Abstract
Upon activation with T-dependent Ag, B cells enter germinal centers (GC) and upregulate activation-induced deaminase (AID). AID+ GC B cells then undergo class-switch recombination and somatic hypermutation. Follicular dendritic cells (FDC) are stromal cells that underpin GC and require constitutive signaling through the lymphotoxin (LT) β receptor to be maintained in a fully mature, differentiated state. Although it was shown that FDC can be dispensable for the generation of affinity-matured Ab, in the absence of FDC it is unclear where AID expression occurs. In a mouse model that lacks mature FDC, as well as other LT-sensitive cells, we show that clusters of AID+PNA+GL7+ Ag-specific GC B cells form within the B cell follicles of draining lymph nodes, suggesting that FDC are not strictly required for GC formation. However, later in the primary response, FDC-less GC dissipated prematurely, correlating with impaired affinity maturation. We examined whether GC dissipation was due to a lack of FDC or other LTβ receptor–dependent accessory cells and found that, in response to nonreplicating protein Ag, FDC proved to be more critical for long-term GC maintenance. Our study provides a spatial-temporal analysis of Ag-specific B cell activation and AID expression in the context of a peripheral lymph node that lacks FDC-M1+ CD35+ FDC and other LT-sensitive cell types, and reveals that FDC are not strictly required for the induction of AID within an organized GC-like environment.
Introduction
T-dependent immune responses take place within germinal centers (GC) located in the follicles of secondary lymphoid organs. GC are clusters of activated B cells that undergo proliferation, class-switch recombination (CSR), somatic hypermutation (SHM), and clonal selection. Activation-induced deaminase (AID) is a DNA-editing enzyme that is expressed by B cells following activation and is required to initiate CSR and SHM (1, 2). B cells upregulate AID as they enter GC so that they may undergo Ab diversification. GC B cells subsequently downregulate AID as they exit GC to become memory or plasma cells (3).
The initiation of GC requires the activation of B cells by binding Ag and by receiving costimulation from CD4+ T cells. Therefore, delivery of Ag into lymphoid follicles is crucial to this process. Although small soluble Ag can enter lymph node (LN) follicles through conduits (4), large Ag can be trapped and transported into follicles by subcapsular sinus macrophages (SCS MΦ) in a complement receptor–dependent manner (5, 6). The nature of an Ag affects the kinetics and magnitude of the response that it elicits: smaller protein Ag accumulates poorly in follicles until the appearance of Ag-specific Ig enables the formation of larger immune complexes, whereas larger multimeric Ag, such as virus-like particles, can be bound early on by natural IgM, enabling them to form complement-containing immune complexes within follicles (7).
Once Ag is transported into the follicle, it is deposited on a type of nonhematopoietic stromal cell known as the follicular dendritic cell (FDC). FDC secrete chemokines and survival factors important for GC structure and function (8). The maintenance of FDC maturation status (phenotype and function) requires constitutive signaling through the lymphotoxin (LT) β receptor (LTβR); FDC are induced to mature in situ from a perivascular precursor through engagement of LTβR by membrane-bound LTαβ expressed on lymphocytes (9, 10). Within the GC, B cells express even higher levels of LTαβ, which is likely important for orchestrating a tight FDC network to support maturation of the Ab response (11, 12). Because LTβR signaling is required to maintain identifiable FDC within lymphoid tissues (13), mice rendered deficient in LTβR signaling by deletion of LTα, LTβ, or LTβR have been studied to query the role of FDC in the GC response. Notably, inhibition of LTβR signaling results in the impaired formation of splenic GC (14–16). Supporting the concept that FDC are required for GC structures, selective ablation of FDC using diphtheria toxin terminates the GC response in peripheral LN (17). However, GC formation in the mesenteric LN was shown to occur normally in LTβ−/− mice (18), and mice deficient in LTβR only exhibit a defect in affinity maturation at low doses of Ag (19). Variable dependency on FDC for a GC response may depend on whether Ag is particulate versus adjuvanted, as demonstrated in TNFRI−/− mice, which also lack FDC (20). Indeed, in the context of viral infection, although immunization of TNFRI−/− mice with viral protein results in a defective Ab response, immunization with live virus induces strong titers of neutralizing IgG (21). Thus, the mode of Ag delivery and the site of Ag encounter (spleen versus peripheral LN versus mucosal LN) are variables that likely affect the dependency on FDC for an effective humoral immune response.
Once formed, GC are the primary location of AID expression (3, 22). Although the mutagenic activity of AID is required for the generation of a high-affinity Ab response, this mutagenic activity carries the collateral risk of the potential generation of autoreactive B cells. For this reason, it would be expected that the expression of AID is restricted to B cells in microenvironments where they are regulated and subjected to tolerance checks. GC, which contain FDC and CD4+ T follicular helper cells, could be such an environment, explaining why AID activity is primarily observed in GC (3, 22). However, somatically mutated autoreactive B cells in MRL/lpr mice have been found outside of GC (23), and class-switched B cells can be aberrantly localized to the T cell zone in TNFRI−/− mice (20). The location and persistence of AID expression in the context of an FDC-less LN is not known. Understanding the minimal spatio-temporal requirements for AID expression has important clinical implications because inflamed nonlymphoid tissues can also support B cell–rich follicle-like structures (24).
We sought to initiate a robust, class-switched humoral response that would generate GC B cells whose AID expression and Ag specificity could be tracked by flow cytometry and immunofluorescent histology in FDC-less LN of LTβ−/− chimeric mice. We observed clusters of AID+ GC B cells in the absence of FDC in the follicles of LN of LTβ−/− chimeric mice. These AID+ GC B cells appeared with surprisingly normal kinetics, and their phenotype was very similar to AID+ GC B cells generated within FDC-sufficient GC. However, following the peak of the response, in the absence of LTβ and FDC, GC disappeared abruptly, and such abortive GC correlated with impaired affinity maturation. During this later phase of the GC, we used a combination of genetic and pharmacological approaches to ablate SCS MΦ and/or FDC and observed a weak correlation between the number of SCS MΦ and the number of GC B cells, whereas a significant decrease in the number of GC B cells late in the primary response occurred only when FDC were ablated. Our findings show that FDC-M1+ CD35+ FDC are not strictly required for the generation or clustering of AID+ GC B cells. In addition, our results suggest that FDC are required for the long-term maintenance of GC and the resulting production of high-affinity Ab.
Materials and Methods
Mice and immunizations
C57BL/6 mice were purchased from Charles River Laboratories. AID-GFP–transgenic mice were obtained from Rafael Casellas (National Institutes of Health, Bethesda, MD). LTβ−/− mice were purchased from B&K Universal and crossed with AID-GFP mice to produce LTβ−/− × AID-GFP mice. LTβR−/− mice with the CD45.1 congenic allele were obtained from Rodney Newberry (Washington University School of Medicine, St. Louis, MO). Mice congenic for the CD45.1 allele on the C57BL/6 background were obtained from The Jackson Laboratory (stock 002014). Bone marrow (BM) chimeras were generated as previously described (25). Chimeric mice were immunized s.c. in both hind flanks with 50 μl of a 1:1 PBS/CFA mixture (CFA from Difco Laboratories) containing 0.4 mg/ml (high-dose) or 0.035 mg/ml (low-dose) R-phycoerythrin (R-PE, Anaspec) or an irrelevant Ag (4-hydroxy-3-nitrophenylacetyl hapten conjugated to chicken gamma globulin [NP-CGG]; Biosearch Technologies). In some experiments, mice were treated by injecting 100 μg LTβR-Ig or control MOPC21 mAb (26) (Biogen-Idec) i.p. at days −2 and 5 postimmunization (p.i.). All mice (both chimeric and nonchimeric) were housed in specific pathogen–free conditions. All animal experiments were performed in accordance with end points and standards of animal care approved by the University of Toronto, Faculty of Medicine Animal Ethics Committee (Protocol # 20008480).
Cell isolation and flow cytometry
Inguinal LN (iLN) from immunized mice were ground between glass slides and suspended in HBSS with 1 mg/ml collagenase D and 60 μg/ml DNase I (Roche Diagnostics) for 30 min at 37°C. Cell suspensions were then filtered through a 70-μm strainer and resuspended in PBS. Cells were labeled with Live/Dead Aqua (Life Technologies) prior to staining or with 7-aminoactinomycin D (BD), and dead cells were gated out. Prior to staining, cells were blocked with mouse serum and 2.4G2 Ab (Fc block). Cells were stained with Abs against B220 (eFluor 450; mAb RA3-6B2), CD1d (PE; mAb 1B1), CD23 (biotin; mAb B3B4), CD11b (PerCPCy5.5; mAb M1/20), CD11c (allophycocyanin; mAb N418), F4/80 (biotin; mAb BM8) (all from eBioscience); with Abs against CD169 (FITC; mAb MOMA-1) (GeneTex); or with Abs against IgM (PerCPCy5.5; mAb R6-60.2), CD21 (FITC; mAb 7G6), GL7 (Alexa Fluor 647), and Fas (PECy7; mAb Jo2) (all from BD). Biotin-conjugated Abs were stained with streptavidin-allophycocyanin or streptavidin–allophycocyanin–eFluor 780 (eBioscience). Ag-specific B cells were labeled with R-PE. Fluorescent-labeled cells were analyzed on a BD FACSCanto II flow cytometer.
Histology
iLN from immunized mice were processed for histology, as previously described (12). Tissue sections were blocked with mouse serum prior to staining. Serial sections were stained with allophycocyanin-conjugated Abs against B220 (mAb RA3-6B2; eBioscience), FDC-M1, CD35 (mAb 8C12), or CD138 (mAb 281-2) (BD) or biotin-conjugated peanut agglutinin (PNA, Vector Labs). Serial sections were also stained with anti-IgD (mAb 11-26; eBioscience) or anti-fibronectin (Sigma). Biotin-conjugated reagents were then stained with streptavidin-allophycocyanin (eBioscience). Unlabeled rat anti-mouse ER-TR7 (BMA Biomedical) was labeled with anti-rat–FITC (Southern Biotech). Ag-specific Ig was detected by applying R-PE directly to the tissue (Anaspec). The AID-GFP reporter transgene was too dim for direct histological visualization and so was amplified by staining with chicken anti-GFP polyclonal Ab (Aves Labs), followed by anti-chicken–Alexa Fluor 488 (Life Technologies). Stained sections were mounted with Dako Fluorescent Mounting Medium and visualized on a Leica DMRA2 microscope at room temperature. Photographs were taken with 10× and 20× objectives using a QImaging Retiga EXi camera with an EXFO X-Cite 120 lamp using Openlab software.
ELISA
Blood was collected from immunized mice and centrifuged to isolate sera. To measure Ag-specific Ab, wells in NUNC MaxiSorp plates (Thermo Scientific) were coated with 10 μg/ml R-PE (Anaspec) overnight at 4°C and then blocked with 2% BSA/PBS for 1 h at 37°C. To measure relative Ab titers, R-PE–coated wells were incubated with a dilution series of serum for 2 h at 37°C and then probed using biotinylated Abs against IgG (polyclonal Ab Poly4053) or IgM (mAb RMM-1) (BioLegend). Wells were then incubated with streptavidin-HRP enzyme, developed with liquid tetramethylbenzidine substrate (BioShop), stopped with 1 M H2SO4, and read at 450 nm. Relative titers were calculated as the serum dilution at 1 OD relative to a serum control. To measure relative Ag affinity, we used a protocol adapted from the literature (27). Briefly, R-PE–coated wells were incubated with a quantity of serum, such that each well for each sample contained the same amount of anti–R-PE Ab (as calculated by relative titer ELISA). Wells were then incubated for 15 min with 0 to 1.2 M ammonium thiocyanate in 0.1 M phosphate buffer at pH 7.15. Wells were then probed and developed as described above. The resulting dilution curve was fitted with a third-order polynomial regression line. The affinity index of each sample was calculated from the regression line as the molar concentration of ammonium thiocyanate required to remove 50% of bound anti–R-PE serum Ab.
Measuring SHM in LN GC B cells
AID-GFP (wild-type [WT]) and AID-GFP × LTβ−/− BM chimeras were immunized with a high dose of R-PE, as described above, and LN were collected at day 9 p.i. LN B cells from three mice of each genotype were pooled together for sorting. B220+AID-GFP+R-PE+GL7+ GC B cells were isolated by FACS on a BD FACSAria flow cytometer. Sorted cells were suspended in TRIzol (Life Technologies), and the JH2–JH4 region of extracted DNA was amplified, as previously described (28), with the following modifications: PCR products were amplified by Q5 High-Fidelity DNA polymerase (New England BioLabs) at 68°C annealing temperature. Amplified DNA was gel purified with QIAquick gel extraction kit (QIAGEN) and cloned into Zero Blunt TOPO plasmid vector (Life Technologies) before transformation into DH5α bacteria. Plasmid DNA was prepared using an E-Z 96 FastFilter plasmid kit (Omega Bio-Tek) and sequenced by an ABI3730XL automatic sequencer (Macrogen).
Results
Phenotypically normal GC B cells are observed in the absence of LTβ
In this study, we developed a system for examining the humoral immune response within LN using tools that allow us to track Ag-specific B cells that express AID. Specifically, we immunized mice with the 240-kDa fluorochrome R-PE in CFA to follow an Ag-specific response within the endogenous polyclonal repertoire because B cells activated by R-PE in vivo can later be stained with R-PE ex vivo (29, 30). For detecting AID expression in Ag-specific B cells, we used mice harboring a transgene with an AID-GFP fusion protein under the control of AID cis-elements. The AID-GFP reporter induces fluorescence in AID-expressing GC B cells, which is subsequently quenched upon downregulation of AID (3). LTβ−/− mice lack peripheral LN. Therefore, we crossed LTβ−/− mice with AID-GFP reporter mice to track AID expression in the absence of LTβ and used these mice (as well as WT AID-GFP control mice) as the source of donor BM to transfer into C57BL/6 hosts. This allowed us to induce GC within draining iLN in the presence or absence of LTβ, creating AID-GFP (WT) BM chimeras and AID-GFP × LTβ−/− BM chimeras (Supplemental Fig. 1A).
We confirmed the ablation of LTαβ-induced LTβR signaling in AID-GFP × LTβ−/− BM chimeras based on the lower cellularity of the peripheral LN (Supplemental Fig. 1B), consistent with a role for LTβR signaling in regulating the maturation status of high endothelial venules (31). AID-GFP × LTβ−/− BM chimeras also had severely reduced populations of splenic marginal zone B cells, consistent with observations that the marginal zone of the spleen collapses in the absence of LTβR signaling (Supplemental Fig. 1C) (32).
We next tested the GC response to R-PE in the draining iLN of AID-GFP (WT) and AID-GFP × LTβ−/− BM chimeras. Validating the specificity of R-PE staining, we observed that mice immunized with an irrelevant Ag (NP-CGG) generated AID-GFP+ B cells that are not stained by R-PE, and B cells from mice immunized with R-PE do not fluoresce in the PE channel when R-PE is not added to the flow cytometry panel (Fig. 1A). We then immunized AID-GFP (WT) and AID-GFP × LTβ−/− BM chimeras s.c. with R-PE in each hind flank to generate humoral responses in both iLN. At day 7 of the response, we observed a population of AID-GFP+R-PE+ B cells in AID-GFP (WT) and AID-GFP × LTβ−/− BM chimeras by flow cytometry (Fig. 1B). Surprisingly, the percentage of AID-GFP+R-PE+ B cells in the iLN of AID-GFP × LTβ−/− BM chimeras was equal to or greater than in the iLN of AID-GFP (WT) chimeric mice (Fig. 1C). We analyzed the expression of other GC B cell markers on AID-GFP+R-PE+ B cells from both types of BM chimeric mice and found that they displayed the same levels of Fas, GL7, and AID-GFP (Fig. 1D). A population of AID−R-PElow cells was also observed (Fig. 1B). It is unclear whether these are B cells with low affinity for R-PE or represent cells staining for FcR-bound anti–R-PE Ab; nevertheless, the frequency and number of these cells were similar for both groups (data not shown). Taken together, these observations indicate that significant numbers of phenotypically normal GC B cells are produced in the draining iLN in the absence of LTβ.
AID-GFP+ GC B cell clusters form in the absence of LTβ and FDC
Given that AID-GFP+ GC phenotype B cells are readily detected in the draining iLN of R-PE–immunized AID-GFP × LTβ−/− BM chimeric mice, we sought to determine the location of AID expression by immunofluorescence microscopy. At day 7 p.i., serial sections of iLN from AID-GFP (WT) BM chimeric mice were generated, and clusters of AID-GFP+ B cells were observed within B cell follicles (Fig. 2A). These clusters of AID-GFP+ cells stained with PNA, indicating that they contained activated GC B cells (Fig. 2A). These GC were populated with FDC-M1+ CD35+ FDC (Fig. 2A). Staining serial sections with R-PE revealed R-PE–specific Ab secreted by CD138+ plasma cells in clusters adjacent to GC and in the medullary cords (Fig. 2A). Surprisingly, iLN from AID-GFP × LTβ−/− BM chimeras also contained clusters of AID-GFP+PNA+ B cells within B cell follicles (Fig. 2B), as well as R-PE+CD138+ cells in clusters adjacent to GC and in the medullary cords (Fig. 2B). However, as expected, these clusters did not contain any FDC, as verified by the absence of FDC-M1 and CD35 staining (Fig. 2B, Supplemental Fig. 2A, 2B). Interestingly, although FDC networks were disrupted in LTβ−/− BM chimeras, ERTR7/fibronectin conduits produced by fibroblastic reticular cells (FRC) (4, 33) remained intact and indeed infiltrated FDC-less GC (Fig. 3).
Taken together, these observations show that clusters of AID+PNA+ GC B cells form in the iLN of AID-GFP × LTβ−/− BM chimeras in the absence of FDC. Because these B cells form clusters situated in the follicle and because they express numerous markers of the GC B cell phenotype (AID, GL7, Fas, PNA), we elected to call these clusters GC, despite the lack of an FDC-M1+ CD35+ FDC network.
LTβ is required for the maintenance, but not the initiation, of GC
We next investigated the kinetics of the GC response in FDC-less BM chimeras to examine whether the absence of LTβ could result in any kinetic or qualitative differences. We immunized AID-GFP (WT) and AID-GFP × LTβ−/− BM chimeric mice as before and quantified AID-GFP+R-PE+ B cells over time. GC B cells were not present in iLN at day 5 p.i., but they appeared at day 6 (Fig. 4A). The initial response in AID-GFP × LTβ−/− BM chimeras appeared delayed; there were slightly fewer AID-GFP+R-PE+ B cells in AID-GFP × LTβ−/− BM chimeras at day 6; however, that number had caught up to or exceeded the number found in AID-GFP (WT) BM chimeras by day 7 (Fig. 4A). GC B cells were still present by day 12 in AID-GFP (WT) BM chimeras (Fig. 4B). In contrast, the response in AID-GFP × LTβ−/− BM chimeras had vanished by day 12 (Fig. 4B). Similar results were observed using immunofluorescence microscopy (Fig. 5).
Because FDC are thought to trap Ag through complement and FcRs for protracted presentation to Ag-specific B cells (8), we considered that high doses of Ag may circumvent the requirement for FDC-mediated Ag capture. To test this, we repeated the kinetic analysis by immunizing with a low dose of Ag (1.75 μg R-PE/injection, as opposed to 20 μg for the high dose). The low-dose immunization resulted in much weaker GC responses that were sustained until day 12 in AID-GFP (WT) BM chimeras, whereas they had completely dissipated by day 12 in the AID-GFP × LTβ−/− BM chimeric mice (Fig. 4C, 4D). Therefore, LTβ is dispensable for the initiation of Ag-specific GC B cell expansion in the draining iLN but is critical for postpeak maintenance of GC B cell numbers both at low and high doses of Ag.
LTβ is required for optimal affinity maturation
Given that GC B cells were not sustained in the draining iLN of AID-GFP × LTβ−/− BM chimeras following immunization, we assessed the impact of GC collapse on the affinity of the anti–R-PE humoral response after high-dose immunization. Using ELISA, we found that the production of serum anti–R-PE IgM and IgG was roughly equal in AID-GFP (WT) and AID-GFP × LTβ−/− BM chimeras (Fig. 6A, 6B). However, AID-GFP × LTβ−/− BM chimeras produced serum IgG with significantly lower affinity for R-PE, indicating a defect in affinity maturation (Fig. 6C). To determine whether the impairment in affinity maturation was due to reduced SHM, we isolated B220+AID-GFP+R-PE+GL7+ GC B cells from the LN of AID-GFP (WT) and AID-GFP × LTβ−/− BM chimeras by FACS and sequenced their JH2–JH4 regions to look for AID-induced mutations. Because Ag-specific AID+ GC B cells are not detectable at day 12 p.i., we elected to look for evidence of SHM at day 9 p.i. At this time point we found that the mutation frequency was ∼30% lower among GC B cells in LTβ−/− versus WT chimeric mice, although the difference was not statistically significant (Table I, Supplemental Fig. 3). Therefore, we conclude that, during the first 9 d of the immune response, SHM occurs in AID-GFP × LTβ−/− BM chimeras. Together, these results indicate that the inability to sustain the GC response in AID-GFP × LTβ−/− BM chimeric mice corresponds with poor affinity maturation.
. | WT . | LTβ−/− . |
---|---|---|
No. sequences analyzed | 47 | 39 |
No. mutations | 19 | 11 |
Mutation frequencya | 2.68 × 10−4 | 1.87 × 10−4 |
Mutations at WRC (%)b | 26.3 | 18.2 |
Mutations at WA (%)b | 47.4 | 81.8 |
. | WT . | LTβ−/− . |
---|---|---|
No. sequences analyzed | 47 | 39 |
No. mutations | 19 | 11 |
Mutation frequencya | 2.68 × 10−4 | 1.87 × 10−4 |
Mutations at WRC (%)b | 26.3 | 18.2 |
Mutations at WA (%)b | 47.4 | 81.8 |
Frequency is defined as mutations/bp sequenced. WT: 19/70,782; LTβ−/−: 11/58,734.
Percentage of mutations calculated over total number of mutations.
W, A/T nucleotides; R, A/G nucleotides.
Premature GC dissipation is primarily due to the absence of FDC
Although the defects of early GC dissipation and poor affinity maturation in AID-GFP × LTβ−/− BM chimeras correlate with the absence of FDC, we could not conclude from those experiments that these defects were caused by the absence of FDC. This is because abrogation of LTβR signaling by removing LTβ has effects other than ablating FDC. Most relevant to the LN GC response is that the abrogation of LTβR signaling results in a significant reduction in the number of SCS MΦ (6). SCS MΦ trap immune complexes in the SCS and then pass them on to noncognate B cells, which, in turn, deliver the immune complexes to FDC for long-term retention (5, 6). Thus, interactions between FDC and SCS MΦ may be important for the generation of Ag deposits within GC that are required to sustain a prolonged GC response.
To determine whether the absence of FDC, SCS MΦ, or both was responsible for the collapse of GC that we observe at day 12 p.i., we generated BM chimeras in which C57BL/6 mice were reconstituted with BM from LTβR−/− mice with the CD45.1 congenic allele. In these LTβR−/− BM chimeras, the lack of LTβR on hematopoietic cells results in a significant deficiency in SCS MΦ (6), whereas FDC, which are radioresistant, are unaffected. Treatment of LTβR−/− BM chimeras with LTβR–Ig, a soluble decoy receptor that effectively blocks LTβ–LTβR interactions in vivo (26, 34), allowed us to ablate FDC and/or SCS MΦ to systematically evaluate the contribution of each cell type to GC maintenance.
Using this approach, we immunized LTβR−/− and control WT (CD45.1) BM chimeras with R-PE as before and treated them with LTβR–Ig or control mAb. iLN from control-treated LTβR−/− BM chimeras had significantly fewer SCS MΦ than did WT BM chimeras (Fig. 7A) but still contained mature FDC (Supplemental Fig. 2C, 2D). On average, 97% of SCS MΦ were CD45.1+ in LTβR−/− and WT BM chimeras, indicating that the vast majority of SCS MΦ were donor derived (data not shown). Consistent with previous reports (6), treatment with LTβR–Ig resulted in a near complete loss of iLN SCS MΦ (Fig. 7A), as well as ablation of FDC in both LTβR−/− and WT BM chimeras. In LTβR−/− BM chimeras treated with control Ab, the number of GC B cells at day 12 p.i. was slightly lower than in WT BM chimeras, although this reduction was not statistically significant (Fig. 7B). A significant decrease in the number of GC B cells at day 12 p.i. was observed only when LTβR−/− BM chimeras were treated with LTβR–Ig (Fig. 7B), and we only observed a weak correlation between the number of SCS MΦ and the number of GC B cells in the iLN at day 12 p.i. (Fig. 7C). Together, these observations suggest that SCS MΦ play a partial role in the prolonged maintenance of GC in response to protein Ag, but that FDC, or the cooperation between FDC and SCS MΦ, are required for GC to persist beyond the peak of the primary response.
Discussion
Studies examining the spleens of LT-deficient mice and nonhuman primates treated with LT pathway inhibitors have consistently demonstrated poor or absent GC formation correlating with a lack of FDC (14–16, 35). Furthermore, FDC within Peyer’s patches play an important role in regulating GC responses to gut-derived Ag (36), and FDC networks are strongly associated with AID expression in ectopic lymphoid structures in the salivary glands of patients with Sjögren’s syndrome (24) and in the synovium of patients with rheumatoid arthritis (37). These previous studies suggest that FDC are critical for the formation of GC and for the concomitant expression of AID by GC B cells. However, the AID-dependent processes of SHM and CSR still occur in immunized LTα−/− mice, which lack FDC and display absent or abortive splenic GC (38). Therefore, we sought to determine where AID expression can occur if not in canonical GC.
Although we initially expected to observe abnormal locations for AID-expressing B cells, we were surprised to see clusters of AID+ GC B cells in the follicles of iLN that resembled GC, despite their lack of FDC. This suggests that FDC-M1+CD35+ FDC are not required to initiate GC or to foster AID expression in Ag-specific B cells within iLN in response to protein Ag. Our results suggest that peripheral LN may have different types of stromal cells that could substitute for some of the functions attributed to mature FDC. Indeed, we observed the presence of an ERTR7+Fibronectin+ network within FDC-less GC. Such matrix elements are produced by FRC and were shown to facilitate the movement of follicular chemokines, such as CXCL13, which could conceivably draw Ag-specific B cells together into a GC niche (4). Both FRC and marginal reticular cells can produce CXCL13 and are a putative source of this chemokine in the absence of FDC (39, 40). Indeed, it was reported that ablation of FDC does not result in a significant decrease in CXCL13 transcripts in peripheral LN (17), suggesting that FDC are not an essential source of CXCL13 in this location and that FRC or marginal reticular cells could potentially organize B cell follicles and GC. Therefore, our findings suggest a role for other stromal cell types in the formation of GC in peripheral LN. Future studies that perform proteomic analysis of putative compensatory non-FDC stromal cells in the LTβ-deficient environment potentially would provide a more complete understanding of LN stroma. Alternatively, noncognate B cells, which also can capture and transport Ag in the follicle, could potentially display Ag to GC B cells in the absence of FDC, providing a partial redundancy for the Ag-trapping role of FDC. The collapse of the GC in FDC-less mice at day 12 may reflect turnover of these particular B cells (5, 41, 42).
Previous studies demonstrating the correlation between a lack of FDC and a lack of GC focused on the splenic lymphoid environment. In this study, we investigated peripheral LN and observed GC forming in the absence of FDC. Our findings highlight the fact that the lymphoid environments in the spleen and in LN are likely structured differently. A previous study reported GC without FDC in mesenteric LN (18). However, mesenteric LN are unique in a few respects. Although peripheral LN in SPF mice experience periods of quiescence during which the follicles are devoid of GC, mesenteric LN are sites of constitutive GC activity as the result of stimulation by gut Ag (43). As well, the ontogeny of mesenteric LN differs from other LN; although peripheral LN require signaling through LTβR for their development, mesenteric LN are able to form in the absence of LTβ (44). For these reasons, mesenteric LN were perhaps thought of as a special case, and the prevailing view that FDC were crucial to GC formation remained. Our findings support these previous observations in the GALT and extend them to peripheral LN where we show that organized GC-like clusters of AID-expressing B cells are effectively induced within the LN B cell follicle at the peak of the response to protein Ag, despite a lack of FDC-M1+ CD35+ FDC.
A more recent study by Wang et al. (17) used a different experimental approach to investigate the role of FDC in supporting GC by eliminating FDC with diphtheria toxin. They found that their model of FDC ablation resulted in the abrupt disappearance of LN GC B cells, demonstrating that FDC are required to maintain normal GC at the peak of the immune response. We found that FDC were dispensable for the formation of GC that support peak numbers of AID+Fas+PNA+GL7+ Ag-specific B cells and that it was not until late in the response that GC were negatively affected by a lack of FDC. One difference between these two studies is that, in our study, FDC-M1+ CD35+ FDC were absent for the entire duration of the primary response, whereas in the study by Wang et al. FDC were ablated only after GC had already formed. Nevertheless, both studies demonstrate that an absence of FDC results in the disappearance of GC later in the primary response.
Although we found that FDC-M1+ CD35+ FDC are dispensable for the initial formation of GC in LN, we determined that the LT pathway is crucial for the long-term maintenance of GC. By using a combination of LTβR−/− BM chimeras and treatment with LTβR–Ig, our study was able to determine the specific contribution of FDC to the formation of GC and to contrast this with the role of other LTβR-dependent cells, such as SCS MΦ. Other effects of abrogating LTβR signaling include loss of a mature high endothelial venule phenotype (31), impaired dendritic cell (DC) function and/or homeostasis (45–47), and loss of SCS MΦ (6). SCS MΦ are important for the optimal transport of Ag into the follicle (5, 6). We observed a weak correlation between the number of SCS MΦ and the number of GC B cells late in the primary response. However, significantly reducing the number of SCS MΦ did not result in a significant decrease in GC B cells at day 12 p.i. Therefore, we concluded that SCS MΦ play a secondary role (or cooperate with FDC) in sustaining GC that form in response to nonreplicating protein Ag, although in the context of a viral infection it is likely that SCS MΦ are more critical (48, 49). With respect to DC function, our BM chimera experiments also suggest that DC-intrinsic LTβR signaling does not play a significant role in the maintenance of the GC microenvironment. Indeed, we showed that the role of LTβR signaling in DC is more important in the context of cross-presentation of protein Ag to CD8+ T cells (45). In contrast, only the treatment of LTβR−/− BM chimeras with soluble LTβR–Ig resulted in a significant decrease in GC B cells late in the response. This evidence suggests that FDC play a primary role in sustaining GC.
Given that we observed a reduction in affinity maturation at days 20 and 48 p.i. in the AID-GFP × LTβ−/− BM chimeras, it is possible that this could be due to impaired survival of Ag-specific GC B cells; indeed, the number of Ag-specific GC B cells rapidly dwindles at day 12 (Fig. 4), concomitant with the collapse of GC structures at this time point (Fig. 5). However, if this were the case, the fittest GC B cells would need to be selectively lost in the later stages of the GC response to account for the reduced Ab affinity observed at days 20 and 48. Moreover, if GC B cell death were to account for the reduction in Ab affinity, we would note an overall reduction in anti-PE titers (irrespective of affinity), which is something that, for the most part, we do not observe (Fig. 6A, 6B). We favor the alternative interpretation that, in the presence of FDC, GC are sustained for a longer period of time. This extra time would allow for additional cycles of mutation and selection that are conducive to the emergence of higher-affinity B cell clones.
In summary, we conclude that FDC-M1+ CD35+ FDC are not required for the formation of GC (i.e., clusters of Ag-specific PNA+GL7+Fas+AID+ B cells) in LNs. This finding highlights differences in the structure and function of follicular stroma in the spleen versus peripheral LN. However, we suggest that FDC are required for the prolonged maintenance of GC, allowing for robust affinity maturation and the production of high-affinity Ab.
Note added in proof.
A recent study using Ccl19-cre conditional ablation of LTβR found that ERTR7+ networks are still formed in LNs in the absence of LTβR on FRC (50).
Acknowledgements
We thank Dr. Rafael Casellas for AID-GFP mice and the Gommerman and Martin laboratories for helpful discussions. We also acknowledge the gift of LTβR–Ig and MOPC21 from Biogen-Idec, as well as Dr. Jeff Browning for helpful feedback on the manuscript.
Footnotes
This work was supported by an operating grant from the Canadian Institutes for Health Research to J.L.G. and A.M. (MOP 89783). A.M. is supported by a Canada Research Chair award and B.B. is supported by a Canadian Institutes of Health Research Doctoral award.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- AID
activation-induced deaminase
- BM
bone marrow
- CSR
class-switch recombination
- DC
dendritic cell
- FDC
follicular dendritic cell
- FRC
fibroblastic reticular cell
- GC
germinal center
- iLN
inguinal lymph node
- LN
lymph node
- LT
lymphotoxin
- LTβR
lymphotoxin receptor
- NP
4-hydroxy-3-nitrophenylacetyl hapten conjugated to chicken gamma globulin
- p.i.
postimmunization
- PNA
peanut agglutinin
- R-PE
R-phycoerythrin
- SCS MΦ
subcapsular sinus macrophage
- SHM
somatic hypermutation
- WT
wild-type.
References
Disclosures
The authors have no financial conflicts of interest.