CD1d-Restricted Immunoglobulin G Formation to GPI-Anchored Antigens Mediated by NKT Cells

See allHide authors and affiliations

Science  08 Jan 1999:
Vol. 283, Issue 5399, pp. 225-229
DOI: 10.1126/science.283.5399.225


Immunoglobulin G (IgG) responses require major histocompatibility complex (MHC)–restricted recognition of peptide fragments by conventional CD4+ helper T cells. Immunoglobulin G responses to glycosylphosphatidylinositol (GPI)- anchored protein antigens, however, were found to be regulated in part through CD1d-restricted recognition of the GPI moiety by thymus-dependent, interleukin-4–producing CD4+, natural killer cell antigen 1.1 [(NK1.1)+] helper T cells. The CD1-NKT cell pathway regulated immunogobulin G responses to the GPI-anchored surface antigens of Plasmodium andTrypanosoma and may be a general mechanism for rapid, MHC-unrestricted antibody responses to diverse pathogens.

NKT cells are unusual CD4+, NK1.1+ lymphocytes (1) that produce interleukin-4 (IL-4) rapidly in response to T cell receptor (TCR) ligation (2). These cells have a skewed VαVβ TCR repertoire (Vα14 and Vβ8 in mice) (3), suggesting that they are positively selected by a limited range of ligands. Murine NKT cells are positively selected by cortical thymocytes expressing the non-MHC–encoded but MHC class I–like molecule CD1d (1). The related human CD1b and CD1c molecules can elicit cytolytic and interferon-γ responses by presenting mycobacterial glycolipid antigens to CD8+ or CD4CD8 T cells (4). Murine Vα14+ NKT cells recognize synthetic α-galactosylceramide in the context of CD1d (5), and murine CD1d in transfected human T2 cells associates with phosphatidylinositol (PI)-containing compounds that may be GPIs (6). Therefore, CD4+ NKT cells may participate in CD1d-restricted recognition of lipid antigens. However, the natural ligand and functional significance of NKT cells in immune responses in vivo remains unclear. The present study was initiated to test the assumption of MHC restriction of immunoglobulin G (IgG) responses to the GPI-anchored surface antigens of important protozoal pathogens. We have found instead that several such responses are controlled in part by CD1d-restricted recognition of GPI moieties by CD4+ NKT cells.

T cell–dependent IgG responses to protein antigens are thought to be exclusively MHC class II–restricted. However, allogeneic bone-marrow irradiation chimeras were similar to syngeneic controls in responding to malaria sporozoites (SPZs) with IgG to the circumsporozoite (CS) protein, despite being unable to respond to the nominal protein antigens tetanus toxoid (TT) or a full-length recombinant Plasmodium falciparum CS protein (recCS) (7). Nude mice cannot respond to SPZs with anti-CS IgG, and passive transfer of depleting antibodies to CD4 into euthymic animals abolishes the anti-CS response to SPZs (8). However, nude mice engrafted with irradiated neonatal allogeneic thymi mounted anti-CS IgG responses to SPZs similar to those of recipients of syngeneic thymi, but did not respond to recCS or TT (7). Mice lacking both class II and class II–restricted CD4+ T cells, and that are unable to respond to T-dependent antigens (9), produced anti-CS IgG (mean log2 reciprocal titer of 10) in response to SPZs (10). Thus, CD4+ T cells are required for the IgG response to the native CS protein, but this may proceed through a MHC class II–independent route.

The native CS protein is posttranslationally modified by a GPI anchor (11), whereas TT and recCS are not. To determine whether the GPI anchor accounts for the difference in immunological behavior of the proteins, we purified COOH-terminal GPIs from affinity-purified GPI-anchored proteins ofP. falciparum (PfGPI) and Trypanosoma bruceimembrane-form variant surface glycoprotein (mfVSG) (12) (Fig. 1A), and nonprotein-linked free GPIs from Leishmania mexicana (12, 13) (Fig. 1B). The compositional purity of these latter molecules was confirmed by gas chromatography–mass spectrometry (GC-MS). In addition, a phosphorylated and lipidated mammalian GPI based on the rat brain Thy-1 GPI (Fig. 1C), and the corresponding inositolphosphoglycan (IPG) lacking a lipid tail, both chemically synthesized by n-pentenyl glucoside strategy and compositionally pure by 1H nuclear magnetic resonance (NMR) analysis (14), were also used. To generate responses to the hapten fluorescein (FLU), native and synthetic GPIs and IPG were exposed to 2-iminothiolane to introduce a sulfhydryl onto free amino groups, desalted, and conjugated in a molar ratio of 1:1 to fluoresceinated, maleimide-activated ovalbumin (OVAFLU). In contrast to sham-OVAFLU alone or sham-OVAFLUmixed with equal molar amounts of free PfGPI, PfGPI-OVAFLU conjugates were able to induce anti-FLU IgG1 formation in MHC class II−/− mice (mean log2reciprocal titer of 8). Similar IgG responses were obtained in class II−/− mice with the mfVSG of T. brucei, but not the deacylated soluble VSG derived by PI-specific phospholipase C (PI-PLC) hydrolysis (Fig. 1A), demonstrating that the GPI lipid domain is required, and the GPI glycan is not sufficient, for the phenomenon. This was confirmed by comparing responses to OVAFLUconjugated to either synthetic Thy-1 GPI or Thy-1 IPG lacking fatty acid (log2 reciprocal titer of 9.75 versus no response, respectively). Thus, IgG responses in class II−/− mice require linkage of antigen to GPI with an intact lipid, which may be composed of diacylglycerol or alkylacylglycerol.

Figure 1

Diagrammatical representation of GPI structures used in this study. Purification and compositional analyses are as described (12–14). (A) COOH-terminal GPIs from T. brucei and P. falciparum. Boxed areas represent modifications found in PfGPI. The cleavage site of mfVSG by phosphatidylinositol-specific phospholipase C (PI-PLC) is indicated. (B) Free iM2, iM4, and EP-iM4 GPIs of L. mexicana. Nomenclature is as described (14), where all isomers contain one mannose in α1-3 linkage, EP indicates ethanolamine phosphate, and M2 and M4 indicate the number of mannose residues, as shown. (C) Chemically synthesized rat brain Thy-1 GPI.

Although lacking conventional T cells, class II−/− mice retain a diverse population of MHC-nonrestricted CD4+α/β TCR+ T cells, including NKT cells and other CD4+ cells selected on CD1 (9, 15). We therefore hypothesized that IgG responses to native and synthetically conjugated GPI-anchored proteins in both wild-type and class II−/− mice proceed from CD1d-restricted presentation of GPIs to nonconventional T cells. To test this hypothesis, we examined the in vitro proliferative and cytokine responses to purified malarial GPI of splenocytes from animals primed with malaria SPZs. There was a marked increase in both the relative and absolute numbers of NK1.1+ CD4+ blastoid cells responding to GPI from both class II−/− (Fig. 2A) and wild-type mice. A high frequency of both Vα14+ CD4+ (Fig. 2A) and Vβ8+ cells was also detected in the responding population. No exogenous cytokines were required for this proliferation, but supplementation of cultures with IL-2 (5 U/ml) increased the level of response.

Figure 2

Response of peripheral NKT cells to purified GPIs in vitro. (A) As determined by forward (FSC) and side (SSC) light scatter, splenocytes from SPZ-primed class II−/−donors proliferate within 48 hours exposure to PfGPI (unshaded) compared with medium controls (shaded). The responding cells are NK1.1+, CD4+ and include a Vα14+, CD4+ subset (25). (B) Splenocytes from class II−/− donors were exposed to various antigens, and [3H]TdR incorporation was determined after 4 days. Other cultures were exposed to PfGPI or Thy-1 GPI for 4 days, washed, and cultured in IL-2 (10 U/ml) for 2 days, followed by replating with irradiated wild-type APCs and restimulation with either PfGPI or Thy-1 GPI for 48 hours. IL-4 levels in the supernatant were determined by capture ELISA.

To examine the fine specificity of responding cells, we exposed splenocytes from wild-type and class II−/− animals primed to P. berghei SPZs to 0.5 μM of the GPI structures shown in Fig. 1, together with dipalmitoyl-PI. The cells responded to a similar degree to most intact GPIs, but only weakly to the iM2 GPI with truncated glycan and not at all to glycans lacking the fatty acid domain, or to PI (Fig. 2B). Thus, both glycan and fatty acids are required for recognition, and NKT cells from SPZ-primed donors respond to a range of GPIs from diverse protozoal and mammalian taxa. However, because the full range of structures presented to the host under these priming conditions is not known, the results may reflect either broad recognition of diverse antigens by the general population of NKT cells or heterogeneous responses from a clonally mixed population. To further distinguish between these possibilities, cells were expanded in the presence of either PfGPI or Thy-1 GPI for 4 days, rested in IL-2, then restimulated with homologous or heterologous antigen. Cells expanded by either antigen responded significantly less well to the heterologous stimulus (Fig. 2B). Analysis of antigen-specific frequencies after in vitro culture revealed no increase above background levels in response to PfGPI in naı̈ve donors, but a clear increase in antigen-reactive NKT cells from SPZ-primed donors (7 to 30% of total) (16). However, naı̈ve and primed donors both mounted significant responses to the iM4 GPI (to 5% of NKT cells) (16). Together, the data are consistent with the reported fine specificity of CD1-restricted CD4CD8 human T cells for the glycan component of glycolipids (4, 17), suggesting that antigen priming expands clonally diverse NKT cells that are able to discriminate among structurally distinct GPIs, but that high precursor frequencies for some GPIs may occur even among naı̈ve animals kept under specific-pathogen–free conditions.

Cells that have been positively selected on NK1.1 and CD4 proliferate and produce cytokines specifically in response to TCR-mediated signals (2, 18). When sorted NK1.1+CD4+ cells from wild-type and class II−/−mice that had been primed to SPZs were cultured with irradiated wild-type or class II−/− antigen-presenting cells (APCs), they responded to purified GPIs, as determined by incorporation of [3H]thymidine ([3H]TdR) and the production of high levels of IL-4 (Fig. 3A). No proliferation in the absence of APCs indicated that GPIs do not provide a direct activation signal to NKT cells that is sufficient to induce cell growth. NKT cells did not respond to GPIs when cultured with irradiated APCs from β2-microglobulin−/−2M−/−) and CD1.1/CD1.2−/−(CD1−/−) (19) donors, or with wild-type and class II−/− APCs in the presence of anti-CD1.1 (1B1) (20), but responded fully in the presence of isotype controls (Fig. 3A). The proliferative and IL-4 response to PfGPI of NKT cells and the Vα14+, CD4+ subset in unfractionated splenocytes could also be blocked by the anti-CD1 monoclonal antibody 1B1 (Fig. 3B). Thus, the recognition of GPIs by NKT cells is MHC-independent and CD1-restricted. In addition, NKT cells produced IL-4 in response to CD1.1-transfected J774 macrophages (20) in the absence of exogenous antigen, but not in response to sham-transfected controls. Nonetheless, the response was enhanced when CD1.1-transfectants were pulsed with PfGPI (Fig. 3A). The response of NKT cells to CD1.1, observed previously (5,18), has been adduced in support of the proposition that this cell population may play a physiological role in the absence of associative recognition of antigen (5, 21). However, as reported (5, 18), in the absence of exogenous antigen no cytokine expression was detected in response to APCs expressing normal levels of CD1.1. Thus, high levels of CD1.1 expression in transfected cells may alone be sufficient to drive proliferation. Alternatively, because NKT cells can respond to mammalian GPIs [for example, synthetic Thy-1 GPI (Fig. 2B)], CD1.1-transfectants may be able to present endogenous GPIs to NKT cells, as suggested by the association of CD1d in transfected T2 cells with PI-containing compounds (5). Thus, CD1 may not be “empty,” and self-reactive NKT cells may arise through incomplete negative selection. Such a possibility may explain the nonantigen-specific regulatory activity of NKT cells (2,22).

Figure 3

The proliferative and IL-4 response of NKT cells to PfGPI is MHC-independent and CD1-restricted. (A) Sorted NK1.1+, CD4+ cells (2 × 104) from wild-type or class II−/−donors were placed in triplicate with or without purified GPI on irradiated splenocyte APCs from wild-type (WT), class II−/−, β2M−/−, or CD1−/− donors, or CD1.1-transfected and sham-transfected J774 macrophages. [3H]TdR incorporation was determined after 3 days, or IL-4 production in the presence or absence of anti-CD1 was determined as in Fig. 2. (B) Splenocytes from SPZ-primed class II−/− donors were exposed to PfGPI in the presence or absence of anti-CD1 or isotype control and taken for flow cytometric analysis after 3 days.

NKT cells can induce Ig class switch in B cells exposed to anti-IgD (2). Extending to antigen-specific systems, NKT cells cooperated with B cells by ELISPOT assay in CD1-restricted IgG formation to GPI-OVAFLU and native P. berghei CS protein formation, but not to OVAFLU (Fig. 4A). To determine, therefore, whether CD1.1- or CD1.2-restricted antibody formation was a major or minor contributor to the IgG response to GPI-anchored proteins and SPZs in vivo, we exposed CD1−/− mice and wild-type controls to mfVSGFLU, SPZs, or recCS. Responses to mfVSGFLUand SPZs were significantly curtailed in CD1−/− mice (Fig. 4B), indicating that under these conditions the CD1-restricted pathway of IgG formation is an important component of responses to the native CS protein. Both groups responded equally to recCS, confirming that class II–dependent responses are unaffected by loss of CD1. It is not yet clear whether the IgG responses to SPZs in CD1−/−mice result from class II–restricted responses to the intact GPI-anchored CS protein or to a proportion of the antigen adventitiously deacylated in these preparations, or from a non–class II, non-CD1 pathway.

Figure 4

CD1-restricted antibody formation to neo-GPI-proteins and malaria SPZs. (A) Donornu/nu mice were primed twice with P. berghei SPZs or twice with LPSFLU. Splenocytes were cultured in the presence of IL-2 (10 U/ml), with and without antigen (0.1 μg/ml sham-OVAFLU, PfGPI-OVAFLU, or 5 × 104 SPZs), anti-class I, anti-class II, and anti-CD1, with 104 NKT cells from SPZ-primed class II−/−donors. Antigen-specific IgG production was quantified by ELISPOT against fluoresceinated dog serum albumin for responses to OVAFLU, and rCS for responses to SPZs. (B) Responses of CD1.1/CD1.2−/− (○) and Balb/c wild-type mice (•) to SPZs, recCS, and mfVSGFLU. IgG titers were determined as described (7).

GPIs are widespread among eukaryotes, and the expression of GPI-anchored proteins and free GPIs is particularly abundant among the parasitic protozoa. Because CD1-restricted NKT cells can recognize GPIs from diverse taxa (Fig. 2B), CD1-restricted IgG formation may represent a general mechanism for rapid responses to the GPI-anchored surface antigens of various pathogens. The native CS protein appears to be more immunogenic than the recombinant version. Inoculation of as few as 104 nonreplicating irradiated SPZs (1 ng of native CS protein, assuming 106 copies per cell) is sufficient to elicit an antibody response comparable to that obtained with 1 to 10 μg of recCS. The rapid responses of NKT cells in vivo (2) and a relatively high precursor frequency of antigen-specific NKT cells may contribute to this phenomenon. Thus, consistent with the “danger model” of pathogen-initiated immune responses (23), CD1-restricted immunity may be intermediate between the innate “pattern recognition” and adaptive immune systems. MHC-restricted nonresponsiveness to malarial surface antigens has been proposed to be a major obstacle to the development of vaccines (24). Because both human and murine CD1 molecules are relatively nonpolymorphic, GPI anchors may provide universal T cell sites, overcoming MHC restriction in antibody responses to various pathogens.

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


Stay Connected to Science

Navigate This Article