, Jagn1^ΔB mice immunized with the hapten NP (4-hydroxy-3-nitrophenyl-acetyl) coupled to the carrier protein KLH also showed a striking reduction in the generation of NP-specific IgG1 antibody–secreting cells, as assessed by ELISPOT (Fig. 1 C). The profound defects in humoral immunity in Jagn1^ΔB mice prompted us to study their B cell response in a physiologically relevant immune response upon infection with vesicular stomatitis virus (VSV; Turner et al., 2008). Upon infection with VSV, Jagn1^ΔB mice exhibited significantly decreased titers of virus-specific total Ig and, even more pronounced, of antiviral IgG antibodies (Fig. 1 D). These data show that deletion of Jagn1 in the B cell lineage results in decreased Ig levels in naive mice and strongly diminished humoral responses upon immunization with specific antigens or when challenged with a viral infection.

To gain insight into the role of Jagn1 in B cell development and homeostasis, we analyzed steady-state levels of developmental and mature B cell subpopulations in the bone marrow (BM) and spleen of Jagn1^ΔB and control littermate mice by flow cytometry. Jagn1^ΔB mice showed apparently normal early B cell development but exhibited markedly reduced levels of BM plasma cells (Lin⁻CD28⁺CD138⁺) when compared with littermate control mice (Fig. 2 A). While we observed a marked effect of Jagn1 deficiency on the plasma cells residing in the BM, which are long-lived and the major source of serum antibody levels (Benner et al., 1981; Slifka et al., 1995; Manz et al., 1997; Slifka et al., 1998), short-lived plasma cells in the spleen (Lin⁻CD28⁺CD138⁺) were only slightly reduced in Jagn1^ΔB mice (Fig. 2 B). Additionally, Jagn1^ΔB mice exhibited enlarged spleens, increased proportions of splenic marginal zone (MZ; CD21⁺CD23^low) B cells, and reduced proportions of follicular (FO; CD23⁺CD21^int) B cells, as detected by flow cytometric analyses and immunohistochemistry, when compared with littermate control mice (Fig. 2, B and C; and Fig. S2 D). Thus, inactivation of Jagn1 results in markedly impaired BM plasma cell numbers and altered splenic B cell populations.

To investigate B cell–intrinsic effects of Jagn1 deletion, we generated mixed BM chimeric mice. To this end, we transplanted a 1:1 mixture of enhanced GFP–labeled (Rosa26-EGFP) Jagn1^+/+ BM plus Jagn1^fl/fl or Jagn1^ΔB BM into lethally irradiated Rag2^−/− recipient mice. The increased MZ/FO B cell ratio observed in Jagn1^ΔB mice was restored to control levels in Jagn1^ΔB BM–derived cells by the presence of wild-type cells (Fig. 2 D). Moreover, in these mixed BM chimeric mice, the presence of Jagn1^+/+ B cells also restored serum IgM to levels observed in control BM chimeric animals (Fig. 2 E). These data indicate that the increased numbers and proportions of MZ B cells observed in Jagn1^ΔB mice are not caused by an intrinsic function of JAGN1 in MZ B cell differentiation or homeostasis but are a secondary effect, most probably caused by reduced Ig serum levels (Tsiantoulas et al., 2017). While Jagn1^fl/fl and Jagn1^ΔB BM–derived cells equally contributed to the reconstitution of mature splenic B cell populations (Fig. 2 F), Jagn1^ΔB BM–derived cells were depleted in CD28⁺CD138⁺ plasma cells both in the spleen and the BM (Fig. 2 G). To further characterize the plasma cell phenotype, we crossed Jagn1^ΔB mice to Blimp1-GFP reporter mice (Kallies et al., 2004), carrying a GFP reporter gene in the Blimp1 locus. The transcription factor BLIMP1 is a central regulator of plasma cell differentiation (Tellier et al., 2016; Minnich et al., 2016). This allowed for the detection of CD138⁺GFP⁺ plasma cells irrespective of their CD28 expression status and confirmed that Jagn1^ΔB mice indeed exhibit severely reduced numbers and proportions of BM plasma cells with lower levels of BLIMP1-GFP per cell (Fig. S2 E). In addition to their decreased numbers, the remaining Jagn1-deficient plasma cells exhibited significantly lower levels of IgM on a per-cell basis as shown by intracellular staining and analysis by flow cytometry (Fig. S2 F). Moreover, intracellular staining revealed reduced BLIMP1 protein levels in Jagn1^ΔB plasma cells (Fig. S2 G). These data indicate that loss of Jagn1 results in a competitive disadvantage to generate plasma cells.

To account for possible early developmental defects contributing to the observed plasma cell phenotype in the Jagn1^ΔB animals, we generated Jagn1^fl/fl-Cd23-Cre mice to delete Jagn1 at a later developmental stage (Kwon et al., 2008). In these mice, we detected a phenotype similar to the one observed for Mb1-Cre–mediated Jagn1 deletion, although slightly weaker for BM plasma cells (Fig. S1, D–F).

Since JAGN1 mutant patients exhibit phenotypic variability (Boztug et al., 2014), presumably caused by their differing genetic backgrounds, we next tested for mouse strain–specific effects on the observed B cell phenotypes. To this end, we crossed Jagn1^ΔB C57BL/6J mice onto C57BL/6J/129/Sv and C57BL/6J/BALB/c mixed backgrounds (Fig. S3 A). In all mixed backgrounds, Jagn1^ΔB mice showed a comparable phenotype of increased splenic MZ B cells, strongly decreased intracellular IgM levels in splenic and BM plasma cells, and markedly reduced numbers of long-lived BM plasma cells, whereas short-lived splenic plasma cells were apparently not affected by the loss of Jagn1 (Fig. S3, B–D). Taken together, these data indicate that JAGN1 is cell-intrinsically required for the generation of plasma cells.

To further analyze the consequences of Jagn1 loss on antibody-secreting cells, we isolated splenic B cells from control and Jagn1^ΔB mice and stimulated them with LPS for 4 d to induce IgM secretion and plasma cell differentiation. LPS induces the sequential differentiation of naive B cells into activated B cells (CD22⁺CD138⁻), preplasmablasts (CD22^lowCD138⁻), and finally plasmablasts (CD22⁻CD138⁺; Minnich et al., 2016). LPS stimulation of Jagn1^ΔB-derived B cells led to a significant albeit small reduction in antibody-producing preplasmablasts and plasmablasts (Fig. 3 A).

To study transcriptional changes in these in vitro–differentiated plasmablasts caused by Jagn1 loss, we FACS sorted CD22⁻CD138⁺ plasmablasts and performed RNAseq, which revealed 521 (289 down and 232 up) differentially expressed genes (padj < 0.01; Fig. S4 A). Gene ontology (GO) enrichment analysis of the up-regulated genes revealed stress response, apoptosis, protein folding, unfolded protein response (UPR), or ER as overrepresented categories in Jagn1^ΔB plasmablasts (Fig. 3 B). The up-regulated UPR/stress-response genes included genes encoding chaperones (such as Hspa1a, Hspa1b, and Hspa8), transcriptional regulators (Atf4, Atf5, Ddit3, and Nupr1), and enzymes (Fkbp11 and Ppp1r15a) and other known UPR target genes (Gadd45a and Sqstm1) to assist in protein folding and relieving ER stress. We therefore measured ER stress in Jagn1^ΔB plasmablasts compared with littermate control-derived cells. Jagn1-deficient plasmablasts indeed showed increased ER stress as evidenced by enhanced X-box binding protein 1 (Xbp1) splicing (Fig. 3, C and D) and increased spliced XBP1 protein levels (Fig. 3 E). Furthermore, we found up-regulated Golgi stress genes including Fut1 and Golgb1, which are involved in maintaining Golgi homeostasis (Serebrenik et al., 2018). To further assess whether Jagn1 loss and the observed ER stress affects the Golgi apparatus, we stained cells with the Golgi Cytopainter dye to measure Golgi content by flow cytometry. We found a decreased Golgi Cytopainter fluorescence intensity in Jagn1^ΔB plasmablasts (Fig. S4 C), suggesting a reduced abundance of Golgi compartments. Additional evidence for an impact on the endomembrane system was an increase in the LysoTracker signal in Jagn1^ΔB plasmablasts, indicative of a higher abundance of acidic compartments (Fig. S4 D). Down-regulated GO categories in our RNA sequencing analysis included sterol and lipid synthesis as well as metabolism (Fig. S4 B). Since plasma cell differentiation requires an extensive increase in lipid metabolism for ER synthesis, we next tested for the impact of Jagn1 deletion on the structure of the ER. JAGN1 inactivation was associated with a massively altered ER structure in plasmablasts as analyzed by electron microscopy (Fig. 3 F). Thus, loss of ER-localized JAGN1 in B cells leads to increased ER stress as well as marked changes in ER architecture in antibody-secreting plasmablasts.

To test for the ability of Jagn1 KO B cells to produce Ig in response to LPS stimulation in vitro, intracellular and secreted IgM levels were quantified for Jagn1-deficient and -sufficient B cells. Loss of JAGN1 led to a substantial reduction in intracellular and secreted Ig when compared with Ig levels of control cells (Fig. 3, G and H; and Fig. S4, E and F). Of note, transcript levels for both membrane (µM) and secreted (µS) mRNA species were comparable in Jagn1-sufficient and -deficient cells (Fig. 3 I). Similar results were obtained in a more “physiological” in vitro differentiation system—the iGB culture—to generate germinal center–like class-switched B cells (Fig. 4 A). Control and Jagn1-deficient B cells were co-cultured on 40LB stromal cells expressing membrane-bound CD40L and producing the pro-survival B cell activating factor to support optimal proliferation of B cells. In addition, cells were stimulated with IL-4 for 4 d followed by 4 d of IL-21 stimulation to generate IgG1 and IgE class-switched plasmablasts (Nojima et al., 2011; Wöhner et al., 2016). Using this assay, Jagn1-deficient B cells generated an altered ratio of early activated B cells and preplasmablasts but the same number of plasmablasts when compared with control cells (Fig. 4 B). This is in accordance with our observation that splenic plasma cells were found at normal levels in Jagn1^ΔB mice (Fig. 2 B and,Fig. S3 C). However, Jagn1^ΔB plasmablasts exhibited a substantial decrease in IgE and IgG1 production as shown by reduction in both spot numbers and sizes in ELISPOT assays (Fig. 4, C–E).

To analyze the contribution of cell death to the observed Jagn1^ΔB plasma cell phenotype, we crossed Jagn1^ΔB mice to Vav-Bcl2 transgenic mice overexpressing the anti-apoptotic BCL-2 protein in the hematopoietic lineage (Ogilvy et al., 1999). Under conditions of severe and sustained ER stress, pro-apoptotic signals trigger the intrinsic apoptosis pathway by inhibiting anti-apoptotic BCL-2, leading to cytochrome C release from mitochondria, activation of caspases, and ultimately apoptotic cell death (Hetz and Papa, 2018; Urano et al., 2000; Tournier et al., 2000; Lei and Davis, 2003). Overexpression of BCL-2 resulted in reduced numbers of early apoptotic cells (Annexin V⁺ propidium iodide [PI]⁻) and late-apoptotic or dead cells (Annexin V⁺PI⁺), thereby strongly increasing the fraction of viable cells (Annexin V⁻PI⁻) after 4 d of LPS stimulation of control as well as Jagn1^ΔB splenic B cells in vitro (Fig. 4 F). Importantly, overexpression of BCL-2 substantially increased the number and proportion of BM plasma cells in control mice, but only moderately in Jagn1^ΔB mice (Fig. 4 G). Overexpression of BCL-2 could also not rescue the decreased serum IgM levels in vivo or intracellular IgM levels of plasmablasts generated by LPS stimulation of Jagn1-deficient splenic B cells (Fig. 4 G). This might be explained by the fact that continuous ER stress leads to an apoptotic signal that at one point overexpression of anti-apoptotic BCL-2 cannot overcome. These data indicate that while the reduced number of plasma cells observed in Jagn1^ΔB mice as compared with non-Bcl2 transgenic mice can be partially rescued, enhancing cell survival does not restore the impaired IgM levels of Jagn1^ΔB antibody-producing cells.

Serum IgG comprises multiple glycoforms as a result of the addition of complex N-glycans to the IgG Fc domain during trafficking through the ER and Golgi (Takahashi et al., 1987; Raju et al., 2000). IgG glycoform profiles have been reported to vary with age, genetic background, sex, pregnancy, and disease states (de Haan et al., 2016; Pucić et al., 2011; Bondt et al., 2013; Seeling et al., 2017; Zhang et al., 2016; Ackerman et al., 2013). However, the specific endogenous mechanisms that control distinct Ig glycosylation patterns in plasma cells are poorly understood. Typically, IgG-attached glycans are di-antennary, complex-type, N-linked glycans, including combinatorial variants of the common fucosylated N-glycan core structure (i.e., two N-acetylglucosamine residues [GlcNAc] extended by a central mannose and two distal branching mannoses) and its extensions by one additional GlcNAc residue on each mannose, followed by one galactose and terminated by one sialic acid residue (Fig. 5 A).

Considering the fact that Jagn1 mutant neutrophils exhibited aberrant N-glycosylation of multiple neutrophil effector proteins in mice and humans (Wirnsberger et al., 2014; Boztug et al., 2014), we speculated that this role of JAGN1 is not limited to neutrophils but that JAGN1 might also be involved in antibody glycosylation. To investigate this hypothesis, we assessed the relative abundance of IgG subclass–specific glycopeptide glycoforms purified from mouse serum by liquid chromatography electrospray ionization mass spectrometry (LC-ESI-MS). IgG peptides from Jagn1^ΔB mice exhibited altered glycoprofiles defined by reduced fucosylation (GnGnF) of IgG2 and IgG3 as well as increased galactosylation (AA) of IgG3 and sialylation (NgAF and NgNgF) of IgG2 and IgG3 when compared with Ig from their control littermates (Fig. 5 B). Additionally, sera of Jagn1^ΔB mice contained nonfucosylated GnGn and AGn glycoforms, which were absent in control littermates. The strongest difference was found in fucosylated GnGnF IgG3 (28.22% in control vs. 13.78% in Jagn1^ΔB mice; Fig. 5 B).

The observed differences in antibody glycosylation could be due to a direct effect of Jagn1 deficiency on the glycosylation machinery or due to the increased ER stress and thereby decreased antibody glycosylation and secretion in Jagn1-deficient plasma cells. To explore these mutually nonexclusive hypotheses, we analyzed mice lacking the ER stress sensor inositol-requiring enzyme 1 (IRE1) in the hematopoietic lineage using Vav1-Cre as a deleter line (Ire1^Δhem). IRE1 is an essential component of the UPR pathway important for sensing and responding to ER stress by splicing of its downstream transcription factor Xbp1 (Bettigole et al., 2015) and has been reported to be required for efficient B cell differentiation (Zhang et al., 2005). In plasma cells, IRE1/XBP1 is not required for differentiation, but to increase the secretory apparatus for Ig synthesis (Taubenheim et al., 2012). When we assayed Ire1^Δhem mice, we observed markedly reduced serum Ig levels (Fig. S4 G), very similar to Jagn1^ΔB mice. Importantly, the antibody glycosylation changes we had observed in Jagn1^ΔB mice were also found in sera of Ire1^Δhem mice (Fig. 5 C). Together, these data reveal an important role for the ER stress response in the altered glycoprofile of serum Ig in Jagn1^ΔB mice.

Reduced IgG fucosylation and increased sialylation, as observed in Jagn1^ΔB Ig, have been shown to strengthen IgG binding to Fcγ receptors (Shields et al., 2002; Lin et al., 2015; Shinkawa et al., 2003; Okazaki et al., 2004; Ferrara et al., 2011), thereby leading to altered effector functions. Analyses of Rag2^−/− mice transfused with serum of control or Jagn1^ΔB mice (normalized to equal amounts of IgG) showed no change in Ig stability (Fig. S5 A). Furthermore, the respective Fc-glycosylation profiles remained stable over the time of analysis (Fig. S5 B). However, incubation of BM cells isolated from Rag2^−/− mice with serum of Jagn1^ΔB mice yielded an increased fraction of surface-bound IgG on neutrophils, indicating enhanced IgG binding to Fcγ receptors (Fig. 5 D). The augmented binding of serum IgG from Jagn1^ΔB mice to CD11b⁺Ly6G⁺ neutrophils was not observed for CD11b⁺Ly6G⁻ myeloid cells and was strongly diminished by blocking the Fcγ receptors with CD16/32 antibodies, indicating Fc receptor–specific binding (Fig. 5 E). Notably, Jagn1^ΔB mice exhibited an increased fraction of Fc receptor–bound IgG on isolated BM neutrophils despite the fact that these animals had severely reduced total IgG serum levels compared with littermate control mice (Fig. 5 F). To assess whether the increased Fc receptor binding of Jagn1^ΔB IgG has functional consequences, we performed neutrophil-mediated ADCC assays with E0771 breast cancer cells. E0771 cells were opsonized with control and Jagn1^ΔB Ig from E0771 tumor–bearing mice, resulting in antibody coating of the E0771 cells (Fig. S5, C and D). Jagn1^ΔB sera or purified IgG resulted in a higher percentage of dead tumor cells than their wild-type counterparts when mixed with increasing E/T ratios of Rag2-deficient BM cells (Fig. 5 G) or purified neutrophils (Fig. 5 H). These data indicate that the enhanced Fc receptor binding results in increased ADCC by neutrophils toward cancer cells coated with Jagn1^ΔB IgG.

Interestingly, after Ig purification of Jagn1^ΔB sera with protein G beads, additional bands were visible on Coomassie-stained protein gels despite the overall reduction of Ig levels (Fig. S5 E). Using tandem MS, these bands were identified as complement protein C3, a central component of the complement system. Immunoblotting for complement C3 revealed no difference in total C3 complement levels between control and Jagn1^ΔB mouse sera (Fig. S5 F), but after Ig purification by protein G we observed increased binding of complement protein C3 to Jagn1^ΔB mouse serum–derived Ig (Fig. S5 G). Together, this indicates that the glycosylation changes in Jagn1^ΔB-derived IgG enhance their binding to Fcγ receptors and complement components.

Mutations in the human JAGN1 gene cause severe congenital neutropenia (Boztug et al., 2014), and a recent report showed that some JAGN1 patients also suffer from hypogammaglobulinemia (Baris et al., 2015). We therefore analyzed sera from patients with homozygous missense mutations (P6: c.59G>A; p.Arg20Glu, P9: c.40G>A; p.Gly14Ser, P12: c.63G>T; p.Glu21Asp) or a nonframeshift deletion mutation in JAGN1 (P14: c.3543del CCGACGGCA; p.Thr12Gly14del) leading to expression of a mutant and likely nonfunctional protein (Boztug et al., 2014). Importantly, even though the serum Ig levels were within normal limits, we detected an altered Ig glycoprofile of the most abundant IgG1 subclass in sera from patients carrying JAGN1 mutations when compared with healthy donors (Fig. 6 A). Moreover, in patient-derived Epstein-Barr virus–transformed B lymphoblastoid cells, we detected increased signs of ER stress including elevated XBP1 splicing and up-regulation of heat shock protein family A member 5 (HSPA5; Fig. 6 B). These data show that loss of functional JAGN1 protein in congenital neutropenia patients is associated with changes in Ig N-glycosylation and alterations in ER homeostasis and function in B cells, resembling the described phenotypes observed in our Jagn1^ΔB mouse model.

In this study, we identify a crucial role for the ER protein JAGN1 in humoral immunity. Deletion of Jagn1 in B cells led to defects in the differentiation and maintenance of long-lived BM plasma cells by causing increased ER stress and decreased antibody production and secretion, as well as marked alterations in the N-glycosylation of Ig. Consequently, Jagn1^ΔB mice exhibited reduced serum Ig levels at steady state and failed to mount an efficient humoral immune response upon infection with VSV or decreased antibody titers upon immunization with model antigens when compared with littermate control animals. Serum antibodies of Jagn1^ΔB mice displayed alterations in complement and Fc receptor binding, indicative of the functional relevance of their altered antibody glycosylation. Importantly, antibody glycosylation changes and increased ER stress in B cells observed in this mouse model were also detected in human patients carrying JAGN1 mutations, highlighting the relevance of these findings for basic B cell biology and extending our understanding of the observed human rare-disease phenotype.

JAGN1 mutations have been described to alter ER function in the process of neutrophil granule formation and the glycosylation of neutrophil effector molecules (Boztug et al., 2014). We therefore hypothesized that JAGN1 might also affect B cells, which require extensive ER remodeling as an adaptation to produce and secrete massive amounts of glycosylated antibodies (Hibi and Dosch, 1986). Our data show that loss of JAGN1 in B cells results in profound alterations in Ig glycosylation. The complex carbohydrate structures linked to Asn297 of IgG do not affect antigen binding but rather modulate downstream effector functions by altering the conformation and biological properties of the Fc region, which acts as the prime interaction interface between the adaptive and innate immune system (Woof and Burton, 2004). Following Fc binding to Fc receptors on innate immune cells, nonclassical Fc receptors, and complement proteins, activation of these receptors can drive ADCC, CDC, phagocytosis, or neutrophil activation, all of which are severely compromised for nonglycosylated IgG (Jefferis, 2009). Our data now identify the ER protein JAGN1 as a novel and critical regulator of antibody glycosylation.

The B cell phenotypes observed in our study regarding Ig levels are reminiscent of genetic impairment of the UPR in Ire1-deficient B cells. We now also found altered N-glycosylation of IgG secreted by Ire1 mutant B cells. The concept that inhibiting N-glycosylation causes an accumulation of mis- or unfolded glycoproteins, thereby leading to ER stress, was established a long time ago (Zinszner et al., 1998; Koizumi et al., 1999; Brewer et al., 1999; Caspersen et al., 2000). Here, for the first time, we provide evidence that this crosstalk is bidirectional, where increased ER stress can also lead to altered N-glycosylation. It is therefore tempting to speculate that the Ig of Jagn1^ΔB mice or antibodies generated during increased ER stress (as observed in Ire1^Δhem mice) might alter their effector functions and lead to increased ADCC and CDC. Glycosylation changes under elevated ER stress might be a regulatory mechanism used, for instance, during severe infections. To relieve ER stress, IRE1 oligomerizes and thereby autoactivates its RNase domain and splices Xbp1 (Yoshida et al., 2001). The spliced Xbp1 (Xbp1s) mRNA encodes a functional transcription factor, which regulates a variety of genes encoding chaperones and enzymes involved in protein folding and lipid synthesis (Lee et al., 2003; Sriburi et al., 2004; McGehee et al., 2009), thereby contributing to the ER expansion and increased Ig production in plasma cells. Of note, deficiency of Jagn1 has more severe phenotypic consequences than deletion of Xbp1, which leads to a reduced density and organization of the ER as well as decreased antibody secretion, but not to a reduction in antibody-secreting cells (Taubenheim et al., 2012). Interestingly, we found that loss of Jagn1 does not only result in increased ER stress, but also affects the Golgi apparatus, where IgG fucosylation and sialylation occur. This is in accordance with a previous study showing that JAGN1 interacts with members of the coat protein complex I, which plays an important role in vesicular trafficking between Golgi and ER (Boztug et al., 2014), which might explain the observed N-glycosylation changes.

Technological advances, especially in DNA sequencing, have vastly increased our understanding of the genetic basis of thousands of monogenic disorders. Mutations in JAGN1 have been identified to be causative for severe congenital neutropenia by affecting the differentiation and survival of mature neutrophils (Boztug et al., 2014). Importantly, mutations in JAGN1 are associated with considerable phenotypic variance. Additionally, defects in humoral immunity have been described for some patients (Baris et al., 2015). This phenotypic variability in genetic disorders could be a consequence of both differences in disease-causing mutations and genetic backgrounds as well as exposure to different pathogens. Indeed, differences in genetic backgrounds have been shown to compensate for deleterious genetic defects in human patients (Chen et al., 2016). Further increasing the complexity of genetic disorders of the immune system is the concerted contribution of different cell types to immune responses. Disentangling cause from consequence is therefore often difficult. In this study, we address these challenges by generating a model allowing us to decipher the phenotypic consequences of Jagn1 mutations in humoral immunity. Importantly, phenotypic defects in humoral immunity—altered antibody glycosylation and increased ER stress in B cells—were observed both in our murine model system as well as in human rare-disease patients. Our data uncover a novel path required for effective antibody-mediated immunity.

Jagn1fl/fl (Jagn1tm1c(EUCOMM)Hmgu), Ire1fl/fl (Ern1tm1Tiw), Mb1-Cre (Cd79atm1(cre)Reth), Cd23-Cre (Tg(Fcer2a-cre)5Mbu9), Vav1-Cre (Tg(Vav1-cre)A2Kio/J), Blimp1-GFP (Prdm1tm1Nutt), and Vav-Bcl2 (Tg(VavBCL2)1Jad) mice have been described previously and were bred on a C57BL/6 genetic background (Wirnsberger et al., 2014; Bettigole et al., 2015; Hobeika et al., 2006; Kwon et al., 2008; Kallies et al., 2004; Ogilvy et al., 1999). Rosa26-EGFP mice were generated by insertion of the gene encoding EGFP into the ubiquitously expressed Rosa26 locus. Mouse genotypes were assessed by PCR (see Table S1 for genotyping primer sequences). Of note, only age- and sex-matched littermates from respective crosses were used for experiments. If not explicitly stated, experiments were performed 10–14 wk after birth. All mice were bred and maintained in accordance with ethical animal license protocols of the respective legal authorities. All animal experiments were approved by the Austrian Federal Ministry of Education, Science, and Research.

For antibody analyses, serum was prepared using serum-separating tubes (BD Biosciences) according to the manufacturer’s instructions. IgG2b, IgG2c, IgG3, IgA, IgE, and IgM serum antibody titers were quantified using the Luminex xMAP technology-based Antibody Isotyping 7-Plex Mouse ProcartaPlex Panel 2 (ThermoFisher Scientific) according to the manufacturer’s instructions. IgG1 serum antibody titers were determined by ELISA by using plates coated with anti-mouse IgG1. Mouse IgG1 antibodies were used in defined concentrations as control standards. For glycosylation analyses, ∼1 µg of human or murine serum IgG was purified using protein G Dynabeads (ThermoFisher Scientific) and separated by SDS-PAGE. The Coomassie-stained SDS-PAGE bands representing the IgG heavy chain polypeptides were excised, destained, reduced, alkylated, and in-gel digested with trypsin (Trypsin Gold sequencing grade; Promega) as described previously (Stadlmann et al., 2008, 2009; Maresch and Altmann, 2016). Tryptic peptides and glycopeptides were separated by reversed-phase nano-HPLC (Acclaim PepMap C18, 75 µm × 50 cm; ThermoFisher Scientific) and analyzed by ESI-MS/MS (Q Exactive Hybrid Quadrupole-Orbitrap; ThermoFisher Scientific). The relative abundances of IgG subclass–specific glycopeptide glycoforms (Maresch and Altmann, 2016) were calculated from the manual integration (FreeStyle 1.3; ThermoFisher Scientific) of the respective extracted ion chromatographs.

Mice were immunized intraperitoneally with 100 µg OVA (Sigma) adsorbed to 200 µl of alum as adjuvant. OVA-specific serum Ig titers were determined by ELISA of serial serum dilutions using 96-well plates (Maxisorp; Nunc) coated with 50 µg/ml OVA. Titers were determined using alkaline phosphatase–linked anti-IgG and anti-IgM detection antibodies (SouthernBiotech). Humoral responses to a T cell–dependent antigen were studied by intraperitoneal injection of 100 µg of 4-hydroxy-3-nitrophenylacetyl–conjugated KLH (NP-KLH)/alum. The frequencies of antibody-secreting cells in the spleen were determined by ELISPOT assays as described previously (Smith et al., 1997). Briefly, NP23-bio–coated plates were used for capturing antibodies secreted by individual cells. Spots were visualized with goat anti-mouse IgM or IgG antibodies conjugated to alkaline phosphatase (Southern Biotechnology Associates), and color was developed by the addition of BCIP/NBT Plus solution (Southern Biotechnology Associates). After extensive washing, spots were counted with an AID ELISPOT reader system (Autoimmun Diagnostika). The same procedure was used for the in vitro–cultured iGB cells, for which the ELISPOT plates were coated with anti-mouse IgE or IgG1 antibodies. 200 cells were sorted directly on the plates, which were either CD19⁺CD138⁺IgG1⁺ or CD19⁺CD138⁺IgE⁺, and were then incubated for 6 h. The area of the ELISPOTs was determined using the Fiji software ImageJ (version 2.0.0-rc65/1.51s).

Mice were infected with 2 × 10⁶ PFU of VSV-containing full-length OVA (seed stock) via tail vein injection (Turner et al., 2008). Mice were bled at 3, 7, and 10 d after infection, and blood samples were centrifuged at 10,000 g for 5 min to collect serum. VSV-neutralizing antibodies in the serum were measured using a VSV neutralization assay. Briefly, 5 µl of serum was left untreated for total Ig or mixed with 5 µl of 0.1 M 2-mercaptoethanol for 1 h for IgG measurement (all complex isoforms were reduced and disassembled under this condition), subsequently diluted 1:40 in MEM containing 2% fetal calf serum, and heat inactivated for 30 min at 56°C. Twofold serial dilutions of serum samples were then mixed with 500 PFU/ml VSV, incubated for 90 min, and then added to monolayers of Vero cells. The cells were infected for 60 min in an incubator at 37°C and 5% CO₂ and then overlaid with methylcellulose and incubated for 24 h at 37°C and 5% CO₂. Subsequently, cells were stained with crystal violet stain, and titers were scored by the highest serum dilution, reducing the plaque numbers by 50% (-log₂x40).

To generate BM chimeric mice, 8-wk-old Rag2-deficient mice were split dose irradiated (2 × 5.5 Gy) and reconstituted 18 h after the second irradiation with mixtures of B cell/T cell/NK cell–depleted donor cells using magnetic cell separation (MACS; Miltenyi Biotec; two million Rosa-EGFP and two million control/Jagn1^ΔB lineage–depleted donor cells per mouse) by intravenous injection. Experiments were performed 4 mo after reconstitution.

Antibody labeling of cells after ammonium-chloride-potassium lysis of red blood cells was performed in FACS buffer (PBS supplemented with 2% fetal calf serum and 2 mM EDTA) on ice for 30 min after blocking Fc receptors with CD16/CD32 Fc block (2.4G2; eBioscience) for 5 min to inhibit unspecific antibody binding. Cell viability was assessed with the Fixable Viability Dye eFluor780 (eBioscience). See Table S2 for a list of antibodies (BD Biosciences, Biolegend, or eBioscience) used in this study for flow cytometric analysis. Intracellular staining for JAGN1, IgM, and BLIMP1 was performed using the Foxp3 staining kit (eBioscience) according to the manufacturer’s instructions. Cells were recorded on an LSR II flow cytometer (BD Biosciences), and data were analyzed using FlowJo v10.0.6 software (Tree Star). For viability analysis, cells were stained with Annexin V-BV605 (BD Biosciences) and PI (eBioScience) in Annexin V binding buffer (BD Biosciences) according to the manufacturer’s instructions. For labeling of lysosomes, living cells were incubated in medium containing 50 nM LysoTracker Deep Red (ThermoFisher Scientific) for 30 min at 37°C, washed, stained with additional antibodies to distinguish the different cell populations, and immediately recorded using flow cytometry. For labeling of the Golgi apparatus, living cells were labeled with the Golgi/ER staining kit Cytopainter (Abcam) according to the manufacturer’s instructions, stained with additional antibodies to distinguish the different cell populations, and immediately recorded using flow cytometry.

We performed a knockdown of endogenous Jagn1 in the IgG2b-producing mouse B lymphocyte plasmacytoma cell line MPC-11 (ATCC CCL-167; BALB/c background). Additional expression of full-length JAGN1 fused to an amino-terminal V5 tag allowed immunofluorescence detection of V5-tagged JAGN1 in these cells. Transfected MPC-11 cells were washed in HBSS and stained with ER-Tracker Green (ThermoFisher Scientific; GenBank accession no. E34251, 1:3,000 in HBSS) for 30 min at 37°C. Following washing, cells were seeded on poly-D-lysine–coated coverslips and fixed for 2.5 min with 4% paraformaldehyde at 4°C. After blocking Fc receptors with CD16/CD32 Fc block (eBioscience) for 5 min, cells were permeabilized with permeabilization buffer (eBioscience) and incubated with the V5-555 antibody (1:100 in permeabilization buffer). After washing in permeabilization buffer and subsequently in FACS buffer, nuclei were counterstained with DAPI, and coverslips were mounted in fluorescent mounting medium (Dako). Images were acquired with an LSM 780 Axio Imager 2 confocal microscope using a 63× objective.

For immunohistochemistry, spleens were fixed in PBS with 4% paraformaldehyde for 1 h at 4°C and cryo-protected in PBS with 30% sucrose before being embedded in optimal cutting temperature compound (Sakura). Frozen spleens were sectioned (14-µm thickness) on a microtome and fixed in acetone. For immunostaining, all incubations were performed in a humidified chamber at room temperature. Sections were washed and blocked in FACS buffer and CD16/CD32 Fc block (eBioscience) for 15 min. The following antibodies and reagents were used for immunostaining: anti–CD21-FITC (BD Biosciences), anti–CD23-APC (Biolegend), anti–MOMA1-biotin (Abcam), and Streptavidin-PE (Biolegend). Images were acquired with an LSM 780 Axio Imager 2 confocal microscope using a 25× objective and a 4 × 4 tile scan.

For LPS-induced plasmablast differentiation assays, splenic B220⁺ B cells were enriched using MACS technology (Miltenyi Biotec) according to the manufacturer’s instructions. Purified B cells were plated at a density of 150,000 cells/ml in IMDM (Life Technologies) supplemented with 10% fetal calf serum (Sigma), 50 µM 2-mercaptoethanol, 2 mM glutamine (Gibco), 100 U/ml penicillin (Sigma), and 100 µg/ml streptomycin sulfate (Sigma) and were subsequently treated for up to 4 d with 25 µg/ml LPS (Sigma). For the iGB cell differentiation system (Nojima et al., 2011; Wöhner et al., 2016), 40LB stromal cells (cultured in DMEM supplemented with 10% fetal calf serum) were inactivated by irradiation. 100,000 splenic B cells, which were enriched using CD43 MACS beads (Miltenyi Biotec), were plated onto the 40LB cells in 6-well plates in RPMI medium (supplemented with 10% fetal calf serum [ThermoFisher Scientific; lot42A0368K], 1 mM glutamine, 50 µM 2-mercaptoethanol, 10 mM Hepes, and 1 mM sodium pyruvate) and stimulated with recombinant IL-4 (20 ng/ml) for 4 d. At day 4, 100,000 stimulated B cells were transferred onto fresh 40LB cells in a 6-well plate and stimulated with recombinant IL-21 (10 ng/ml) for another 4 d until cells were analyzed by flow cytometry.

Tissues and cells were isolated and homogenized in TRIzol reagent (Invitrogen). Total RNA was isolated according to the manufacturer’s instructions. RNA was reverse transcribed with the iScript cDNA synthesis kit (Bio-Rad). Real-time PCR analysis was performed with GoTaq qPCR master mix (Promega) on a CFX384 system (Bio-Rad). Data were normalized to values for the housekeeping gene Gapdh for mouse samples and ACTB for human samples. All primers used in this study are listed in Table S1. Mouse Xbp1s and Xbp1u mRNA levels were monitored by semiquantitative RT-PCR using cDNA synthesized from RNA as described above with OneTaq Hot Start PCR Reaction Mix (New England Biolabs) for 26 cycles (see Table S1 for RT-PCR primer sequences). PCR products were resolved by 3% agarose gel electrophoresis.

cDNA libraries for each of the samples were generated from 500 ng of total RNA using the QuantSeq 3′ mRNA-Seq Library Prep Kit (FWD) for Illumina (Lexogen) following the manufacturer’s instructions. QuantSeq generates highly strand-specific next-generation sequencing libraries close to the 3′ end of polyadenylated RNA. Briefly, the first cDNA strand was generated through reverse transcription initiated by oligo(dT) priming. Synthesis of the second cDNA strand was performed by random priming in a manner that DNA polymerase was efficiently stopped when reaching the next hybridized random primer, so only the fragment closest to the 3′ end got extended until the end and got both the adapter sequences necessary for PCR amplification. SR100 sequencing was performed on HiSeq 2500 at the Vienna BioCenter Core Facilities (VBCF) Next-Generation Sequencing Facility using v4 SBS reagents. Trimming was performed using BBDuk v36.92 (ref = polyA.fa.gz,truseq.fa.gz k = 13 ktrim = r forcetrimleft = 11 useshortkmers = t mink = 5 qtrim = t trimq = 10 minlength = 30). Reads mapping to mouse ribosomal RNA transcripts (refseq) were removed using bwa v0.7.12 alignment and samtools v1.3.1. Remaining reads were aligned to the mouse genome mm10 using TopHat v2.1.1 (parameter: –library-type fr-secondstrand), and reads in genes were counted with htseq-count v0.6.1 (parameter: -s yes; reference data: illumina iGenomes, UCSC; mm10). DESeq2 v1.18.1 in R v3.4.0 was used to perform differential expression analysis using recommended minimal gene prefiltering (keeping genes with at least 10 reads total). Counts per million were computed using edgeR v3.20.1; counts per million gene expression values were transformed into log space and scaled (by gene) to mean 0 and unit standard deviation before heatmap visualization. Enriched GO terms were determined using DAVID v6.7 (https://david.ncifcrf.gov). The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO Series accession no. GSE141629.

For protein extraction, cells or tissues were manually homogenized in Hunt buffer (20 mM Tris-HCl, pH 8.0, 100 mM sodium chloride, 1 mM EDTA, and 0.5% NP-40) supplemented with Halt protease/phosphatase inhibitor cocktail (ThermoFisher Scientific). After full-speed centrifugation, the supernatant containing the soluble protein fraction was further used. Equal amounts of 20–30 µg of proteins were separated by SDS-PAGE and transferred onto polyvinylidene fluoride membranes (Immobilion-P, Merck Millipore) according to standard protocols. Blots were blocked for 1 h with 5% milk in TBST (1× Tris-buffered saline and 0.1% Tween-20) and were then incubated overnight at 4°C with primary antibodies diluted in 5% milk in TBST. Blots were washed three times in TBST for 5 min and were then incubated with HRP-conjugated secondary anti-mouse-IgG-H&L chain (Promega) or anti-rabbit-IgG-F(ab′)2 (GE Healthcare) antibody for 1 h at room temperature, washed three times in TBST for 5 min, and visualized using enhanced chemiluminescence (GE Healthcare). See Table S2 for a list of antibodies used in this study. β-Actin was used to control for protein loading. IgM was detected by incubating blots for 1 h with HRP-conjugated secondary anti-mouse-IgG-H&L chain (GE Healthcare).

Electron microscopy was performed by the EM Facility of the VBCF, a member of the Vienna BioCenter, Austria. B220⁺ B cells were stimulated for 4 d with LPS and then pelleted for 3 min at 500 g, washed in prewarmed PBS, and then fixed using prewarmed 2.5% glutaraldehyde in in 0.1 mol/liter sodium phosphate buffer, pH 7.4, for 1 h. Samples were then rinsed with the same buffer, post-fixed in 1% osmium tetroxide in 0.1 mol/liter sodium phosphate buffer, pH 7.4, on ice, dehydrated in a graded series of acetone on ice, and embedded in Agar 100 resin. 70-nm sections were cut parallel to the substrate and post-stained with 2% uranyl acetate and Reynolds lead citrate. Sections were examined with a FEI Morgagni 268D operated at 80 kV. Images were acquired using an 11-megapixel Morada CCD camera (Olympus-SIS).

For serum antibody-binding assays, BM from femurs and tibias of Rag2-deficient mice was isolated. After red blood cell lysis (ammonium-chloride-potassium buffer incubation), two million cells were incubated with the indicated dilutions from control or Jagn1^ΔB sera for 10 min on ice with and without previously blocking Fc receptors with Fc-blocking antibodies (eBioscience). After two washing steps in PBS, cells were stained with anti-mouse IgG F(ab)₂-PE (eBioscience; 1:1,000 in PBS) for 15 min and then washed again in PBS. After blocking Fc receptors with CD16/CD32 Fc block (eBioscience) in all samples, cells were stained with Ly6G-FITC or Ly6G-PE-Cy7 and CD11b-BV510 (all Biolegend; 1:500), washed in FACS buffer, filtered, and analyzed for IgG binding to Fc receptors on CD11b⁺Ly6G⁺ neutrophils and CD11b⁺Ly6G⁻ myeloid cells, including monocytes.

E0771 breast cancer cells were orthotopically injected in control and Jagn1^ΔB mice as previously described (Ewens et al., 2005). In brief, cells were harvested for injection into mice by trypsin digestion for 5 min, washed in HBSS, counted, diluted in this salt solution, and orthotopically injected into the fat pad of the mammary gland (250,000 cells/100 µl 1:1 diluted in Matrigel [Corning]). Sera of tumor-bearing mice were harvested before they were sacrificed following the ethical guidelines. For the ADCC experiments, 10,000 E0771 cells were plated in a total of 50 µl RPMI-based medium (supplemented with 10% fetal calf serum, 1 mM glutamine, 50 µM 2-mercaptoethanol, 10 mM Hepes, 1 mM sodium pyruvate, 100 U/ml penicillin, and 100 µg/ml streptomycin sulfate) per well in 96-well plates containing either 20 µl serum or 1 µg purified IgG from control and Jagn1^ΔB E0771 tumor–bearing mice. After 2-h opsonization, either Rag2-deficient BM cells (containing 50% neutrophils) or freshly isolated Ly6G-enriched BM neutrophils using MACS (Miltenyi Biotec) were added at increasing E/T ratios for 4 h. Cell viability of Ly6G⁻ FSC^high tumor cells was assessed with the Fixable Viability Dye eFluor780 (eBioscience), and IgG binding to tumor cells was confirmed by staining with anti-mouse IgG F(ab)₂-PE (eBioscience).

To generate Epstein-Barr virus–transformed lymphoblastoid cell lines, B cells from healthy individuals and JAGN1 mutant patients were isolated and infected with Epstein-Barr virus as described previously (Boztug et al., 2014). Experiments with human biological samples were done upon informed consent and approval by the Ludwig Maximilians University ethics board, respecting current regulatory frameworks.

The genomic proportion of C57BL/6J, 129/Sv, and BALB/c was determined by simple sequence length polymorphisms (also called microsatellites)–based screening of the mouse genetic background. The PCR screening was achieved by using sets of ∼92 markers evenly dispersed at a distance of 20 cM (∼20–40 Mbases) over the 19 autosomal chromosomes. To analyze PCR products, capillary-based electrophoresis runs on a fragment analyzer were used.

All values are given as means ± SD. Comparisons between two groups were analyzed using unpaired Student’s t tests and corrected for multiple comparisons using the Holm-Sidak method. Comparisons between groups over different time points (Fig. 1 D and,Fig. S5 A) or dilutions (Fig. 1 B) were analyzed by two-way ANOVA with Bonferroni correction. Comparisons between four groups (Fig. 4 G) were analyzed by one-way ANOVA followed by Bonferroni correction. P values were calculated with GraphPad Prism software: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Fig. S1 excludes Mb1-Cre effects and confirms the Jagn1 KO phenotype using Cd23-Cre–mediated deletion of Jagn1.Fig. S2 shows assessment of Jagn1 deletion, JAGN1 subcellular localization, and characterization of Jagn1^ΔB plasma cells. Fig. S3 illustrates phenotypic effects of Jagn1 deletion on B cell homeostasis and plasma cell levels in mixed genetic backgrounds. Fig. S4 reports differentially expressed genes, affected organelles, and IgM levels in in vitro–generated Jagn1-deficient plasmablasts. Fig. S5 shows in vivo stability, target cell binding, and complement C3 binding of control versus Jagn1^ΔB antibodies. Table S1 lists primers used for genotyping and gene expression analysis in this study. Table S2 provides details of the antibodies used in this study.

We thank all members of the Penninger laboratory for helpful discussions and technical support. Moreover, we thank all members of the Research Institute of Molecular Pathology–Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA) bio-optics service facility for assistance in cell sorting as well as the Molecular Biology Service, VBCF Electron Microscopy Facility, the VBCF Next-Generation Sequencing Facility, and the VBCF Histopathology Facility. We also thank Laurie Glimcher (Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA) and her laboratory for kindly providing the Ire1^Δhem mouse sera. Moreover, we would like to thank Sabine Jurado, Rene Rauschmeier, and other members of the Busslinger laboratory for their assistance and advice. Furthermore, we thank Christian Theussl for cryopreservation of mouse lines, in vitro fertilization, and rederivation of mice from frozen sperm.

J.M. Penninger is supported by grants from IMBA, the Austrian Ministry of Sciences, the Austrian Academy of Sciences, the T. von Zastrow Foundation, the FWF Wittgenstein award (Z 271-B19), and a Canada 150 Chair.

Author contributions: Immune phenotyping, generation of BM chimeras, and flow cytometry experiments were performed and analyzed by A. Hagelkruys and G. Wirnsberger. The OVA immunization was carried out and analyzed by L. Tortola and G. Wirnsberger, the NP-KLH immunization by M. Wöhner and A. Hagelkruys, and the VSV infections by B. Vilagos and A. Bergthaler. J. Stadlmann performed all the MS analyses. Ig isotyping, immunofluorescence, immunohistochemistry, quantitative PCR analysis, immunoblots, and in vitro LPS-stimulated plasmablast differentiations were performed by A. Hagelkruys. M. Wöhner and A. Hagelkruys carried out the iGB experiments, and M. Wöhner performed the ELISPOT assays and IgG1 ELISAs. M. Horrer, G. Jonsson, and M. Kogler helped with immunophenotyping and flow cytometry. R. Koglgruber assisted with mouse genotyping. D. Hoffmann prepared the libraries for the QuantSeq experiment; M. Novatchkova analyzed the sequencing data. P. Bönelt and M. Busslinger generated the Rosa26-EGFP mice used in this study. U. Steffen and G. Schett performed experiments with human macrophages. C. Klein designed the experiments with human patients and provided samples. A. Hagelkruys, G. Wirnsberger, and J.M. Penninger conceptualized and designed the experiments and wrote the manuscript with input from the co-authors.