Apolipoprotein-mediated lipid antigen presentation in B cells provides a pathway for innate help by NKT cells

Unlike peptide antigens, which are presented to conventional T cells via major histocompatibility complex (MHC) molecules, lipid antigens are presented to T cells by the MHC-like molecule, CD1. Humans express several nonpolymorphic CD1 molecules, including CD1d, which presents lipids to a unique subset of T cells termed natural killer T (NKT) cells. NKT cells are innate-like lymphocytes defined by their characteristic semi-invariant T-cell receptor that recognizes the potent glycolipid antigen α-galactosylceramide (αGalCer). In addition to this nonphysiologic antigen, NKT cells have recently been shown to respond to exogenous bacterial-derived lipid antigens^1-3  as well as endogenous lipids presented by antigen-presenting cells (APCs) responding to innate stimuli.^2,4,5  Resulting in part from their inherent memory phenotype and rapid cytokine secretion after stimulation, NKT cells complement innate signals and promote adaptive immunity.⁶ 

αGalCer has been shown to have adjuvant-like properties that enhance T-dependent and T-independent humoral immune responses.^7-10  To generate antibodies to T-dependent antigens, B cells require cognate T-cell help from conventional T helper (Th) cells that recognize the same antigen or antigenic complex, which has been internalized by the B-cell receptor (BCR). αGalCer has been shown to boost dendritic cell (DC)–Th interactions that augment Th priming and indirectly promote B-cell help.¹¹  However, B cells also express CD1d and have been shown to present lipid antigens and recruit cognate help from NKT cells.^8,9,12,13 

Recent work has shown that BCR-mediated uptake of model B-cell antigens linked to αGalCer elicits cognate NKT help for lipid-specific B cell responses in vivo and in vitro.^12-14  However, B cells also internalize and present lipid antigens to NKT cells by pathways that are independent of the BCR,¹⁵  and the mechanisms by which this occurs are unknown. Our previous studies have shown that DCs use the low-density lipoprotein (LDL)–receptor pathway to endocytose apolipoprotein E (apoE)–bound lipid antigens for subsequent presentation to NKT cells.¹⁶  ApoE, found in serum very low density lipoproteins or secreted locally by DCs and macrophages, rapidly binds exogenous lipid antigens. ApoE-lipid antigen complexes are efficiently captured by the LDL-R on DCs and delivered to intracellular CD1d where the lipid antigen is loaded. Whereas DCs use multiple nonspecific pathways of antigen uptake, B-cell antigen uptake is thought to be restricted to the BCR. An LDL-R-mediated pathway in B cells would provide a means for innate help by NKT cells stimulating polyclonal B-cell activation. We thus investigated whether B cells may also use an apoE-mediated pathway for lipid antigen presentation.

Two CD1d-restricted human NKT cell clones (BM2a.3 and J3N.5, previously described¹⁷ ) and 1 CD1d-restricted human NKT line (M0) were used in these studies, and similar results were obtained. M0 was derived by successive rounds of αGalCer-CD1d-tetramer sorting and expansion with αGalCer, with a resulting population of T cells that are more than 99% CD4 positive, more than 95% Valpha24/V beta 11 (double positive), and more than 97% αGalCer-CD1d-tetramer–positive. NKT cells were maintained in RPMI media (supplemented with 10% fetal bovine serum, HyClone Laboratories; 2% human AB serum, MP Biomedical; penicillin/streptomycin, and 200 U/mL interleukin-2) and restimulated every 17 to 30 days using phytohemagglutinin (PHA) and irradiated peripheral blood mononuclear cells obtained from healthy volunteers.

Tonsils were obtained from anonymous patients undergoing tonsillectomy at British Columbia Children's Hospital with ethics approval obtained from the University of British Columbia and Children's and Women's Health Center of British Columbia Review Board and with informed consent obtained in accordance with the Declaration of Helsinki. Tonsils were disrupted mechanically and lymphocytes were isolated by Ficoll separation before B-cell purification. B cells were purified either by sorting CD19⁺ cells on BD FACSAria or by magnetic selection using a B-cell enrichment kit (StemCell Technologies) according to the manufacturer's instructions. All preparations yielded more than 99% CD19⁺ cells.

Purified B cells were cultured in 96-well plates at a 1:1 ratio with NKT cells (5 × 10⁵ cells/well in triplicate) in serum-free AIM V media, media with αGalCer, or media with αGalCer plus 2.5 μg/mL fractionated apoE. Control wells were stimulated with PHA, anti-CD3, anti-Ig, or Staphylococcus Aureus Cowan I with or without apoE. On day 4, wells from each treatment were pooled, counted, and stained as indicated in “Antibodies and flow cytometry.” The percentage of CD19⁺ and CD3⁺ cells was used to calculate absolute B cells and NKT-cell numbers from total cell counts.

On day 10, immunoglobulin concentration in supernatants was determined by enzyme-linked immunosorbent assay (ELISA) using goat antihuman IgG polyclonal antibody (unlabeled for capture and horseradish peroxidase–conjugated for detection from Southern Biotechnology). Standards were obtained from Bethyl Laboratories.

Purified B lymphocytes were cultured for 18 to 24 hours in either media alone (RPMI plus 10% fetal bovine serum) or stimulated with anti-Ig (10 μg/mL; Southern Biotechnology) to cross-link BCRs, or with CD40L provided by a monolayer of irradiated (3000 cGy) CD40L-expressing L cells (mouse fibroblast cell line). The next day, B lymphocytes were washed and cultured at a 1:1 ratio with NKT cells (5 × 10⁵ cells/well in triplicate in 96-well plates) overnight in serum-free media (SFM) consisting of RPMI supplemented with l-glutamine and penicillin/streptomycin and 0.4% bovine serum albumin (used as a nonspecific carrier) or AIM V (Invitrogen) for long-term cultures. Lipid antigens (αGalCer, synthesized and kindly provided by Dr Huib Ovaa, Amsterdam; and αGalGalCer, synthesized and kindly provided by Dr Petr Ilarionov and Dr Gurdyal Besra) were incubated for 10 minutes with SFM alone or SFM plus 2.5 μg/mL fractionated apoE (Meridian), or recombinant apoE (rApoE2, rApoE3, or rApoE4 from Invitrogen), or with 5% fresh human serum before addition to cells. PHA (Sigma-Aldrich) was used as a positive control. The concentration of apoE used was optimized as described previously¹⁶  and reflects the concentration found in typical serum containing media. Where indicated, blocking antibodies, including anti-CD1d (12 μg/mL, clone 12.1.1.1¹⁸ ), goat polyclonal anti–LDL-R (5 μg/mL, R&D Systems), monoclonal anti–LDL-R (4 μg/mL clone IgG-C7; Biodesign International), or equal amounts of corresponding isotype control antibodies (mIgG₁, goat IgG, and mIgG_2b, respectively) were used. The next day, culture supernatants were harvested and T-cell interferon-γ (IFN-γ) production was measured by capture ELISA using anti–IFN-γ capture antibody (2G1) and biotin-labeled anti–IFN-γ (B133.5) detection antibody (both from Thermo Electron). Lipid uptake experiments were performed by pulsing CD40L-activated B cells for 4 hours in media alone or with lipid antigens with or without 2.5 μg/mL apoE. Cells were washed before being cultured with NKT cells in SFM overnight as previously described in this paragraph.

Fluorescein isothiocyanate, phycoerythrin (PE), peridinin chlorophyll protein, and allophycocyanin-conjugated monoclonal anti-CD86, CD19, CD3, and CD154 were purchased from BD Biosciences PharMingen. Monoclonal anti–LDL-R (clone IgG-C7; from Biodesign International) antibody was detected with PE-conjugated anti–mouse IgG PE secondary antibody (Southern Biotechnology). Isotype-matched antibodies and secondary reagents were used as controls. For proliferation studies, cells were first surface stained with anti-CD3 and anti-CD19 antibodies and then washed, fixed, and permeabilized using Cytofix/Cytoperm solutions (BD Biosciences) according to the manufacturer's instructions. Fluorescein isothiocyanate-labeled anti–Ki-67 antibody (Dako Denmark) was used to detect proliferating cells (late G₁, S, G₂, and M phases of cell cycle).

Flow cytometry was performed on FACScalibur (BD Biosciences), and data were analyzed using FlowJo software (TreeStar).

Statistical analysis was performed using the paired Student t test or by analysis of variance using Excel Version 12.0 (Microsoft) or Prism Version 4 (GraphPad) software. The fold enhancement by apoE was calculated by determining the half-maximal stimulation of the dose-response curve as calculated with Prism software.

We previously demonstrated that apoE could enhance lipid antigen presentation to NKT cells by DCs.¹⁶  Because αGalCer has been shown to promote humoral immunity in a CD1d-dependent manner, we tested whether apoE-delivery of αGalCer to B cells enhances NKT help for B-cell proliferation and immunoglobulin production. Purified tonsillar B cells were cultured with αGalCer and NKT cells in the presence or absence of human apoE. Addition of apoE substantially enhanced both NKT-cell and B-cell proliferation in response to αGalCer, as shown by expression of the nuclear proliferation antigen Ki-67 (Figure 1A,C,E) and total cell number (Figure 1B,D). In the absence of apoE, αGalCer was ineffective at inducing B-cell and NKT-cell proliferation even at relatively high concentrations (500 ng/mL). With apoE, proliferation was observed with αGalCer concentrations as low as 1 ng/mL. This enhancement by apoE did not occur in the absence of antigen, or with controls using nonspecific stimuli, including B-cell agonists (anti-Ig and Staphylococcus Aureus Cowan I) or T-cell agonists (PHA and anti-CD3; Figure 1 and data not shown).

Next, we analyzed B-cell activation and antibody production to assess the influence of apoE on efficient αGalCer presentation. B cells and NKT cells were cocultured with αGalCer in the presence or absence of apoE, and CD86 expression was measured. Whereas αGalCer alone had little effect on B-cell CD86 expression, apoE/αGalCer-treated cells expressed 2- to 4-fold higher levels of CD86 on day 2 and day 4, respectively (Figure 2A). Likewise, antibody production was entirely dependent on apoE/αGalCer (Figure 2B). ApoE/αGalCer elicited significantly (7-fold) higher IgG than media, αGalCer, apoE, or B cells only (Figure 2B; and data not shown). Together, these data suggest that B cells are highly dependent on apoE for efficient uptake of lipid antigens to recruit NKT-cell help.

The LDL-R is the predominant cell surface receptor that binds apoE and mediates lipoprotein internalization; however, other receptors can also capture apoE. Given that DCs primarily use LDL-R mediated uptake for antigenic lipids,¹⁶  we chose to investigate its role in lipid antigen uptake by B cells. Ex vivo, B cells express low levels of surface LDL-R¹⁹  (and data not shown). However, expression was found to be up-regulated after B-cell stimulation.¹⁹  A correlation between B-cell activation and LDL-R expression suggests that activated B cells may be more capable of uptake and presentation of apoE-bound lipid antigens. B cells were stimulated overnight with either anti-Ig to cross-link surface BCR or with CD40L to model conventional T-cell help. As expected, CD86 expression on B cells was increased by anti-Ig and CD40L treatment (Figure 3A). Both anti-Ig and CD40L (but not media) also induced LDL-R expression on approximately 40% to 50% of B cells by 24 hours, thus demonstrating that activated B cells up-regulate LDL-R.

Next, the ability of resting and activated B cells to present αGalCer to NKT cells was compared with or without apoE. We found that B cells activated by either anti-Ig or CD40L were modestly better at presenting free αGalCer than resting B cells. However, the delivery of αGalCer with apoE significantly enhanced their ability to activate NKT cells, shifting the dose-response curve 30-fold for anti-Ig and 105-fold for CD40L (vs just 3-fold enhancement for media alone; Figure 3B). Thus, the degree of enhancement by apoE is closely associated with LDL-R expression on B cells.

Although optimal loading of αGalCer onto CD1d requires trafficking through the endosomal-lysosomal pathway before presentation to NKT cells, αGalCer does not require processing for loading onto CD1d.²⁰  Given that protein antigens need to be processed into peptides for MHC loading, it is possible that some pathogen-derived lipid antigens also require processing to generate antigenic lipids for NKT-cell presentation. We investigated the ability of B cells to process lipid antigens using αGalGalCer, a lipid that requires uptake and processing in lysosomes where it is converted to the NKT agonist αGalCer.²⁰  Because CD40L-activated B cells were consistently the most potent APCs when assayed in vitro with apoE-αGalCer, we used these in subsequent experiments. Like αGalCer, αGalGalCer completely requires apoE for CD40L-activated B cells to activate NKT cells (Figure 3C), indicating that B cells are able to process lipid antigens internalized with apoE.

To rule out any effect of apoE directly on T cells, CD40L-activated B cells were pulsed for 4 hours with either αGalCer or αGalGalCer in the presence or absence of apoE, and the cells were then washed before incubation with NKT cells. B cells pretreated with lipid antigen alone did not promote NKT-cell activation, whereas addition of apoE allowed B cells to efficiently acquire αGalCer and αGalGalCer for presentation to NKT cells (Figure 3D). Although we and others have observed that all lymphocytes may up-regulate the LDL-R after activation (not shown),¹⁹  we found no direct effect when NKT cells were stimulated in the presence of αGalCer and apoE in the absence of B cells (Figure 3D). Thus, efficient uptake and presentation of αGalCer require B-cell uptake of apoE-bound lipid antigens.

To determine whether uptake of apoE-bound lipid antigens was mediated by the LDL-R, we pulsed CD40L-activated B cells with apoE-αGalCer in the presence of LDL-R blocking antibodies. NKT activation was blocked by 2 LDL-R blocking antibodies and a CD1d-blocking antibody, but not by corresponding control antibodies (Figure 4A). To further implicate LDL-R–mediated uptake, we compared the ability of different apoE isoforms to enhance αGalCer presentation by CD40L-activated B cells. Enhanced NKT activation was observed with apoE3 and apoE4 but not with apoE2, a variant defective in LDL-R binding (Figure 4B).^21,22  Importantly, compared with free αGalCer, apoE2 actually diminished αGalCer activity, suggesting that apoE2 binds and sequesters the antigen but is unable to be taken up by the B cells.

Our previous work demonstrated that αGalCer incubated in human serum distributes in the very low density lipoprotein fraction and that DC presentation to NKT cells was apoE-dependent. We wished to confirm that B-cell uptake of αGalCer from human serum, like purified apoE-bound αGalCer, uses the LDL-R pathway. LDL-R–specific antibodies abolished αGalCer stimulation incubated with either total human serum, or purified apoE, but had no effect on the PHA-stimulated control (Figure 4C). These findings indicate that αGalCer uptake in B cells from human serum requires the LDL-R.

In summary, apoE significantly enhances proliferation and activation of NKT cells induced by B cells and the reciprocal proliferation and activation of B cells. This enhancement is dependent on the LDL-R, which is expressed on activated B cells. Thus, in addition to the BCR-mediated uptake described for specific lipid antigens,^12-14  B cells can use an apolipoprotein-mediated pathway of lipid antigen uptake for presentation to NKT cells. This might be the predominant mechanism used by B cells for the uptake of αGalCer when used as an adjuvant for enhancing antibody responses to T-dependent antigens.^8-11  Although part of this effect can also be attributed to increased priming of antigen-specific Th cells,¹¹  αGalCer also induces specific antibody production to coadministered T-dependent antigen in the absence of class II-restricted Th cells, thus demonstrating a role for direct NKT help.^8-10 

Our previous work demonstrated the role for apoE and the LDL-R in human DCs presenting CD1-restricted antigens.¹⁶  Unlike DCs, however, our current findings indicate that B cells are much more dependent on apolipoprotein-mediated antigen capture and presentation. The degree of augmentation of B-cell antigen presentation by apoE (100- to 250-fold) is substantially greater than the enhancement seen in DCs (up to 50-fold enhancement)¹⁶  and is similar to that seen when antigens are internalized specifically by the BCR.^23,24  In addition, DCs were still able to capture and present lipid antigens associated with apoE2, albeit with a significantly reduced efficiency, whereas B cells were almost completely unable to present apoE2-bound αGalCer. It thus seems that B cells are dependent on the LDL-R for lipid antigen uptake, whereas DCs may use additional receptor pathways or macropinocytosis. Our findings suggest that, beyond BCR recognition, B cells may rely on the LDL-R for lipid antigen presentation in vivo. However, given that other APCs also contribute to the adjuvant effect of αGalCer observed in vivo, additional effects independent of apoE and LDL-R on B cells may also play a role.

This non–BCR-mediated pathway of antigen presentation in B cells elicits a form of innate help for B cells by NKT cells by providing help for B cells regardless of their specificity. The physiologic context by which NKT cells provide such help in vivo is unclear. Our finding that the LDL-R is up-regulated during B-cell activation suggests that NKT cells may augment B-cell maturation initiated by other stimuli, including antigen recognition or Th-cell interactions. It is also possible that, in addition to activated B cells, other B-cell subsets express the LDL-R. Interestingly, αGalCer has also been reported to enhance antibody responses to T-independent antigens that are typically generated by B-1 cells and marginal zone B cells.⁸  It will be important to determine whether these cell or other cell subsets also express the LDL-R.

The role that apolipoproteins play in the capture of microbial lipid antigens for presentation to NKT cells is currently unknown. Although αGalCer serves as a powerful tool for studying NKT-cell function, it is a nonphysiologic agonist. Whether bacterial-derived lipid antigens, such as the model NKT agonist αGalCer, are acquired by B cells via an LDL-R-mediated pathway to directly elicit NKT help for generation of protective antibodies remains to be determined. We previously demonstrated that bacterial antigens could be transferred by apoE from infected macrophages to bystander APCs.¹⁶  It has also been shown that NKT cells recognize lipids, such as α-glucuronylceramide and galactosyl diacylglycerols derived from Sphingomonas and Borrelia, respectively.^1-3  These CD1d-presented lipids may also contain B-cell eptiopes but, aside from containing specific targets for antibody production, such lipids may also serve as natural adjuvants that stimulate NKT cells to help B-cell antibody secretion.

Effective defense against Borrelia infection requires Borrelia-specific antibodies, a response that is impaired in the absence of CD1d indicating a role for NKT cells.²⁵  Interestingly, marginal zone B cells, an innate B-cell subset characterized by high expression of CD1d in mouse, are responsible for generating this early antibody response against the Borrelia spirochete.²⁶  Their ability to respond rapidly to blood-borne pathogens during infection and promoting subsequent adaptive immune response makes them attractive candidates for further exploring direct B-cell–NKT-cell collaboration. Other systems demonstrating cognate NKT-cell help for B cells, such as in Novosphingobium-induced primary biliary cirrhosis, where α-glucuronylceramides induce pathogenic autoantibodies,²⁷  are also potential models that will allow us to explore the role of apoE and the LDL-R in antimicrobial responses.

We propose that the LDL-R-mediated pathway for uptake of lipid antigens by B cells provides a mechanism for the adjuvant effects of αGalCer and potentially for the uptake of bacterial-derived lipid antigens or endogenous lipid antigens that have been recently described.^1-4,28  By providing polyclonal B-cell stimulation, this pathway of innate help could bolster humoral responses to infection. However, without shared antigen specificity, such promiscuous antigen presentation by B cells may also promote the generation of pathogenic autoantibodies, and potentially, NKT cell-mediated self-reactivity.

The authors thank Dr Michael Gold for critical reading of the manuscript, Ms Pamela Lutley for assistance with tonsil collection, and Dr Elizabeth Leadbetter for helpful suggestions.

L.L.A. was supported by the Canadian Institute for Health Research and Canadian Blood Services. P.v.d.E. was supported by the Multiple Sclerosis Society of Canada, the National Multiple Sclerosis Society, the Michael Smith Foundation for Health Research, and the Canadian Institute for Health Research.

Contribution: L.L.A. and P.v.d.E. designed research, analyzed data, and wrote the paper; L.L.A., K.H., and D.-J.Z. performed research; and B.K.C., F.K.K., and R.T. provided reagents/tissues and input.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Peter van den Elzen, Department of Pathology and Laboratory Medicine, University of British Columbia, Child and Family Research Institute, 950 W 28th Ave, Rm A4-145, Vancouver, BC, Canada; e-mail: pvde@interchange.ubc.ca.