Expression Hierarchy of T Cell Epitopes from Melanoma Differentiation Antigens: Unexpected High Level Presentation of Tyrosinase-HLA-A2 Complexes Revealed by Peptide-Specific, MHC-Restricted, TCR-Like Antibodies¹

Akey advance in tumor immunology in the past decade has been the elucidation of the antigenic basis of tumor cell recognition and destruction. The ultimate effector cells that mediate the immune activity against tumors are the CTLs, which recognize through their clonotypic and specific TCR peptide fragments bound within the major clefts of MHC class I molecules on the cell surface. Tumor Ags have been conceptually grouped into categories based on the genes that encode them. Some tumor Ags, such as an epitope from a mutated β-catenin gene (1), arise de novo and are unique to individual cancer cells. Other tumor Ags, such as NY-ESO-1, are derived from the aberrant expression of nonmutated genes, the products of which are normally expressed only in testes or fetal tissues (2, 3). Another very important group of tumor Ags, such as Mart-1, gp100, and tyrosinase, which are expressed by melanoma cells, are derived from nonmutated, cell lineage-specific proteins also termed differentiation Ags (4). However, the existence of tumor-reactive CD8⁺ T cells in the peripheral circulation of a patient or an experimental animal is not sufficient to cause the rejection of an established tumor. Clinical trials using peptide Ags for vaccination succeeded in routinely generating tumor-reactive CTLs in patients (5, 6), but vaccination alone only sporadically induced tumor regression in patients with metastatic disease. Even in transgenic mice that were engineered to enable every T cell to express a tumor-reactive TCR, tumors still grew progressively.

The lack of inflammatory rejection of tumors by immunized patients and TCR-transgenic mice is not well understood at the cellular and molecular level. Many mechanisms could account for the failure of Ag-specific CD8⁺ T cells to eliminate Ag-expressing tumor cells in vivo. For instance, the tumor Ag-specific T cells themselves could be functionally deficient, rendered anergic or unable to fully differentiate in the tumor environment (7). The tumor environment could lack a “danger signal” or other innate immune stimulation, preventing a general inflammatory reaction from evolving (8, 9). Alternatively, active immune regulatory mechanisms such as CD4⁺CD25⁺ T cells might impede any endogenous immune reaction to cancer cells (10, 11). Whatever the mechanism(s), without an inflammatory immune response, the CD8⁺ T cells of the adaptive immune system are rendered ineffective. As a tumor grows and metastasizes, additional systemic immune suppression could develop, and Ag-escape variants of the tumor could arise (9). This bleak picture of tumor immune evasion has engendered many strategies for immunotherapeutic intervention.

All immunotherapeutic approaches are designed to induce and enhance T cell reactivity against tumor Ags. Intensive research on cancer peptides has culminated in many clinical trials involving therapeutic vaccination of cancer patients with antigenic peptides or proteins (12, 13, 14). However, without knowledge regarding the mechanisms of Ag presentation, these studies are difficult in interpretation. Moreover, several studies demonstrated that the inability of the patient’s immune system to elicit an effective immune response against the tumor is often due to poor Ag presentation (15).

The advent of MHC-peptide tetramers has provided a new tool for studying Ag-specific T cell populations in health and disease, even when they are very rare, by monitoring tetramer-T cell binding via flow cytometry (16, 17, 18). However, to date there are very few tools available to detect, visualize, count, and study Ag (MHC-peptide) presentation. Abs with TCR-like specificity could enable measuring the Ag presentation capabilities of such tumor or APCs, e.g., by direct visualization of the specific MHC-peptide complex on the cell surface. Therefore, TCR-like Abs would serve as a valuable tool to obtain precise information about the presence, expression pattern, and distribution of the target tumor Ag, i.e., the MHC-peptide complex, on the tumor cell surface, on tumor metastases, in lymphoid organs, and on professional APCs. Abs that specifically recognize class I MHC-peptide complexes have already been used in murine systems to study Ag presentation, to localize and quantify APCs displaying a T cell epitope, or as a targeting tool (19).

In this study, for the first time, we have used phage-derived human TCR-like Abs to study the expression hierarchy of HLA-A2-peptide complexes (T cell epitopes) derived from the three major differentiation Ags of human melanoma, gp100, Mart-1, and tyrosinase. We found that the presentation of these T cell epitopes does not correlate with their gene expression profile and surprisingly we show that HLA-A2-peptide complexes derived from tyrosinase are expressed and presented at unexpectedly high numbers on the surface of melanoma cells in culture as well as on melanoma tissue frozen sections. Evidence for the molecular mechanism responsible for this unexpectedly high presentation suggests that protein stability plays a major role in the high efficiency presentation of tyrosinase-derived epitope in comparison to the other T cell epitopes derived from gp100 and Mart-1.

Single-chain MHC (scMHC)³-peptide complexes were produced by in vitro refolding of inclusion bodies produced in Escherichia coli upon isopropyl β-d-thiogalactoside (IPTG) induction, as described previously (20). Briefly, a scMHC, which contains the β₂-microglobulin and the extracellular domains of the HLA-A2 gene connected to each other by a flexible linker, was engineered to contain the BirA recognition sequence for site-specific biotinylation at the C terminus (scMHC-BirA). In vitro refolding was performed in the presence of peptides as described. Correctly folded MHC-peptide complexes were isolated and purified by anion exchange Q-Sepharose chromatography (Pharmacia), followed by site-specific biotinylation using the BirA enzyme (Avidity).

Selection of phage Abs on biotinylated complexes was performed as described recently (21). Briefly, a large human Fab library containing 3.7 × 10¹⁰ different Fab clones was used for the selection. Phages were first preincubated with streptavidin-coated paramagnetic beads (200 μl; Dynal) to deplete the streptavidin binders. The remaining phages were subsequently used for panning with decreasing amounts of biotinylated scMHC-peptide complexes. The streptavidin-depleted library was incubated in solution with soluble biotinylated scHLA-A2-tyrosinase complexes (500 nM for the first round, and 100 nM for the following rounds) for 30 min at room temperature. Streptavidin-coated magnetic beads (200 μl for the first round of selection and 100 μl for the following rounds) were added to the mixture and incubated for 10–15 min at room temperature. The beads were washed extensively 12 times with PBS/0.1% Tween 20 and an additional two washes were with PBS. Bound phages were eluted with triethylamine (100 mM, 5 min at room temperature), followed by neutralization with Tris-HCl (1 M, pH 7.4), and used to infect E. coli TG1 cells (OD = 0.5) for 30 min at 37°C. The diversity of the selected Abs was determined by DNA fingerprinting using a restriction endonuclease (BstNI), which is a frequent cutter of Ab V gene sequences.

Fab Abs were expressed and purified, as described recently. TG1 or BL21 cells were grown to OD₆₀₀ = 0.8–1.0 and induced to express the recombinant Fab Ab by the addition of IPTG for 3–4 h at 30°C. Periplasmic content was released using the B-PER solution (Pierce), which was applied onto a prewashed TALON column (Clontech). Bound Fabs were eluted using 0.5 ml of 100 mM imidazole in PBS. The eluted Fabs were dialyzed twice against PBS (overnight, 4°C) to remove residual imidazole. Specificity of the produced Fabs was verified by ELISA analysis as previously described (21).

EBV-transformed B lymphoblast JY cells were washed with serum-free RPMI 1640 medium and incubated overnight with medium containing 10 μM tyrosinase_369–377 YMDGTMSQV peptide or control peptides: TyrN₃₆₉ (YMNGTMSQV), gp100₂₀₉ (ITDQVPFSV), gp100_209-2M s(IMDQVPFSV), gp100₁₅₄ (KTWGQYWQV), gp100₂₈₀ (YLEPGPVTA), MART-1_{27 A27L}: (LAGIGILTV), HIV: Gag₇₇ (SLYNTVATL), HTLV-1 TAX₁₁ (LLFGYPVYV), hTERT₅₄₀ (ILAKFLHWL), and hTERT₈₆₅ (RLVDDFLLV). Cells (10⁶) were incubated with 1–2 μg of specific Ab for 1 h at 4°C, followed by incubation with PE-labeled anti-human Ab for 45 min at 4°C. Cells were finally washed and analyzed by a FACStar flow cytometer (BD Biosciences). Melanoma cells were examined for endogenous Ag expression by staining with 2–5 μg of specific Ab.

mRNA was isolated from melanoma cell lines with oligo(dT) magnetic beads by using a Dynabeads mRNA DIRECT Kit (Dynal) according to the manufacturer’s instructions. The mRNA was converted into cDNA by using cloned AMV reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. Real-time PCR was performed by using Assays-on- Demand Gene Expression Assays (Applied Biosystems). Assay identification nos. Hs00165976_m1, Hs00173854_m1, and Hs00194133_m1 were used for tyrosinase, gp100, and Mart-1 expression assays, respectively. For all real-time PCR, the total volume of 20 μl contained 15 ng of mRNA template from which cDNA had been reverse transcribed and 10 μl of TaqMan Universal PCR Master Mix (Applied Biosystems). Primers and FAM-labeled probes were added to each reaction at the final concentration of 0.9 and 0.25 μM, respectively. A no-template control that contained all of the above reagents was also included to detect the presence of contaminating DNA. The cDNA reverse transcribed from 5 ng of GAPDH mRNA was used as an internal control gene for mRNA expression and for analyzing relative expression (Assay identification no. Hs99999905_m1). Amplification and fluorescence detection was conducted in an Applied Biosystems Prism 7700 Sequence Detector with a program of 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 1 min.

The amounts of target genes were determined from the comparative threshold cycle (C_T) method. The target genes were normalized to GAPDH and were expressed as 2^−ΔCT(ΔC_T of target gene − C_T of GAPDH).

The cDNA sequences of the L and H chains of the Fab were PCR amplified and cloned separately into pET-based expression vectors. The C terminus of the Fab L chain was fused to the BirA tag for site-specific biotinylation. Each of the vectors was transformed into E. coli BL21 cells and expressed upon IPTG induction as inclusion bodies. The inclusion bodies, containing L or H chains of the Fab, were isolated and solubilized, and the L and H chains were refolded with each other and purified by ion exchange chromatography. The recombinant Fab was biotinylated and tetramers were generated by adding fluorescently labeled streptavidin (22).

The H and L Fab genes were cloned for expression as human IgG1 κ Ab into the eukaryotic expression vector pCMV/myc/ER. For the H chain, the multiple cloning site, the myc epitope tag, and the endoplasmic reticulum (ER) retention signal of pCMV/myc/ER were replaced by a cloning site containing recognition sites for BssHI and NheI followed by the human IgG1 constant H chain region cDNA isolated by RT-PCR from human lymphocyte total RNA. A similar construct was generated for the L chain. Each shuttle expression vector carries a different antibiotic resistance gene. Expression was facilitated by cotransfection of the two constructs into the human embryonic kidney HEK293 cell by using the FuGENE 6 Transfection Reagent (Roche).

After cotransfection, cells were grown on selective medium. Clones that reacted specifically with JY cells pulsed with tyrosinase 369–377 peptide were adapted to growth in 0.5% serum and were further purified using protein A affinity chromatography. SDS-PAGE analysis of the purified protein revealed homogenous, pure IgG with the expected molecular mass of ∼150 kDa.

Cells were fixed with 0.1% formaldehyde for 10 min at room temperature, rinsed with 0.1% BSA-PBS, incubated with primary Ab at 4°C for 1 h, followed by incubation with the fluorescence-labeled secondary Ab goat anti-human Alexa Fluor 488 or goat anti-mouse Alexa Fluor 594 (Molecular Probes). For staining of nuclear DNA, DRAQ5 (Alexis Biochemicals) was used. A Bio-Rad MRC1024 confocal microscope was used for analysis.

A frozen melanoma tissue with HLA-type A0201 and good tyrosinase protein expression was sectioned (4 μm) and mounted on slides. Biotinylated anti-HLA-A2-Tyr 369 Fab, biotinylated anti HLA-A2-HIV-gag Fab, and biotinylated anti-HLA ABC Ab W6/32 (eBioscience) were tetramerized with the use of peroxidase-labeled streptavidin (Jackson ImmunoResearch Laboratories) and IHC diluent (DakoCytomation). Slides were fixed with acetone (cold 99% acetone for 2 min), blocked with Sniper blocking solution (Biocare) and biotin-avidin block (Biocare), incubated with tetramerized, biotinylated Abs, and washed with TBS with Tween 20 (Biocare). The staining was developed with DAB peroxidase substrate (Biocare) and counterstained with hematoxylin (Biocare).

Melanoma 624.38 cells (10⁹) were lysed in buffer containing: 1% octyl glycoside, 0.25% deoxycholate, 33 μg/ml iodoacetamide, 2 mM EDTA, 0.1 mM PMSF, and protease inhibitor mixture. The lysate was incubated with protein A-Sepharose beads conjugated to anti-HLA class I mAb W6/32. The column was washed and HLA and peptides were eluted at room temperature with 0.1 N acetic acid adjusted to pH 3.0. Peptides were separated from HLA and Ab by ultrafiltration through 3-kDa cutoff Microcon centrifugation filters. The peptide pools were concentrated by C18 column, followed by vacuum evaporation. For MS, the peptides were resolved by reverse-phase chromatography on 0.075 × 200-mm fused silica capillaries (J&W) packed with Reprosil reversed-phase material (Dr. Maisch, GmbH, Ammerbuch, Germany). The peptides were eluted with linear 95-min gradients of 5–40% and 10 min at 95% acetonitrile with 0.1% formic acid in water at flow rates of 0.25 μl/min. MS was performed by an ion-trap mass spectrometer (Orbitrap XL; Thermo) in a positive mode using repetitively full MS scan followed by collision induces dissociation of the seven most dominant ions selected from the first MS scan. Following this sample, a mixture of synthetic HLA peptide ligands (500 fmol each) was analyzed using the same methodology. The MS data were clustered and analyzed using the Sequest software (Thermo), searching against the human part of the non-redundant National Center for Biotechnology Information database.

501A melanoma cells were incubated with cycloheximide (100 μg/ml; Sigma-Aldrich) to inhibit protein synthesis. The cells were lysed after 0, 2, 4, and 6 h in PBS containing 10% Nonidet P-40, 1% deoxycholate, 2 mM EDTA, 10 mM Tris (pH 7), 1 mM PMSF, and protease inhibitor mixture. The lysate was passed through a 21-gauge needle and incubated for 10 min on ice. Cells were centrifuged at 14,000 rpm for 5 min at 4°C and the supernatant was collected. Equal amounts of sample (15 μg) were loaded on SDS-PAGE and electroblotted onto nitrocellulose. The blots were probed with T311 mouse anti-tyrosinase, HMB-45 mouse anti-gp100, or A103 mouse anti-Mart-1 followed by incubation with a secondary HRP-conjugated Ab and detection by chemiluminescence. For a negative control, Panc-1 cells, which do not express tyrosinase, gp100, and Mart-1, were used. The resulting bands were quantified with Multi Gauge version 2.2 software. The degradation rate is expressed as half-life (t_½), the time for degradation of 50% of the sample. The degradation rate of each protein was evaluated by three to eight independent determinations of t_½. The data are expressed as mean ± SD.

TAP-deficient T2 cells were incubated with the indicated peptides at 37°C overnight. Cells were then washed, incubated with 10 μg/ml brefeldin A at 37°C for the indicated times, and analyzed by flow cytometry with the TCR-like Abs. The half-life (t_½) of the pMHC was calculated by linear regression analyses of semilogarithmic decay plots.

All experiments performed under this study are presented as independent assays which are representative of three to eight independent experiments. Cytotoxicity assays were performed in triplicates with SD bars indicated. For real-time PCR analysis of expression of Ags, r Pearson correlation coefficients were used with a confidence level of 95%.

We have shown previously the ability to generate recombinant Abs with peptide-specific, HLA-A2-restricted specificity to an array of tumor and viral T cell epitopes using large Ab phage libraries. We termed these molecules TCR-like Abs. A similar strategy was used to isolate such Abs to HLA-A2/tyrosinase_369–377 in which a large naive Ab phage display library was screened against recombinant A2-tyrosinase_369–377 complexes. Specific clones were detected by an ELISA in which binding was tested with specific and nonspecific complexes. Fig. 1 A shows the reactivity of several Fab clones in an ELISA with purified HLA-A2-Tyr complexes as well with control HLA-A2 complexes displaying other HLA-A2-restricted peptides. The soluble Fabs reacted specifically with the complex containing the TyrD_369–377 peptide but not with HLA-A2 complexes folded with either of the other six control peptides. DNA fingerprint analysis revealed, by BstNI restriction reaction, a single pattern indicating one positive anti-tyrosinase 369–377 clone isolated from the library (data not shown). The final clone used further in this study was designated TA2.

We next tested the ability of the anti-HLA-A2/Tyr TA2 Fab to bind the target in its native form as expressed on APCs (Fig. 1,B). EBV-transformed B lymphoblasts HLA-A2⁺ JY cells were loaded with tyrosinase 369–377 YMDGTMSQV peptide or control peptides. Peptide-loaded cells were incubated with the soluble, purified Fab, followed by incubation with FITC-labeled anti-human Ab. As shown in Fig. 1,B, Fab TA2 binds to cells loaded with the tyrosinase peptide but not to cells loaded with the control peptides. In addition, Fab TA2 reacted poorly with HLA-A2⁺ cells loaded with the unmodified tyrosinase N 369–377 peptide containing asparagine rather than aspartic acid in position 371 (Fig. 1 B, pink histogram), compared with cells loaded with the TyrD peptide. This further emphasizes the fine specificity of TA2 Fab for the HLA-A2-tyrosinase 369–377 peptide complex.

Because Fabs isolated from phage libraries are monovalent, we wished to improve the reactivity and sensitivity of the Fab by increasing its avidity. This was achieved by using two strategies, generating Fab tetramers, as was previously shown for other TCR-like Abs (22), and transforming a TCR-like Fab into a whole bivalent IgG molecule (see Materials and Methods for details). The binding intensity and specificity of the TA2 Fab tetramer was assessed by flow cytometry using peptide-loaded JY cells (Fig. 1, C and D). As shown in Fig. 1,C, TA2 Fab tetramer binds tyrosinase peptide-loaded JY cells with greater intensity compared with TA2 Fab monomer. In addition, the Fab tetramer maintained its specificity (Fig. 1 D), because no binding to cells loaded with control peptides was observed.

For the second approach, we implanted the Ab Fab domains onto an IgG1 Ab scaffold as described in Materials and Methods. The reactivity and specificity of the purified TA2-IgG were assessed by flow cytometry. JY cells were loaded with specific or control peptides and incubated with the Ab, followed by incubation with PE-labeled anti-human Ab. As shown in Fig. 1,E, IgG Ab-bound cells loaded with tyrosinase peptide but did not bind cells loaded with control peptides (Fig. 1 F). Hence, TA2-IgG maintained the specificity of TA2-Fab. Furthermore, a whole IgG Ab showed improved sensitivity compared with the Fab. When JY cells were pulsed with low peptide concentrations, a 60-fold lower concentration of TA2-IgG Ab compared with TA2-Fab was required to detect the same number of tyrosinase-specific MHC complexes on the surface of peptide-pulsed cells (data not shown).

Previously reported TCR-like Fab Abs, which recognize HLA-A2 in complex with peptides derived from the melanoma differentiation Ags gp-100 (209, 280, and 154) and MART-1 (Ref. 23 and data not shown), were transformed into Fab tetramers or whole IgG molecules as described above, while maintaining the very unique fine specificity toward the particular HLA-A2-peptide MHC complex. Thus, we demonstrate the feasibility of transforming phage-derived Fab Abs into whole IgGs displaying the same specificity and increased binding reactivity.

To study expression of melanoma differentiation-derived HLA-A2-peptide complexes, we used five lines derived from melanoma patients. The gene expression of the melanoma differentiation Ags Melan-A/Mart-1, Pmel17/gp100, and tyrosinase was analyzed by RT-PCR. As show in Fig. 2 A, the amplification results show that the lines 624.38, 501A, TC-2224, and TC-1352 express all three melanoma differentiation Ags, while the cell line 1938 fails to express any of these three Ags. GAPDH was used as a control and showed positive in all five cell lines.

To explore whether the HLA-A2/tyrosinase TCR-like TA2 Ab is capable of binding endogenously derived MHC-tyrosinase complexes on the surface of tumor cells, we performed flow cytometry analysis on lines derived from melanoma patients. Cells were incubated with TA2 anti-tyrosinase 369–377/HLA-A2 Ab followed by incubation with PE-labeled anti-human Ab. As shown in Fig. 2, B–D, the TA2 Fab recognized tyrosinase-positive and HLA-A2-positive cells with a very high intensity, which may indicate that large numbers of HLA-A2-tyrosinase complexes are presented on the surface of the melanoma cells. The staining with TA2 was very homogeneous; intracellular staining of these melanoma cells (TC1352, TC2224, 624.38, and 501A) with Ab against the tyrosinase protein revealed that ∼95% of the cells in each line tested express the tyrosinase protein (data not shown). No reactivity was detected with tyrosinase-negative or HLA-A2-negative cells (Fig. 2, E and F). The specificity of the anti-tyrosinase/HLA-A2 TCR-like Ab was verified by extensive flow cytometry analysis of multiple cell lines of various histological origins which are HLA-A2 positive and Ag (tyrosinase) negative such as Saos-2, Panc-1, MDA231, MCF7, SW624, JY (data not shown), as well as the analysis of 31 primary melanoma cultures from patients (see below). The overall conclusion from these studies as well as further analysis presented herein is that the TCR-like Abs are highly specific and they recognize only the specific peptide-MHC complex presented on the cell surface when the adequate combination of HLA allele and Ag exist.

Interestingly, using the tyrosinase TA2 TCR-like Ab, the expression of HLA-A2-tyrosinase complexes on the surface of normal melanocytes was low to undetectable, although HLA-A2 expression was high (supplemental Fig. 1⁴). This observation, however, is limited to one normal primary melanocyte culture line and therefore should be expanded. Expression of tyrosinase-HLA-A2 complexes was also observed on melanoma HLA-A2-positive and tyrosinase Ag-negative cells after transfection with the tyrosinase gene (supplemental Fig. 2). However, the density of tyrosinase-HLA-A2 complexes on these transfected cells was relatively low compared with authentic HLA-A2 and tyrosinase-positive melanoma cells (Fig. 2), which may reflect significant differences in tyrosinase processing between transfected and native cells.

Further evidence for the high reactivity of TA2 with melanoma cells stems from experiments in which the reactivity of the TA2 Fab and whole IgG were compared and when titration experiments were performed. As shown in Fig. 2, G and H, the amount of whole IgG required to achieve a comparable intensity to that observed with the Fab was 5-fold lower (1 μg of IgG vs 5 μg of Fab). In addition, the whole IgG molecule could be titrated down to 4 ng and still demonstrate clear reactivity with the melanoma cells. Staining cells with such a low amount of Ab can be achieved only if sufficient complexes are present on the surface of the cells. Such intense reactivity was not observed before for TCR-like Abs in our studies (further comparison to flow cytometry reactivity will follow) as well by other groups.

To further verify this observation made with the TA2 TCR-like Ab that high numbers of HLA-A2-Tyr complexes are expressed on the surface of melanoma cells, we performed cytotoxicity experiments with CTLs that specifically recognize the HLA-A2-Tyr 369–377 epitope. To this end, we used two melanoma lines: 624.38, which expresses high levels of the HLA-A2-Tyr complexes on the surface, as shown by flow cytometry via the reactivity of the TA2 Ab, and TC-2183, which expresses low levels of these complexes (Fig. 3,A). For cytotoxicity assays, radioactive chromium-labeled target melanoma cells were incubated overnight in the presence or absence of 10 μM tyrosinase peptide and subsequently exposed to increasing CTL E:T ratios. As shown in Fig. 3 B, the addition of the tyrosinase peptide significantly increased by 2.5-fold specific lysis of TC-2183 melanoma target cells, which express low numbers of HLA-A2-tyrosinase complexes according to our flow cytometry analysis, compared with target cells without prior pulsing with Tyr peptide. Thus, the pulsing with peptide increased the number of complexes on the surface and resulted in enhanced killing. In contrast, the addition of peptide to 624.38 melanoma target cells, which express high levels of HLA-A2-Tyr complexes, did not significantly affect the degree of lysis accomplished with the anti-HLA-A2-Tyr CTLs. These results suggest that the 624.38 target cells express endogenously derived high levels of the HLA-A2-Tyr complexes and, in contrast to TC-2183 cells, which express low levels, additional pulsed peptide does not contribute to increased killing. These results further indicate the native endogenously derived high level expression of the HLA-A2/Tyr epitope on the surface of melanoma target cells.

Further evidence for the specificity of these assays is shown in Fig. 3,C in which the CTL-mediated killing was blocked completely and in a dose-dependent manner with the HLA-A2/tyrosinase-specific TCR-like Ab similarly to blocking with W6/32 (which binds total HLA molecules). TA2 at 100 μg/ml partially blocked tyrosinase-specific CTL lysis of melanoma cells (data not shown). In contrast, the affinity-matured clone MC1 efficiently and completely inhibited lysis at 50 μg/ml (Fig. 3,C, left panel). This effect was dose dependent. Moreover, inhibition of lysis was specific; a control TCR-like Ab was not able to block killing (Fig. 3 C, right panel).

To further study the high HLA-A2-tyrosinase presentation, we attempted to visualize these complexes on the surface of melanoma cells by confocal microscopy. The melanoma cells were reacted with the TA2 anti-HLA-A2-Tyr Ab as well as with anti-HLA-A2 BB7.2 mAb and examined by confocal microscopy. As shown in Fig. 4, the cells were stained very intensely for HLA-A2 expression and highly stained with the TA2 Ab of HLA-A2-Tyr complexes, indicating the large number of tyrosinase-derived complexes expressed on the surface of these cells. Interestingly, the specific HLA-A2-Tyr complexes are organized in unique clusters on the surface of the melanoma cells. Colocalization of HLA-A2 staining (BB7.2 mAb) and HLA-A2-tyrosinase complexes (TA2 TCR-like mAb) was only partially observed (Fig. 4). However, this was explained by results from experiments (data not shown) in which BB7.2 and TCR-like Abs were shown to interfere with each other, indicating that the binding of one Ab disrupts the other due to steric interference. These findings may indicate that the epitopes of these two mAbs are structurally close.

To investigate further the high level presentation of HLA-A2-tyrosinase complexes on the surface of melanoma cells, we examined the relative expression of the three major melanoma differentiation Ags: tyrosinase, gp100, and Mart-1. This was performed to exclude the possibility that the high numbers of HLA-A2-Tyr complexes are due to significant overexpression of tyrosinase mRNA in melanoma cells compared with the other Ags. For this purpose, cDNA was produced from melanoma cells, followed by quantitative real-time PCR analysis. GAPDH was used as a control gene for verification of cDNA production and for normalizing the relative expression of the three target genes. The relative expression of the target genes was determined as 2^−ΔCT (ΔC_T is determined by C_T target gene − C_T of GAPDH). Fig. 5 A presents the relative gene expression as determined by the real-time PCR analysis. The results show no correlation between the relative gene expression and the presentation of HLA-A2-tyrosinase complexes. On the contrary, the expression of tyrosinase is the lowest in the expression hierarchy of the three genes with gp100 being the highest in three of four cell lines and tyrosinase last. In TC-2224 cells, tyrosinase was somewhat higher in expression levels compared with the two other genes.

We expanded the studies on the relative expression of tyrosinase, gp100, and Mart-1 by performing quantitative real-time PCR from 31 primary melanoma cell lines isolated from patients (gift from the Surgery Branch at the National Cancer Institute, National Institutes of Health). GAPDH was used as a control gene for verification of cDNA production and for normalizing the relative expression of the three target genes. Several studies have demonstrated that the classic reference genes can vary extensively and display fluctuating expression levels upon different conditions. Thus, they are unsuitable for normalization purposes. However, since no absolute quantification or comparison between different cell lines was made in this study, GAPDH serves as an appropriate control gene. The relative expression of the target genes was determined as 2^−ΔCT (Table I). As determined by real-time PCR analysis, 23 (75%) of 31 melanoma cell lines express all three melanoma differentiation Ags. The expression level of the genes is presented in Fig. 5 B. The results show a strict correlation between the Mart-1 and tyrosinase expression levels (r Pearson correlation coefficients = 0.8 with confidence level of 95%). An average of 2.6-fold higher transcript number for Mart-1 compared with tyrosinase was observed. The expression levels of gp100 were significantly higher than those of Mart-1 and tyrosinase, but no correlation was observed.

We next wished to study the expression hierarchy of HLA-A2-derived epitopes generated from the three major melanoma Ags: tyrosinase, gp100, and Mart-1. We therefore used, in addition to TA2 anti-HLA-A2/tyrosinase, the TCR-like Ab CLA12, which recognizes the Mart-1-derived epitope 27–35 and the gp100-specific TCR-like Abs 1A7 and 2F1, which recognize the gp100-derived epitopes 209 and 280, respectively. All of these TCR-like Abs were transformed into whole IgG molecules for improved reactivity and detection capabilities (data not shown). To monitor total HLA-A2 expression, we used mAb BB7.2. The apparent binding affinity of the various TCRL IgG Abs to their respective HLA-A2-peptide complex was determined by using surface plasmon resonance BIAcore analysis (data not shown) and found to be similar in the range of 50 nM. This observation excludes the possibility that differences in expression levels/hierarchy of the melanoma-derived HLA-A2-peptide complexes results from differences in the detection capabilities (i.e., affinities) of the TCR-like Abs used for their detection.

The reactivity of these TCR-like Abs with melanoma cell lines was analyzed by flow cytometry. Indeed, staining of three melanoma cells lines with the various TCR-like whole IgGs revealed a distinctive expression hierarchy (Fig. 5, C–E). All three lines were stained intensely with BB7.2, indicating a high expression level of HLA-A2 molecules on their cell surface. As shown, anti-HLA-A2/tyrosianse TA2 Ab displayed the highest reactivity compared with Mart-1- or gp100-specific TCR-like Abs, indicating high expression of HLA-A2-Tyr complexes on the surface of melanoma cells. The expression levels of HLA-A2-Mart-1 complexes were low (on TC-2224 cells) to modest (as expressed on 501A and 624.38 cells). The expression levels of HLA-A2/gp100 complexes were very low on all examined cell lines. No reactivity by these indicated TCR-like Abs was observed on control melanoma cells which were HLA-A2 negative (Fig. 5,F). All four melanoma lines used for staining express the three genes as shown by RT-PCR (Figs. 2,A and 5 A).

We then wished to confirm this observation on the 31 primary lines that were characterized for the relative expression of tyrosinase, Mart-1, and gp100 by real-time PCR as described above (Table I and Fig. 5 B). Flow cytometry analysis performed with the three TCR-like Abs, recognizing HLA-A2 in complex with tyrosinase-, Mart-1-, or gp100-derived peptides, on 23 melanoma cell lines (which express the three differentiation Ags, as determined by real-time PCR) further support our findings. Of 23 melanoma cell lines, 18 (75%) express HLA-A2 and 15 (83%) of the 18 which express tyrosinase and HLA-A2 were recognized by the TA2 Ab. Most of these were recognized with very high reactivity, which implies a high presentation number of HLA-A2-Tyr complexes on the surface of these cells. In contrast, only 8 (44%) of 18 were recognized by the anti-Mart-1 or anti-gp100 Abs. The low reactivity of these Abs with the melanoma cell lines implies that low numbers of Mart-1 and gp100 complexes are presented on these cells.

The analysis by real-time PCR does not support the possibility that the high amount of HLA-A2-Tyr complexes is the result of overexpression of the tyrosinase protein in melanoma cells. In fact, the tyrosinase transcript was expressed at lower levels than Mart-1 and gp100 in most of the cells examined. Thus, there is no correlation between the relative gene expression and the presentation of the specific HLA-A2 complexes.

The data obtained from these experiments suggest a clear expression hierarchy of the three major Ags with the number of HLA-A2-tyrosinase complexes presented on melanoma cells significantly higher than the number of HLA-A2-Mart-1 and gp100 complexes. Moreover, we attempted to expand these studies also to melanoma tissues from patients. Thus, frozen, tyrosinase-positive melanoma tissues from two patients with HLA-type A0201 were sectioned and stained with the tyrosinase-specific TCR-like Ab. As shown in Fig. 5,G, the melanoma tissues were stained intensely with the anti-HLA-A2/tyrosinase Fab Ab (Fig. 5,G, a and c) but not with anti-HLA-A2/Gag Fab that was used as a control (Fig. 5,G, b and d). These sections were stained also with W6/32 (Fig. 5 Ge) to monitor for their total expression of HLA class I molecules. These results strengthen our observations made with cultured melanoma cell lines, because the high level of presentation of the tyrosinase epitope exists in the authentic melanoma tissue from patients as evident from the high intensity of reactivity with the tyrosinase-specific TCR-like Ab.

To corroborate the results described herein with respect to relative abundance of the tyrosinase vs gp100-Mart-1 complexes on the surface of melanoma cell lines, we performed direct MS analysis of HLA-A2-derived peptides eluted from melanoma cells (Fig. 6). Isolation of HLA molecules from melanoma 624.38 cells using a W6/32 affinity column and peptide elution combined with MS analysis confirmed our results, with the TCR-like Abs, that the tyrosinase HLA-A2-restricted peptide 369–377 is one of the most abundant peptides bound by the A2 molecules of these cells (Fig. 6,A and B, d–f), while gp100 (Fig. 6,Cd) and Mart-1 (data not shown) are presented at a very low abundance. The identity of the tyrosinase/gp100/ Mart-1 peptide eluted from 624.38 melanoma cells was confirmed by using synthetic tyrosinase (Fig. 6,B), gp100 (Fig. 6 C), and Mart-1 (data not shown) peptides analyzed in the same LC-MS/MS conditions and compared with the tandem MS analysis of the melanoma-eluted HLA peptides.

Overall, these results suggest that large numbers of HLA-A2-tyrosinase complexes are expressed on the surface of melanoma cells with no direct correlation to the expression of other major Ags such as gp100 and Mart-1 and with no correlation to the gene expression level of these Ags.

We next wished to quantify, using TA2 Ab, the number of tyrosinase-HLA-A2 complexes on the surface of melanoma cells, as previously described (24). The minimal number of specific HLA-A2 complexes on the surface of the cells was determined by comparing the fluorescence intensity of a specific PE-labeled TCR-like Fab tetramer or IgG with a standard curve (see Fig. 7) generated by measuring the fluorescence intensity of calibration beads with known numbers of PE molecules per bead (QuantiBRITE PE beads). Using this strategy, we determined the number of HLA-A2-Tyr complexes on the surface of cells using TA2 Fab tetramer. As shown in Fig. 7, the minimal number of HLA-A2-tyrosinase complexes displayed on 501A Tyr⁺/HLA-A2⁺ melanoma cells was determined as ∼3700 ± 265 complexes/cell. The number of total HLA-A2 complexes expressed on the surface of the melanoma lines tested was measured by staining with PE-labeled anti-HLA-A2.1 mAb BB7.2 and was determined to be in the range of 10–14,000 molecules/cell. Thus, HLA-A2-tyrosinase complexes account for ∼20% of the total number of HLA-A2 complexes on the surface of the cell. Similar measurements were performed with the TA2 Fab tetramer on melanoma cells with high, medium, and low reactivity with the TCR-like Ab. The minimal number of HLA-A2-Tyr complexes displayed on high expressing cells such as 624.38 Tyr⁺HLA-A2⁺ melanoma cells was determined as 4100 ± 338, which is ∼17% of the total number of HLA-A2 complexes (Fig. 7,B). TC-2224 cells, which were also stained with high reactivity, revealed similar results (data not shown). TC-2207 cells that were stained at a moderate level displayed a minimal number (1120 ± 201) of HLA-A2-Tyr complexes. Cells, such as TC-1760, which exhibited low reactivity with the TA2 Ab, displayed 120 ± 50 HLA-A2-Tyr complexes (Fig. 7 B). Presentation of tumor Ags at such magnitude was not described until now.

We further attempted to investigate the mechanisms leading to such a high level of HLA-A2-tyrosinase complex presentation on the surface of melanoma cells. Following verification that the high presentation of HLA-A2-Tyr complexes on the surface of melanoma cells was not derived from overexpression of the tyrosinase gene, we further examined the influence of tyrosinase protein stability on complex presentation. Since our melanoma cells exhibit an amelanotic phenotype, we hypothesized that the tyrosinase protein is inactive, thus large amounts of the protein are being degraded by the proteasome and thus high presentation is accomplished. It was shown elsewhere that 3,4-dihydroxy-l-phenylalanine (l-DOPA) can induce conformational changes favorable for the exit of tyrosinase from the ER to the Golgi and restore melanin synthesis, thus inducing activity of inactive tyrosinase (25). To test the effect of l-DOPA on the presentation of HLA-A2/tyrosinase, we first eluted peptides presented on MHC molecules by using citrate-phosphate buffer as previously described (26) to eliminate endogenously derived complexes and treated the cells with 1 mM l-DOPA for 20 h. Cells were then analyzed by flow cytometry for the presentation of HLA-A2-tyrosinase complexes using TA2 Ab. The data in Fig. 8, from one experiment, are representative of the data from five experiments showing a 50–75% decrease in the mean fluorescence intensity (MFI) of cells treated with l-DOPA compared with untreated cells. This dramatic decrease indicates that a much lower number of HLA-A2-tyrosinase complexes are presented on l-DOPA-treated cells. Thus, high HLA-A2-Tyr complex presentation on the surface may be the result of the tyrosinase protein instability in these cells.

To further investigate the relationship between protein stability and Ag presentation, we determined the rate of protein degradation in melanoma cells. To this end, we have used 501A melanoma cells which express HLA-A2 and the three melanoma Ags as determined by RT-PCR (Fig. 2,A) and reactivity with the TCR-like Abs (Fig. 5,D). To determine protein half-life, protein synthesis was arrested by treatment with cycloheximide, followed by cell harvesting at various time points. Protein degradation was monitored by running SDS-PAGE of cell extracts at a fixed protein amount followed by Western blotting and detection of the amount of each Ag by using Abs specific for gp100 (Fig. 9,C), Mart-1 (Fig. 9,B), and tyrosinase (Fig. 9,A) whole proteins followed by anti-mouse peroxidase Ab. Ab reactivity on blots was quantified by scanning densitometry using a densitometry computer program. As shown in Fig. 9, A–C, there was a difference in the rate of degradation of the various proteins. Mart-1 (Fig. 9,B) was the most stable with a t_½ of 5.62 ± 1.1 h, gp100 (Fig. 9,C) had moderate stability (t_½ of 4.34 ± 0.2 h), and tyrosinase (Fig. 9 A) was the least stable of the three proteins with a t_½ of 2.5 ± 0.9 h. These results further indicate the possible relationship between protein stability and the presentation hierarchy of the three melanoma Ags as observed by staining with the specific TCR-like Abs. It is possible that rapid degradation of the protein results in more efficient intracellular processing and presentation, as observed for tyrosinase, which demonstrated a relatively short t_½ compared with the other Ags, but extremely high level presentation on the cell surface.

In addition to protein stability, we tested the decay of HLA-A2 molecules complexed with Tyr_369–377, Mart-1_27–35, gp100_209–217, and gp100_280–288 peptides. This was performed on brefeldin A-treated, peptide-pulsed T2 cells by flow cytometry with the TCR-like Abs. As shown in Fig. 9 E, the most stable complex was HLA-A2-tyrosinase, exhibiting a t_½ of ∼10 h. HLA-A2-Mart-1 complexes displayed a moderate half-life of ∼3.5 h, and HLA-A2-gp100 209 and 280 complexes displayed a t_½ of 1.8–2 h. Thus, the hierarchy of the pepitide-HLA-A2 complex decay/stability correlates with the level of the specific peptide-HLA-A2 complex presentation.

We have demonstrated previously our ability to generate recombinant TCR-like Abs, which can specifically recognize HLA-A2 in complex with peptides derived from tumor-associated or viral Ags (23, 27). The specificity and binding properties of a newly isolated TCR-like Ab directed to HLA-A2-tyrosinase complexes are described herein. We demonstrate in this study the importance of TCR-like Abs for studies aiming to measure and quantify Ag presentation of MHC class I complexes on cancer cells.

Using TCR-like Abs directed to the three major differentiation Ags expressed by human melanoma, we identified an expression hierarchy which demonstrated a surprisingly high level presentation of HLA-A2-tyrosinase complexes on the surface of melanoma cells compared with the two other Ags gp100 and Mart-1. Tyrosinase is a membrane-associated N-linked glycoprotein and it is the key enzyme in melanin synthesis. It is expressed in all healthy melanocytes and in nearly all melanoma tumor samples (28, 29). Peptides derived from this enzyme are presented on MHC class I molecules and are recognized by autologous CTLs in melanoma patients (30).

The tyrosinase HLA-A2-associated epitope 369–377, YMDGTMSQV, is generated by posttranslational conversion of the sequence YMNGTMSQV. Only YMDGTMSQV and not YMNGTMSQV has been reported to be presented by HLA-A*0201 on cells expressing full-length tyrosinase (31). The tyrosinase protein is synthesized and folded in the ER. Properly folded tyrosinase is transported via the trans-Golgi network to melanosomes (32). The proposed model for the tyrosinase epitope presentation process includes glycosylation on the N₃₇₃ residue during synthesis of the full-length protein in the ER, followed by reverse translocation of the enzyme to the cytosol, deglycosylation accompanied by deamination (thus conversion of N₃₇₃ to D), degradation by the proteasome, and TAP-mediated transport of the resulting peptide fragments into the ER for HLA-A2 binding (33).

Loss of pigmentation is frequently observed in human melanoma cells. In these amelanotic melanoma cell lines, tyrosinase failed to reach the melanosome and was retained in the ER. The aberrant accumulation of tyrosinase in the ER of melanoma cells results from tumor-induced metabolic changes. The acidification of the ER-Golgi boundary of melanoma cells, which is hostile to tyrosinase maturation, is the cause of the amelanotic phenotype (34). It has been shown that the substrates DOPA and tyrosine can induce conformational changes favorable for the exit of tyrosinase from the ER to the Golgi and restore melanin synthesis (25).

In this study, we show that the TCR-like Ab, specific to the tyrosinase epitope 369–377 presented on melanoma cells, is capable of recognizing melanoma cell lines with a high reactivity, as a consequence of the unusually elevated number of the HLA-A2-tyrosinase peptide complexes presented on the surface of melanoma cells. The unexpected presentation of the tyrosinase-derived epitope was confirmed in five independent assays: flow cytometry on multiple melanoma lines generated from patients, confocal microscopy immunostaining of melanoma lines, staining of authentic melanoma tissue from patients, cytotoxicity assays using tyrosinase-specific CTLs, and finally direct elution of HLA peptide ligands from melanoma cells and their analysis by MS. We propose that the inactive tyrosinase, which results in the abundant amelanotic phenotype, is the cause of the very high presentation of tyrosinase-HLA-A2 complexes on the surface of melanoma cells. We postulate that other tyrosinase-derived peptides are presented at high density on MHC class I molecules; using new TCR-like Abs against these complexes preliminarily confirms this hypothesis (Y. Michaeli and Y. Reiter, unpublished data).

We further show that the use of l-DOPA, which causes active tyrosinase and melanin synthesis, prevents the high presentation of tyrosinase-HLA-A2 complexes. We observed that among the three Ags analyzed, Tyrosinase, which is highly presented on the cell surface, is the fastest in degradation, thus its relative low stability may contribute to the high level of presentation that was observed. However, protein stability may be only one reason for this finding. Other possibilities may include efficiency of processing that relates to the particular composition of the Ag. Decay experiments which measure peptide-HLA-A2 complex stability on the surface indicated that HLA-A2-tyrosinase complexes are more stable compared with gp100 and Mart-1 epitopes. This may also contribute to the high level presentation of HLA-A2-tyrosinase complexes on the surface of melanoma cells. The binding affinity of the three differentiation Ag-derived peptides to HLA molecules is also quite different, which may influence complex stability. The binding score for the TyrD peptide, analyzed by BIMAS (http://www-bimas.cit.nih.gov/molbio/hla_bind/), is 212.58 while the score for the gp100-209 and Mart-1 peptides are 3.8 and 2.2, respectively.

Mart-1_27–35 and the gp100₂₀₉ and gp100₂₈₀ epitopes are very common immunogenic epitopes for HLA-A2-restricted melanoma-specific tumor-infiltrating lymphocytes. They are known as immunodominant epitopes and CD8⁺ CTLs specific for these epitopes are frequently found in melanoma patients (35, 36, 37). In contrast, the generation of a tyrosinase-specific response in melanoma patients is a relatively infrequent event. In several studies, tyrosinase-specific T cells were hardly detectable in the examined tumor-infiltrating lymphocytes (36). In addition, flow cytometric analysis of PBMCs stained with tetramers showed that tyrosinase peptide 369–377-specific CD8⁺ T cells were hardly detectable in peripheral blood of melanoma patients. However, significant numbers of such cells were detected after short-term stimulation of CD8⁺ lymphocytes with tyrosinase peptide (16). The results presented herein may explain the low immunogenicity of the tyrosinase epitope. Continual exposure of T cells to Ag maintains an unresponsive state and results in adaptive tolerance (38, 39). Preliminary data shown herein suggest that HLA-A2-Tyr_369–377 complexes are not presented on healthy melanocyte membranes. This may be because in these cells the tyrosinase protein is stable and melanin synthesis is accomplished; however, this hypothesis has to be tested on additional samples of normal melanocytes.

Recently, we have shown that properly primed activated CD8⁺ T cells acquire an unresponsive phenotype when they are exposed to high numbers of peptide-MHC complexes. This phenotype is accompanied by a significant change in gene expression profiles and presence of an anergic gene signature (40). We thus postulate that the large numbers of HLA-A2-tyrosinase complexes presented on the surface of melanoma may induce unresponsiveness in specific CD8⁺ T cells.

We present herein new information describing the unique presentation hierarchy of melanoma differentiation tumor Ags. According to our results, tyrosinase_369–377 is presented as thousands of copies on melanoma cell lines. Presentation of a tumor Ag with such magnitude was not described until now. This information is particularly important when targets for immunotherapy are considered.

The most important question with respect to immunotherapeutic and diagnostic applications of TCR-like Abs relates to the low density and turnover of the specific epitope on the target cell surface. With regard to the density and targeted killing of cells, we have previously shown in a murine model that to achieve efficient killing with a TCR-like immunotoxin molecule a density of several hundreds to a thousand MHC-peptide complexes is required for selective elimination of APCs (41). The present data suggest that the concept that T cell epitopes are expressed at low numbers on target tumor cells is not general and there are probably many exceptions represented by Ags such as tyrosinase. These targets may be ideal for Ab-mediated drug delivery, as well as tumor cell lysis achieved by Ab-dependent cell-mediated cytotoxicity or complement-dependant cytotoxicity mechanisms. Exploring these highly expressed T cell epitopes may shed new light on the biology of Ag processing and presentation. For the clinical aspects, such epitopes that are tumor associated and are expressed at high levels open new opportunities for therapeutic interventions, particularly in cases where T cell responses fail. In such cases, TCR-like Abs can replace cellular immunity and constitute an attractive therapeutic modality.

We thank Céline Beauverd for excellent technical assistance.

The authors have no financial conflict of interest.