Simultaneous Generation of CD8⁺ and CD4⁺ Melanoma-Reactive T Cells by Retroviral-Mediated Transfer of a Single T-Cell Receptor

Adoptive transfer of T cells with in vitro antitumor activity has been used to treat cancer patients to circumvent difficulties in generating an antitumor immune response in vivo (1). However, isolating and/or expanding sufficient quantities of tumor reactive T cells is a difficult and laborious process for melanoma patients (2, 3) and extremely difficult for patients with other cancers. We recently described an approach to generating large numbers of antitumor T cells whereby retroviral vectors encoding the T-cell receptor (TCR) genes from tumor-reactive T cells are transferred to normal blood-derived T cells (4). We and others have shown that retroviral-mediated TCR gene transfer can confer tumor antigen (TA)-specific reactivity to the new T cells (4–7). This approach would provide an “off the shelf” reagent capable of engineering a patient's T cells to recognize their cancer. The key step in this process is to identify TCRs with desirable properties that make them candidates for genetic transfer. We previously reported isolating a tumor-infiltrating lymphocyte (TIL) culture (TIL 1383I) from a melanoma patient that is specific for the melanoma antigen tyrosinase (8). TIL 1383I recognizes an MHC class I (HLA-A2*01) restricted epitope (368-376) of tyrosinase despite being a CD4⁺ T cell. This violates the standard paradigm of TCR/peptide/MHC recognition whereas CD4⁺ T cells recognize peptides bound to MHC class II molecules, whereas CD8⁺ T cells recognize peptides bound to MHC class I molecules. TCRs that can mediate MHC class I–restricted tumor cell recognition in the absence of CD8 have properties that are distinct from TCRs expressed by the majority of T cells found in the normal repertoire in that they violate principles of antigen recognition by CD4⁺ and CD8⁺ T cells. In the current study, we describe the transfer of the TIL 1383I TCR to peripheral blood lymphocyte (PBL)-derived normal human T cells. The resulting TCR-transduced T cells recognized antigen-presenting cells loaded with tyrosinase peptide, as well as melanoma cells in an HLA-A2 restricted manner. Furthermore, clones isolated from these cultures showed both CD8⁺ and CD4⁺ T cells recognized HLA-A2⁺/tryosinase⁺ melanoma cells, giving us the possibility of engineering class I MHC–restricted effector and T helper cells against melanoma.

Cell Lines. Tumor cell lines and TILs were established from melanoma patients at the Surgery Branch, National Cancer Institute. Tumor cells, T2 cells, and Jurkat cells were maintained in RPMI 1640 (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (Invitrogen Life Technologies, Carlsbad, CA) 100 units/mL penicillin, 100 μg/mL streptomycin, and 2 mmol/L l-glutamine (Mediatech; cRPMI). 293GP retroviral producer cells were maintained in DMEM supplemented as above (cDMEM). TILs were maintained in X-Vivo 15 (Biowhittaker, Walkersville, MD) supplemented with 10% heat inactivated, pooled human AB serum (Valley Biomedical, Winchester, VA) and 6,000 IU/mL rhIL-2 (Chiron Co., Emeryville, CA).

Retroviral Vector Construction. Full-length cDNAs encoding the TIL 1383I TCR were amplified by PCR using cloning primers designed from genomic sequences of the α and β chain genes as described (9) and ligated into the pCR2.1 cloning vector (Invitrogen Life Technologies). Their sequences were determined using BigDye Terminator Cycle Sequencing kits and analyzed using an ABI Prism 310 Genetic Analyzer (Applied Biosystems, Foster City, CA). The TIL 1383I TCR genes were subcloned into the SAMEN CMV/SRα retroviral vector (9) to create the A9.3 retrovirus.

Retroviral Production and Transduction. 293GP cells (70-90% confluent) were transiently cotransfected in 10-cm tissue culture dishes with 3 μg A9.3 retroviral plasmid and 3 μg of a plasmid containing the vesicular stomatatis virus envelope gene using LipofectAMINE and PLUS reagents (Invitrogen Life Technologies) as described (9); 10 mL of fresh retroviral supernatants (cRPMI for Jurkat cells or cDMEM for human PBL) were harvested at 48 and 72 hours post-transfection.

PBL from four normal donors (designated B, D, F, and V) were isolated by density gradient centrifugation using Lymphocyte Separation Medium (Mediatech), suspended at 10⁶ cells/mL in X-Vivo 15 containing 10 ng/mL OKT3 (Ortho Biotech, Raritan, NJ) and 600 IU/mL rhIL-2, and cultured for 3 days to activate T cells. PBL or Jurkat cells were then resuspended at 10⁶ cells/mL in fresh retroviral supernatant containing 8 μg/mL polybrene (Sigma, St. Louis, MO) with 600 IU/mL rhIL-2 being added to PBL. All cells were transduced by spinoculation in a centrifuge at 1,000 × g for 90-minute at 32°C in 24-well tissue culture plates (1 mL per well). Four hours post-spinoculation, 1 mL of fresh medium supplemented as above was added to wells and the cells cultured overnight. This process was repeated once 24 hours later for Jurkat cells or twice at 24 and 48 hours later for primary T cells. Transduced Jurkat cells were selected by the addition of 2.0 mg/mL G-418 (Research Products Intl., Hanover Park, IL) for 7 days then assayed for anti-tyrosinase reactivity. Transduced T cells were selected by addition of G-418 (0.5 mg/mL for donors D and V, and 1.0 mg/mL for donors B and F) for 3 to 5 days then expanded by culturing 2.5 × 10⁵ cells with 4 × 10⁷ irradiated (5,000 rads) allogeneic peripheral blood mononuclear cells in the presence of 30 ng/mL OKT3 and G-418 (concentrations as above) in upright 25-cm² tissue culture flasks. 300 IU/mL rhIL-2 was added day 2. Cultures were replenished with fresh IL-2 and G-418 containing medium on days 5, 8, and 11 and were assayed and frozen for future experiments on day 14.

T Cell Cloning. TIL 1383I TCR transduced T cells were cloned in limiting dilution as described (4) or by single cell sorting of T cells stained with FITC conjugated anti-BV12 (Pierce, Rockford, IL) and PE conjugated anti-CD4 or CD8 (BD PharMingen, San Diego, CA) monoclonal antibodys using a MoFlo-HTS cell sorter (DakoCytomation, Carpinteria, CA) at a density of 1 cell/well of 96-well tissue culture plates. Growth positive clones were tested for antigen recognition and reactive clones were expanded in the presence of OKT3 and IL-2 as above (excluding G-418) for later experiments.

Antigen Recognition Assays. Transduced cells were tested for anti-tyrosinase reactivity by coculturing responder cells with stimulator cells in a 1:1 ratio overnight in 200 μL cRPMI in 96-well U-bottomed plates. Stimulator cells included tumor cells or peptide loaded T2 cells. T2 cells were peptide pulsed by incubating 10⁶ cells/mL with MART-1:27-35 (AAGIGILTV) or tyrosinase:368-376 (YMDGTMSQV) peptide for 2 hours at 37°C. 5 ng/mL phorbol 12-myristate 13-acetate (Sigma) was added to Jurkat cell cocultures to enhance signaling. The amount of cytokine released [IFN-γ, granulocyte macrophage colony-stimulating factor (GM-CSF), IL-4, tumor necrosis factor-α or IL-2] was measured by ELISA (Pierce; R&D Systems, Minneapolis, MN).

Cell Lysis Assays. Transduced bulk cultures and clones were tested for their ability to lyse peptide loaded T2 cells or melanoma cells. T2 cells were loaded with 2 μg/mL tyrosinase or MART-1 peptide as above. Tumor cells and peptide loaded T2 cells were then labeled for 1 hour with ⁵¹Cr (200 μCi ) at 37°C. Labeled target cells were then washed thrice with PBS then plated in triplicate wells of a 96-well tissue culture plate in a volume of 100 μL cRPMI per well. Effector cells were added in a volume of 100 μL cRPMI per well at E/T ratios as described in the figures. 1% SDS was added to selected target wells to determine maximum ⁵¹Cr release. Plates were incubated 4 hours at 37°C; 50 μL supernatant were harvested and plated onto LumaPlate-96 well scintillation plates (Perkin-Elmer, Boston, MA) and dried overnight. Sample plates were counted the next morning on a TopCount analyzer (Perkin-Elmer). The % specific lysis was determined using the following formula [(Experimental cpm − spontaneous cpm)/(Maximum cpm − spontaneous cpm)] × 100.

Immunofluorescence. Transduced bulk cultures and clones were stained for 30 minutes on ice with the following antibodies. FITC conjugated anti human BV12 antibody (Pierce), FITC labeled isotype control, PE and/or FITC labeled anti human CD8 and CD4 and isotype controls (BD PharMingen). Analysis was done on a FACScan flow cytometer (BD Biosciences, San Jose, CA).

The A9.3 Retrovirus Functionally Transfers the TIL 1383i TCR to Alternate Effectors. We constructed the A9.3 retrovirus encoding the α (AV4s1) and β (BV12s2) chains of the TIL 1383I TCR (Fig. 1). The functionality of the A9.3 retrovirus was first confirmed by transducing Jurkat cells and assaying them for peptide and/or tumor cell recognition by measuring IL-2 secretion. G-418 resistant cells expressed TCR BV12 as detected by flow cytometry (data not shown) and secreted IL-2 when cocultured with T2 cells loaded with tyrosinase peptide or with HLA-A2⁺/tyrosinase⁺ melanoma cells (Fig. 2). As expected, transduced cells did not recognize T2 cells loaded with a control peptide (MART-1) or HLA-A2⁻/tyrosinase⁺ melanoma cells. These results show the A9.3 retrovirus can transfer the TIL 1383I TCR and confer HLA-A2 restricted antigen specificity for tyrosinase to alternate effectors. Furthermore, tumor cell recognition was independent of CD8 since Jurkat cells lack CD8 expression. Therefore, the full specificity and function of the original CD4⁺ TIL 1383I was conferred by genetic transfer of the TCR.

Transfer of the TIL 1383i TCR to Primary T Cells. The ability to transfer the HLA-A2 anti-tyrosinase reactivity from TIL 1383I to normal human T cells was evaluated by transducing peripheral blood mononuclear cells from four normal donors (designated B, D, F, and V). Activated T cell cultures were transduced with A9.3 retroviral supernatants then cultured and expanded in the presence of G-418 to select for TCR gene–modified cells. Transduced T cell bulk cultures had very low expression of TCR BV12 by flow cytometry (data not shown). Cultures were assayed for reactivity with tyrosinase peptide loaded T2 cells and tumor cell lines in cytokine release assays. Significant cytokine release (defined as >100 pg/mL and twice the specific negative control value) was observed from transduced T cells from all donors when stimulated with tyrosinase loaded T2 cells compared with culture with negative control peptide (Fig. 3A). Three of four transduced cultures (B, D, and F) specifically secreted significant amounts of both GM-CSF and IFN-γ, although only donor B secreted IFN-γ levels comparable to GM-CSF. Transduced donor V cells secreted significant but low levels of GM-CSF but not IFN-γ. No antigen-specific cytokine release was detected from untransduced control donor cells in response to peptide loaded T2 cells (data not shown). Transduced cells from donors B, D, and F secreted 5- to 7-fold more GM-CSF and/or IFN-γ in response to culture with HLA-A2⁺ melanoma lines (Mel) 1300 and 624 compared with HLA-A2⁻ 624-28 Mel or an HLA-A2⁺ renal cell carcinoma (RCC) UOK131 which is tyrosinase negative (Fig. 3B). No specific cytokine release was detected from transduced donor V cells in response to tumor cells. IFN-γ secretion by transduced donor D and F cultures was detected by ELISA but below the threshold value of 100 pg/mL in all but one sample (D with 624 MEL), whereas transduced donor B cells secreted both GM-CSF and IFN-γ at significant levels. No specific cytokine secretion was detected from untransduced donor cultures stimulated with tumor cells (data not shown). Transduction of primary T cells with A9.3 retrovirus therefore resulted in cultures specific for tyrosinase:368-376 and in three of four cases, cultures that recognized HLA-A2⁺ melanoma cells. Low cytokine release by G-418 resistant donor V cells in response to peptide, in conjunction with no significant cytokine release in response to tumor cells indicate a poor transduction efficiency resulting in a low frequency of cells expressing adequately high levels of transferred TCR.

CD4⁺ and CD8⁺ Transduced T-Cell Clones Recognize Melanoma Cells. We generated a panel of T-cell clones from transduced donor cultures by limiting dilution or single cell sorting for CD4⁺/BV12⁺ or CD8⁺/BV12⁺ and further expanded those clones found reactive to antigen. Expanded clones were positive for TCR BV12 expression by flow cytometry (Fig. 4) and were stained for CD4 and CD8 to confirm the coreceptor phenotype of each clone.

We tested clones for sensitivity to antigen by coculture with T2 cells loaded with decreasing concentrations of tyrosinase and the amount of cytokine released measured by ELISA (Fig. 5A). Both CD4⁺ and CD8⁺ clones secreted increasing levels of cytokine in response to increasing tyrosinase peptide stimulation. TIL 1383I begins to secrete cytokine when stimulated with T2 cells loaded with 1 to 10 μg/mL tyrosinase, whereas TCR-transduced clones secreted cytokine when stimulated by T2 cells loaded with 10 to 100 μg/mL tyrosinase. GM-CSF was the predominant cytokine released from all clones except CD4⁺ clone D2F4 (donor D derived), which secreted comparable amounts of IFN-γ.

We tested clones for recognition of a panel of tumor cells and again detected cytokine release from both CD4⁺ and CD8⁺ cells in response to HLA-A2⁺ melanomas 1300 and 624, but not the negative control tumors (624-28 Mel or RCC UOK131; Fig. 5B). Indeed, several clones were highly reactive to HLA-A2⁺ melanoma cells, secreting ng/mL quantities of IFN-γ and/or GM-CSF. Clones V8B4, V8C4, and V8B11, which secrete GM-CSF in response to tyrosinase peptide and HLA-A2⁺ melanoma cells, were derived from transduced donor V cells. This shows that although transduction efficiency in donor V was too low for detection of tumor cell recognition in bulk cultures, highly reactive cells could be isolated from the culture. Furthermore, clone D2F4 (donor D) secreted tumor necrosis factor-α (100-500 pg/mL) in specific response to tyrosinase loaded T2 cells and HLA-A2⁺ melanoma cells, whereas clones F4D71, and F4E91 (donor F) secreted 100 to 200 pg/mL IL-4 in specific response to both antigen and HLA-A2⁺ melanoma cells (data not shown).

Lysis of Melanoma Cells by TCR-Transduced T Cells. We further tested both transduced bulk cultures and selected clones for the ability to kill target cells in lysis assays. Bulk cultures were coculured with ⁵¹Cr labeled T2 cells preloaded with 2 μg/mL tyrosinase peptide or control MART-1 peptide (Fig. 6A). A lack of sufficient numbers of transduced donor V cells coupled with an inability to obtain additional cells from this donor prevented us from determining the lytic capacity of this culture. Specific lysis between 17% and 50% of tyrosinase loaded T2 cells was observed by transduced donor B, D, and F cells at 100:1 E/T ratio. Bulk cultures were also cultured with ⁵¹Cr labeled HLA-A2⁺ and HLA-A2⁻ melanoma lines 624 and 624-28, respectively, with specific lysis of the HLA-A2⁺ melanoma between 8% and 20% at 100:1 E/T ratio (Fig. 6B). Selected CD8⁺ clones were also tested for CTL capability by coculture with peptide loaded T2 cells and melanoma lines as above (Fig. 6C). Clones V8C4 and V8B11 were both able to specifically lyse tyrosinase loaded T2 cells and HLA-A2⁺ 624 Mel.

From these data, we concluded that through transduction of primary human T cells with the A9.3 retrovirus encoding the TIL 1383I TCR, we can engineer antigen-specific T-cell responses to melanoma in an HLA-A2 restricted manner. The resulting transduced cultures contain both CD8⁺ and CD4⁺ T cells that are highly reactive and can secrete a variety of cytokines, including Th1 and Th2 cytokines, upon interaction with tumor cells and/or kill HLA-A2⁺ melanoma cells.

Recent reports of the functional transfer of TCRs specific for various tumor antigens show the interest in TCR gene transfer for adoptive immunotherapy for cancer (4–6, 10, 11). However, numerous factors must be considered when choosing a TCR for gene transfer. Antigen specificity alone may be insufficient criteria for choosing a TCR. Factors such as MHC restriction of the TCR must be considered as this will define the patient pool to be treated. We have chosen HLA-A2 restricted T cells because roughly 50% of melanoma patients in the United States express this HLA allele. Therefore, an HLA-A2 restricted TCR has the potential to treat the largest number of patients.

Avidity of T cells is important to consider since high functional avidity has been shown to correlate with tumor recognition in vitro (12) and tumor regression in vivo (13). Therefore, it is reasonable to conclude that TCRs expressed by high avidity T cells would be superior to those expressed by low avidity T cells. However, Jurkat and normal T cells expressing the TIL 1383I TCR have intermediate to low avidity for antigen yet still efficiently recognize tumor cells representing an exception to this rule. It is difficult to fully characterize the functional avidity of the T-cell clones tested in this study. Transduced clones stained positive for BV12 yet this may not accurately reflect the number of functional transferred TCRs on the cell surface because the introduced β chain may pair with the endogenous α chain and vice versa, diluting the number of functionally paired TCRs. However, our previous work with a TCR⁻ murine T-cell lymphoma showed that the expression level of introduced TCR did correlate with functional avidity (9). In this case the avidity of the clones tested was not widely variable. Most clones secreted >100 pg/mL cytokine at peptide concentrations between 10 and 100 ng/mL. Clones V8C4 and V8B11 however, seemed to have slightly lower avidity with higher level BV12 expression than clone F8B5 which expressed far lower levels of BV12 (Fig. 4) indicating some variability in clonal response to activation mediated through the transferred TCR. The use of high titer stable retroviral producer cell lines, necessary for any clinical application, should increase the infections titer of retrovirus over the transient transfection method employed in this initial analysis. We are also beginning to investigate modifications of the A9.3 retroviral vector in an attempt to increase retroviral production and functional expression of the TIL 1383I TCR with the goal of increasing avidity.

Another factor in considering a TCR for gene transfer is the dependence on CD8 for antigen recognition. CD8 enhances the stability of the TCR/peptide/MHC complex and promotes signal transduction through the recruitment of the protein kinase lck to the CD3 complex (14). It has been suggested that CD8-independent antigen recognition by T cells indicates expression of a high affinity TCR (15). However, the vast majority of MHC class I restricted T cells are CD8 dependent and would therefore have intermediate to low affinity TCRs. We have previously shown that the TIL 5 TCR could transfer anti-MART-1:27-35 reactivity to CD8⁺ and CD4⁺ T cells. The CD8⁺ T cells were highly avid and recognized peptide loaded antigen-presenting cells and tumor cells, whereas the CD4⁺ T cells were low avidity and recognized peptide loaded antigen-presenting cells only (4). A more recent study described the transfer of a gp100:209-217 reactive TCR to normal T cells (5). The resulting CD4⁺ T cells had high avidity for peptide loaded antigen-presenting cells, but no evidence of tumor cell recognition was provided suggesting this TCR was similar to others that lack sufficient affinity to activate T cells upon encountering physiologic levels of antigen on tumor cells in the absence of CD8. Here, we report both CD4⁺ and CD8⁺ PBL-derived T cells and CD8⁻ Jurkat cells expressing the TIL 1383I TCR can efficiently recognize tumor cells. This CD8-independent tumor cell recognition by TIL 1383I TCR transduced CD4⁺ T cells supports the notion that the TIL 1383I TCR has sufficient affinity for the peptide/MHC complex to provide the necessary signals to fully activate the T cell.

The TIL 1383I TCR has advantages over other TCRs currently being developed for patient treatment. Animal studies have shown the importance of CD4⁺ T helper cells in combating malignancies (16). A role for CD4⁺ cells in cross priming of CD8⁺ effector cells has been shown (17–19), whereas Hung et al. (20) showed a need for both Th1 and Th2 responses for maximal tumor immunity. It has been difficult to confirm these observations in humans however, due in part to the lack of shared helper epitopes and the extreme diversity of the MHC class II genes in humans which limits the number of patients eligible for clinical trials studying T-cell help. One clinical trial has shown the importance of T-cell help for viral clearance whereby adoptively transferred CMV pp65 reactive CTL clones persisted longer in only those patients with detectable anti-CMV T-cell help (21). Other reports have described the importance of CD4⁺ T cells in the generation and maintenance of CD8⁺ T-cell memory (22, 23). Our ability to engineer HLA-A2 restricted CD4⁺ T cells offers the opportunity to further evaluate the role of T-cell help in up to 50% of all melanoma patients regardless of their immune status.

Engineering CD4⁺ T cells to become tumor cell reactive raises a possibility of generating tumor-specific CD4⁺/CD25⁺ regulatory T cells (Treg). These cells exist in the periphery and have been shown to impair the normal proliferative and effector functions of antigen-specific T cells and are believed to play a role in preventing autoimmunity (24). Generating a large number of Tregs specific for cancer cells could negate any benefit obtained from TCR gene transfer. However, T cells must be activated and rapidly proliferating for incorporation of retrovirus into the T-cell genome to occur. Protocols to reliably expand large numbers of Tregs in vitro have been elusive. We thus feel the activation and cell growth observed during the transduction protocol as well as the subsequent reexpansion/selection of transduced cells shows the generation of significant numbers of tumor-specific Tregs to be unlikely. Because our transduction protocol requires T-cell activation, it would be difficult to differentiate Tregs in the transduced bulk culture by simple phenotypic analysis. The possibility of generating tumor-specific Tregs may in fact be an argument against the use of lentiviral vectors for therapeutic TCR gene transfer. Lenitviral vectors have begun to be investigated for gene transfer to T cells precisely because they do not require cell division for viral incorporation (25–27). This would increase the likelihood of transducing TCRs with CD8-independent function into Tregs. In either case, protocols could be developed to remove these cells from the lymphocyte population before transduction. If protocols for expanding Tregs in vitro become more developed the TIL 1383I TCR may provide an interesting opportunity to generate Tregs with known antigen specificity for the study of Treg function.

It is worth noting the majority of the T cells expressing the TIL 1383I TCR secreted GM-CSF but not IFN-γ. Of significance was that both the CD4⁺ and CD8⁺ T-cell clones from all three donors exhibited this cytokine secretion pattern suggesting a feature of the 1383I TCR that may influence the phenotype of the T cells. Whereas most investigators favor T cells secreting type I cytokines such as IFN-γ, immune monitoring of clinical trials have generally failed to find a correlation between IFN-γ and clinical responses. Indeed, clones V8C4 and V8B11 failed to secrete IFN-γ in response to tumor cells yet both were able to lyse tumor cell targets. This is consistent with our previous data involving transduction of a MART-1-specific TCR into normal T cells (4). In that study, some transduced clones lysed tumor cells while also secreting IFN-γ whereas other clones lysed targets while failing to secrete IFN-γ. Still other clones only secreted cytokine and did not lyse tumor cells. However, a correlation between GM-CSF secretion by adoptively transferred TIL and clinical response has been reported (28). GM-CSF secreting tumor cells have been shown to have enhanced immunogenicity in animal models leading to effective antitumor immunity (29) and tumor regression has been observed in clinical trials of GM-CSF secreting tumor cell vaccines (30, 31). Therefore, it is possible that T cells engineered to express the TIL 1383I TCR will not only provide potent antitumor effectors, their secretion of GM-CSF may enhance the endogenous antitumor immunity. In addition, the secretion of the Th2 cytokine IL-4 by some transduced CD4⁺ clones shows we may be able to generate immune responses beyond the function of the original TIL 1383I which was not shown not to secrete IL-4 in response to antigen stimulation (8).

Because the goal of immunotherapy for cancer is to eliminate malignant cells, we consider the CD8-independent tumor cell recognition exhibited by TIL 1383I TCR gene modified T cells to be a critical feature of a cloned TCR. TCRs with this property make it possible to treat patients with MHC class I restricted T helper cells and possibly more potent CTL. Given this TCR was isolated from an MHC class I restricted CD4⁺ T cell, cells bearing this TCR may represent a class of T cell not usually expected to be present in our normal repertoire because they would likely be clonally deleted or rendered immunologically tolerant during T-cell development.

Grant support: NIH grant CA90873, CA97296, CA100240, and CA102280 (M.I. Nishimura).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.