Rhabdomyosarcomas are the most frequent malignant soft tissue tumors of childhood; however, because current multimodality treatments fail to improve the poor survival rate of children with metastatic rhabdomyosarcoma, new treatments are required. We previously identified the γ-subunit of the fetal acetylcholine receptor (fAChR) as a specific cell surface target in rhabdomyosarcoma. Here, we engineered human T lymphocytes to express chimeric receptors composed of the antigen-binding domain of a human anti-fAChR antibody joined to the signaling domain of the human T-cell receptor ζ-chain. The interaction of fAChRζ-transduced T cells with fAChR-positive rhabdomyosarcoma cell lines, but not with fAChR-negative control cells, induced T-cell activation characterized by strong secretion of IFN-γ and delayed lysis of tumor cells. Importantly, we found that in six of six rhabdomyosarcoma patients, chemotherapy increased fAChR expression on residual tumor cells in vivo. Our observations suggest that these fully human chimeric fAChRζ-transduced T cells, which should be well tolerated by the patient, have potential use in vivo both as a primary treatment for rhabdomyosarcoma and as a complementary approach to eradicate residual tumor cells after chemotherapy. (Cancer Res 2006; 66(1): 24-28)
- chimeric T cells
- acetylcholine receptor
Rhabdomyosarcomas are the most common soft tissue sarcomas of childhood. Despite intensified, multimodality treatments, the overall survival for high-risk populations has remained at 5% to 20% over the last decades ( 1, 2). Therapeutic targeting of cancers with adoptively transferred cytotoxic T cells recognizing tumor-specific antigens has had encouraging success in lymphomas ( 3) and melanoma ( 4). T cells may be a therapeutic option for rhabdomyosarcoma, and the PAX3-FKHR fusion protein in alveolar rhabdomyosarcoma ( 5) or MAGE antigens ( 6) are potential targets. However, rhabdomyosarcomas express only low levels of MHC molecules and resist the MHC-restricted recognition by T cells ( 6). To bypass immune resistance, T cells have been engineered to express tumor-specific, chimeric receptors (chRec) for antigen recognition ( 7). In these recombinant receptors, an antibody-derived, single-chain scFv fragment is linked to the cytoplasmic signaling domain of the TCRζ chain. ChRec gene–modified T cells exhibit specific, MHC-independent in vitro lysis and cytokine secretion in response to antigen-expressing target cells and were immunoprotective in murine tumor xenograft models ( 7). A critical requirement is the presence of a tumor-specific target and availability of a specific antibody. A candidate target in rhabdomyosarcoma is the fetal type of the nicotinic acetylcholine receptor (fAChR; ref. 8). During development of the neuromuscular junction, a change from the fetal type (α2βγδ) to the adult type (α2βεδ) of the AChR occurs, with replacement of the γ-subunit by the ε-subunit ( 9). After birth, expression of fAChR is almost completely lost from mature muscle but is maintained in thymic myoid cells, some extraocular muscle fibers, and denervated muscle ( 8, 10). In addition, fAChRs are highly expressed in rhabdomyosarcoma, distinguishing them from normal muscle ( 8, 10). Moreover, cloned human Fab antibodies specific for fAChR are available ( 11). Here, we created chimeric T cells with specificity for fAChR and show that they can mediate specific T-cell lysis of rhabdomyosarcoma cells.
Materials and Methods
Cell lines and antibodies. Cell lines were provided by Ewa Koscielniak (Olgahospital, Stuttgart, Germany; rhabdomyosarcoma: FL-OH, Ax-OH-1), Paul H. Sorensen (University of British Columbia, Vancouver, British Columbia, Canada; rhabdomyosarcoma: Birch), Gary P. Nolan (Massachusetts Institute of Technology, Cambridge, MA; Phoenix-eco), and American Type Culture Collection (Manassas, VA; rhabdomyosarcoma: TE-671, A-204; leiomyosarcoma: SK-LMS-1). The DNA encoding adult AChR (α12, β, δ, ε) and fetal AChR (α12, β, δ, γ) was generated as described ( 9). Murine monoclonal antibody B8 (MIB8) binds to the γ-subunit of human AChR as shown previously ( 12). Santa Cruz Biotechnology (Heidelberg, Germany) provided the rabbit polyclonal antiserum (SC-1453) against the fAChR γ-subunit.
Cloning of chRec genes. The human recombinant Fab fragment, Fab 35, specific for the extracellular domain of the fetal AChR γ-chain, was isolated by phage display technology from a pregnant woman whose fetus was affected by in utero myasthenia gravis ( 11). The variable Fab domains VH and VL were assembled as scFv fragment by PCR. To generate the fAChRζ chRec gene, the DNA encoding VH and VL were amplified by PCR and ligated into pRSV (provided by Zelig Eshhar, Rehovot, Israel) in frame with the DNA coding for the human IgG1 hinge domain and transmembrane and intracellular domain of the human TCR ζ-chain. The chimeric gene was subcloned into the retroviral vector SFG ( 13). Expression of the chRec in SFG-transduced cells was proved by reverse transcription-PCR (RT-PCR). The negative control CD19ζ chRec was cloned as reported previously ( 14).
Production of recombinant retrovirus. Fresh retroviral supernatants 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 hours at 32°C. After overnight incubation at 37°C in fresh culture medium, infected cells were subjected to a second identical round of infection. Supernatants were generated on the resulting bulk producer cell lines by incubation for 24 hours at 32°C and used to transduce T cells.
Isolation and transduction of lymphocytes. Fresh peripheral blood mononuclear cells from five healthy donors were prestimulated with CD3- and CD28-specific antibodies and transduced as described ( 15). Bulk-transduced T cells containing >98% CD3+ T cells (coexpressing CD8 in 55-86%), but no detectable CD3−CD56+ natural killer cells, were used for all subsequent experiments without selection of chRec+ cells.
RT-PCR. Expression of chRec mRNA in transduced T cells was semiquantitatively assessed by RT-PCR using insert-specific primers. Amplification was done using a 5′ primer annealing to the scFv domain of the chRec, combined with primer 3′ Hinge, annealing to an 18 bp region within the spacer domain. Fγ-AChR and Rγ-AChR primers were used to determine the fAChR-specific γ-subunit and the common α-subunit mRNA as surrogate markers for fAChR expression ( 8, 10).
Measurement of cytokine production. Triplicate samples of transduced effector T cells (5 × 104 per well) were cocultured with various tumor 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 72 hours, supernatants were analyzed for IFN-γ (Hölzel Diagnostika, Cologne, Germany) by ELISA following the instructions of the manufacturer.
51Cr release assays. Various numbers of T cells were coincubated in triplicate with 2,500 target cells labeled with 100 μCi 51Cr / 0. 5 × 106 cells (PE Applied Biosystems, Foster City, CA) in a volume of 200 μL in a V-bottomed 96-well plate. After 4 hours of incubation at 37°C and 5% CO2, radioactivity of supernatants was counted in a γ-counter. Maximum release was determined by Triton X lysis of target cells.
Long-term cytotoxicity assays. Target cells (rhabdomyosarcoma cell line cells and control SK-LMS-1 cells at 2 × 105 per well) were cocultured in 24-well plates with fAChRζ-transduced or negative control (nontransduced or CD19-transduced) T cells at various ratios. Other negative control wells contained target cells alone. During coculture, the percentage of tumor cells was repeatedly determined by cytometry analysis after intracellular desmin staining. Nonvital cells were excluded by propidium iodide staining.
Granzyme B secretion assay. T cells (1 × 105 per well) and tumor target cells (5 × 105 per well) were added to triplicate wells coated with antihuman granzyme B capture antibody (ELISPOT Human Granzyme B kit, Becton Dickinson, Heidelberg, Germany). Negative controls consisted of T cells alone, target cells without T cells, and media alone. Following 4 hours of coincubation, granzyme B was detected following the recommendations of the manufacturer.
Biopsies. Fresh frozen prechemotherapeutic and postchemotherapeutic biopsies were available from six patients (two males, four females; age range 3-17 years, mean = 10.7 years) with either an alveolar (n = 2) or embryonal (n = 4) rhabdomyosarcoma. Chemotherapies followed CWS-96-based protocols ( 16).
In vitro chemotherapy. Subconfluent TE-671 cells grown on histologic slides or in cell culture flasks were subjected to single-agent chemotherapy for 3 days using either doxorubicin (250 nmol/L), etoposide (5 μmol/L), or cisplatinum (140 nmol/L). TE-671 cells on glass slides were fixed with acetone before immunofluorescence analysis. For RT-PCR and flow cytometry studies, TE-671 cells were released from cell culture flasks by trypsinization.
Immunohistochemistry and immunofluorescence studies. Immunohistochemistry and immunofluorescence studies on fresh frozen sections and rhabdomyosarcoma cells grown on histologic slides, respectively, followed standard protocols. Percentages of MIB8-immunoreactive cells were calculated from tumor cells counted in at least 10 high-power fields (at ×400 magnification).
Statistical analysis. Student's t test was used to test for significance in each set of values, assuming equal variance. Mean values ± SD are given unless otherwise stated.
FAChRζ-transduced T cells produce IFN-γ in response to fAChR-expressing tumor cells. To show specific recognition of fAChR-expressing rhabdomyosarcoma cells by chR-transduced T cells, we measured IFN-γ secretion after 72 hours of coculture. ChRec composed of the scFv domains of the anti-fAChR Fab 35 in VL-VH orientation fused to the signaling moiety of the ζ-chain (designated fAChRζ) induced strong secretion of IFN-γ (up to 19 ng/mL/106 cells) when cocultured with fAChR-positive rhabdomyosarcoma cell lines (TE-671, Birch, and FL-OH), but not with the fAChR-negative cell line A-204 ( Fig. 1A ). Nontransduced T cells (not shown) and T cells transduced with an unrelated anti-CD19-reactive scFv ( Fig. 1B) secreted much less IFN-γ. Thus, it seems that the interaction between fAChRζ and fAChR triggers a functional and specific T-cell response associated with IFN-γ secretion.
Increasing fAChR density on rhabdomyosarcoma cells confers higher susceptibility to fAChRζ-mediated immediate cytotoxicity. Although fAChRζ+ T cells exhibited fAChR-specific IFN-γ production ( Fig. 1A), they failed to mediate significant lysis of these rhabdomyosarcoma cell lines within the 4-hour incubation period typically used in 51Cr release assays ( Fig. 2A ). Therefore, we transiently transfected TE-671 cells with fetal AChR (α12, β, δ, and γ; henceforth called TE-671/γ) to increase the surface fAChR expression level ( Fig. 2B). Indeed, TE-671/γ cells were consistently lysed better than native TE-671 cells ( Fig. 2A) or TE-671 cells transfected with adult AChR (α12, β, δ, ε; TE-671/ε). Furthermore, the native fAChRhigh rhabdomyosarcoma cell line Ax-OH-1 was even better lysed than TE-671/γ cells ( Fig. 2A). Moreover, secretion of granzyme B, an important mediator of immediate T-cell cytolysis, was lower in cocultures with fAChRlow (TE-671; TE-671/ε; Birch; FL-OH-1) compared with fAChRhigh (TE-671/γ; Ax-OH-1) rhabdomyosarcoma cell lines ( Fig. 2C).
FAChRζ-expressing T cells efficiently lyse fAChR-expressing tumor cells during long-term coincubation. The above findings showed resistance of low AChR–expressing rhabdomyosarcoma cells toward fAChR-specific cytolysis in short-term culture. Coincubation of rhabdomyosarcoma targets with fAChRζ-transduced T cells for 7 days, however, resulted in gradual elimination of tumor cells ( Fig. 3A-C ). Tumor cell numbers declined after 24 hours of coculture, dropped steeply up to day 3, and were almost undetectable by day 7. FAChR-expressing tumor cells incubated with nontransduced T cells were not lysed during long-term cocultures ( Fig. 3A and B) nor were fAChR-negative control SK-LMS-1 cells (not shown).
FAChR expression in rhabdomyosarcoma after chemotherapy. It has been reported in a single case of rhabdomyosarcoma that AChR expression was increased after chemotherapy ( 17). To confirm this finding, we studied biopsies from six individual patients, both before and after chemotherapy, by immunohistochemistry and RT-PCR, using simultaneous amplification of the α- and γ-subunit of the AChR as previously described ( 10). Whereas the primary biopsies exhibited variable percentages of fAChR+ tumor cells (35-60%, mean = 48.3%; Fig. 4A and B ), postchemotherapy tumor cells showed strong fAChR expression in >95% of the cells ( Fig. 4D and E), with increased fAChR γ-subunit mRNA levels in all biopsies ( Fig. 4C and F). In addition, in vitro, single-agent, short-term treatment with rhabdomyosarcoma-directed chemotherapeutics (doxorubicine, etoposide, and cisplatinum) resulted in an increased expression of AChR γ-subunit protein and mRNA in fAChRlow TE-671 rhabdomyosarcoma cells ( Fig. 4G-K; confirmatory flow cytometry findings not shown). Neither agent induced fAChR expression in fAChR-negative A-204 rhabdomyosarcoma cells (not shown).
We show that a chimeric T-cell receptor consisting of a recombinant human-derived scFv, specific for human fAChR and fused to the T-cell receptor ζ-chain, specifically recognizes fAChR-expressing rhabdomyosarcoma cells as indicated by short-term IFN-γ secretion and granzyme B release, and cytolysis of >90% of fAChR+ rhabdomyosarcoma cells after prolonged coculture. During short-term cultures, substantial cytolysis occurred only when rhabdomyosarcoma cells expressed higher fAChR levels, either spontaneously or by transfection. Moreover, we found that fAChR expression was increased after chemotherapy in biopsies of residual tumors of rhabdomyosarcoma patients compared with the respective tumors before chemotherapy. Based on these two findings, the fully human fAChRζ-transduced T cell has some potential as a future treatment option for rhabdomyosarcoma and could, in addition, be helpful in eradicating residual cells after chemotherapy.
We chose to target the fAChR γ-chain because it is the first antigen shown to be overexpressed on both embryonal and alveolar rhabdomyosarcoma ( 8, 10). Due to its restricted expression on only a few extraocular muscle fibers and thymic myoid cells but not on innervated skeletal muscle ( 10), immune targeting of fAChR in patients with rhabdomyosarcoma could be tumor selective and the risk of side effects by cross-reactivity with normal tissues should be limited. This is supported by the observation that maternal autoantibodies specific for the fAChR can be present in asymptomatic mothers while causing paralysis in their developing fetuses ( 18).
Besides tumor specificity, the successful immunotherapeutic use of chRec-redirected T cells requires two more antigen features. First, accessible extracellular domains and, second, stable expression on all or most tumor cells. The fAChR is a transmembrane protein with an extracellular domain that is a proved target for autoantibodies ( 9). In addition, although primary rhabdomyosarcoma tumor biopsies ( 8, 10) and rhabdomyosarcoma cell lines ( Fig. 2B) show variable fAChR expression, most persisting rhabdomyosarcoma cells following chemotherapy exhibit strong fAChR expression ( Fig. 4; ref. 17). Thus, the use of fAChR-specific T cells as an adjuvant to conventional chemotherapy should be considered.
One potential obstacle for the successful therapeutic use of chRec-expressing T cells is the limited survival of adoptively transferred xenogeneic T cells. The antigen-binding domains of most tumor-specific chRecs described to date are derived from murine antibodies, including the recently described immunotoxin that reacts with another target specifically expressed on rhabdomyosarcoma cells ( 19). Human anti-mouse antibody responses mediate rapid clearance of murine antibodies from the circulation, especially after repeated administration ( 20), and may thus limit the life span of T cells bearing murine antibody-derived recognition domains. Furthermore, humanization of mouse antibodies by complementarity determining region grafting does not completely abrogate immunogenicity. Our observation that the killing process in vitro requires prolonged contact times between sarcoma targets and effector T cells indicates that extended in vivo survival of chimeric T cells following adoptive transfer may be important for effective treatment of rhabdomyosarcoma. In addition, whereas immunotherapies against leukemias typically follow short-term treatment protocols ( 21), due to the easy accessibility of tumor cells, efficient T-cell homing to solid tumors may require prolonged survival or repeated application of the chimeric T cells ( 19, 20). Because the chimeric T-cell receptor described here is fully human, it should show enhanced persistence in vivo.
The molecular mechanisms underlying long-term cytotoxicity of fAChRζ-transduced T cells toward fAChRlow rhabdomyosarcoma cell lines remain unclear. However, the Fas/FasL system seems not to be involved: Although we found virtually no expression of FAS on TE-671 or FL-OH-1 cells by flow cytometry, substantial FAS expression occurred on Birch or Ax-OH-1 cells (not shown). Nevertheless, fAChRζ-transduced T cells killed the various rhabdomyosarcoma cell lines to a similar extent ( Fig. 3).
In summary, we show for the first time that fAChRζ-transduced T cells are specifically activated by tumor target cells expressing fAChR, and specifically eliminate rhabdomyosarcoma cells from long-term cocultures. Furthermore, we show that fAChR expression on rhabdomyosarcoma can be increased by chemotherapeutic agents in vitro and confirm the previous observation ( 17) that fAChR hyperexpression is characteristic of rhabdomyosarcoma cells surviving chemotherapy in rhabdomyosarcoma patients. Therefore, our findings open new perspectives for clinical trials to evaluate fAChRζ-transduced T cells for potential immunotherapy of rhabdomyosarcoma and as a tool to eradicate chemoresistant residual disease in rhabdomyosarcoma patients.
Grant support: Bundesministerium für Bildung und Forschung (C. Rossig), Dr. Mildred-Scheel-Stiftung der Deutschen Krebshilfe grant 10-1859 (C. Rossig), and Wilhelm-Sander-Stiftung grant 99.112.1 (S. Gattenlöhner and A. Marx), all in Germany.
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.
We thank Malcolm K. Brenner and Cliona A. Rooney for helpful discussions and Erwin Schmitt and Margrit Bonengel for expert technical assistance.
Note: S. Gattenlöhner and A. Marx share first authorship.
- Received February 18, 2005.
- Revision received September 12, 2005.
- Accepted November 11, 2005.
- ©2006 American Association for Cancer Research.