Mature Follicular Dendritic Cell Networks Depend on Expression of Lymphotoxin β Receptor by Radioresistant Stromal Cells and of Lymphotoxin β and Tumor Necrosis Factor by B Cells

Mice.

LTβ- and LTβR-deficient mice were generated as described previously (21, 22). BCR-deficient mice (30) and TCR β/δ–deficient mice (31) were purchased from The Jackson Laboratory. Ly5.1⁺ C57BL/6 mice were supplied by Dr. H.R. Rodewald (Basel Institute for Immunology, Basel, Switzerland). TNF-deficient mice (19) were provided by Dr. G. Kollias (Hellenic Pasteur Institute, Athens, Greece). Mice were maintained and bred in a conventional mouse facility in isolated cages according to German guidelines for animal care. 8–12-wk-old mice were taken for experiments.

Bone Marrow Transfer.

Bone marrow (BM) cells were harvested by flushing femurs and tibias of donor mice with cold RPMI 1640 medium (Seromed) supplemented with 10% heat-inactivated FCS (Seromed), 2 mM l-glutamine (Seromed), 50 μM 2-ME (GIBCO BRL), 50 μg/ml streptomycin (Seromed), and 100 U/ml penicillin. Cells were washed and depleted of mature T cells using magnetic beads coated with anti-Thy1.2 mAb (Dynal) according to the manufacturer's protocol. After the depletion, cells were counted, washed, and resuspended in PBS. Recipient mice were irradiated with 9.5 Gy using a ¹³⁷Cs irradiator (Buchler) and injected intravenously with 2 × 10⁶ BM cells in 0.2 ml PBS. For mixed BM transfer, 1.5 × 10⁶ cells of each genotype were injected in a total volume of 0.2 ml PBS.

Immunization.

6–8 wk after the BM transfer, the mice received an intraperitoneal injection of 5 μg of alum-precipitated (4-hydroxy-3-nitrophenyl-acetyl)-chicken gamma globulin (molar ratio 19:1; NP₁₉-CG) in 0.2 ml PBS.

Evaluation of Chimerism.

10 d after immunization, mixed BM chimeric mice were killed, and BM, spleens, lymph nodes, and sera were collected. One half of each spleen was frozen for immunohistology, and the other half and all lymph nodes were used to purify B lymphocytes. Single cell suspensions were prepared using nylon cell strainers (Becton Dickinson). Cells were washed and incubated in Tris-buffered ammonium chloride solution to lyse erythrocytes. Cells were then washed in RPMI 1640 medium and in PBS containing 0.5% BSA. At this stage, chimerism of some mice was determined by flow cytometry on a FACScan^®, using mAbs directed against the Ly5.1 (CD45.1) and Ly5.2 (CD45.2) isoforms conjugated to FITC or PE (PharMingen). B cells were purified by magnetic cell sorting using mouse CD45R (B220) microbeads, MACS^® VS⁺ separation columns and a MACS^® magnet (Miltenyi Biotec) according to the instructions provided by the manufacturer. The selected fractions were additionally depleted of remaining T cells by magnetic cell sorting using anti-Thy1.2–coated Dynabeads (Dynal). BM cells were depleted of T cells and of IgM⁺/IgG⁺ cells using a cocktail of Dynabeads coated with anti-Thy1.2 or anti-IgM/IgG antibodies (Dynal). The purity of cell populations was confirmed by flow cytometry. The percentage of LTβ^−/− cells in the purified populations was determined by Southern blot hybridization of BamHI-digested genomic DNA using an SphI-PstI fragment of the murine LTβ promoter (nucleotides 2977–3411, sequence data available from EMBL/GenBank/DDBJ under accession no. U06950) as a probe. Similarly, the percentage of TNF^−/− cells was determined by Southern blot hybridization of EcoRI- digested genomic DNA using a PCR-generated fragment of exon 4 of the TNF gene (PCR primers 5′-AGGTCACTGTCCCAGCATCT and 5′-GTCAGCCGATTTGCTATCTCA) as a probe. Quantifications of chimerism were performed by densitometry using a PhosphorImager (Molecular Dynamics).

Immunohistochemistry.

Tissue samples were embedded in tissue-freezing medium (Leica) and snap-frozen in 2-methylbutane (Merck) prechilled by liquid nitrogen. Cryostat sections (7 μm) were fixed for 8 min in acetone (Merck), dried, and preincubated for 30 min with PBS containing 5% (vol/vol) goat serum, 1% (wt/vol) BSA, and 0.15% (vol/vol) hydrogen peroxide (Sigma). Blocking of endogenous biotin was performed using an avidin-biotin blocking kit (Vector) according to the manufacturer's protocol. For double labeling, sections were incubated for 30 min with (a) biotinylated PNA diluted 1:500 (Vector) and rat anti– mouse CR1 (CD35) diluted 1:100 (clone 8C12; PharMingen); (b) biotinylated PNA diluted 1:500 (Vector) and rat anti–mouse B220 diluted 1:100 (clone RA3-6B2; PharMingen); and (c) biotinylated mouse anti–mouse IgD diluted 1:50 (clone 1.3-5), rat anti–mouse CD4 diluted 1:100 (GK1.5; PharMingen), and rat anti–mouse CD8 used as a 1:2 diluted hybridoma supernatant (clone 53.6-72; American Type Culture Collection). Rat IgG2a and IgG2b (PharMingen) were used as isotype controls. Single labeling was performed with FDC-M1 mAb (clone 4C11). After washing, alkaline phosphatase (AP)-conjugated streptavidin (Sigma) and/or horseradish peroxidase–coupled mouse anti–rat IgG (Dianova) were added. After 30 min incubation and washing, color development for bound AP and horseradish peroxidase was consecutively performed with an AP reaction kit (Vector) according to the manufacturer's instructions and with 3-aminoethyl-carbazole (Sigma) as described (32). In addition, fluorescent microscopy was used for analysis of sections labeled with PNA and FDC-M2 mAb (clone 209) as described previously (25).

Measurement of Antigen-specific IgG.

10 d after immunization, NP-specific IgG antibodies were detected using sandwich ELISAs with NP₁₂-BSA– or NP₅-BSA–conjugated ELISA plates (10 μg/ml in carbonate buffer [pH 9.5]). Murine NP-specific IgG antibodies were detected with an AP-conjugated goat anti– mouse IgG antiserum (Dianova). The substrate used was p-nitrophenyl phosphate (Sigma). For calculation of arbitrary binding units of NP-specific IgG antibodies, the standard NP-reactive mAb, N1G9 (33), was included on each ELISA plate.

GC Formation and Intact Splenic Architecture Require Expression of LTβR on Radioresistant Cells.

To address which cell types have to express LTβ or LTβR to initiate and maintain GC reactions, a reciprocal BM transplantation approach was used. BM from C57BL/6 wild-type donors was transferred into myeloablatively irradiated LTβR^−/− recipients (B6 → LTβR^−/−) and vice versa (LTβR^−/− → B6). In the same manner, LTβR^−/− → LTβ^−/−, B6 → LTβ^−/−, and LTβ^−/− → B6 chimeric mice were generated. As a control, BM from C57BL/6 wild-type donors was taken to repopulate irradiated C57BL/6 wild-type recipients (B6 → B6). After 8 wk, mice were immunized intraperitoneally with 5 μg NP₁₉-CG adsorbed to alum, and 10 d later spleens and sera were taken for analysis. For detection of GCs, a double labeling of spleen sections was performed using the plant lectin PNA and the mAb 8C12. PNA binds to centroblasts/centrocytes, whereas the mAb 8C12 is directed against the murine complement receptor CR1, which is highly expressed on FDCs and at lower levels by B lymphocytes (34). As shown in Fig. 1, E–H, in the spleens of B6 → LTβR^−/− animals only a few, small PNA⁺ cell aggregates were observed, and FDC networks were completely absent. A similar phenotype was found in immunized LTβR^−/− mice (21). An aliquot of the wild-type BM used to reconstitute the LTβR^−/− recipients was also transferred into LTβ^−/− recipients. In these mice, the wild-type BM-derived cells were capable of restoring GC formation, proving the intrinsic competence of the donor BM cells (data not shown). Morphologically intact GCs, i.e., PNA⁺ cell clusters in association with FDC networks, were also detectable in spleens from LTβR^−/− → LTβ^−/− (Fig. 1, I and J), LTβR^−/− → B6 (data not shown), and B6 → B6 (Fig. 1, A and B) chimeras. In contrast, wild-type mice reconstituted with LTβ^−/− BM had fewer and smaller PNA⁺ cell clusters with almost undetectable FDC networks (Fig. 1, M and N). Presence or absence of FDC networks in association with PNA⁺ cell clusters was confirmed by double labeling with PNA and FDC-M2 mAbs or by single labeling with FDC-M1 mAbs (data not shown).

The anatomical localization of GCs was determined by labeling splenic sections with either PNA and anti-B220 mAbs or anti-IgD, anti-CD4, and anti-CD8 mAbs. Downregulation of IgD on most GC B cells and relatively low numbers of T cells scattered in GCs compared with compact T cell areas in periarteriolar lymphoid sheaths served as criteria for the identification of GCs. In LTβR^−/− → LTβ^−/− (Fig. 1, K and L), LTβR^−/− → B6 (data not shown), and B6 → B6 (Fig. 1, C and D) chimeric mice, all GCs were correctly localized in the B cell areas. Conversely, in spleens of LTβR^−/− mice reconstituted with wild-type BM, PNA⁺ cell aggregates were found around central arterioles (Fig. 1, G and H). T and B cells segregated normally in LTβR^−/− → LTβ^−/− (Fig. 1 L), LTβR^−/− → B6 (data not shown), and B6 → B6 (Fig. 1 D) chimeric mice, forming distinct periarteriolar lymphoid sheaths and B cell follicles. In contrast, in B6 → LTβR^−/− mice, T and B cells were mixed despite the presence of hematopoietically derived LTβR^+/+ donor cells (Fig. 1 H). Taken together, the failure of LTβR^+/+ BM-derived cells to restore GCs and an intact splenic architecture in LTβR^−/− recipients provides evidence that LTβR on radioresistant stromal cells is required for these functions. However, LTβR^−/− BM-derived cells were capable of establishing GCs in LTβ^−/− recipients. This shows that for GC development in adult mice the presence of LTβR on radiosensitive BM-derived cells is dispensable, whereas the presence of LTβ on these cells is necessary. The latter conclusion is further supported by the finding that transfer of LTβ^−/− BM into wild-type C57BL/6 recipients severely impaired GC formation.

LTβ and TNF from B Cells Are Required for Formation of Mature FDC Networks.

Expression of LTβ and TNF by hematopoietic cell lineages is required for GC formation (results above, and reference 35). Yet it is unclear whether a single hematopoietic lineage is necessary and, perhaps, sufficient for the production of LTβ and/or TNF, or whether different cellular sources can redundantly provide these ligands in GC reactions. In particular for TNF, the latter appears to be possible, since a great variety of cell types (e.g., macrophages, granulocytes, T cells, B cells, dendritic cells) can produce both soluble and membrane-bound TNF. Similarly, LTβ can be synthesized by three distinct cell types, namely T, B, and NK cells (15, 16, 36). Thus, to address the question of which cell type is required to express LTβ and TNF for GC establishment, compound BM chimeric mice were made. BM cells from BCR-deficient donors were mixed in a 1:1 ratio with BM cells from TNF^−/− or LTβ^−/− donors and transferred into myeloablatively irradiated BCR^−/− recipients (LTβ^−/− + BCR^−/− → BCR^−/−; TNF^−/− + BCR^−/− → BCR^−/−). Control groups were established by reconstituting myeloablatively irradiated BCR^−/− recipients with mixed BM from Ly5.1⁺ C57BL/6 wild-type donors and LTβ^−/− or TNF^−/− donors (LTβ^−/− + B6 → BCR^−/−; TNF^−/− + B6 → BCR^−/−). Since BCR^−/− BM cannot give rise to mature B cells, peripheral B cells in the LTβ^−/− + BCR^−/− → BCR^−/− and TNF^−/− + BCR^−/− → BCR^−/− chimeras were genetically deficient in LTβ and TNF, respectively. All other radiosensitive BM-derived cell populations consisted of a mixture of wild-type and gene-targeted cells. In control groups (LTβ^−/− + B6 → BCR^−/−; TNF^−/− + B6 → BCR^−/−), all BM-derived cell populations—including B cells—were composed of wild-type and genetically deficient cells. After 6–7 wk, mice were immunized intraperitoneally with 5 μg of alum-precipitated NP₁₉-CG, and 10 d later BM, spleen, lymph nodes, and serum were taken for analysis. BM chimerism and B cell chimerism were determined by Southern blotting and/or flow cytometric analysis of the CD45 isoforms Ly5.1 and Ly5.2 (see Materials and Methods, and Table I). BM chimerism was found to be comparable between experimental and control groups (Table I). Immunohistochemical analysis of splenic sections from LTβ^−/− + BCR^−/− → BCR^−/− chimeras revealed the presence of few PNA⁺ cell clusters. Associated FDC networks were absent or significantly reduced in size (Fig. 2, E and F). Most of the PNA⁺ cell clusters were localized within B cell areas (Fig. 2, E–H). Albeit segregated from the T cell areas, the B cell areas did not represent well-defined B cell follicles. Spleens of LTβ^−/− + B6 → BCR^−/− control mice showed numerous PNA⁺ cell clusters in association with large FDC networks, most of them correctly localized within B cell follicles (Fig. 2, A–D). With regard to FDC network formation, the absence of TNF production by B cells in TNF^−/− + BCR^−/− → BCR^−/− chimeras resulted in a phenotype similar to the one found in LTβ^−/− + BCR^−/− → BCR^−/− chimeras: few PNA⁺ cell clusters contained considerably underdeveloped FDC networks, and most clusters lacked immunohistochemically detectable networks altogether (Fig. 2, M and N). However, in these mice, approximately two thirds of the PNA⁺ cell clusters were located around central arterioles in T cell areas (Fig. 2, O and P) and only one third was found in B cell areas (not shown). B cells segregated from T cells and some B cell follicles were observed (not shown). In TNF^−/− + B6 → BCR^−/− control chimeras, most PNA⁺ cell clusters associated with FDC networks were readily detectable in distinct B cell follicles (Fig. 2, I–L). All results regarding splenic FDC network formation in compound BM chimeras were confirmed by double labeling with PNA and FDC-M2 mAbs (data not shown). Taken together, the results demonstrate that expression of both LTβ and TNF by B cells is required for the development of mature splenic FDC networks, but not for the formation of PNA⁺ cell clusters.

Specific Primary IgG Responses.

To investigate whether the observed abnormal GC reactions correlated with impaired specific IgG responses, NP-specific IgG titers were quantified in the sera of animals 10 d after immunization. Densely or sparsely haptenated BSA (NP₁₂-BSA or NP₅-BSA) was used for coating of ELISA plates, allowing detection of both low and high affinity NP-specific IgG. B6 → B6 and LTβR^−/− → B6 chimeras responded to immunization with a comparable production of specific IgG, whereas B6 → LTβR^−/− animals did not mount a significant primary IgG response (Fig. 3, A and B). Moreover, significant defects were not observed in B6 → LTβ^−/−, LTβ^−/− → B6, or LTβR^−/− → LTβ^−/− chimeras (data not shown). In B6 → LTβR^−/− chimeras, the impaired primary IgG response correlated with multiple defects in the organization of peripheral lymphoid tissues such as lack of lymph nodes and Peyer's patches, disruption of T–B cell segregation, and aberrant PNA⁺ cell clusters without FDC networks in the spleen (macroscopic examination, and Fig. 1, E–H). In contrast, LTβ^−/− + BCR^−/− → BCR^−/− and TNF^−/− + BCR^−/− → BCR^−/− mice contained lymph nodes, PNA⁺ cell clusters, and distinct T and B cell areas, yet differed from their control groups regarding splenic FDC network formation (Fig. 2). Here, differences in NP-specific IgG titers between experimental and control groups were not statistically significant (Fig. 3, C and D), implying that a lack or strong reduction of splenic FDC networks does not necessarily lead to impaired specific primary IgG responses.