Disruption of the Bcl6 Gene Results in an Impaired Germinal Center Formation

C57BL/6 mice were purchased from Japan SLC Co. Ltd. (Hamamatsu, Japan). Bcl6^−/− (Yoshida, T., manuscript in preparation) and RAG1^−/− (33) were described in detail previously.

mAbs used for cell surface staining were FITC–anti-B220 (RA3-6B2; PharMingen, San Diego, CA), –anti-CD8 (Ly-2; PharMingen), biotinylated anti-IgM (II/41; PharMingen), PE–anti-CD4 (L3T4; PharMingen), and –anti-CD3ε (145-2C11; PharMingen). For flow cytometric analysis, cells were stained with antibodies for 30 min on ice. The cells were further incubated with a PE–streptavidin (PharMingen) for 15 min on ice. The stained cells were analyzed by FACSCalibur^® (Becton Dickinson, San Jose, CA).

RAG1^−/− (8–16 wk old) were sublethally irradiated (3.5 Gy) and injected with 2–5 × 10⁶ BM cells from Bcl6^−/− or Bcl6^+/+ littermates (3–5 wk old). The reconstitutions were confirmed 8–12 wk after transplantation by identifying CD3⁺ and B220⁺ mature lymphocytes in peripheral blood from Bcl6^+/+RM and Bcl6^−/−RM with flow cytometry.

DNP–OVA was prepared by coupling of OVA (Sigma Chemical Co., St. Louis, MO) with 2,4-dinitrophenylbenzensulfonic acid under alkaline condition (34). NP– chicken γ globulin (NP–CG) was made by reaction of succinimide ester of 4-hydroxy-3-nitrophenyl acetyl (NP; Genosys Biotechnology, Inc., Woodlands, TX) and CG (Sigma Chemical Co.). Mice were immunized intraperitoneally with 100 μg of alum-precipitated DNP–OVA or NP–CG. Sera were collected from the mice 7 d (for IgM) and 14 d (for IgG1) after immunization. Amounts of DNP-specific Abs in the sera were measured by ELISA as described previously (34).

A SmaI–XhoI fragment of murine Bcl6 cDNA (mBcl6SX) was inserted into pGEX4T-2 vector (Pharmacia Biotech, Piscataway, NJ). The resulting glutathione-S-transferase (GST)–mBcl6SX fusion protein was used to immunize rabbits to raise anti-Bcl6 antibodies. The immune sera were precleared by passage through GST–Sepharose columns and affinity purified with antigen columns on which GST–mBcl6SX protein was coupled to activated CH Sepharose (Pharmacia Biotech).

Spleens were isolated from C57BL/6, Bcl6^−/−RM, or Bcl6^+/+ RM after immunization, one third were used for flow cytometric analysis, and two thirds were embedded in O.C.T. compound (Miles, Incorporated, Elkhart, IN) and frozen in liquid nitrogen. Serial frozen sections (6 μm) were fixed in cold acetone and the activity of endogeneous peroxidase was quenched with 3% H₂O₂ in methanol. Sections were stained with anti-Bcl6 Ab, PNA coupled to horseradish peroxidase (HRP) (EY Laboratories, San Mateo, CA), biotinylated anti-B220 Ab (PharMingen), or alkaline phosphatase (AP)–labeled goat anti–mouse λ light chain Ab (Southern Biotechnology Associates, Inc., Birmingham, AL) as the first reagents. A biotinylated goat anti–rabbit antibody (Nichirei, Tokyo, Japan) and streptABComplex–HRP (Dako, Carpinteria, CA) were used as the second- and third-phase reagents, respectively. Bound HRP and AP activities were visualized with the diaminobenzidine detection kit (DAB kit; Nichirei) and the fast blue kit (Nichirei), respectively.

Splenic B cells were enriched by depleting non–B lineage cells from spleen cells. In brief, spleen cells were incubated with mixture of biotinylated mAbs to CD3, CD4, CD8, TER119 (PharMingen), and CD11b (Mac-1) (M1/70; PharMingen). These cells were subsequently reacted with streptavidin-coated immunomagnetic beads (BioMag; PerSeptive Diagnostics, Cambridge, MA). Labeled cells were removed by applying them in a magnetic field. The resulting B cell fraction contained >95% of B220⁺ cells and <0.5% of CD3⁺ cells. Splenic T cells were also enriched by a similar manner with mAbs to B220, TER119, and Mac-1. Resulting T cells contained <0.5% of B220⁺ cells. The mixture of separated B and T cells (10⁷ cells each) were injected intravenously into RAG1^−/−. The mice were immunized with DNP–OVA 1 d after transplantation. Spleens were isolated on day 14 after immunization.

Proliferating cells in spleen were histologically detected by 5-bromo-2′-deoxyuridine (BrdU) incorporation assay (31). BrdU (2 mg; Boehringer Mannheim, Indianapolis, IN) was injected intraperitoneally in Bcl6^−/−RM or Bcl6^+/+RM on day 3 after immunization with DNP–OVA. Spleens were isolated 2 h after BrdU injection and their sections were incubated with PNA–HRP for identifying GC B cells. After color development with DAB, the sections were incubated in 2N HCl at 37°C for 1 h, and then reacted with mouse mAb to BrdU (Boehringer Mannheim). The sections were further incubated with biotinylated rabbit anti–mouse Ig Ab (Dako), followed by staining with streptABComplex–AP (Dako). The fast blue kit (Nichirei) was used for chromogen. Therefore, blue color staining indicates both BrdU uptake in nuclei and Ig on B cells.

Splenic B cells (5 × 10⁴/well) from Bcl6^−/− RM were cultured with LPS (10 μg/ml; Sigma Chemical Co.), IL-4 (10² U/ml; culture supernatant of X63Ag8-653 cells transfected with murine IL-4 gene; reference 35), goat anti–IgM Ab (anti-μ; 10 μg/ml; Jackson ImmunoResearch Laboratories, West Grove, PA), or CD40 ligand (CD40L)–transfected Chinese hamster ovary cells (a gift from H. Yagita, Juntendo University, Tokyo, Japan; 2 × 10⁴ cells fixed with paraformaldehyde per well) in 96-well flat-bottomed microplates for 48 h. The cultures were pulsed with 1 μCi of [³H]thymidine (Amersham International, Aylesbury, UK) for 6 h, and [³H]thymidine uptake was measured in a scintillation counter.

We have made polyclonal antibodies specific for mouse Bcl6 and immunohistologically analyzed the expression in GCs from spleens of immunized mice. The GC reaction in spleen reaches its maximum by days 10–12 and keeps the same high levels until day 21 after immunization (31, 36). Therefore, expression of Bcl6 was examined in spleens of normal C57BL/6 mice immunized with T cell– dependent antigen, DNP–OVA, on day 14 after immunization (Fig. 1, A and B). Bcl6 was detected strongly in nuclei of most of GC B cells and weakly in those of other lymphocytes in white pulps.

After activation by helper T cells in PALS, antigen-reactive B cells migrate into follicles to form GCs or generate Ab-producing foci in PALS. The PALS-associated foci can be identified by immunohistological staining of spleens from C57BL/6 mice immunized with an immunogenic conjugate of NP with anti-λ light chain antibody (36). The peak of the focus formation occurs on about day 10 after immunization (36). Thus, Bcl6 expression in PALS-associated foci from the mice immunized with NP–CG was analyzed on day 7 after immunization. Although strong Bcl6 expression was detected in GCs, it was not detected in the PALS-associated foci (Fig. 1, C and D). Anti-λ light chain antibody used in this staining also distinguished centroblasts (sIg⁻) from centrocytes (sIg⁺) in GCs, in both of which Bcl6 expression was strongly upregulated.

Since Bcl6^−/− mostly died at young age, the function of Bcl6 in lymphocyte development was investigated in RAG1^−/− reconstituted with BM cells from Bcl6^−/− (Bcl6^−/−RM). Flow cytometric analysis of BM cells and thymocytes from Bcl6^−/−RM 3 mo after transplantation revealed no abnormality in early lymphocyte development (Fig. 2, A and B). Percentages of B220⁺ B cells, CD4⁺, or CD8⁺ T cells in spleens from the Bcl6^−/−RM were also comparable to those from Bcl6^+/+RM (Fig. 2, C and D). Although Bcl6 expression can be detected in murine BM cells, thymocytes, and splenocytes (9), these data indicate that Bcl6 is not essential for lymphogenesis in primary lymphoid tissues.

We then analyzed the function of mature lymphocytes in Bcl6^−/−RM by immunizing them with DNP–OVA. Titers of primary IgM and IgG1 anti-DNP Abs in sera from the mice on day 7 (IgM) and on day 14 (IgG1) after immunization were measured by ELISA. Bcl6^−/−RM produced the primary IgM Abs and IgG1 Abs at levels comparable to those of Bcl6^+/+RM (Fig. 3). These results indicate that Bcl6 is not essential for differentiation of mature B cells into IgG1-producing plasma cells.

GC formation was histologically analyzed in spleens from mice immunized with DNP–OVA on day 14 after immunization. PNA-binding GC B cells were clearly identified in all of the Bcl6^+/+RM analyzed (n = 8, Fig. 4, A and C). In contrast, none of the Bcl6^−/−RM spleens (n = 8) showed PNA-binding B cells in follicles (Fig. 4, B and D), although primary B cell follicles, marginal zones, and PALS were histologically identified (Fig. 4 F).

Because the great majority of Abs produced in the first 12 d after immunization are derived from PALS-associated foci (37), the focus formation in PALS was histologically analyzed in spleens from Bcl6^−/−RM immunized with NP–CG on day 7 after immunization with anti-λ light chain antibody. PALS-associated foci and PNA-binding GCs could be detected in the Bcl6^+/+RM (Fig. 5,A). Although PNA-binding GCs were absent in the Bcl6^−/−RM, PALS-associated foci could be identified (Fig. 5 B). Therefore, Bcl6 is essential for GC formation in follicle but not for focus formation in PALS.

Functions of T cells in GCs are essential for GC formation of B cells (38, 39) and BCL6 is detected in a small fraction of CD4⁺ T cells in GCs (7). To determine whether defects in GC formation in Bcl6^−/−RM are due to defects in B or T cells or both, splenic T and B cells were isolated from Bcl6^−/−RM and Bcl6^+/+RM and then co-transferred into RAG1^−/− 1 d before immunization with DNP–OVA. GC formation in the spleen was histologically analyzed 2 wk later. GC formation was detected in spleens from mice reconstituted with B cells from Bcl6^+/+RM and T cells from Bcl6^−/−RM (Fig. 6,A). In contrast, co-transfer of B cells from Bcl6^−/−RM and T cells from Bcl6^+/+RM did not restore GC formation (Fig. 6 B), indicating that Bcl6 expression in B but not in T cells is essential for the GC formation.

GC first appears in splenic follicles by day 4 and begins to wane by day 21 after immunization (31). It might be possible that GC formation occurs in Bcl6^−/−RM shortly after immunization but cannot be maintained. To examine the possibility, the presence of PNA-binding GCs was examined in spleens from Bcl6^−/−RM immunized with DNP– OVA at an early stage after immunization. Nascent PNA-binding GCs could be identified in the Bcl6^+/+RM spleen on day 3 after immunization (Fig. 7,A) and Bcl6 was detected strongly in nuclei of these GC B cells (data not shown). However, the nascent GCs were not detected at all in the Bcl6^−/−RM on day 3 (Fig. 7 B) and day 4 (data not shown) after immunization.

To exclude the possibility that nascent GC B cells from Bcl6^−/−RM cannot obtain the PNA-binding capacity, the presence of blast cells in S phase in splenic follicles from the Bcl6^−/−RM on day 3 after immunization was histologically analyzed using BrdU incorporation assay, because nascent GC B cells undergo massive clonal expansion in follicles (21, 22). About half the numbers of GC cells in spleen from the Bcl6^+/+RM were positive for BrdU uptake 2 h after pulse (Fig. 7,A). However, no BrdU-positive cells were detected in B cell follicles from the Bcl6^−/−RM spleen (Fig. 7 B), suggesting that antigen-reactive B cells activated by helper T cells in PALS cannot proliferate at all in splenic follicles.

We then examined the capacity for proliferation of mature B cells from Bcl6^−/−RM in vitro. Splenic B cells from Bcl6^−/−RM were stimulated with LPS, LPS plus IL-4, anti-IgM Ab, CD40L, or anti-IgM plus CD40L. DNA synthesis of these stimulated cells was analyzed on day 2 after stimulation. Splenic B cells from Bcl6^−/−RM proliferated well against these stimuli similar to those from Bcl6^+/+RM (Fig. 8). Therefore, antigen-reactive B cells in Bcl6^−/−RM may have a capacity to proliferate by antigen stimulations and signals from helper T cells in PALS.

Antigen-reactive B cells activated by helper T cells in PALS further differentiate into AFCs in PALS or GC B cells in follicles. We show here that Bcl6 is essential for differentiation of GC B cells (Fig. 4). However, Bcl6 is not required for the focus formation in PALS by antigen-reactive B cells (Fig. 5). This is also supported by the evidence that Bcl6 expression is not upregulated in PALS-associated foci (Fig. 1). Because primary AFCs and GC B cells are thought to be derived from a common clonal origin (40), Bcl6 may be an essential transcription factor for the differentiation pathway into GC B cells. Although it is still argued that these cells originate from distinct cell lineages (41, 42), Bcl6 is a regulator for the differentiation of GC B cells.

Bcl6 expression is upregulated in nascent GC B cells (data not shown) and detected in both centroblasts and centrocytes in GCs from normal mice (Fig. 1). Furthermore, GC formation is impaired in Bcl6^−/−RM (Fig. 4). Therefore, signals that upregulate Bcl6 expression in antigen-reactive B cells are essential for the differentiation of GC B cells. Stimulation of mature B cells with CD40L plus anti-IgM Ab induces phenotypic markers on GC B cells (43), and LPS plus IL-4 induces RAG1 and RAG2 expression (44) and Ig isotype class-switching in B cells like GC B cells (45). However, these signals cannot make upregulation of Bcl6 expression in mature B cells (data not shown; reference 46). Interactions of antigen-reactive B cells with helper T cells in PALS through cell surface molecules such as CD40–CD40L and CD86(B7-2)–CD28 are required for the differentiation (47–51). Because the blocking of these interactions inhibits both primary Abs production and GC formation, signals from these interactions may be required for both differentiation pathways. Follicular dendritic cells are unique cells in follicles (19). Helper T cells in GCs have also unique characters that are distinct from those of helper T cells in PALS (52, 53). These cells seem to be required for GC formation by antigen-reactive B cells. Interactions of the B cells with these dendritic cells or GC-helper T cells may be a possible origin of the signals that upregulate Bcl6. Further study on the signals will be able to elucidate the molecular mechanism of differentiation into GC B cells.

GC formation is also impaired in mice lacking genes other than Bcl6. Lymphotoxin-α– (54) or TNF-α–deficient mice (55) can produce primary IgG1 Abs without GC formation like Bcl6^−/−RM. The defect in GC formation may be due to the abnormality of microarchitecture of spleens (56). Chemokine receptor, Blr1-deficient mice also produce normal levels of primary IgG1 Abs without functional GCs (57). Although PNA-binding B cells are detected in PALS from the Blr1-deficient mice after immunization, these B cells fail to migrate from PALS into B cell follicles in the spleen. However, mature B cells from Bcl6^−/−RM can migrate into follicles to form the normal microarchitecture of spleens (Fig. 4,F). Mice lacking CD40 (47, 48), complement receptors (CR1 and CR2) (58, 59), or CD28 (51) display impairments of both GC formation and primary IgG1 production to T cell–depedent antigens, but Bcl6^−/−RM produced normal levels of the primary IgG1 Abs (Fig. 3). Therefore, the mechanism of the defect in GC formation in Bcl6^−/−RM is distinct from the mechanisms in these deficient mice.

There are two distinct types of B cells in GCs: centroblasts and centrocytes. Several unique molecular events such as somatic hypermutation of the Ig gene, Ig isotype class-switching, selective cell death, and differentiation into memory B cells occur separately in each type of GC B cells (60). Bcl6 expression is continuously upregulated in both types of GC B cells (Fig. 1), suggesting that Bcl6 is required not for each specific molecular event in GC B cells but for more fundamental events throughout GC reactions. These events may not be related with proliferation of GC B cells because Bcl6 is strongly detected in both cell cycling centroblasts and resting centrocytes (Fig. 1) and splenic B cells lacking Bcl6 can proliferate to various stimuli (Fig. 8).

BCL6 was identified as an oncogene involved in chromosomal rearrangements in B cell lymphoma. Bcl6 expression is ubiquitous in mature tissues and is upregulated in cardiac myocytes, keratinocytes, and sperm at their terminal stages. Furthermore, Bcl6-deficient mice die with spontaneous death of mature cardiac myocytes, suggesting that mature heart muscle without Bcl6 cannot maintain its differentiation stage. Although the GC stage is not a terminal stage of B cell differentiation, Bcl6 is continuously upregulated in GC B cells (Fig. 1). If the specific function of Bcl6 is shared between B cells at the GC stage and heart muscle at the terminal stage, Bcl6 may be essential for protection of the GC B cells against specific stresses such as somatic hypermutation and selective cell death. Furthermore, this protective function may also contribute to lymphomagenesis of GC B cells by deregulation of BCL6.

We thank Dr T. Takemori for discussion, Dr H. Yagita for providing CD40L-transfected CHO cells, Y. Iwata for skillful technical assistance, and E. Furusawa for secretarial services.

This work was supported in part by Grants-in-Aid for Cancer Research from the Ministry of Education, Science, Sports and Culture of Japan and a Research Grant of the Princess Takamatsu Cancer Research Fund (No. 96-22807).