Antitumor and Immunotherapeutic Effects of Activated Invasive T Lymphoma Cells That Display Short-Term Interleukin 1α Expression¹

A significant percentage of lymphoid tumors in humans are non-Hodgkin’s T-cell lymphomas, which are weakly or nonimmunogenic, highly aggressive, and frequently resistant to conventional therapeutic approaches. Until now, only limited attempts have been made to treat T-cell malignancies using immunotherapeutic methods; the feasibility of recombinant cytokines and techniques of gene transfer has opened new and promising avenues for such approaches (1, 2, 3, 4, 5, 6, 7, 8, 9).

IL³-1 is a pleiotropic cytokine with multiple antitumor activities that are potentially more effective than those of other cytokines used in immunotherapy (10, 11, 12, 13, 14). Thus, IL-1 activates nonadaptive immune surveillance cells (i.e., natural killer cells and macrophages) that have the potential to limit tumor growth before specific immunity develops, and it also facilitates the development of antitumor specific immune responses, primarily by affecting IL-2 secretion or the expression of high-affinity IL-2 receptors, thus serving as a costimulatory signal for T-cell activation (reviewed in Refs. 15 and 16). In addition, IL-1 is an inducer of several cytokines in diverse cells with the potential to amplify and sustain antitumor immune responses. Antitumor effects of IL-1 have been described in experimental tumor systems and in Phase I and Phase II clinical trials (17, 18, 19, 20, 21, 22, 23, 24). In those trials, treatment with IL-1 has consisted of repeated bolus injections of the cytokine, which result in unfavorable side effects for the patients (i.e., hypotension, fever, and so on), and studies were terminated before beneficial antitumor effects could be observed.

We have initiated studies to demonstrate the antitumor effects of IL-1α expression by malignant cells (1, 10, 25, 26, 27, 28, 29). IL-1α-transduced fibrosarcoma cells, which constitutively express the cytokine, lose their tumorigenicity; they either do not grow at all in mice, or they start to grow and thereafter regress. Regression of IL-1α-positive fibrosarcomas mainly involves the activation of T cell-mediated immune responses and the development of a long-term specific immune memory, which protects against a challenge of violent parental cells (IL-1α negative). On the basis of IL-1α expression by fibrosarcoma cells, we developed immunotherapeutic approaches in which small existing tumors of violent fibrosarcoma regressed following treatment with syngeneic cells expressing IL-1α (1, 10, 25, 26, 27, 28, 29).

Gene transfer approaches, although widely used to intervene in the growth of tumors in laboratory animals (1, 2, 3, 4, 5, 6, 7, 8, 9), have not yet been adequately applied in clinical trials. Induction of short-term expression of cytokines in tumor cells at critical intervals for mounting antitumor immune responses may serve as an additional immunotherapeutic approach. Indeed, we have previously shown that fibrosarcoma cells that express short-term activity of IL-1α following in vitro activation with cytokines and bacterial LPS lose their tumorigenicity and can serve as vaccine cells in treating violent tumors (1, 10, 25, 26, 28, 29). In fibrosarcomas, similar levels of IL-1α expression, regression kinetics and immunotherapeutic effects are manifested following expression of IL-1α in constitutive or short-term manners. It was of interest to assess whether induction of short-term activity of IL-1α in malignant cells is also feasible in tumors of other lineages and whether it may be effective in reducing metastasis. In this study, we have evaluated the antitumor effects of short-term expression of IL-1α by activated T lymphoma cells and its possible immunotherapeutic potential in treating metastatic T-cell malignancies.

BALB/c female mice were purchased from Harlan Laboratories (Jerusalem, Israel) and were housed at the Animal Facilities of the Faculty of Health Sciences, Ben-Gurion University of the Negev (Beer-Sheva, Israel). Six- to 8-week-old mice were used for experimental studies.

Cells were grown in RPMI 1640 containing 100 IU/ml penicillin, 100 μg/ml streptomycin, 2 mml-glutamine, 2-mercaptoethanol (5 × 10⁻⁵m), and 5% FCS, hereafter referred to as complete medium. All media and supplements were purchased from Biological Industries (Beth-Haemek, Israel). Disposable tissue culture flasks and plates were purchased from Greiner Labortechnik (Frickenhausen, Germany).

The IL-1Ra was kindly provided by Dr. Daniel Tracey (Upjohn, Kalamazoo, MI). The M 20 IL-1 inhibitor was purified from the M20 cell line as described (30, 31). Briefly, crude supernatants of serum-free cultures were concentrated by vacuum ultrafiltration using dialysis tubing and purified on a high-performance liquid chromatography-DEAE column. The conditioned medium was also purified by molecular sieving on a Sephacryl S-300 (Pharmacia Biotech Inc., Uppsala, Sweden) column, followed by isoelectric focusing in free solution, using the Rotofor cell (Bio-Rad Laboratories, Richmond, VA). Fractions were collected, dialyzed, and bioassayed. The active fractions corresponding to a M_r ∼52,000 (± 4000) and pI 4.1–4.2 were pooled and stored at −70°C before use. One unit of M 20 is characterized as the amount of high-performance liquid chromatography-DEAE partially purified material that causes 50% inhibition of mouse thymocyte proliferation stimulated by 1 unit of recombinant IL-1β, as described previously (30, 31, 32). LPS from Escherichia coli 055:B5 and mitomycin C were purchased from Sigma Chemical Co. (St. Louis, MO). Fluid thioglycollate medium was purchased from Difco Laboratories (Detroit, MI).

The RO1 and RO2J T lymphoma cell lines were derived from tumors recovered from the thymus of adult mice that were neonatally infected with the murine Moloney leukemia virus. The establishment and characteristics of these cell lines were described previously (33). The chemically induced EL4 cell line was obtained from the American Tissue Culture Collection (Manassas, VA). Cells were incubated at 37°C in an atmosphere of 5% CO₂-95% air.

The optimal conditions for inducing IL-1α activity in T lymphoma cells consist of activation of malignant cells in culture with macrophages (adherent PECs) and LPS. PECs were harvested on day 3 following thioglycollate injection, thoroughly washed, suspended in complete medium, and cultured (at 10⁴ cells/ml) for 2 h to allow cell adherence to the tissue culture Petri dishes. Thereafter, nonadherent cells were removed by vigorous washings and T lymphoma cells (at 10⁶ cells/ml) and LPS (1 μg/ml) were added to the adherent cells. At indicated intervals, nonadherent cells were collected and separated from adherent cells by two cycles of adherence on tissue culture Petri dishes at 37°C (2 h each), followed by collection of nonadherent T lymphoma cells.

Cytokine activity was assessed in supernatants and cell lysates. Supernatants from activated T lymphoma cells were obtained following reculture of nonadherent T lymphoma cells. Thus, T lymphoma cells were recultured (at 10⁶ cells/ml) in complete medium for various intervals; the medium was collected and centrifuged; and supernatants were collected, filtered through 0.45-μm syringe filters (Corning Glass Works, Corning, NY), and kept at −20°C before assay. For the preparation of lysates, cells were collected, thoroughly washed three times in PBS, counted, and resuspended (at 10⁶ cells/ml) and thereafter lysed by three freeze-thaw cycles at −70°C. Lysates were cleared from cell debris by centrifugation, and supernatants were collected, filtered as described, and kept at −20°C before assay.

Murine IL-1α ELISAs were performed using commercial pairs of antibodies and recombinant cytokine (Genzyme Immunodiagnostics, Cambridge, MA). The following antibodies were used: a primary anti-IL-1α capture monoclonal antibody (1 μg/ml; code 1837-01), which recognizes the IL-1α precursor and secreted and membrane-associated forms of natural mouse IL-1α; biotinylated anti-IL-1α polyclonal antibodies (1:1000 dilution; code 1P-110); and for detection, peroxidase-conjugated affinity pure donkey antirabbit IgG polyclonal antibodies and their substrate, o-phenylenediamine dihydrochloride (Sigma). The ELISA procedure was performed according to the manufacturer’s instructions. The amount of IL-1α in samples was extrapolated from a standard curve consisting of recombinant IL-1α, ranging from 15 to 405 pg/ml.

BALB/c mice were injected i.v., into the tail vein, with violent or activated RO1 cells (10⁶ cells/mouse), and tumorigenicity was scored by development of hind legs paralysis and mortality rate. Immunotherapeutic treatment consisted of single i.v. injections of activated cells (10⁶ cells/mouse) at different intervals after the inoculation of violent RO1 cells. Vaccinating cells were treated with mitomycin C; 5 × 10⁶ cells/ml were exposed to the drug (100 μg/ml) for 1 h at 37°C and then thoroughly washed four times before use.

Mice were treated with the IL-1Ra or M 20 IL-1 inhibitor before and after the inoculation of activated T lymphoma cells. The protocols of treatment with the IL-1Ra or the M 20 IL-1 inhibitor consisted of multiple inoculations of the inhibitors, and their exact schedules are mentioned in “Results.”

Violent and activated T lymphoma cells were plated in 96-well plates at a concentration of 5 × 10⁴ cells/ml (0.1 ml/well) in medium containing 1% FCS for different intervals, and thereafter, proliferation was measured by the colorimetric MTT assay (34).

Formalin-fixed and paraffin-embedded samples, from the vertebral region of mice injected with violent and activated RO1 cells, were cut into 4-μm-thick sections and stained with H&E.

Each experiment was repeated at least three to five times, unless otherwise indicated, with similar patterns of results. In vivo experiments consisted of groups of 6–10 mice. The results shown are from individual experiments. ELISAs were performed in triplicates, and the results represent the means. Triplicate values did not differ from the mean by >20%.

None of the T lymphoma cell lines screened expressed constitutive activity of IL-1α. Direct activation of T lymphoma cells with T-cell mitogens, such as phytohemagglutinin, concanavalin A, and phorbol 12-myristate 13-acetate, also did not result in the expression of IL-1α (results not shown).

Significant levels of IL-1α were induced following activation of the malignant T cells in culture with macrophages and LPS. Optimal levels of IL-1α were observed in mixed cultures of T lymphoma cells (10⁶ cells/ml) and adherent PECs (10⁴ cells/ml) that were cultured for 24 h in the presence of LPS (1 μg/ml). IL-1α activity was assessed in nonadherent lymphoma cells that were separated from macrophages and recultured in plain medium for an additional 24 h, as described in the “Materials and Methods.” In Fig. 1, the patterns of IL-1α induction are shown in the RO1 T lymphoma cell line. In activated RO1 cells, IL-1α is expressed at similar levels in cell lysates (Fig. 1,A) and supernatants (Fig. 1,B). Untreated T lymphoma cells and lymphoma cells stimulated with either LPS or macrophages alone generated only low background levels of IL-1α. Fig. 1 A also shows IL-1α levels in macrophage controls, 10⁴ cells/ml fresh LPS-stimulated macrophages, and the adherent fraction obtained after separation of nonadherent RO1 cells in the activation cultures. Thus, macrophages, at the same concentration as that added to T cells, generate only small amounts of IL-1α (∼30–50 pg/ml). From these results, it can be concluded that the IL-1α activity detected in our experiments is of T cell lymphoma origin. The presence of contaminating macrophages in the nonadherent fraction of activated cells was excluded by negative staining with anti-Mac-1 antibodies (results not shown).

Induction of IL-1α expression could also be demonstrated in other malignant T cell lines using the same activation protocol. Table 1 demonstrates the IL-1α content in lysates of two additional T cell lymphoma lines, EL4 and RO2j, which were generated and assayed as described above (Fig. 1). In general, activated T lymphoma cells generate IL-1α in amounts ranging from 150 to 500 pg/ml by 10⁶ malignant cells in 24 h.

T lymphoma cells stimulated in culture with macrophages and LPS to express IL-1α in a short-term manner will be hereafter referred to as activated T lymphoma cells. All subsequent experiments described herein were performed with the RO1 T cell lymphoma line.

To use activated T lymphoma cells for in vivo immunotherapeutic purposes, we found that it was, thus, essential to assess whether IL-1α activity in malignant T cells persists for some time after removal of the activators from the culture. Thus, RO1 cells were stimulated in culture with macrophages and LPS for 24 h, as indicated above. Then nonadherent activated RO1 cells were separated from macrophages, extensively washed to remove residual activators, and recultured in plain tissue-culture medium (without the activators). At various intervals, supernatants were collected and lysates from samples of 10⁶ cells/ml were prepared and assayed for IL-1α content. As can be seen in Fig. 2,A, IL-1α expression in activated RO1 cells persists for a few days in supernatants and lysates; 75% and 50% of the original levels of IL-1α in activated T lymphoma cells were retained in cell lysates after 48 and 96 h, respectively. In supernatants, high levels of IL-1α are detected for longer periods, up to 120 h following activation. The kinetics of IL-1α expression in supernatants and cell lysates of activated T cells, compared to cell viability, indicate that, at early intervals (∼ up to 48 h), IL-1α levels in supernatants reflect active secretion, whereas at later intervals they may also represent IL-1α released from disintegrating tumor cells. Mitomycin C-treated activated RO1 cells retain their ability to express IL-1α expression for a few days, similarly to untreated activated cells (Fig. 2 B). This is of importance because only inactivated cells can be used in tumor vaccines.

The in vitro proliferation patterns of activated RO1 cells were compared to those of violent cells. RO1 were activated in culture with macrophages and LPS, according to the protocol demonstrated in Fig. 1, thoroughly washed, and thereafter recultured in plain medium, and their proliferative capacity was assessed by the MTT colorimetric assay. The proliferation of RO1 cells (5 × 10⁴ cells/ml) was assessed in medium containing 1% FCS, to assess whether IL-1α expression provides some “proliferative advantage” to cells cultured at limiting conditions. As shown in Fig. 3, violent and activated RO1 cells manifested similar proliferative rates.

The effects of short-term IL-1α expression on tumorigenicity of the malignant cells were studied in the RO1 experimental tumor model. RO1 cells manifest exceptionally high patterns of malignancy upon i.v. inoculation. The tumor cells disseminate into the vertebral column, resulting in compression of the spinal cord and in paralysis of the hind legs, followed by death of 100% of the injected mice (within 4–7 weeks). Lymphoid organs and the liver are also affected. The invasion of violent lymphoma cells into the vertebral column is demonstrated in Fig. 4. Sections from the vertebral column of mice injected with violent cells are shown in Fig. 4, A and B, 30 days after tumor cell inoculation and illustrate invasiveness of lymphoma cells to the spinal cord (Fig. 4,A) and peripheral nerves (Fig. 4,B). The cervical region of mice injected with activated T lymphoma cells, 60 days after injection of the malignant cells, demonstrated normal histological features (Fig. 4 C).

The tumorigenicity of activated RO1 cells injected i.v. is depicted in Fig. 5. Survival rates of mice injected with activated and violent RO1 cells in a single experiment are shown. One hundred % of the mice injected with violent cells died within 25 days, whereas 75% of the mice injected with activated cells remained alive after 3 months. Similar patterns of tumorigenicity were observed in 10 additional experiments with groups of 6–10 mice. In each experiment, 50–100% of the mice injected with activated RO1 cells survived 3–6 months, until the termination of the experiment, whereas each mouse that was injected with cells not expressing IL-1α died.

Similar retardation of tumor growth and increase in survival rate were observed with activated cells of other T lymphoma cell lines (results not shown).

Recombinant IL-1α at the range of concentrations expressed by activated RO1 cells and even at much higher concentrations (in the range of nanograms), when applied as a single treatment or in multiple injections did not impair the tumorigenicity patterns of violent RO1 cells (results not shown). This points to the immunotherapeutic significance of tumor cell-associated IL-1α in activated RO1 cells.

We next assessed whether IL-1, expressed in a short-term manner in activated RO1 cells, is the moiety responsible for reduced tumorigenicity. IL-1 activity was reduced by using either the IL-1Ra or the M 20 IL-1 inhibitor. The IL-1Ra was injected on days −5, −3, and −1; before tumor cell inoculation; on the day of tumor cell inoculation; and subsequently, on days 1, 4, 10, and 19 (i.p. injections; 125 μg/0.1 ml per injection). The protocol of treatment with the M 20 IL-1 inhibitor consisted of one i.v. injection, one day after tumor cell inoculation, followed by biweekly i.p. inoculations of 2 units of M 20 per mouse. As seen in Fig. 6, in the group of mice injected with activated cells alone, a 75% rate of survival was recorded after 105 days (termination of the experiment), whereas 100% of the mice injected with untreated (violent) RO1 cells died within 25 days. Treatment of mice with M 20 IL-1-inhibitor completely abrogated the reduced tumorigenicity patterns of activated RO1 cells and 100% of the mice died within 30 days, whereas only 25% of the mice injected with activated RO1 cells together with the IL-1Ra survived after 105 days (Fig. 6). The use of IL-1-specific inhibitors did not alter the tumorigenicity patterns of violent cells, possibly ruling out the involvement of nonspecific metabolic effects induced by the inhibitors in the malignant cells (results not shown). It is noteworthy that treatments with the IL-1Ra were performed only until day 19 after the inoculation of the malignant cells, whereas the M 20 IL-1 inhibitor was applied throughout the experiment. This may explain why the M 20 IL-1 inhibitor exerted more pronounced effects on tumorigenicity. Our results on the degree of in vivo neutralization of IL-1-mediated effects with the IL-1Ra or the M 20 inhibitor are comparable to those obtained with these agents in other biological systems (13, 30, 31, 32). These results suggest a dominant role of IL-1 in determining reduced tumorigenicity of activated RO1 cells.

The ability of activated T lymphoma cells to intervene in the growth of violent RO1 cells after tumor establishment was assessed. Mice were inoculated i.v. with violent RO1 cells, and at different intervals thereafter, activated cells (treated with mitomycin C) were inoculated i.v. All mice injected with violent RO1 cells died within 60 days (Fig. 7,A). As can be seen in Fig. 7,B, inoculation of activated cells at critical intervals (day 6 or 9) improved the survival rate of the mice. Thus, in this experiment, all mice injected only with violent cells died within 30 days, whereas 30% and 66% of the mice treated with activated cells on days 6 and 9, respectively, survived for 3 months (termination of the experiment). In contrast, mice treated on day 14 did not manifest reduced tumorigenicity patterns; they displayed the patterns of violent cells (Fig. 7 B). When using IL-1α-expressing cells in various experimental tumor systems, we always characterize a specific “therapeutic window” in which treatment is efficient (usually between days 5 and 12). However, the peak day for achieving maximal immunotherapeutic effects may slightly vary among experiments.

We have shown here that short-term expression of IL-1α by activated malignant T cells endows reduced tumorigenicity patterns. In addition, we have demonstrated that tumor cells expressing IL-1α can serve as a treatment when they are used to intervene in the growth of violent metastatic T-cell lymphomas. IL-1α activity was optimally induced by activating malignant T cells in culture with macrophages in the presence of LPS, in agreement to other reports on induction of IL-1α in normal and malignant T cells following stimulation with mitogens or antigens in the context of antigen-presenting cells (35, 36, 37). IL-1α generated by T lymphoma cells was detected in cell lysates, on the membrane of activated T lymphoma cells as well as in its secreted form. The mechanism of induction of short-term activity of IL-1α in T lymphoma cells is still under investigation. Activated T lymphoma cells manifest similar proliferative patterns to those of the violent cells, even in “limiting conditions,” i.e., in low concentrations of serum in which IL-1α or cytokines induced by IL-1α may potentiate the growth of the cells, thus excluding a possible autocrine role of endogenous IL-1α in the proliferation of RO1 malignant T cells.

The antitumor effects observed in this study possibly stem from IL-1α, generated by activated T lymphoma cells, because systemic neutralization by IL-1-specific inhibitors reversed the tumorigenicity patterns of activated T lymphoma cells. The M 20 IL-1 inhibitor was shown to specifically abrogate IL-1-dependent effects in various in vitro and in vivo biological systems, i.e., T-cell proliferation, inflammation in rheumatoid arthritis, and the proliferation of chronic myelogenous leukemia, in an as yet unknown mechanism (30, 31, 32), whereas the IL-1Ra binds to IL-1 receptors, without transmitting activation signals, and thus serves as a blocking agent that protects cells from external IL-1 (reviewed in Ref. 13).

Tumor cell-derived IL-1α possibly serves as an adjuvant to increase the immunogenicity of malignant T cells and to potentiate specific antitumor immune responses. Cell-mediated immune responses, i.e., development of cytotoxic cells and the secretion of T_H1 cytokines, are involved in reduced tumorigenicity patterns of activated RO1 cells (results not shown). In addition, an immune memory is established following the rejection of activated cells expressing IL-1α. Thus, 75% of the mice preimmunized with activated RO1 cells (treated with mitomycin C) are protected from a challenge with violent cells, whereas no protection was observed with violent cells (results not shown). Similarly, a significant proportion of mice that had survived following injection with activated RO1 cells develop an immune memory, which protects them from a challenge with violent RO1 cells. The nature of immune mechanisms that control reduced tumorigenicity patterns of activated RO1 cells is currently under investigation and will be described elsewhere.

The effects of IL-1α may be direct, due to its pleiotropic antitumor effects, or indirect, through its ability to induce the generation of a cytokine cascade by the malignant cells in their microenvironment or in the draining lymph nodes. It is noteworthy that, in supernatants and cell lysates of activated T lymphoma cells, in which IL-1α is abundant, we could not detect “classical” T cell cytokines, such as IL-2, IL-4, IL-10, or IFN-γ. Thus, we assume that tumor cell-derived IL-1α possibly initiates antitumor T cell-mediated immunity in the host. As indicated, IL-1 is a unique cytokine with significant potential as an antitumor immunotherapeutic agent (10, 11, 12, 13, 14, 15). The unique subcellular compartmentalization of IL-1α in tumor cells is important for exerting its efficient antitumor effects and differs from IL-1β and most cytokines, which are active only in their secreted form. Thus, the membrane-associated form of IL-1α (M_r 23,000) may serve as an adhesion molecule, allowing efficient cell-to-cell interactions between the malignant and immune effector cells, and as a focused cytokine with strong adjuvant activity. In addition, the active cytosolic form of IL-1α (M_r 31,000) may be released from disintegrating tumor cells and may be present in the tumor’s microenvironment during tumor eradication by the immune system. The expression of active IL-1α in multiple compartments may allow efficient activation of the immune system at levels below those that are toxic to the host. Indeed, relatively moderate amounts of IL-1α generated by activated malignant T cells, i.e., ∼150–500 pg per 10⁶ cells over 24 h, significantly reduces their tumorigenicity, as we showed previously in fibrosarcoma experimental tumors (1, 10, 25, 26, 27, 28, 29). Short-term expression of IL-1α persists in culture (Fig. 2) and possibly also in vivo for a few days; this is sufficient for stimulation of antitumor immune responses, as was also shown in our studies on fibrosarcoma experimental tumor systems, in which malignant cells expressing IL-1α in a short-term manner regress due to the development of T cell-mediated immune responses against the malignant cells (1, 10, 25, 26, 27, 28, 29).

We showed previously that IL-1α-positive fibrosarcoma cells can serve as a treatment by intervening in the growth of existing violent tumors. Successful immunotherapeutical intervention in the growth of violent tumors, with IL-1α-positive fibrosarcoma cells, occurs at certain critical intervals after tumor cell inoculation, usually 6–12 days after the injection of the malignant cells. At that critical time, an inefficient antitumor immune response develops, which is potentiated by IL-1α, resulting in tumor regression. At later intervals, the tumor’s mass is possibly too large for intervention. A similar immunotherapeutic window was characterized here for the treatment of mice previously inoculated with violent RO1 cells (Fig. 7). The immunotherapeutic manipulation described here is minimal (a single treatment) and its efficiency (66% rate of survival for treatment on day 9) can possibly be improved by multiple vaccinations or in combination with other debulking treatments. The effects of tumor cell-associated IL-1α, that are described here, are remarkable, especially considering that T lymphomas represent disseminated metastatic tumors. The antimetastatic effectiveness of short-term expression of IL-1α by T lymphoma cells, is comparable and, in some cases, even better than effects observed with constitutive expression of IL-1α or other cytokines by tumor cells (gene transfer; Refs. 1, 2, 3, 4, 5, 6, 7, 8, 9 and 25, 26, 27, 28, 29). Short-term expression of cytokines in tumor cell vaccines may represent a novel immunotherapeutic modality of treatment, which may supplement the current cytokine gene transfer approaches and may facilitate the better use of cell-associated IL-1α in tumor cell vaccines.