A Critical Role for IL-21 in Regulating Immunoglobulin Production

See allHide authors and affiliations

Science  22 Nov 2002:
Vol. 298, Issue 5598, pp. 1630-1634
DOI: 10.1126/science.1077002


The cytokine interleukin-21 (IL-21) is closely related to IL-2 and IL-15, and their receptors all share the common cytokine receptor γ chain, γc, which is mutated in humans with X-linked severe combined immunodeficiency disease (XSCID). We demonstrate that, although mice deficient in the receptor for IL-21 (IL-21R) have normal lymphoid development, after immunization, these animals have higher production of the immunoglobulin IgE, but lower IgG1, than wild-type animals. Mice lacking both IL-4 and IL-21R exhibited a significantly more pronounced phenotype, with dysgammaglobulinemia, characterized primarily by a severely impaired IgG response. Thus, IL-21 has a significant influence on the regulation of B cell function in vivo and cooperates with IL-4. This suggests that these γc-dependent cytokines may be those whose inactivation is primarily responsible for the B cell defect in humans with XSCID.

The receptor for the lymphoid-specific cytokine IL-21 is expressed on T, B, and NK cells (1, 2). IL-21 was initially reported to have a costimulatory T cell proliferative effect, to augment NK cell expansion and differentiation, and to augment B cell proliferation in response to CD40-specific antibodies, but to inhibit proliferation of B cells stimulated with the combination of IL-4 and IgM-specific antibodies (2). Subsequently, IL-21 was reported not to be required for NK cell development or expansion from murine splenocytes and to oppose certain actions of IL-15 on activated NK cells (3). The receptor for IL-21 contains IL-21R (1, 2) and also shares the common cytokine receptor γ chain (γc) with IL-2, IL-4, IL-7, IL-9, and IL-15 (4, 5). Mutations in γcresult in XSCID (4, 6), a disease characterized by an absence of T and NK cells and nonfunctional B cells (7). Although the absence of T and NK cells can be explained by defective responses to IL-7 (8–10) and IL-15 (11–13), no cytokine has been linked to the B cell defect. To determine whether defective signaling by IL-21 might contribute to this defect, we generated IL-21R–/– mice by using a targeting construct that eliminates the IL-21R extracellular and transmembrane domains (fig. S1). IL-21R–/– and wild-type mice are similar in the number and phenotype of thymocytes, splenocytes, peritoneal cells, and bone marrow cells, as detected using a panel of antibodies to multiple surface markers (3, 14). IL-21R–/– and wild-type splenocytes exhibited similar T cell proliferation in response to CD3-specific antibodies. Similarly, they showed nearly equal B cell proliferation in response to LPS, to CD40-specific antibodies, and to the combination of IL-4 and IgM-specific antibodies (14). Serum levels of IgG2a, IgG3, and IgM from naı̈ve mice were normal, but the amounts of IgG1 and IgG2b were lower and of IgE were higher in IL-21R–/– mice (14). After immunization with ovalbumin, the normal increase in IgG1 seen in wild-type mice was markedly impaired in IL-21R–/– mice; total serum IgG1 was 4 to 5% of wild type (Fig. 1A), and antigen-specific IgG1 antibody was about 0.1% of that in wild-type mice (Fig. 1B). There was little, if any, difference in total serum IgG2a, IgG2b, IgG3, IgM, or IgA (Fig. 1A, left), but ovalbumin-specific IgG2b and IgG3 were significantly lower in IL-21R–/– mice (Fig. 1B), and total serum IgE levels were variable although markedly higher in some IL-21R–/– mice (Fig. 1A, right). Immunization of IL-21R–/– mice with keyhole limpet hemocyanin (KLH) also revealed an impaired IgG1 response (Fig. 1C), and KLH-specific IgG1, IgG2b, and IgG3 levels were reduced to about 1/10th of the values for these subclasses in wild-type mice (Fig. 1D, top). It was noteworthy that IgE production in IL-21R–/– mice was markedly increased (50- to 55-fold) (Fig. 1C), as was KLH-specific IgE (Fig. 1D). Normal mice injected with IgD-specific antibodies respond by day 8 with marked elevations of serum IgG1 and IgE (15). In IL-21R–/– mice, IgG subclasses were similar to those in wild-type mice (IgG1 tended to be lower, but the difference was not statistically significant); however, IgE responses of IL-21R–/– mice were increased about 10-fold (fig. S2).

Figure 1

Total and antigen-specific immunoglobulin induced by immunization with ovalbumin (A and B) or KLH (C and D) in wild-type and IL-21R knockout (KO) mice. Shown are the total immunoglobulin for each subclass (A and C) and relative amount of antigen-specific immunoglobulin subclasses (B and D) induced by immunization. In (A) and (B), ovalbumin was injected intraperitoneally twice, and after an additional week, serum immunoglobulin levels were assessed by enzyme-linked immunosorbent assay (ELISA). For (C) and (D), KLH was injected once into foot pads, bottom of tail, and back, and at day 9, serum immunoglobulin levels were determined. In (A) and (C), shown are the means ± SEM of eight and six mice, respectively, in each group. (B) and (D) each show results from one of three experiments. (E) Diminished IgG subclasses and increased IgE after infection with Toxoplasma gondii. T. gondii (ME-49 strain) was intraperitoneally injected at a dose of 20 cysts per mouse, and serum immunoglobulin was determined by ELISA at day 100 (six survivors of seven mice in each group were analyzed).

Infection with Toxoplasma gondii induces a striking T helper cell type 1 (TH1) CD4+ T cell–dependent interferon-γ (IFN-γ) response that is required for survival during the acute phase of infection (16). Serum IFN-γ levels were similar in wild-type and IL-21R–/–mice 5 days after infection, as were IFN-γ levels after in vitro culture with Toxoplasma antigen (17). Correspondingly, six of seven mice in both wild-type and IL-21R–/– groups survived more than 100 days. Despite similar survival, at day 100 in the IL-21R–/–mice, serum IgG1 was about one-third that seen in wild-type mice (Fig. 1E, left), whereas IgE was increased about 20- to 25-fold (Fig. 1E, right). This increased IgE is noteworthy given that there is not normally an IgE response in mice to Toxoplasma gondii(18). IgG2a, IgG2b, and IgG3 were lower than in wild-type mice (Fig. 1E).

To further evaluate the impaired IgG1 and augmented IgE response, splenic CD4+ T cells were stimulated with plate-bound CD3-specific antibody under TH1- or TH2-polarizing conditions, but neither IFN-γ nor IL-4 production was diminished (fig. S3A), suggesting that these differentiation pathways are not impaired in these mixed background (129 × C57BL/6) IL-21R–/– mice. Moreover, when mice were immunized with KLH and spleen cells were restimulated in vitro, IFN-γ (fig. S3B) and IL-4 (fig. S3C) levels, as well as proliferation (fig. S3D), were relatively similar in IL-21R–/– and wild-type mice. Spleen cells from IL-21R–/– and normal mice generated comparatively similar levels of IFN-γ (fig. S3E) and IL-4 (fig. S3F) after 4 days of stimulation on plates coated with 2C11 monoclonal antibody to CD3ɛ.

To determine whether the impaired IgG1 response and augmented IgE production resulted from an obvious B cell defect, we cultured purified B cells with either the combination of CD40-specific antibodies plus IL-4 or LPS plus IL-4. As expected, production of IgG1 (Fig. 2A) and IgE (Fig. 2B) was relatively similar in IL-21R–/– and wild-type B cells, consistent with the lack of a requirement of IL-21R for stimulation of B cells by these stimuli. However, the combination of CD40-specific antibodies plus IL-4 and IL-21 boosted IgG1 production in purified wild-type B cells to a significantly higher level than in B cells from IL-21R–/– mice (Fig. 2C). Thus, IL-21 affects IgG1 production in normal murine B cells, and IL-21R–/– mice have an intrinsic B cell defect.

Figure 2

In vitro IgG1 and IgE production in IL-21R–/– mice. (A and B) B cells were stimulated with IL-4 plus either CD40-specific antibodies or LPS. Production (means ± SEM) of IgG1 at day 4 (A) and IgE at day 7 (B) were measured by ELISA. (C) B cells from wild-type and IL-21R–/– mice were stimulated with the combination of CD40-specific antibodies + IL-4 + IL-21. IgG1 was measured at day 4. In B cells from wild-type mice, the combination of CD40-specific antibodies + IL-4 + IL-21 also gave higher IgG1 production than did CD40-specific antibodies + IL-4 without the addition of IL-21 (14). (D) NP-KLH was mixed with Ribi adjuvant and injected intraperitoneally. Twelve days later, the number of NP-specific IgG1-, IgM-, and IgE-forming cells in splenocytes was determined by ELISPOT. Shown are means ± SEM from four IL-21R–/– and three wild-type mice. The levels of IgE-forming cells were below the limits of detection.

To evaluate the ability of IL-21R–/– mice to develop antibody-forming cells and memory B cells, we immunized mice with a T cell–dependent antigen, nitrophenyl (NP)-conjugated KLH. After exposure to this antigen, we evaluated the number of antibody-forming cells as hapten (NP)-specific antibody–secreting cells and memory B cells as NP-specific antibody–expressing cells that did not express IgD, CD4, CD8, or F4/80 and that did not stain with propidium iodide (19, 20). In wild-type and IL-21R–/– mice, there was at most only a modest difference in the number of splenic memory cells (fig. S4), but the number of hapten-specific IgG1-forming cells in the spleen was much lower in the IL-21R–/– mice (Fig. 2D). Thus, the decreased IgG1 could not be explained simply by a decrease in memory cells and, instead, likely reflects diminished antigen-specific IgG1-producing cells. Serum NP-specific IgG1 was also diminished (14).

Mice lacking γc exhibit a marked defect in B cell development (21, 22) as a result of defective IL-7 signaling (8, 9), whereas humans with XSCID have normal numbers of B cells (4), presumably because of IL-7–independent signals that mediate human B cell development (4, 10). Nevertheless, humans with XSCID exhibit markedly defective B cell function (23–26). IL-21R–/– murine B cells are functionally abnormal but not as defective as in human XSCID. We hypothesized that if IL-7 signaling were left intact, but if signaling by another γc-dependent cytokine in addition to IL-21 were also inactivated, it might be possible to generate a phenotype in mice more closely resembling that found in human XSCID. Given the importance of IL-4 for B cell function, we explored the effect that deficient IL-4 signaling might have in the context of IL-21R deficiency using IL-4–/–IL-21R–/–double-knockout (DKO) mice. As with IL-21R–/– mice, no gross developmental defects were observed compared with wild-type mice (fig. S5, A and B). However, naı̈ve IL-4–/–IL-21R–/– mice exhibited low serum levels of IgA and IgG subclasses, most strikingly in IgG1 (Fig. 3A). In contrast, serum IgM levels were relatively normal, which is analogous to observations made in humans with XSCID.

Figure 3

Defective immunoglobulin production in IL-4–/– IL-21R–/– mice correlates with disorganized germinal center formation. (A) Defective immunoglobulin levels in IL-4–/–IL-21R–/– naı̈ve mice. (B) Serum TNP-specific immunoglobulin levels were measured by ELISA 10 days after peritoneal injection of TNP-CGG in complete Freund's adjuvant (CFA) in wild-type, IL-21R–/–, IL-4–/–, and IL-4–/– IL-21R–/– (DKO) mice. Shown is a representative result of four independent experiments. IL-4–/– mice on the C57BL/6 background were from Jackson Laboratory. (C) Serum KLH-specific immunoglobulin levels were measured by ELISA 10 days after a peritoneal injection of KLH in CFA. Shown is one of three independent experiments; overall, one of six IL-4–/–IL-21R–/– mice did not show a decrease in IgG2a, whereas the other five mice were as shown in the Figure. (D) Defective germinal center formation in IL-4–/–IL-21R–/– mice. Mice were injected with 100 μg TNP-CGG with CFA. The wild-type (WT) lymph node contains a well-organized germinal center (outlined) with blast cells and tingible body macrophages containing apoptotic bodies (arrows). The IL-4–/–IL-21R–/– DKO lymph nodes had barely recognizable, poorly organized germinal center–like areas with scattered apoptotic cells (arrowheads). The wild-type and DKO lymph nodes were readily distinguished by a histopathologist in a blinded fashion on the basis of the abnormal germinal centers in the DKO mice.

To further investigate immunoglobulin production by these animals, we immunized them with trinitrophenyl-conjugated chicken γ-globulin (TNP-CGG), and the IL-4–/–IL-21R–/– mice showed severely impaired production of TNP-specific IgG1, IgG2a, IgG2b, and IgG3, and only about 10 to 20% of normal levels of IgM production (Fig. 3B). Strikingly, the strong up-regulation of IgE seen in the IL-21R–/–mice completely disappeared in IL-4–/–IL-21R–/– mice (Fig. 3B, lower right panel), which indicated that the increased IgE in IL-21R–/– mice was strictly dependent on IL-4. We further evaluated the ability of the IL-4–/–IL-21R–/– mice to respond to two other antigens. KLH induced a similar immunoglobulin profile to TNP-CGG (Fig. 3C), showing severely impaired production of KLH-specific IgG1, IgG2a, and IgG3, and only about 10 to 20% of normal levels of IgG2b and IgM production. The response to NP-KLH was also impaired, although the defects in IgG1, IgG2a, and IgG3 responses were less severe (14). Although germinal centers were relatively normal in mice lacking either IL-4 or IL-21R (14), in IL-4–/–IL-21R–/– mice germinal centers were disorganized, with decreased numbers of apoptotic bodies in macrophages (Fig. 3D), consistent with diminished somatic mutation and defective affinity maturation. These results indicate an important cooperative role for IL-4 and IL-21 for normal germinal center function.

Although IL-4 plays an essential role for IgE class switching, residual IgG1 production in IL-4–/– or Stat6–/– mice suggests the existence of additional regulator(s) of IgG1 production. We demonstrate that IL-21 is such a molecule. It is noteworthy that, in IL-21R–/– mice, IgE production is elevated but IgG1 is diminished. This potentially results from favored utilization of a γ1-independent mechanism of μ to ɛ immunoglobulin class switch, or alternatively, class switching might proceed so rapidly from μ via γ1 to ɛ that IgG1 levels are diminished. Other mechanisms for the increased IgG could also exist.

Humans or dogs with XSCID can mount minimal specific IgM responses but almost no IgG response (27). In some patients lacking evidence of donor B cell engraftment, B cell function can be normalized after bone marrow transplantation (28). However, it is conceivable that small numbers of donor B cells may in fact have engrafted in this situation, perhaps outside the peripheral blood (e.g., in lymph nodes) and can thus respond to γc-dependent cytokines, particularly since the detection of B cell chimerism can be difficult. Other data strongly support an intrinsic B cell defect in XSCID. First, in XSCID carrier females, who have normal T cell function, immature surface IgM+ B cells exhibit random X chromosome inactivation, but more mature, terminally differentiated IgM B cells preferentially exhibit nonrandom X inactivation (29), which suggests that only B cells expressing the wild-type X chromosome become antibody-producing cells. Second, other studies have concluded that effective humoral reconstitution can only be achieved by the engraftment of normal donor B cells (30), so that if donor B cells do not engraft, patients typically require long-term intravenous γ-globulin (31). In this regard, in data collated in 2002, of 43 SCID patients who received hematopoietic stem cell transplantation at Duke University Medical Center, the 20 who exhibited normal B cell function either had documented donor B cell engraftment (18 patients) or have not been adequately evaluated (2 patients); moreover, effective response to immunization with ΦX174 was only found when donor B cells engrafted (31). This is in contrast to the situation in humans with SCID resulting from IL-7R deficiency, in which intrinsic B cell function is readily restored by engrafted T cells (32), and no intrinsic B cell defect has been noted. As noted above, in humans, IL-7 signaling can be inactivated without loss of B cell development, whereas IL-7 is vital for B cell development in mice (4). By generating mice defective in IL-4 and IL-21 signaling but which retain IL-7 signaling and thus B cell development, we have established in mice a phenotype that appears to closely resemble that of B cells from patients with XSCID, suggesting that defective signaling by IL-4 and IL-21 might explain the B cell defect in XSCID.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

  • * Present address: Institute of Medical Science, University of Tokyo, Tokyo, Japan.

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


View Abstract

Navigate This Article