Human γδ T cells as mediators of chimaeric-receptor redirected anti-tumour immunity

Authors


Claudia Rossig, Department of Paediatric Haematology and Oncology, University Children's Hospital Münster, Albert-Schweitzer-Str. 33, D-48149 Münster, Germany.
E-mail: rossig@uni-muenster.de

Summary

Human peripheral blood γδ T cells (Vγ9+ Vδ2+) can be selectively expanded in vivo by the systemic administration of aminobisphosphonates without prior antigen priming. To assess the potential of human γδ T cells to serve as effector cells of specific anti-tumour immunity, we expanded peripheral blood-derived γδ T cells and transduced them with recombinant retrovirus encoding GD2- or CD19-specific chimaeric receptors. Flow cytometric analysis of T cells from four individual donors cultured in the presence of zoledronate at day 14 of culture showed selective enrichment of the γδ T cell population (Vγ9+ Vδ2+ CD3+ CD4 CD8) to 73–96% of total CD3+ T cells. Retroviral gene transfer resulted in chimaeric receptor surface expression in 73 ± 12% of the population. Transduced γδ T cells efficiently recognized antigen-expressing tumour cell targets, as demonstrated by target-specific upregulation of CD69 and secretion of interferon-α. Moreover, transduced γδ T cells efficiently and specifically lysed the antigen-expressing tumour targets. They could be efficiently expanded in vitro and maintained in culture for prolonged periods. Zoledronate-activated human γδ T cells expressing chimaeric receptors may thus serve as potent and specific anti-tumour effector cells. Their responsiveness to stimulation with aminobisphosphonates may enable the selective re-expansion of adoptively transferred T cells in vivo, permitting long lasting anti-tumour immune control.

The targeting of cancer cells with adoptively transferred specific T cells has high potential for effective immunotherapy of cancer. Clinically administered antigen-specific T cells have been shown to specifically and efficiently eradicate Epstein–Barr virus (EBV)-infected target cells in post-transplant lymphoproliferative disease (Rooney et al, 1995, 1998; Roskrow et al, 1998). Attempts to extend the use of this strategy to non-viral malignancies have been hampered by the poor immunogenicity of most cancers that allows them to evade major histocompatibility complex (MHC)-restricted T cell-mediated immune recognition. Redirecting T cells to tumour surface antigens by genetic modification with tumour-specific chimaeric T cell receptors (chRec) is one means of bypassing immune resistance (Sadelain et al, 2003). Chimaeric receptors combine antigen recognition and signal transduction in a single molecule by linking antibody-derived ligand-binding domains to signalling components of the T cell receptor (TCR). ChRec-modified T cells have shown therapeutic potential in model systems (Kershaw et al, 2002) and were shown to lack significant toxicity in phase I clinical studies (Hwu et al, 1993; Altenschmidt et al, 1997; McGuinness et al, 1999). However, in vivo functional persistence by adoptively transferred T cells is limited, probably due to the inadequacy of chimaeric receptor stimulation to induce activation responses in gene-modified T cells in vivo (Brocker & Karjalainen, 1995; Brocker, 2000; Rossig et al, 2002). Indeed, the first clinical studies have failed to show objective clinical responses in patients receiving transfusions of chimaeric receptor gene-modified T cells (Mitsuyasu et al, 2000).

Defining a long-lived effector T cell population that specifically recognizes tumour cells via stable markers would be very useful for providing therapeutic T cell surveillance in haematopoietic and other malignancies. T lymphocytes bearing γδ TCRs represent a unique population of T cells, which constitute a small proportion (1–5%) of the lymphocytes that circulate in human blood. In contrast to T cells bearing αβ TCRs, γδ T cells have the potential for polyclonal expansion without prior priming (Morita et al, 1995). Whereas activation via αβ TCRs requires interaction with receptor-specific antigenic peptides bound to surface protein encoded by MHC genes, antigen recognition by γδ TCRs is not constrained by the requirement to bind MHC, thereby allowing γδ T cells to recognize a wide array of protein and non-protein antigens (Schild et al, 1994; Burk et al, 1995; Morita et al, 1995; Tanaka et al, 1995). Due to structural homologies with γδ TCR ligands, aminobisphosphonate compounds (pamidronate, zoledronate), which are effectively used to treat conditions with excessive bone resorption (Berenson et al, 1996), can induce the activation and interleukin-2 (IL-2)-dependent proliferation of peripheral blood γδ T cells in vitro and in vivo (Kunzmann et al, 2000). Based on the observation that the survival rates in multiple myeloma are improved in patients receiving pamidronate in addition to salvage chemotherapy (Berenson et al, 1996, 1998; Dhodapkar et al, 1998; McCloskey et al, 1998), it was hypothesized that selective expansion of γδ T cells may contribute to the anti-tumour effects of these drugs. A pilot study to evaluate the concomitant use of pamidronate and low-dose IL-2 in patients with multiple myeloma and low-grade Non-Hodgkins lymphoma was recently performed and showed significant in vivo activation and proliferation of γδ T cells in five of nine patients (Wilhelm et al, 2003). Thus, systemic administration of aminobisphosphonates provides a means of selectively stimulating a T cell subset in vivo in the absence of specific target recognition and costimulation.

Here, we determined the potential of human peripheral blood γδ T cells to exert tumour-specific effector functions by transducing them with chimaeric receptors specific for tumour antigen. γδ T cells could be genetically engineered to specifically recognize tumour cells and commit to target-induced cytokine release and cytotoxicity. ChRec-transduced γδ T cells may represent a renewable source of tumour-specific effector T cells for adoptive immunotherapy of cancer that can be selectively re-expanded by systemic administration of aminobisphosphonates.

Methods

Cell lines and antibodies

The amphotropic packaging cell line Phoenix (Kinsella & Nolan, 1996) was provided by Gary P. Nolan (Stanford, CA, USA). PG-13 [obtained from American Type Culture Collection (ATCC), Manassas, VA, USA] is a retrovirus packaging cell line that produces virus pseudotyped with the gibbon-ape leukaemia virus (GALV). The neuroblastoma cell line LAN-1 was provided by Robert C. Seeger's laboratory (UCLA; University of California, Los Angeles), and JF by Malcolm Brenner, Houston, USA. K-562 (ATCC) is a human erythroleukaemia line that is sensitive to lysis by lymphokine-activated killer (LAK) cells. Raji and Daudi (ATCC) are Burkitt's lymphoma cell lines, and Reh (ATCC) is an acute lymphocytic leukaemia cell line. A-204 (ATCC) is a human rhabdomyosarcoma cell line. Lymphoblastoid cell lines (LCLs) were generated by infection of peripheral blood-derived mononuclear cells (PBMC) with supernatant from the EBV producer cell line B95-8. Anti-14.G2a idiotypic antibody 1A7 (Sen et al, 1997) was provided by Titan Pharmaceuticals Inc., South San Francisco. The mouse antihuman TCR Vγ9-specific antibody clone 7A5 was a kind gift of Dieter Kabelitz, Kiel, Germany.

Chimaeric receptor genes

The chimaeric receptor gene CD19ζ contains the variable domains of monoclonal antibody FMC-63 (Nicholson et al, 1997) assembled as a single-chain Fv (scFv) molecule, cloned in frame with a sequence encoding the human IgG1 hinge domain and the transmembrane and cytoplasmic domains of the human TCR ζ chain. In 14.G2aζ, the single-chain antibody domain is derived from the GD2-specific antibody 14.G2a (Rossig et al, 2001). The chimaeric genes were subcloned into the BamHI and NcoI sites of the retroviral vector SFG (Riviere et al, 1995) (provided by R. C. Mulligan, Cambridge, MA, USA).

Production of recombinant retrovirus

Fresh retroviral supernatants collected from transiently transfected Phoenix-eco cells were used to infect the packaging cell line PG-13 in the presence of polybrene (8 μg/ml) for 48 h at 32°C. The infected cells were incubated overnight at 37°C in fresh culture medium and then subjected to a second round of infection under the same conditions. Viral supernatants were generated on the resulting bulk producer cell lines by adding Iscove's modified Dulbecco medium (IMDM; BioWhittaker, Taufkirchen, Germany) supplemented with 20% fetal bovine serum (FBS; Perbio Science, Erembodegem-Aalst, Belgium). Following incubation for 24 h at 32°C, the supernatants were filtered through a 0·45 μm filter, and used directly to transduce the T cells.

Expansion and transduction of γδ T cell lines

Peripheral blood-derived mononuclear cells (2 × 106) were resuspended in RPMI 1640 medium (Perbio Science), containing recombinant human IL-2 (rhIL-2) at 100 IU/ml and rhIL-15 (10 ng/ml), and supplemented with 10% FBS and 2 mmol/l L-glutamine in a 24-well plate. Zoledronate (1 μg/ml) was added to the cultures on day 1. After 48 h, the cells were resuspended at 1 × 106 cells/ml in culture medium and incubated with equal volumes of freshly generated viral supernatant for 36 h at 37°C. Transductions were carried out in 24-well non-tissue culture-treated plates (Becton Dickinson, Franklin Lakes, NJ, USA), coated with recombinant fibronectin fragment FN CH-296 (Hanenberg et al, 1996) (Retronectin, Takara Shuzo, Otsu, Japan) at a concentration of 4 μg/cm2. Two weekly doses of rhIL-2 (100 IU/ml) and rhIL-15 (10 ng/ml) were added to the cultures. For control experiments with αβ T cells, PBMC were prestimulated on a 24-well plate precoated with OKT-3 (1 μg/ml; Ortho Pharmaceuticals, Raritan, NJ, USA) and anti-CD28 antibody (1 μg/ml; Pharmingen, San Diego, CA, USA) at 1 × 106 cells per well for 48 h, followed by transduction with viral supernatant and expansion in the presence of rhIL-2 (100 U/ml).

Flow cytometry

For immunophenotyping, cells were stained with fluorescein isothiocyanate (FITC)-conjugated monoclonal antibodies (Becton Dickinson, San Jose, CA, USA) directed against CD3, CD4, CD8, CD56, TCRγδ, TCRαβ and Vγ9 surface proteins. For each sample, 10 000 cells were analysed on a FACSCalibur with the Cell Quest Software (Becton Dickinson).

Measurement of cytokine production

Triplicate samples of transduced effector cells (5 × 104/well) were cocultured with various tumour cells at a stimulator-to-effector ratio of 3:1 in the presence of rhIL-2 (100 IU/ml) in 96-well round-bottomed plates. After 24 h, the supernatants were harvested and analysed for human interferon (IFN)-γ (Hölzel Diagnostika, Köln, Germany) by enzyme-linked immunosorbent assay according to the manufacturer's instructions.

Intracellular cytokine assay

Non-transduced and 14.G2aζ-transduced γδ T cells were seeded at 1 × 106 cells per well in a 24-well plate and were stimulated with 2·5 × 105 irradiated Ginline image tumour target cells. Controls consisted of γδ T cells cultured without target cells for 4 h. Cytokine secretion was blocked with 10 μg brefeldin A (Sigma, Deisenhofen, Germany) per 2 × 106 cells. Permeabilization of the cells was performed using a proprietary solution (Becton Dickinson). Cells were stained according to the manufacturer's recommendations, and isotype-matched negative controls were used for all antibodies.

Cytotoxicity assays

Cytotoxic specificity was determined in standard 51Cr release assays. Various numbers of T effector cells were coincubated in triplicate with 2500 target cells labelled with 3·7 MBq 51Cr. 5 × 106 cells (PE Applied Biosystems) in a total volume of 200 μl, in a V-bottomed 96-well plate. At the end of a 4-h incubation period at 37°C with 5% CO2, supernatants were harvested, and radioactivity was counted in a gamma counter. Maximum release was determined by lysis of target cells with Triton X. To determine human leucocyte antigen (HLA) class I or II restriction of cytolysis, target cells were preincubated for 30 min with W6/32 or CR3/43 antibody (Dako, Carpinteria, CA, USA). For blocking experiments, target cells were preincubated with various concentrations of anti-idiotype-specific monoclonal antibody 1A7.

Re-expansion experiments

Cells were plated at 1 × 106 cells/well of a 24-well plate and cocultured with 1 × 106 irradiated PBMC and 0·1 × 106 irradiated allogeneic Epstein–Barr virus-transformed lymphoblastoid cells (EBV-LCL) in growth medium containing phytohaemagglutinin (PHA; 0·5 μg/ml). Cultures were fed twice-weekly with growth medium supplemented with IL-2 (100 IU/ml) and IL-15 (10 ng/ml).

Statistical analysis

The student T test was used to test for significance in each set of values, assuming equal variance. Mean values ± SD are given unless otherwise stated. P values < 0·05 were considered significant.

Results

TCR γδ+ T cells (Vγ9+) can be efficiently expanded from peripheral blood in the presence of zoledronate

Zoledronate induced significant expansion of γδ T cells in PBMC of four healthy donors, resulting in a relative increase of T cells expressing the γδ TCR. On day 7–10, 90 ± 5% (73–96%) of cultured T cells expressed the γδ TCR (Fig 1). The γδ T cell stimulatory capacity of zoledronate was superior to pamidronate (not shown). Immunophenotyping showed the selective expansion of a Vγ9-expressing population. 52 ± 7% of the γδ T cell cultures coexpressed CD56. Fewer than 5% of the cells had an immunophenotype characteristic of natural killer (NK) cells (CD3 CD56+). The majority of the cells were CD4 CD8 (not shown). For all subsequent experiments, T cell cultures containing <90%γδ T cells underwent one round of positive selection using TCR γδ-specific magnetic beads, resulting in >90%γδ T cell purity. Control αβ T cell populations expanded in the presence of rhIL-2 after a 48-h prestimulation with CD28- and CD3-cross-linking antibodies contained 97 ± 1% TCR αβ+ T cells.

Figure 1.

Immunophenotype of non-transduced (NT) and chRec-transduced T cells on day 10 after culture initiation. Mean of 10 (NT), 7 (14.G2aζ) and 3 (CD19ζ) individual cell cultures from 4 donors. a, α; b, β; d, δ; g, γ.

γδ T cells can be efficiently transduced with chimaeric receptor genes

Ten γδ T cell cultures, generated from four different healthy donors, were transduced with the chimaeric receptor genes 14.G2aζ, which recognize the ganglioside antigen GD2 present on tumours of neural crest origin (Schulz et al, 1984; Mujoo et al, 1989), including neuroblastoma and small cell lung cancer, or CD19ζ, specific for the B cell surface marker CD19. Retroviral transduction was performed during the initial expansion phase of γδ T cells. Maximum efficiencies of transduction were achieved when the populations were exposed to recombinant retrovirus within 48 and 72 h after initial stimulation with zoledronate (not shown). Flow cytometric analysis of γδ T cells stained with anti-14.G2a idiotype-specific antibody identified chimaeric receptors on 73 ± 12% of the γδ T cells (Fig 2). The efficiencies of transduction were superior to those of αβ+ T cell cultures expanded from the same donors and transduced in parallel, resulting in chimaeric receptor surface expression in up to 30% of cells. Chimaeric receptor expression was maintained over the entire period of γδ T cell culture of up to 70 d without any apparent downregulation. Transduction of T cells did not shift cellular immunophenotypes by comparison with non-transduced cells (Fig 1). Bulk transduced γδ T cells were used for all subsequent experiments without positive selection of cells expressing the chimaeric receptor.

Figure 2.

Chimaeric receptor surface expression on a γδ T cell culture, determined by staining of non-transduced (A) and 14.G2aζ transduced (B) γδ T cells with monoclonal anti-idiotype antibody and fluorescence-conjugated secondary antibody. One representative experiment out of seven is shown.

Transduced γδ T cells produce intracellular IFN-γ in response to stimulation with antigen-expressing tumour cells

To assess functional γδ T cell anti-tumour responses mediated by the chimaeric receptor, expression of the activation marker CD69, as well as intracellular IFN-γ production and secretion, by 14.G2aζ-expressing γδ T cells were assessed in response to chRec-specific stimulation. By flow cytometric gating on 1A7+ CD3+ cells within the bulk populations, only cells that expressed surface 14.G2aζ were included in the analysis.

Following stimulation with the GD2-expressing neuroblastoma target cell line LAN-1, 98 ± 2% of zoledronate-activated γδ T cell cultures transduced with the GD2-specific chimaeric receptor 14.G2aζ coexpressed the activation marker CD69, as opposed to 77 ± 23% of unstimulated cell (P = 0·198). Median fluorescence intensities obtained after staining with phycoerythrin (PE)-labelled CD69 antibody increased from 190 ± 37 in unstimulated γδ T cells to 377 ± 28 (IFN-γ-negative cells, P = 0·002) and 649 ± 46 (IFN-γ-secreting cells, P < 0·001) respectively. The cultures that were stimulated with GD2+ tumour cells contained 33 ± 3% IFN-γ-producing cells (Fig 3), whereas only 1·5 ± 0·5% of unstimulated chRec-expressing γδ T cell cultures were CD69+/IFN-γ+ (P < 0·001). Background cytokine production by LAN-1-stimulated non-transduced γδ T cells, selected by gating on 1A7-negative cells within the GD2-stimulated population, was 5·7 ± 1·2%.

Figure 3.

CD69 upregulation and intracellular IFN-γ secretion by 14.G2aζ-transduced cells in response to the GD2+ target cell line LAN-1 (B) or in the absence of chimaeric receptor stimulation (A). To exclude non-transduced γδ T cells within the cultures from analysis, the gate was set on 1A7+ cells. One representative experiment of three is shown.

Accordingly, culture supernatants of 14.G2aζ-transduced γδ T cells contained high amounts of IFN-γ (up to 45 ng/ml per 1 × 106 cells) in response to stimulation with antigen-expressing tumour cells (Fig 4) in the absence of significant background IFN-γ secretion by non-transduced γδ T cells or by 14.G2aζ+γδ T cells exposed to GD2-negative tumour targets.

Figure 4.

Enzyme-liked immunosorbent assay quantification of IFN-γ in a 72 h coculture of 14.G2aζ-transduced γδ T cells with GD2+ (JF) or GD2-negative (A-204) tumour cells (A), and of CD19ζ-transduced γδ T cells with CD19+ (Daudi, Raji, Reh) or CD19-negative (A-204) tumour cells (B). One representative experiment of two is shown.

Exposure of CD19ζ-transduced γδ T cells to the CD19+ Burkitt's cell lines Daudi or Raji, or the B precursor acute lymphoblastic leukaemia (ALL) cell line Reh also resulted in substantial IFN-γ secretion (up to 414 ng/ml per 1 × 106 cells) (Fig 4), comparable with that obtained with CD19ζ transduced αβ T cell lines (not shown). Whereas cytokine release by non-transduced γδ T cells in response to Raji or Reh cells was <3% of that obtained with transduced cells, coincubation of non-transduced cells with Daudi cells induced IFN-γ secretion of >20% of that obtained with CD19ζ+ cells.

ChRec-expressing γδ T cells efficiently and specifically lyse antigen-expressing tumour cells

To investigate whether transduction with the chimaeric receptor gene endows the γδ T cells with the ability to lyse antigen-expressing tumour cells, their cytotoxic activity was compared in standard 51Cr release assays. 14.G2aζ-transduced γδ T cells reproducibly showed specific cytolysis of all GD2-positive target cell lines in the absence of significant background cytotoxicity against GD2-negative cells (Fig 5). Specific cytolysis was comparable with that obtained with chRec-transduced αβ+ T cells (not shown). γδ T cells lacking the 14.G2a-specific chimaeric receptor failed to lyse the GD2+ tumour targets. CD19ζ-transduced γδ T cells specifically lysed Raji, Daudi and Reh cells in the absence of cytotoxicity against the CD19-negative erythroleukaemia cell line K-562 (Fig 5). Non-transduced γδ T cells had substantial cytotoxicity against Daudi cells, but failed to lyse any of the other B cell targets.

Figure 5.

Chimaeric receptor-transduced γδ T cells specifically lyse antigen-expressing tumour cells in 4-h 51Cr release assays. (A) Non-transduced and CD19ζ transduced γδ T cells were tested against 51Cr-labelled CD19+ (Raji, Reh) and against CD19-negative tumour cells (A-204, K-562). (B) Non-transduced and 14.G2aζ transduced γδ T cells were tested against 51Cr-labelled GD2+ (LAN-1, JF) and against GD2-negative tumour cells (A-204, K-562). (C) 51Cr-labelled GD2+ LAN-1 tumour cells were preincubated monoclonal antibodies recognizing monomorphic determinants of HLA class I or HLA class II or with anti-14.G2a idiotype-specific monoclonal antibody 1A7 (5 μg/ml), then coincubated for 4 h with 14.G2aζ-transduced γδ or αβ T cells at a 20:1 effector-to-target ratio. No Ab, no antibody (control). Shown is one representative experiment of two (B) or three (A and C).

To confirm the specificity of the chimaeric receptor interaction with target antigen, we performed antibody inhibition experiments. Preincubation of 14.G2aζ-transduced γδ T cells with 14.G2a idiotype-specific monoclonal antibody 1A7 resulted in up to 78% inhibition of target cell lysis (Fig 5), indicating a chimaeric receptor-mediated mechanism of target cell recognition. Preincubation with HLA class I and II blocking antibodies did not affect the lysis of tumour targets by transduced γδ T cells, excluding an MHC-dependent mechanism of lysis of the tumour target cells by the transduced γδ T cells (Fig 5). A comparable pattern of inhibition was observed with 14.G2a-ζ-transduced αβ+ T cell cultures.

Transduced γδ T cells can be efficiently expanded and maintained in culture at high purity

After the initial stimulation with zoledronate, γδ T cells expanded in vitro for up to 5 weeks in the presence of IL-2 and IL-15 without any further stimulation. Non-specific restimulation with irradiated feeder cells and PHA at three-weekly intervals produced continued expansion of both transduced and non-transduced γδ T cells during prolonged culture periods of up to 9 weeks (Fig 6). The γδ T cell purity of the cultures was maintained without further exposure to bisphosphonates. Transduction efficiencies, assessed by repeated surface staining with 14.G2a idiotype-spedific antibody, were stable over time.

Figure 6.

Superexpansion of non-transduced and 14.G2aζ-transduced γδ T cell populations. Three weeks after initial stimulation with zoledronate, non-transduced and 14.G2aζ-transduced γδ T cell cultures were restimulated at 3-weekly intervals with irradiated feeder cells and PHA (0·5 μg/ml) and maintained in IL-2 and IL-15 containing medium.

Discussion

The aim of this study was to generate potent anti-tumour effector cells for adoptive immunotherapy of cancer by genetically modifying human peripheral blood γδ T cells to specifically recognize tumour cells via surface ligands. Peripheral blood γδ T cells were efficiently expanded and transduced with the chimaeric receptor genes and maintained in culture for prolonged periods of time. They efficiently recognized antigen-expressing tumour cell targets and exhibited target-specific effector functions, including secretion of IFN-γ and non-MHC-restricted tumour cell lysis.

Effective adoptive cellular immunotherapy with chimaeric-receptor expressing T cells is currently limited by the failure of adoptively transferred T cells to functionally persist long-term in vivo. Their efficient reactivation in response to residual tumour cells in vivo would require coordinated triggering of both the chimaeric receptor and an independent costimulatory receptor (Bretscher & Cohn, 1970). However, most tumour cells lack costimulatory molecule expression and thus fail to provide adequate second signals. Human γδ T cells represent effector cells at the interface between innate and adaptive immunity, using both TCR- and NK cell receptor-signalling. While their costimulatory requirements have not been defined, unlike those of αβ T cells, second signals for γδ cells appear to be dispensable for full activation. Furthermore, ligands of the best-defined γδ T cell activating receptor, NKG2D-DAP10, are expressed on many tumour cells and have been attributed an important role in innate anti-tumour responses (Bauer et al, 1999). Thus, we suggest exploiting the potent immunostimulatory properties and minimal costimulatory requirements of γδ T cells receptor by using them as effector cells of chimaeric receptor-mediated adoptive immunotherapy of cancer.

A critical requirement for the effective targeting of tumour surface antigens via chimaeric receptors is the efficient gene transfer of chimaeric receptor genes into immune effector cells. No previous experience with the genetic modification of human γδ T cells has been reported. Using retroviral gene transfer, we consistently achieved γδ T cell transduction efficiencies of >50%, resulting in stable surface expression of the chimaeric receptors that was sufficient to obtain efficient target cell recognition and lysis even without positive selection of transduced cells. The transduction efficiencies were superior to those we and others obtained with polyclonal T cell populations, non-specifically activated by cross-linking CD3 and CD28 surface receptors (Maher et al, 2002; Rossig et al, 2002), or of EBV-specific cytotoxic T cell cultures following specific stimulation with EBV antigen (Rossig et al, 2002). The efficiency of retroviral transduction was highly dependent on the time after initial stimulation with aminobisphosphonates at which cells were exposed to retrovirus, and was highest during the initial expansion of γδ T cells, which proliferated at a very high rate to expand from 1 to 5% on day 1 to yield up to 96% of the culture on day 10. Thus, the potent activation stimulus provided by aminobisphosphonates that results in rapid cell division and proliferation appears to render the cells highly susceptible to retroviral transduction and stable insertion of recombinant genes. Following retroviral transduction with chimaeric receptors, no major changes in cellular phenotypes were observed, and the ability of the cells to persist and expand in vitro was comparable with that of non-transduced cells.

Adoptive immunotherapy using gene-modified γδ T cells further requires their continued and robust ex vivo expansion to yield sufficient numbers of viable cells for retransfusion. Human γδ T cells, when compared with αβ T cells, undergo apoptosis more readily after stimulation (Janssen et al, 1991; Ferrarini et al, 1995), which, together with the relative infrequency of γδ T cells in peripheral blood, may represent a serious obstacle to developing γδ T cell-based adoptive cellular immunotherapy. We have shown here that a single initial stimulation with zoledronate enabled gene-modified γδ T cells to be maintained in culture for up to 9 weeks in the absence of overgrowth by other lymphocyte populations. Expansion of retrovirally-transduced T cell cultures was not impaired when compared with non-transduced cells. Thus, sufficient numbers of chRec-gene modified γδ T cells for adoptive immunotherapy can be obtained by ex vivo culture of peripheral blood lymphocytes under the described conditions.

Efficient immune responses against tumours rely on the tumour-specific induction of potent effector functions, such as secretion of stimulatory cytokines and specific cytolysis. Tumour-induced chimaeric receptor signalling was shown to partially mimic the activation signals provided by αβ TCR engegement by agonist ligand/MHC. In polyclonal as well as EBV-specific oligoclonal αβ T cells, chimaeric receptor engagement was shown to induce upregulation of the early activation marker CD69, secretion of cytokines, such as IFN-γ, tumour necrosis factor-α, and granulocyte-macrophage colony-stimulating factor, as well as efficient and rapid target cell lysis (Roessig et al, 2002). We studied the consequences of chimaeric receptor ligation in γδ T cells and found an activation response similar to that obtained with αβ T cells. High amounts of IFN-γ were secreted following exposure with target antigen, and antigen-expressing tumour target cells were efficiently lysed in a chimaeric-receptor-dependent, non-MHC restricted manner in a short coculture assay. Thus, the potency of γδ T cells to mediate anti-tumour effector functions via chimaeric receptors is comparable with that of T cells bearing αβ TCRs.

γδ T cells may possess an innate surveillance function against tumours (Zocchi et al, 1990; Nanno et al, 1992; Watanabe et al, 1995; Chen et al, 2001). γδ T cells from both healthy individuals and tumour patients are cytotoxic to a variety of tumour cell lines and primary tumour cells, including Burkitt's lymphoma, multiple myeloma, and renal cell carcinoma (Fisch et al, 1990; Bukowski et al, 1995; Choudhary et al, 1995; Boismenu & Havran, 1998; Kunzmann et al, 2000; Lopez et al, 2000; Kato et al, 2001; Sicard et al, 2001; Zheng et al, 2001; Gober et al, 2003). Other tumour entities, such as osteosarcoma, were essentially resistant to the cytotoxicity by γδ T cell lines (Weber et al, 1992). The mechanism of γδ T cell mediated tumour cell lysis has not yet been clarified. It was suggested that tumour-specific non-peptide antigens can serve as specific ligands for γδ TCR (Thomas et al, 2000; Gober et al, 2003). Furthermore, NKG2D ligands expressed on tumour cells were considered to provide activating signals for Vγ9Vδ2 T cells (Das et al, 2001a; Wu et al, 2003). Others suggested that γδ T cell-mediated cytotoxicity against tumour cell lines may represent TCR-independent LAK-like activity associated with cytotoxic T cell lines maintained in IL-2 containing medium (Weber et al, 1992). The ability of γδ T cells to mediate natural immune responses against human malignant tumours in vivo is still poorly defined. A recent study showed that, in contrast to NK cells and Vδ1+γδ T cells, intravenous infusions of Vγ2Vδ9+ T cells were not capable of preventing or inhibiting the growth of autologous melanoma in severe combined immunodeficient (SCID) mice (Lozupone et al, 2004). Based on the observation that the survival rates in multiple myeloma are improved in patients receiving pamidronate in addition to salvage chemotherapy (Berenson et al, 1996, 1998; Dhodapkar et al, 1998; McCloskey et al, 1998), it was hypothesized that selective expansion of Vγ9Vδ2+ T cells may contribute to the anti-tumour effect of these drugs. A recent study reported treatment of refractory myeloma or low-grade Non-Hodgkin lymphoma with low dose IL-2 in combination with pamidronate (Wilhelm et al, 2003). Objective tumour responses, which were achieved in three of nine patients, correlated with in vivo proliferation of γδ T cells, strongly suggesting a γδ T cell mediated anti-lymphoma effect.

By endowing Vγ9Vδ2+ T cells with chimaeric receptor-mediated tumour specificity, we were able to compare their ‘innate’ and antigen-specific anti-tumour activity against neuroblastoma and malignant B cell targets in vitro. We confirm potent cytolysis of the Burkitt's lymphoma cell line Daudi by unmodified γδ T cells, which was further increased by expression of a CD19-specific chimaeric receptor. IFN-γ secretion in response to Daudi cells was also significantly enhanced by additional chimaeric receptor-mediated signals. Background activity of unmodified γδ T cells against other malignant B cell lines, including the Burkitt's lymphoma line Raji and the B precursor ALL cell line Reh was limited. The Daudi cell line has previously been shown to be particularly susceptible to γδ T cell-mediated lysis (Fisch et al, 1990). In this regard, endogenously produced phosphorylated metabolites of the mevalonate metabolic pathway have been identified as ligands recognized by tumour-reactive human Vγ9Vδ2 γδ T-cells (Fisch et al, 1990). At least in the case of Daudi cells, the recognition by human γδ T cells additionally depends on the missing expression of HLA class I, due to deficient β2 microglobulin synthesis (Rothenfusser et al, 2002). In our experiments, neuroblastoma cell lines were not lysed by γδ T cells unless these expressed a tumour-specific chimaeric TCR, indicating that such neuroblastomas did not produce sufficient amounts of endogenous mevalonate metabolites to be recognized by unmodified γδ T-cells. In one report, anti-neuroblastoma γδ T cell activity was shown by stimulation with a non-aminobisphosphonate (clodronate) (Schilbach et al, 2001), however, in other studies, non-aminobisphosphonates were ineffective in activating γδ T cells, even at high concentrations (Kunzmann et al, 2000; Das et al, 2001b).

Thus, although an innate anti-tumour effect by γδ T cells may exist against selected tumours, chimaeric receptor expression can significantly enhance the potency and specificity of the anti-tumour response and extend the therapeutic use of γδ T cells to all malignancies expressing antigens against which a chimaeric receptor has been generated.

Recent knowledge regarding the biology of γδ T cells as well as clinical experiences support our hypothesis that ex vivo gene-modified and expanded populations of these cells can indeed be efficiently re-expanded in vivo. Like for αβ T cells, naïve, memory and effector γδ T cell subsets have been identified and shown to display different functional activities (Eberl et al, 2002; Gioia et al, 2002). Treatment of cancer patients with zoledronate induced Vγ9Vδ2 cells to mature toward an IFN-γ producing effector phenotype (Dieli et al, 2003) while maintaining naïve and memory subsets at decreased levels. Thus, although γδ T cell stimulation with aminobisphosphonates results in polyclonal expansion without TCR priming, the populations appear to represent a functional range of strong proliferative and immediate effector functions essential for their anti-tumour activity as well as their capacity for re-expansion.

Some non-peptide antigens were shown to function as agonists during their first contact with γδ T cells but fail to induce an activating response after repetitive stimulation (Burk et al, 1997). In multiple myeloma patients, however, previous aminobisphosphonate treatment was not correlated with the failure to stimulate bone marrow γδ T cells (Kunzmann et al, 2000). Repeated infusions of zoledronate at 3-weekly intervals resulted in a continuing increase in the effector subset of Vγ9Vδ2 cells in vivo (Dieli et al, 2003). Thus, γδ effector T cells can be expected to respond to repeated administrations of aminobisphosphonates.

The side effects of clinical bisphosphonate treatment are limited. Pamidronate combined with low dose IL-2 was shown to be well tolerated in patients with refractory myeloma or low-grade Non-Hodgkin lymphoma, with common clinical side effects being limited to thrombophlebitis at the injection site and fever (Wilhelm et al, 2003).

One potential clinical application of this approach would be the adoptive transfer of donor-derived, chRec gene-modified CD19-specific γδ T cells to patients with pediatric ALL, a malignancy that generally does not respond to the administration of unmodified donor lymphocytes (Collins et al, 2000). Whereas some cancer patients fail to respond to aminobisphosphonates with efficient γδ T cell proliferation, probably due to severe immunodeficiency caused by prior chemotherapy, γδ T cell expansion is reliably induced in healthy donors (Wilhelm et al, 2003). Furthermore, the preferential localization of bisphosphonates to skeletal sites, including bone marrow, may result in selective targeting of gene-modified T cells that have homed to the bone marrow compartment, resulting in the eradication of minimal residual leukaemic blasts.

In conclusion, our results demonstrate that γδ T cell effector function can be redirected against haematopoetic and solid tumour cells. γδ T cells that express chimaeric anti-tumour receptors may represent a new source of effector cells for the adoptive transfer of tumour-specific immunity to cancer patients. Their responsiveness to stimulation with systemic aminobisphosphonates may enable the selective re-expansion of adoptively transferred T cells in vivo, preventing their functional inactivation and permitting long lasting anti-tumour immune control.

Acknowledgements

We thank Heddy Zola for providing the CD19 single chain gene and Dieter Kabelitz for technical advice and for providing the Vγ9-specific antibody clone 7A5. This work was supported by grants from the Bundesministerium für Bildung und Forschung (to C.R.) and from the Dr Mildred-Scheel-Stiftung der Deutschen Krebshilfe (to C.R.).

Ancillary