Extension of HLA-A*0201-Restricted Minimal Epitope by N^ε-Palmitoyl-lysine Increases the Life Span of Functional Presentation to Cytotoxic T Cells¹

The efficiency of vaccines against many intracellular infectious agents requires the stimulation of an immune response resulting in the induction of CTL that recognize specific peptide epitopes in the context of MHC class I molecules at the surface of infected cells. MHC molecules bind, with high affinity, peptides that conform to particular structural parameters in terms of sequence motifs and length; minimal cytotoxic epitopes are usually 8–11 residues long and are mostly produced by the cytoplasmic endogenous processing pathway for protein degradation and presentation (1). In addition, selectivity based on size and on some sequence characteristics has been demonstrated in the activity of TAPs (2, 3), the endoplasmic reticulum chaperone gp96, or cytoplasmic heat shock proteins HSP70 and HSP90 (4, 5, 6). Efficient T cell responses have been shown to correlate with the affinity of binding to the relevant MHC molecules (7, 8) or the stability of the corresponding MHC-peptide complex (9, 10, 11) and the frequency of serial TCR engagement with the MHC-peptide complex (12, 13).

Intervention strategies aimed at inducing CTLs have been elaborated in two opposite directions. The first one is based on the use of small peptides, administered in water-in-oil formulation; peptides of 8–11 residues, corresponding to the “optimal” ligands of the target MHC class I receptor (14) are presumably able to associate directly with some MHC molecules on the cell surface without needing further intracellular processing (15). An alternate approach is based on the expression or introduction of peptide or protein Ags into the cytoplasm of APCs; CTL induction depends on the actual delivery of the Ag into the cytoplasmic processing pathway, where antigenic peptides are generated by proteolytic degradation (1, 16, 17, 18).

Among different vehicles that have been designed for Ag delivery into the cytoplasm, palmitoyl-modified peptides (lipopeptides) were able to induce CTL responses in vivo in mice (19, 20). High CTL responses compatible with protection were induced by immunizing primates and humans with lipopeptides derived from SIV or HBV sequences (21, 22, 23, 24). Recently, we and others have reported (25, 26, 27, 28, 29) that soluble palmitoyl-modified peptides (1000–5000 Da) were able to target and modulate the activity of cytoplasmic enzymes (protein kinase C-α, -ε or -ζ) or receptors (intracytoplasmic domain of integrins or IFN-γ receptor) in intact nonphagocytic cells. The palmitoyl group allowed a rapid and passive membrane translocation of the associated peptide, leading to differential patterns of intracellular distribution, depending on the localization of the pharmacological targets. We have hypothesized that antigenic lipopeptides comprising 9–12 residues obtained by the introduction of a single N^ε-palmitoyl-lysine residue in one of the extremities could favor internalization and cytoplasm delivery in APC, thereby functioning as immediate precursors of the epitope.

As a model peptide, we have selected the HLA-A*0201-restricted HIV-1 polymerase (pol)⁴ conserved epitope 476–484 (30). This choice was governed by the frequency of CTL responses to this epitope among HIV-1-positive individuals and by the availability of structural studies of five peptides bound to HLA-A*0201 allowing a rational choice of peptide modifications that would presumably be compatible with MHC binding (31, 32). Moreover, previous studies had demonstrated the efficiency of peptide recognition by CTL after the replacement of the Ile in position P1 by the relatively bulky Tyr. This position is buried within the MHC class I binding cleft, allowing alteration of the N-terminal side chain without changing the interaction with the TCR (33). We have focused on the P0, P10 (external to the binding cleft), or P1 position of the parent pol_476–484 peptide, in which we introduced a lysine modified or not by N^ε-palmitoylation. To allow direct microscopic observation of the binding of the pol_476–484-derived peptide or lipopeptides, we also synthesized a series of fluorescent probes in which the His in position P7 was replaced by a rhodamine-labeled residue.

The peptide and lipopeptides were built up by solid phase synthesis using the fluorenylmethoxycarbonyl-t-butyl strategy (34). The side chains of the lysine residues to be modified by palmitoylation or selective fluorescent labeling were protected with a 4-methyltrityl group (Novabiochem, Läufelfingen, Switzerland), allowing selective deprotection by 0.5% TFA in dichloromethane, followed by on-resin acylation with palmitic acid or 5(6)-carboxytetramethylrhodamine (Fluka, Buchs, Switzerland). After deprotection and cleavage, the peptide and lipopeptides were purified by reverse phase HPLC on a Nucleosil C₁₈ column. All peptides and lipopeptides were >95% pure. Identity was confirmed by determination of amino acid composition after total acid hydrolysis, and molecular mass determination was recorded on a Bio-Ion 20 plasma desorption mass spectrometer (Bio-Ion, Uppsala, Sweden). Sequences and molecular mass of all peptides and lipopeptides are indicated in Table I. For cell surface stabilization assay, peptides or lipopeptides were dissolved at 10 mM in 10% DMSO-water and added to culture medium at a final concentration of 10–300 μM per well. The concentration of DMSO in the cell suspension never exceeded 0.3–0.4%.

T2 is a somatic cell hybrid of human B and T lymphoblastoid cell lines deficient in TAP proteins and expressing low levels of surface HLA-A*0201 and undetectable levels of HLA-A5 (35). T2 cells were maintained in RPMI 1640 (Life Technologies, Courbevoie, France) supplemented with 10% FCS (Life Technologies), penicillin (100 IU/ml), streptomycin (100 μg/ml), l-glutamine (2 mM), sodium pyruvate (1 mM), and 2-ME (20 μM; Sigma, St. Quentin Fallavier, France). T1 (wild-type counterpart of T2 cell line) and Jurkat cells (T lymphoblastoïd, HLA type: A9, A25, B7, B41) were grown in the same medium without 2-ME.

HIV RT pol_476–484-specific CTLs lines were generated from PBMC of HIV-seropositive HLA-A*0201-positive individuals donors Z1 and Z55 from cohort studies etablished with the approval of the ethics committee of the Cochin Hospital) as previously described (36). Briefly, cells were cultured in RPMI 1640-Glutamax (Life Technologies), containing antibiotics and sodium pyruvate as above and 5% heat-inactivated human AB serum (Valbiotech, Paris, France). They were stimulated weekly with HLA-A2-matched B lymphoblastoid cell lines preincubated with 1 μM parent pol_476–484 peptide and irradiated. They were maintained between 0.7 and 1 × 10⁶/ml, fed with IL-2 (10 U/ml; Boehringer, Meylan, France) twice a week, and used after 2 or 3 weeks of culture.

T2 cells were suspended in complete medium supplemented with different concentrations of peptide or lipopeptide. The cells were incubated overnight at 26°C in the presence of the test peptide or lipopeptide. Then, after 6 h of incubation at 37°C, cells were stained for conformationally correct HLA-A*0201 with the monoclonal mouse Ab (mAb) BB7.2 (10 μg/ml), followed by FITC-labeled goat anti-mouse IgG Ab (Immunotech, Marseille, France). Ten thousand cells were analyzed on a flow cytometer (FACscan, EPICS II; Coulter, Margency, France). The mean fluorescence (MF) is the mean channel number of 10,000 gated cells. The fluorescence index (FI) was calculated as follows: FI = (MF − MF₀)/MF₀, where MF₀ corresponds to the mean fluorescence in the absence of exogenous peptide. Peptides or lipopeptides with a FI ≥0.5 were considered to bind to HLA-A*0201 molecules. By interpolation, we determined the peptide concentration that resulted in 50% of the maximal up-regulation of HLA-A*0201 expression in the presence of the known HLA-A*0201 binding peptide and its derivatives.

T2 cells (5 × 10⁴) were incubated with 100 μM peptide or lipopeptide as described for the cell surface stabilization assay. After overnight incubation at 26°C followed by extensive washes, cells were incubated in RPMI medium containing 4% FCS and brefeldin A (BFA) (Sigma) at 10 μg/ml to block egress of new class I molecules. After 1 h incubation at 37°C in the presence of BFA, the block of Golgi to cell surface egress was maintained in medium containing BFA at 0.5 μg/ml. All Abs and wash solutions contained BFA at 0.5 μg/ml. The same experiments were repeated without BFA in 4% FCS. At the indicated times, aliquots were stained with the BB7.2 mouse Ab and analyzed by flow cytometry. The half-life of HLA/peptide complexes was calculated as the time required for 50% of the molecules to decay.

Confocal epifluorescence microscopy was performed with a Zeiss LSM 410 inverted confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany) and a 100×/1.3 oil objective. The confocal pinhole was kept small to obtain thin optical slices. As exciting light source, a helium-neon laser was used at 543 nm. The emitted light was collected through a 575–640 nm bandpass filter (Carl Zeiss). Cells (T1, T2, or Jurkat) were incubated with either peptides or lipopeptides, at 26°C or 37°C for 1 h. Excess ligand was removed by a single wash in medium, and cells were then mounted on a microscope slide and covered.

T1 cells incubated with soluble peptides at 1 μM in complete RPMI and then washed were used as stimulating cells in the ELISPOT assay. For kinetics studies, T1 cells were fixed with 4% paraformaldehyde for 20 min at 37°C and washed with 0.1 M glycine after the peptide pulse. CTL lines at serial dilutions (1–20 × 10³ cells/well) were seeded overnight in triplicate with 10⁴ stimulating cells/well on 96-well nitrocellulose plates (MultiScreen HA, Millipore, Bedford, MA) that had been coated with 2 μg/ml capture mouse anti-human IFN-γ mAb (Genzyme, Russelheim, Germany) in 0.1 M carbonate buffer, pH 9.6. The IFN-γ ELISPOT assay was performed as previously described (37). Spots were visualized with a stereomicroscope (Leica MZ6, Heerbrugg, Switzerland; magnification, ×40).

Responses were considered significant if 1) a minimum of five spot-forming cells (SFCs) were present per well, 2) this number was at least 2-fold that obtained with the negative control at the cell concentration used, and 3) the same result was obtained using at least two different effector cell numbers. Negative control cells pulsed with the HLA-A*0201-restricted peptide HTLV-1 Tax 11-19 did not elicit a specific response from the effector cells. The positive control consisted of 500 effector cells plated with 50 ng/ml PMA and 500 ng/ml ionomycin. Peptide-responding cells were exclusively CD8 positive, i.e., class I restricted, as assessed by double labeling by flow cytometry (not shown).

The influence of the modification of P0, P1, or P10 of the parent peptide (termed I-9-V) on peptide-HLA-A*0201 interaction was analyzed by a binding assay on TAP-deficient human T2 cell line. This assay is based on the fact that in T2 cells, a large proportion of MHC class I molecules are devoid of endogenous peptides and are unstable at 37°C. This can be prevented by the addition of exogenous specific peptides that bind within the MHC class I Ag-presenting groove, stabilizing the three-dimensional structure of the complex (33, 38). T2 cell surface HLA-A*0201 levels are measured by Ab staining and can be quantified by flow cytometry (Fig. 1). Fig. 1,A shows a flow cytometer profile which represents the increase of the mean fluorescence of T2 cells incubated with different concentrations (0, 10, 100, and 300 μM) of the I-9-V model peptide. Similar features were obtained with T2 cells incubated with the peptides or lipopeptides evaluated. The results are presented in Fig. 1, B and C. N-terminal acetylation (Ac-K(Pam)I-9-V), or P10 modification (I-9-V K(Pam)) totally prevented the formation of class I complexes of the attached nonapeptide. Stabilization of MHC class I complexes was observed with the parent peptide, with the K(Ac)L-8-V peptide or the K(Pam)L-8-V lipopeptides, and was 17 times stronger with the K(Pam)I-9-V lipopeptide. This stabilization assay is more stringent and might be a more accurate reflection of the in situ process in comparison with competition assays (38).

The association of the lipopeptides with the HLA-A*0201 molecules was assessed by confocal microscopy, using fluorescent analogues of the peptide or lipopeptides. Direct comparison of the fluorescent labeling after a 1-h incubation of human cells expressing or not HLA-A*0201 molecules is shown in Fig. 2. Surface labeling was observed after incubation of T2 cell with the fluorescent I-9-V[K*₄₈₂] model peptide (Fig. 2,A) or the K(Pam)I-9-V[K*₄₈₂] lipopeptide (Fig. 2,B). No surface labeling was obtained after T2 cells were incubated with the I-9-V K(Pam)[K*₄₈₂] lipopeptide, whereas an intracellular staining was detected (Fig. 2,C), showing that the association of the model peptide and the P0 modified lipopeptide on the cell surface was not due to rhodamine. The absence of nonspecific, rhodamine-mediated interactions with the cell membrane or its proteic components was assessed by the specific inhibition of this phenomenon by a 100-fold molar excess of nonlabeled lipopeptides (Fig. 2,D). Furthermore, only an intracellular labeling was observed when Jurkat cells (which do not express HLA-A*0201) were incubated under the same conditions with the fluorescent lipopeptides (Fig. 2,E). Down-regulation of the surface expression of empty receptors by incubation of the T2 cells at 37°C permitted both intracellular delivery of the fluorescent lipopeptide and a surface labeling (Fig. 2 F). Thus, the lipopeptides are able both to associate directly with HLA-A*0201 and to enter the cytoplasm of intact cells.

The kinetics of complex dissociation was evaluated by measuring the time necessary for the complex to disappear from the cell surface. The peptide-pulsed T2 cells were carefully washed before being immunolabeled with a conformation-specific mAb (39), and the experiment was performed in the presence of BFA, which blocks the exocytosis pathway (40) (Fig. 3 A). Under these conditions, the quantification of conformationally correct HLA-A*0201 could not be influenced by the reloading of intra- or extracellular peptides onto newly egressing molecules during the progress of the experiment. The time of dissociation of the complex formed with the K(Pam)L-8-V lipopeptide was relatively short, with a half-life of ∼5 h, whereas the half-life of the complex formed with the P0-modified lipopeptide K(Pam)I-9-V was found to be equivalent to the value observed with the parent peptide (about 12 h).

To take into account the eventual egress of intracellularly formed HLA-peptide complexes emerging from the exocytosis pathway, or a possible recycling process (41), the experiments were repeated in the absence of BFA (Fig. 3 B).

Our observations revealed a marked increase in the persistence of the antigenic complexes formed with both lipopeptides. The kinetics of dissociation of the complexes formed with the lipopeptides in the absence of BFA were significantly different for the two lipopeptides. The half-life of the HLA-A*0201/peptide complexes (Fig. 3,C) was approximately doubled after cell incubation with K(Pam)I-9-V and was extended over the limits of the experiment with K(Pam)L-8-V; 49 and 100% of the initial immunolabeling level was still present after 32 h of cell culture following incubation with K(Pam)I-9-V and K(Pam)L-8-V, respectively, whereas the antigenic complexes formed after incubation with the nonlipidic peptides had totally disappeared. Also, complex stability was maintained for ∼5 h with all peptides and lipopeptides before it decreased (Fig. 3 B). This indicated a role for internalization and recycling mechanisms, which are inhibited by BFA, in this stabilization.

Therefore, the experiments performed with the lipopeptides of this study analyzed not only the ligand-receptor dissociation but also its superposition with other exocytosis-dependent processes. This suggested that the “minimal” lipopeptide of this study gained access into the MHC-loading compartment without the help of the TAP transport systems.

On the basis of our binding and stability analyses, we selected three P0- and P1-modified analogues of the parent I-9-V peptide for functional analysis. A sensitive ELISPOT assay (42) for single IFN-γ-secreting cells was used to compare the CTL recognition of the different constructs (Fig. 4). CTLs from donors Z55 and Z1 recognized equally the parent peptide or the P0-modified lipopeptide (K(Pam)I-9-V). The shorter K(Ac)L-8-V peptide analogue was also recognized by CTLs from one donor (Fig. 4, ∗), but not the other. The K(Pam)L-8-V lipopeptide analogue was not recognized by the cells from either of the two donors.

This suggested that the K(Pam)I-9-V lipopeptide could sensitize target cells for epitope-specific CTL recognition, either as a direct ligand of HLA-A*0201 or as a prodrug able to generate in situ the parent nonapeptide. To detect only direct binding of the peptide analogue on the cell surface, the experiment was repeated with the use of T1 cells fixed with PFA before incubation with the peptide or lipopeptide, to detect only direct binding of the peptide analogue on the cell surface (data not shown). In this case once again, the complexes formed after cellular incubation with the K(Pam)I-9-V lipopeptide or the parent peptide were similarly recognized. Therefore, P1 substitution of the parent peptide inhibited peptide recognition at the cell surface, whereas P0 substitution did not, and the latter lipopeptide was functionally recognized at a level similar to that of the parent peptide.

T1 cells were incubated with the parent peptide or its K(Pam)I-9-V analogue for 2 to 48 h before fixation and contact with CD8⁺ T cells specific for I-9-V (Fig. 5). Under these conditions, no or very few IFN-γ-secreting cells could be detected using T1 cells incubated for >24 h with the parent I-9-V peptide, attesting its complete degradation. On the contrary, CTL recognition of the lipopeptide was maintained for as long as 48 h. This suggests that N-terminal modification of the parent peptide by a N^ε-palmitoyl-lysine increases the life span of cell sensitization for recognition by CD8⁺ T cells specific for the parent peptide.

Several attempts have been proposed using minimal cytotoxic epitopes to induce CTLs. In terms of pharmaceutical development, this approach would be relatively simple, given that the synthesis, handling, and characterizations of nonamer peptides are usually straightforward. However, this approach revealed the risk of an inadequate setting of the balance between T cell stimulation and tolerization; tolerance induction was observed after immunization with optimal peptide ligands (43) and was associated with rapid diffusion of the peptides from the s.c. depot. With the exception of protocols based on ex vivo-generated dendritic cells used as vaccine vehicles (44), the risk of tolerization seems inherent to any optimal-peptide-based formulations and demonstrates the importance of the mode of Ag delivery.

The use of lipidated peptides, initially proposed by the groups of Jung and Rammensee (19), represents an alternative that is gaining interest. The first lipopeptide vaccine able to induce CTL response associated an efficient immunostimulating lipotripeptide (tripalmitoyl-S-glyceryl-cysteinyl-seryl-serine) to an epitope peptide. The biological activity of this lipopeptide could result from different mechanisms: stimulation of particular populations of immunocompetent cells, a depot effect and/or the formation of aggregates that would be efficiently captured by phagocytic or micropinocytic cells with an inherent good Ag-presenting capacity. In this case, the hydrophobicity of the lipidic component would favor escape from the endosome-lysosome compartments, or even transmembrane delivery into the cytoplasmic processing pathway. The use of palmitoyl peptide vaccines for primate (21, 45, 46) or human (22, 23) vaccination afforded very encouraging results but revealed difficulties during the scale-up of the synthesis, purifications, and analytical characterizations (47) of long lipopeptides, justifying the search for simpler structures.

To date, minimal sized lipopeptides used for CTL induction have been described by Diamond et al. (48), who used nonamer peptides modified by a palmitoyl-lysyl-seryl-seryl sequence. A palmitoyl chain was introduced on the N^α extremity, eventually associated to a second palmitoyl chain introduced on the lysine side chain. Under these experimental conditions, the structural criteria for direct MHC class I binding such as length and presence of an N-terminal ionizable end group were not satisfied (1).

As a first step toward the definition of minimal lipopeptides for the modulation of CTL activities, our goal was to delineate the minimal requirements for efficient delivery of a functional cytotoxic epitope into APC. In former experiments, we have observed the ability of different lipid-associated functional cargoes to accumulate at the sites of their respective intracellular pharmacological targets into living cells (49), suggesting that ligand-receptor equilibrium could drive their cytoplasmic diffusion. In this context, we hypothesized that short lipidated epitopes might be delivered toward cytosolic processing and TAP transport, or even enter passively the endoplasmic reticulum, where they would find the shelter of chaperone proteins before being further trimmed to the size that efficiently binds to MHC class I.

Our observations suggest that two palmitoylated derivatives of a minimal cytotoxic epitope HIV-1 pol_476–484 for HLA-A*0201 molecules possessing ionizable extremities fulfill both the size and charge criteria for MHC class I loading. Both compounds are able to stabilize the expression of peptide-HLA complexes with apparent efficacies equal or superior to that of the complexes formed with the parent peptide. The direct surface association of the lipopeptide, which clearly depends on the presence of the appropriate HLA molecule, was observed only under artificial conditions that favored the surface expression of “empty” molecules. When assaying cells that do not express HLA-A*0201, or when using temperature conditions that down-regulate the surface expression of the MHC, the principal observation was the intracellular delivery of the lipopeptides, followed by the formation of intracellular stores in as yet unidentified sites.

The presence of the lipid tail modifies considerably the apparent life span of expression of the HLA-peptide complexes formed after incubation of the TAP-deficient T2 cells with the lipopeptides. This phenomenon depends on a compensatory mechanism of the dissociating complexes that would have been indistinguishable from a kinetic stabilization of the complexes without the help of an artificial interruption of the exocytosis pathway by BFA. This dynamic system might depend either on a recycling process and/or on the egress of newly formed, intracellularly loaded HLA-peptide complexes.

Whatever the mechanism involved, this process is faster than the complex dissociation; this is particularly striking for the nonamer K(Pam)L-8-V, which was found to be able to up-regulate a maximal expression of conformationally correct HLA/peptide complexes, despite their low intrinsic stability. This observation also argues in favor of an intrinsic higher metabolic stability of the lipopeptide or influence of additional chaperone molecule into special storage vesicles that protect the construct from proteolytic degradation. The situation is probably different for the decamer lipopeptide: 49% of the complexes were still present after 32 h (instead of the 100% observed with the K(Pam)L-8-V), suggesting a mechanism of compensation (recycling or new class I complexes synthesis) much slower than observed with K(Pam)L-8-V, and/or a progressive metabolic degradation of the lipopeptide.

A 5-h delay in HLA-A*0201 dissociation was observed in the absence of BFA for all peptides and lipopeptides, as if they had all entered into the cells; however, the peptides without a lipid tail were not found inside the cells, at least not in amounts measurable by confocal microscopy. This delay may be explained by internalization and the recycling of peptides bound to class I molecules (rather intracellularly for lipopeptides), during overnight incubation before the chase.

Functional TCR-MHC/peptide recognition was a key parameter of the evaluation of these “minimal” lipopeptide and was determined using pol_476–484-specific CTL lines obtained from naturally infected HLA-A*0201 individuals. Despite the known compatibility of the immunogenicity of the pol_476–484 derived peptides with replacement of the Ile in P1 position by other amino acids (33), the K(Ac)L-8-V peptide was recognized by only one of the two donors, perhaps because of different T cell repertoires, because this was also found with unstimulated PBMCs as well as with CTL lines (data not shown). Also, no agonist activity was found with the complexes derived from the nonamer K(Pam)L-8-V lipopeptide, despite its strong ability to bind HLA-A*0201 molecules intracellularly; the lipid moiety was probably unprocessed and inhibited T cell binding.

Conversely, both CTL lines were activated to produce IFN-γ in response to the stimulation by the complexes formed after incubation of T1 cells with the P0-modified decamer lipopeptide. In the first hour of the experiment, the frequency of IFN-γ-producing cells was similar to that found when tested against target cells pulsed with the parent peptide or with K(Pam)I-9-V peptide. A clear difference was noted after longer periods of incubation. Although after 48 h of incubation the agonist HLA/peptide complexes had almost disappeared from the surface of cells incubated with the parent peptide, functional presentation persisted for at least 48 h with cells sensitized with K(Pam)I-9-V. This prolonged functional presentation is consistent with the prolonged stabilization of the expression of HLA-A*0201 molecules at the surface of T2 cells, as previously described (11). These data suggest that the ligand had been stored in the cell and had been protected from complete inactivating proteolytic degradation, at least for the duration of the experiment. This might involve the contribution of a chaperone protein, such as gp96, which has been reported to shuttle efficiently the associated antigenic material into the class I Ag presentation pathway (50).

Additional experiments will be required to study in detail not only the subcellular distribution of the lipopeptides and its evolution with time but also the quantitative and qualitative consequences of functional presentation of such minimal lipopeptide-derived HLA/peptide complexes. Another question would be to test the applicability of this approach to other epitopes, recognized in the context of other HLA class I molecules. The use of such minimal lipopeptides in the formulation of immunizing mixtures might be an interesting alternative to the use of minimal cytotoxic epitopes (51), because the presence of the lipid tail could not only affect their presentation but also limit diffusion of the vaccine preparation away from the s.c. site and favor physical associations with other components of the mixture, such as T-helper lipopeptides or immunoadjuvants.

We thank Dr. J. G. Guillet for important initial and ongoing suggestions and for reviewing the manuscript; Dr. B. Georges for interesting discussion and help; Dr. S. Brooks for proofreading the manuscript; and E. Legue, J. F. Desoutter, and Dr. M. Dallod for their help.