Efficient Presentation of Naturally Processed HLA Class I Peptides by Artificial Antigen-Presenting Cells for the Generation of Effective Antitumor Responses

Tumor-reactive lymphocytes can target and kill human cancer cells, and infiltration of tumor masses by lymphocytes has been correlated with improved survival in patients with certain types of cancer (1–6). Furthermore, it has also been shown that, at least in some diseases, the presence of intratumoral CD8⁺ T cells most strongly correlates with survival (3, 4). These CD8⁺ T lymphocytes are thought to have tumor-specific cytotoxicity by recognizing tumor-associated antigen (TAA)–derived peptides that are processed and presented by HLA class I molecules on tumor cells (7).

In recent years, several TAA and TAA-derived peptides have been identified (8), and a variety of immunotherapeutic clinical trials using vaccine or adoptive cell transfer strategies have been conducted to determine the optimal method of inducing clinically relevant immune responses to these antigens (9–11). Although immune responses can be detected in many of these trials, improvement in the magnitude of clinical antitumor activity is needed (11). For example, vaccination with synthetic peptides has sometimes induced ineffective CTL responses that do not persist or localize to tumor sites. These CTL are often of low avidity and are not able to recognize tumor cells. Furthermore, the CTL generated by immune responses in some trials are specific for cryptic epitopes and not for tumor cell targets. In fact, in some studies, responses to cryptic epitopes were dominant, and little or no response to the natural epitope was detected (12, 13).

Recently, however, significant tumor regressions by the transfer of highly avid antitumor lymphocytes have been shown in heavily pretreated patients with metastatic melanoma (14, 15). Immunotherapy by adoptive cell transfer is based on the ex vivo expansion and reinfusion of TAA-specific T cells to tumor-bearing patients (16). Dudley and Rosenberg have done adoptive cell transfer therapy in melanoma patients where tumor-infiltrating lymphocytes from surgical specimens are expanded ex vivo with general T-cell stimulatory signals and then reinfused into tumor-bearing patients. Although this method has the advantage of expanding T cells that are primed in vivo by professional antigen-presenting cells (APC) endogenously presenting TAA, tumor-infiltrating lymphocytes are not readily available or expandable in many patients and/or tumor types.

Alternatively, patient T cells can be stimulated by APC that present a particular TAA-derived peptide. Compared with ex vivo activated tumor-infiltrating lymphocytes, where antigen specificity is not controlled, this strategy can generate peptide-specific T cells by stimulating with APC that present a particular antigen. To generate an effective T-cell response, however, peptide must be appropriately presented in the context of a given HLA molecule in a manner that is identical to that found on tumor cells. This does not always occur, particularly when APC are exogenously pulsed with synthetic peptide, because pulsed peptide can generate cryptic epitopes that may induce ineffective immune responses (17). This failure is partly due to the formation of complexes between pulsed synthetic peptides and HLA molecules that have different conformations compared with that formed naturally by intracellular peptides processed and loaded onto HLA molecules (17). In addition, cryptic T-cell epitopes may be formed through the alteration of exogenously added peptides by cleavage with endopeptidases and exopeptidases that can be derived from the APC or found in culture medium (18, 19).

Several APC, such as autologous dendritic cells, CD40 B cells, EBV-LCL, and several artificial APC (aAPC), have been shown to be capable of stimulating and expanding functional antigen-specific CD8⁺ T cells (20–26). Typically, these APC are pulsed with synthetic peptides when used as stimulators. An alternative strategy that is designed to generate effective antigen-specific T-cell responses is to stimulate T cells using APC that not only express the optimal combination of immunoaccessory molecules but also endogenously expressed antigens. In this case, antigenic peptides are endogenously processed and presented, forming HLA/peptide complexes that are presumably identical to those present on tumors. Recently, we reported that our aAPC, which expresses HLA-A2, CD80, and CD83, is able to support the priming and prolonged expansion of peptide-specific CD8⁺ cytotoxic T cells (27). In this article, we report on additional modifications of our aAPC so that it can naturally process virtually any antigenic peptide of choice and present that peptide to CD8⁺ T cells via transduced A2 molecules. Furthermore, we show that these aAPC can be used to generate CTL specific for the melanoma-associated antigen, MART1, and that these CTL can efficiently recognize and kill tumor cell targets in a HLA-A2-restricted way. Because the induction of appropriate immune responses that can recognize tumor cells is essential, our aAPC may serve as a versatile tool in the generation of anticancer immunity.

cDNAs. Enhanced green fluorescent protein (EGFP)-mini-MP1 cDNA was generated by fusing EGFP cDNA (Clontech, Palo Alto, CA) to the DNA sequence of ⁵⁵LTKGILGFVFTL⁶⁶, which is derived from the matrix antigen of influenza virus, preceded by an intervening “LTK” sequence. EGFP-mini-MART1 cDNA was produced by replacing the mini-MP1 cDNA sequence, ⁵⁸GILGFVFTL⁶⁶, with the mini-MART1 sequence, ²⁷AAGIGILTV³⁵. Puromycin N-acetyltransferase (pac)-mini-MART1 (pac MART1) cDNA was produced by replacing the EGFP moiety of mini-MART1 with the pac cDNA. All the constructs were verified by DNA sequencing.

Cells. Peripheral blood samples from normal donors were collected in compliance with protocols approved by the institutional review board of the Dana-Farber Cancer Institute. aAPC were generated by transducing K562 cells (American Type Culture Collection, Manassas, VA) with HLA-A*0201, CD80, and CD83 as described previously (27). cDNAs in an expression vector, pMX (a gift from Dr. T. Kitamura, University of Tokyo, Tokyo, Japan), were transfected into a 293GPG packaging cell line, and replication-defective virus supernatants were harvested. Each infection was done under conditions so that at least 10⁵ independent transduced cells were obtained. Positive cells were collected by flow cytometry–guided sorting or selected by puromycin (2.5 μg/mL; InvivoGen, San Diego, CA). Where indicated, aAPC were treated with the indicated concentration of lactacystin or 100 ng/mL IFN-γ.

Extraction of cell surface peptides. Peptides were isolated from the cell surface as described elsewhere with the following modifications (28). aAPC-derived cells (200-500 million) were washed twice with ice-cold PBS or 10 mmol/L Tris (pH 7.4) supplemented with 0.9% NaCl. Cell surface peptides were extracted by exposing cells for 5 minutes to citrate buffer [0.06 mol/L Na₂HPO₄, 0.13 mol/L citric acid (pH 3.0)]. Extracts were spun and the peptide containing supernatant was membrane filtered with a size cutoff of 5 kDa and frozen until further analysis.

High-performance liquid chromatography separation and matrix-assisted laser desorption ionization-time of flight mass spectrometry of peptides. Filtered peptide extracts or 100 μg synthetic peptide were separated by reverse-phase high-performance liquid chromatography (HPLC) on a C18 column (Source 5RPC St 4.6/150, Amersham Pharmacia, Piscataway, NJ). The peptides were separated using an acetonitrile gradient (2-100%). The specific HPLC fractions corresponding to MP1 and MART1 peptides were collected, lyophilized, and resuspended in 0.1% trifluoroacetic acid. Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MS) analysis was done on a Voyager Biospectrometry Workstation (Perseptive Biosystems, Framingham, MA) in the positive, delayed reflector mode. All mass spectra were collected with the acquisition range (0.8-3 kDa) following the same instrumental variables. All mass spectra were acquired by 100 shots evenly distributed on each spot. The next-well external calibration was used to calibrate each sample plate and sample preparation. The peptide sequences were confirmed using electrospray ionization ion trap liquid chromatography/tandem MS (Finnigan LCQDeca, Thermo Electron, Waltham, MA).

ELISPOT analysis. ELISPOT analysis for IFN-γ was done as described previously (27). Where indicated, aAPC derivatives were treated with 10 μmol/L lactacystin (Calbiochem, La Jolla, CA) or β-lactone (Calbiochem) for 4 hours and then peptide stripped with citrate buffer. Stripped cells were further treated with 10 μmol/L of a proteasome inhibitor for 10 hours and then used as stimulators in the presence of the inhibitor. Control cells were similarly treated with MeSO₄ (Sigma, St. Louis, MO).

Production of HLA class I peptide-specific CD8⁺ T cells. Peptide-specific cytotoxic CD8⁺ T-cell lines were generated as described previously (27). Where indicated, aAPC were pulsed with graded concentrations (0.1, 1, and 10 μg/mL) of synthetic peptide (⁵⁸GILGFVFTL⁶⁶ of the influenza virus matrix antigen or ²⁷AAGIGILTV³⁵ of MART1; New England Peptides, Fitchburg, MA) for 6 hours at room temperature. aAPC were then irradiated, washed, and added to the CD8⁺ responder cells at 20:1. On the next day (day 1) and on days 4, 7, and 10, interleukin (IL)-2 (10 IU/mL; Chiron, Emeryville, CA) and IL-15 (10 ng/mL; R&D, Minneapolis, MN and Peprotech, Rocky Hill, NJ) were added to the cultures. Where indicated, T-cell cultures were repeatedly stimulated on a weekly basis and supplemented with IL-2 between the stimulations.

Tetramer analysis and flow cytometric analysis. Tetramer analysis was done as described previously (27). EGFP⁺ aAPC-derived cells were subjected to flow cytometric analysis using the FL1 channel.

Western blot analysis. Western blot analysis was done as described previously (29). Anti-EGFP antibody (Clontech) and antiproteasome subunit antibodies (Affiniti Research Products, Devon, United Kingdom) were used to detect specific proteins.

aAPC/mini-MP1 can naturally process and present MP1^58-66 peptide derived from transduced mini-MP1 gene. K562 lacks the expression of any HLA class I or II molecules on the cell surface (30) but highly expresses the adhesion molecules CD54 (intercellular adhesion molecule-1) and CD58 (LFA-3). aAPC were generated by stably transfecting K562 with HLA-A2, CD80, and CD83. We have shown previously that peptide-pulsed aAPC can prime and expand antigen-specific CD8⁺ T cells from HLA-A2⁺ healthy donors (27). We hypothesized that aAPC can endogenously process and present HLA-A2-restricted peptides and can induce functional CD8⁺ T-cell immunity. To test this, aAPC were transduced with an EGFP-mini-MP1 fusion gene to generate aAPC/mini-MP1, which consists of EGFP joined to an antigenic peptide at the COOH terminus via a LTK linker sequence (Fig. 1A). aAPC/mini-MP1 and control aAPC-expressing EGFP were established by flow cytometry–guided sorting (Fig. 1B).

We examined whether MP1 peptide is naturally processed and presented by HLA-A2 molecules on the surface of aAPC/mini-MP1. Freshly isolated CD8⁺ T cells from antigen-experienced A2⁺ healthy donors were stimulated by either aAPC/EGFP or aAPC/mini-MP1, and an IFN-γ ELISPOT was done. As shown in Fig. 1C, aAPC/mini-MP1 induced IFN-γ secretion to the same extent as aAPC/EGFP pulsed with MP1^58-66 peptide, although aAPC/EGFP did not induce any IFN-γ secretion. IFN-γ secretion was identical when T cells were stimulated with aAPC/mini-MP1 alone or with aAPC/mini-MP1 pulsed with either MP1^58-66 or irrelevant HIV-pol peptide. These results suggest that aAPC/mini-MP1 are able to naturally process and present MP1^58-66 peptides derived from transduced mini-MP1 gene and that the antigenic density of the MP1^58-66 peptide in A2 groove is sufficient to provoke effector function of MP1^58-66-specific CD8⁺ T cells.

MP1^58-66 peptide is predominantly processed by the proteasome of aAPC/mini-MP1. The COOH terminus of class I–restricted peptides is believed to be generated solely by proteasome machinery (31–33). However, it is yet to be determined to what extent nonproteasomal components participate in producing the NH₂ terminus of the class I–restricted peptides in the endoplasmic reticulum or the cytosol (34). To address this issue, aAPC were pretreated with the proteasome inhibitor lactacystin and used as stimulators in an IFN-γ ELISPOT. As shown in Fig. 1D, endogenous processing and presentation of MP1^58-66 peptides were completely abrogated by treating aAPC/mini-MP1 with lactacystin, suggesting that the NH₂ terminus of mini-MP1 was mainly processed by the proteasome under the conditions examined. The fact that the exogenous addition of MP1^58-66 but not HIV-pol peptides successfully restored the induction of IFN-γ secretion suggests that the expression of A2 molecules were not substantially affected by lactacystin. Another proteasome inhibitor β-lactone also showed inhibition of MP1^58-66 peptide processing to the same extent as lactacystin (data not shown). These results indicate that, in aAPC, EGFP-mini-MP1 is predominantly processed by proteasome machinery and not by nonproteasome enzymes.

EGFP-mini-MP1 protein is ubiquitinated and degraded by the proteasome in aAPC. Most proteins that are degraded by the proteasome are ubiquitinated and marked by polyubiquitin chains (35, 36). Therefore, treatment of cells with a proteasome inhibitor should lead to the accumulation of ubiquitinated proteins. aAPC/EGFP and aAPC/mini-MP1 were treated with or without lactacystin for 24 hours and then subjected to Western blot analysis. EGFP-specific antibody detected a ladder pattern of polyubiquitinated EGFP in both lactacystin-treated and untreated aAPC/EGFP, reflecting the fact that EGFP is a very stable protein with a half-life of ∼24 hours. On the other hand, ubiquitinated EGFP mini-MP1 was only detectable in lactacystin-treated cells but not in untreated aAPC/mini-MP1, suggesting that EGFP-mini-MP1 is less stable and more rapidly degraded by the proteasome (Fig. 1E). Next, we quantitatively measured the effects of a proteasome inhibitor on the expression level of EGFP-mini-MP1 protein. aAPC/mini-MP1 was treated with graded concentrations of lactacystin and then subjected to flow cytometric analysis. As shown in Fig. 1F, lactacystin increased the expression of EGFP-mini-MP1 protein in a dose-dependent manner, supporting that the EGFP-mini-MP1 fusion protein is degraded via the proteasome in aAPC/mini-MP1.

aAPC expresses intact proteasome machinery and can be induced to express the immunoproteasome. aAPC were cultured in the presence or absence of IFN-γ, a potent inducer of the immunoproteasome, and total cell lysates were prepared. Western blot analysis revealed that aAPC constitutively express all conventional proteasome subunits examined and, following exposure to IFN-γ, up-regulate the expression of the immunoproteasome subunits (Fig. 2).

Acid elution and biochemical identification of MP1^58-66 peptide directly from the HLA-A2 groove on aAPC/mini-MP1. We biochemically confirmed the presence of MP1 peptide in the HLA-A2 groove on the surface of aAPC/mini-MP1. aAPC/mini-MP1 cells were washed and then acid treated to elute cell surface peptides and proteins. Peptide fractions were collected as a flow-through following filtration through a membrane with a 5-kDa cutoff. They were further separated by reverse-phase HPLC, and the peptide fraction corresponding to synthetic MP1^58-66 peptide was collected and subjected to MS. As shown in Fig. 3, MP1^58-66 peptide was identified in fraction 15. Peptide sequencing of the purified peptide completely matched MP1^58-66 (data not shown). Because HLA-A2 is the only HLA expressed by aAPC, these results provide further support that mini-MP1 gene is transcribed, translated, processed, and presented in the groove of the A2 molecule on the surface of aAPC/mini-MP1.

Determination of the immunogenicity of MP1^58-66 peptides endogenously processed and presented by aAPC. Using an immunologic method, we next evaluated the density of MP1^58-66 peptides naturally processed and presented by HLA-A2 on aAPC/mini-MP1. HLA-A2⁺ CD8⁺ T cells from normal healthy donors were stimulated once by either aAPC/mini-MP1 or parental aAPC exogenously pulsed with graded concentrations of synthetic MP1^58-66 peptide. T-cell cultures were supplemented with IL-2 and IL-15 every 3 days. After 2 weeks, MP1^58-66 peptide specificity of generated CTL was examined by MP1 tetramer staining. The comparison of tetramer staining positivity of CD8⁺ T cells stimulated by aAPC/mini-MP1 or peptide-pulsed aAPC revealed that the tetramer positivity obtained with aAPC endogenously processing and presenting MP1^58-66 corresponded to that obtained with aAPC exogenously pulsed with >10 μg/mL MP1^58-66 peptide (Fig. 4A).

MP1-specific CD8⁺ T cells generated by aAPC/mini-MP1 show potent effector function. HLA-A2⁺ CD8⁺ T cells from healthy donors were stimulated with irradiated unpulsed aAPC/mini-MP1 at weekly intervals and cultures were supplemented with IL-2 twice weekly. At the indicated time, peptide specificity of generated T cells was determined by tetramer analysis. After five stimulations, 50% of the T-cell population was MP1^58-66 tetramer positive. These CD8⁺ T cells showed potent effector function in an antigen-specific manner by the cytotoxicity assay using MP1^58-66 peptide-pulsed T2 cells as a target. They also exhibited antigen-specific secretion of IFN-γ by the ELISPOT assay (Fig. 4B and C; data not shown).

MART1-specific CD8⁺ T cells generated by aAPC/mini-MART1. To extend this strategy to a TAA, we generated aAPC/mini-MART1 by using the mini-gene, EGFP-mini-MART1 (amino acids 27-35; Fig. 5A and B). CD8⁺ T cells from A2⁺ healthy donors were stimulated at weekly intervals by either aAPC/mini-MART1 or parental aAPC exogenously pulsed with graded concentrations of synthetic MART1^27-35 peptide. Comparison of MART1 tetramer staining suggested that MART1^27-35 peptide-specific immunogenicity of aAPC/mini-MART1 was more potent than that of aAPC pulsed with 10 μg/mL MART1^27-35 peptide (Fig. 5C). HPLC and MS analysis revealed the presence of the MART1^27-35 peptide in eluted material obtained from acid stripping of aAPC/mini-MART1 (data not shown). These results suggest that regardless of which A2-restricted peptide is fused to EGFP via the three–amino acid LTK linker, efficient processing and presentation of peptide by aAPC occurs at a density greater than 10 μg/mL. A2⁺ CD8⁺ T cells from healthy donors stimulated by irradiated unpulsed aAPC/mini-MART1 on a weekly basis generated MART1^27-35-specific CTL lines with positive tetramer staining and potent antigen-specific effector function as measured by cytotoxicity (Fig. 5D and E).

LTK linker enables the processing of the downstream A2-restricted peptide regardless of the upstream moiety. We next studied whether the LTK sequence can serve as the universal linker that promotes processing of the COOH-terminal peptide sequence regardless of the NH₂-terminal sequence. We generated aAPC expressing a puromycin resistance gene fused to the MART1 9-mer sequence via the intervening LTK linker, aAPC/pac MART1. CD8⁺ T cells purified from A2⁺ normal donors were stimulated by aAPC/pac MART1 as described above. Between the stimulations, T-cell cultures were supplemented with IL-2 and IL-15. As shown in Fig. 6A, the MART1 CTL line underwent considerable expansion even with the low concentrations of IL-2 and IL-15, and after four stimulations, the total number of CD8⁺ T cells increased by >70-fold. The tetramer staining analysis revealed 7.81% MART1 peptide specificity. When aAPC/pac (control) was used as a stimulator, no MART1 tetramer positivity was observed (data not shown). Considering that the tetramer positivity of MART1 before stimulation was 0.08% (data not shown), the expansion rate of MART1-specific T cells was >5,000 times. To date, we have successfully generated a total of six MART1 CTL lines from healthy donors using aAPC/pac MART1 as a stimulator, and the expansion rate of MART1-specific CTL has been >2,000 times for every donor (data not shown). These results suggest that aAPC/pac MART1 was capable of endogenously processing and presenting MART1 peptide derived from the transduced pac MART1 fusion gene and that the LTK linker enables the processing of the downstream A2-restricted peptide regardless of the upstream moiety.

To examine the cytotoxicity of generated MART1 CTL, peptide-pulsed T2 cells and melanoma cell lines were used as targets. MART1 peptide-pulsed T2 cells but not TAX peptide-pulsed T2 cells were specifically killed by the MART1 CTL stimulated by aAPC/pac MART1 (Fig. 6B). Furthermore, A2⁺ MART1⁺ melanoma cells (Malme-3M) but not A2⁺ MART1⁻ melanoma cells (A375) were effectively killed by the CTL (Fig. 6C), suggesting that the generated MART1 CTL line had sufficiently high T-cell receptor avidity to recognize melanoma tumor cells. Blocking experiments using anti-A2 monoclonal antibody further confirmed that the observed killing by the CTL was restricted by HLA-A2 (Fig. 6D). IFN-γ ELISPOT analysis revealed that the MART1 CTL could secrete IFN-γ when stimulated by MART1 peptide-pulsed T2 cells or Malme-3M but not A375 (data not shown). These results show that MART1 CTL generated by aAPC/pac MART1 were able to recognize MART1⁺ tumor cells and possessed potent effector functions.

Using both immunologic and biochemical methods, we have shown that our aAPC can naturally process and present a HLA-A2-restricted peptide derived from a transduced mini-gene. The peptide density and the expression level of HLA-A2 on the cell surface of aAPC were sufficient not only to support the expansion of memory T cells specific for a recall response but also to deliver a strong antigen-specific stimulus that supported the priming of naive T cells. aAPC constitutively expressed the proteasome subunits tested and also up-regulated the expression of immunoproteasome subunits on IFN-γ treatment.

EGFP-mini-gene, which is composed of EGFP fused with a three–amino acid LTK linker and HLA-A2-restricted peptide at the COOH terminus, was efficiently translated and predominantly degraded by the proteasome, which resulted in efficient peptide presentation via HLA-A2 on the aAPC cell surface. We have found that in addition to the human erythroleukemia cell line, K562/HLA-A2, several other cell lines, such as COS/HLA-A2 (African green monkey kidney fibroblast), 293/HLA-A2 (human kidney fibroblast), U2-OS (human osteosarcoma cell), Jurkat/HLA-A2 (human T-cell leukemia cell), and DU145/HLA-A2 (human prostate cancer cell), were all able to process and present MP1^58-66 peptides following transduction with EGFP-mini-MP1 (data not shown). In addition, preliminary experiments have revealed that, in addition to MP1^58-66 and MART1^27-35, the EGFP fusion construct was applicable to every HLA-A2-restricted peptide tested, including ³⁷³ILKEPVHGV³⁸¹ of HIV-pol, ¹⁹⁰FLDPRPLTV¹⁹⁸ and ²³⁹SLVDVMPWL²⁴⁸ of CYP1B1, ⁵⁴⁰ILAKFLHWL⁵⁴⁸ of hTERT, and ³⁶⁹KIFGSLAFL³⁷⁷ of HER-2/neu (8). Furthermore, the pac-mini-MART1 gene, which is composed of a puromycin resistance gene fused to the MART1 9-mer peptide sequence at the COOH terminus via the LTK linker, was also processed and presented via HLA-A2 on the aAPC cell surface. Therefore, our results strongly suggest that these mini-gene constructs can be used with a variety of host cells, antigens, and selection strategies. This versatility could be unique to this particular mini-gene construct with the LTK linker sequence, because it has been reported that different tissues have different processing machinery and because there is substantial evidence that nonproteasomal machinery is required for the cleavage of the NH₂ terminus of class I peptides (34).

Our system also offers the opportunity to test multiple HLA subtypes. We have initially focused on HLA-A*0201 because it is one of the most common HLA alleles and because many HLA-A2-restricted peptides have been identified previously. Given that the parental cell line, K562, is easily transfected, HLA-A2 can be substituted with any HLA class I subtype. In fact, we already established aAPC-expressing HLA-A*0101, HLA-A*2402, or HLA-A*2601, and the characterization of these aAPC is ongoing.

Several cell lines based on K562 have been developed and used for the stimulation of T cells. The K562-based APC by Thomas et al., which is loaded with anti-CD3 and CD28 antibodies, was shown to expand CD4⁺ T cells nonspecifically (37). Likewise, the K562-based APC described by Maus et al. was used to confer nonspecific expansion signals to preselected CD8⁺ T cells. In their system, CD8⁺ T cells were initially primed by dendritic cells and then selected for antigen specificity by tetramer staining (38). APC, such as those described by Britten et al., are engineered to express HLA-A2, but no additional costimulatory molecules, and served as a target of effector T cells (39). As we have shown recently, however, these APC are not able to efficiently prime or expand antigen-specific CD8⁺ T cells (27). Because our aAPC can prime CD8⁺ T cells without the need for other cells or reagents, our system may bring advancement to the generation of an immune response in vitro.

Using a relatively simple and direct method of acid stripping (28), column filtration, and HPLC separation followed by MS analysis, we were able to biochemically confirm the presence of MP1^58-66 and MART1^27-35 peptides on the cell surface of aAPC/mini-MP1 and aAPC/mini-MART1, respectively. Because aAPC expresses only HLA-A2 and is relatively resistant to acid treatment, our method can isolate HLA-bound peptides without the generation of large-scale cell lysates with detergent followed by immunoprecipitation using specific antibody before acid stripping of peptides (40, 41). There is an increased probability that peptides identified by our method are truly presented on the cell surface, because it avoids the procedures that may increase the possibility of contaminating peptide pools with intracellular “junk” peptides. We are currently investigating whether aAPC can naturally process and present an A2-restricted peptide derived from full-length genes as well as mini-gene constructs. If this is the case, it is possible that this versatile aAPC-based methodology could be used to identify unknown A2-restricted peptides from novel TAA.

Although aAPC constitutively expresses the conventional proteasome, it can also up-regulate the expression of the immunoproteasome in response to IFN-γ. Because it has been shown that peptides produced by the conventional proteasome and the immunoproteasome differ not only in quantity but also in quality (42–44), it is plausible that immunoproteasome-specific peptides could be isolated using our aAPC treated with IFN-γ. Our aAPC might serve as a unique platform to perform a comparative analysis and isolate naturally presented class I–restricted peptides endogenously processed by either the constitutively expressed proteasome or the immunoproteasome.

For the purpose of T-cell-mediated immunotherapy for cancer, cytotoxic T cells must be able to recognize TAA that are naturally presented by tumors. Exogenously pulsed peptides occasionally elicit ineffective immune responses due to a variety of mechanisms (17–19, 45, 46), resulting in the induction of CTL responses with heterogeneous functional avidity and poor antitumor responses (47). Using our aAPC transduced with mini-gene constructs, the immune response is generated with a stimulus that may more closely correspond to the tumor cell because the HLA/peptide complexes of the aAPC are formed through endogenous processing. Because aAPC can be manufactured under current good manufacturing practice conditions we conclude that aAPC could serve as an alternative “off-the-shelf” APC that is able to constitutively present class I–restricted TAA peptides and has the potential to induce clinically relevant T-cell responses.

We thank G. Dranoff, R. Mulligan, and T. Kitamura for providing the valuable reagents and J. Daley and S. Lazo-Kallanian for excellent technical help.