Immunologic Research 2000;21/2–3:279–288
Targeted Cytokines for Cancer Immunotherapy
Holger N. Lode Ralph A. Reisfeld
Targeting of cytokines into the tumor microenvironment using antibody-cytokine fusion proteins, called immunocytokines, represents a novel approach in cancer immunotherapy. This article summarizes therapeutic efficacy and immune mechanisms involved in targeting interleukin-2 (IL-2) to neuroectodermal tumors using ganglioside GD2-specific antibody-IL-2 fusion protein (ch14.18IL-2). Treatment of established melanoma metastases with ch14.18IL-2 resulted in eradication of disease followed by a vaccination effect protecting mice from lethal challenges with wild-type tumor calls. In a syngeneic neuroblastoma model, targeted IL-2 was effective in the amplification of a weak memory immune response previously induced by IL-12 gene therapy using an engineered linear version of this heterodimeric cytokine. These findings show that targeted IL-2 may provide an effective tool in cancer immunotherapy and establish the missing link between T cell–mediated vaccination and objective clinical responses.
Neuroblastoma Melanoma Immunotherapy Immunocytokines Interleukin-2 T cell memory
Introduction The discovery of antibodies by Emil von Behring in 1890 was followed by Paul Ehrlich’s proposal in 1906 to apply them as “magic bullets” and “poisoned arrows” to specifically direct toxic substances to pathogenic targets (1). However, it was nearly a century before antibody-based therapies became
Ralph A. Reisfeld, PhD The Scripps Research Institute Department of Immunology, R218, IMM13 10550 N. Torrey Pines Road La Jolla, CA 92037 E-mail: [email protected]
© 2000 Humana Press Inc. 0257–277X/00/ 21/2–3:279–288/$12.50
Department of Immunology The Scripps Research Institute La Jolla, CA 92037
established. This development was aided considerably by the discovery of monoclonal antibodies (Mabs) directed against well-characterized antigens by Kohler and Milstein (2) in 1975. Subsequently, the introduction of recombinant DNAtechnologies and their rapid advances were key factors in accelerating the development of an array of bioengineered antibodies. This article focuses on one example
of bioengineered antibodies—recombinant antibody-cytokine fusion proteins—and summarizes data obtained with these immunocytokines in preclinical studies of tumor immunotherapy in the our laboratories. Immunocytokines Recombinant antibody-cytokine fusion proteins are immunocytokines that achieve sufficient concentrations in the tumor microenvironment to effectively stimulate a cellular immune response against tumors. This contrasts with passive immunotherapy by antibodies directed against tumor-associated antigens, which utilizes the natural effector mechanisms of antibodies to destroy tumor cells. More important, immunocytokines provide a tool for active tumor immunotherapy by increasing the cytokine concentration in the tumor microenvironment and thereby potentiating immunogenicity of syngeneic tumors followed in some cases by T cell activation and a subsequent memory immune response. Immunocytokines are neither limited by a patient-specific modus operandi nor by the antigenic heterogeneity of tumor cells, because only a limited number of antigen sites are required as their docking sites. Once placed into the tumor microenvironment, immunocytokines are capable of activating and expanding a variety of immune effectors, including T lymphocytes, natural killer (NK) cells, macrophages, and granulocytes, and thereby eradicate tumor cells and their metastases. This effect can amplify insufficient T cell immune responses previously induced by a cancer vaccine and lead to effective tumor eradication followed by a long-lasting protective memory. The construction of immunocytokines involves the generation of sequences coding for the cytokines by reverse transcriptase polymerase chain reaction (RT-PCR) with primers that include designated restriction sites used
for cloning purposes. These cytokine genes are fused with the human Cγ1 gene at the C-terminal end of the heavy chain of an antibody. Gillies et al. (3) inserted the fused gene of a human/mouse chimeric antiganglioside GD2 antibody (ch14.18) into the vector pd HL2 encoding the dihydrofolate reductase gene. The same vector carried the gene encoding for the light chain of ch14.18 in a separate expression unit. The expression plasmid was transduced in the Sp2/0-Ag14 hybridoma cell line by protoplast fusion and selected in the presence of increasing concentrations (100–5 µM) of methotrexate. Purification of fusion proteins makes use of the Fc portion of the antibody molecule that selectively binds Protein-A Sepharose (3). The biologic activities of the ch14.18-IL-2 immunocytokine were fully maintained, because its interleukin-2 (IL-2) activity was equivalent to that of commercially available rhIL-2 and its antibody binding affinity remained in the nanomolar range. Initial experiments in xenograft models of human melanoma and neuroblastoma tumors in severe combined immunodeficiency disease (SCID) mice reconstituted with human LAK cells clearly demonstrated the superior effect of the immunocytokine when compared to IL-2 alone or to mixtures of antibody and IL-2 (4). The following discussion is devoted entirely to treatment effects and mechanisms of the ch14.18-IL-2 immunocytokine in syngeneic tumor models of melanoma and neuroblastoma. Melanoma Human melanoma is a neuroectodermal tumor characterized by the upregulated expression of various gangliosides, including disialoganglioside GD2 (GD2). Because murine B16 melanoma fails to express ganglioside GD2, these cells were transfected with human genes encoding for two enzymes catalyzing the last
Lode and Reisfeld
steps of GD2 biosynthesis: β-1,4-N-acetylgalactosaminyltransferase and α-2,8-sialyltransferase. The transduced B16 cells stably expressed GD2, bound the ch14.18-IL-2 fusion protein, and formed experimental hepatic and pulmonary metastases in syngeneic C57BL/6 mice following intrasplenic or iv injection, respectively. Biodistribution experiments with 125I-labeled ch14.18-IL-2 demonstrated specific tumor targeting by effective localization in the lungs and livers of tumor-bearing mice (5). Therapeutic Effects Immunotherapy of established hepatic and pulmonary melanoma metastases in C57BL/6J mice with ch14.18-IL-2 fusion protein 1 wk after tumor cell inoculation completely eradicated established metastases in the vast majority of mice. This was indicated by histologic examination of tissue sections. The treatment effect proved to be specific since a nonspecific fusion protein directed against the human epidermal growth factor (EGF) receptor, ch225-IL-2, did not show any antitumor effect (5). The efficacy of fusion protein therapy was further documented by a twofold increase in lifespan of only ch14.18-IL-2 immunocytokine-treated mice. Immunocytokine therapy also overcame heterogeneity of the tumor docking sites, as indicated by a successful eradication of melanoma metastases achieved with only a small percentage of tumor cells expressing the docking site when GD2+ cells admixed with GD2– cells were used as in vivo targets (6). Mechanisms of Immunocytokine-Induced T Cell Responses ch14.18-IL-2 immunocytokine treatment of melanoma induced a T cell–dependent immune response, followed by a long-lasting protective tumor immunity against a lethal challenge with wild-type (WT) tumor cells.
Immunocytokines for Cancer Immunotherapy
Immunocytokine-treated animals exhibited an inflammatory response with heavy infiltrates of lymphocytes intermixed with few granulocytes and macrophages. A strong staining of CD8+ T cells and, to a lesser extent, CD4+ T cells infiltrating the tumor microenvironment was observed whereas occasional NK cells were located primarily in the tumor periphery (6). More rigorous proof for a T cell–mediated mechanism was obtained in C57BL/6 scid/scid mice, which lack mature T and B cells, and in C57BL/6 beige/beige mice devoid of functional NK cells. The absence of NK cells did not hinder the successful treatment of established pulmonary melanoma metastases, but the lack of T lymphocytes abrogated this effect. This was further documented by selective in vivo depletion of distinctive T cell subpopulations and in vitro cytotoxicity assays indicating that CD8+ T cells are mediators of the ch14.18-IL-2-induced, major histocompatibility complex (MHC) class I–restricted, immune response in this tumor model (5). Interestingly, the depletion of CD4+ T cells partially diminished the efficacy of the immunocytokine-mediated immune response. In fact, the absence of cytolytic activity in the CD4+ T cell compartment against melanoma target cells in vitro indicates a helper function of CD4+ T cells for most optimal activation of a CD8 + T cell response. The mechanism of help provided by CD4+ T cells was shown to be mediated by CD40/CD40 ligand (CD40L) interactions, and not by the release of IL-2 from this T cell subpopulation. This finding is based on the observation that ch14.18-IL-2-mediated antimelanoma activity is partially abrogated in CD40L knockout, but not in IL-2 knockout mice. Furthermore, successful therapy with the ch14.18-IL-2 immunocytokine translated into a long-lived and transferable tumor immunity. Thus, mice cured of established sc melanoma or pulmonary metastases by the
IL-2 immunocytokine completely rejected a subsequent lethal iv challenge with melanoma cells in at least 50% of all animals up to 4 mo after the initial treatment (7). More important, challenges with an unrelated syngeneic tumor cell line (EL4) expressing the docking site GD2 induced fulminant metastases in the same mice that could be fully protected against challenges of murine melanoma cells. These data indicate that as-yet undefined tumor antigens recognized by T cells are required to induce a tumor-protective immunity that is completely independent of the GD2 docking antigen and that is simply used to deliver IL-2 to the tumor microenvironment (7). In summary, the data obtained in the melanoma model provide proof of a concept for an immunocytokine to direct IL-2 to the tumor microenvironment, eradicate established metastases, and induce a T cell–mediated memory immune response, suggesting that this approach may be used as a nonindividualized tumor vaccine. Neuroblastoma An effective treatment of stage IV neuroblastoma presents a major challenge in pediatric oncology (8). Thus, it was encouraging that passive immunotherapy in adjuvant settings with human/mouse chimeric antiganglioside GD2 antibody ch14.18 resulted in a response rate of >50% in phase I and phase I/II clinical trials, including several long-term and complete remissions of stage IV patients (9). Data from a phase I/Ib clinical trial using a combination therapy with murine anti-GD2 Mab 14.G2a and rhIL-2 also showed antitumor activity (10). Consequently, preclinical experiments with a recombinant ch14.18-IL2 immunocytokine in a syngeneic immunocompetent model were a logical consequence, which was established by using GD2-positive NXS2 cells in A/J mice (11). More important, following sc and iv injections, NXS2 cells
metastasized both spontaneously and experimentally to sites typical for human neuroblastoma, including bone marrow, liver, lymph nodes, and adrenal glands. Therapeutic Effect and Mechanisms of an IL-2 Immunocytokine Against Murine Neuroblastoma The ch14.18-IL-2 immunocytokine suppressed both spontaneous and experimental bone marrow and liver metastases induced by NXS2 neuroblastoma cells, whereas mixtures of antibody and rhIL-2 at an equivalent concentration proved ineffective. Immunocytokine-treated mice also revealed no macroscopic liver disease and lacked detectable metastases in their bone marrow as determined by tyrosine hydroxylase RT-PCR capable of detecting one tumor cell in 100,000 naïve bone marrow cells (11). In contrast to our previously reported T cell–mediated immune response following immunocytokine therapy of murine melanoma, several lines of evidence indicate that the effector mechanism involved in the syngeneic neuroblastoma model is exclusively mediated by NK cells. First, treatment of metastases with ch14.18-IL-2 was completely effective in T cell–deficient scid/scid mice and was only abrogated in scid/beige mice that lack both T and NK cells. However, reconstitution of such animals with NK cells reestablished the therapeutic effect of ch14.18-IL-2. Second, depletion of NK and CD8+ T cells in immunocompetent, syngeneic A/J mice eliminated the immunocytokine’s treatment effect only in NK cell–depleted mice (12,13). A role for CD8+ T cells was further excluded when mice successfully cured of liver and bone marrow metastases by ch14.18IL-2 were challenged by sc injection of 1 × 106 NXS2 cells and showed equal sc tumor growth as naïve mice, indicating the absence of a memory response, a typical feature of CD8+ T cells but not NK cells. Third, immuno-
Lode and Reisfeld
histochemical analyses of inflammatory infiltrates in livers of mice successfully treated with ch14.18-IL-2 showed positive staining for NK cells but not CD8+ T cells (12). Amplification of Suboptimal CD8+ T Memory Cells by Targeted IL-2 In contrast to the absence of a T cell memory following IL-2 immunocytokine treatment in this model, gene therapy with a single-chain IL-12 (scIL-12) demonstrated efficacy in mounting a CD8+ T cell–mediated immune response against NXS2 neuroblastoma (14). This was accomplished by genetically fusing the two chains of heterodimeric IL-12, p35 and p40, using a flexible protein linker, thereby generating a linearized IL-12 that maintained one-sixth of specific IL-12 activity, as determined by the release of interferon-γ compared to commercial IL-12 standard. The introduction of scIL-12 into NXS2 cells, generation of NXS2 clones with stable scIL-12 secretion, and subsequent vaccination of syngeneic A/J mice with this cellular vaccine was completely effective in protecting mice from a lethal challenge with WT NXS2 cells, in contrast to empty vector controls (14). The antitumor effect was clearly mediated by CD8+ T cells, as demonstrated by the absence of immune protection in mice depleted of CD8+ T cells in vivo and by MHC class I–restricted NXS2 target cell lysis in vitro. Interestingly, transfection of IL2 into NXS2 cells and subsequent vaccination of A/J mice proved ineffective in activating CD8+ T cells, consistent with the findings obtained when using immunocytokine targeted IL-2. Surprisingly, the memory immune response elicited by scIL-12 gene therapy was only short, as indicated by the reappearance of liver and bone marrow metastases following WT tumor cell challenge 4 wk after initial vaccination. We analyzed the hypothesis that a decrease in functional T cell memory over time is ampli-
Immunocytokines for Cancer Immunotherapy
fied by targeted IL-2 in this model, which is an ideal system because IL-2 alone was shown to be inefficient in activating a T cell response. This was examined by vaccinating animals with scIL-12-producing NXS2 cells followed by a lethal challenge with NXS2 WT tumor cell challenge and iv injection of small, noncurative doses of ch14.18-IL-2 (Table 1). Only animals receiving scIL-12 NXS2 cellular vaccines followed by injections with tumor-specific ch14.18-IL-2 immunocytokine revealed an absence of liver and bone marrow metastases in the majority of cases (Table 1). This finding was consistent with the presence of a CD8+ T cell memory phenotype, reactivation of CD8+ T memory cells as determined by upregulation of CD25, and effective NXS2 target cell killing only in this group of mice (Fig. 1). All these results were in contrast to control experiments with irrelevant vaccines, i.e., mice vaccinated with NXS2 cells transfected with empty vector and control experiments using nonspecific immunocytokine ch225-IL-2 recognizing the human EGF receptor not expressed on NXS2 cells and equivalent mixtures of antibody and cytokine (15). More important, tumor-specific delivery of IL-2 was necessary to accomplish effective amplification of CD8+ T cell memory, since nonspecific ch225-IL-2 immunocytokine was ineffective in this respect. Effective reamplification of CD8 + T cell memory was also achieved up to 3 mo after initial vaccination (Table 2), which indicates the strength of this principle. Figure 2 illustrates the putative mechanisms involved in these events. Accordingly, the production of scIL-12 in the tumor microenvironment increases the immunogenicity of the syngeneic tumor, leading subsequently to the induction of a CD8+ T cell response in conjunction with tumor antigen peptides being presented via MHC class I on the surface of tumor cells and via MHC class II by antigen-
Table 1. Effect of a tumor-specific antibody IL-2 fusion protein on amplification of tumor-protective immunity induced by scIL-12 gene therapy
Mice were vaccinated by sc injection of 5 × 106 NXS2 cells genetically engineered to produce scIL-12 and challenged by a lethal iv injection of 5 × 104 NXS2 WT cells 28 d after the initial vaccination. †Treatment was initiated at d 5 after tumor cell challenge by five daily iv injections of phosphate buffered saline (PBS), 10 µg of ch14.18 antibody + 30,000 IU of rhIL-2, 10 µg of the nonspecific ch225-IL-2 fusion protein, or 10 µg of the tumorspecific ch14.18-IL-2 fusion protein. ‡Bone marrow metastases were staged according to results obtained by high- and low-sensitivity tyrosine hydroxylase RT-PCR as described previously (11). *Liver metastases were staged according to the percentage of metastatic liver surface: 0 = 0%, 1 = 75%. ¶Differences in bone marrow staging, liver metastases, or liver weights among fusion protein–treated mice and all control groups were statistically significant (p < 0.05). §T cells were depleted by ip injection of 500 µg of anti-CD4 and anti-CD8 antibodies at d 26, 31, 39, and 46.
Fig. 1. (see facing page) Effect of ch14.18-IL-2 treatment on reactivation of CD8+ T cells following initial vaccination with scIL-12 NXS2 cells. (A) Phenotype of CD8+ T cells after vaccination of A/J mice with NXS2 cells genetically engineered to secrete scIL-12. Splenocytes of mice (n = 4) injected with scIL-12 NXS2, NXS2 cells carrying the empty vector, or naïve mice were analyzed 28 d after vaccination by twocolor flow cytometry. Differences between scIL-12 NXS2 vaccinated and naïve mice or mice receiving the irrelevant vaccine were statistically significant (*p < 0.02). (B) Detection of activated CD8+ T cells after vaccination, challenge, and boost with ch14.18-IL-2 fusion protein. Splenocytes of naïve mice (n = 4), scIL12 NXS2–vaccinated mice, and mice injected with NXS2 cells carrying the empty vector were analyzed for CD25+/CD8+ T cells 5 d after five daily iv injections of 10 µg of ch14.18-IL-2, an equivalent mixture of 10 µg of ch14.18 antibody, and 30,000 IU of IL-2 or 10 µg of nonspecific ch225-IL-2 fusion protein.
Fig. 1. (continued) Differences between scIL-12 NXS2–vaccinated and ch14.18-IL-2 boosted mice and all control groups were statistically significant (*p