Dendritic cell maturation enhances CD8⁺ T-cell responses to exogenous antigen via a proteasome-independent mechanism of major histocompatibility complex class I loading

Abstract

Immune stimulating complexes (ISCOMS) containing the saponin adjuvant Quil A are vaccine adjuvants that induce a wide range of immune responses in vivo, including strong class I major histocompatibility complex (MHC) -restricted cytotoxic T-lymphocyte activity. However, the antigen-presenting cell responsible for the induction of these responses has not been characterized. Here we have investigated the role of dendritic cells (DC) in the priming of antigen-specific CD8⁺ T cells in vitro by ISCOMS containing ovalbumin. Resting bone marrow DC pulsed with ovalbumin ISCOMS efficiently prime resting CD8⁺ T cells through a mechanism that is transporter associated with antigen processing (TAP) dependent, but independent of CD40 ligation and CD4⁺ T-cell help. Lipopolysaccharide-induced maturation of DC markedly enhances their ability to prime CD8⁺ T cells through a mechanism which is also independent of CD4⁺ T-cell help, but is dependent on CD40 ligation. Furthermore, DC maturation revealed a TAP-independent mechanism of CD8⁺ T-cell priming. Our results also show that class I MHC-restricted presentation of ovalbumin in ISCOMS by DC is sensitive to chloroquine and brefeldin A but insensitive to lactacystin. We suggest that DC may be the principal antigen-presenting cells responsible for the priming of CD8⁺ T cells by ISCOMS in vivo and that targeting these vectors to activated DC may enhance their presentation via a novel pathway of class I antigen processing.

Introduction

An important goal of vaccine research is to develop a single-dose, orally administered, recombinant vaccine that induces long-lasting mucosal and systemic immunity.¹ As recombinant proteins are poorly immunogenic, particularly for mucosal and major histocompatibility complex (MHC) class I-restricted immune responses, their use in vaccines requires co-administration of an adjuvant.² A number of mucosal vaccine adjuvants have been described, including live vectors such as attenuated Salmonella and non-viable agents such as cholera toxin or the heat-labile enterotoxin of Escherichia coli (LT).³ Our approach has been to use immune-stimulating complexes (ISCOMS) containing the saponin adjuvant Quil A. ISCOMS are rigid cage-like structures that form spontaneously when mixing cholesterol, phosphatidylcholine and Quil A.⁴ Proteins incorporated in ISCOMS become highly immunogenic in vivo, inducing a wide range of immune effector responses, including delayed-type hypersensitivity, class I MHC-restricted cytotoxic T-lymphocyte activity, serum antibody production, T-cell proliferation, and secretion of both T helper type 1- (Th1) and Th2-dependent cytokines.^5–10 Of great importance is the fact that ISCOMS are equally effective by most parenteral and mucosal routes and therefore it would be useful to investigate their modes of action, with the aim of exploiting these for improved efficacy.

It is now appreciated that one of the most important properties of an adjuvant is how they interact with professional antigen-presenting cells (APC) to enhance presentation and co-stimulation.¹¹ Of these APC, the dendritic cell (DC) is now recognized as the most potent at priming both CD4⁺ and CD8⁺ T cells in vivo and activation of DC enhances their ability to present antigen to CD4⁺ T cells.¹² Thus adjuvants or pathogens that target and activate DC are likely to be highly immunogenic.

ISCOMS recruit a number of accessory cells in vivo, including DC,¹³ but very little is known of which APC presents the ISCOMS-associated antigen and of how ISCOMS are processed by defined APC subsets. In particular, it is not known how ISCOMS can prime CD8⁺ T-cell responses so efficiently in vivo, an unusual, but important, property for a vaccine adjuvant. In addition, unlike CD4⁺ T-cell responses, where the role of DC maturation is well characterized,¹² there is little information on the effects of DC maturation on class I-restricted presentation to CD8⁺ T cells, particularly when exogenous antigens have been used. Here we have tested the hypothesis that presentation by DC underlies the priming of CD8⁺ T cells by ISCOMS and have examined the mechanisms responsible for this phenomenon.

Our results show that bone marrow-derived DC are extremely effective at presenting ISCOMS-associated antigen to resting CD8⁺ T cells in vitro and this is greatly enhanced when ISCOMS-pulsed resting DC are induced to differentiate with lipopolysaccharide (LPS). The priming of CD8⁺ T cells is independent of cognate CD4⁺ T-cell help and the intracellular processing involves elements of both the class I and II pathways. Interestingly, the presentation by resting DC is dependent on TAP, but independent of CD40–CD40 ligand (CD40L), whereas the maturation effect reveals a CD40L-dependent, TAP-independent pathway. These findings show for the first time that ISCOMS can be presented very efficiently to CD8⁺ T cells by DC and also suggest the existence of a novel pathway of class I MHC-restricted antigen processing which may be an important target for vaccine adjuvants.

Materials and methods

Mice

Female C57BL/6 (B6) (H-2^b) mice were obtained from Harlan Olac, Bicester, UK. OT-1 mice transgenic for a T-cell receptor specific for ovalbumin (OVA) 257–264⁺ (H-2K^b), TAP^–/– and MHC class II^–/– mice on the B6 background were obtained from Dr M Merkenschlager, Clinical Sciences Centre, Hammersmith Hospital, London, UK. All these mice were then bred under specific pathogen-free conditions at the University of Glasgow. CD40^–/– and β₂-microglobulin^–/– mice on the B6 background were obtained from Prof. David Gray and Prof. R Maizels, respectively (Institute of Cell, Animal and Population Biology, University of Edinburgh, UK). All mice were first used from 8 weeks of age.

Antigens

Both native OVA and the synthetic OVA peptide 257–264 were obtained from Sigma Genosys, Cambridge, UK. ISCOMS containing diphtheria toxoid (DT-ISCOMS) were formulated in this department by Antonio Aguila.

Preparation of OVA ISCOMS

ISCOMS containing palmitified OVA (Sigma, Poole, Dorset) were prepared using phosphatidylcholine, cholesterol and Quil-A (obtained from Dr K Lövgren-Bengtsson, Department of Virology, Biomedicum, Uppsala, Sweden) as described previously.⁵ The integrity of the OVA ISCOMS was ensured by electron microscopy, and the protein content was determined by the Bradford reaction (BioRad, Hemel Hempstead, Hertfordshire, UK). The ISCOMS used in this study contained OVA and Quil A at a ratio of 10 : 1 and were approximately 40 nm in diameter.

Preparation of spleen and lymph node cells

Spleen and lymph nodes were harvested from OT-1 mice and single-cell suspensions prepared in sterile RPMI-1640 medium (Gibco BRL, Paisley, UK) by homogenization through Nitex mesh (Cadisch, London, UK). After centrifugation at 400 g for 5 min, the cells were resuspended and viable cell counts were performed using phase-contrast microscopy. Where required, purified CD8⁺ T cells were obtained from OT-1 spleen and lymph node preparations by magnetic cell sorting (MACS) using negative selection according to the manufacturer's instructions (Miltenyi Biotec Ltd, Surrey, UK) with the following antibodies, anti-CD4, -CD19, -CD11b and -CD16/32 (all PharMingen, Oxford, UK). CD8⁺ T-cell purity was assessed by flow cytometry and was typically >98%.

Bone marrow-derived dendritic cells

To obtain DC, bone marrow cells were washed out of the femurs of adult mice in RPMI-1640 using a syringe and a 21-gauge needle. Aliquots of 3 × 10⁶ bone marrow cells were seeded into 90-cm Petri dishes (Bibby Sterilin, Middlesex, UK) and grown in RPMI-1640 containing 2 mm l-glutamine, 100 μg/ml penicillin, 100 μg/ml streptomycin, 1·25 μg/ml fungizone (all from Gibco BRL) and 10% fetal calf serum (FCS; Harlan Sera Labs, Crawley Down, UK) (complete medium) at 37° in 5% CO₂. On days 0, 3, 6 and 8 of culture, the medium was supplemented with 10% supernatant from the X-63 cell line transfected with the murine granulocyte–macrophage colony-stimulating factor (GM-CSF) gene.¹⁴ After 10 days, the non-adherent DC were harvested by gentle washing and were typically >85% CD11c⁺, class II^moderate, CD40^low, B7-1^low and B7-2^low when compared to LPS-activated DC (10 μg/ml for 24 hr) (Fig. 2).

Figure 2.

Phenotypic analysis of bone-marrow-derived DC. Flow cytometric analysis of the expression of class II MHC, CD40, B7-1 and B7-2 by freshly isolated (solid line) and LPS-stimulated (10 μg/ml for 24 hr) bone marrow DC (dotted line).

Antigen presentation in vitro

After harvest, aliquots of 4 × 10⁶ resting DC were plated in 24-well ultra-low adherence plates (Costar, New York, NY) and pulsed with OVA ISCOMS (0·5–10 μg/ml) or 10 μg/ml OVA peptide 257–264 for 2 hr at 37° in a total volume of 1 ml. For the last hour of culture, 50 μg/ml mitomycin C (Sigma) was added, before the DC were removed by gentle resuspension and washed four times in RPMI-1640. In some assays, 10 μg/ml LPS (Escherichia coli Serotype 026:B6, Sigma) was added for 24 hr following the antigen pulse period to induce maturation. After washing, the DC were plated in triplicate at 1 × 10⁵ cells/well in 96-well flat-bottomed microtitre plates (Costar) together with 2 × 10⁵ lymph node cells from OT-1 mice in a total volume of 200 μl. To assess T-cell proliferation, 1 μCi/well thymidine ([³H]TdR; West of Scotland Radionucleotide Dispensary, Western Infirmary, Glasgow, UK) was added for the last 16 hr of culture and cell-bound DNA was harvested onto glass-fibre filter mats (Wallac, Turku, Finland). [³H]TdR uptake was counted on a Betaplate counter (Wallac).

Intracellular processing

To assess the intracellular mechanisms of antigen processing brefeldin A, lactacystin, chloroquine, or cytochalasin D (all from Sigma) were added to the DC for 30 min prior to and throughout the 2 hr antigen pulse. Following inhibitor treatment DC were fixed in 0·8% paraformaldehyde (Sigma) for 5 min before being quenched for 5 min in a solution of Gly-Gly (Sigma) at a final concentration of 0·06%. Fixed DC were then washed four times with increasing volumes of cold RPMI-1640 before counting and plating.

Maintenance of cell lines

EG7.OVA cells were originally obtained from Dr M Bevan (University of Washington, Seattle, USA). These cells were derived from EL4 cells transfected with a single copy of a plasmid containing a cDNA copy of the chicken OVA mRNA and a neomycin resistance gene. OVA-expressing cells were selected for culture by maintenance in complete medium supplemented with 400 μg/ml Geneticin (G418 sulphate, Gibco BRL).

Two- and three-colour flow cytometry

Aliquots of 1 × 10⁵ −10 × 10⁵ cells were washed in fluorescence-activated cell sorter (FACS) buffer (PBS/2%FCS/0·05% sodium azide) and incubated for 30 min at 4° in 100 μl anti-CD16/CD32 antibody (Fc Block, PharMingen) to block non-specific binding. After washing in FACS buffer, combinations of biotinylated and fluorochrome-labelled monoclonal antibodies were added in 100 μl FACS buffer and incubated for 30 min at 4° in the dark. After washing twice, the cells were incubated for 30 min at 4° in 100 μl of phycoerythrin (PE) or peridinin chlorophyll-a protein (PerCP) conjugated to streptavidin (both PharMingen) to visualize the biotinylated antibodies. The monoclonal antibodies used were, PE-anti-CD8, biotinylated anti-Vα2, fluorescein isothiocyanate (FITC)-anti-CD69 and FITC-anti-CD25 (all PharMingen). To assess class I surface expression aliquots of 5 × 10⁵ DC were first incubated for 30 min at 4° with Fc block before the addition of anti-CD11c PE and biotinylated H-2K^b (PharMingen). After washing, streptavidin-FITC was added and the cells were incubated for a further 30 min at 4° before being washed twice. Immediately prior to acquisition propidium iodide (PI; Sigma) was added at 5 μg/ml in 200 μl. The cells were analysed on a Becton Dickinson FACScalibur flow cytometer and analysed using cellquest software.

Measurement of OVA-specific interferon-γ production

Supernatants from cultures of antigen-stimulated T cells were harvested after 48 hr of culture, centrifuged at 13 000 g and stored at −20° until required. As described previously,¹³ interferon-γ (IFN-γ) levels were determined using a sandwich enzyme-linked immunosorbent assay, with a pair of IFN-γ-specific antibodies (PharMingen) and standardized with known concentrations of recombinant murine IFN-γ (PharMingen).

Statistics

Where appropriate, results are expressed as the mean from triplicate cultures ±1 SD and were compared using Student's t-test (P < 0·02 considered significant); n = 3 or more related experiments.

Results

DC present ISCOMS-associated antigen to antigen-specific CD8⁺ T cells in vitro and this is enhanced by DC maturation

In the first experiments, we examined whether DC would present ISCOMS-associated antigen to OVA-specific CD8⁺ T cells in vitro. Bone marrow-derived DC were incubated with OVA ISCOMS, OVA 257–264 peptide, native OVA protein, or with ISCOMS containing an irrelevant protein (diphtheria toxoid) for 2 hr, washed and cultured with OT-1 lymphocytes.

OVA-specific CD8⁺ T cells stimulated with DC pulsed with ISCOMS containing 10 μg/ml OVA protein (equivalent to 0·22 μg/ml OVA peptide) showed a proliferative response greater than 20-fold above background (Fig. 1a,b), which was often equivalent to that seen using 10 μg/ml of the epitope peptide (Fig. 1c). The stimulation of OT-1 T cells was antigen specific, as there was no T-cell activation when DC were pulsed with DT-ISCOMS (Fig. 1a). No stimulation of CD8⁺ T cells was found when DC were pulsed with 10 μg/ml native OVA protein (data not shown), confirming that OVA at this concentration does not have an inherent ability to enter the class I MHC processing pathway.

Figure 1.

Presentation of ISCOMS-associated antigen to CD8⁺ T cells. (a) DC were untreated, pulsed with either OVA ISCOMS or DT ISCOMS for 2 hr, or pulsed with OVA ISCOMS or DT ISCOMS and stimulated with 10 μg/ml LPS for 24 hr, washed and co-cultured with OT-1 lymphocytes for 48 hr. (b) DC were untreated, pulsed with OVA ISCOMS, or pulsed with OVA ISCOMS following replating in medium or after activation with LPS, washed and co-cultured with OT-1 lymphocytes for 48 hr. (c) DC were untreated, pulsed with OVA peptide 257–264 for 2 hr, or pulsed with OVA peptide 257–264 and stimulated with LPS for 24 hr, washed and co-cultured with OT-1 lymphocytes for 48 hr. Results shown are mean [³H]TdR incorporation ±1 SD for triplicate cultures. (** P < 0·01 versus OVA ISCOMS pulse only, ††P < 0·01 versus peptide pulse only).

The ability of freshly isolated bone marrow DC to present ISCOMS OVA to OVA-specific CD8⁺ T cells suggested that immature DC are involved in this activity (Fig. 2). Inflammatory agents such as LPS are known to enhance their ability to present antigen to CD4⁺ T cells¹² and we therefore examined the effects of DC maturation on the presentation of ISCOMS OVA to CD8⁺ T cells. LPS induced the expression of increased levels of class II MHC, CD40, CD80 and CD86 (Fig. 2) and the presentation of both OVA in ISCOMS and of OVA peptide, but not DT-ISCOMS to OT-1 cells was enhanced when DC were treated with LPS for 24 hr after the pulse with antigen (Fig. 1a,c). DC that had simply been replated in fresh medium for 24 hr after culture with ISCOMS also showed an enhanced ability to stimulate CD8⁺ T cells, but this effect was much less than that of LPS (Fig. 1a). Replating or activating the DC for 24 hr before the pulse with OVA ISCOMS significantly reduced their ability to stimulate the OT-1 lymphocytes (Fig. 1b).

As the OT-1 cells used as responders in these experiments were heterogeneous in nature, we confirmed that the cells being stimulated by OVA ISCOMS-pulsed DC were indeed antigen-specific CD8⁺ T cells by examining the expression of CD69 and CD25 on transgenic OT-1 cells 24 hr after culture with OVA ISCOMS-pulsed DC. Antigen-stimulated OT-1 cells showed a dose-dependent increase in the expression of CD69 and CD25 and this was enhanced when DC were activated with LPS, when almost 100% of the OVA-specific CD8⁺ T cells expressed both markers (Fig. 3a,b). In parallel, the activated OT-1 cells produced antigen-specific IFN-γ, indicating that presentation of OVA ISCOMS by DC induced functional differentiation of CD8⁺ T cells (Fig. 3c). Again this was enhanced by LPS treatment of DC after uptake of OVA ISCOMS.

Figure 3.

Antigen dose dependence of CD8⁺ T-cell responses to ISCOMS-associated antigen. DC were pulsed with 0–10 μg/ml OVA ISCOMS with or without LPS activation, washed and co-cultured with OT-1 lymphocytes. (a,b) The expression of CD69 and CD25 on CD8⁺ Vα2⁺ T cells was assessed after 24 hr in culture, and (c) IFN-γ production was assessed after 48 hr in culture. Results shown are mean ±1 SD for triplicate cultures. (*P < 0·02; **P < 0·01 versus OVA ISCOMS pulse only).

Intracellular processing events in the presentation of OVA in ISCOMS to CD8⁺ T cells by DC

As it is relatively unusual for exogenous antigens to enter a class I processing pathway, we next investigated some of the intracellular parameters determining the presentation of OVA in ISCOMS to OVA-specific CD8⁺ T cells by DC. In these experiments the DC were treated with inhibitors for 30 min prior to and during the 2 hr incubation with ISCOMS and then fixed immediately, so that the responses obtained reflected only the processing events that occurred within the period of antigen exposure. As a result, the proliferative responses of the OT-1 cells were much reduced compared with those using the non-fixed DC described above. To control for the effects of the inhibitors on total class I MHC levels, we assessed its expression by flow cytometry and also measured proliferative responses to a limiting dose of OVA 257–264 peptide, which in preliminary experiments was shown to be 10 pg/ml (Fig. 4a).

Figure 4.

Effects of brefeldin A on presentation of ISCOMS OVA to CD8⁺ T cells. (a) DC were pulsed with a range of OVA peptide concentrations for 2 hr before being fixed, washed and co-cultured with OT-1 lymphocytes for 72 hr. (b) DC were treated with 5 μg/ml brefeldin A for 30 min before and throughout the 2 hr pulse with OVA ISCOMS, or (c) 10 pg/ml OVA peptide, before being fixed, washed and co-cultured with OT-1 lymphocytes for 72 hr. Results shown are mean [³H]TdR incorporation ±1 SD for triplicate cultures. (d) Class I MHC expression on DC was assessed by flow cytometry in the presence or absence of brefeldin A. (**P < 0·01 versus OVA ISCOMS pulse only).

First we examined the effects of inhibiting the cytosolic class I MHC processing pathway. The presentation of OVA in ISCOMS to OVA-specific CD8⁺ T cells by resting DC required trafficking of class I MHC from the endoplasmic reticulum, as it was completely inhibited by brefeldin A (Fig. 4b). This was not the result of a non-specific effect on total class I MHC levels, as the presentation of OVA 257–264 peptide was significantly enhanced under these conditions (Fig. 4c) and surface class I MHC expression was unaltered (Fig. 4d). TAP-mediated transport into the endoplasmic reticulum was also required, as TAP^–/– DC were completely unable to present OVA in ISCOMS to OVA-specific CD8⁺ T cells (Fig. 5a). Interestingly, LPS-induced maturation of TAP^–/– DC revealed a TAP-independent mechanism of class I MHC loading, suggesting that DC maturation may promote an alternative mechanism of class I MHC loading (Fig. 5a). This TAP-independent mechanism was antigen specific, as LPS stimulation of TAP^–/– DC loaded with DT ISCOMS promoted much reduced stimulation of OVA-specific CD8⁺ T cells (Fig. 5b). The presentation of OVA in ISCOMS was not affected by the specific inhibitor of proteasome function, lactacystin (Fig. 6a). However, lactacystin significantly inhibited the presentation of endogenously expressed OVA in EG7.OVA cells (Fig. 6b), while having no effect on the presentation of OVA peptide (Fig. 6a) or on total class I MHC expression (Fig. 6c). Importantly, B6 DC that were fixed before the ISCOMS pulse failed to stimulate OVA-specific CD8⁺ T cells, but retained their ability to present OVA peptide (Fig. 6d), indicating that uptake of free or regurgitated peptide was not responsible for the stimulation of CD8⁺ T cells by ISCOMS-associated OVA.

Figure 5.

Role of TAP in presentation of ISCOMS-associated antigen to CD8⁺ T cells by bone marrow DC. Wild-type and TAP^–/– DC were untreated, pulsed with OVA ISCOMS for 2 hr, or stimulated with LPS for 24 hr following the antigen pulse period, washed and co-cultured with OT-1 lymphocytes for 72 hr. (b) TAP^–/– DC were untreated, pulsed with OVA ISCOMS or DT ISCOMS for 2 hr, or stimulated with LPS for 24 hr following the antigen pulse period, washed and co-cultured with OT-1 lymphocytes for 72 hr. Results shown are mean [³H]TdR incorporation ±1 SD for triplicate cultures. (**P < 0·01 versus wild-type DC, *P < 0·01 versus DT ISCOMS plus LPS stimulation).

Figure 6.

Effects of lactacystin on presentation of ISCOMS OVA to CD8⁺ T cells. (a) DC were treated with 20 μm lactacystin for 30 min before and throughout the 2 hr pulse with OVA ISCOMS or 10 pg/ml OVA peptide, before being fixed, washed and co-cultured with OT-1 lymphocytes for 72 hr. (b) EG7 OVA cells were untreated, or treated with 20 μm lactacystin for 2 or 4 hr before being fixed, washed and co-cultured with OT-1 lymphocytes for 72 hr. (c) Class I MHC expression on DC was assessed by flow cytometry in the presence or absence of lactacystin. (d) DC were fixed and then incubated with OVA ISCOMS or 10 μg/ml OVA peptide, washed and co-cultured with OT-1 lymphocytes for 72 hr. Results shown are mean [³H]TdR incorporation ±1 SD for triplicate cultures. (**P < 0·01 versus no lactacystin).

Because proteasomal processing did not appear to be necessary by fixed DC for the generation of class I MHC-restricted peptides from OVA ISCOMS, we examined alternative pathways by which these peptides might arise. The presentation of OVA in ISCOMS by DC to OT-1 lymphocytes was completely inhibited by chloroquine (Fig. 7a) which prevents the acidification of the endocytic vesicles normally associated with class II MHC-restricted processing. Chloroquine had no effect on the presentation of OVA 257–264 or class I MHC expression (Fig. 7b,c). Actin-dependent phagocytosis played some role in allowing ISCOMS to gain access to processing compartments within DC, as treatment of DC with cytochalasin D partially inhibited the presentation of OVA in ISCOMS (data not shown). As ISCOMS themselves are too small to be taken up by classical phagocytosis, we considered the possibility that cross-presentation of cell-associated antigen could be occurring. To examine this idea, β₂-microglobulin^–/– DC, which cannot themselves present OVA to OT-1 cells (Fig. 7d), were pulsed with OVA ISCOMS or the OVA peptide 257–264, washed and then co-cultured overnight with wild-type B6 DC, which were then added to OT-1 lymphocytes. OVA-specific CD8⁺ T-cell proliferation was observed when peptide loaded β₂-microglobulin^–/– DC were added to the B6 DC, indicating that the assay was capable of detecting cross-priming (Fig. 7d). However, no cross-presentation was seen when OVA ISCOMS-pulsed β₂-microglobulin^–/– DC were used to load B6 DC (Fig. 7d). Thus, the cytochalasin D-dependent processing of ISCOMS does not involve cross-presentation of ISCOMS-loaded DC by new DC.

Figure 7.

Effects of chloroquine and the role of cross-priming on presentation of ISCOMS OVA to CD8⁺ T cells. (a) DC were treated with 0·1 mm chloroquine for 30 min before and throughout the 2 hr pulse with OVA ISCOMS or (b) 10 pg/ml OVA peptide, before being fixed, washed and co-cultured with OT-1 lymphocytes for 72 hr. (c) Class I MHC expression on DC was assessed by flow cytometry in the presence or absence of chloroquine. (d) β₂-microglobulin^–/– DC were pulsed with OVA ISCOMS or OVA 257–264 and then co-cultured with B6 DC for 24 hr, before secondary co-culture with OT-1 lymphocytes for 48 hr. OVA ISCOMS pulsed β₂-microglobulin^–/– or B6 DC were co-cultured directly with OT-1 lymphocytes as negative and positive controls, respectively. Results shown are mean [³H]TdR incorporation ±1 SD for triplicate cultures. (**P < 0·01 versus OVA ISCOMS pulse only, *P < 0·01 versus background).

Role of CD4⁺ T cells and CD40 in the presentation of OVA ISCOMS to CD8⁺ T cells by DC

In view of the unusual intracellular processes which appeared to be involved in the priming of OVA-specific CD8⁺ T cells by ISCOMS-pulsed DC, we thought it important to study some of the cellular events involved in this presentation, including the role of CD4⁺ T-cell help and the engagement of CD40 on DC by CD40L, which are important in other models of CD8⁺ T-cell priming.^15,16

First, we compared B6 and class II^–/– DC for their ability to present OVA in ISCOMS to OT-1 lymphocytes. As shown in Fig. 8(a), freshly isolated class II^–/– DC were fully able to stimulate OVA-specific CD8⁺ T cells after pulsing with OVA ISCOMS. After activation with LPS, these DC also showed enhanced presentation of OVA in ISCOMS to the same extent as found with B6 DC (Fig. 8b). OVA ISCOMS pulsed class II^–/– DC were also able to stimulate highly purified CD8⁺ T cells (data not shown), excluding the possibility that residual class II⁺ APC or CD4⁺ T cells present in the responding population of OT-1 lymphocytes may have contributed to this response.

Figure 8.

Role of CD4⁺ T cells and CD40–CD40L in presentation of ISCOMS-associated antigen to CD8⁺ T cells by bone marrow DC. (a,b) Class II^–/– and B6 DC were untreated, pulsed with OVA ISCOMS for 2 hr, or stimulated with LPS for 24 hr following the antigen pulse and co-cultured with OT-1 lymphocytes for 48 hr. (c) B6 and CD40^–/– DC were treated as above and co-cultured with OT-1 lymphocytes for 48 hr. Results shown are mean [³H]TdR incorporation ±1 SD for triplicate cultures. (**P < 0·01 versus LPS-activated CD40^–/– DC).

Next we investigated the role of CD40. Resting CD40^–/– DC presented OVA in ISCOMS to OT-1 lymphocytes to the same degree as B6 DC (Fig. 8c). However, the CD40^–/– DC could not be activated by LPS to give the enhanced presentation of OVA in ISCOMS seen using wild-type B6 DC (Fig. 8c). Together, these results show that class II MHC-restricted CD4⁺ T-cell help mediated by CD40–CD40L is not required for the activation of OVA-specific CD8⁺ T cells by OVA ISCOMS-pulsed resting DC. However, CD40–CD40L is necessary for the enhanced presentation by mature DC.

Discussion

The current results show that bone marrow DC pulsed with antigen in ISCOMS efficiently prime resting CD8⁺ T cells in vitro by a mechanism that appears to involve acidic endosomal vesicles as well as cytosolic processing, and that this is enhanced by LPS. Our findings that DC could present ISCOMS-associated antigen to CD8⁺ T cells extends previous work showing that ISCOMS prime class I MHC-restricted cytotoxic T lymphocytes in vivo,^5,17–19 as well as our more recent evidence that these responses are greatly enhanced by expanding DC numbers with the cytokine flt3 ligand (Beacock-Sharp et al. manuscript submitted). DC have been shown to prime CD8⁺ T cells to other exogenous antigens in vivo and in vitro,²⁰ but relatively little is known about the processing and presentation mechanisms involved. Here we show that OVA ISCOMS-pulsed DC grown from bone marrow can prime resting OVA-specific CD8⁺ T cells efficiently in the absence of additional maturational signals to the DC. This was not the result of contaminating peptides as there was no presentation of OVA ISCOMS by DC which had been fixed before the ISCOMS pulse. Importantly, soluble OVA was not presented under the same circumstances and the maximum response obtained with OVA ISCOMS was equivalent to that found using saturating amounts of epitope peptide. This emphasizes the unusual ability of ISCOMS to allow relatively small amounts of native antigen to gain access to the class I processing pathway.

The presentation of OVA ISCOMS by DC was inhibited by chloroquine, indicating that acidic endosomes are required for generation of the peptide epitopes which are to be loaded on to class I MHC molecules. This is consistent with the ability of ISCOMS to prime class II-restricted CD4⁺ T cells in vivo^8,21,22 and with the fact that other exogenously administered antigens can be presented to class I MHC-restricted T cells after egress from endocytic or phagocytic compartments.^23–26 However, the activation of CD8⁺ T cells also required access to the conventional site of class I MHC loading in the endoplasmic reticulum, as it was inhibited by brefeldin A and dependent on functional TAP molecules.

Despite apparently entering the classical class I MHC pathway in the endoplasmic reticulum, the presentation of ISCOMS OVA to CD8⁺ T cells was not inhibited by lactacystin-treated DC that had been fixed, indicating that proteasomal degradation of the antigenic peptides themselves was not required. These findings are paradoxical in view of the requirement for other components of the class I cytosolic pathway of processing and we propose the following hypothesis to explain this apparent contradiction. The effects of brefeldin A and TAP knockout show a need for continuous export of TAP-loaded class I MHC from the endoplasmic reticulum for the presentation of OVA peptides. However, these class I MHC molecules may not necessarily be loaded with OVA peptides when leaving the endoplasmic reticulum, but could be loaded with endogenous self peptides on their way to the cell surface. Our hypothesis is that the OVA peptides gain access to these endogenously loaded class I MHC molecules in endocytic compartments into which surface class I molecules are being continuously recycled. This is consistent with complete inhibition of processing by chloroquine and the partial effects of cytochalasin. In addition, lactacystin may have had no effect on presentation by fixed DC as the brief period of inhibition we used would be unlikely to have reduced the existing and relatively stable pool of endogenous peptides which would then be sufficient to sustain class I synthesis during the study. Interestingly, we have found that lactacystin markedly inhibits the class I MHC restricted presentation of ISCOMS if the DC are not fixed after treatment. This would be consistent with eventual depletion of the pool of endogenous self peptides required for maintenance of sufficient class I MHC molecules to be available for loading after recycling, as our idea predicts.

The enhanced presentation of OVA ISCOMS to CD8⁺ T cells by LPS-treated DC is consistent with other recent work showing the effects of DC maturation on CD8⁺ T cells²⁷ and with the well-known ability of LPS to stimulate class II MHC-restricted presentation by DC. This may partly reflect its effects on the increased expression of co-stimulatory molecules by DC, although it should be noted that the co-stimulatory requirements for CD8⁺ T cells are not fully known. However, we have also extended these findings by showing that the maturational effects of LPS on class I presentation were independent of CD4⁺ T cells, but were CD40 dependent and that the maturation of DC revealed a TAP-independent pathway of processing. Thus, the maturation of DC not only enhances the endoplasmic reticulum-dependent pathway of class I MHC presentation through increased co-stimulation, but also allows DC to develop an alternative, non-TAP-dependent mechanism for the generation of class I epitopes. Another study has shown that class I MHC-restricted epitopes of the measles virus F protein can be presented in a TAP-independent and ammonium chloride-sensitive manner by Epstein–Barr virus-transformed B lymphoblastoid cell lines.²⁸ As we have postulated above, this appears to reflect the ability of a fraction of internalized cell surface class I molecules to be internalized and intersect the acidic MHC class II compartments, where the recycled MHC class I complexes can release their peptides, allowing the binding of new peptide epitopes in transit to the cell surface. Together, our results suggest that DC are capable of utilizing this pathway and that it is enhanced when DC are activated by agents such as LPS. The reasons for this are uncertain but could involve increased membrane turnover and/or the induction of the acidic proteases involved in processing exogenous antigen and peptide exchange by internalized class I MHC.²⁹ The potency of this maturation effect is underlined by the fact that it was partly TAP independent, indicating that in the presence of LPS, sufficient class I MHC molecules may be released from the endoplasmic reticulum in these cells to allow effective endocytic loading of class I MHC after recycling from the cell membrane. Alternatively, LPS may allow access of peptides directly to the endoplasmic reticulum by some novel, but unknown mechanism.

The presentation of ISCOMS OVA to CD8⁺ T cells by DC did not require help from CD4⁺ T cells, as priming by class II MHC^–/– DC was entirely normal, even when highly purified CD8⁺ T lymphocytes depleted of endogenous APC were used. Previous studies have shown that some, but not all CD8⁺ T-cell responses in vivo are dependent on CD4⁺ T cells, especially under conditions which directly induce increased expression of CD40 by DC.¹⁵ In our hands, the priming of OT-1 lymphocytes by immature DC was independent of CD40 expression by the DC, indicating that these APC may express other, as yet unidentified, accessory molecules required for the priming of CD8⁺ T cells. In addition our results suggest that the ISCOMS-primed CD8⁺ T cells can produce sufficient interleukin-2 to sustain their own clonal expansion. In contrast, the maturation effects of LPS on DC priming of CD8⁺ T cells were completely dependent on CD40, confirming the role of this molecule in the APC function of activated DC. However, these effects were also independent of class II MHC, indicating that the CD40L necessary for ligating CD40 on the DC must be present on the CD8⁺ T cell itself.³⁰ Overall, these results are consistent with our recent findings that adoptively transferred OT-1 cells can be primed normally by OVA ISCOMS in vivo in the absence of CD4⁺ T cells (Beacock-Sharp et al. manuscript submitted) suggesting that similar, direct activation of CD8⁺ T cells by ISCOMS-loaded DC may occur in vivo. However, it is important to note that both these studies have used populations of responding CD8⁺ T cells containing a high frequency of antigen-specific T cells. Under these circumstances, CD8⁺ T cells may be able to produce sufficient interleukin-2 and co-stimulatory molecules in an autocrine manner, which may not occur with the lower frequencies seen in the wild-type immune system.³¹ These issues need to be addressed directly in vivo.

In summary, our results indicate that DC are extremely potent at processing and presenting ISCOMS-associated antigen to CD8⁺ T cells. Together with the in vivo evidence that DC are central to the effects of ISCOMS, we propose that targeting DC may be a major mechanism underlying the adjuvant activity of ISCOMS on cytotoxic T-lymphocyte priming. Our results also indicate that ISCOMS may allow exogenous antigens to be processed in an alternative, endosomal pathway of class I MHC antigen processing. By investigating the cellular basis of these processes, it may be feasible to target ISCOMS better to DC in vivo and hence improve further the usefulness of ISCOMS as potential vaccine vectors for protection against viral and related infections.