Dependence of Germinal Center B Cells on Expression of CD21/CD35 for Survival

See allHide authors and affiliations

Science  24 Apr 1998:
Vol. 280, Issue 5363, pp. 582-585
DOI: 10.1126/science.280.5363.582


Affinity-driven selection of B lymphocytes within germinal centers is critical for the development of high-affinity memory cells and host protection. To investigate the role of the CD21/CD35 coreceptor in B cell competition for follicular retention and survival within the germinal center, either Cr2+ or Cr2nulllysozyme-specific transgenic B cells were adoptively transferred into normal mice immunized with duck (DEL) or turkey (TEL) lysozyme, which bind with different affinities. In mice injected with high-affinity turkey lysozyme, Cr2null B cells responded by follicular retention; however, they could not survive within germinal centers. This suggests that CD21 provides a signal independent of antigen that is required for survival of B cells in the germinal center.

The murine Cr2 locus encodes complement receptors CD21 (CR2; 150 kD) and CD35 (CR1; 190 kD) that are expressed primarily on B cells and follicular dendritic cells (FDCs) (1). Cr2null mice have impaired immune responses to T-dependent antigens (2, 3); the defect is in the B cell compartment (2, 4). The natural ligands for CD21 and CD35 are activation products of complement C3 (C3d and C3b, respectively) (5) that are covalently coupled to antigen (6); mice deficient in C3 have impaired humoral responses similar to those observed in Cr2null animals (7). In studies in which fusion proteins of C3d and hen egg lysozyme (HEL) were used, B cell activation in vitro and in vivo was enhanced and dependent on the presence and density of C3d (8). Thus, CD21/CD35 may be a potent coreceptor for complement-decorated antigens that can raise the intensity of suboptimal activating signals.

To examine directly the fate of B cells deficient in CD21/CD35 in an immune recipient, we bred Cr2null mice with mice expressing a transgenic (tg) immunoglobulin (Ig) consisting of both heavy and light chains, which bind avian egg lysozymes with very high affinity (9). The importance of CD21/CD35 as a coreceptor in B cell activation by antigen ligands that bind with increasingly substantial affinities was examined by comparing the response of Cr2+and Cr2null lysozyme-specific Ig tg B cells to DEL and TEL, respectively (10, 11). As expected, 10- to 100-fold less TEL than DEL was required for the induction of proliferative responses and up-regulation of the CD86 costimulator molecule in both groups of tg B cells (12). Thus, activation through the B cell receptor (BCR) in the absence of complement is comparable in Cr2+ and Cr2null tg B cells, and B cell activation and proliferation is proportional to antigen affinity.

To examine the importance of B cell expression of CD21/CD35 in combination with low-affinity antigen (DEL) in vivo, we adoptively transferred splenic B cells isolated from either Cr2+ or Cr2null lysozyme-specific Ig tg mice with DEL into wild-type (WT) recipients that had been primed 7 days earlier with DEL (13, 14). Cr2null tg B cells from the spleen declined over the 5-day period in the DEL-primed recipients (Fig. 1A); by day 5 after transfer, there were 75% fewer Cr2null than Cr2+ tg B cells (0.26 ± 0.04% versus 1.14 ± 0.26%, respectively;P < 0.01) in the recipient spleens (15). An increase in Cr2null tg B cells in the blood was observed by day 5 (Fig. 1B); however, this did not account for the loss of cells from the spleen given the relatively low number of B cells in blood compared with spleen. Loss of Cr2null tg cells from the spleen was confirmed by immunohistochemical analysis (Fig.2, A to D, and Table1) and correlated with a low frequency of expression of CD86 (B7-2) at day 1 after transfer (12). Although similar numbers of Cr2+ and Cr2null tg B cells were present on day 1 in the primary splenic follicles of immune recipients (Fig. 2, A and C), by day 5 less than 2% of follicles held Cr2null tg B cells compared with 85% of follicles positive for Cr2+ tg B cells (Fig. 2, B and D, and Table 1) (16). The loss of the Cr2null tg B cells in DEL-immune recipients was greater than that in nonimmune controls in which over 50% of the follicles were positive for tg B cells 5 days after transfer (Table 1). Thus, binding of intermediate affinity antigen without coligation of CD21/CD19 coreceptor might deliver a subthreshold signal leading to B cell elimination. Cook et al. (17) have reported that subthreshold BCR stimulus of tg B cells in the splenic outer periarteriolar lymphoid sheath (PALS) can result in an abortive response in the follicle in the presence of T cell help. Whether specific elimination or migration out of the white pulp accounts for the loss of Cr2null tg B cells cannot be determined from these results.

Figure 1

Flow cytometric analysis of splenic (MNCs) and peripheral blood mononuclear cells (PBMCs) of recipients after adoptive transfer of Cr2+ or Cr2null lysozyme-specific Ig tg B cells. (A toD) Frequency of Cr2null or Cr2+HEL-specific tg B cells in the spleen or blood 1, 3, and 5 days after transfer into nonirradiated MHC class II–matched WT recipients that had been immunized 7 days previously with 50 μg of soluble DEL (A and B) or TEL (C and D) antigen. Values represent the mean ± SD of HEL-binding B cells within the total B cell population (B220+) of three to four experiments with pooled cells from two mice for each group in each experiment. Closed and open circles represent Cr2+ HEL+ tg B cells with and without antigen; closed and open triangles represent Cr2nullHEL+ tg B cells with and without antigen.

Figure 2

Immunohistochemical analysis of splenic follicles of recipients after adoptive transfer of either Cr2null or Cr2+ lysozyme-specific Ig tg B cells. Recipients were immunized either with DEL (A toD), TEL (E to H), or saline (day 5 only) (12). On day 7 mice received 1 × 107 splenocytes of Cr2+ (A and B, E and F) or Cr2null HEL-specific tg mice (C and D, G and H). Spleens were harvested on day 1 (A, C, E, and G) or day 5 (B, D, F, and H) after transfer. Splenic cryosections were analyzed by two-color immunohistochemistry demonstrating HEL-binding cells (blue) and PNA-positive GCs (crimson). (I and J) Injection of TEL-immunized WT mice with sCR2 (24).

Table 1

Morphometric analysis of splenic sections of recipients on day 5 after transfer of tg B cells. HEL-binding B cells and PNA-positive GC B cells were identified by immunostaining as described (16). Total numbers of follicles or follicles and GCs with more than 10 tg B cells were counted for six recipients each for Cr2null and Cr2+. Values represent the mean ± SEM.

View this table:

To rule out down-regulation of the Ig receptor, we labeled splenocytes with a fluorescent dye (BCECF AM-2′7′-bis-2-carboxyethyl-5 and 6-carboxy-fluorescein) before transfer (15). On day 5 after transfer, spleen cells were harvested and fluorescent cells analyzed for expression of the B cell marker CD45R (B220), and HEL binding. Similar numbers of BCECF+B220+ and HEL+B220+ tg cells were observed, confirming that the majority of the HEL+ tg cells were accounted for (12). Thus, most Cr2null tg B cells in vivo remained inactivated by DEL antigen despite its substantial affinity and activating potential in vitro. This finding supports the threshold model (18) that CD21/CD35 (with CD19 and CD81) is a potent coreceptor for antigen and its coligation with the BCR lowers the affinity threshold for B cell activation (19).

To determine if the binding of high-affinity antigens by Cr2null tg B cells could drive in vivo activation, follicular retention, and participation in the germinal center (GC) response, we transferred tg cells into TEL-primed recipients. Analysis of the frequency of CD86+ Cr2null tg B cells in TEL-immune versus nonimmune recipients (26% versus 17%, respectively;P < 0.05) indicated a significant increase in expression of this activation marker. By day 5 after transfer, an ∼30-fold difference was observed in the number of follicles in TEL-immune versus DEL-immune recipients that held Cr2nulltg B cells (60% versus 1.5%, respectively) (Fig. 2, H and D, respectively, and Table 1). This difference is not likely due to the differential availability of T cell help because (i) similar numbers of GCs were observed in spleens of both DEL- and TEL-primed recipients (Table 1); (ii) mice immunized with TEL or DEL showed similar antibody responses (12); and (iii) T cells isolated from both groups of immune recipients on day 3 after transfer expressed similar amounts of the CD25 activation marker and proliferated comparably in vitro when cultured with the relevant antigen (12).

Despite their activation and persistence within the splenic follicles, few GCs of TEL-primed recipients contained Cr2nullHEL-specific tg B cells. Two-color immunohistochemical analyses of splenic sections on day 5 after transfer demonstrated that less than 7% of GCs included Cr2null tg B cells as compared with greater than 53% positive for Cr2+ tg B cells (Fig. 2, H and F, respectively, and Table 1). Although interaction with a very high affinity antigen was sufficient for the activation and follicular retention of coreceptor-deficient tg B cells, it was insufficient for their entrance or survival (or both) within GCs. The relative absence of GCs that include Cr2null tg B cells is not explained by the migration of tg B cells out of the spleen because less than a twofold difference was observed in the number of follicles (85% versus 60%) occupied by Cr2+ and Cr2null tg cells, respectively (Table 1). To examine if the Cr2+ and Cr2null tg B cells were actively dividing in GCs, we pulsed both groups of TEL-primed recipients with bromodeoxyuridine (BrdU) on day 5 after transfer and harvested their spleens 2 hours later. Similar proportions of BrdU-labeled endogenous B cells and Cr2+ tg B cells were present in the GCs of mice immunized with TEL (Fig.3A) (20). However, few Cr2null tg B cells incorporated BrdU (Fig. 3B). Thus, high-affinity B cells that lack CD21 are activated by antigen and enter GCs but do not proliferate there.

Figure 3

Cr2+ lysozyme-specific Ig tg B cells localized within the GCs of TEL-primed recipients are actively dividing. BrdU uptake of dividing (A) Cr2+(n = 4) or (B) Cr2null(n = 5) HEL-specific tg B cells in splenic follicles 5 days after transfer into nonirradiated MHC class II–matched WT recipients that had been immunized 7 days previously with TEL antigen. B cells are stained with B220-HRP (crimson), tg cells are stained with HEL-biotin-strepavidin-AP (blue), and proliferating cells that incorporated BrdU were revealed by sequential incubation with BU20a, goat anti-mouse IgG-biotin, and strepavidin-AP (pink).

The reduction in the number of Cr2null tg B cells within the GCs of TEL-primed mice suggests that CD21/CD35 expression is required for survival in GCs. Given the extraordinary affinity of TEL binding to the BCR of Cr2null tg cells, it seems unlikely that coreceptor amplification of BCR signals alone could account for the failure of tg B cells to participate in the GC reaction. Instead, expression of CD21 by GC B cells may be critical for contact with C3d-antigen complexes retained by FDCs and may deliver a distinct survival signal. Whether this signal represents the synergistic effect of coligation of CD21/CD19 with the BCR as reported by Tooze et al. (21) is not known. Because the efficient retention of antigen by FDCs is mediated by complement (22), attachment of C3d-antigen complexes by way of CD21/CD35 receptors may provide not only a source of antigen for GC B cell activation and selection, but also a ligand (C3d) necessary for B cell differentiation along the memory pathway. Therefore, WT mice immunized with TEL were injected intravenously with a soluble form of CD21 ([CR2]2-IgG1) (23) at the peak period of GC formation (days 8, 9, and 10) and their spleens harvested 48 hours later. As a control, TEL-immune mice were injected with a similar concentration of IgG1. Soluble CD21 treatment resulted in an 87% reduction in total GC area within the splenic white pulp (Fig. 2, I and J) (24). [CR2]2-IgG1 has been well characterized and is specific for iC3b and C3d (23); unlike human CD21, mouse CD21 does not bind CD23. This reduction in GC area is most likely explained by the blockade of C3d on FDCs within the GC, which would disrupt contact between activated B cells and FDCs as well as coligation of the CD21/CD19 coreceptor. Although [CR2]2-IgG1 treatment would also inhibit activation of naı̈ve B cells at other sites such as the PALS, this would not explain the elimination of preexisting GCs at day 8. Coculture of FDCs with primed B and T cells promoted B cell survival, and the effect was dependent on expression of CD21 and CD21L (C3d) by B cells and FDCs, respectively (25). Thus, expression of the CD21 and CD19 coreceptor provides not only an enhancing signal for lowering the threshold of B cell activation, but also mediates an unanticipated survival signal for B cells within the GC.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: mcarroll{at}


Stay Connected to Science

Navigate This Article