Gene-Engineered Varicella-Zoster Virus–Reactive CD4⁺ Cytotoxic T Cells Exert Tumor-Specific Effector Function

Introduction

T cells are potent immune effectors and have shown promise for the elimination of tumors in vivo ( 1). In recent clinical trials, adoptive T-cell transfer has been shown to mediate successful tumor regression in cancer patients ( 2, 3). These trials have further shown the importance of survival, in vivo growth, and functional maintenance of adoptively transferred antitumor T cells for achieving clinical antitumor responses ( 4). One of the major challenges in the field remains the in vivo or ex vivo generation of large numbers of tumor-specific T cells capable of surviving long-term in a functional memory state. Serious obstacles are the scarcity of T cells with high T-cell receptor avidity for tumor-associated antigens within the patient's lymphocyte repertoire, and defective antigen presentation by tumor cells can potentially result in tumor escape from immune killing. Both requirements can be bypassed by genetic modification with recombinant chimeric receptors (chRec), which redirect T cells toward tumor surface antigens independent of antigen presentation ( 5). ChRecs consist of antibody-derived ligand-binding domains linked to stimulatory and costimulatory T-cell signaling pathways and thereby combine antigen recognition and signal transduction in a single molecule. Indeed, cross-linking of chRecs with tumor antigen induces immediate effector functions, and immunoprotection against tumor growth was shown in murine model systems ( 6). However, the T-cell activation response induced by chRec engagement is insufficient to allow clonal reexpansion on encounters with residual tumor cells in vivo, and T cells with grafted specificities have failed to ensure long-term immune control in vivo ( 7).

By contrast, the sustained clinical benefit of transfusions of herpes virus–specific cytolytic T cells in reconstituting protective immunity in bone marrow transplant recipients is now well documented ( 8, 9). Effector T cells with native specificity for herpes viruses, including cytomegalovirus (CMV) and EBV, naturally arise during primary infection. Due to the failure to eradicate the virus from the host and the establishment of life-long latent infection by herpes viruses, mechanisms exist that preserve protective virus-specific active immunity for a lifetime. Indeed, adoptively transferred T cells with native specificity for herpes virus antigens receive potent reactivation stimuli in vivo, resulting in persistence for extended periods and establishment of a long-lasting functional memory ( 8, 9).

In an attempt to exploit the mechanisms of persistent antiviral immune control for the immunotherapy of cancer, we and others have proposed a strategy based on retargeting of virus-specific T cells to tumor cells. Dual-specific T cells with native specificity for EBV ( 10), CMV ( 11), and influenza ( 12) should receive strong reactivation and survival signals in response to virus antigen while doing tumor antigen-specific effector functions via chimeric T-cell receptors. However, this requires that the patient is infected with the appropriate virus. EBV and CMV infections occur in a limited proportion of children before adolescence, hampering the use of specific T cells against these viruses in childhood malignancies. Even in adults, latent viruses are tightly controlled by endogenous immunity and infused T cells may receive insufficient antigenic stimulation. In the setting of stem cell transplantation (SCT), lymphopenia and virus reactivation account for the massive expansion of transfused virus-specific cytotoxic T cells (CTL) observed. By contrast, in patients who are not lymphodepleted, endogenous stimulation by virus may be insufficient. For example, EBV-specific T cells did not expand in patients with nasopharyngeal carcinoma whose T-cell compartment was replete ( 13).

Here, we propose that both natural immunity to varicella zoster virus (VZV) and the availability of a VZV vaccine can be exploited for T-cell immunotherapy of cancer. Natural primary infection with VZV usually occurs early in childhood and is associated with the rapid induction of virus-specific T cells that specifically lyse infected cells and are crucial for persistent antiviral control in vivo ( 14). Alternatively, memory CTL responses comparable with those of naturally immune subjects can be induced by immunization with a live attenuated VZV vaccine ( 15), which is now part of routine vaccination programs in many countries. Importantly, the VZV-specific T-cell memory subsets persist in vivo for a lifetime and can be efficiently reexpanded either by reexposure or endogenous reactivation of the virus or by booster doses of the vaccine ( 16). Therefore, we speculate that VZV-specific T cells meet all the apparent requirements for serving as potent and long-lasting effector cells of chRec-mediated antitumor immunity.

We show here that human VZV-reactive T cells with cytolytic properties can be efficiently expanded in vitro and engineered to recognize and lyse tumor cell targets via chRec while maintaining their ability to functionally interact with VZV antigen-presenting cells (APC).

Materials and Methods

Cell lines. The ecotropic packaging cell line Phoenix-eco ( 17) was provided by Gary P. Nolan (Stanford University, Stanford, CA). PG-13 [American Type Culture Collection (ATCC)] is a retrovirus packaging cell line that produces virus pseudotyped with the gibbon-ape leukemia virus. The amphotropic FLYRD18 (provided by E. Vanin, Baylor College of Medicine, Houston, TX) cell line provides viral recombinants with the RD114 envelope. Reh (ATCC) is an acute lymphoblastic leukemia cell line, K-562 (ATCC) is a human erythroleukemia line, and A-204 (ATCC) is a human rhabdomyosarcoma cell line. The neuroblastoma cell line LAN-1 and LAN-5 were provided by Robert C. Seeger's laboratory at University of California at Los Angeles (Los Angeles, CA), and JF was obtained from Malcolm Brenner of Baylor College of Medicine. Lymphoblastoid cell lines (LCL) were generated by infection of peripheral blood mononuclear cells (PBMC) with supernatant from the EBV producer cell line B95-8 ( 9). Autologous fibroblasts immortalized by retroviral transduction with hTERT ( 18) were obtained from Silvia Puttmann (University of Muenster, Münster, Germany).

chRec genes. The chRec gene CD19ζ contains the variable domains of monoclonal antibody (mAb) FMC-63 ( 19) assembled as an extracellular single-chain antibody Fv fragment (scFv) molecule, cloned in frame with a sequence encoding the human IgG1 hinge domain and the transmembrane and cytoplasmic domains of the human T-cell receptor ζ chain ( 20). 14.G2aζ contains the single-chain antibody domain of the G_D2-specific antibody 14.G2a ( 10). In 14.G2a-CD28ζ, the human CD28 transmembrane and cytoplasmic domain was subcloned upstream of the TCRζ gene. The chimeric genes were subcloned into the BamHI and NcoI sites of the retroviral vector SFG (provided by R.C. Mulligan, Massachusetts Institute of Technology, Cambridge, MA; ref. 21).

Production of recombinant retrovirus. Fresh retroviral supernatants collected from transiently transfected Phoenix-eco cells were used to infect the packaging cell line PG-13 (for CD19ζ), and supernatants from Phoenix-ampho cells were used to infect FLYRD18 (for 14.G2aζ and 14.G2a-CD28ζ), respectively, in the presence of polybrene (8 μg/mL) for 48 h at 32°C. Viral supernatants were generated from the resulting bulk producer cell line by adding Iscove's modified Dulbecco medium (BioWhittaker) supplemented with 20% fetal bovine serum (Perbio Science). After 24 h of incubation at 32°C, the supernatants were filtered through a 0.45-μm filter and used directly to transduce the T cells.

Generation of VZV lysates. Cells (2 × 10⁷) of the human lung fibroblast cell line MRC-5 (ATCC) were grown in monolayers and infected with one vaccination dose of the VZV vaccine strain Oka (Varilrix, GlaxoSmithKline), containing 2,000 plaque-forming units. When reaching 50% to 80% cytopathic effect, cells were harvested. Following three freeze/thaw cycles, cells were sonicated and then centrifuged for 10 min at 900 × g to remove cell debris. Finally, the lysates were heat inactivated at 56°C for 30 min. Control lysate was generated in an analogous manner from noninfected MRC-5 cells. Each batch of lysate was tested separately for its capacity to stimulate peripheral blood T-cell proliferation in a [³H]thymidine assay.

Expansion and transduction of VZV-reactive T-cell cultures. PBMCs from seropositive donors were resuspended in RPMI 1640 (Perbio Science) and supplemented with 2% autologous serum and 2 mmol/L l-glutamine at 2 × 10⁶ cells per well in a 24-well plate, and 10 to 30 μL VZV lysates. Starting on day 7, the cells were restimulated weekly with irradiated autologous PBMC (30 Gy) at a 1:1 responder-to-stimulator ratio and 10 to 30 μL VZV lysate. Two weekly doses of recombinant human interleukin (IL)-2 (rhIL-2; 25 IU/mL) were added from day 11. Twenty-four hours after their third stimulation with VZV lysates, CTLs were transferred to a 24-well nontissue culture-treated plates (Becton Dickinson) precoated with recombinant fibronectin fragment FN CH-296 (Retronectin, Takara Shuzo) at a concentration of 4 μg/cm² at 1 × 10⁶ cells/mL and incubated with equal volumes of freshly generated viral supernatant for 48 h at 37°C and 5% CO₂.

Flow cytometry. For immunophenotyping, cells were stained with fluorescein-conjugated mAbs (Becton Dickinson) directed against CD3, CD4, CD8, CD27, CD28, CD45RA, CD45RO, CD56, and CD25 surface proteins. For each sample, 20,000 cells were analyzed with FACSCalibur and CellQuest Software (Becton Dickinson). CD19ζ was detected by staining with IgG F(ab)2 antibody (Dianova) followed by streptavidin/phycoerythrin or streptavidin/APC (Becton Dickinson). Surface expression of 14.G2aζ and 14.G2a-CD28ζ was analyzed by staining with 14.G2a anti-idiotypic antibody 1A7 (200 ng/5 × 10⁵ cells; ref. 22) followed by incubation with fluorescein-labeled goat anti-mouse antibody (Becton Dickinson). For intracellular granulysin staining, the cells were fixed in 4% paraformaldehyde (Sigma-Aldrich) for 30 min; then washed in permeabilizing buffer (PBS; Life Technologies), 0.1% saponin (Sigma-Aldrich), and 1% FCS (Perbio Science); and resuspended in 100 μL permeabilizing buffer. To each sample, 5 μL of granulysin-specific mAb DH4, kindly provided by Alan Krensky (Stanford University), was added for 20 min on ice, followed by washing and 30 min of incubation with 1 μL APC-labeled goat anti-mouse secondary antibody. Intracellular staining with phycoerythrin-labeled forkhead box P3 (FoxP3) antibody (eBioscience) was done according to the manufacturer's instructions.

Generation and loading of dendritic cells. Freshly isolated PBMCs were seeded at 10⁷ per well into a six-well plate in RPMI 1640 supplemented with 10% FCS and 2 mmol/L l-glutamine and allowed to adhere for 2 h. After removing the nonadherent fraction, the adherent cells were resuspended in growth medium supplemented with 800 units/mL granulocyte macrophage colony-stimulating factor and 1,000 units/mL IL-4 (both cytokines were obtained from ImmunoTools) and incubated at 37°C and 5% CO₂. Immature dendritic cells were harvested on day 5 and assessed by flow cytometry. For use as stimulators of VZV-reactive cytokine responses, dendritic cells were loaded with 100 μL lysate/well for 16 h.

Intracellular cytokine assay. CTLs were seeded at 1 × 10⁶ cells per well in a 24-well plate and stimulated with 1 × 10⁶ irradiated antigen-expressing tumor target cells for 2 h. Cytokine secretion was blocked with 10 μg brefeldin A (BFA; Sigma-Aldrich)/2 × 10⁶ cells for 4 h when tumor cells were used as stimulators and for 16 h when dendritic cells were used. To measure VZV-specific cytokine secretion, 10⁶ VZV-CTLs per well were cocultured with 10⁵ lysate-loaded dendritic cells. Permeabilization of the cells was done using a proprietary solution (Becton Dickinson), followed by staining according to the manufacturer's recommendations. To determine HLA class I or II restriction of cytolysis, target cells were preincubated for 30 min with W6/32 or CR3/43 antibody (DakoCytomation).

Cytotoxicity assays. Cytotoxic activity was determined in standard ⁵¹Cr release assays. Various numbers of T effector cells were coincubated in triplicate with 2,500 target cells labeled with 100 μCi ⁵¹Cr/0.5 × 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 and 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.

Proliferation assays. Freshly isolated PBMCs were coincubated in triplicate at 1 × 10⁵ cells per well with various concentrations of VZV lysate. For restimulation assays, 1 × 10⁵ T cells per well were coincubated with 1 × 10⁵ tumor cells. After a 5-day coincubation period, wells were pulsed with 1 μCi/well [³H]thymidine for 16 h, and the samples were harvested onto glass fiber filter paper for β scintillation counting.

Expansion experiments. Cells were plated at 1 × 10⁶ cells per well of a 24-well plate and cocultured with various irradiated tumor cell targets or irradiated autologous PBMC and VZV lysate at a 1:1 stimulator-to-responder ratio. Cultures were fed twice weekly with growth medium supplemented with IL-2 (25 IU/mL), and weekly counts of viable cells were done by trypan blue exclusion.

Statistical analysis. The 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.

Results

Stimulation of peripheral blood T cells with VZV antigen induces expansion of a virus-reactive CD4⁺ effector T-cell population. Stimulation of PBMC with lysates generated from VZV-infected fibroblasts induced a significant (P < 0.01) and specific proliferative response in eight VZV-seropositive donors ( Fig. 1A ) but in none of three adult VZV-seronegative donors. However, reanalysis of peripheral blood T cells from two of the donors 6 weeks after active immunization with a live attenuated VZV vaccine showed a strong proliferative response comparable with that obtained in donors after natural infection, indicating a selective response of in vivo generated VZV-reactive effector cells to lysate stimulation ( Fig. 1B). This showed that our VZV lysate was able to reactivate T cells from both naturally infected and vaccinated donors. For all subsequent experiments, PBMCs from healthy donors with naturally acquired VZV seropositivity were used.

To expand VZV-specific T cells, PBMCs were cocultured with VZV lysates in the presence of feeder cells and rhIL-2 (25 IU/mL). Within 6 weeks, 55 ± 29–fold (45–96) total lymphocyte expansion was obtained ( Fig. 1C). Cultures could be maintained for up to 14 weeks, reaching total cell numbers of 10⁹ to 10¹⁰ per 1 × 10⁶ cells initially plated.

After three stimulations with VZV lysate-pulsed PBMCs, T-cell cultures from three of four donors were dominated by a CD3⁺CD4⁺ T-cell population (mean, 90 ± 5%; range, 90–99%; Fig. 2A ). CD3⁺ and CD8⁺ were coexpressed on 2% to 7% of cells and <5% had a natural killer cell phenotype (CD3⁻CD56⁺). T cells from the fourth donor differed from the others by reproducibly containing a higher proportion of CD3⁺CD8⁺ cells (12–20%) and >5% γδ T cells, with only 50% to 70% CD4⁺ cells. Repeated immunophenotyping of the T-cell cultures during expansion revealed an increasing proportion of CD45RO⁺ T cells coexpressing CD28 in the absence of CD27 ( Fig. 2B). CD25 was up-regulated on a substantial proportion of the expanded T cells. The CD4⁺CD25⁺ phenotype is shared between an activated T helper subset of peripheral blood effector T cells and a regulatory T-cell population with immunosuppressive properties [regulatory T cells (Treg); ref. 23]. Intracellular staining with FoxP3-specific mAb failed to detect a population with the characteristic phenotype of Tregs (CD4⁺CD25^highFoxP3⁺), whereas an expected proportion of 1% to 5% Tregs was found among the CD4⁺CD25^high population of peripheral blood lymphocytes ( Fig. 2C and D). Neither the CD25^low nor the CD25^int subpopulations contained any FoxP3⁺ cells. Furthermore, the expanded VZV-reactive T cells were not suppressive in mixed lymphocyte reactions (data not shown).

The VZV reactivity of the expanded T cells was shown in intracellular cytokine staining assays ( Fig. 3 ). Coincubation of T cells from five donors with autologous dendritic cells in the presence of VZV lysate reproducibly resulted in an increased proportion of IFN-γ–secreting T cells above varying background cytokine secretion obtained in the absence of lysate ( Fig. 3A and B). Cytokine release in response to antigen presented by HLA-mismatched allogeneic dendritic cells was minimal ( Fig. 3C). VZV-specific T-cell responses were restricted by MHC class II, as shown by blocking with MHC class II–specific inhibitory antibodies, which resulted in a reduction of cytokine-secreting cells to background levels ( Fig. 3D). To exclude autoreactivity of VZV-specific T cells, their reactivity toward autologous fibroblasts was assessed in a ⁵¹Cr release assay, and no cytolysis of these targets was observed ( Fig. 5B).

Thus, stimulation of peripheral blood T cells of VZV-seropositive donors with VZV lysates induced expansion of a population of CD4⁺ effector memory T cells that contained high numbers of virus-specific effector T cells.

VZV-CTLs are efficiently transduced with retroviruses expressing chRec genes. Twelve VZV-CTL cultures were generated from four different healthy seropositive donors and transduced with chRecs recognizing the surface marker CD19, expressed on normal and malignant cells of the B lineage ( 20), or the ganglioside antigen G_D2 present on tumors of neural crest origin, including neuroblastoma ( 24). ChRecs with two different intracellular signaling domains were compared. CD19ζ and 14.G2aζ signal only through the TCRζ chain, whereas 14.G2a-CD28ζ contains an additional signaling domain derived from the costimulatory molecule CD28 ( Fig. 4A ; ref. 25). Retroviral transduction was done after the third stimulation with VZV lysates. Transduction efficiencies were determined by flow cytometric analysis of CTL stained with antimouse Fab antibody and revealed chRec surface expression on 46 ± 14% (29–74%) of the cells ( Fig. 4B).

The distribution of CD4⁺ and CD8⁺ cellular subsets within the T-cell cultures was not affected by modification with chRec genes, and the ability for expansion for at least 5 to 8 weeks was maintained when compared with nontransduced CTL ( Figs. 1 and 6). Furthermore, transduction with chRec genes did not affect the cytokine and cytolytic response to stimulation with VZV antigens nor did it further increase the non-VZV–specific background reactivity within the cultures (data not shown). Thus, VZV-reactive CTL can be activated in vitro for efficient retroviral transduction.

chRec-modified VZV-CTLs produce cytokines in response to antigen-expressing tumor targets. To investigate functional antitumor responses mediated by the chRec, we first analyzed intracellular cytokine production in response to antigenic stimulation using flow cytometry. Exposure of 14.G2aζ-transduced VZV-CTL to the G_D2⁺ neuroblastoma cell line LAN-5, but not G_D2⁻ target cells, resulted in secretion of both IFN-γ and IL-4 ( Fig. 4C). Secretion of the inhibitory cytokine IL-10 was minimal in all cocultures including those of transduced VZV-CTL with antigen-expressing target cells (data not shown). To show that individual cells within cultures of gene-modified VZV-reactive CTLs possess dual-receptor specificity, we stimulated chRec-transduced CTL with VZV lysate, followed by costaining with Fab-specific antibody, which interacts with the grafted receptor, and with IFN-γ– and tumor necrosis factor-α–specific antibodies, which detect the cytokine response induced by the native receptor. A substantial proportion of T cells within transduced bulk VZV-CTL cultures were found to coexpress both receptors, consistent with functional bispecificity toward VZV and CD19 ( Fig. 4D).

chRec-modified VZV-reactive T cells specifically and efficiently lyse target-expressing tumor cells. For optimal antitumor activity, it is important that the chRec-expressing T cells exert cytolytic function. Because the majority of the expanded cells had a CD4 phenotype, we evaluated their ability to kill tumor cells in standard ⁵¹Cr release assays. Both CD19ζ- and 14.G2aζ-transduced T cells (CTL) showed specific and potent cytolysis of tumor cells expressing the respective target antigen, with no killing of antigen-negative cell lines ( Fig. 5A and B ). No lysis of tumor cells was observed by nontransduced CTL. Efficient blocking of cytotoxicity by concanamycin and BFA confirmed that CD4⁺ VZV-CTLs exert their cytolytic function via granule exocytosis (data not shown). As perforin is a key molecule in the granzyme-mediated cytotoxic pathway, we investigated the expression of perforin in VZV-CTL. Whereas intracellular staining detected perforin in control EBV-specific CD8⁺ CTL lines (data not shown), all VZV-reactive CD4⁺ cell lines investigated were perforin negative ( Fig. 5C). Exocytosis of granulysin was described recently as an alternative mechanism of T-cell cytolysis ( 26, 27). Analysis of intracellular granulysin expression by flow cytometry revealed a high percentage of granulysin-positive cells of 38.0 ± 4.6 (32–43.3%) among the VZV-CTL cultures ( Fig. 5C).

Virus-specific stimulation but not chRec engagement induces proliferation and expansion of gene-modified VZV-CTL. Whereas repeated stimulation with VZV lysates induced a potent proliferative response, resulting in expansion of gene-modified VZV-CTL for prolonged periods and with kinetics similar to those of unmodified CTL, exposure to chRec antigen-expressing tumor target cells failed to induce T-cell proliferation ( Fig. 6A and B ). One potential reason for the limited activation response to chRec engagement is the lack of CD28 costimulatory signals along with TCRζ-mediated primary signaling. However, including the CD28 signaling domain in a chRec expressed in VZV-CTL failed to restore their ability to proliferate and expand in response to tumor cells ( Fig. 6C and D). Thus, a potent proliferation stimulus to gene-modified VZV-CTL is provided only by the native receptor. Appropriate CD28 signaling in the construct we used had been confirmed earlier in experiments with nonspecifically prestimulated T cells, in which inclusion of the CD28 signaling domain significantly improved antigen-induced expansion over chimeric TCRζ signaling alone ( 25). Therefore, the diverse potential of the endogenous and grafted receptors to mediate expansion of VZV-CTL cannot be explained by the presence or absence of CD28 costimulatory signaling alone.

Discussion

In this article, we present a strategy for exploiting the potency of VZV-specific memory T-cell reactivation for immunotherapy of cancer. We show that VZV-reactive, CD4⁺ cytotoxic T-cell lines can be efficiently expanded in vitro and functionally redirected to tumor cells by genetic engineering with antigen-specific chRecs while maintaining functional, MHC-restricted interaction with VZV antigen through their native receptor.

The major advantage of T cells with predefined specificity for a strong viral antigen compared with cells with exclusive tumor specificity is the potential for therapeutic reinduction of antitumor immune control. Primary infection with VZV or immunization establishes life-long viral latency, which supports the indefinite persistence of high levels of functional virus-specific T cells in vivo ( 28). Booster doses of VZV vaccine were shown to mimic naturally occurring subclinical reactivation of VZV ( 16), resulting in highly efficient clonal reexpansion of memory T cells. Thus, administration of varicella vaccine can be expected to induce specific proliferation of the therapeutic T cells, thereby increasing the number of circulating effector memory T cells with grafted reactivity against leukemia cells. This is a significant advantage over previously suggested viral model systems based on CMV ( 11) or EBV ( 10)-specific T cells that rely on endogenous restimulation of adoptively transferred T cells.

One potential application of gene-modified VZV-reactive T cells is the prevention of leukemia relapse following SCT in high-risk patients. Indeed, the lymphopenic state early after SCT was shown to provide an ideal platform for restoring virus-specific CTLs, and lymphodepletion deliberately induced by lymphotoxic drugs allowed the in vivo expansion of tumor-specific T cells ( 3, 29, 30). In addition to their antileukemic properties, adoptively transferred T cells with native specificity for VZV have the potential to reduce the significant morbidity associated with this virus in stem cell transplant recipients ( 31). The potency of adoptively transferred T cells to control VZV infection in vivo was shown in four children with disseminated varicella after SCT who were successfully treated with donor lymphocyte transfusions ( 32), and in a liver-transplanted child with chronic VZV infection ( 33). Outside the SCT setting, the tight immune control of herpes viruses may limit the in vivo boosting of the anticancer immune responses by virus-infected cells. In these patients, iatrogenic boosting using the VZV vaccine can bypass the requirement for lymphodepletion and its inherent toxicities. Thereby, the use of VZV-dual specific T cells extends the spectrum of malignancies amenable to the sustained effects of adoptive cellular immunotherapy.

The major requirements for the use of VZV-specific T cells as antitumor effector cells are (a) in vitro expansion of high numbers of T cells with native receptor specificity for VZV, (b) efficient genetic modification with tumor-specific chRec, (c) exertion by these cells of potent and specific cytolytic effector functions against the tumor target while maintaining functional interaction with VZV-infected cells, and (d), most importantly, the ability to proliferate in vivo after infusion.

While establishing an in vitro system for expanding VZV-specific CTL, we confirm previous experiences that virus-specific memory T cells within the peripheral blood of donors with a history of varicella infection or active immunization show strong proliferative responses to viral antigen lysates ( 28). The fact that lysates obtained from identically grown, noninfected fibroblasts did not induce T-cell proliferation argues against nonspecific reactivity against nonviral epitopes. Furthermore, proliferative responses were restricted to seropositive donors and could be evoked specifically by active immunization. The high proliferative activity of VZV-specific T cells in response to viral lysates allowed transduction efficiencies reliably exceeding 50%, sufficient to obtain efficient target cell recognition and lysis.

Maintaining VZV-CTLs in culture for longer than 4 weeks has not been reported previously. Here, repeated stimulations with lysate-pulsed autologous APC allowed us to maintain CTLs in culture for prolonged periods and generate sufficient cell numbers for adoptive immunotherapy compared with successful clinical trials ( 9), within weeks. To increase the safety of the clinical application of the strategy, we plan to modify the culture conditions in a manner allowing exclusive expansion of T cells specifically interacting with virus-derived peptides. This will involve specific stimulation with epitopes derived from immunogenic VZV proteins (e.g., IE-62 or gpI), followed by positive selection with VZV-specific tetramers or pentamers. Currently, this strategy is limited by the lack of available VZV peptides and their corresponding pentamers.

A major determinant of the therapeutic potential of the antitumor response is the potency of chRec-mediated effector functions. Because exogenous antigens have only limited access to class I pathways, the generation of T cells using lysate-pulsed APCs has the potential disadvantage of favoring expansion of poorly cytolytic, class II restricted CD4⁺ T cells ( 34). Indeed, VZV-reactive T-cell cultures are dominated by CD4⁺ T cells. However, chRec-retargeted CD4⁺ VZV-reactive T cells mediate potent cytolysis of tumor targets comparable with that obtained with CD8⁺CTL ( 10). Our observation agrees with previous reports on the induction of CD4⁺ T-cell populations with cytolytic properties by vitro stimulation with viral lysates ( 35, 36). A physiologic role in humans is suggested by the detection of cytotoxic CD4⁺ T cells in various viral infections, including human immunodeficiency virus, CMV, and EBV ( 37), and in malignant disease ( 38). They seem to represent a subset of antigen-experienced or memory cells that have acquired cytolytic activity during differentiation in response to strong or chronic immune activation. The phenotype of the VZV-CTL expanded in our system, characterized by lack of expression of CD27 and CD45 RA, as well as CD25 and CD45 R0 positivity ( 37), is indeed consistent with a highly activated, late effector memory T-cell subset.

A potential mechanism of tumor cell escape from chRec-expressing T lymphocytes is the development of antigen-negative clonal loss variants. Thus, the successful therapeutic use of tumor antigen-specific T cells relies on the choice of an adequate target structure. Both antigens used in the present study, CD19 and G_D2, are stably expressed on the respective malignant cells throughout the stages of the disease and following relapse ( 39, 40). Furthermore, G_D2-specific immunotherapy has not resulted in down-regulation of G_D2 expression in the vast majority of patients, supporting the stability of antigen expression even on its specific targeting ( 41). Importantly, B-cell precursor acute lymphoblastic leukemia was shown to arise in the CD19-positive lymphoid compartment ( 42, 43), possibly allowing for CD19-mediated eradication of the leukemia-initiating clone.

The field of tumor immunology has been further complicated by the discovery of T cells with regulatory function that are capable of suppressing the activation and function of tumor antigen-specific T cells ( 44). The best-known subset of Treg is characterized by high coexpression of CD4 and CD25 and therefore phenotypically indistinguishable from the VZV-CTL populations. Functional characteristics as well as lack of expression of the transcription factor Foxp3, however, categorize VZV-reactive CD4⁺ T cells as a nonregulatory, activated effector T-cell subset.

To specify the tumor-specific functional response of CD4⁺ VZV-CTL, we investigated the cytokine pattern triggered by tumor antigen, as well as the mechanism of tumor cytolysis. The parallel increase of IFN-γ and IL-4 in response to stimulation with tumor antigen was unexpected because they are T_H1 and T_H2 cytokines with opposing effects on CTL function ( 45). Recently, however, IL-4 has been shown to have important contributions to the development of CD8⁺ T-cell responses to tumor and viral challenges in mice ( 46). It is conceivable that via IL-4, VZV-CTLs provide help for chRec-induced activation while maintaining potent cytolytic function. Cytolysis by CD4⁺ CTL has been reported to be mediated either by perforin-dependent mechanisms ( 37) or alternatively by induction of apoptosis via Fas/CD95 on target cells ( 47). VZV-CTLs, however, lack perforin-containing lytic granules, and the CD19⁺ tumor targets are Fas negative (data not shown) and thereby resistant toward FasL. Instead, granulysin was identified in VZV-CTL. This molecule was found previously to be involved in granule-mediated cytotoxicity by perforin-negative EBV-specific CD4⁺ CTL ( 27) and may thus potentially contribute to the mechanism by which VZV-CTL lyse their targets.

Both the cytokine production profile and the cytolytic mechanism of CD4⁺ VZV-CTLs are comparable with that described in ex vivo expanded therapeutic EBV-specific T-cell cultures ( 27). Because EBV-CTLs have been safely and successfully used for adoptive immunotherapy in a large number of transplant recipients, the functional similarity between EBV-CTL and VZV-CTL supports our hypothesis that VZV-CTLs are capable of establishing a lasting immune response.

Important safety concerns are raised by the adoptive transfer of gene-modified T cells and the active immunization of immunocompromised patients. Extensive experience now exists with the therapeutic use of gene-modified T cells in vivo ( 9, 29). Due to their predetermined specificities for viral antigen, clinically relevant alloreactions are highly unlikely, and indeed, no evidence of alloreactivity was found in vitro or in vivo ( 9, 13). The risks associated with potential insertional mutagenesis can be minimized by the use of novel vector systems containing suicide safety switches. Booster doses of varicella vaccine are not associated with toxic side effects even in persons with preexisting natural immunity and in the elderly whose immunity is fading ( 25). In spite of the compromised immune system in stem marrow transplant recipients, the risk of inducing varicella symptoms by boosting adoptively transferred VZV immunity is low because VZV-specific precursor T cells would have been infused into these patients. Furthermore, a recent clinical trial has shown that enhanced memory T-cell responses against viruses can be safely induced in lymphopenic patients by combined vaccine therapy and adoptive T-cell transfer ( 30). As an alternative to live vaccines, heat-inactivated varicella vaccines with a high safety profile even in patients with severely impaired cellular immunity may be used ( 48).

In conclusion, our results show that VZV-reactive cytolytic T-cell effector functions can be efficiently redirected against tumor cells. Gene-modified VZV-CTL may represent a new source of effector cells for adoptive transfer of antitumor immunity to cancer patients. Because VZV infects only humans, a generally accepted animal model does not exist. Therefore, clinical experience will be required to determine the duration of protection, the efficacy of repeated doses of the vaccine, and the determination of safe intervals.