Reprogramming of Virus-specific T Cells into Leukemia-reactive T Cells Using T Cell Receptor Gene Transfer

Donor lymphocyte infusion (DLI) into patients with a relapse of their leukemia or myeloma after allogeneic stem cell transplantation has been proven to be a successful treatment strategy (1–3). The beneficial graft versus leukemia effect of donor lymphocytes, however, is often accompanied by GVHD. Adoptive transfer of antigen-specific T cell lines may eradicate the relapsed hematological malignancy and can separate the antileukemic effect from GVHD (4). The main drawback of adoptive therapy is the difficulty to produce sufficient quantities of antigen-specific T cells. In addition, the specificity of the infused T cells is difficult to control.

An alternative strategy might be to equip donor T cell populations with a TCR of defined antitumor specificity. Dembic et al. (5) were the first to show that transfer of the TCR into recipient T cells resulted in redirected antigen specificity of these T cells. Since then several groups have demonstrated that transfer of virus-specific or tumor antigen–specific TCRs resulted in T cell populations with redirected antigen specificity (6–12). Transfer of TCRs specific for class I–restricted antigens into CD8⁺ T cells and transfer of TCRs of class II–restricted specificity into CD4⁺ T cells leads to functional redirected T cells, exerting antigen-specific cytolytic activity as well as antigen-specific cytokine secretion. Stanislawski et al. (11) demonstrated that TCR gene transfer can be used to circumvent self-tolerance of autologous T cells to tumor-associated antigens by transferring a humanized MDM2-specific TCR selected from HLA-A201 transgenic mice. Furthermore, TCR transfer studies have demonstrated that retroviral transfer of high avidity TCRs into T lymphocytes is accompanied by the maintenance of the parental T cell avidity and that TCR-transferred T cells exhibit no change in the peptide-fine specificity in comparison to the parental T cell clones (12, 13). Recently, the feasibility of cellular therapy using TCR-modified T cells was demonstrated in a mouse model system (14). In this study it was demonstrated that T cells that were redirected by TCR gene transfer can be activated in vivo, home to effector sites, and contribute to tumor clearance.

Although the approach of TCR transfer appears to be feasible, various potential problems may occur. First, the pairing of the retroviral-introduced TCR chains with the endogenously expressed TCR chains can lead to the formation of new TCR complexes with undesired specificities. Modification of highly polyclonal T cell populations will result in a diverse set of new specificities. This may lead to the production of T cells that have self-reactive TCRs that may lead to autoimmunity. Second, TCR transfer of ignorant self-specific T cells may lead to autoaggressive T cells after triggering through the exogenous TCR. In addition, if antigen-specific TCRs are transferred into unselected primary T cells not only might a potentiating immune response develop, but also suppressive T cell reactivity may develop, due to TCR transfer into naive T cells or suppressor T cells. Furthermore, end-stage effector cells with limited proliferative capacity that will be transduced with the exogenous TCR may only exhibit short-term cytolytic activity.

Based on these various reasons, we propose to redirect the specificity of virus-specific T cells toward antileukemic reactivity using TCR gene transfer. T cells directed against viruses with a latent persistence in vivo like CMV and EBV would be attractive candidates to use. These virus-specific T cells will have an additive advantage because they will also be triggered by the endogenous virus-specific TCR and therefore we hypothesize that these T cells will survive for prolonged periods of time in vivo.

In this study we redirected virus-specific T cells into leukemia-reactive T cells by the transfer of TCRs derived from T cell clones specific for the minor histocompatibility antigen (mHag) HA-2. mHags are derived from genetically polymorphic proteins that can be differentially expressed between donor and recipient (15). Because the HA-2 mHag is exclusively expressed on cells of the hematopoietic lineages (16), T cells directed against HA-2 can be used to treat hematological malignancies relapsing after allogeneic stem cell transplantation when the patient is HA-2⁺ and the donor is HA-2⁻ (17). T cells directed against HA-2 will eliminate the hematopoietic cells from the patient, including the malignant cells, without affecting donor-derived hematopoiesis or directly affecting nonhematopoietic tissue of the patient.

Because a high percentage of the human population express the HA-2 (18), this mHag is not frequently useful for induction of HA-2–specific T cells in transplantations between HLA-identical individuals. However, in HLA-A2 mismatch transplants in which patient hematopoietic cells are HLA-A2⁺ and express the mHag HA-2, whereas the donor cells do not express the HLA-A2 restriction molecule, donor T cell responses might be generated against mHag HA-2 peptide presented in HLA-A2. Several studies have shown the in vitro generation of antigen-specific T cells that are restricted by nonself HLA molecules (19–23). However, stimulation with nonself HLA molecules also induced undesired allo-HLA–specific T cells, which can cause damage of all HLA-A2⁺ hematopoietic and nonhematopoietic tissues (20, 24). Transfer of mHag-specific TCRs into virus-specific T cells from the donor would be a strategy to circumvent the undesired induction of allo-HLA–specific T cells. Because the mHag-specific TCRs were isolated from T cells exhibiting an immune response in an HLA-identical setting, the risk of transfer of TCRs that not only have mHag specificity but also anti–HLA-A2 reactivity will be minimized.

In this study we explored the possibility to transfer HA-2–specific TCRs into various HLA-restricted virus-specific T cells derived from HLA-A2⁺ and HLA-A2⁻ individuals. We demonstrated that HA-2-TCR–transduced virus-specific populations exerted significant virus-specific as well as antileukemic reactivity without the appearance of anti–HLA-A2 reactivity.

The HLA-A2–restricted T cell clones HA2.5 and HA2.27 specific for the mHag HA-2 were isolated from the peripheral blood of a patient with chronic myeloid leukemia (CML) during an ongoing graft versus leukemia response after DLI (17). The allo-HLA-A2–specific T cell clone MBM13 was derived from a mixed lymphocyte reaction with peripheral blood from haploidentical siblings by limiting dilution analysis (9). The HLA-A2⁺ HFF fibroblast cell line was derived from the American Type Culture Collection (ATCC). COS and 293 EBNA cell lines (ATCC) were transduced with a retroviral vector encoding for the HLA-A2 molecule (9). Mesenchymal stem cells (MSCs) were derived from the bone marrow of CML patients by culturing the adherent cells for several weeks on DMEM supplemented with l-alanyl-l-glutamine, sodium pyruvate, 1 mg/ml glucose, pyridoxine (GIBCO BRL), and 10% FBS.

By RT-PCR, using primers that cover the complete repertoire of known TCR chains (25), the TCR α and β usage of the mHag HA-2–specific HA2.5 was determined. The T cell clone expressed two in-frame gene transcripts, TCR AV15S1 and TCR BV18S1. The different TCR α and β chains of the T cell clone were individually cloned into retroviral vectors. The Moloney murine leukemia virus–based retroviral vector LZRS and packaging cells φ-NX-A were used (26). Two bicistronic retroviral vectors were constructed in which the multiple cloning site is linked to the downstream internal ribosome entry sequence and the marker gene enhanced green fluorescent protein (eGFP; reference 27) or truncated form of the nerve growth factor receptor (ΔNGF-R; reference 28). The TCR α chain was cloned into the retroviral vectors in combination with eGFP and the TCR β chain was cloned in combination with the ΔNGF-R. Retroviral vectors encoding eGFP or ΔNGF-R alone were used as control vectors in the experiments. In addition, the gene product of the lower matrix protein pp65 of HCMV AD169 (provided by E. Wiertz, Leiden University Medical Center, Leiden, Netherlands) was cloned into the retroviral vector in combination with eGFP. The constructs were transfected into φ-NX-A cells using calcium phosphate (Life Technologies) and 2 d later 2 μg/ml puromycin (CLONTECH Laboratories, Inc.) was added. 10–14 d after transfection, 6 × 10⁶ cells were plated per 10-cm Petri dish in 10 ml IMDM supplemented with 10% FBS without puromycin. The next day the medium was refreshed and the day after retroviral supernatant was harvested, centrifuged, and frozen in aliquots at −70°C.

Virus-specific T cells were isolated from the peripheral blood of healthy individuals. After informed consent PBMCs were harvested and labeled with tetrameric complexes for 2 h at 4°C in RPMI without phenol, supplemented with 2% FBS, washed three times, and sorted at 4°C using the FACS Vantage™ (Becton Dickinson). Tetramer⁺ T cells were stimulated with either 800 ng/ml phytohemagglutinin (PHA; Murex Diagnostics) and 100 IU/ml IL-2 (Chiron Corp.), or with virus peptide–pulsed autologous PBMCs and 100 IU/ml IL-2. The autologous PBMCs were pulsed with 10 μM CMV peptide for 1 h, irradiated (30 Gy), and washed, and 10⁶ peptide-loaded autologous PBMCs were added to each well of a 24-well plate containing 2 × 10⁵ virus-specific T cells. After 2 d of culture, the T cells were transduced with retroviral supernatant. The transduction procedure used for the peripheral blood T cells was based on the use of recombinant human fibronectin fragments CH-296 (29) and has been described (27). In brief, 10⁶ T cells were cultured on CH-296–coated 24-well nonculture-treated plates (Falcon) together with 1 ml thawed retroviral supernatant for 6 h or overnight at 37°C, washed, and transferred to 24-well tissue culture plates. The transduction efficiency as measured by the expression of the markers eGFP and ΔNGF-R was analyzed by flow cytometry 3–5 d after transduction. In addition, flow cytometric analyses were performed with the relevant tetrameric complexes. HA-2-TCR–transduced virus-specific T cells were FACS^® sorted on bases of marker gene expression or HA-2 tetramer⁺ staining, and cultured at 1 cell/well (clones) or 25 cells/well (cell lines) in IMDM supplemented with 10% human serum. Sorted T cells were nonspecifically stimulated every 2 wk with feeder cell mixtures containing irradiated allogeneic PBMCs (30 Gy), irradiated EBV-transformed B cells (EBV-LCLs; 50 Gy), 800 ng/ml PHA, and 100 IU/ml IL-2, or stimulated with virus peptide–pulsed irradiated autologous PBMCs (30 Gy) and 100 IU/ml IL-2.

PE- or APC-conjugated tetrameric complexes were constructed as previously described (30) with minor modifications. Tetrameric HLA-A2 molecules in complex with CMV pp65–derived peptide NLVPMVATV (CMV^A2 tetramer) and HA-2–derived peptide YIGEVLVSV (HA-2^A2 tetramer) were constructed. In addition, tetrameric HLA-B7 molecules in complex with CMV pp65–derived peptide TPRVTGGGAM (CMV^B7 tetramer) were constructed. For flow cytometric analyses as well as FACSort™ experiments, cells were labeled with tetrameric complexes for 2 h at 4°C in RPMI without phenol, supplemented with 2% FBS, and washed three times. During the last 30 min, mAbs directed against the various cell surface molecules were added. The mAbs used were anti-CD4 (FITC), anti-CD8 (FITC or PECy5), anti-CD40 (FITC), and anti–NGF-R (PE; all from Becton Dickinson).

Target cells were harvested, labeled with 50 μCi Na₂⁵¹ CrO₄ for 60 min at 37°C, washed three times, and added to various numbers of effector T cells in a final volume of 150 μl IMDM supplemented with 10% FBS in 96-well U-bottomed microtiter plates. HA-2⁺ and HA-2⁻ EBV-LCLs, HFF, COS, 293 EBNA cells, and leukemic cells harvested from patients after informed consent were used as target cells. In addition, EBV-LCLs transduced with a retroviral vector encoding for the gene product of the lower matrix protein pp65 of HCMV AD169 were used (EBV-Z pp65). In some experiments, ⁵¹Cr-labeled target cells were loaded by preincubation for 1 h at 37°C with different concentrations of synthetic peptides, washed, and added to the effector cells. After 4 or 18 h of incubation of target and effector cells at 37°C and 5% CO₂, 25 μl supernatant was harvested and measured in a luminescence counter (Topcount-NXT; Packard Instrument Co.). The mean percentage of specific lysis of triplicate wells was calculated as follows: specific lysis = [(experimental release − spontaneous release)/(maximal release − spontaneous release)] × 100.

The PIA was performed as previously described (31, 32). T cells at a concentration of 10⁴ cells/well were cocultured with 10⁴ cells/well of cell suspensions containing malignant hematopoietic progenitor cells (HPCs) in 96-well U-bottomed microtiter plates in a volume of 200 μl IMDM supplemented with 10% FBS and multiple hematopoietic growth factors. T cells were irradiated with 15 Gy before use, preventing their proliferation in the PIA. HPCs collected from different CML patients were used as target cells. After 5 d of culture, the wells were pulsed with 1 μCi [³H]thymidine for 18 h. Cells were harvested and the incorporated [³H]thymidine was measured by luminescence counter to determine the capacity of the T cell clones to inhibit the growth of the HPCs. The mean percentage of growth inhibition of triplicate wells was calculated as follows: [(1 − experimental cpm/mean cpm HPC only) × 100%].

To determine whether virus-specific T cells could be isolated, retrovirally transduced and expanded HLA-A2–restricted CMV pp65–specific T cells (CMV^A2-specific T cells) were isolated from the peripheral blood of HLA-A2⁺ HA-2⁻ individuals by FACSort™ using CMV^A2 tetrameric complexes, and T cells were subsequently stimulated with PHA, IL-2, and allogeneic feeder cells. 2 d after stimulation, the T cells were transduced with the HA-2-TCR complex derived from the HA-2–specific HA2.5 T cell clone, and sorted at day 5 on the basis of HA-2^A2 tetramer positivity. In addition, T cells were transduced with the control retroviral vectors and sorted on the basis of marker gene expression. T cells were sorted in bulk or single cell and expanded by nonspecific stimulation. Fig. 1 shows the HA-2^A2 and CMV^A2 tetramer staining of the HA-2-TCR–transduced and control-transduced CMV^A2-specific T cell lines 1 wk after sorting. The results demonstrated that the CMV^A2-specific T cells transduced with the HA-2-TCR and sorted on the basis of HA-2^A2 tetramer staining expressed both TCR complexes at the cell surface. The mean fluorescence intensity (MFI) of the CMV^A2 tetramer staining was lower on the HA-2-TCR–transduced T cells compared with the control-transduced T cells. The MFI decreased from 265 on control-transduced to 86 on HA-2-TCR–transduced CMV-specific T cells, indicating competition between the two TCR complexes for cell surface expression.

To determine whether the HA-2-TCR–transduced CMV^A2-specific T cells were able to recognize the HA-2 mHag as well as CMV pp65 antigen, we analyzed the cytotoxicity against a panel of different target cells. As demonstrated in Fig. 2 A, the HA-2-TCR–modified CMV^A2-specific T cells exerted cytolytic activity against HA-2–expressing EBV-LCLs as well as CMV^A2 peptide–loaded EBV-LCLs. In contrast, control-transduced CMV^A2-specific T cells exerted only cytolytic activity against CMV^A2 peptide–loaded target cells. By peptide titration on T2 cells we illustrated that the CMV^A2 peptide reactivity of the HA-2-TCR–modified T cell populations did not significantly differ from the CMV^A2 peptide reactivity of the control-transduced T cells (Fig. 2 B). The HA-2-TCR–transferred T cells displayed similar cytolytic efficiencies toward HA-2 peptide–loaded target cells as the original HA-2–specific T cell clone HA2.27. These results demonstrated the dual specificity of the HA-2-TCR–modified CMV^A2-specific T cells.

In addition to the HA-2-TCR–modified T cell lines, we generated HA-2-TCR–modified CMV-specific T cell clones to analyze the expression of the two TCR complexes in more detail, and to correlate this with the functional activity of the individual TCR-transferred T cell clones. As demonstrated in Fig. 3, the HA-2-TCR–modified CMV-specific T cell clones exhibited differences in HA-2^A2 and CMV^A2 tetramer staining. Some HA-2-TCR–modified T cell clones preferentially expressed the HA-2–specific TCR (type A), whereas other T cell clones preferentially expressed the CMV-specific TCR (type C). Interestingly, the majority of the T cell clones showed intermediate expression of both TCR complexes (type B). Although the HA-2^A2 and CMV^A2 tetramer staining in the TCR-transferred T cell clones varied, the overall TCR α and β expression was similar, indicating that the composition of cell surface TCRs containing the endogenous and introduced TCR chains was variable in the different T cell clones, and not the absolute numbers of TCRs expressed. These results indicated that the different TCR complexes competed for cell surface expression. Importantly, the expression of the HA-2-TCR and CMV-TCR complexes on the different T cell clones as measured by HA-2^A2 and CMV^A2 tetramer staining, respectively, was stable during a culture period of 2 mo. Furthermore, as illustrated in Table I, which shows representative examples of each clone type, the lytic activity of the dual-specific T cell clones correlated in general with the TCR expression. Type A and type B T cell clones exerted efficient HA-2–specific cytolytic activity directed against HA-2 peptide–loaded as well as endogenous HA-2–expressing target cells. Importantly, these type A and type B T cells were CMV specific because CMV pp65 peptide–loaded target cells as well as target cells endogenously expressing the CMV pp65 protein were recognized by these TCR-transduced T cell clones (Table I), demonstrating the dual specificity of these T cell clones. In agreement with the bright CMV^A2 tetramer staining, type C T cells exerted efficient cytolytic activity directed against the CMV pp65⁺ target cells. Although only minimal HA-2-TCRs could be visualized by HA-2^A2 tetramers on these type C T cell clones, T2 cells loaded with HA-2 peptides could be lysed by these T cells (Table I).

To predict whether stimulation in vivo with CMV-infected cells would lead to the stimulation and expansion of all three types of TCR-transferred virus-specific T cells, HA-2-TCR–transferred CMV-specific T cells were stimulated twice in vitro with CMV pp65–expressing stimulator cells over a period of 30 d. As demonstrated by tetramer staining shown in Fig. 4, the total population of HA-2-TCR–transduced CMV-specific T cells expressed after 30 d marked levels of CMV-TCRs as well as HA-2-TCRs at the cell surface. Stimulation of the HA-2-TCR–transferred CMV-specific T cells with stimulator cells endogenously expressing the HA-2 mHag also resulted in expansion of the TCR-transduced CMV-specific T cells. Tetramer staining demonstrated the marked expression of both TCRs at the cell surface, with a tendency of lower levels of CMV-TCR expression. In parallel we stimulated control-transduced CMV-specific T cells with stimulator cells either expressing the CMV or the HA-2 antigen. Stimulator cells endogenously expressing the HA-2 mHag were unable to stimulate and expand the control-transduced CMV-specific T cells, whereas stimulator cells expressing CMV pp65 induced expansion of the control-transduced CMV-specific T cells. In a cytotoxicity assay we demonstrated that the HA-2-TCR–transferred T cells stimulated with CMV pp65–expressing stimulator cells exerted, comparable to the HA-2-TCR–transferred T cells stimulated with HA-2–expressing stimulator cells, specific cytolytic activity directed against HA-2–expressing target cells as well as CMV pp65–expressing target cells (Table II), illustrating that the dual specificity of the TCR-transferred virus-specific T cells is preserved after stimulation of either the endogenous or the introduced TCR.

To determine whether the mHag HA-2 specificity of the HA-2-TCR–modified CMV-specific T cells could be translated into antileukemic reactivity, the TCR-transferred T cells were tested against a panel of CML cells in a cytotoxicity assay. The HA-2-TCR–modified CMV-specific T cell clones were able to exert cytolytic activity against HA-2–expressing CML cells and not against HA-2⁻ CML cells (Fig. 5 A). The cytolytic activity of the TCR-transferred T cells was for most TCR-transferred T cell clones as efficient as the original HA-2–specific T cell clone HA2.27. The HA-2–specific TCR density, measured by HA-2^A2 tetramer staining, correlated with the ability of the TCR-modified T cell clones to lyse the HLA-A2⁺ HA-2⁺ leukemic cells. In addition, we tested two HA-2-TCR–modified CMV-specific T cell clones in the liquid hematopoietic PIA, and demonstrated that these T cell clones were able to specifically inhibit the growth of HLA-A2⁺ CML progenitor cells expressing the HA-2 mHag, and not the growth of HLA-A2⁺ HA-2⁻ or HLA-A2⁻ CML progenitor cells (Fig. 5 B). The specific growth inhibition of the HA-2-TCR–transferred CMV-specific T cells was as efficient as the original T cell clone HA2.27.

To explore the possibility of an alternative strategy to treat hematological malignancies of HLA-A2⁺ HA-2–expressing patients transplanted with HLA-A2⁻ donors, we isolated CMV-specific T cells from an HLA-A2⁻ individual and transferred these T cells with HA-2-TCRs. For this purpose CMV-specific HLA-B7–restricted T cells (CMV^B7-specific T cells) were FACS^® sorted using CMV^B7 tetramers stimulated with CMV^B7 peptide–loaded, irradiated autologous PBMCs and transduced with the HA-2-TCR derived from the HA-2–specific T cell clone HA2.5, or with control retroviral vectors. By FACS^® analyses we demonstrated that a high percentage of CMV^B7-specific T cells was transduced (49–58%) and that the majority of HA-2-TCR α and β chain transduced T cells stained with the HA-2^A2 tetramer (Fig. 6). Similar to the HA-2-TCR–transduced CMV^A2-specific T cells, the CMV^B7 tetramer staining was decreased on the HA-2-TCR–transduced CMV^B7-specific T cells compared with the control-transduced CMV^B7-specific T cells. The double marker gene⁺ CMV^B7-specific T cells transduced with either the HA-2-TCR or with the control vectors were sorted and cultured for an addition 10 d in IMDM supplemented with 10% human serum and 100 IU/ml IL-2. The expanded HA-2-TCR–transferred CMV^B7-specific T cells were tested in a cytotoxicity assay against a panel of target cells. As can be seen in Fig. 7 A, HA-2-TCR–transduced T cells were able to efficiently lyse the HA-2–expressing EBV-LCLs (EBV-RZ) and not the HA-2–negative EBV-LCLs (EBV-Z). To investigate whether the HA-2-TCR–transduced CMV^B7-specific T cells exerted dual HA-2 and CMV specificity, similar to the HA-2-TCR–transferred CMV^A2-specific T cells, HLA-B7–transduced EBV-Z were loaded with the CMV^B7 peptide or were additionally transduced with the pp65 gene of CMV. As can be seen in Fig. 7 A, the HLA-B7⁺ EBV-Z target cells either loaded with CMV^B7 peptide or endogenously expressing the pp65 protein of CMV were efficiently lysed by the control-transduced as well as the HA-2-TCR–transduced CMV^B7-specific T cells, illustrating that the HA-2-TCR–transduced CMV^B7-specific T cells exerted dual specificity. In addition, HLA-A2⁺ CML cells expressing the HA-2 mHag (CML-T) were efficiently lysed by the HA-2-TCR–modified T cells (Fig. 7 B). In contrast, HLA-A2⁺ CML cells negative for HA-2 or HLA-A2⁻ CML cells were not lysed by the HA-2-TCR–modified T cells. No anti–HLA-A2 alloreactivity was observed because HLA-A2⁺ target cells negative for HA-2, including the HLA-A2⁺ HFF fibroblast cell line, HLA-A2–transduced COS and EBNA cells, and the HLA-A2⁺ MSCs derived from the patients with CML (MSC-T and MSC-Z), were not lysed by the HA-2-TCR–transduced CMV^B7-specific T cells (Fig. 7 C). In contrast, the allo-HLA-A2–restricted control T cell clone MBM13 lysed all of these HLA-A2–expressing target cells efficiently. In agreement with the expression pattern of the HA-2 mHag, the MSCs of patient T were not lysed by the HA-2–specific T cells because MSCs in contrast to the cells derived from the hematopoietic lineage do not express the HA-2 mHag. These results illustrate that reprogramming of virus-specific T cells into antileukemic reactive T cells by transfer of mHag-specific TCRs without the occurrence of allo-HLA reactivity is feasible.

In this study we demonstrated the efficient reprogramming of virus-specific T cells into leukemia-reactive T cells using TCR gene transfer. HLA-A2–restricted as well as HLA-B7–restricted CMV-specific T cells derived from HLA-A2⁺ and HLA-A2⁻ individuals transferred with the HA-2-TCR were able to exert efficient antileukemic reactivity, without signs of anti–HLA-A2 alloreactivity. The virus-specific T cells were isolated by tetrameric complexes, expanded by specific or aspecific stimulation, and transferred with the genes encoding for the TCR directed against the mHag HA-2 in the context of HLA-A2. The CMV-specific T cells isolated by CMV^A2 or CMV^B7 tetrameric complexes were >95% pure, could be efficiently transduced (40–60%), and expanded vigorously. The HA-2–specific cytolytic activity of most HA-2-TCR–transferred T cell populations was similar to the original HA-2–specific T cell clones and in addition, the TCR-redirected virus-specific T cells exerted efficient cytolytic activity against CMV peptide–loaded target cells or target cells that endogenously express the CMV pp65 protein, demonstrating the dual specificity of the HA-2-TCR–transferred CMV-specific T cells.

No anti–HLA-A2 alloreactivity was observed because HLA-A2⁺ nonhematopoietic cell lines, HA-2⁻ HLA-A2⁺ EBV-LCLs, and HLA-A2⁺ MSCs of patients from which the HA-2⁺ CML cells were derived were not recognized by the HA-2-TCR–transferred T cells. The HA-2-TCR–transferred T cells only exerted specific CMV as well as leukemia reactivity. Therefore, we speculate that these T cells will be useful in HLA mismatch transplantations because after these transplantations administration of unmodified HLA class I disparate DLI has a high risk of inducing GVHD.

Regardless of whether the HA-2-TCR–transferred virus-specific T cells were stimulated with HA-2 or CMV, no selective decrease in the HA-2-TCR expression was observed. Based on these results and the latent persistence of herpes viruses like CMV and EBV in vivo, we hypothesize that the TCR-transferred dual specific T cells exerting both leukemia as well as virus specificity will be triggered continuously with low doses of viral antigens via their endogenous virus-specific TCR in vivo. Therefore, we speculate that due to this low dose triggering of these TCR-transferred dual specific T cells, the T cells will survive for prolonged periods of time in vivo. Furthermore, we hypothesize that by virus-specific peptide vaccinations we will be able to promote the survival of these leukemia-reactive, TCR-transferred virus-specific T cells in vivo.

During CMV reactivation, there might be preferential proliferation of TCR-transduced T cells with the highest activity against CMV. Although the TCR-transduced CMV-specific T cells with the highest HA-2 activity may not expand as vigorously during CMV disease, we speculate that these T cells will persist in the patient and during reappearance of the malignancy, these T cells with the highest HA-2–specific avidity are likely to preferentially expand.

Because in vivo the compartments in which specific viral antigens are expressed may not be identical to the site of the malignancy, the homing characteristics of the engineered T cells might be a relevant factor. By selecting for TCR transfer T cells specific for not only CMV but also for other target antigens like EBV, reprogrammed T cells might be directed to several tissues in search of the malignancy.

An advantage of using virus-specific T cells for TCR gene transfer is that these polyclonal T cell responses consist of a restricted TCR repertoire. The formation of mixed TCR dimers due to pairing of the endogenous TCR chains with the introduced TCR chains that may have unpredictable specificities possibly leading to harmful immune reactions, will therefore be limited. In addition, TCR transfer of virus-specific T cells will minimize the induction of autoaggressive T cells that may develop due to triggering of ignorant self-specific T cells through their introduced TCR.

Furthermore, if antigen-specific TCRs are transferred into unselected primary T cells not only might a potentiating immune response develop, but also suppressive T cell reactivity. If naive T cell populations that are transduced with the specific TCR encounter their stimulatory antigen in vivo in the context of nonprofessional antigen-presenting cells, the T cell response may abort or antigen-specific suppression may even occur. In addition, if circulating suppressor T cells are transduced with the specific TCR, suppression of the desired T cell response may develop. Furthermore, end-stage effector cells with limited proliferative capacity that will be transduced with the exogenous TCR may only exhibit short-term cytolytic activity. In contrast, virus-specific memory T cells are likely to expand rapidly in vivo, and are probably destined to undergo multiple cell divisions. Therefore, as host cells for TCR gene transfer, it may be desirable to select for memory T cell populations that apparently have been successfully triggered in the past by professional antigen-presenting cells, and are capable of further large scale expansion in response to specific activation.

Transfer of TCR α and β chains into T cells results in cell surface expression of the introduced TCR, the endogenous TCR, and mixed TCR dimers, which consist of endogenous and exogenous TCR chains. If the individual TCR α and β chains pair with similar affinity to each other and no competition for cellular components between the individual TCR α and β chains occurs, the introduced TCR and endogenous TCR density on individual T cells would be ∼25% of the total TCR expression for each TCR complex. In addition, ∼50% of the total TCR αβ expression would consist of mixed TCR dimers. The majority of the HA-2-TCR–transferred T cell clones demonstrated intermediate expression of both the virus- and the HA-2–specific TCR complexes, representative for the type B T cell clones (Fig. 3). Because the TCR αβ cell surface expression on these HA-2-TCR–transferred T cell clones is similar to the TCR αβ expression of the type A and type C TCR-transferred T cell clones, these results indicated that formation of chimeric TCR complexes is occurring. Furthermore, we demonstrated that the cell surface expression of the introduced HA-2-TCR complex in relation to the endogenous CMV-specific TCR varied between T cells. This might be due to differences in expression levels of the retroviral vectors encoding for the TCR α and β chains or due to intrinsic properties of the TCR-transferred T cells. Preferential pairing of particular TCR α and β chains might be responsible for the variations in TCR expression. Hence, further selection of certain virus-specific T cells with a particular TCR α and β chain usage unable to pair with the introduced mHag-specific TCR would then be relevant.

In conclusion, these results demonstrated that it is possible to redirect the specificity of virus-specific T cells toward antileukemic reactivity. Based on the dual specificity of these TCR-redirected virus-specific T cells and the minimization of autoimmunity, TCR transfer of virus-specific T cells will open new possibilities for the treatment of hematological malignancies of HLA-A2⁺ HA-2–expressing patients transplanted with HLA-A2–matched or –mismatched donors.

The authors thank Linda Cox, Reinier van der Linden, and Maarten van de Keur for expert technical assistance, and Edith van der Meijden for providing the HLA-A2–transduced COS and 293 EBNA cells.