Tumor Ablation by Gene-Modified T Cells in the Absence of Autoimmunity

Strategies, including the use of cytokines, antibodies, and various vaccine regimens, aimed at harnessing the power of the immune system have long been investigated for the treatment of cancer (1–3). However, the generation of effective and long-lasting antitumor responses has remained a challenge. In recent studies, the genetic modification of T cells with scFv chimeric receptors that have the ability to target any characterized cell surface tumor antigen has been investigated as a new avenue of redirecting immune effector function (4–9). Adoptive transfer of scFv receptor gene–modified T cells in immunocompromised mouse models has shown great potential and demonstrated long-term protection against primary and secondary tumor challenge (8, 10).

Although the results achieved in the immunocompromised setting have been encouraging, these studies have been performed in models in which the targeted antigen has been expressed exclusively in tumor, which is rarely the case in a clinical setting. Consequently, this has not allowed investigation of whether host immunoregulatory factors may impact on the activity of transferred T cells or whether these cells may also damage normal tissue expressing the target antigen. The latter point is particularly important, given that adoptive cell therapies utilizing tumor-infiltrating lymphocytes (TILs) or gene-modified T cells have in some cases resulted in on-target toxicity, manifested as “autoimmunity” in patients including thyroiditis, uveitis, and respiratory and liver toxicity (11–14).

The generation of human Her-2 transgenic mice represents an ideal model to investigate this issue. These mice have been reported to constitutively express human Her-2 as a self-antigen both in brain tissue and in the mammary gland during lactation under the control of the whey acidic protein promoter (15). Her-2⁺ tumor cells grow progressively when injected into these mice. Thus, Her-2 transgenic mice offer a physiologically relevant model that can be utilized for the preclinical evaluation of gene-modified T cells in a setting that more closely mimics the situation in patients. In the current study, we utilized this Her-2 transgenic mouse model to evaluate the antitumor efficacy of Her-2-reactive, gene-modified T cells and assess whether these cells induce autoimmune pathology.

The MHC class I–negative C57BL/6 mouse 24JK sarcoma cell line (kindly provided by Dr. Patrick Hwu, National Institutes of Health, Bethesda, MD), 24JK cells retrovirally transduced with the human Her-2 antigen (24JK-Her-2) and GP+E86 retroviral packaging cells were cultured as described previously. The 24JK tumor cell lines were routinely tested and authenticated in the last 6 months by flow cytometry and found to be MHC class I negative as per the original line acquired. C57BL/6 human Her-2 transgenic mice and congenic Thy1.1⁺ Her-2 mice were bred at the Peter MacCallum Cancer Centre, Victoria, Australia, and used for experimentation at 6 to 16 weeks of age. All animal experimentation was approved in advance by the Peter MacCallum Cancer Centre Animal Experimentation Ethics Committee.

The GP+E86 packaging line expressing the scFv-anti-Her-2 CD28-ζ receptor that contained the extracellular scFv-anti- Her-2 monoclonal antibody (mAb) region (kindly provided by Dr. Winfried Wels) (16) was generated as previously described (17, 18). For transduction, splenocytes from C57BL/6 Her-2 transgenic mice were gene modified as previously described (19). Following transduction, splenocytes were cultured in RPMI, supplemented with 100 U/mL of IL-2 and 2 ng/mL of IL-7, and utilized for in vitro and in vivo experiments on days 7 and 8.

Expression of the chimeric anti-Her-2 receptor on the surface of T cells was determined by direct immunofluorescence with a c-myc tag Alexa Fluor 488–conjugated immunoglobulin (Cell Signaling Technology, Beverly, MA). Background immunofluorescence was assessed using an IgG2a Alexa Fluor 488–conjugated mouse isotype immunoglobulin (Invitrogen). Phenotypic characterization of transduced cells was determined using direct staining with fluorescein isothiocyanate–conjugated anti-CD4 (clone RM4-5; BD Pharmingen, San Jose, CA) and phycoerythrin-conjugated anti-CD8 (clone 53-6.7; BD Pharmingen).

The ability of transduced anti-Her-2 T cells to secrete IFN-γ in response to Her-2 antigen stimulation was analyzed by enzyme-linked immunosorbent assay (ELISA) following overnight coculture with 24JK-Her-2 or 24JK parental cells as previously described (20). The proliferative potential and capacity of transduced T cells to mediate antigen-specific tumor cell lysis was assessed by [³H]-thymidine incorporation and 5-hour chromium release assays, respectively, as described previously (21).

Groups of 5 to 10 Her-2 transgenic mice were intravenously injected with 1 × 10⁶ 24JK-Her-2 cells followed by transfer of scFv-anti-Her-2–transduced T cells or control LXSN–transduced T cells (1 × 10⁷ per dose on days 0 and 1) or left untreated, and survival of mice was monitored. In some experiments, recipient mice received transduced donor T cells from congenic Thy1.1⁺ Her-2 mice. For experiments involving lymphodepletion, Her-2 transgenic mice were irradiated (5 Gy) followed by intravenous injection of 24JK-Her-2 tumor cells (1 × 10⁶) and adoptive transfer of transduced T cells at 1 × 10⁷ per dose at days 0 and 1, days 5 and 6, or days 10 and 11. Mice were also given twice-daily intraperitoneal injections of recombinant human IL-2 (Biological Resource Branch, National Cancer Institute, Frederick, MD) involving 9 doses of 50,000 IU/200 μL given subsequent to T-cell transfer. For all experiments, mice were monitored for survival. Mice displaying overt signs of distress were culled, and lungs were harvested and weighed to record tumor burden. To investigate the trafficking and persistence of adoptively transferred T cells in vivo, 1 × 10⁷ donor T cells from congenic Thy1.1⁺ Her-2 mice were transduced and transferred into irradiated 24JK-Her-2 tumor bearing Her-2 recipient mice (Thy1.2⁺) on days 0 and 1 and received IL-2 as described previously. Flow cytometry was carried out to determine the frequency of all T cells and Thy1.1⁺ T cells in the organs.

Lung, mammary (nonlactating female), and brain tissues (female and male) from Her-2 transgenic mice were removed, fixed with 10% neutral buffered formalin, embedded in paraffin, and sectioned. To examine tumor burden in lungs of mice (days 5 and 10 after 24JK-Her-2 tumor inoculation) and toxicity to mammary and brain tissues expressing Her-2 antigen following therapy, sections were stained with hematoxylin and eosin (H&E) and analyzed by a pathologist. To examine expression of the human Her-2 antigen in mammary and brain tissues from Her-2 transgenic mice and the 24JK-Her-2 and 24JK cell lines, sections were stained with a polyclonal rabbit anti-human Her-2 oncoprotein antibody (A 0485; DAKO Corp., Real Carpinteria, CA) following antigen retrieval in 10 mmol/L of citrate buffer (pH 6.0). Images were visualized with an Olympus BX51 microscope (Olympus Corporation, Tokyo, Japan), acquired using RT SE Diagnostics Instruments SPOT camera (Diagnostic Instruments, Sterling Heights, MI) in conjunction with SPOT Advanced Version 4.6 (Diagnostic Instruments), and compiled with Adobe Photoshop and Illustrator CS4 (Adobe systems, San Jose, CA).

Differences between mice survival distributions were analyzed using log-rank (Mantel–Cox) test to determine statistical significance. P2 < 0.05 was considered significant.

Using flow cytometry, we observed reproducibly high-level receptor expression on transduced T cells (Fig. 1A) (46 ± 3.2% SEM, n = 16). There was no detectable anti-tag staining on the surface of T cells transduced with an empty LXSN retroviral vector (Fig. 1B). These transduced T cells were predominantly CD8⁺ (79.5 ± 4.4% SEM, n = 8), with only a small percentage of CD4⁺ T cells present in the culture (7.3 ± 2.7% SEM, n = 8) (Fig. 1C and D).

The functional capacity of transduced transgenic T cells was tested in several in vitro assays. Following stimulation by overnight coculture with the C57BL/6 sarcoma cell line 24JK expressing human Her-2 antigen (24JK-Her-2), transgenic anti-Her-2 T cells secreted high levels of IFN-γ as determined by ELISA (Fig. 2A), whereas coculture with the Her-2⁻ 24JK parental tumor cell line demonstrated no significant release of IFN-γ. Anti-Her-2 T cells could also significantly lyse 24JK-Her-2 target cells compared with Her-2⁻ 24JK tumor cells (Fig. 2B). The proliferative capacity of transduced transgenic T cells was measured by a [³H]-thymidine incorporation assay following stimulation with plate-bound anti-c-myc antibody. Anti-Her-2 T cells could specifically proliferate following stimulation with anti-c-myc antibody but not with isotype control antibody (Fig. 2C). Taken together, these results showed that transgenic T cells could be effectively transduced with the scFv-anti-Her-2 receptor and mediate antigen- specific cytokine secretion, killing, and proliferation.

We have previously shown that adoptive transfer of T cells gene-modified with the scFv-anti-Her-2 receptor could specifically eradicate 5-day established lung metastases in immunocompromised SCID mice (10). Thus, we wanted to assess whether transduced T cells could impact on Her-2⁺ tumor in an immunocompetent setting in a similar way. In this experiment, we compared the survival of Her-2 transgenic mice bearing 24JK-Her-2 lung metastases treated with anti-Her-2 or control LXSN–transduced T cells administered intravenously on days 0 and 1. A significant increase in survival of mice (∼20% long-term, tumor-free survivors) was observed following transfer of anti-Her-2 T cells compared with mice that received control T cells or no T cells, which all succumbed to lung disease (P2 < 0.05; Fig. 3A). It has been reported both in mouse studies and in patients that the use of lymphodepletion as a preconditioning treatment prior to adoptive T-cell transfer can enhance their therapeutic efficacy (22–25). To test whether this was the case in our model using gene-engineered T cells, Her-2 transgenic mice were irradiated prior to tumor challenge and transfer of transduced T cells and administered IL-2 between days 0 and 4. In this experiment, a striking increase in survival was observed in recipient mice treated with anti-Her-2 T cells in combination with irradiation and IL-2 (∼70% long-term, tumor-free survivors; Fig. 3B) compared with mice given anti-Her-2 T cells with no preconditioning (only 20% survivors; Fig. 3A). Interestingly, the combination of both irradiation and IL-2 along with anti-Her-2 T cells was critical for this therapeutic effect, given the decreased survival that was observed in groups of mice that received anti-Her-2 T cells with either IL-2 or irradiation. There was no survival of mice that received control T cells with the combined regimen (Fig. 3B). Importantly, treatment of mice with irradiation or IL-2 alone did not significantly alter the growth kinetics of the 24JK-Her-2 tumor in recipient mice and no overt toxicity was observed in mice following therapy (Fig. 3B). To determine whether transduced T cells could be detected following transfer, mice were challenged with 24JK-Her-2⁺ tumor and irradiated before receiving transduced donor T cells from congenic Thy1.1⁺ Her-2 mice and IL-2. In this experiment, we observed a significant increase in percentage of Thy1.1⁺ donor T cells in lungs of anti-Her-2 T-cell–treated mice compared with the mice that received control T cells at day 18 following T-cell transfer (Fig. 4A). There was no significant difference in Thy1.1⁺ T cells observed in spleens of mice treated with anti-Her-2 or control T cells. We could also detect Thy1.1⁺ donor T cells in lungs and spleen of mice at day 53 in mice treated with anti-Her-2 T cells, indicating persistence of these cells (Fig. 4B). Collectively, these set of experiments showed that adoptively transferred anti-Her-2 T cells could be detected at the tumor site and significantly impact on Her-2⁺ tumor growth in immunocompetent mice and that combined treatment with irradiation and IL-2 could further increase the therapeutic effect similar to that observed in a patient setting receiving TIL and lymphodepletion with IL-2.

Given that the majority of patients will present with established disease, we next assessed whether our therapy could impact on advanced lung metastases in mice. In 2 separate experiments, the antitumor ability of anti-Her-2 T cells in combination with irradiation and IL-2 was examined against day 5 and day 10 established disease. Immunohistochemical staining using H&E demonstrated increased 24JK-Her-2 metastases in lung sections of mice at days 5 and 10 compared with untreated mice (Fig. 5A). In the first experiment, mice were irradiated and challenged with 24JK-Her-2 tumor on day 0 followed by transfer of transduced T cells on days 5 and 6 and administered IL-2 between days 5 and 9. The majority of mice that received anti-Her-2 T cells combined with irradiation and IL-2 survived (∼80% long-term, tumor-free survivors; Fig. 5B). This effect was dependent on T cells modified with the scFv anti-Her-2 receptor, as mice treated with control LXSN–transduced T cells succumbed to lung metastases similar to mice challenged with 24JK-Her-2 tumor alone (Fig. 5B). In the next experiment, mice were irradiated on day 10 and challenged with 24JK-Her-2 tumor followed by transfer of transduced T cells on days 10 and 11 and administered IL-2 between days 10 and 14. Strikingly, a significant increase in survival of mice was observed in mice with day 10 established disease that received anti-Her-2 T cells in combination with irradiation and IL-2 compared with control treated mice (Fig. 5C). Again, no overt toxicity was observed in mice following therapy. These experiments showed that gene-modified T cells could significantly impact on established disease in immunocompetent Her-2 transgenic mice, which has important implications for development of this approach for the treatment of patients with advanced cancer.

Given that tumor relapse can often occur in patients following treatment, the ability for cancer immunotherapies to induce a strong recall response is an important issue. To test whether this was the case for our therapy, long-term surviving mice (120 days after primary 24JK-Her-2 tumor inoculation) that were irradiated and treated with anti-Her-2 T cells and IL-2 were rechallenged with either 24JK-Her-2 or parental 24JK tumor cells. A significant increase in survival of mice was observed following rechallenge with 24JK-Her-2⁺ compared with 24JK-Her-2⁻ tumor cells (Fig. 6). These experiments demonstrated for the first time in an immunocompetent setting that adoptively transferred, gene-engineered T cells could induce a strong recall response in an antigen-specific manner.

An important issue with regard to adoptive immunotherapy using gene-engineered T cells is whether these cells may cause severe autoimmune effects against normal tissue expressing the target antigen. In Her-2 transgenic mice, the human Her-2 antigen has been shown to be expressed as a self-antigen in parts of the brain and mammary tissue under lactating conditions (15). We confirmed these results and showed with a specific antibody recognizing Her-2 positive staining for this antigen in the cerebellum of brain sections from transgenic mice but not in nontransgenic littermates (Fig. 7A). Interestingly, we could also detect expression of Her-2 antigen in the mammary gland of nonlactating Her-2 transgenic mice but not in nontransgenic littermates for which staining was confined predominantly in adipocytes surrounding the ducts (Fig. 7A). As a positive control for antibody recognition of Her-2 antigen, we showed intense staining of 24JK-Her-2 tumor cells but not 24JK parental cells (Fig. 7A). The expression of Her-2 antigen in transgenic mice approximates the situation in humans in whom Her-2 can be expressed on a range of normal tissues including epithelial cells of the breast, neuronal cells, and gastrointestinal tract (26, 27). To determine whether genetically redirected T cells caused autoimmunity against normal tissue, we performed H&E staining of brain and mammary sections from Her-2 transgenic mice that were irradiated and challenged with 24JK-Her-2 tumor and received anti-Her-2 engineered T cells and IL-2 as previously. Brain and mammary tissues were then assessed for any indication of T-cell–mediated damage. Interestingly, all mice treated with T cells expressing the scFv-anti-Her-2 receptor demonstrated no tissue damage, as shown by H&E staining of mammary or brain sections from representative mice at days 18 and 50 following T-cell transfer (Fig. 7B and C). The morphologic appearance of sections from these tissues was comparable with sections either from mice treated with control T cells or from untreated mice (Fig. 7B). Given that the presence of Her-2⁺ tumor may act as a sink for transferred T cells, we also treated Her-2 transgenic mice with transduced T cells in the absence of tumor and found no evidence of autoimmune damage to brain and mammary tissues in these mice (data not shown). Thus, the transfer of genetically redirected T cells did not seem to induce any significant autoimmunity against normal host tissue expressing Her-2 target antigen at the time points analyzed.

The above-described model closely approximates the clinical setting in which the Her-2 antigen is overexpressed in ∼30% of breast cancer patients as well as several other cancers including ovarian, lung, and head and neck cancer (28–30) while also present on normal tissue including those of epithelial, mesenchymal, and neuronal origin (26, 27). In this study, we showed that anti-Her-2 T cells could significantly impact on 24JK-Her-2 lung metastases in Her-2 transgenic mice compared with control T cells.

An important consideration for the therapeutic use of adoptively transferred, gene-engineered T cells is whether they may induce extensive autoimmune damage to normal tissues expressing the target antigen (31). In this study, we investigated whether the transfer of gene-modified T cells in combination with lymphoablation and IL-2 into tumor-bearing Her-2 transgenic mice caused any autoimmune toxicity to brain and/or mammary tissues expressing the Her-2 target antigen. In these experiments, no autoimmune damage to these tissues was observed by engineered T cells at early and late time points analyzed. The fact that we did not observe any autoimmunity may be in part due to differences in Her-2 antigen levels between tumor and normal brain and mammary tissues. Our immunohistochemical analysis indicated comparably stronger Her-2 antigen staining on 24JK-Her-2 tumor cells relative to brain and mammary tissues, which is generally the case seen in the patient setting. Our results correlate well with a recent report showing no overt toxicity in mice following adoptive transfer of scFv-transduced T cells targeting mouse CD19 antigen, although in this study, scFv-transduced T cells did not persist long term in mice. This may have been due to lack of costimulation with the use of a chimeric receptor that contained only the CD3-ζ signaling domain (19). In the present study, we used an scFv receptor containing both the CD28 costimulatory and CD3-ζ domains. Our results are also consistent with another study that showed that nonmyeloablative preconditioning and adoptive transfer of carcinoembryonic antigen–specific, MHC-restricted, T cells mediated antitumor activity without associated autoimmunity (32). However, autoimmunity was observed in this model following increased lymphoablation at a myeloablative dose (9.5 Gy) or the addition of anti-IL-10 receptor blocking antibody, indicating that the type of immunomodulation employed must be treated with caution.

Despite these reports, several studies have reported mild to severe autoimmune toxicity following transfer of tumor reactive T cells (33–35) including liver toxicity in patients involving transfer of T cells gene-modified with an anti-CAIX scFv receptor (13). Toxicity, manifested by increased levels of transaminase in serum, was thought to be due to CAIX expression on bile duct epithelium. Resolution of toxicity was achieved by cessation of therapy and administration of corticosteroids.

A recent trial involving adoptive transfer of gene-modified T cells targeting the Her-2 antigen resulted in the death of a patient 5 days after treatment; this might have been due to a cytokine storm resulting from recognition of Her-2 on normal tissue (14). Pulmonary edema occurred within 1 hour of cell infusion requiring intubation. Hypotension, bradycardia, and gastrointestinal bleeding led to cardiac arrest and death. The herceptin-based scFv receptor utilized in this trial incorporated the CD28, 4-1BB, and CD3-ζ domains, which may have led to increased secretion of cytokines from T cells following antigen recognition and subsequent toxicity (14). Interestingly, in another study, no toxicity was reported in patients following transfer of autologous anti-Her-2 CTL clones targeting the same antigen (36).

In another phase I clinical study, a serious adverse event involving death of a patient was observed following adoptive transfer of T cells gene-modified to respond against CD19 (37). The patient, with chronic lymphocytic leukemia, received (1.2–3.0) × 10⁷ T cells/kg following cyclophosphamide conditioning. Hypotension and acute renal failure led to respiratory arrest. However, a definitive role for the infused T cells was not indicated in this case and death was thought to be as a result of sepsis due to infection leading to hypotension.

Collectively, studies both in preclinical mouse models and in patients have indicated that the location and levels of antigen expressed on normal tissue, the number of T cells transferred, the type of signaling domains incorporated into chimeric receptors, and the level of immune preconditioning used have to be carefully considered for adoptively transferred T-cell therapy. Given the potential severity of autoimmunity that has been reported in some studies, it would be prudent in the future to use scFv chimeric receptors targeting carefully selected antigens that are either tumor-specific or expressed at relatively low levels on normal tissue compared with tumor. In support of this idea, we have found that T cells engineered with an scFv-anti-Lewis^Y chimeric receptor could strongly react with tumor cells expressing high Lewis^Y antigen but did not react against endogenous neutrophils expressing low levels of Lewis^Y (8). The possibility of redirecting T cells to multiple tumor-associated antigens expressed on tumor cells may further enhance their specificity for tumor while minimizing their activity against healthy tissue. Another alternative may be the inclusion of a regulated “suicide” gene function such as hsv-tk or the cytoplasmic domain of Fas or an inducible caspase incorporated into genetically engineered T cells to abort any aberrant T-cell responses (38–40).

In summary, this study has demonstrated that transfer of gene-engineered T cells combined with lymphoablation and IL-2 could mediate significant antitumor effects in a self-antigen setting without causing damaging autoimmune effects. These results suggest that redirected T-cell therapy has great potential as a safe and viable option for the treatment of cancer patients in the future.

No potential conflicts of interest were disclosed.

The authors thank the Peter MacCallum Cancer Centre Experimental Animal Facility technicians for animal care. They also thank pathologist Fenella Long for analysis of H&E sections of brain and mammary tissues following therapy.

Grant Support This work was funded by Project Grants from the National Health and Medical Research Council (NHMRC), Cancer Council of Victoria, Bob Parker Memorial Fund, and the Susan Komen Breast Cancer Foundation. M.H. Kershaw and P.K. Darcy were supported by an NHMRC Senior Research Fellowship and Career Development Award, respectively. M.J. Smyth and J.A. Trapani were supported by an NHMRC Australia Research Fellowship and Senior Principal Research Fellowship, respectively.