Abstract
Memory B cell responses are vital for protection against infections but must also be regulated to prevent autoimmunity. Cognate T cell help, somatic hypermutation, and affinity maturation within germinal centers (GCs) are required for high-affinity memory B cell formation; however, the signals that commit GC B cells to the memory pool remain unclear. In this study, we identify a role for IgG-immune complexes (ICs), FcγRs, and BAFF during the formation of memory B cells in mice. We found that early secretion of IgG in response to immunization with a T-dependent Ag leads to IC–FcγR interactions that induce dendritic cells to secrete BAFF, which acts at or upstream of Bcl-6 in activated B cells. Loss of CD16, hematopoietic cell–derived BAFF, or blocking IC:FcγR regions in vivo diminished the expression of Bcl-6, the frequency of GC and memory B cells, and secondary Ab responses. BAFF also contributed to the maintenance and/or expansion of the follicular helper T cell population, although it was dispensable for their formation. Thus, early Ab responses contribute to the optimal formation of B cell memory through IgG-ICs and BAFF. Our work defines a new role for FcγRs in GC and memory B cell responses.
Introduction
Adaptive immunity requires the commitment of activated B cells to either the memory or the plasma cell (PC) compartment, the differentiation of CD4+ T cells to follicular helper T (Tfh) cells, and coordinated expression of chemoattractant receptors to position T and B cells within the follicle for cognate interactions (1, 2). The specialized microenvironment of the germinal center (GC) provides a site for rapid expansion and selection of B cell clones whose somatically mutating Ig V regions compete for a limiting amount of Ag displayed on follicular dendritic cells (DCs) and limited availability of T cell help (3–5). Although many steps in the cyclic process of somatic hypermutation and clonal selection are defined, the events that dictate activated B cell fate to the GC or to the memory B cell pool are incompletely understood.
During the GC response, Tfh cells are critical effectors that provide help to B cells (6, 7). Tfh cells engage activated B cells at the T:B border, and their secreted cytokines promote Ig isotype switching and the selection of cells with high-affinity BCRs in GCs (1, 5, 8). The expression of CXCR5, ICOS, and PD-1 and the secretion of IL-21 distinguish Tfh cells from other CD4+ T cell subsets (6, 9). The formation of Tfh cells is dependent on the expression of Bcl-6, a process linked to ICOS expression on CD4+ T cells (10) and influenced by IL-2 (11, 12). This process commits primed T cells to the Tfh pool and inhibits their differentiation to other T cell subsets (13–16). Bcl-6 is also required for GC B cell formation (17–19). In activated B cells, Bcl-6 downregulates Blimp-1, directing B cells away from PC differentiation and toward the memory pathway (20, 21). Cytokines such as IL-6 and IL-21 have been shown to affect Bcl-6 expression in B and T cells (22–24); however, the loss of either cytokine is not enough to eliminate GCs and memory B cells, and a more complete picture of the events upstream of Bcl-6 expression is of interest in understanding B and T cell differentiation in GC responses.
BAFF plays an essential role in controlling the development and survival of B2 and marginal zone B cells (25, 26), enhancing the survival of plasmablasts (27) and affinity-matured B cells in the GC (5). Earlier studies in which BAFF was neutralized or deleted suggest BAFF plays a role in the GC response; however, interpretations of those results are complicated by the global loss of B cells associated with BAFF depletion (28–31). Others have shown that BAFF and anti-CD40 increase ICOSL expression on B cells (32, 33), and that TACI serves to limit the expression of ICOSL and the expansion of Tfh cells and GC B cells (34). Thus, BAFF has been implicated in events that contribute to GC responses; however, how BAFF is induced and where it acts in the GC response remain unclear.
In this study, we identify a previously unrecognized role for IgG-immune complexes (ICs), CD16 (FcγRIII), and BAFF in the formation of B cell memory. We found that early production of anti-nitrophenol (NP)–IgG promotes the formation of ICs that activate DCs through CD16. This process induces the secretion of BAFF, which acts at or upstream of Bcl-6 to promote the formation of GC B cells and proper memory cell formation. Although BAFF is not involved in the formation of Tfh cells, it plays a role in stabilizing and/or expanding the population at the peak of the GC response. Thus, IgG-ICs and CD16, through BAFF, act at or upstream of Bcl-6 expression in GC B cells and in the maintenance and/or expansion of Tfh cells to support the optimal formation of B cell memory during NP-specific immune responses.
Materials and Methods
Animals
B6-Ly5.2 congenic mice were purchased from the National Cancer Institute; BAFF−/− mice (29) and CD16-2−/− (FcγRIV) mice (35) were obtained from Glenn Matsushima and Charles Jennette at University of North Carolina at Chapel Hill. CD16−/− and CD64−/− mice (36) were obtained from Dr. Anne Sperling at the University of Chicago, and tissue from BAFF transgenic (Tg) mice (37) was obtained from Jeffrey Rathmell at Duke University (MMRRC strain #36508, B6.Cg-Tg[CD68-Tnfrsf13c]MB21Nemz/Mmucd). CD32−/− mice (38) were purchased from The Jackson Laboratory. Mice were used at 8–12 wk of age and maintained in an accredited animal facility.
Reagents and Abs
Abs against mouse CD4, CD19, CD95, GL-7, ICOS, ICOSL, PD-1, and B220-647 were purchased from BioLegend; CXCR5, B220, IgG1, IgG2a, IgG2b, and IgG3 Abs from BD Biosciences; Bcl-6, XBP-1, and IRF-4 from Santa Cruz Biotechnology; and BAFF (1C9) from Enzo. Anti–μ F(ab)2 was purchased from Jackson ImmunoResearch Laboratories. Anti-μ (clone B7.6), anti-NP (clones H33L and B1-8), 2.4G2 (FcγRIIb/FcγRIII block), and Ac38 idiotype Abs were purified from hybridoma supernatants. (Ac38 is an idiotype Ab that recognizes B-1-8 specificities generated during NP immunization.) Recombinant mouse BR3-Fc and isotype control protein were generated using mammalian expression systems and standard purification protocols. H33L and B1-8 were gifts from Dr. Garnett Kelsoe (Duke University). IL-4 and IL-5 were purchased from PeproTech, recombinant BAFF from R&D Systems, NP-OSu from Biosearch Technologies, KLH and PNA-biotin from Sigma-Aldrich, alum from Thermo Scientific, and streptavidin–Alexa 488 and –Alexa 647 from Invitrogen. Streptavidin–alkaline phosphatase and anti-IgG alkaline phosphatase were purchased from Southern Biotech. The Fc-binding TG19320 peptide was synthesized as described (39, 40).
B cell purification and bone marrow derivation of macrophages and DCs
Splenic B cells were isolated from B6 mice by negative selection (StemCell Technologies) and were 95–99% pure, as determined by flow cytometry. Splenic DCs were purified by positive selection of CD11c+ cells (Miltenyi) from enriched low-density cells (OptiPrep; Sigma-Aldrich). Purified cells were 80% CD11c+.
Bone marrow–derived DCs (BMDCs) and bone marrow–derived macrophages (MFs) (BMMFs) were prepared from single-cell suspensions from the tibias and femurs of B6, CD64−/−, CD32−/−, CD16−/−, and BAFF−/−mice. Following RBC lysis, cells were cultured 7 d in a 24-well low-cluster plate (Costar 3471) with 10 ng/ml GM-CSF and IL-4 to derive DCs and in 20 ng/ml M-CSF to derive MFs.
Cell culture
We preformed ICs by stimulating B6 B cells (1.5 × 105) with an excess of anti-μ (B7.6; IgG1; 30 μg/ml). The polyclonal IgM produced after 7 d forms a complex, with the excess anti-μ resulting in IgG1-IgM ICs in the supernatant. These B cell supernatants were used as a source of ICs in preparing DC–conditioned medium (DC CM).
DC CM was prepared by culturing 2 × 104 BMDCs (day 7) in a 96-well plate in the presence of IC-containing supernatants (see above; 20% of volume), IL-4 (25 ng/ml), and IL-5 (25 ng/ml). After 7 d, supernatants were harvested and frozen at −80°C.
For in vitro cocultures, 1.5 × 105 purified B6 B cells were cocultured with 1 × 104 BMDCs or ex vivo DCs in a 96-well plate stimulated with IL-4 (25 ng/ml), IL-5 (25 ng/ml), and 30 μg/ml anti-μ with or without recombinant murine BAFF (5 ng/ml) or DC CM (20% of total volume). Intracellular Bcl-6 was assessed by flow cytometry after 48 h.
ELISAs
NP-specific IgG levels were quantitated from serum using microtiter plates coated with NP13BSA and blocked with 0.5% BSA. Serially diluted serum samples were incubated overnight at 4°C. Anti-NP was detected using an alkaline phosphatase–conjugated rabbit anti-mouse IgG Ab (1/1000 dilution) followed by phosphatase substrate. OD values were converted to concentration based on standard curves using the H33L (anti-NP) hybridoma.
ELISPOT
For the analysis of NP-specific B cells, multiscreen ELISPOT plates (Millipore) were coated with NP13BSA in PBS and blocked with 1% BSA. Single-cell suspensions of spleen were prepared from immunized or naive B6 mice. After RBC lysis, cells were plated in serial dilutions on washed ELISPOT plates. Anti-NP IgG-secreting spots were detected with anti-IgG–biotin and streptavidin-HRP (BD Biosciences). Plates were developed with 3-amino 9-ethylcarbazole.
To enumerate BAFF-secreting DCs, CD11c+ cells (1 × 106) were purified from spleens and cultured for 60 h on BR3-Fc–coated ELISPOT plates. BAFF-secreting cells were detected using anti-BAFF (clone 1C9). To enumerate BAFF-secreting cells from BMDCs, day 7 cells (2.5 × 105) were plated on ELISPOT plates as above and incubated 24 h with preformed ICs (IgM + anti-μ or NP-OVA + anti-NP IgG mAb, H33L) prior to addition of 1C9. Anti-μ ICs were made by combining the supernatant from stimulated B cells (20 ng IgM) with anti-μ (5 μg) or by combining anti-NP IgG with NP-OVA. In some experiments, TG19320 was added at 50 μg/ml to inhibit IgG binding to FcγRs.
Bone marrow chimeras
B6-Ly5.2 congenic mice (6–8 wk of age) were lethally irradiated (10.5 Gy; 1050 rad) and reconstituted with 8 × 106 bone marrow cells from either B6 (B6 control chimeras) or BAFF−/− (BAFF−/− chimeras) mice. After 8 wk, we monitored reconstitution by assessing the frequency of CD45.1+ and CD45.2+ splenocytes by flow cytometry.
Immunization and adoptive transfers of BMMF/BMDCs
B6, BAFF−/− bone marrow chimeras, and CD16−/− mice (8–10 wk of age) were immunized by i.p. or s.c. injection with 100 μg NP14KLH precipitated in an equal volume of alum (Imject; Thermo Scientific). Mice were boosted by i.v. injection with the same dose of soluble NP14KLH at day 35. To assess the contribution of DCs or MFs in the secretion of BAFF, 8 × 106 BAFF Tg or BAFF−/− BMDCs or BMMFs were injected at the time of s.c. immunization. Draining lymph nodes were harvested on day 7 for flow cytometry analysis.
TG peptide injections
B6 mice were immunized with 100 μg NP14KLH in alum (1:1) via i.p. injection and were administered three (i.p.) injections (15–30 mg/kg) of Fc blocking peptide (TG19320) or an equal amount of unrelated control peptide over the course of 7 d.
Flow cytometry
GC B cells and Tfh were analyzed on day 7 postimmunization and were defined as CD19+, GL-7+, CD95+ and CD4+, CXCR5+, PD-1+. Ac38 was used to define NP-specific GC B cells. NP-specific memory B cells were defined as Ac38+ IgG+ double positive CD19+ lymphocytes. The lymphocyte gate was determined by forward and side scatter properties. To gate on Tfh populations, we initially used isotype control Ab staining for CXCR5. To identify GC B cells (CD19+CD95+GL-7+), we used fluorescence minus one where the CD95 gate was established using the CD19+GL-7+ population. Similarly, the GL-7 gate was established using the CD19+CD95+ population. All subsequent gating was based on untreated B6 mice. To quantitate expression of intracellular IRF-4, Bcl-6, and XBP-1, splenocytes from immunized B6, CD16−/−, B6 control chimera, and BAFF−/− chimera mice were washed, fixed (4% paraformaldehyde), and permeabilized with methanol for ≥24 h at −20°C. Fixed cells were washed and blocked with 2.4G2 before staining. Data are expressed as fold change in mean fluorescence intensity/isotype control mean fluorescence intensity.
Real-time PCR
Splenic B cells from B6 and BAFF−/− chimeras were purified after NP14KLH immunization. mRNA was isolated from 5–10 × 106 purified B cells and cDNA synthesized using SuperScript VILO cDNA Synthesis Kit (Invitrogen). DNA was subsequently amplified using FastStart Universal SYBR Green Master Mix (Roche). Relative values were compared using the 2−ΔΔCΤ method. In all experiments, 18S rRNA was used as an internal control. Primers included the following: murine Aicda forward 5′-GGGAAAGTGGCATTCACCTA-3′, murine Aicda reverse 5′−GAACCCAATTCTGGCTGTGT-3′; murine 18S rRNA forward 5′-TCAAGAACGAAAGTCGGAGGTT-3′, murine 18S rRNA reverse 5′-GGACATCTAAGGGCATCACAG-3′.
GC staining and counting
Spleens were harvested from B6 or CD16−/− mice on days 7, 14, and 21 after immunization and were flash frozen in OCT (Optimum Cutting Temperature; Fisher). Tissue sections (6 μm) were fixed in 1:1 MeOH/acetone; blocked with 10% FBS in PBS containing 2.4G2; and stained with PNA-biotin, B220–Alexa 647, and streptavidin–Alexa 488. GCs were defined as PNA+ cell clusters within B220+ follicles (41). The number of germinal centers per square millimeter of B220+ area was determined by dividing the number of GCs counted in a field by the area of B220+ follicles in the same field. This method accounted for follicles that were only partially represented in a given field (42). This process was done for 10–30 fields per mouse, totaling 30–100 follicles per mouse at each time point.
Microscopy
Macroscope images were obtained on a Leica MX16FA fluorescence stereo microscope/macroscope (×0.63 objective; numerical aperture of 1.0). Other images were obtained using an Olympus Fluoview 500 (×10 objective; numerical aperture of 0.45).
Statistics
The one-sided, or one-sample, t test was used to compare changes in transcription factor levels and ELISPOT. The two-sample Student t test was used to assess statistical differences between cell populations measured by flow cytometry and serum Ab secretion. Analyses were performed in GraphPad Prism.
Results
BAFF−/− bone marrow chimeras exhibit reduced secondary responses
Previous studies have linked BAFF with an enhanced response to vaccination, suggesting that it plays a role in memory responses (43–45). To assess this, we generated BAFF−/− bone marrow chimeras by engrafting irradiated B6 mice with B6 (BAFF+/+) or BAFF−/− bone marrow. This approach limits BAFF deficiency to hematopoietic cells, allowing other sources of BAFF to maintain the peripheral B cell population (46). No differences in the spleen cellularity between B6 and BAFF−/− bone marrow chimeras were observed (Supplemental Fig. 1A), and the basal level (day 0) of Tfh and GC B cells were not different. In BAFF−/− chimeras, we found that the primary IgG response to NP14KLH (Fig. 1A) was comparable to that in B6 control chimeras. However, 7 d after secondary immunization (day 42), BAFF−/− bone marrow chimeras showed a 1.4-fold reduction in the levels of IgG, and 14 d after secondary immunization (day 49) the levels of IgG were reduced 2-fold compared with B6 chimeras (Fig. 1B). Diminished production of IgG during the secondary response could reflect diminished class switch because BAFF can induce AID expression (47, 48). However, AID mRNA levels in B cells from B6 control and BAFF−/− chimeras were not different (Supplemental Fig. 1B), suggesting that BAFF has a role other than in class switch.
BAFF−/− chimeras have defective memory and GC B cells, diminished Bcl-6 levels, and decreased frequency of Tfh cells. (A) B6 and BAFF−/− chimeric mice were immunized (i.p.) with 100 μg NP14KLH in alum. Serum IgG anti-NP responses in B6 and BAFF−/− chimeric mice were measured by ELISA on days 7, 14, 21, 28, and 35 postimmunization. n = 4–6 mice per group over three experiments. (B) Serum IgG anti-NP levels measured on days 39, 42, and 49 (4, 7, 14 d) following a boost of 100 μg of soluble NP14KLH on day 35. n = 3–6 mice over three experiments. (C) The frequency of CD19+Ac38+IgG+ B cells was determined by flow cytometry from the spleens of B6 and BAFF−/− chimeras immunized for 28 d. The number of splenocytes from B6 and BAFF−/− chimeras was comparable. n = 4–6 mice over four experiments. (D and E) The frequency of CD4+CXCR5+PD-1+ T cells from B6 and BAFF−/− chimeric mice was measured on days 0, 3, and 7 following immunization. n = 3–6 mice per time point over two experiments. (F and G) The frequency of CD19+, GL-7+, CD95+ GC B cells was measured by flow cytometry on day 7 postimmunization. n = 3–5 mice over three experiments. (H) The frequency of CD19+Ac38+CD95+GL-7+ B cells was measured on day 7 postimmunization. n = 3–5 mice over three experiments. (I and J) Relative expression of Bcl-6 in GC B cells (CD19+CD95+GL-7+) from B6 and BAFF−/− chimeras measured 7 d after NP14KLH immunization (i.p.). n = 4–6 mice per group over two experiments. Bars display mean (C, E, G, H, and I); error bars indicate SD (A and B). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
BAFF−/− chimeras have defective memory and GC B cells, diminished Bcl-6 levels, and decreased frequency of Tfh cells. (A) B6 and BAFF−/− chimeric mice were immunized (i.p.) with 100 μg NP14KLH in alum. Serum IgG anti-NP responses in B6 and BAFF−/− chimeric mice were measured by ELISA on days 7, 14, 21, 28, and 35 postimmunization. n = 4–6 mice per group over three experiments. (B) Serum IgG anti-NP levels measured on days 39, 42, and 49 (4, 7, 14 d) following a boost of 100 μg of soluble NP14KLH on day 35. n = 3–6 mice over three experiments. (C) The frequency of CD19+Ac38+IgG+ B cells was determined by flow cytometry from the spleens of B6 and BAFF−/− chimeras immunized for 28 d. The number of splenocytes from B6 and BAFF−/− chimeras was comparable. n = 4–6 mice over four experiments. (D and E) The frequency of CD4+CXCR5+PD-1+ T cells from B6 and BAFF−/− chimeric mice was measured on days 0, 3, and 7 following immunization. n = 3–6 mice per time point over two experiments. (F and G) The frequency of CD19+, GL-7+, CD95+ GC B cells was measured by flow cytometry on day 7 postimmunization. n = 3–5 mice over three experiments. (H) The frequency of CD19+Ac38+CD95+GL-7+ B cells was measured on day 7 postimmunization. n = 3–5 mice over three experiments. (I and J) Relative expression of Bcl-6 in GC B cells (CD19+CD95+GL-7+) from B6 and BAFF−/− chimeras measured 7 d after NP14KLH immunization (i.p.). n = 4–6 mice per group over two experiments. Bars display mean (C, E, G, H, and I); error bars indicate SD (A and B). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
BAFF−/− chimeras exhibit defects in the frequency of memory B, Tfh cells, and GC B cells
Rapid, high-titer secondary immune responses require the activation of memory B cells (49). Although most IgG memory B cells do not require BAFF for maintenance (50), it is not known whether BAFF is important for their formation. To determine whether BAFF affects the frequency of memory B cells, we immunized (i.p.) BAFF−/− chimeras with NP14KLH and measured the frequency of NP-specific memory B cells (CD19+IgG+Ac38 Id+) on day 28 postimmunization. We found that immunization significantly increased the frequency of memory B cells in B6 and BAFF−/− bone marrow chimeras; however, the magnitude of the response was significantly lower in BAFF−/− chimeras (Fig. 1C).
Tfh cells are critical in the early GC response and required for the differentiation of memory B cells (51, 52). It was possible that BAFF affected memory responses by influencing Tfh cells, which in turn affected GC responses. To assess whether BAFF affects formation and/or maintenance of Tfh cells, B6 and BAFF−/− chimeras were immunized and the frequencies and numbers of Tfh cells (CXCR5+PD-1+CD4+) were quantitated on days 3 and 7 postimmunization (Fig. 1D, 1E, Supplemental Fig. 1C). On day 3, the frequency of Tfh cells in B6 chimeras increased by 1.5-fold, whereas in BAFF−/− chimeras it increased 1.2-fold. This finding suggests that BAFF does not play a significant role in the formation of Tfh cells. However, on day 7 postimmunization, the frequency and number of Tfh cells in B6 chimeras increased an additional 2-fold, whereas their frequency in BAFF−/− chimeras did not change. This observation suggests that BAFF may support the expansion of Tfh cells after pre-Tfh cells transition to Tfh.
GCs are necessary for the formation of high-affinity, class-switched memory B cells (53, 54). To determine whether BAFF influenced the frequency of GC B cells, we enumerated CD19+GL-7+CD95+ GC B cells 7 d after immunization. In BAFF−/− chimeras, the frequency and number of total GC B cells (Fig. 1F, 1G, Supplemental Fig. 1D), and the frequencies of Ag-specific (Fig. 1H; CD19+Ac38+GL-7+CD95+) GC B cells in BAFF−/− chimeras, were significantly lower than those in B6 chimeras. Thus, BAFF significantly contributes to optimal Ag-specific GC responses.
BAFF acts at, or upstream of, Bcl-6 expression in B cells
Bcl-6 plays a critical role in initiating GC responses and committing activated B cells to a memory cell phenotype (17, 19). Thus, one possibility was that the loss of BAFF negatively affected Bcl-6 expression. To test this, we measured Bcl-6 levels in GC B cells after immunization. We found that on day 7, the levels of Bcl-6 in GC B cells from immunized BAFF−/− chimeras were decreased 40% compared with those in B6 chimeras (Fig. 1I, 1J). The data show that hematopoietic cell–derived BAFF acts at, or upstream of, Bcl-6 expression in B cells. Collectively, our data indicate that BAFF has an impact on the formation of GC and memory B cells by increasing the expression of Bcl-6 in GC B cells, and indirectly through maintaining the Tfh cell populations during the GC response.
DC-derived BAFF regulates the frequency of GC B and Tfh cells
The BAFF−/− chimera model is characterized by the absence of BAFF in all hematopoietic cells. Among bone marrow–derived cell types that are capable of producing BAFF, myeloid cells are a major source of BAFF following infection or immunization (27, 43, 55). To determine whether myeloid cells are a sufficient source of BAFF during GC B cell and memory B cell fate decisions, we adoptively transferred BMDCs and BMMFs from BAFF Tg mice into BAFF−/− chimeras by s.c. injection at the time of immunization. We previously established that 70% of s.c. injected BMDCs migrated to the inguinal lymph nodes, and that the magnitude of the s.c. anti-NP response was comparable to that of i.p. immunization (data not shown). We found that constitutive expression of BAFF by Tg DCs, but not Tg MFs, restored the frequencies of GC B cells (Fig. 2A) and Tfh cells (Fig. 2B), as well as the levels of Bcl-6 in GC B cells (Fig. 2C), in BAFF−/− chimeras. Because these BAFF Tg MFs secrete more BAFF than do BAFF Tg DCs (37), the responses were not due to higher production of BAFF by Tg DCs. Conversely, transfer of BAFF−/− DCs did not increase the frequencies of GC B cells (Fig. 2A), indicating that the effects of the transfer on GC B cells were not due to more DCs, or to secretion of other cytokines made by the transferred DCs. This finding suggests that in addition to DCs presenting Ag during GC responses (56), DC-derived BAFF is also be important in directing the differentiation of GC cell populations.
Secretion of BAFF by DCs restores Bcl-6 levels and the frequency of GC B and Tfh cells. (A–C) BAFF Tg BMDCs, BMMFs, or BAFF−/− BMDCs (8 × 106) were injected (s.c.) into B6 or BAFF−/− chimeric mice that were simultaneously immunized (s.c.) with NP14KLH. On day 7, inguinal lymph nodes were harvested and (A) the frequency of GC B cells (GL-7+CD95+ cells in CD19+ B cells), (B) the frequency of Tfh cell (CXCR5+PD-1+ cells in CD4+ T cells), and (C) Bcl-6 expression in GC B cells from inguinal lymph nodes were measured by flow cytometry. n = 3–5 over three experiments. (D) On days 2, 3, and 7 following immunization, 1 × 106 splenocytes from B6 mice were plated on NP-BSA–coated ELISPOT plates. After 18 h, NP-specific ASCs were quantitated. n = 3–4 mice per time point over three experiments. (E) A total of 1 × 106 purified CD11c+ cells from B6 mice immunized for 2, 3, and 7 d were plated on BR3-Fc–coated ELISPOT plates. After 60 h, the number of BAFF-secreting cells was enumerated. n = 3–6 mice per time point over three experiments. BAFF ELISPOTs ranged from 32 to 137 [day (D)0–D7]. (F) BAFF-secreting cells enumerated from B6 BMDCs (2.5 × 105) cultured for 24 h in the presence or absence of anti-μ ICs, with or without Fc blocking peptide (TG19320; 50 μg/ml). n = 3–10 over six experiments. BAFF ELISPOTs ranged from 531 to 1051 (untreated and treated with IgG-ICs ±TG19320). (G and H) B6 mice were immunized with NP14KLH and dosed with 15–30 mg/kg of Fc blocking peptide (TG19320) or an unrelated scrambled peptide (control) via i.p. injection. On day 7, the frequency of Tfh cells (CD4+CXCR5+PD-1+) (G) and GC B cells (CD19+GL-7+CD95+) (H) was determined. n = 3–6 over two experiments. Bars display mean (A, B, D, G, and H), and error bars indicate SEM (C, E, and F). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
Secretion of BAFF by DCs restores Bcl-6 levels and the frequency of GC B and Tfh cells. (A–C) BAFF Tg BMDCs, BMMFs, or BAFF−/− BMDCs (8 × 106) were injected (s.c.) into B6 or BAFF−/− chimeric mice that were simultaneously immunized (s.c.) with NP14KLH. On day 7, inguinal lymph nodes were harvested and (A) the frequency of GC B cells (GL-7+CD95+ cells in CD19+ B cells), (B) the frequency of Tfh cell (CXCR5+PD-1+ cells in CD4+ T cells), and (C) Bcl-6 expression in GC B cells from inguinal lymph nodes were measured by flow cytometry. n = 3–5 over three experiments. (D) On days 2, 3, and 7 following immunization, 1 × 106 splenocytes from B6 mice were plated on NP-BSA–coated ELISPOT plates. After 18 h, NP-specific ASCs were quantitated. n = 3–4 mice per time point over three experiments. (E) A total of 1 × 106 purified CD11c+ cells from B6 mice immunized for 2, 3, and 7 d were plated on BR3-Fc–coated ELISPOT plates. After 60 h, the number of BAFF-secreting cells was enumerated. n = 3–6 mice per time point over three experiments. BAFF ELISPOTs ranged from 32 to 137 [day (D)0–D7]. (F) BAFF-secreting cells enumerated from B6 BMDCs (2.5 × 105) cultured for 24 h in the presence or absence of anti-μ ICs, with or without Fc blocking peptide (TG19320; 50 μg/ml). n = 3–10 over six experiments. BAFF ELISPOTs ranged from 531 to 1051 (untreated and treated with IgG-ICs ±TG19320). (G and H) B6 mice were immunized with NP14KLH and dosed with 15–30 mg/kg of Fc blocking peptide (TG19320) or an unrelated scrambled peptide (control) via i.p. injection. On day 7, the frequency of Tfh cells (CD4+CXCR5+PD-1+) (G) and GC B cells (CD19+GL-7+CD95+) (H) was determined. n = 3–6 over two experiments. Bars display mean (A, B, D, G, and H), and error bars indicate SEM (C, E, and F). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
The binding of IC to FcγRs induces BAFF secretion
Previous studies showed that exogenous ICs induce BMDCs to secrete a number of cytokines, including BAFF (57, 58). In another study, mice lacking the common γ-chain of the FcγRs (Fcγc) exhibited diminished secondary immune responses (59). Because our data suggest that DCs may promote secondary responses via BAFF, we postulated that ICs formed by the early IgG Ab response might induce DCs to secrete BAFF. This model requires that the early IgG response occur concurrently with, or precede, BAFF secretion. To test this, we harvested spleens from B6 mice on days 2, 3, and 7 following NP14KLH immunization and used ELISPOT to measure the numbers of Ab (IgG)-secreting cells (ASCs) and BAFF-secreting DCs. NP-specific (IgG) ASCs were increased 7-fold by day 2 postimmunization and expanded to 90-fold over the course of 7 d (Fig. 2D). Similarly, the number of splenic CD11c+ DCs that secreted BAFF increased 6-fold between days 2 and 7 (Fig. 2E). Thus, secretion of Ig by B cells and production of BAFF by DCs occur concomitantly, beginning ∼2 d following immunization. This finding supports the idea that IgG-ICs formed early in the immune response contribute to the production of BAFF, which is required to optimize GC responses.
To further test the idea that IgG-ICs induce DCs to secrete BAFF, we blocked IgG-Fc:FcγRs interactions in vitro and assessed whether this affected BAFF secretion by DCs. Stimulation of B6 BMDCs with preformed IgG-ICs (IgM bound by anti-μ of IgG1 isotype) induced a 1.8-fold increase in BAFF secretion (Fig. 2F). This result was not unique to IgM/IgG ICs because preformed anti-NP ICs (NP-OVA bound by anti-NP of IgG1 isotype) induced a dose-dependent 2.5-fold increase in BAFF secretion (Supplemental Fig. 1E). Coculture with a tetrameric tripeptide (TG19320) that blocks IgG Fc regions (39, 40) reduced the number of BAFF-secreting DCs to levels indistinguishable from those of unstimulated cells. To assess whether blocking Fc:FcγR interactions affected the adaptive immune response in vivo, we administered TG19320 to B6 mice at the time of immunization and measured the frequencies of GC B and Tfh cells. We found that coadministration of TG19320 with NP14KLH blocked an increase in GC B and Tfh cells on day 7 (Fig. 2G, 2H). This finding indicates that the interactions between IgG-ICs and FcγRs are necessary for optimal GC responses and for the maintenance and/or expansion of newly formed Tfh cells in response to immunization. Collectively, these data identify a mechanism wherein IgG-ICs formed early in the immune response ligate FcγRs on DCs to induce BAFF secretion, which in turn contributes to optimal GC responses.
BAFF regulates the expression of Bcl-6 in activated B cells in vitro
To define whether BAFF plays a role in committing activated B cells to the memory compartment, we established an in vitro reconstitution system using the expression of Bcl-6 as a marker of memory B cell commitment. In this in vitro system, we used conditioned medium prepared from B6 BMDCs (DC CM) stimulated with preformed IgG1-ICs (Fig. 2F) as a source of BAFF. This approach was based on in vivo findings that GC responses were dependent on BAFF produced by DCs (Fig. 2A–C). To generate activated B cells, we used purified splenic B cells (B6) stimulated with anti-μ in combination with IL-4 and IL-5 to induce a low level of Bcl-6 expression (Fig. 3A, 3B; B cell activation). Thus, changes in Bcl-6 as a consequence of DC CM or recombinant BAFF could be measured. We found that DC CM increased Bcl-6 expression in activated B cells ∼2-fold, compared with cells cultured in the absence DC CM (Fig. 3A, 3B). These levels were comparable to those achieved with recombinant BAFF. Further, CM from BMDCs stimulated with F(ab′)2-containing ICs was not as efficient as CM stimulated with IgG-ICs (intact Fc regions) at inducing Bcl-6. In this in vitro system, Bcl-6 expression was not elevated in B cells cultured with DC CM wherein BAFF was neutralized with BR3-Fc, or wherein DC CM was made from BAFF−/− BMDCs (Fig. 3C). These data showed that DC-derived BAFF, induced by IgG-ICs, directly affects Bcl-6 expression in B cells activated in vitro.
DC-derived BAFF regulates the expression of Bcl-6 in cultured B cells. Purified B6 B cells (1 × 105) stimulated with anti-μ (30 μg/ml), IL-4 (25 ng/ml), and IL-5 (25 ng/ml) (designated as B cell activation) were cocultured with DC CMs. Intracellular levels of key transcription factors (Bcl-6, XBP-1, IRF-4) were measured by flow cytometry on day 2. (A) Bcl-6 expression in B cells cocultured with B6 DC CM or recombinant BAFF (5 ng/ml). Flow analysis was restricted to live cells. (B) Representative histogram of Bcl-6 expression in B cells activated by DC CM, as in (A). n = 3–4 mice over four experiments. (C) Purified B6 B cells were cocultured with CM generated from B6 BMDCs treated with ICs containing intact IgG (B6 DC CM) or F(ab′)2 of IgG [B6 DC CM F(ab′)2], BAFF−/− DC CM, or B6 DC CM neutralized with BR3-Fc (10 μg/ml) or control IgG-Fc (10 μg/ml). n = 4–7 over four experiments. (D–G) B cells were cocultured with B6 DC CM. Intracellular levels of XBP-1 (D and E) and IRF-4 (F and G) were quantitated. n = 3 mice per group over three experiments. Histograms are representative of three experiments, and all flow analysis was done after gating on live cells. (H) Numbers of splenic PCs (CD138+B220−) were enumerated by flow cytometry in B6, B6 chimera, or BAFF−/− chimeras at day 5 following immunization. n = 1–3 mice per group over two experiments. (I) Bcl-6 expression in B cells cultured with DC CM from B6, CD64−/−, CD32−/−, CD16−/−, or CD16-2−/− (FcγRIV) mice. n = 6–7 mice per group over four experiments. Bars display mean (H), and error bars indicate SEM (A, C, D, F, and I). *p ≤ 0.05, **p ≤ 0.01.
DC-derived BAFF regulates the expression of Bcl-6 in cultured B cells. Purified B6 B cells (1 × 105) stimulated with anti-μ (30 μg/ml), IL-4 (25 ng/ml), and IL-5 (25 ng/ml) (designated as B cell activation) were cocultured with DC CMs. Intracellular levels of key transcription factors (Bcl-6, XBP-1, IRF-4) were measured by flow cytometry on day 2. (A) Bcl-6 expression in B cells cocultured with B6 DC CM or recombinant BAFF (5 ng/ml). Flow analysis was restricted to live cells. (B) Representative histogram of Bcl-6 expression in B cells activated by DC CM, as in (A). n = 3–4 mice over four experiments. (C) Purified B6 B cells were cocultured with CM generated from B6 BMDCs treated with ICs containing intact IgG (B6 DC CM) or F(ab′)2 of IgG [B6 DC CM F(ab′)2], BAFF−/− DC CM, or B6 DC CM neutralized with BR3-Fc (10 μg/ml) or control IgG-Fc (10 μg/ml). n = 4–7 over four experiments. (D–G) B cells were cocultured with B6 DC CM. Intracellular levels of XBP-1 (D and E) and IRF-4 (F and G) were quantitated. n = 3 mice per group over three experiments. Histograms are representative of three experiments, and all flow analysis was done after gating on live cells. (H) Numbers of splenic PCs (CD138+B220−) were enumerated by flow cytometry in B6, B6 chimera, or BAFF−/− chimeras at day 5 following immunization. n = 1–3 mice per group over two experiments. (I) Bcl-6 expression in B cells cultured with DC CM from B6, CD64−/−, CD32−/−, CD16−/−, or CD16-2−/− (FcγRIV) mice. n = 6–7 mice per group over four experiments. Bars display mean (H), and error bars indicate SEM (A, C, D, F, and I). *p ≤ 0.05, **p ≤ 0.01.
As Bcl-6 levels become elevated, the PC program is attenuated (60, 61). To further validate the in vitro model, we measured intracellular IRF-4 and XBP-1 and found that DC CM diminished the levels of both transcription factors by 2.2-fold (Fig. 3D–G), indicating that BAFF acts at, or upstream of, Bcl-6, directing B cell differentiation away from a PC fate. To test whether BAFF reduces PC differentiation in vivo, we enumerated splenic PCs from immunized B6 chimeras and BAFF−/− chimeras at day 5 (Fig. 3H), a time when PCs normally appear in the spleen (62). We found that CD138+B220− cell numbers in BAFF−/− chimeras were increased 2.7-fold compared with those in B6 chimeras. This finding indicates that in vivo, BAFF decreases PC differentiation as the memory response initiated.
IgG-ICs bind CD16 during the anti-NP response
We reasoned that if FcγR stimulation promoted BAFF secretion, loss of the FcγR that binds IgG1–anti-NP ICs would negatively affect the ability of DC CM to promote Bcl-6. To assess this, we tested whether CMs from BMDCs derived from B6 mice, or mice deficient in CD64 (FcγRI−/−), CD32 (FcγRIIb −/−), CD16 (FcγRIII −/−), or CD16-2 (FcγRIV−/−), induced Bcl-6 in the in vitro reconstitution system described above. DC CM from CD16−/− mice failed to induce Bcl-6 expression, whereas CMs from all other FcγR-deficient mice induced Bcl-6 to levels comparable to, or above, those induced by B6 DC CM (Fig. 3I). This observation suggests that ligation of CD16 on DCs is required for the expression of Bcl-6 in activated B cells during the anti-NP response.
Impaired secondary responses in CD16−/− mice
Our in vitro data indicate that CD16 is responsible for inducing DCs to secrete BAFF after NP14KLH immunization. To test this in vivo, we quantitated the number of splenic BAFF-secreting DCs in B6 and CD16−/− mice 7 d after immunization (Fig. 4A). In the absence of CD16, we found that the number of BAFF-secreting DCs was markedly diminished compared with the number in wild-type mice. This finding is consistent with the idea that CD16 is the FcγR responsible for initiating the IgG1-dominant immune response to NP14KLH (54). It is likely that other Fc receptors would be used during immune responses that generate Abs of IgG subclasses other than IgG1.
CD16−/− mice have defective BAFF secretion, secondary Ab responses, and frequency of memory B cell and Tfh cells. (A) BAFF-secreting cells enumerated from purified CD11c+ cells that were isolated from B6 and CD16−/− mice on day 7. n = 3 over three experiments. B6 and CD16−/− mice were immunized (i.p.) with 100 μg NP14KLH in alum. Serum IgG anti-NP (B) primary responses measured by ELISA on days 0, 7, 14, 21, 28, and 35; serum IgG (C) secondary Ab levels measured on days 39, 42, and 49 (days 4, 7, and 14) following a boost of 100 μg soluble NP14KLH on day 35. n = 4–8 mice over three experiments. (D and E) The frequency of CD19+Ac38+IgG+ memory B cells was determined by flow cytometry from the spleens of B6 and CD16−/− mice immunized for 28 d. The number of splenocytes in B6 and CD16−/− mice was comparable. n = 5–6 mice over three experiments. Data are expressed as the percentage of CD19+ cells that are IgG+ and Ac38+. (F) The frequencies of CD4+CXCR5+PD-1+ T cells from immunized B6 and CD16−/− mice on days 0, 3, 7, or 10. n = 2–8 mice per time point over four experiments. (G) B6, BAFF Tg, and BAFF−/− BMDCs (8 × 106) were injected into CD16−/− mice that were simultaneously immunized with NP14KLH. On day 7, the frequencies of CD4+CXCR5+PD-1+ T cells from inguinal lymph nodes were enumerated by flow cytometry. n = 3–7 mice over three experiments. Bars display mean (E–G), and error bars indicate SEM (A) or SD (B and C). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
CD16−/− mice have defective BAFF secretion, secondary Ab responses, and frequency of memory B cell and Tfh cells. (A) BAFF-secreting cells enumerated from purified CD11c+ cells that were isolated from B6 and CD16−/− mice on day 7. n = 3 over three experiments. B6 and CD16−/− mice were immunized (i.p.) with 100 μg NP14KLH in alum. Serum IgG anti-NP (B) primary responses measured by ELISA on days 0, 7, 14, 21, 28, and 35; serum IgG (C) secondary Ab levels measured on days 39, 42, and 49 (days 4, 7, and 14) following a boost of 100 μg soluble NP14KLH on day 35. n = 4–8 mice over three experiments. (D and E) The frequency of CD19+Ac38+IgG+ memory B cells was determined by flow cytometry from the spleens of B6 and CD16−/− mice immunized for 28 d. The number of splenocytes in B6 and CD16−/− mice was comparable. n = 5–6 mice over three experiments. Data are expressed as the percentage of CD19+ cells that are IgG+ and Ac38+. (F) The frequencies of CD4+CXCR5+PD-1+ T cells from immunized B6 and CD16−/− mice on days 0, 3, 7, or 10. n = 2–8 mice per time point over four experiments. (G) B6, BAFF Tg, and BAFF−/− BMDCs (8 × 106) were injected into CD16−/− mice that were simultaneously immunized with NP14KLH. On day 7, the frequencies of CD4+CXCR5+PD-1+ T cells from inguinal lymph nodes were enumerated by flow cytometry. n = 3–7 mice over three experiments. Bars display mean (E–G), and error bars indicate SEM (A) or SD (B and C). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
To test whether loss of CD16 in vivo impaired adaptive immune responses, we measured primary and secondary Ab responses in B6 and CD16−/− mice following NP14KLH immunization. As in the BAFF−/− chimeric mice, the primary IgG response in CD16−/− mice was comparable to that in B6 mice (Fig. 4B). However, in the secondary response, the levels of IgG in B6 mice increased 1.7-fold on day 42 (7 d after boost) and 4.5-fold on day 49 (14 d after boost), whereas IgG levels in the CD16−/− mice did not increase on days 42 or 49 (Fig. 4C). This result was not an indirect consequence of altered spleen cellularity or changes in the splenic cell populations owing to CD16 deficiency because the frequencies of DCs, T cells, and B cells in CD16−/− mice were not different from those in B6 mice, and the number of splenocytes from B6 and CD16−/− mice were comparable (Supplemental Fig. 1F). This finding indicates that CD16 plays a role in generating memory B cells and secondary immune responses to NP14KLH.
CD16−/− mice exhibit defects in forming GC and memory B cells and maintaining Tfh cells
The loss of DC-derived BAFF in CD16−/− mice supports a role for CD16 in the memory response to NP14KLH. To assess whether the diminished secondary response in CD16−/− mice reflects a reduction in memory B cells, we assessed the frequency of NP-specific memory B cells (CD19+Ac38+IgG+) in B6 and CD16−/− mice 28 d following immunization. CD16−/− mice showed a 2.4-fold decrease in the Ac38 Id+ IgG memory B cell population compared with immunized B6 mice (Fig. 4D, 4E), indicating that CD16 regulates memory responses in part through BAFF production.
Data from immunized BAFF−/− chimeras suggest that BAFF contributes to B cell memory by affecting the GC B and Tfh cell populations. If the binding of IgG-ICs to CD16 were the predominant source of BAFF, then loss of CD16 would also diminish the GC B and Tfh pools. To assess this possibility, we quantitated Tfh cells (CD4+CXCR5+PD-1+) from immunized B6 and CD16−/− mice. After 3 d, CD16−/− mice had a comparable frequency of Tfh cells compared with immunized B6 mice (Fig. 4F). However, on day 7, CD16−/− mice had 3-fold fewer Tfh cells, similar to the defect in maintaining Tfh cells observed in immunized BAFF−/− chimeras (Fig. 1E). After 10 d, frequencies of Tfh cells decreased 2-fold in B6 mice, whereas the frequencies in CD16−/− mice were not changed compared with day 7. This finding suggests that BAFF plays a role in the expansion and/or maintenance of Tfh cells during early GC responses, but that other factors also contribute. Adoptive transfer of BAFF Tg, but not BAFF−/− BMDCs, into immunized CD16−/− mice restored the frequency of Tfh cells on day 7 to levels seen in B6 mice (Fig. 4G), suggesting that the defect in CD16−/− mice was a consequence of reduced BAFF. These data demonstrate that CD16 ligation by IgG-ICs induces DCs to secrete BAFF and that CD16 is necessary for memory responses. Although our findings collectively show that binding of IgG-ICs to CD16 contributes to B cell memory through the effects of BAFF on Bcl-6 expression in GC B cells, and maintaining and/or expanding Tfh cells, sources of BAFF other than CD16—or factors other than BAFF—may also play a role because loss of either BAFF or CD16 did not completely ablate these populations.
Loss of CD16 diminishes GC responses
To assess whether loss of CD16 diminished GCs, we measured the frequency of splenic GC B cells (CD19+GL-7+CD95+) and Ac38 Id+GC B cells 7 d after immunization. We found that both populations of GC B cells were significantly reduced in CD16−/− mice (Fig. 5A–C). Consistent with the diminished frequency of GC B cells, we found that the GCs in CD16−/− mice were 54% smaller on day 7 postimmunization, but as the size of the B6 GCs declined (by days 14 and 21) the sizes became similar (Fig. 5D, Supplemental Fig. 2). The smaller size of the CD16−/− GC on day 7 was not the result of delayed kinetics because by day 28, the GCs in both B6 and CD16−/− mice were diminished, suggesting a comparable duration of the response (data not shown). We also found that immunized CD16−/− mice displayed a 2.3-fold reduction in the number of GCs on day 7; however, the differences were less apparent on day 14. By day 21, the CD16−/− mice had a similar number of GC structures (Fig. 5E). Thus, although GCs form in the absence of CD16, they are reduced in number and size early during the immune response. This finding suggests that stimulation of CD16 is an early event that enhances the GC response and subsequently optimizes the formation of memory B cells.
CD16−/− mice have diminished GC responses. (A and B) The frequency of splenic CD19+GL-7+CD95+ GC B cells was measured by flow cytometry from B6 and CD16−/− mice 7 d after immunization. n = 4–11 mice per group over three experiments. (C) The frequency of NP-specific GC B cells (CD19+Ac38+GL-7+CD95+) measured in B6 and CD16−/− mice 7 d after immunization. n = 3–10 mice over 3 experiments. (D) The size of GCs and (E) the number of GCs were enumerated from B6 and CD16−/− mice 7, 14, and 21 d after immunization, using confocal microscopy. n = 3–6 mice per time point over four experiments. Bars display mean (B and C), and error bars indicate SEM (D and E). *p ≤ 0.05, **p ≤ 0.01.
CD16−/− mice have diminished GC responses. (A and B) The frequency of splenic CD19+GL-7+CD95+ GC B cells was measured by flow cytometry from B6 and CD16−/− mice 7 d after immunization. n = 4–11 mice per group over three experiments. (C) The frequency of NP-specific GC B cells (CD19+Ac38+GL-7+CD95+) measured in B6 and CD16−/− mice 7 d after immunization. n = 3–10 mice over 3 experiments. (D) The size of GCs and (E) the number of GCs were enumerated from B6 and CD16−/− mice 7, 14, and 21 d after immunization, using confocal microscopy. n = 3–6 mice per time point over four experiments. Bars display mean (B and C), and error bars indicate SEM (D and E). *p ≤ 0.05, **p ≤ 0.01.
CD16 and BAFF are required for the expression of Bcl-6 and the formation of GC and memory B cells
Our data indicate that CD16 and IgG-ICs are required for DCs to make BAFF in response to NP14KLH (Fig. 4A). To address whether DC-derived BAFF was sufficient to restore the GC and memory B cell pools in the CD16−/− mice, we adoptively transferred BMDCs from BAFF Tg mice at the time of immunization. We found that BAFF Tg DCs restored the frequency of GC B cells on day 7 (Fig. 6A) and the frequency of Ac38 Id+ memory B cells on day 28 (Fig. 6B). This observation suggests that the lack of B cell memory responses in CD16−/− mice was due to lack of BAFF, and not to intrinsic defects in B cells. In contrast, BMDCs derived from BAFF−/− chimeras were unable to restore GC or memory B cells (Fig. 6A, 6B). These results also emphasize that the defects observed in CD16-deficient mice are mediated by BAFF because DCs from BAFF−/− chimeras have intact CD16.
Adoptive transfer of BAFF-expressing DCs rescues GC and memory B cell populations and restores Bcl-6 levels in CD16−/− mice B cells. (A) BMDCs (8 × 106) from BAFF Tg or BAFF−/− mice were injected (s.c.) into CD16−/− mice that were simultaneously immunized (s.c.) with NP14KLH. On day 7, the frequency of GC B cells (CD19+GL-7+CD95+) was measured from inguinal lymph nodes. n = 3–7 mice per group over three experiments. (B) Same as (A) but on day 28, the frequency of CD19+Ac38+IgG+ memory B cells was measured. n = 3–4 over two experiments. (C) BAFF Tg and BAFF−/− BMDCs (8 × 106) were injected into CD16−/− mice that were simultaneously immunized with NP14KLH. n = 4 mice over three experiments. On day 7, Bcl-6 expression in GC B cells from inguinal lymph nodes was measured by flow cytometry. Bars display means (A and B), and error bars indicate SEM (C). *p ≤ 0.05, **p ≤ 0.01.
Adoptive transfer of BAFF-expressing DCs rescues GC and memory B cell populations and restores Bcl-6 levels in CD16−/− mice B cells. (A) BMDCs (8 × 106) from BAFF Tg or BAFF−/− mice were injected (s.c.) into CD16−/− mice that were simultaneously immunized (s.c.) with NP14KLH. On day 7, the frequency of GC B cells (CD19+GL-7+CD95+) was measured from inguinal lymph nodes. n = 3–7 mice per group over three experiments. (B) Same as (A) but on day 28, the frequency of CD19+Ac38+IgG+ memory B cells was measured. n = 3–4 over two experiments. (C) BAFF Tg and BAFF−/− BMDCs (8 × 106) were injected into CD16−/− mice that were simultaneously immunized with NP14KLH. n = 4 mice over three experiments. On day 7, Bcl-6 expression in GC B cells from inguinal lymph nodes was measured by flow cytometry. Bars display means (A and B), and error bars indicate SEM (C). *p ≤ 0.05, **p ≤ 0.01.
Our data suggest that DC-derived BAFF acts at or upstream of Bcl-6 (Figs. 1I, 1J, 2C, 3A–C) and downstream of CD16 (Fig. 4A). Thus, the absence of CD16 should also lead to diminished Bcl-6 expression after NP immunization, and restoring DC-derived BAFF by BMDC transfer should restore Bcl-6 levels. To test this, we adoptively transferred BAFF Tg or BAFF−/− BMDCs into CD16−/−mice at the time of immunization. We found that 7 d after immunization, the expression levels of Bcl-6 in CD16−/− GC B cells were 60% lower compared with B6 controls. Transfer of BAFF Tg BMDCs, but not BAFF−/− BMDCs, restored Bcl-6 levels in CD16−/− GC B cells (Fig. 6C). Thus, loss of CD16 reduces BAFF, which affects Bcl-6 expression in GC B cells. Collectively, the data show that IgG-ICs induce DC-derived BAFF through ligation of CD16. BAFF promotes optimal B cell memory responses by inducing Bcl-6 expression in GC B cells and by maintaining and/or expanding Tfh cells.
Discussion
Interactions between T cells, B cells, and DCs are necessary for the proper execution of the adaptive immune response. This study identifies a previously unappreciated role for FcγRs and BAFF in the early events of the GC response. We show that ICs formed during the early IgG response to NP14KLH induced the production of BAFF by DCs. BAFF acted at, or upstream of, Bcl-6 to promote the optimal formation of GC B cells and to maintain and/or expand newly formed (day 7) Tfh cells. This series of events depends on the formation of IgG-ICs, suggesting that productive early Ab responses contribute to the optimal formation of B cell memory. This mechanism may ensure that an Ag-specific Ab response has occurred prior to initiating the events that promote B cell memory.
In the absence of CD16, the diminished formation of GC and memory B cells, and reduced expansion of Tfh cells (Figs. 4–6), coupled with observations that mice deficient in the Fc common γ-chain (Fcγc) exhibit diminished secondary Ab responses (59) imply that IgG is required for optimal adaptive immune responses. We detected Ag-specific IgG by ELISPOT within 2–3 d of immunization. The source of this IgG could be extrafollicular PCs because these cells class switch (63) and produce significant local levels of IgG early in the immune response (64, 65). By days 14 and 21, the sizes and numbers of GCs were similar in B6 and CD16-deficient mice, suggesting that this mechanism is involved in pre-GC or early GC responses. This observation may also reflect a reduction in available Ag, which would reduce IC formation and make any differences mediated by IC:FcγR interactions less apparent. Consistent with a role for early GC events in the formation of B cells memory, we also found that in the absence of CD16, the average GC size and the number of GCs were diminished on day 7. Further, others showed that in the absence of soluble IgG, the kinetics of secondary GCs were disrupted (66). Thus, IgG plays a role in both primary and secondary GC responses.
Temporally, anti-NP IgG production by ASCs was coincident with BAFF production by DCs (Fig. 2D, 2E). Although both DCs and MFs are among the major producers of BAFF (46, 67–69), we found that transfer of BAFF-producing DCs, but not MFs, restored the numbers of GC B cells in BAFF−/− chimeras and CD16−/− mice. This finding may reflect the tripartite colocalization and interactions between DCs, T cells, and B cells during adaptive immune responses. Although they are most well known for presenting Ag to and activating T cells, DCs also directly modulate B cells through production of IC-induced cytokines like IFN, BAFF, and IL-12 (55, 56, 70). Our data do not rule out a role for other immune cell types expressing CD16, like neutrophils, but highlight a key role for DCs as BAFF producers in T-dependent adaptive immune responses.
Because GC B cells, Tfh cells, or memory B cells were not completely absent in the BAFF−/− chimeras or the CD16−/− mice, it is also possible that BAFF from non–hematopoietic cell sources contributes to the response, or that BAFF is not the only factor induced by IgG-ICs that is involved in early GC events. Early in the adaptive immune response, B cells interact with cognate T cells and costimulatory signals induce AID activation and class switch (71). These signals include CD40L, which is also required for GC formation (70). We found that BAFF−/− chimeras had normal AID expression and primary IgG production, suggesting that cognate interactions mediated by CD40L were intact, possibly accounting for the small number of GC and memory B cells that remain despite the absence CD16 or BAFF. It is unlikely that BAFF and CD40L are completely redundant because diminished memory B cell responses were noted in both BAFF−/− chimeras (Figs. 1, 2) and CD40L-deficient mice (72, 73).
It is unclear whether the observed role of BAFF occurs before, during, or after the initial interaction between cognate B and T cells. Because AID induction is intact in BAFF−/− chimeras, BAFF likely acts downstream of initial T:B interactions. Our in vitro data show that recombinant BAFF directly induced Bcl-6 in activated B cells, and DC CM (elicited under conditions that induce BAFF) diminished the expression of XBP-1 and IRF-4. This finding is supported by our in vivo data, in which increased numbers of PCs were evident in BAFF−/− chimeras. This observation suggests that BAFF acts on activated B cells at the time the PC phenotype is being diminished. Therefore, BAFF may act on B cells after cognate T cells help to reduce the PC phenotype, and either commit cells to the memory or GC pathways or support the survival of precursors that will enter those pathways. This idea is consistent with studies showing that DC secretion of the PC-inducing cytokine IL-12 is dampened by IC signaling (74).
Another signal required for a productive GC reaction is IL-21. It is possible that BAFF acts in concert with IL-21 to induce and/or sustain expression of Bcl-6 in GC B cells (23, 24). T cell–secreted IL-21 acts on B cells both during initial T:B interactions and after GCs are formed, to promote either Blimp-1 or Bcl-6 expression, depending on the context (75). BAFF may serve as a contextual signal early in the adaptive response to direct B cells toward a GC fate. BAFF may also maintain Bcl-6 in B cells destined for the GC, allowing IL-21 from Tfh cells to take over as the response progresses.
Previous studies showed that cells with a Tfh phenotype appear by day 3 after immunization (10, 76). These cells migrate toward follicles, where they interact with B cells at the T:B border (77) to promote continuous expression of Bcl-6 and entry of Tfh cells into the GC (15, 78, 79). Our data suggest that B cells and IgG-ICs are not involved in the formation of Tfh cells (day 3). This suggestion is consistent with previous studies showing that on days 1–3 postimmunization, the expression of Bcl-6 and CXCR5 in CD4+ T cells is independent of B cells, and hence independent of IgG (10, 56). Instead, BAFF acted downstream of the formation of Tfh cells, and stabilized and/or expanded the population between days 3 and 7. Thus, previous studies showing a role for B cells in maintaining Tfh cells (22, 76, 80) might reflect the need for B cell–elicited Ig, ICs, and DC-derived BAFF. The role of BAFF in maintaining the Tfh population may be direct or may involve GC B cells. One possibility is that BAFF promotes the expression of ICOSL on B cells. Previous studies showed that signaling through BAFF-R regulates the expression of ICOSL on B cells (32, 33), thereby sustaining interaction between Tfh cells and B cells at the T:B border and within GCs (10, 11). This interaction could also stabilize the expression of Bcl-6 and the downstream molecules CXCR5 and PD-1 (14, 15, 81) to maintain the Tfh phenotype. This possibility is supported by studies showing that the absence of ICOSL on B cells reduces the frequency of CXCR5+ CD4+ cells after immunization (15), that ICOS:ICOSL interaction prolongs the engagement between B cells and Tfh cells (78), and that follicular bystander B cells support the formation and/or maintenance of Tfh cells by providing ICOSL in an Ag-independent manner (82).
Because the interaction between GC B cells and Tfh cells is bidirectional, the maintenance of Tfh cells may have also had a role in sustaining GC B cells. Recent work described Tfh as a source of local BAFF within GC B cells that is required for affinity maturation (5). Tfh-derived BAFF, however, is not required for GC initiation or maintenance (5), and its role in memory B cell development remains unknown. Our work indicates that early production of BAFF is upstream of GC formation and memory B cell development. Of interest, we observed a more complete loss of Ag-specific GC B cells (Ac38+) than total GC B cells in BAFF−/− chimeras and CD16−/− mice. This observation may reflect a loss of local BAFF within the GC owing to diminished maintenance and/or expansion of Tfh cells. However, more work would be needed to determine whether affinity maturation is altered in these models.
Overall, our studies highlight a novel role for IgG-ICs and DC-derived BAFF in the GC response. Elucidating the events that initiate GC responses may have an impact on our understanding of ICs and BAFF in autoimmunity. Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by elevated levels of BAFF, autoantibody/autoantigen ICs, and multiorgan disease. The formation of autoreactive memory is thought to be instrumental in driving long-lived PCs and sustaining autoantibody production (83, 84); however, the mechanisms that regulate memory formation to self-antigens are unclear. Our findings suggest that chronically high levels of ICs containing self-antigens could contribute to a break in B cell tolerance at the GC checkpoint. In SLE, elevated levels of circulating ICs may elevate BAFF and promote the GC response (83, 85, 86). This finding suggests that neutralization of BAFF in patients with SLE may affect both B cell survival and GC responses that are necessary in the formation of autoreactive B cells, T cells, and memory cells (67, 87, 88).
Acknowledgements
We thank J. Sjef Verbeek for the CD16- and CD64-deficient mice; Jeff Ravetch for CD16-2–deficient mice; Jin Kim (Centers for Disease Control/Office of Infectious Diseases/National Center for Immunization and Respiratory Diseases) for helpful discussions; the Flow Cytometry Core (National Cancer Institute P30CA06086); and the Microscopy Services Laboratory (CA 16086-26) for support.
Footnotes
This work was supported by National Institutes of Health Grants R01 AI070984 and R21 AI105613 and the Lupus Research Institute. A.B.K. and S.Z.J. were supported by National Institutes of Health Grant 5T32AI07273-27, and S.Z.J. was also supported by a minority supplement to National Institutes of Health Grant AI070984.
The online version of this article contains supplemental material.
Abbreviations used in this article:
References
Disclosures
The authors have no financial conflicts of interest.