Tumor-Specific CD8⁺ T Cells Expressing Interleukin-12 Eradicate Established Cancers in Lymphodepleted Hosts

CD8⁺ T cells migrate through tumor with high instantaneous velocity but arrest on contact with cognate antigen (1, 2). As a result, we reasoned that T cells might be able to overproduce an antitumor agent directly into a suppressive microenvironment that includes negative regulatory factors such as myeloid-derived suppressor cells and regulatory T cells (3–10). We chose to genetically engineer tumor antigen–specific CD8⁺ T cells to secrete interleukin (IL)-12, a potent proinflammatory and antiangiogenic cytokine capable of activating multiple aspects of innate and adaptive antitumor immunity (11–20). The mechanisms of antitumor action are complex, as IL-12 can directly augment the functionality of multiple end effectors such as CD4⁺ (type 1), CD8⁺, and natural killer (NK) cells (12, 21–25).

In clinical trials, attempts to give IL-12 systemically have been restricted by toxicities, limiting the benefits seen in preclinical models (12, 20, 21, 24, 26, 27). Due to systemic dosing restrictions, the ability to reach therapeutic levels of IL-12 at the tumor site likely represents a major barrier for treatment of solid tumors. Several preclinical and clinical studies have shown successful tumor regression by direct injection of IL-12 into lesions or genetic manipulations to induce tumor cells to produce IL-12 in situ (27–30). However, the majority of patients with metastatic cancer present with widespread disease not accessible to direct application at all sites, and even if lesions are approachable, uniform distribution remains problematic. Several studies have engineered tumor cells, dendritic cells, and fibroblasts to overproduce a single-chain functional IL-12, but achieving high local concentrations through a systemic approach remains elusive (26, 30–34).

In this study, we examined the effects of delivering IL-12 directly to the tumor site by adoptively transferring tumor-specific gene-engineered CD8⁺ T cells. We found that transferring small numbers of IL-12–engineered T cells into lymphodepleted hosts resulted in the destruction of large vascularized tumors without the need for vaccine or IL-2. We show that T cells engineered to produce IL-12 infiltrate tumors in higher numbers than nontransduced cells and continue to secrete supraphysiologic levels of IL-12 directly at the tumor site just before regression. T-cell receptor (TCR) specificity was critical for tumor regression, suggesting that arrested migration of T cells at the tumor site was essential for successful treatments. These results describe a systemic approach to achieve high local concentrations of IL-12, leading to dramatic improvements in adoptive cell therapies.

Female thy1.1⁺ pmel-1 TCR transgenic mice were generated in our laboratory (35) and made available online (http://www.jax.org). Female C57BL/6 mice and C57BL/6 TCR-α^−/− mice (obtained from The Jackson Laboratory) bearing B16, an H-2^b+/gp100⁺ murine melanoma, established and maintained as previously described (36), were used as recipients for adoptive immunotherapies. Experiments were conducted with the approval of the National Cancer Institute Animal Use and Care Committee.

The cDNAs for the p40 (Il12b) and p35 (Il12a) genes for IL-12 were linked by a sequence encoding a 15-amino-acid (Gly₄Ser)₃ flexible linker and inserted into a MSGV-1–based retroviral vector with a woodchuck hepatitis virus posttranscriptional regulatory element to help increase transgene expression.

The codon-optimized pmel-1 TCR-α (Tcra) and TCR-β (Tcrb) cDNA genes were linked by a sequence encoding the foot and mouth disease picornavirus 2A “self-cleaving” linker and inserted into a MSGV-1 vector (37).

Platinum Eco 293 based cells (Cell Bio Labs) were plated on poly-d-lysine–coated 100-mm plates (BD Biosciences) and transfected with 6 μg of pCL-Eco helper plasmid (Imgenex) and 9.3 μg of the MSGV-1^IL-12 or the MSGV-1 pmel-1 TCR vector with Lipofectamine 2000 (Invitrogen) overnight in antibiotic-free complete medium. Viral supernatants were harvested 36 to 48 hours after transfection.

Pmel-1 splenocytes were cultured in the presence of 1 μmol/L hgp100_25–33 and complete medium containing 60 IU/mL of recombinant human (rh) IL-2 (Chiron). For transduction of C57BL/6 splenocytes, 1 μg/mL of soluble anti-CD3 (BD Biosciences) and 1 μg/mL of soluble anti-CD28 (BD Biosciences) were used to stimulate bulk splenocytes. Two days later, splenocytes were collected and resuspended in retroviral supernatant with 60 IU/mL rhIL-2 and 10 μg/mL protamine sulfate (Abraxis Pharmaceutical Products) and spun at 1,000 × g at 37°C for 90 minutes in 24-well plates. Cultured cells were adoptively transferred 3 to 5 days after transduction (>90% CD8⁺ T cells).

Six- to twelve-week-old mice (n = 5 for all groups) were injected s.c. with 5 × 10⁵ B16 melanoma cells. After 10 to 14 days, they were irradiated with 5 Gy of total-body irradiation (TBI) and given IL-12–transduced pmel-1 CD8⁺ T cells or pmel-1 TCR– and IL-12–double-transduced C57BL/6 cells by tail vein. Tumors were measured using digital calipers and the tumor area was calculated as the product of perpendicular diameter by investigators in a blinded manner. All experiments were performed independently at least twice with similar results and all tumor curve data are shown as mean ± SEM.

Before transfer, cells were characterized by flow cytometry for CD8α, CD62L, CD44, IL-7Rα, IL-2Rα, and Sca-1 (BD Biosciences). IL-12, IFN-γ, tumor necrosis factor-α (TNF-α), and IL-2 were analyzed using intracellular staining kits (BD Biosciences) with or without a 4-hour stimulation with 50 ng/mL phorbol 12-myristate 13-acetate (PMA) and 1 μg/mL ionomycin (Sigma). Following transfer, tumor samples were harvested and lymphocytes were isolated using lympholyte cell separation media (Cedarlane Laboratories) and enumerated by flow cytometry. Transferred pmel-thy1.1⁺ cells were analyzed by flow cytometry for thy1.1 (CD90.1), NK1.1, CD4, CD8α, and IL-12 expression. H&E staining was performed on paraffin-fixed tumor samples and analyzed at ×100 magnification by an Olympus IX-FLA microscope. Real-time reverse transcription-PCR was done as previously described (38).

Tumor growth slopes were compared using Wilcoxon rank sum test. One-way ANOVA and Student's t tests were used to test for significant differences in enumeration assays. P < 0.05 was considered significant.

To assess the ability of tumor-specific T cells to overproduce IL-12, we constructed a retrovirus based on a derivative of the murine stem cell virus, MSGV-1 (39), encoding a single-chain bioactive IL-12 obtained by fusing the p35 and p40 subunits with a flexible (Gly₄Ser)₃ linker (40–42) and designated the construct as MSGV-1^IL-12 (Fig. 1A, left). We transduced pmel-1 CD8⁺ T cells, which express a transgenic TCR specific for the melanoma-associated antigen, gp100, with this vector to express high levels of IL-12 (designated as pmel-1^IL-12-TD; Fig. 1A, right). We next examined the phenotype of pmel-1^IL-12-TD T cells used for adoptive transfer experiments and showed several distinct characteristics compared with mock-transduced cells (Supplementary Fig. S1). IL-12–engineered cells also expressed higher levels of IFN-γ (Fig. 1B and C) and TNF-α (Supplementary Fig. S2A) on secondary stimulation, but expressed less IL-2 (Supplementary Fig. S2B). Two critical T-box transcription factors for CD8⁺ T cells were altered with increased relative T-bet expression and downregulation of eomesodermin (Supplementary Fig. S3; ref. 43). On the other hand, IL-12–engineered cells underwent apoptosis when prolonged in culture beyond 7 days, failed to proliferate on secondary stimulation (Supplementary Fig. S4), and exhibited a small decrease in cytotoxic ability (Supplementary Fig. S5).

To assess the in vivo efficacy and toxicity of pmel-1^IL-12-TD cells, we adoptively transferred decreasing doses of the gene-modified T cells cultured in vitro for 5 to 7 days into sublethally irradiated mice (5-Gy TBI) bearing large vascularized B16 tumors established for 14 days. Surprisingly, we found that as few as 10,000 pmel-1^IL-12-TD cells mediated significant tumor destruction (Fig. 2A, left). This potent antitumor response was observed in the absence of high-dose recombinant IL-2 (rIL-2) and a gp100 vaccine, which we previously found were required for effective antitumor immunity (35). It is noteworthy that we observed tumor destruction without significant weight loss (Fig. 2A, right) or pathologic end-organ damage (Supplementary Fig. S6). Although all the experiments in this study were completed without observable toxicities, administration of doses >500,000 pmel-1^IL-12-TD cells given alone or in concert with IL-2 and vaccination produced weight loss and decreased survival (Supplementary Fig. S7). These toxicities were clearly related to the increased number of cells transferred and, hence, likely correlated to the amount of IL-12 produced systemically, a well-described clinical phenomenon (44). Experiments in mice treated with <500,000 cells did not show any clear signs of systemic side effects. Thus, small numbers of pmel-1 CD8⁺ T cells engineered to secrete IL-12 safely enabled the destruction of large established tumors.

Due to pmel-1^IL-12-TD cells undergoing apoptosis after 1 week in culture, we were surprised by the potency of these cells and next assessed the in vivo persistence of transferred cells, a characteristic classically associated with improved tumor regression (38). We adoptively transferred thy1.1 gene–marked, pmel-1^IL-12-TD cells into sublethally irradiated (5 Gy) tumor-bearing mice and detected increased engraftment of pmel-1^IL-12-TD cells compared with nontransduced cells on days 3 and 7, but found that the persistence of IL-12–engineered cells dramatically dropped on day 14 (Fig. 2B).

We were surprised by the degree of tumor destruction despite the lack of persistence and therefore hypothesized that there may be differences at the tumor site contributing to the marked enhancement in response. Similar to previous experiments, we adoptively transferred thy1.1 gene–marked, pmel-1^IL-12-TD cells into sublethally irradiated tumor-bearing mice and harvested the tumors 1 week after transfer. We also observed an increase in cellular infiltration within tumors in association with higher degrees of tumor necrosis in mice treated with pmel-1^IL-12-TD cells (Fig. 3A). We next quantified the characteristics and numbers of tumor-infiltrating cells and found that mice treated with pmel-1^IL-12-TD cells had a significant increase in the number of CD8⁺ thy1.1⁺ adoptively transferred T cells (Fig. 3B, left). Importantly, the transgene expression of IL-12 in transferred cells remained robust (Fig. 3B, right), indicating the ability of these cells to secrete supraphysiologic amounts of IL-12 directly into the tumor microenvironment just before regression. We next assessed infiltration of tumors by endogenous cells (thy1.1⁻) and observed a significant increase in NK cell infiltration and an increased trend in endogenous CD8⁺ T-cell infiltration (Fig. 3C). On the other hand, we did not observe any differences in the number of endogenous CD4⁺ T cells infiltrating the tumors (data not shown). It is important to note that all mice received a 5-Gy TBI lymphodepleting regimen just before adoptive transfer, indicating that tumors were infiltrated by reconstituting endogenous cells. Thus, pmel-1^IL-12-TD cells were capable of delivering IL-12 at therapeutic levels directly into the tumor microenvironment, leading to an increase in infiltration of transferred cells along with endogenous NK and CD8⁺ T cells.

Based on the results from cells infiltrating the tumor, we next assessed the ability of nontransduced pmel-1 CD8⁺ T cells (mock) to assist in tumor eradication. We hypothesized that cotransferring nontransduced pmel-1 CD8⁺ T cells in combination with pmel-1^IL-12-TD cells may affect antitumor responses. We adoptively transferred pmel-1^IL-12-TD cells alone or in combination with mock-transduced pmel-1 CD8⁺ T cells into sublethally irradiated tumor-bearing mice. We did not observe any significant differences in antitumor responses by adding nontransduced pmel-1 CD8⁺ T cells, most likely due to the potency of pmel-1^IL-12-TD cells alone (Fig. 4A). To further understand the nature of these antitumor responses, we tested the importance of local versus systemic IL-12. We sought to ascertain whether we could replicate the successful treatments using exogenously provided IL-12. We compared pmel-1^IL-12-TD cells with mock-transduced pmel-1 CD8⁺ T cells adoptively transferred into sublethally irradiated tumor-bearing mice in combination with increasing doses of exogenous rIL-12 given at doses up to 2 μg/d. Unlike pmel-1^IL-12-TD, exogenously provided rIL-12 given together with pmel-1 CD8⁺ T cells alone did not induce major tumor regression (Fig. 4B). We also compared pmel-1 CD8⁺ T cells exposed to rIL-12 ex vivo (Tc1) with pmel-1 CD8⁺ T cells engineered to secrete IL-12 and found that pmel-1 Tc1 cells failed to exhibit antitumor responses similar to those of pmel-1^IL-12-TD cells when transferred in similar low numbers (Fig. 4B). Thus, IL-12 produced by the transferred T cells seemed to be more effective than that provided exogenously.

The reason for the selective retention and efficacy of IL-12–producing tumor-specific T cells was unclear, but the requirement of host immunodepletion was unambiguous in this model (Fig. 5A). Therapy failed when endogenous cells were not depleted by a preparative dose of 5-Gy TBI. We hypothesized that host immunodepletion functioned to create a “niche” that facilitated engraftment of adoptively transferred cells, but we did not measure significantly greater numbers of pmel-1^IL-12-TD cells in the tumors or the spleens of mice receiving 5-Gy TBI (Fig. 5B, left). Instead, we observed that lymphodepleted mice had significantly lower numbers of intratumoral CD4⁺ T cells (Fig. 5B, right) and that these remaining T cells from mice receiving 5-Gy TBI had lower expression of Foxp3, the transcription factor marking regulatory T cells (Fig. 5B, right). To further test the negative functional effect of endogenous T cells, we transferred pmel^IL-12-TD cells into nonablated wild-type (WT) versus TCRα^−/− (no endogenous T cells) tumor-bearing mice (Fig. 5C). We observed a potent antitumor response in nonablated TCRα^−/− mice, indicating that endogenous T cells were indeed imparting a negative or regulatory effect on the adoptively transferred pmel^IL-12-TD cells. Several studies by our laboratory and others have described the importance of a preparative ablative regimen for improving adoptive T-cell transfer experiments, and once again, we show a similar requirement in this model (36, 45–47).

The one aspect of the therapeutic response that still remained unclear was the importance for a tumor antigen–specific TCR. To address the question of specificity using “open-repertoire” CD8⁺ T cells, we developed a retrovirus that encoded the α/β pmel-1 TCR. We created three distinct cell types from a common precursor pool of CD8⁺ splenocytes expressing (a) the pmel-1 TCR alone, (b) IL-12 alone, or (c) both the pmel-1 TCR and IL-12 (Fig. 6A). We found that the amounts of IL-12 produced by single- and double-transduced cells were similar (Fig. 6A). Interestingly, open-repertoire CD8⁺ T cells expressing either the single-chain IL-12 or the pmel-1 TCR failed to induce tumor regression in sublethally irradiated tumor-bearing mice (Fig. 6B). The treatment was also ineffective when mice were given a 1:1 mixture of CD8⁺ T cells expressing the pmel-1 TCR mixed with cells expressing the single-chain IL-12. Only double-transduced CD8⁺ T cells coexpressing the pmel-1 TCR and IL-12 were able to induce potent antitumor responses (Fig. 6B). We also found that CD8⁺ T cells possessing both the pmel-1 TCR and IL-12, but not either component alone, accumulated at the tumor site in significantly higher numbers (Fig. 6C). These results indicated that IL-12 must be produced by T cells with specificity for the tumor to increase local infiltration and induce antitumor responses. Furthermore, our experiments revealed that enhanced engraftment of the gene-engineered T cells possessing both the pmel-1 TCR and IL-12 occurred significantly at the tumor site but not within the spleens or lymph nodes of the same recipient mice (Fig. 6C). These results were consistent with the observations seen previously with transgenic pmel-1 CD8⁺ T cells transduced with IL-12.

The adoptive transfer of autologous tumor-infiltrating lymphocytes (TIL) or TCR-redirected peripheral blood lymphocytes (PBL) is a promising treatment for patients with metastatic cancer (48, 49). The experiments described in this report show that small numbers of tumor antigen–specific CD8⁺ T cells engineered to produce high levels of IL-12 can lead to the regression of large vascularized tumors without the need for exogenous IL-2 or vaccine in lymphodepleted hosts. The marked improvement in treatment was associated with an increase in tumor infiltration by adoptively transferred T cells, along with an increase in endogenous NK cells and endogenous CD8⁺ T cells from the reconstituting host. Based on the delayed kinetics of tumor destruction, modulation of endogenous host immune cells likely plays an important role in facilitating tumor destruction, but the effect of host NK cells and CD8⁺ T cells remains unclear.

Nevertheless, one of our key findings in this study included isolating gene-engineered T cells from tumors 1 week following adoptive transfer and observing the continued stable expression of high levels of IL-12. At the same time point, we also observed an increase in the degree of necrosis within tumors treated with pmel-1^IL-12-TD T cells, which was not surprising given the overall decrease in tumor size. We found that the expression of a tumor-specific TCR was critical for the ability of T cells to deliver IL-12 to the tumor microenvironment, suggesting that the arrested migration of T cells on recognizing cognate antigen played a mechanistic role in the local delivery of IL-12.

Interestingly, a preparative lymphodepleting regimen, known to deplete suppressive factors such as regulatory T cells, was necessary for successful tumor treatments. We were surprised by the requirement for lymphodepletion, largely due to the observation that the transfer of IL-12–engineered T cells into a lymphodepleted host resulted in an increase in endogenous NK and CD8⁺ cells within the tumor. Our laboratory and others have described the ability of host cells to function as cytokine sinks, limiting the ability for homeostatic activation of adoptively transferred cells (36). Lymphodepletion likely removes endogenous regulatory T cells, such as CD4⁺ Foxp3⁺ cells, along with other T cells that function as suppressors and “sinks,” allowing for transferred IL-12–engineered T cells to arrest within the tumor on recognizing cognate antigen. Once the host immune system reconstitutes, the changes within the tumor microenvironment induced by IL-12 likely trigger multiple downstream factors contributing to the profound changes observed. Thus, the reconstituting endogenous immune response to IL-12 secretion within the tumor may play a significant role in the antitumor effects described in our experiments.

In regard to treatment-related toxicities, none of the mice in this study had observable toxicities attributed to the transferred T cells. However, during initial pilot experiments, we observed decreased survival when we transferred >500,000 pmel^IL-12-TD cells. This observation likely correlates directly to the level of systemic IL-12 produced by the constitutive expression of IL-12 in retrovirally transduced T cells. When we transferred 500,000 cells or less, thus decreasing the amount of systemic IL-12, we did not observe the previously witnessed toxicities. Whereas TCR mispairing, which leads to off-target recognition, has been raised as a theoretical concern, we did not observe any graft-versus-host–like toxicities in experiments following double transduction of open-repertoire CD8⁺ T cells with the pmel TCR and IL-12 (50). The inability of IL-12–engineered cells to persist long-term likely contributed to mitigating any toxicities related to TCR redirection. In regard to efforts to translate our findings into human clinical trials, a careful phase I dose escalation trial will likely be necessary to obtain a safe therapeutic treatment cell number. Our group has taken additional measures to reduce the level of systemic IL-12 toxicity by developing retroviral vectors driven by an NFAT (nuclear-factor of activated T cells) promoter, capable of producing the single-chain functional IL-12 only on TCR stimulation.³

In conclusion, engineering TILs or antigen-specific PBLs with IL-12 may dramatically improve cellular therapies against metastatic cancer. Furthermore, decreasing the number of cells required for adoptive transfer and eliminating the need for systemic IL-2 may help facilitate the widespread acceptance of a promising treatment modality. We recently completed the construction and amplification of a γ-retrovirus encoding a single-chain human IL-12 using good manufacturing process conditions. This retrovirus will be used for the transduction of naturally occurring and receptor-engineered T cells in the treatment of patients with advanced cancers. In the future, we may be able to engineer tumor-specific lymphocytes to locally deliver a wide array of immunomodulating payloads.

There are no potential conflicts of interest.

We thank Richard Lee (Department of Pathology, National Cancer Institute) for blinded pathologic scores and Dorina Frasheri for excellent technical assistance.

Grant Support: Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.

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.