CD83 is a maturation marker for dendritic cells. In the B cell lineage, CD83 is expressed especially on activated B cells and on light zone B cells during the germinal center (GC) reaction. The function of CD83 during GC responses is unclear. CD83−/− mice have a strong reduction of CD4+ T cells, which makes it difficult to analyze a functional role of CD83 on B cells during GC responses. Therefore, in the present study we generated a B cell–specific CD83 conditional knockout (CD83 B-cKO) model. CD83 B-cKO B cells show defective upregulation of MHC class II and CD86 expression and impaired proliferation after different stimuli. Analyses of GC responses after immunization with various Ags revealed a characteristic shift in dark zone and light zone B cell numbers, with an increase of B cells in the dark zone of CD83 B-cKO mice. This effect was not accompanied by alterations in the level of IgG immune responses or by major differences in affinity maturation. However, an enhanced IgE response was observed in CD83 B-cKO mice. Additionally, we observed a strong competitive disadvantage of CD83-cKO B cells in GC responses in mixed bone marrow chimeras. Furthermore, infection of mice with Borrelia burgdorferi revealed a defect in bacterial clearance of CD83 B-cKO mice with a shift toward a Th2 response, indicated by a strong increase in IgE titers. Taken together, our results show that CD83 is important for B cell activation and modulates GC composition and IgE Ab responses in vivo.
The CD83 protein was introduced in 1992 as an Ig-like molecule that is expressed on Langerhans cells and activated lymphocytes (1). Even though CD83 was mainly used as a selective marker for mature dendritic cells (DCs) (2, 3), it is also expressed on a variety of other cell types, including thymus epithelial cells (4), T cells (5), and especially regulatory T cells (6, 7) and B cells (8). Interestingly, it was shown that CD83 expression on thymus epithelial cells is essential for CD4+ T cell development (4). The detailed mechanism for this observation is still unknown, but CD83−/− animals show a strong reduction in the CD4+ T cell compartment in the thymus as well as in the periphery (4).
On developing B lymphocytes CD83 expression is detectable beyond the pre–B cell stages once they express a functional BCR (9, 10). Naive B cells only express low levels of CD83. In contrast, after stimulation with anti-IgM or TLR ligands, CD83 is rapidly upregulated on B lymphocytes in vitro and also upon immunization in vivo (10, 11). By generation of CD83-deficient mice, Fujimoto and colleagues (10) showed that CD83 expression is not essential for B cell development or tissue localization, but it is required for B cell longevity. When stimulating CD83−/− B cells with LPS and anti-IgM in vitro, an impaired CD86 and MHC class II (MHC II) upregulation was observed, whereas no differences regarding their proliferative potential were detected (4, 12). In contrast to studies using CD83−/− mice, animals overexpressing CD83 under the control of an MHC class I (MHC I) promoter (CD83 transgenic [CD83tg] mice) show increased CD86 and MHC II expression as well as higher amounts of IL-10 in supernatants of LPS-stimulated B cells (8). In general, the CD83tg mice showed stronger B cell phenotypes in vivo compared with the CD83−/− mice. These phenotypes included B cell maturation defects, defects in B cell Ca2+ signaling, suppression of serum Ig levels, reduced humoral responses to thymus-dependent and thymus-independent Ags, and reduced Ig responses during infections. Only some of these phenotypes were shown to be B cell intrinsic by mixed bone marrow chimeras. The functional role of CD83 on B cells during humoral immune responses to thymus-dependent Ags could not be studied in CD83−/− mice owing to their strong Th cell defect caused by decreased CD4+ T cell numbers. No influence of a B cell CD83 deficiency on the humoral responses was detected when transfer experiments with CD83−/− bone marrow were performed (13).
Expression of CD83 on germinal center (GC) lymphocytes was first shown in 1992 (1), and since 2012 it has been used as a marker for light zone (LZ) B cells during GC reactions (14). GC formation takes place after Ag stimulation in lymphoid organs (15). GCs are the site of somatic hypermutation (SHM) and class switch recombination leading to the production of high-affinity Ab-secreting plasma and memory B cells (16). One anatomical feature of the GC is its division into dark zone (DZ) and LZ. The DZ is the site where B cells strongly proliferate and SHM occurs, whereas the LZ is the zone where selection for B cells with high-affinity BCRs with the help of follicular DCs and T follicular helper (TFH) cells takes place (15). Victora et al. (14) showed a distinct upregulation of CD83 specifically in LZ B cells, whereas DZ B cells express high CXCR4 levels. Additionally, a recent study analyzing secondary GCs on a transcriptional level revealed CD83 expression in two of four individual clusters of class-switched B cells in the LZ. CD83 as an LZ marker and the DNA polymerase Polη as a DZ marker segregated the secondary GC transcriptional program into four stages that regulated divergent mechanisms of memory BCR evolution. Distinct changes in expression pattern including CD83 may be important for distinguishing reentry into the DZ from ongoing BCR rediversification (17). CD83 on LZ B cells is thought to be involved in Ag presentation to TFH cells, a process that is important for selection (18).
CD83 expression is not only important as a transmembrane protein on different immune cells, but also a soluble CD83 (sCD83) molecule can be found. sCD83 is generated from the extracellular domain of the membrane-bound CD83 and can be detected in human serum (19). This form of sCD83 possesses immunomodulatory effects, as recombinant sCD83 inhibits DC maturation and subsequently DC-mediated T cell stimulation (20). Application of sCD83 into mice reduces severity of autoimmune diseases and induces prolonged graft survival (21–23).
However, knowledge regarding distinct mechanisms of the CD83 molecules expressed on different cell types remains still elusive. Previous mouse models were hampered by the fact that in the reported CD83tg mice, a broad CD83 overexpression was obtained on almost all cell types due to the fact that CD83 expression was driven under the control of an MHC I promoter (11). Even in bone marrow chimeras it cannot be excluded that strong B cell phenotypes are also caused by increasing sCD83 levels in the mice. In vivo B cell functions are hard to study in CD83−/− mice owing to the strong CD4+ T cell defect that affects humoral Ig responses. To gain deeper insights in the role of CD83 on B cells, we therefore generated B cell–specific CD83 knockout (KO) animals. In these mice, we observed characteristic B cell developmental blocks affecting both B1 cells as well as marginal zone (MZ) B cells. In vitro, CD83−/− B cells showed a strong impact of loss of CD83 on CD86 and MHC II expression after stimulation. Additionally, during GC responses, changes in GC B cell numbers, as well as a characteristic shift of DZ and LZ B cell numbers, were observed in B cell–specific CD83 KO animals. Furthermore, infection of mice with Borrelia burgdorferi revealed a defect in bacterial clearance of B cell–specific CD83 conditional KO (CD83 B-cKO) mice with a shift toward a Th2 response, underlined by a strong increase in IgE titers.
Materials and Methods
Generation of B cell CD83-deficient mice
A targeting vector containing floxed exons 1 and 2 of the cd83 gene was generated by PCR cloning. Three DNA fragments were cloned into the pRAPIDflirt vector (24). The following primes were used for PCR amplifying the CD83 fragments: short arm, 5′-TGCGGCCGCATGTCCAGTAAGACAAAC-3′ and 5′-TGGATCCTGAAACCCAGGGCTGTG-3′; exon 1 and 2 fragment, 5′-TCTCGAGGAGCTGCCCACCCTATC-3′ and 5′-TGTCGACGAGCCAATAGCGAGCCT-3′, long arm, 5′-TGTCGACTGCCACCTTGCTTCTTCATG-3′ and 5′-TCTCGAGTTCCAGGCAGTGACAGAACC-3′. All primer contained restriction sites at their 5′ end. JM8A3 embryonic stem (ES) cells were electroporated with the linearized targeting vector.
Four positive ES cell clones were identified and injected into blastocysts. Blastocysts were transferred into pseudopregnant females, and resulting offspring was analyzed with PCR to identify correct germline mutations. Those were bred with Flp mice to deplete the neo-cassette and then bred to homozygosity. For B cell–specific depletion of the cd83 gene, floxed CD83 animals were crossed with CD19-cre mice (CD83 B-cKO) (25). All mice were on a C57BL/6 background. Age-matched control and B cell–specific CD83−/− mice of 8–14 wk of age were used for analysis, except for adoptive transfer experiments where 20-wk-old recipient mice were used.
RNA preparation and PCR
RNA from single-cell suspension was prepared using an RNeasy Plus mini kit (Qiagen) according to the manufacturer’s instructions. RNA yield was measured using a spectrophotometer (NanoDrop 2000c, Thermo Scientific). For cDNA synthesis, 1 μg RNA was transcribed using a first-strand cDNA synthesis kit (Fermentas) as specified by the manufacturer. CD83 gene expression was determined using PCR, and PCR products were analyzed using gel electrophoresis.
Single-cell suspensions from spleen, bone marrow, peritoneal cavity, and blood were stained in PBS. Staining was done in ice-cold PBS containing Live/Dead fixable aqua dead cell stain (Life Technologies) using the following Abs (conjugated with biotin, FITC, PE, PE-Cy7, allophycocyanin, PerCP, allophycocyanin-Cy7, BV421, and Pacific Blue): anti-B220 (RA3-6B2; BD Biosciences), anti-CXCR4 (2B11; eBioscience), anti-Fas (15A7; eBioscience), anti-GL7 (GL7; BioLegend), anti-IgM (RMM1; BioLegend), anti–MHC II (I-A/I-E; M5/114; BD Biosciences), anti-CD4 (RM4-5; BD Biosciences), anti-CD5 (53-7-3; BD Biosciences), anti-CD8 (53-6.7; BD Biosciences), anti-CD19 (6D5; BioLegend) anti-CD21 (7E9; BioLegend), anti-CD23, anti-CD45.1 (104; BioLegend), anti-CD45.2, anti-CD69 (H1.2F3; BioLegend), anti-CD83 (Michel-19; BD Biosciences), and anti-CD86 (Gl-1; BD Biosciences). Biotinylated Abs were further stained with streptavidin-FITC or PerCP-Cy5.5. Detection of cell surface marker expression was performed with a flow cytometer and analyzed with FlowJo (Tree Star). Living lymphocytes were gated for further analyses.
TLR ligand stimulation and cytometric bead array analysis
Splenic cells were sorted with anti-CD19 beads (Miltenyi Biotec), and obtained B cells were cultivated at a density of 106 cells/ml for FACS analyses or 5 × 105 cells/ml for cytokine analysis. Cells were stimulated for 24–72 h with LPS, CpG, R848, and anti-IgM. Cells were then harvested for further FACS analyses and supernatants were used for measuring cytokine levels using cytometric bead array (BD Biosciences) according to the manufacturer’s instructions.
B cells were purified from RBC-lysed splenic single-cell suspensions by magnetic separation with anti-CD19 beads (Miltenyi Biotec), labeled with CellTrace Violet dye (Molecular Probes/Thermo Fisher Scientific) according to the manufacturer’s instructions and cultivated at a density of 105 cells/ml. After 72, 96, and 120 h of stimulation with different concentrations of LPS, R848, and CpG (InvivoGen), proliferation was analyzed using a flow cytometer and FlowJo (Tree Star).
In vitro class switch analysis
Splenic cell were sorted with anti-CD19 beads (Miltenyi Biotec), and obtained B cells were cultivated at a density of 106 cells/ml with 10 μg/ml LPS and 10 ng/ml IL-4 for 6 d. Supernatants were analyzed for Ig secretion using ELISA.
Immunizations and ELISA
CD83fl/fl × CD19cre/+ and control mice were i.p. immunized with 100 μg 4-hydroxy-3-nitrophenylacetyl (NP)–keyhole limpet hemocyanin (KLH) or 50 μg NP– chicken γ-globulin (CGG) in aluminum hydroxide (alum) with PBS for analysis of T cell–dependent immune responses. Blood was taken at days 0, 7, 14, 21, and 42. At day 63 after first immunization, mice were immunized again with 100 μg NP-KLH and blood was taken at days 70, 77, and 84. For analysis of GC responses against SRBCs, mice were injected i.p. with 109 SRBCs/mouse. Animals were sacrificed 6, 10, or 12 d after immunization (as indicated in figures). Ig serum titers were measured using standard ELISA methods (26). In brief, MaxiSorp plates (Nunc) were coated with Ag (5 μg/ml NP-BSA) for detection of NP-specific Igs in sera of immunized mice. The sera were added in serial dilutions.
For bone marrow reconstitutions, a 1:1 mixture of purified wild-type (WT; CD45.1+) and CD83fl/fl × CD19cre/+ (CD45.2+) bone marrow cells was i.v. injected into lethally irradiated (11 Gy) Rag1−/− mice. Successful reconstitution of lymphocytes was analyzed 5 wk after transfer using blood samples by flow cytometry. Seven weeks after transfer, mice were immunized with SRBCs, and at days 6 and 12 following immunization, mice were sacrificed and the contribution of different donor cells in several lymphocyte populations was analyzed using anti-CD45.1 and anti-CD45.2 staining in combination with other markers.
Next generation sequencing and SHM analyses
To prepare samples for next generation sequencing (NGS), synthesized cDNA was used in a seminested PCR approach. As a forward primer, VH558 primers (5′-GRGCCTGGGRCTTCAGTGAAG-3′) containing multiplex identifier (MID) tags for individual mice and the A-Key for 454 sequencing were used. In the first PCR round, a Cγ1 outer reverse primer (5′-GGAAGGTGTGCACACCGCTGGAC-3′) and in the second round a Cγ1 inner primer (5′- GGCTCAGGGAAATAGCCCTTGAC-3′) with the B-Key for 454 sequencing were used. Protocols were established as previously described (27, 28). For purification of PCR products, a QiaQuick PCR purification kit (Qiagen) was used according to the manufacturer’s instructions. DNA yield was assessed using Qubit (Invitrogen) according to the manufacturer’s instructions. The 454 sequencing was performed on a Roche 454 platform. After splitting for MID tags for the different samples, individual FASTA files for each individual sample containing 4500–8000 high-quality reads were analyzed on the IMGT/HighV-QUEST platform (29). Sequences using the canonical V186.2 gene (resembling the VH1-72 gene according to the IMGT nomenclature) were subjected to a Bayesian estimation of Ag-driven selection using the BASELINe algorithm (30).
Infection with B. burgdorferi
Infection of mice with B. burgdorferi as well as measurement of ankle swelling, B. burgdorferi–specific Ab response, and bacterial load in organs were performed as previously described (31). Cytokine production was assessed using a LEGENDplex assay (BioLegend) following the manufacturer’s instructions.
Immunofluorescence images of cryosections of the spleen have been generated using the multi-epitope ligand cartography technique. Sample preparations from tissue, data generation, and analysis were performed as described previously (22). ImageJ software was used to determine the size of DZ and LZ within the GC.
Statistical analysis was carried out using the Mann–Whitney U test or the Student t test.
Calcium mobilization assay in primary mouse cells
Splenic cells were loaded with Indo-1 as described (32). Cells were stained with anti-CD5 and anti-B220 and analyzed using an LSR II (Becton Dickinson). Cells were stimulated with anti-IgM [F(ab′)2] (Jackson ImmunoResearch Laboratories).
Ag presentation assay
B cells were purified from RBC-lysed single-cell suspensions by magnetic separation with anti-CD19 beads (Miltenyi Biotec), and OT-II CD4+ T cells were purified with a CD4 purification kit (Miltenyi Biotec) according to the manufacturer’s instructions. The assay was performed as described (33). In brief, B cells were stained with anti-IgM F(ab′)2-biotin and subsequently with OVA Ag delivery agent (Miltenyi Biotec), containing an anti-biotin Ab conjugated to OVA and FITC, to deliver Ag via the BCR. Ag-loaded B cells were cocultured with OT-II CD4+ T cells, and T cell activation was assessed via the IL-2 production in the supernatants of cocultures using cytometric bead array (BD Biosciences).
Generation of B cell–specific CD83 KO animals
To delete CD83 in specific cell types, floxed CD83 mice have been generated. Therefore, exons 1 and 2 of the five exons of the CD83 gene were flanked with two loxP sites (floxed) and inserted into the genome of ES cells. Positive ES cell clones were injected into embryonic blastocysts and chimeras were generated (Supplemental Fig. 1). After germline transmission and deletion of the neomycin resistance cassette, floxed mice were used for further matings with CD19-Cre mice for B cell–specific deletion of CD83 (Fig. 1A). The successful KO of CD83 in B cells was examined using mRNA and FACS analysis. Because CD83 is weakly expressed on naive B cells, but upregulated on activated B cells (9), LPS-activated B cells were examined. The CD83 mRNA of isolated LPS-stimulated CD19+ B cells of CD83fl/fl × CD19Cre/+ (CD83 B-cKO) mice was strongly reduced. Quantification of the band intensity using the Bio1D software revealed a 30% level of the CD83 cDNA in CD83 B-cKO animals compared with CD83fl/fl control animals (Fig. 1B). FACS analysis by Ab staining after LPS stimulation of CD19+ B cells showed loss of CD83 surface expression on B cells of CD83 B-cKO mice, whereas floxed CD83 mice showed comparable upregulation of CD83 to WT animals (Fig. 1C). CD83 floxed mice were used as controls throughout all experiments, as it is known that CD19Cre/+ mice are phenotypically normal (25, 34). The remaining CD83 mRNA in CD83 B-cKO mice could be due to a noncomplete deletion of CD83 in CD19+ B cells by CD19-Cre, however the CD83 surface expression was lost.
Reduced splenic MZ B cell and reduced B1a cell numbers in peritoneal cavity of CD83 B-cKO mice
First, B cell development was examined in CD83 B-cKO mice. In the bone marrow, no changes in pro–B, pre–B immature, or mature B cells were detected (not shown). However, in the spleen we observed reduced transitional B cells type 2 and MZ B cell numbers in CD83 B-cKO mice compared with control animals (Fig. 2A). In contrast, mature follicular zone B cell numbers were normal in the spleen (Fig. 2B). The subpopulations of B-1 cells were found in normal numbers in the spleen (not shown), but in the peritoneal lavage of CD83 B-cKO mice a significant decrease in cell numbers of B1a and B2 cells could be detected compared with control mice (Fig. 2C). Thus, B1 or MZ B cells were affected by the loss of CD83, whereas mature follicular B cells were observed in normal numbers in the spleen.
Defective upregulation of MHC II and CD86 on activated CD83−/− B cells in vitro
We studied B cell signaling in CD83 B-cKO mice, because it was reported that the level of CD83 on B cells affects BCR signaling, including Ca2+ signaling (8). When we stimulated B cells of CD83 B-cKO mice with anti-IgM, we did not detect any changes in Ca2+ signaling (Supplemental Fig. 2A). As it was shown previously that CD83 expression affects the expression of the B cell activation markers MHC II and CD86 on B cells (11, 12), we separated splenic CD19+ cells using MACS technology and analyzed the expression of MHC II, CD86, and CD69 after stimulation with different stimuli, including the TLR ligands LPS, CpG, R848, or anti-IgM. In this study, a significant defective upregulation of MHC II and CD86 on CD83 B-cKO B cells 24 after stimulation independent of used stimuli could be detected (Fig. 3A). This observation was specific for MHC II and CD86, as CD69 expression was not affected. Additionally, the observed effect was still consistent after 48 h (not shown) and 72 h after stimulation (Fig. 3B), independent of the trigger for B cell stimulation.
CD19+ B cells were stimulated with LPS and IL-4 for 6 d and supernatants were examined for IgM and IgG1 levels using ELISA. Addition of IL-4 to B cell culture will lead to a class switch and to a shift to IgG1 Ab secretion compared with only stimulation with LPS. We could detect slightly reduced levels of IgM in supernatants of CD83 B-cKO B cells after LPS stimulation, but no changes in IgG1 levels after stimulation with both LPS and IL-4 (Fig. 3C). Additionally, serum Ig levels of the different Ig classes were unchanged in CD83 B-cKO mice (Supplemental Fig. 2B).
Higher IL-10 secretion and increased proliferative capacity of TLR ligand–stimulated B cells of CD83fl/fl × CD19cre/+ mice compared with control B cells
The loss of CD83 on B cells may also affect cytokine secretion and proliferation after TLR stimulation. Therefore, sorted splenic CD19+ B cells were analyzed after 72 h for secreted cytokines in their supernatants and the proliferation was measured using CellTrace Violet proliferation dye. Interestingly, we could show increased IL-10 secretion by CpG-stimulated CD83 B-cKO B cells compared with controls, whereas IL-6 and TNF-α production was not influenced (Fig. 4A). Not only is the cytokine secretion affected by loss of CD83, but also the proliferative capacity of B cells after stimulation with LPS, R848, and CpG at different concentrations and different time points was significantly increased (Fig. 4B, 4C).
Increased GC B cells and characteristic shift of DZ and LZ B cells numbers in CD83fl/fl × CD19cre/+ mice after immunization with SRBCs
Previous work has shown that CD83 is an important marker for LZ B cells within the GC reactions (14). This led us to hypothesize that CD83 itself might have a specific function during the GC reaction. To study a potential effect of B cell–specific loss of CD83 during the GC reaction, mice were immunized with SRBCs and at days 6 and 10, splenocytes were analyzed using flow cytometry. Remarkably, we could observe a significant increase in GC B cell numbers of CD83 B-cKO mice compared with controls at day 6 (Fig. 5A, 5B). Moreover, a more detailed investigation of the different zones within the GC revealed a relative increase in DZ B cells (CXCR4+) in CD83 B-cKO animals, whereas LZ B cells (CD86+) are relatively decreased at day 6 (Fig. 5C–E). This abnormality is in line with a significant increase in the DZ/LZ ratio. These changes were restricted to B cell populations within the GCs, whereas TFH cells were not influenced (not shown). GCs were also analyzed at day 10 after SRBC immunization, and the results of a shifted DZ/LZ ratio were similar to day 6 (Fig. 5H–J). Histology of the spleen confirmed these findings. In spleen sections of SRBC-immunized mice, the DZ of the GC was identified by the marker GL-7high, whereas the LZ was identified as CD21highGL-7low (Fig. 5F). When the size of the DZ and the LZ area was quantified with ImageJ software, a characteristic shift in the ratio of DZ/ LZ size in CD83 B-cKO mice was detected as well (Fig. 5G).
After immunization with NP-KLH, CD83 B-cKO mice show normal IgG but increased IgE responses and a characteristic shift of DZ and LZ B cells numbers
To study GC formation after NP-KLH immunization, animals were immunized with NP-KLH in alum and sera were tested for anti-NP–specific IgM and IgG1 by ELISA. Additionally, animals were boosted with NP-KLH at day 63 after first immunization and sera were analyzed by ELISA to study secondary immune responses. No alterations in the Ag-specific IgM or IgG1 response of CD83 B-cKO mice compared with control mice could be observed at the indicated time points. Also, no significant differences could be detected in the secondary response after boost at day 63 (Fig. 6A). Interestingly, total IgE titers and NP-specific IgE titers (Supplemental Fig. 4A) of CD83 B-cKO mice after NP-KLH were significantly increased at indicated time points and a general higher IgE response could be observed.
SRBC immunization leads to a very strong GC response, so the question arose whether immunization with a thymus-dependent Ag will lead to similar characteristic changes GC composition. To investigate this issue, mice were immunized with NP-CGG in alum and analyzed at days 6 (data not shown) and 12 after immunization. After immunization, GC B cell populations were characterized. FACS analysis studying the distribution of DZ and LZ B cells exposed similar findings in line with immunization with SRBCs. Again, DZ B cells are relatively increased and so is the DZ/LZ ratio (Fig. 6B). We also observed an impaired MHC II expression on GC B cells of CD83 B-cKO mice (Fig. 6C). The impaired MHC II expression may affect Ag presentation of CD83-deficient B cells. To study this, an Ag presentation assay was performed. B cells of CD83 B-cKO and control mice were loaded via the BCR with OVA and the response of OVA-specific OT-II T cells was measured. However, no difference was detected in IL-2 production or in T cell proliferation after Ag presentation by CD83-deficient or control B cells (Supplemental Fig. 3).
Normal Ag-driven selection of CD83 B-cKO GC B cells after immunization with NP-CGG
To understand whether differences in DZ and LZ B cell numbers have an influence on the Ag-driven selection within the GC, we analyzed the Ab VDJ repertoire of sorted GC cells using NGS. CD83 B-cKO and control mice were immunized with NP-CGG and 12 d after immunization, NP binding GC B cells were sorted. After RNA extraction and cDNA synthesis, PCR using a specific primer for the J558 VH gene family and Cγ1 primer was performed. The NGS sequence data are deposited at the National Center for Biotechnology Information SRA database (http://www.ncbi.nlm.nih.gov/sra, accession no. PRJNA310174). The obtained results were analyzed using IMGT/V-QUEST software (29). We focused our analysis on productive sequences of the VH186.2 gene for which a characteristic high-affinity anti-NP W33L mutation has been described repeatedly (35). On average, ∼60% of all sequences on day 12 GC B cells from five control and five CD83 B-cKO mice contained the W33L mutation, suggesting a similar affinity maturation in both types of mice (Fig. 6D). For a more detailed analysis, we subjected all VH186.2 sequences to a quantitative analysis of Ag-driven selection recently described (30). As shown in Fig. 6E, cumulative analysis of all sequences from individual mice clearly shows positive selection in the CDR regions as indicated by positive selection strength (Σ) values and negative or no selection in the framework regions. As the Σ values were comparable in control and CD83 B-cKO mice, Ag-driven selection is apparently not affected by the altered DZ/LZ ratio.
Competitive disadvantage of CD83−/− B cells after bone marrow transfer in RAG−/− animals
Despite the absence of significantly altered Ag-driven selection in the GCs of CD83 B-cKO mice, we wanted to gain further insight into the role of CD83 during GC reactions in a competitive setting between CD83−/− and WT B cells. Thus, BM chimeras were generated by mixing BM cells from both CD83 B-cKO and Ly5.1 control mice. CD83-cKO cells are positive for the congenic marker CD45.2, whereas Ly5.1 cells are positive for CD45.1. Bone marrow cells from both groups were mixed in a 50:50 ratio and injected i.v. in RAG−/− recipients. Five weeks after cell transfer, blood was taken and reconstitution of lymphocyte populations was analyzed using FACS. Interestingly, even in unchallenged, naive mice, a competitive disadvantage of CD83-cKO B cells in the blood could be observed. In contrast, T cell populations, that is, CD4+ and CD8+ T cells of the KO bone marrow, did not show reduced reconstitution (Fig. 7A). To examine whether this effect would persist after challenge by immunization, reconstituted bone marrow chimeras were immunized with SRBC, and B cell populations, as well as the GC reaction, were analyzed 6 and 12 d after immunization by flow cytometry. Strikingly, a disadvantage for CD83-cKO B cells could still be observed and the effect was even stronger within the GC B cell population (Fig. 7B). A strong reduction of CD83-cKO B cells within the GC B cell population was also found on day 12 after immunization, whereas the percentages of B220+ B cells are unaltered at the later time point (Fig. 7C). Within the GC B cells, there is also a strong decrease in both LZ as well as DZ B cells from CD83-cKO cells. A clear shift in the DZ/LZ ratio could not be observed. Only at day 12 could marginal differences be detected, with a tendency of more CD83-cKO B cells being present in the DZ compared with WT B cells (Fig. 7D). CD83 B-cKO bone marrow–derived CD4+ and CD8+ T cells in the spleen are significantly increased both at day 6 and day 12 postimmunization.
CD83 B-cKO mice show normal clearance of B. burgdorferi infection with a shift toward a Th2 response
To obtain insight in the role of CD83 on B lymphocytes during a bacterial infection in which Abs play an important role for protection, CD83 B-cKO mice were analyzed in the murine model of Lyme arthritis. Because B. burgdorferi elimination is completely dependent on adequate Ab production and because bacterial cells are strong B cell mitogens, this infection model is ideal for the study of B cell activation, GC reactions, and especially Ab responses (36). CD83 B-cKO and control animals were infected with B. burgdorferi and the clinical manifestation, bacterial clearance, and B cell responses were analyzed. Although CD83 B-cKO animals show comparable relative ankle swelling and thus no worsening in disease symptoms (Fig. 8A), their ability to control the pathogen is significantly impaired in the infected (i.e., right) tibiotarsal joint and by tendency is impaired in the heart (Fig. 8B). FACS analyses of GCs in the spleen 45 d postinfection revealed again a characteristic shift in the DZ/LZ ratio (Fig. 8C). Splenocytes, which were taken from infected mice at day 45 postinfection, were restimulated with Con A and supernatants were analyzed for cytokine secretion. Interestingly, a specific reduction of INF-γ was observed, whereas IL-6 or TNF-α, for example, was not altered (Fig. 8D). Induction of IL-4 or IL-5 could not be detected. Concomitant with decreased INF-γ production, analysis of sera of B. burgdorferi–infected mice using ELISA showed a strong increase in total IgE titers in CD83 B-cKO mice compared with controls at all eight time points after infection was observed (Fig. 8E). In contrast, specific IgG1 and IgG2c levels were comparable between CD83 B-cKO and control mice (Supplemental Fig. 4B).
CD83 B-cKO animals showed an overall normal B cell development in the bone marrow and normal B cell numbers in the periphery. CD83 is only weakly expressed on the surface of naive B cells (9) but is strongly upregulated on the surface of activated B cells, and therefore no strong changes on B cell maturation were expected upon the loss of CD83. Nevertheless, the reduction of MZ and B1a cell numbers has so far not been observed in mice with a total CD83 KO (10). However, mice carrying a point mutation in the CD83 gene, which affects CD83 expression, also show decreased MZ B cell numbers, whereas transgenic mice overexpressing CD83 show increased MZ numbers (9). Thus, the number of MZ B cells seems to be directly affected by the expression level of CD83.
MZ B cells belong to the innate-like B cells that produce natural IgM and are also involved in thymus-independent type 2 responses. Immunizations of CD83 B-cKO mice with thymus-independent type 2 Ags did not lead to changes in specific Ab responses (results not shown). Additionally, B1 cells from the spleen and peritoneal cavity also possess pronounced innate functional features and also cooperate with MZ B cells (37). Similar to MZ B cells, B1 cells produce natural Abs and they are the main producers of natural IgM (38). Despite the reduced numbers of MZ B cells and peritoneal B1a cells, we did not observe decreased levels of natural IgM in the serum of CD83 B-cKO mice. Apparently the normal numbers of B1 cells in spleen and other organs produce unchanged IgM levels.
It has been reported in CD83tg mice that higher expression of CD83 leads to impaired BCR signaling with impaired proximal tyrosine kinase signaling and decreased Ca2+ mobilization (11, 39). This led to the hypothesis that CD83 might be a negative regulator of BCR signaling. Our demonstration of normal Ca2+ signaling in B cells of CD83 B-cKO mice does not support this hypothesis. One possibility to explain these different results is the fact that an MHC I promoter was used for the CD83tg mice. Overexpression on other cell types may influence B cell functions or general overexpression of CD83 may lead to increased levels of sCD83, which is immunomodulatory (20, 23)
The defective upregulation of CD86 and MHC II after TLR stimulation in CD83 B-cKO mice was similarly observed in total CD83 KO mice (4, 12, 18), and CD83tg mice showed increased CD86 and MHC II expressions on LPS-stimulated B cells (8). A potential role of CD83 on MARCH1-dependent ubiquitination of MHC II and CD86 has been reported. In this study, the transmembrane domain of CD83 enhances MHC II and CD86 expression by blocking association of MHC II with MARCH1 (18). This mechanism could be an explanation for accelerated cell surface turnover of CD86 and MHC II in CD83 KO cells (12). As CD83 is important for stabilizing MHC II and CD86 on the cell surface, its loss would lead to diminished MHC II and CD86.
TLR stimulation not only affected upregulation of activation markers, but it also led to increased proliferation and increased IL-10 secretion. Changes in IL-10 levels were also observed in CD83tg mice, because LPS-stimulated B cells of these mice secrete higher amounts of IL-10 (8). In our CD83 B-cKO B cells, LPS-induced cytokine secretion was normal, whereas CpG-induced IL-10 was found in higher levels. Surprisingly, an increased proliferation was not observed in B cells of total CD83 KO mice (4). We conclude that B cells of CD83 B-cKO mice have a B cell activation defect after BCR or TLR stimulation, resulting in impaired upregulation of activation markers, a cytokine deviation but proliferation advantage.
CD83 is used as a marker for LZ B cells during GC reactions (14). CD83-cKO mice showed a characteristic shift in the DZ/ LZ ratio not only after SRBC immunization, but also when a soluble Ag such as NP-CGG was used or during B. burgdorferi infection. As the function of CD83 on LZ GC B cells is unknown, it can only be speculated about the mechanism for the relative expansion of DZ cells in GCs from CD83 B-cKO mice. Because a role of CD83 in Ag presentation has been detected (18) and because CD83 is highly expressed on LZ GC B cells, CD83 may be involved in Ag presentation to TFH cells in the LZ and thereby in the selection process of GC B cells. Surprisingly, Ag presentation was normal, even though not only CD83 is missing, but also MHC II is expressed at lower levels on CD83−/− GC B cells. It is possible that the in vitro Ag-presenting assay is either not sensitive enough for subtle alterations in the Ag-presenting capacity of B cells or might not reflect the interaction of GC B cells and TFH cells in the LZ of the GC. As it has been described that B cell division in the DZ is directly proportional to Ag presentation and selection by TFH cells in the LZ (40), the higher number of DZ B cells in CD83 B-cKO mice is unexpected. This finding would indicate a more efficient selection in the LZ by TFH cells, that is, a negative role of CD83 in this process. Clearly, we expected that an accumulation of B cells in the DZ of CD83 B-cKO mice would affect the rate of SHM and possibly also the magnitude of the Ab response.
It was reported before that total CD83 KO mice showed impaired Ab responses after immunization with the thymus-dependent Ag DNP-KLH (4). The reduced Ig response in total CD83 KO animals can be explained by a strong reduction of CD4+ Th cells, which normally trigger Ig class switch in T cell–dependent immune responses. This is in accordance with findings from bone marrow chimeras using transfer of total CD83 KO bone marrow in WT animals, which show no influence on the humoral responses (13). In contrast, overexpression of CD83 in CD83tg mice led to reduced humoral responses to thymus-dependent and thymus-independent Ags (13). Thus, Abs were not affected in CD83 B-cKO mice.
Despite the shift in DZ and LZ numbers of CD83 B-cKO mice, we could not find differences in the frequency of tryptophan to leucine exchange at position 33, which is the most frequently observed somatic mutation in VH186.2 leading to a 10-fold higher affinity after NP-CGG immunizations (35). Furthermore, the detailed quantification of positive selection using the BASELINe algorithm did also not reveal any significant difference between WT and CD83 B-cKO mice. These results are surprising, but we cannot exclude that a different Ag dose or analysis of other immunization time points may have a stronger effect on affinity maturation. We observed a strong disadvantage of CD83−/− B cells in participation of the GC response in a competitive cellular setting of mixed bone marrow chimaeras. Because CD83−/− B cells have an activation defect, detected by defective upregulation of activation markers needed for B–T cell interactions, this may lead to strong competitive disadvantage compared with WT B cells in the induction or maintenance of GCs.
Interestingly, IgE responses were significantly increased in CD83 B-cKO animals compared with controls during an immunization with both NP-KLH and B. burgdorferi. Class switching to IgE can be direct or subsequent to IgG1 class switching, as revealed recently by IgE reporter mice (41–43). These studies showed that IgE-producing plasma cells derived from IgE+ GC B cells are of a more transient nature compared with plasma cells derived from IgG1+ GC B cells and leave the GC after fewer selection rounds (41, 43, 44). Thus, when GC B cells of CD83 B-cKO mice are present to a higher extent in the DZ of GCs, a result of the GC reaction could be increased differentiation of IgE+ GC B cells into plasma cells, with the result of higher IgE titers. In support of this, a shift of the LZ/DZ ratio was observed in GFP reporter mice for IgE+ GC B cells. In these studies, IgE+ GC B cells showed an increased proportion of DZ GC cells compared with IgG1+ GC cells (41). Furthermore, a lack of MHC II expression on B cells was shown to be associated with higher IgE titers (45). Alternatively, or in addition, a shift toward enhanced Th2 differentiation in CD83 B-cKO mice could explain the enhanced IgE production. Recently, a novel so-called “rogue” GC B cell population was described (46). Rogue GC B cells accumulate in the case of impaired or deficient FAS expression with the result of an increase in plasma cell differentiation associated with a large number of IgE+ plasma cells (46). The authors already implicated that it could not only be FAS but also other molecules that are important for the emergence of the described rogue GC B cells, and one of those molecules could potentially be CD83.
During infection with B. burgdorferi, an adaptive immune response is controlled by both T cells and B cells. Lipoproteins of B. burgdorferi are strong inducers of humoral immune responses (47, 48). B cell responses are enhanced by CD4+ T cells, which secrete high INF-γ levels (49) and which are also located in the inflamed joint (50, 51). Additionally, in the inflamed synovial tissue high levels of Borrelia-specific Abs are detected, which are essential for bacterial elimination and of some importance for arthritis resolution (52–54). Reduction of INF-γ production along with a strong increase in IgE titers in CD83 B-cKO mice revealed a change in the immune response toward a Th2 response. However, Th2 cytokines such as IL-4 are generally very hard to detect in B. burgdorferi infections in C57BL/6 mice, and the Th1/Th2 cytokine ratio is mouse strain specific (55, 56). B. burgdorferi induces an expansion of MZ B cells and peritoneal B1 cells in WT mice (57). These B cells, which were found in lower abundance in CD83 B-cKO mice, might be activated via TLR2 by lipoproteins of B. burgdorferi. Secreted natural Abs are assumed to contribute to reduction of early bacterial replication (38, 58). Thus, the defect in bacterial clearance observed in CD83 B-cKO could be a cumulative result of imperfect GC B cell as well as innate B cell activation and proliferation.
In summary, we could show that CD83 on B cells is an activation marker, which is important for B cell activation. B cells lacking CD83 fail to upregulate specific costimulatory molecules on the B cell surface, but they show enhanced proliferation. CD83 B-cKO mice show a characteristic shift in composition of the GC with higher B cell numbers in the DZ and lower numbers in the LZ. Although this shift does not drastically impair the amount and affinity of Ag-specific IgG, it leads to increased IgE responses. Furthermore, a characteristic shift to Th2 responses with higher IgE production accompanied by impeded bacterial clearance is observed during B. burgdorferi infections in CD83 B-cKO mice.
We thank S. Angermüller and C. Draßner for technical help.
This work was supported by Deutsche Forschungsgemeinschaft Grants GK 1660 and SFB1181 Projects B03 and B06.
The sequences presented in this article have been submitted the National Center for Biotechnology Information SRA database (http://www.ncbi.nlm.nih.gov/sra) under accession number PRJNA310174.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- CD83 B-cKO
B cell–specific CD83 conditional knockout
keyhole limpet hemocyanin
- MHC I
MHC class I
- MHC II
MHC class II
next generation sequencing
T follicular helper
The authors have no financial conflicts of interest.