Articles

Invariant natural killer T cells direct B cell responses to cognate lipid antigen in an IL-21-dependent manner
Irah L King1, Anne Fortier1, Michael Tighe1, John Dibble1, Gerald F M Watts2, Natacha Veerapen3, Ann M Haberman4, Gurdyal S Besra3, Markus Mohrs1, Michael B Brenner2 & Elizabeth A Leadbetter1
2011 Nature America, Inc. All rights reserved. Mouse invariant natural killer T cells (iNKT cells) provide cognate and noncognate help for lipid and protein-specific B cells, respectively. However, the long-term outcome for B cells after cognate help is provided by iNKT cells is unknown at present. Here we found that cognate iNKT cell help resulted in a B cell differentiation program characterized by extrafollicular plasmablasts, germinal-center formation, affinity maturation and a robust primary immunoglobulin G (IgG) antibody response that was uniquely dependent on iNKT cellderived interleukin 21 (IL-21). However, cognate help from iNKT cells did not generate an enhanced humoral memory response. Thus, cognate iNKT cell help for lipid-specific B cells induces a unique signature that is a hybrid of classic T celldependent and T cellindependent type 2 B cell responses. B cell responses fall into two general categories, T cell dependent and T cell independent. T celldependent responses require the engagement of antigen through the B cell antigen receptor (BCR) and cognate help from CD4+ T cells via major histocompatibility complex class IIrestricted antigen presentation. B cell activation in this context results in either the extrafollicular proliferation of B cells as plasmablasts or the entry of B cells into germinal centers (GCs) for the subsequent development of memory or plasma cells1. Extrafollicular plasmablasts cluster in the bridging channels and red pulp of the spleen, and although some class-switch recombination may occur, these cells do not undergo affinity maturation. In contrast, GC reactions that occur in the follicles involve class-switch recombination, somatic hypermutation and affinity maturation, which produces plasma and memory cells of higher affinity2. Both memory B cells and plasma cells are important for an enhanced memory response after subsequent reexposure to antigen. T cellindependent responses by a B cell do not require any direct interaction with a helper T cell and can be one of two subtypes: type 1 or type 2. Type 1 T cellindependent responses result from the stimulation of B cells by ligands that activate without engaging the BCR, such as the Toll-like receptor ligands lipopolysaccharide and CpG. Type 2 T cellindependent responses involve ligands that engage the BCR with multivalent epitopes such as polysaccharides or 4-hydroxy3-nitrophenylacetyl (NP) bound to Ficoll (NP-Ficoll). Both types of T cellindependent ligands stimulate an innate-like response that is more transient than the T celldependent response and does not lead to an enhanced recall response. T cellindependent responses generally stimulate extrafollicular foci rather than GCs, do not generate antibodies with enhanced affinity and produce few plasma cells and atypical memory cells1.
1Trudeau 3School

Well-characterized T celldependent B cell responses to protein antigen depend on conventional CD4+ T cells. However, invariant natural killer T cells (iNKT cells) also provide help for B cells3,4. Mouse iNKT cells express a restricted T cell antigen receptor (TCR) repertoire composed of the -chain variable region 14-chain joining region 18 (V14-J18) paired with V8.2, V7 or V2 TCR -chains5. The iNKT cell TCR recognizes CD1d, a 2-microglobulin-associated nonpolymorphic antigen-presenting molecule expressed mainly on professional antigen-presenting cells such as dendritic cells, monocytes and B cells but also on other cells such as T cells and hepatocytes6,7. The CD1 family of antigen-presenting molecules is unique in that its members have deep hydrophobic channels on their surfaces that are able to bind and present lipid molecules to T cells. Many bacterial CD1d ligands have been identified8, but the most-studied ligand is -galactosylceramide (-GalCer), a glycosphingolipid isolated from marine sponges that is now available in synthetic form. It is known that -GalCer binds CD1d with high affinity and rapidly activates nearly all iNKT cells to proliferate and simultaneously secrete large amounts of T helper type 1 and T helper type 2 cytokines. Like other innate-type cells, iNKT cells exist in a preactivated state with higher expression of the activation markers CD44, CD69 and CD25 on their surface and have a lower activation threshold than that of naive adaptive CD4+ T cells9,10. Thus, iNKT cells can regulate and activate myriad different cell types (macrophages, dendritic cells, B cells and T cells) early during infection and have an important role in defense against many bacterial, parasitic and autoimmune diseases8. A role for iNKT cells in the production of antibodies important for defense against infection is most commonly demonstrated through comparison of infection of intact versus CD1d- or iNKT celldeficient mice with live organisms. This approach has characterized a role for iNKT cells in the production

Institute, Saranac Lake, New York, USA. 2Division of Rheumatology, Brigham & Womens Hospital, Harvard Medical School, Boston, Massachusetts, USA. of Biosciences, University of Birmingham, Edgbaston, Birmingham, UK. 4Department of Laboratory Medicine, Yale University School of Medicine, New Haven, Connecticut, USA. Correspondence should be addressed to E.A.L. (eleadbetter@trudeauinstitute.org). Received 5 July; accepted 25 October; published online 27 November 2011; doi:10.1038/ni.2172

nature immunology aDVaNCE ONLINE PUBLICaTION

Articles a b e
NP-GC 0.57 16.7

NP-GalCer

NP-KLH + -GalCer

f
NP-specific CD138 5 B cells (10 )

12.5 10.0 7.5 5.0 2.5 0

** **

**

**

NP-KLH + alum

PBS-BSA-DMSO NP-KLH + alum

0.16

g
IgG+ anti-NP spots (per 106 splenocytes)

P-

17.2

KL - H G + C N PKL al H + um

P -

5 4 3 2 1 0

* **

0.31
10
5

10.9

104

DMSO lgD

10

KL - H G + C N PKL al H + um

2011 Nature America, Inc. All rights reserved.

Figure 1 Stimulation of B cells with cognate antigen (lipid) or 0 102 103 104 105 NP-APC CD138 noncognate antigen (lipid plus protein) induces splenic extrafollicular foci. (ad) Confocal microscopy of spleens obtained from B1-8 mice 5 d after immunization with 5 g NP-GalCer (a), 100 g NP-KLH plus 5 g -GalCer (b), 100 g NP-KLH plus alum (c) or PBS-BSA-DMSO (d) and labeled with antibody to B220 (anti-B220; green), anti-CD138 (red) and NP-APC (blue) to identify CD138 + NP-specific plasmablasts. Scale bars, 100 m. (e) Flow cytometry of splenic plasmablast B cells from wild-type C57BL/6 mice immunized as in ad. Numbers below outlined (gated) areas indicate percent NP-specific IgD cells (left) or B220loCD138+ cells (right). (f) Summary of results in e. (g) ELISPOT analysis of NP-specific IgG-secreting splenic B cells from mice immunized as in ad. (h) Flow cytometry of TCR+ iNKT cells binding the CD1d tetramer (CD1d-tet+), from the mice in f. Each symbol represents an individual mouse; small horizontal lines indicate the mean (fh). *P 0.05 and **P 0.001 (unpaired two-tailed t-test (f,h) or Mann-Whitney test (g)). Data are representative of two independent experiments with three to four mice per group (ad) or are representative of (e,g) or pooled from (f,h) two independent experiments with five mice per group.

P -

0.02

2.3

P-

15 10 5 0

** *

**

TCR+ CD1d-tet+ iNKT cells (106)

**

KL - H G + C N PKL al H + um

P G

of antipathogen responses during infection with Borellia hermsii11,12, Streptoccocus pneumoniae13 or Plasmodium falciparum14 and has indicated marginal zone B cells are a likely partner for iNKT cells in the spleen3,12,15,16. Activated iNKT cells are appreciated as having a role as both cognate and noncognate helpers of lipid and peptide-specific B cells. Noncognate studies have characterized an adjuvant-like effect of administering -GalCer together with haptenated proteins or influenza virus peptides17,18. The provision of noncognate help by iNKT cells to protein-reactive B cells has been shown to lead to humoral memory responses, plasma-cell development, affinity maturation and long-term maintenance of antibody responses17,18. Although cognate iNKT cell help has been demonstrated for B cells3, the outcome for B cells after cognate help is unknown at present. Here we found that cognate iNKT cell help for lipid antigenspecific B cells induced a robust primary immunoglobulin G (IgG) antibody response characterized by early extrafollicular plasmablast formation, GCs, antibody affinity maturation and a dependence on iNKT cellderived interleukin 21 (IL-21). However, cognate iNKT cell help failed to drive classical T celldependent aspects of humoral responses, including the humoral memory response and population expansion of antigenspecific antibody-forming cells. We propose that the provision of cognate iNKT cell help to B cells induces a constellation of traits that is representative of a previously unknown class of B cell response: the type 2 T celldependent response. RESULTS Induction of extrafollicular foci and GCs To determine whether the help provided by iNKT cells for lipid and protein antigens induces similar B cell differentiation patterns, we first assessed the extrafollicular plasmablast response at 5 d after


immunization of mice with antigens. For this we used B1-8 mice, in which ~5% of B cells express a transgene encoding a BCR specific for NP. We visualized antigen-specific extrafollicular foci in the red pulp and bridging channels of the spleen by confocal microscopy. We identified these splenic architectural structures as clusters of cells that bound NP tagged with the fluorescent label allophycocyanin (NP-APC) and expressed the plasmablast marker CD138. Mice immunized with the haptenated lipid antigen NP-GalCer (Supplementary Fig. 1) or with haptenated protein antigen mixed with lipid (NP linked to keyhole limpet hemocyanin (NP-KLH) plus -GalCer) developed numerous CD138+NP-APC+ cells clustered in small groups in extrafollicular T cell areas of the spleen (Fig. 1a,b). The mice developed only a few NP-specific CD138+ foci in their red pulp or bridging channels after immunization with NPKLH with aluminum hydroxide (alum) as the adjuvant (Fig. 1c), whereas no NP-APC+ foci developed when we administered the vehicle PBS-BSA-DMSO alone (0.1% BSA in PBS containing 0.25% dimethyl sulfoxide; Fig. 1d). Flow cytometry analysis of spleens from C57BL/6 wild-type mice showed that the NP-specific IgD B220 loCD138 + plasmablast B cell population had notably expanded in all groups, but this population was much larger in the group immunized with NP-GalCer (Fig. 1e,f). In addition, enzyme-linked immunospot analysis showed that spleens from mice immunized with NP-GalCer, but not those from mice immunized with NP-KLH plus -GalCer, contained significantly more B cells that produced NP-specific IgG than did those from mice immunized with vehicle (Fig. 1g) despite having similarly greater iNKT cell numbers than mice immunized with vehicle (Fig. 1h). These results indicated that cognate iNKT cell help to B cells resulted in a robust early plasmablast population expansion typical of the splenic response to T cellindependent antigens such as NP-Ficoll.
aDVaNCE ONLINE PUBLICaTION nature immunology

P-

SO

M SO

102 0

B220

G C

M SO

G C

NP-KLH + -GC

Articles a
NP-GalCer

NP-KLH + -GalCer

NP-KLH + alum

PBS-BSA-DMSO

e
GCs per spleen section

f
B220 CD95 GL7 NP-specific germinal center B cells (104)

g
8 6 4 2 0
SO PKL - H G + C N PKL al H + um P G C M D
lgDCD38+ NP-specific B cells (104)

50 40 30 20 10 0

lgG+ anti-NP spots per 106 splenocytes

h
40 30 20 10 0
PKL - H G + C N PKL al H + um P G C SO M D

** **

**

**

**

**

100 80 60 40 20 0

* * **

**

PKL - H G + C N PKL al H + um

PKL - H G + N C PKL al H + um
NP68-Ficoll 7 61

SO

P G C

P G C

2011 Nature America, Inc. All rights reserved.

** Figure 2 Stimulation of B cells with cognate antigen (lipid) or noncognate antigen (lipid plus protein) leads * 3 to the development of splenic GCs. (ad) Confocal microscopy of spleens obtained from B1-8 mice 12 d after immunization with 5 g NP-GalCer (a), 100 g NP-KLH plus 5 g -GalCer (b), 100 g NP-KLH plus 2 alum (c) or PBS-BSA-DMSO (d), and labeled with GL7 (green), anti-CD3 (blue) and anti-B220 (red). Scale 1 bars, 500 m. (e) Total GL7+ GCs per spleen section of the mice as in ad. (f,g) Flow cytometry of B220+ CD95+GL7+ GC cells (f) and IgDCD38+ NP-specific B cells (g) from spleens of C57BL/6 wild-type mice at 0 day 12 after immunization as in ad. (h,i) ELISPOT analysis of NP-specific IgG-secreting splenic B cells (h) and flow cytometry of TCR+CD1d tetramerpositive iNKT cells (i) from wild-type C57BL/6 mice immunized as in ad. Each symbol represents an individual mouse; small horizontal lines indicate the mean (ei). *P 0.05 and **P 0.001 (Mann-Whitney test). Data are representative of two to three independent experiments with duplicate sections from four mice per group (ad) or are representative of (f,h,i) or pooled from (e,g) two to three independent experiments with four to five mice per group.
CD1d-tet TCR iNKT cells (106)
+ +

KL - H G + C N PKL al H + um

P -

Next we compared GC formation in the spleen after immunization with lipid alone or lipid plus protein. Immunofluorescence labeling showed that at 12 d into the response, B1-8 mice with a greater frequency of NP-specific B cells, immunized with either NP-GalCer or NP-KLH plus -GalCer, developed frequent cell clusters positive for GL7, an antibody clone specific for an as-yet-unidentified T cell and B cellactivation antigen that labels GC B cells (Fig. 2a,b), in a manner similar to that of mice immunized with NP-KLH plus alum (Fig. 2c). Mice immunized with vehicle had low background number of GCs of approximately 10 per spleen section (Fig. 2d,e). There was an average of 24 GCs per spleen section in mice immunized with NP-KLH plus -GalCer and 21 GCs per spleen section in mice immunized with NP-GalCer (Fig. 2e), a notable but not significant difference. We also counted, by flow cytometry, splenic B cell and iNKT cell populations 12 d after immunization (Fig. 2fi). All groups of immunized C57BL/6 wild-type mice had a greater number and frequency of NPspecific (B220+CD95+GL7+) GC B cells than did mice immunized with vehicle, but the GC B cells were significantly more numerous in mice immunized with NP-KLH and -GalCer than in mice immunized with NP-GalCer (Fig. 2f). The number of splenic iNKT cells positively identified by the CD1d tetramer was similarly higher, showing greater population expansion after immunization with lipid plus protein but not after immunization with lipid only (Fig. 2i). These results suggested that when iNKT cells recognizing the lipid component of NP-GalCer provided cognate help to NP-specific B cells, the B cells were induced to produce GCs, although they were smaller than the GCs derived after noncognate iNKT cell help. Consistent with those data, the number of NP-specific IgG-producing cells was significantly greater in both groups of protein-immunized mice than in lipid-immunized mice at this later time point (Fig. 2h). Of note, we also observed more NPspecific CD38+IgD memory-phenotype B cells in all groups of antigenimmunized mice at day 12 (Fig. 2g), but differences between the groups
nature immunology aDVaNCE ONLINE PUBLICaTION

were not significant. Thus, noncognate iNKT cell help seemed to induce and maintain conventional GCs containing protein-specific B cells. In contrast, cognate iNKT cell help recruited by a lipid-only immunization strategy induced smaller GCs and was unable to sustain antigen-specific B cell population expansion and antibody production. Induction of BCR affinity maturation GCs provide an environment for B cell maturation that enables selection for BCRs of higher affinity19. Given that both noncognate and
NP-KLH + alum NP -GalCer NP-KLH + -GalCer

2.0 1.5
NP4/NP25

1.0 0.5 0 7 61 7 61 7 Time (d) 61

Figure 3 Cognate and noncognate iNKT cell help induces antigen-specific antibody affinity maturation. ELISA of affinity maturation in serum from C57BL/6 wild-type mice immunized with 100 g NP-KLH plus alum, 0.5 g NP-GalCer, 100 g NP-KLH plus 0.5 g -GalCer, or 30 g NP68-Ficoll, collected 7 d after the primary challenge (day 7) and 7 d after the secondary boost (day 61) and assessed on plates coated with BSA conjugated to NP (four molecules (NP4-BSA) or twenty-five molecules (NP25-BSA)), presented as the ratio of binding to NP4 to binding to NP25 (NP4/NP25). Each symbol represents an individual mouse; small horizontal lines indicate the mean. *P 0.0001 (unpaired t-test). Data are representative of three independent experiments with eight to ten mice per group.

P-

SO

SO

Articles a 10
IgG anti-NIP (g/ml) WT (day 7) IL-21R-KO (day 7) IgG anti-NIP (g/ml) WT (day 14) IL-21R-KO (day 14) WT (day 21) IL-21R-KO (day 21)

104 10
3

104 IgG anti-NIP (g/ml) 10 10

3

b 10
IgM anti-NIP (g/ml)

WT (day 7) IL-21R-KO (day 7) IgM anti-NIP (g/ml)

104 103 10 10
2

103 102 10
1

**

**

IgM anti-NIP (g/ml)

103 102 101 100

**

WT (day 14) IL-21R-KO (day 14)

104 10 10
3

WT (day 21) IL-21R-KO (day 21)

102 101 100

101 100

101 100

N P N -GC PKL - H N GC + PKL H N alu + P m 68 -F ic ol l

N -GC PKL H N -GC + PKL H N alu + P m 68 -F ic ol l

P N -G P- C KL - H N GC + PKL N alu H + P m 68 -F ic ol l

Figure 4 IL-21R signaling is required for cognate iNKT cellmediated anti-NP responses. (a,b) Titers of NP-specific IgG (a) and IgM (b) in serum obtained from C57BL/6 wild-type (WT) mice and IL-21R-deficient (IL-21R-KO) mice on days 0, 7, 14 and 21 after immunization with 0.5 g NP-GalCer, 100 g NP-KLH plus 0.5 g -GalCer, 100 g NP-KLH plus alum, or 30 g NP68-Ficoll. *P 0.05 and **P 0.001, wild-type versus IL-21R-deficient (Mann-Whitney test). (c) Real-time RT-PCR analysis of IL-21 mRNA expression in iNKT cells sorted by flow cytometry (as TCR+CD19 CD1d tetramerpositive iNKT cells) from mice transgenic for expression of V14, at 1 d, 1 week (days 57) and 2 weeks (days 1113) after intraperitoneal immunization with NP-GalCer (0.5 g per mouse) or PBS-BSA-DMSO; results are presented relative to the expression of mRNA from the housekeeping gene GAPDH. *P 0.001 (Mann-Whitney test). Data are from two to five independent experiments with three to five mice per group in each (a,b) or are pooled from two to three experiments with 511 mice per group (c; mean and s.e.m.).

IL-21 mRNA (relative)

c 40
30 20 10

trl

7 5

Time (d)

2011 Nature America, Inc. All rights reserved.

cognate lipid antigens stimulated the formation of GCs, we next investigated antibody affinity maturation driven by immunization with each form of lipid antigen. We used a standard enzyme-linked immunosorbent assay (ELISA) to assess the binding of serum antibody to sparsely haptenated proteins versus highly haptenated proteins20. In comparing serum from mice collected 7 d after primary immunization (day 7) with serum from the same mice collected 7 d after a secondary boost (day 61), we found that both cognate and noncognate lipid antigens induced a significant increase in antibody affinity after a boost immunization (Fig. 3). As expected, the known T celldependent antigen NP-KLH in alum induced antibodies of higher affinity after a secondary boost, whereas the well-described T cellindependent antigen of Ficoll haptenated with 68 molecules of NP (NP68-Ficoll) failed to induce significant affinity maturation (Fig. 3). Thus, just as both forms of iNKT cell help (cognate and noncognate) stimulated GCs, they also both induced affinity maturation of BCRs. IL-21 is critical component of iNKT cell help for B cells Follicular helper T cells (TFH cells) have been reported to enter the B cell follicle specifically to provide cognate help for protein-specific B cells21,22. Notably, mature iNKT cells are reported to share many of the characteristics of traditional protein-specific TFH cells23; that is, they migrate in response to the chemokine CXCL13 (via the chemokine receptor CXCR5) but not CCL21 (via CCR7)24, express ICOS (data not shown) and secrete IL-21 (ref. 25). To determine if iNKT cells provide B cell help similarly to protein-specific TFH cells, we assessed the importance of signaling via IL-21 and its receptor, IL-21R, for cognate lipidspecific and noncognate lipidenhanced

antibody responses in this system. We immunized IL-21Rdeficient mice and wild-type mice with lipid antigen (NPGalCer), protein antigen (NP-KLH in alum) or protein plus lipid antigens (NP-KLH plus -GalCer). In all three cases, early NPspecific IgG antibodies were less abundant in IL-21R-deficient mice than in wild-type mice (Fig. 4a). As a negative control, intraperitoneal administration of the T cellindependent antigen NP-Ficoll induced IL-21-independent IgG, with no differences between IL-21R-deficient and wild-type mice. In all groups tested, there were no consistent differences between IL-21R-deficient and wild-type mice in anti-NP IgM titers (Fig. 4b). These data suggested that IL-21 was required for antibody class switching, not merely for antibody production. We confirmed IL-21 expression by iNKT cells by real-time RT-PCR at early time points (days 57) and later time points (days 1113) after the administration of 0.5 g NP-GalCer (Fig. 4c). Next we sought to determine whether iNKT cells were the critical source of IL-21. To address this, we generated mixedbone marrow chimeras in which we selectively deleted Il21 in iNKT cells. We created these chimeras by reconstituting irradiated J18-deficient hosts with a mixture of 25% IL-21-deficient bone marrow and 75% J18-deficient bone marrow. We created controls with IL-21-sufficient iNKT cells through the use of J18-deficient hosts reconstituted with a mixture of 25% wild-type bone marrow and 75% J18-deficient bone marrow. Immunizing these mice with NP-KLH in alum, NP-KLH plus GalCer or the lipid NP-GalCer showed that only cognate iNKT cell help depended entirely on iNKT cellderived IL-21. Specifically, chimeric mice with IL-21-deficient iNKT cells had less NP-specific IgG at all time points than did chimeras with IL-21-sufficient iNKT
Figure 5 IL-21 produced by iNKT cells is required for cognate lipid antigen help. (ac) NP-specific IgG titers in blood from mixed bone marrow chimeras with an iNKT cell compartment able (wild-type) or unable (IL-21-deficient) to produce IL-21, obtained on day 8 (a), day 14 (b) or day 21 (c) after immunization with 0.5 g NP-GalCer, 100 g NP-KLH plus 0.5 g -GalCer, or 100 g NP-KLH plus alum. *P 0.05, wild-type versus IL-21-deficient (Mann-Whitney test). Data are from two independent experiments with three to five mice per group in each experiment (mean and s.e.m.).

a
IgG anti-NIP (g/ml)

104 103 102 101 100

WT (day 8) IL-21-KO (day 8)

b
IgG anti-NIP (g/ml)

104 103 102 101 100

WT (day 14) IL-21-KO (day 14)

c
IgG anti-NIP (g/ml)

104 103 102 101 100

WT (day 21) IL-21-KO (day 21)

KL - H + G C N PKL al H + um

KL - H + G C N PKL al H + um

KL - H + G C N PKL al H + um

P-

P-

P-

aDVaNCE ONLINE PUBLICaTION

nature immunology

11

N GC PK LH N -GC + PKL H N alu + P m 68 -F ic ol l

N -GC PKL - H N GC + PKL H N alu + P m 68 -F ic ol l

N -GC PKL H N -GC + PKL N alu H + P m 68 -F ic ol l

100

100

Articles a
2.0 IgG anti-NIP (mg/ml) 1.5 1.0 0.5 0 PBS + alum; PBS + alum NP-KLH + alum; NP-KLH PBS; NP-KLH + alum

b
200 IgG anti-NIP (g/ml) 150 100 50 0

PBS + alum; PBS + alum NP-GalCer; NP-GalCer PBS; NP-GalCer 1 2

* **
170 180 190 Time (d) 200

data are consistent with published reports demonstrating that the humoral memory response to protein immunization is the same whether the adjuvant used is alum or the lipid -GalCer. However, we found that responses to a haptenated lipid antigen, despite eliciting a robust primary antibody response, failed to generate a memory B cell antibody response. DISCUSSION Our studies here have shown that cognate and noncognate iNKT cell help for B cells led to very different B cell outcomes. We demonstrated that after immunization with either cognate or noncognate lipid, mice generated strong primary anti-NP IgG responses characterized by early extrafollicular foci and, later, GCs dependent on signals via IL-21R. However, B cells that received noncognate iNKT cell help made a greater humoral memory response after rechallenge, whereas those B cells that received cognate iNKT cell help made a secondary response of the same magnitude as the primary response. Our results are consistent with other studies of noncognate iNKT cell help17,18 and support the proposal that iNKT cells are memory-like innate lymphocytes able to stimulate a rapid, robust response from the time of their initial activation. In the context of the cognate-help studies, iNKT cells may be functioning as a previously unknown TFH cell population that specializes in helping lipid-specific B cells to generate GC responses. In response to immunization with protein antigens, IL-21 from conventional TFH cells acts directly on GC B cells to support plasma-cell differentiation2628. Data from our bone marrowchimera studies demonstrated that iNKT cells provided cognate lipidspecific T cell help through the production of IL-21. Thus, iNKT cells are able to function in part as iNKTFH cells. It is known that iNKT cells express many of the same surface costimulatory molecules that TFH cells express (for example, CD40L and ICOS)29,30 but, as we have shown here, differ from TFH cells in their ability to generate a memory B cell population. Our imaging studies showed that both the cognate iNKT antigen NP-GalCer and the noncognate mixture of NP-KLH plus -GalCer stimulated similar antigen-specific extrafollicular foci and

170

180 190 Time (d)

200

2011 Nature America, Inc. All rights reserved.

Figure 6 Only noncognate iNKT cell help induces antibody memory response after day 177 rechallenge. Anti-hapten ELISA of NP-specific IgG in blood (obtained periodically up to day 166) from C57BL/6 wild-type mice immunized (1) intraperitoneally on day 0 with PBS plus alum, 2.2 g NP-KLH plus alum or PBS alone (a), or PBS plus alum, 0.5 g NP-GalCer or PBS alone (b), then given a secondary (2) intraperitoneal boost on day 177 (arrow) of PBS plus alum, 2.2 g NP-KLH, or NP-KLH plus alum (a), or an intraperitoneal boost of PBS plus alum or 0.5 g NP-GalCer, followed by additional sampling of blood on days 3, 7, and 14 after the boost (key: primary challenge; secondary boost); full time course, Supplementary Figure 3. *P 0.05 and **P 0.001, versus mice immunized with PBS and boosted with NP-KLH plus alum (Mann-Whitney test). Data are from one experiment with nine mice per group (mean s.e.m.).

cells only after immunization with the cognate iNKT cell antigen NP-GalCer (Fig. 5). Thus, noncognate iNKT cell help elicited by NP-KLH plus -GalCer required IL-21, but it did not need to come from iNKT cells. Cognate versus noncognate humoral memory responses Given the finding that cognate iNKT cell help resulted in GC formation and affinity-matured antibody responses, we next sought to determine whether cognate iNKT cell help could generate B cell memory responses similar to noncognate iNKT cell help. To address this, we immunized wild-type C57BL/6 mice with either NP-GalCer or NP-KLH plus -GalCer. Using classical CD4+ T cell help as a positive control, we immunized some mice with NP-KLH in alum. After that primary intraperitoneal immunization, we allowed the mice to rest for 177 d to let the initial hapten-specific antibody response wane, then boosted the mice with a secondary intraperitoneal challenge of lipid antigen or protein antigen in PBS. As expected, boosting of mice with NP-KLH in PBS after previous immunization with NP-KLH in alum resulted in a distinct memory response (Fig. 6a); that is, the anti-NP titers after the boost were much higher than the antibody titers that resulted from the initial primary challenge. Titers after the boost were also much higher than the antibody titers in age-matched mice that received their primary protein antigen in alum challenge on day 177 (Fig. 6a). Mice that received NP-GalCer in vehicle as a primary challenge and again as a secondary boost developed the same antibody titers after the boost as those of mice that received a primary challenge with NP-GalCer at day 177 (Fig. 6b). We obtained results similar to those above with a shorter delay between challenges (46 d rather than 177 d) and boosting via an intravenous route (Fig. 7), a protocol more commonly used to demonstrate anti-protein memory responses. The response to lipid antigen was most similar to the anti-NP response generated by the T cellindependent antigen NP-Ficoll (Fig. 7b). We obtained similar results after challenge and boost with higher doses of lipid antigen (5 g per mouse; Supplementary Fig. 2), which indicated that antigen availability was not a confounding factor in our studies. Together these
nature immunology aDVaNCE ONLINE PUBLICaTION

a
6 IgG anti-NIP (mg/ml) 4 2 0

PBS; PBS NP-KLH + alum; NP-KLH PBS; NP-KLH + alum NP-KLH + -GalCer; NP-KLH PBS; NP-KLH + -GalCer

b
IgG anti-NIP (mg/ml) 1.0 0.8 0.6 0.4 0.2 0

PBS; PBS NP-GalCer; NP-GalCer PBS; NP-GalCer NP68-Ficoll; NP68-Ficoll PBS; NP68-Ficoll

** ** ** *
50 60 Time (d) 70

**

* **

80

50

60 Time (d)

70

80

Figure 7 Only noncognate iNKT cell help induces an antibody memory response after rechallenge on day 46. Anti-hapten ELISA of NP-specific IgG in blood obtained (periodically up to day 45) from C57BL/6 wild-type mice immunized intraperitoneally on day 0 with PBS, 2.2 g NP-KLH in alum or 2.2 g NP-KLH plus 0.5 g -GalCer (a), or with PBS, 0.5 g NP-GalCer, or 30 g NP68-Ficoll (b), then given a secondary intravenous boost on day 46 with PBS or the same dose of NP-KLH or NP-KLH plus -GalCer, or intraperitoneal boost of NP-KLH plus alum (a), or intravenous boost of PBS, NP-GalCer or NP68-Ficoll (b), followed by additional sampling of blood on days 3, 7, 14 and 29 after the boost (key: primary challenge; secondary boost); full time course, Supplementary Figure 4. *P 0.05 and **P 0.001, versus the corresponding primary immunization group (Mann-Whitney test). Data are from one experiment with eight mice per group (mean s.e.m.).

Articles
numbers of GCs. However, the B cell outcome of cognate help from iNKT cells was notably different from the outcome after help from conventional CD4+ T cells that benefited from enhanced antigenpresenting function secondary to iNKT cell activation. In part, iNKT cell help for cognate haptenated lipid did not entirely mimic extrafollicular focidominated responses to T cellindependent antigens but instead reflected a response based more on T celldependent extrafollicular foci and GCs. The noncognate-immunization approach with NP-KLH plus -GalCer did stimulate more GC B cells than did strictly lipid immunization but resulted in similar numbers of total GCs. That finding was consistent with our observation that spleens from mice immunized with NP-KLH plus -GalCer had relatively larger GCs, which suggested that in this case, activation of iNKT cells functioned more as an adjuvant than stimulating a lipid-specific response in parallel. Given that mice immunized with NP-GalCer or with NP-KLH plus -GalCer induce affinity maturation (probably via GCs) and have similar numbers of memory-phenotype CD38+ antigen-specific B cells31, it is possible that such GCs are present but inadequate or prematurely involute. Many elements have been identified as being critical for sustaining a T celldependent B cell GC response, including Dock8 expression in B cells, the formation of immune synapses between B cells and T cells, and integrin signaling32. The quality and/or quantity of a BCR signal can be a critical aspect of GC persistence, particularly later in a GC response when the way in which antigen is presented, such as in the form of immune complexes, may be different33. Expression of the costimulatory molecule PD-1 on T cells has also been shown to be important for the survival of B cells in GCs and subsequent long-lived plasma cell output without affecting affinity maturation34. Heterogeneity in memory B cell development may also affect the humoral recall response 35. The contribution of these factors to iNKT cellB cell interactions and their relevance to iNKT cellinduced formation of GCs remain to be determined. Finally, iNKT cells have been proposed to localize outside of the splenic white pulp under homeostatic conditions36. Thus, how iNKT cellB cell interactions occur in situ is not clear. One possibility supported by a published study may be that iNKT cells preferentially interact with marginal zone B cells37. These B cells are situated along the marginal sinus at the primary point for the entry of blood-borne particulate antigens into the spleen. This positions them to specialize in responding to type 2 T cellindependent antigens38. Marginal zone B cells express abundant CD1d and secrete mainly IgM and IgG3 (ref. 39), the antibody isotypes produced in response to the pure synthetic lipid antigen NP-GalCer3. Thus, they may also be one of the main B cell subpopulations to receive both cognate and noncognate iNKT cell help16. In conclusion, we propose lipid-specific, type 2 T celldependent responses as a new subcategory of B cell antigen-specific responses that are a hybrid of the other three established categories of B cell antigens. According to our assessment, the characteristics of type 2 T celldependent responses include the formation of extrafollicular foci and GCs accompanied by affinity maturation and the absence of functional humoral memory responses. This category of antigenspecific responses remains to be characterized in the context of live infection but may have an important role early in infection when rapid iNKT cell help could provide a unique advantage. Future studies should investigate humoral immunity to live pathogens induced by cognate and noncognate lipid antigens. METHODS Methods and any associated references are available in the online version of the paper at http://www.nature.com/natureimmunology/.


Note: Supplementary information is available on the Nature Immunology website. AcKNoWLEDGMENTS We thank M. Nussenzweig (Rockefeller University) for B6.SJL B1-8hi mice; M. Exley (Dana Farber Cancer Institute) for C57BL/6 V14-transgenic mice and C57BL/6 J18-deficient mice; K. Rajewsky (Center for Blood Research) for B1-8f mice; M. Rincon (University of Vermont) for IL-21-deficient mice; and the US National Institutes of Health Tetramer Core for mouse CD1d-PBS57 tetramers and unloaded CD1d tetramers. Supported by the Trudeau Institute (E.A.L.), the US National Institutes of Health (AI028973-23 and AI06342806 to M.B.B. and T32 A1049823-10 to I.L.K.), J. Bardrick (G.S.B.), the Royal Society (G.S.B.), The Wellcome Trust (084923/B/08/Z to G.S.B.) and the Medical Research Council (G.S.B.). AUTHoR coNTRIBUTIoNS I.L.K. designed and did experiments, analyzed data and edited the manuscript; A.F., M.T., J.D. and G.F.M.W. designed and did experiments; A.M.H. did experiments, edited the manuscript and provided technical advice; N.V. and G.S.B. synthesized and provided lipid antigens; M.M. and M.B.B. provided conceptual advice and E.A.L. initiated and directed the research, did experiments, analyzed the data and wrote the manuscript. coMPETING FINANcIAL INTERESTS The authors declare no competing financial interests.
Published online at http://www.nature.com/natureimmunology/. reprints and permissions information is available online at http://www.nature.com/ reprints/index.html.
1. MacLennan, I.C. et al. Extrafollicular antibody responses. Immunol. Rev. 194, 818 (2003). 2. Jacob, J., Kelsoe, G., Rajewsky, K. & Weiss, U. Intraclonal generation of antibody mutants in germinal centres. Nature 354, 389392 (1991). 3. Leadbetter, E.A. et al. NK T cells provide lipid antigen-specific cognate help for B cells. Proc. Natl. Acad. Sci. USA 105, 83398344 (2008). 4. Barral, P. et al. B cell receptor-mediated uptake of CD1d-restricted antigen augments antibody responses by recruiting invariant NKT cell help in vivo. Proc. Natl. Acad. Sci. USA 105, 83458350 (2008). 5. Godfrey, D.I., MacDonald, H.R., Kronenberg, M., Smyth, M.J. & Van Kaer, L. NKT cells: whats in a name? Nat. Rev. Immunol. 4, 231237 (2004). 6. Roark, J.H. et al. CD1.1 expression by mouse antigen-presenting cells and marginal zone B cells. J. Immunol. 160, 31213127 (1998). 7. Mandal, M. et al. Tissue distribution, regulation and intracellular localization of murine CD1 molecules. Mol. Immunol. 35, 525536 (1998). 8. Cohen, N.R., Garg, S. & Brenner, M.B. Antigen presentation by CD1 lipids, T cells, and NKT cells in microbial immunity. Adv. Immunol. 102, 194 (2009). 9. Bendelac, A., Matzinger, P., Seder, R.A., Paul, W.E. & Schwartz, R.H. Activation events during thymic selection. J. Exp. Med. 175, 731742 (1992). 10. Uldrich, A.P. et al. NKT cell stimulation with glycolipid antigen in vivo: costimulationdependent expansion, Bim-dependent contraction, and hyporesponsiveness to further antigenic challenge. J. Immunol. 175, 30923101 (2005). 11. Kumar, H., Belperron, A., Barthold, S.W. & Bockenstedt, L.K. Cutting edge: CD1d deficiency impairs murine host defense against the spirochete, Borrelia burgdorferi. J. Immunol. 165, 47974801 (2000). 12. Belperron, A.A., Dailey, C.M. & Bockenstedt, L.K. Infection-induced marginal zone B cell production of Borrelia hermsii-specific antibody is impaired in the absence of CD1d. J. Immunol. 174, 56815686 (2005). 13. Kobrynski, L.J., Sousa, A.O., Nahmias, A.J. & Lee, F.K. Cutting edge: antibody production to pneumococcal polysaccharides requires CD1 molecules and CD8+ T cells. J. Immunol. 174, 17871790 (2005). 14. Schofield, L. et al. CD1d-restricted immunoglobulin G formation to GPI-anchored antigens mediated by NKT cells. Science 283, 225229 (1999). 15. Bialecki, E. et al. Role of marginal zone B lymphocytes in invariant NKT cell activation. J. Immunol. 182, 61056113 (2009). 16. Muppidi, J.R. et al. Cannabinoid receptor 2 positions and retains marginal zone B cells within the splenic marginal zone. J. Exp. Med. 208, 19411948 (2011). 17. Galli, G. et al. Invariant NKT cells sustain specific B cell responses and memory. Proc. Natl. Acad. Sci. USA 104, 39843989 (2007). 18. Devera, T.S., Shah, H.B., Lang, G.A. & Lang, M.L. Glycolipid-activated NKT cells support the induction of persistent plasma cell responses and antibody titers. Eur. J. Immunol. 38, 10011011 (2008). 19. MacLennan, I.C. Germinal centers. Annu. Rev. Immunol. 12, 117139 (1994). 20. Herzenberg, L.A., Black, S.J. & Tokuhisa, T. Memory B cells at successive stages of differentiation. Affinity maturation and the role of IgD receptors. J. Exp. Med. 151, 10711087 (1980). 21. Breitfeld, D. et al. Follicular B helper T cells express CXC chemokine receptor 5, localize to B cell follicles, and support immunoglobulin production. J. Exp. Med. 192, 15451552 (2000). 22. Schaerli, P. et al. CXC chemokine receptor 5 expression defines follicular homing T cells with B cell helper function. J. Exp. Med. 192, 15531562 (2000).

2011 Nature America, Inc. All rights reserved.

aDVaNCE ONLINE PUBLICaTION

nature immunology

Articles
23. Fazilleau, N., Mark, L., McHeyzer-Williams, L.J. & McHeyzer-Williams, M.G. Follicular helper T cells: lineage and location. Immunity 30, 324335 (2009). 24. Johnston, B., Kim, C.H., Soler, D., Emoto, M. & Butcher, E.C. Differential chemokine responses and homing patterns of murine TCR NKT cell subsets. J. Immunol. 171, 29602969 (2003). 25. Coquet, J.M. et al. IL-21 is produced by NKT cells and modulates NKT cell activation and cytokine production. J. Immunol. 178, 28272834 (2007). 26. King, I.L., Mohrs, K. & Mohrs, M. A nonredundant role for IL-21 receptor signaling in plasma cell differentiation and protective type 2 immunity against gastrointestinal helminth infection. J. Immunol. 185, 61386145 (2010). 27. Linterman, M.A. et al. IL-21 acts directly on B cells to regulate Bcl-6 expression and germinal center responses. J. Exp. Med. 207, 353363 (2010). 28. Zotos, D. et al. IL-21 regulates germinal center B cell differentiation and proliferation through a B cell-intrinsic mechanism. J. Exp. Med. 207, 365378 (2010). 29. Vinuesa, C.G., Tangye, S.G., Moser, B. & Mackay, C.R. Follicular B helper T cells in antibody responses and autoimmunity. Nat. Rev. Immunol. 5, 853865 (2005). 30. Hayakawa, Y. et al. Differential regulation of Th1 and Th2 functions of NKT cells by CD28 and CD40 costimulatory pathways. J. Immunol. 166, 60126018 (2001). 31. Ridderstad, A. & Tarlinton, D.M. Kinetics of establishing the memory B cell population as revealed by CD38 expression. J. Immunol. 160, 46884695 (1998). 32. Randall, K.L. et al. Dock8 mutations cripple B cell immunological synapses, germinal centers and long-lived antibody production. Nat. Immunol. 10, 12831291 (2009). 33. Vinuesa, C.G., Linterman, M.A., Goodnow, C.C. & Randall, K.L. T cells and follicular dendritic cells in germinal center B-cell formation and selection. Immunol. Rev. 237, 7289 (2010). 34. Good-Jacobson, K.L. et al. PD-1 regulates germinal center B cell survival and the formation and affinity of long-lived plasma cells. Nat. Immunol. 11, 535542 (2010). 35. Anderson, S.M., Tomayko, M.M., Ahuja, A., Haberman, A.M. & Shlomchik, M.J. New markers for murine memory B cells that define mutated and unmutated subsets. J. Exp. Med. 204, 21032114 (2007). 36. Stetson, D.B. et al. Constitutive cytokine mRNAs mark natural killer (NK) and NK T cells poised for rapid effector function. J. Exp. Med. 198, 10691076 (2003). 37. Muppidi, J.R. et al. Cannabinoid receptor 2 positions and retains marginal zone B cells within the splenic marginal zone. J. Exp. Med. 208, 19411948 (2011). 38. Cyster, J.G. B cells on the front line. Nat. Immunol. 1, 910 (2000). 39. Lopes-Carvalho, T., Foote, J. & Kearney, J.F. Marginal zone B cells in lymphocyte activation and regulation. Curr. Opin. Immunol. 17, 244250 (2005).

2011 Nature America, Inc. All rights reserved.

nature immunology aDVaNCE ONLINE PUBLICaTION

ONLINE METHODS

Mice. C57BL/6 wild-type mice, B6.SJL B1-8hi mice40 (heterozygous for targeted insertion of a mutated variable 186.2 immunoglobulin heavy chain bearing specificity for the hapten NP), C57BL/6 V14-transgenic mice with a greater frequency of iNKT cells 41 and C57BL/6 J18-deficient mice lacking J18+ iNKT cells42 were housed and bred at the Dana-Farber Cancer Institute and the Trudeau Institute according to standards of the animal care and use committees of each institution. C57BL/6 B1-8f mice43 (heterozygous for targeted insertion of variable region 186.2 of the immunoglobulin heavy chain specific for NP) were bred and housed at Yale University School of Medicine according to standards of the institutional animal care and use committee. C57BL/6 IL-21R-deficient mice were generated as described44, and IL-21-deficient mice came from the Mutant Mouse Regional Resource Center at the University of California, Davis, and were provided by M. Rincon. All live animal experimental protocols were approved by the Dana-Farber Cancer Institute Institutional Animal Care and Use Committee or the Trudeau Institute Animal Care and Use Committee. Flow cytometry. Single-cell suspensions were prepared from the spleen and stained with the following monoclonal antibodies for flow cytometry: Alexa Fluor 450IgD (11-26c.2a; BD), phycoerythrin-indotricarbocyanine anti-B220 (RA3-6B2; Biolegend), phycoerythrinanti-Fas (Jo2; BD), fluorescein isothiocyanateGL7 (GL7; BD), fluorescein isothiocyanateanti-CD38 (90; BD), phycoerythrinanti-CD138 (281-2; BD), biotin-IgG1 (A85-1; BD), streptavidin-allophycocyanin-indotricarbocyanine (BD) and Pacific Blueanti-TCR (H57-597; Biolegend). The iNKT cells were identified with tetramers of mouse CD1d-GalCer (PBS57; US National Institutes of Health) conjugated to allophycocyanin. Samples were acquired on a FACSCanto II (BD) and were analyzed with FlowJo software (TreeStar). Bone marrow chimeras. Recipient C57BL/6 J18-deficient mice were irradiated twice with 500 rads and were allowed to rest for a few hours or overnight. Recipient mice were reconstituted with 5 106 bone marrow cells. Donor bone marrow included either 75% J18-deficient bone marrow mixed with 25% wild-type bone marrow, or 75% J18-deficient bone marrow mixed with 25% IL-21-deficient bone marrow. These reconstitution mixtures resulted in mice in which only the iNKT cells were deficient in IL-21 production or all cells were normal. Reconstitution of iNKT cell, T cell and B cell lineages in the spleen and liver were confirmed by flow cytometry after 7 weeks. Antigens, immunization and serum collection. NP-KLH, NP-BSA, KLH, NP68-Ficoll (Biosearch Technologies), NP-GalCer and -GalCer (synthesized by N.V. and G.S.B. as described3) were administered intraperitoneally in a volume 200 l unless noted otherwise. Immunizations included 2.2 g or 100 g protein precipitated in alum or suspended in PBS plus 0.1% (wt/vol) BSA, or 0.55 g lipid antigen solubilized in 0.25% (vol/vol) DMSO, then suspended in PBS with 0.1% (wt/vol) BSA. NP-KLH in these studies contained 20 NP haptens per KLH protein, and NP-GalCer lipid contained a single NP hapten per -GalCer molecule. In serum, the protein is most probably monomeric, whereas the lipid is more probably micellular or bound to a lipid binding protein, which makes equal comparisons of molar quantities or hapten quantities challenging. The quantity 2.2 g NP-KLH is the molar equivalent of 0.5 g NP-GalCer divided by the number of haptens on the KLH protein (20). Serum was collected retro-orbitally or submandibularly and was stored at 20 C until assessment by ELISA. ELISA. NP-specific IgG and IgM in serum was assessed by heteroclitic ELISA specific for NIP (4-hydroxy-5-iodo-3-nitrophenyl) as described 3.

Affinity was assessed by ELISA on plates simultaneously coated with NP4-BSA or NP25-BSA (presented as a ratio of binding to NP4 to the binding to NP25). ELISPOT. Eight half-log serial dilutions of primary spleen cell suspensions from C57BL/6 mice were cultured in duplicate overnight at 37 C on MultiScreen-HA ELISPOT plates (Millipore) coated with NIP 15 BSA (Biosearch Technologies), and nonspecific binding was blocked by incubation with a solution of 1% (wt/vol) BSA in PBS. Spots were detected with horseradish peroxidaseconjugated anti-mouse IgG (1030-05; Southern Biotech) and plates were developed with an AEC staining kit (Sigma). Spots were scanned and counted on an Immunospot analyzer (CTL Analyzers). Real-time RT-PCR. TCR+CD19 CD1d tetramerpositive iNKT cells were isolated from V14-transgenic mice with a BD Influx high-speed cell sorter. RNA was extracted from TRIzol-fixed cells with an RNeasy mini kit according to the manufacturers instructions (Qiagen). A High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) plus RNase inhibitors were used for the production of cDNA. Primers and probes from Applied Biosystems and the TaqMan 7500 Fast System and software (Applied Biosystems) were used for realtime RT-PCR of cDNA samples; expression was calculated by the change-inthreshold method (CT) with GAPDH mRNA (encoding glyceraldehyde phosphate dehydrogenase) as a reference. Confocal fluorescence microscopy. Frozen spleens cut into sections 7 m in thickness and embedded in optimum cutting temperature compound were labeled with fluorescein isothiocyanateanti-B220 (RA3-6B2; BD) plus antibody to fluorescein isothiocyanateAlexa Fluor 488 (A11096; Invitrogen), phycoerythrinanti-CD138 (281-2; BD), rabbit anti-CD3 (145-2C11; BD) plus Alexa Fluor 647anti-rabbit (A21245; Invitrogen), phycoerythrin anti-B220 (RA3-6B2; eBioscience), fluorescein isothiocyanateGL7 (GL7; BD) plus antibody to fluorescein isothiocyanateAlexa Fluor 488 (A11096; Invitrogen) and NP-APC. NP-APC was conjugated as reported45. Fluorescent images were obtained with a TE2000-U inverted microscope with a C1 Plus Confocal System (Nikon; Partners Confocal Microscopy Core) and an Axiovert 200M fluorescence microscope (Zeiss; Trudeau Institute). Final stitched high-resolution whole-spleen confocal images were obtained with a TCS SP5 confocal microscope with LAS AF 2.2.1 software (Leica; Trudeau Institute). Statistics. GraphPad PRISM 5 software was used for nonparametric two-tailed t-tests for normally distributed data sets with ten or more samples. The twotailed nonparametric Mann-Whitney test was used for smaller data sets for which normality could not be determined.
40. Shih, T.-A.Y., Roederer, M. & Nussenzweig, M.C. Role of antigen-receptor affinity in T cell-independent antibody responses in vivo. Nat. Immunol. 3, 399406 (2002). 41. Bendelac, A., Hunziker, R.D. & Lantz, O. Increased interleukin 4 and immunoglobulin E production in transgenic mice overexpressing NK1 T cells. J. Exp. Med. 184, 12851293 (1996). 42. Cui, J. et al. Requirement for V14 NKT cells in IL-12-mediated rejection of tumors. Science 28, 16231626 (1997). 43. Lam, KP., Kuhn, R. & Rajewsky, K. In vivo ablation of surface immunoglobulin on mature B cells by inducible gene targeting results in rapid cell death. Cell 90, 10731083 (1997). 44. Ozaki, K. et al. A critical role for IL-21 in regulating immunoglobulin production. Science 298, 16301634 (2002). 45. Eaton, S.M., Burns, E.M., Kusser, K., Randall, T.D. & Haynes, L. Age-related defects in CD4 T cell cognate helper function lead to reductions in humoral responses. J. Exp. Med. 200, 16131622 (2004).

2011 Nature America, Inc. All rights reserved.

nature immunology

doi:10.1038/ni.2172