Impaired Infiltration of Tumor-specific Cytolytic T Cells in the Absence of Interferon-γ despite Their Normal Maturation in Lymphoid Organs during CD137 Monoclonal Antibody Therapy¹

CD137 (4-1BB) is a member of the tumor necrosis factor receptor superfamily expressed by activated T cells, NK³ cells, monocytes, and dendritic cells (1, 2). The ability of both agonistic anti-CD137 antibodies and CD137L to stimulate T-cell proliferation and cytokine secretion, including IFN-γ, has been demonstrated in previous studies (3, 4, 5, 6); and experiments performed in both CD137 and CD137L-deficient mice have additionally demonstrated the importance of CD137 costimulation in the generation of a fully competent T-cell response (7, 8, 9, 10). Therefore, the CD137/CD137L interaction represents an attractive target for immunotherapy, as demonstrated in various tumor models (11, 12, 13). For example, we have shown that the administration of agonistic anti-CD137 mAb in tumor-bearing mice stimulates tumor-specific CTL, leading to the eradication of established tumors in several tumor models (14, 15). The goal of the present study is to characterize the effector mechanism involved in the potent antitumor CTL response generated after CD137 stimulation. During the effector phase of the CTL response, tumor-specific CTL may kill tumor cells in a contact-dependent fashion that is likely mediated by perforin or FasL. Alternatively, CTL may prevent tumor growth via the release of soluble mediators.

IFN-γ, a pleiotropic cytokine secreted by activated T cells and NK cells, plays a central role in both the innate and adaptive immune response to a variety of pathogens and transformed cells (16, 17). Whereas CD4⁺ and CD8⁺ T cells have been shown to mediate tumor rejection in an IFN-γ-dependent fashion, data have been obtained to support a number of potential mechanisms. On binding a tumor target, IFN-γ may up-regulate the expression of molecules critical for antigen processing and presentation (17). IFN-γ may also be required for the differentiation of fully competent effector cells (18). In contrast, IFN-γ may stimulate the secretion of angiostatic chemokines by stromal cells present within a tumor (19). Additionally, recent studies have implicated IFN-γ in regulating T-cell migration to the tumor site (20). Therefore, IFN-γ may mediate tumor regression through a variety of mechanisms on binding its receptor that can be expressed by tumor cells, hematopoietic cells, or nonhematopoietic stromal cells.

In the present study, we attempt to identify the effector mechanism used by CTL after treatment of an established tumor with a tumor-specific human papillomavirus-16 E7 peptide and CD137 mAb. Tumors failed to regress in IFN-γ-depleted mice. The results presented suggest that IFN-γ plays a central role in the accumulation of specific T cells to the site of tumor growth after anti-CD137 administration.

Female C57BL/6 (B6) mice were purchased from the National Cancer Institute (Frederick, MD). Female IFN-γ-deficient mice (C57BL/6) were purchased from the Jackson Laboratory (Bar Harbor, ME).

C3 cells, generated from human papillomavirus-16-/EJras-transformed BL/6 mouse embryo cells (21), were a gift from Dr. W. Martin Kast (Loyola University, Chicago, IL). The EL4 T-cell lymphoma was purchased from the American Type Culture Collection (Rockville, MD). All of the cell lines were maintained in a complete medium of RPMI 1640 (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (HyClone, Logan, UT), 25 mm HEPES, 2 mm glutamine, 100 units/ml penicillin G, and 100 μg/ml streptomycin sulfate. For in vivo experiments, mice were inoculated s.c. in the flank with 1 × 10⁶ C3 cells in 100 μl HBSS (Mediatech Cellgro) or Matrigel (BD Biosciences, Bedford, MA). Isolation of TILs from C3 tumors grown in Matrigel was performed as described previously (22).

The mock and mγRΔ1C constructs described previously (23) were a generous gift from Dr. Robert Schreiber (Washington University, St. Louis, MO). C3 transfectants were generated according to the manufacturer’s instructions using FuGENE 6 Transfection Reagent (Roche Molecular Biochemicals, Indianapolis, IN). Transfected cells were grown in complete RPMI 1640 supplemented with 10 mm l-histidinol (Sigma, Gilbertsville, PA) and were screened based on their inability to up-regulate MHC class I expression in response to recombinant mouse IFN-γ (R & D Systems, Minneapolis, MN).

Total RNA was isolated from excised tumors after tumor homogenization in Tri-Reagent (Sigma, St. Louis, MO). First strand cDNA synthesis was performed using a (dT)₁₅ primer, and the resulting cDNA was amplified using the following primers: hypoxanthine phosphoribosyltransferase, 5′-CTAATGATCAGTCAACGGGGGAC-3′ and 5′-CCAGCAAGCTTGCAACCTTAACCA-3′; CD8, 5′-AACTGTTTTCTGCCGTGAGG-3′ and 5′-AGGGAGTTCGCAGCACTGG-3′; and IFN-γ, 5′-TTCCTCATGGCTGTTTCTGG-3′ and 5′-AAGACTTCAAAGAGTCTGAGG-3′. The cDNA was subjected to an initial melting step (94°C for 10 min) before 30 cycles of amplification (94°C 30 s, 55°C 30 s, and 72°C 40 s). PCR was performed using Ampli-Taq Gold (Perkin-Elmer, Foster City, CA) according to the manufacturer’s instructions.

A rat IgG2a mAb against CD137, 2A, was generated as described previously (15). The hybridoma 2A was grown in RPMI 1640 supplemented with 10% low IgG fetal bovine serum (Life Technologies, Inc.) and 25 mm HEPES. Supernatant was harvested, concentrated using a tangential flow miniplate concentrator (Millipore, Bedford, MA), and purified using a 5-ml HiTrap protein G-Sepharose column (Amersham Pharmacia Biotech, Piscataway, NJ). Purified mAbs were dialyzed against PBS and concentrated using a Centriprep concentrator (Millipore). The Rat IgG control antibody was purchased from Sigma. The neutralizing anti-IFN-γ mAb (R4–6A2) was purified using a similar method. FITC-conjugated antimouse CD8 was purchased from PharMingen (San Diego, CA).

The H-2D^b-E7 and H-2D^b-Vp2 tetramers were prepared as described previously (24). Briefly, H-2D^b and human β₂-microglobulin were isolated from a bacterial expression system and subsequently folded in excess of peptide. The folded monomeric complexes were desalted and biotinylated. After cation exchange purification, the monomeric complexes were conjugated with R-phycoerythrin-streptavidin, forming a tetrameric complex. The H-2D^b tetramers generated were purified by size exclusion gel filtration.

B6 mice were given 1 × 10⁶ C3 cells resuspended in Matrigel (Becton Dickinson, Bedford, MA) according to the manufacturer’s instructions. No difference in tumor growth kinetics was observed between C3 tumors that had been injected in HBSS compared with those injected in Matrigel (data not shown). After harvesting the tumors from euthanized mice, the tumor was disintegrated with a forceps and wire mesh. The suspension was filtered through a nylon mesh, and the collected cells were washed. Viable cells were separated from dead tumor cells and cell debris by centrifugation on a lympholyte M (Cedarlane Laboratories Limited, Hornsby, Ontario, Canada) gradient according to the manufacturer’s instructions. Viable cells were then used for tetramer staining. The fold increase in total CD8⁺ T cells (as determined by anti-CD8 mAb) after treatment was multiplied with the fold increase of E7-specific T cells (as determined by tetramer analysis) to yield the net increase in E7-specific CD8⁺ T cells in total CD8⁺ CTL.

Tumor DLNs from immunized mice were harvested on day 7 and stained with phycoerythrin-conjugated H-2D^b/E7 or H-2D^b-Vp2 tetrameric complexes and FITC-conjugated CD8 as described previously. TILs isolated from C3 tumors were stained in a similar fashion. CTL activity was measured using a standard 4-h ⁵¹Cr release assay with tumor cell targets at the indicated E:T ratios. Effector cells were generated by incubating 5 × 10⁶ LN cells from E7-immunized mice with 2.5 × 10⁵ UV-irradiated C3 cells for 4 days.

To determine the role of IFN-γ in the regression of established C3 tumors in mice after CD137 mAb treatment, we first sought to detect the expression of IFN-γ and CD8 transcripts by RT-PCR in freshly isolated C3 tumors after treatment with anti-CD137 mAb or a combination of E7 peptide and anti-CD137 mAb. Neither CD8 nor IFN-γ could be detected in mice treated with a control peptide and either a control rat IgG or anti-CD137 (Fig. 1,A, Lanes 1 and 2). This may not be surprising, as we have shown that C3 tumors are immunologically ignored and CD137 mAb alone does not stimulate T-cell activation (15). In contrast, both CD8 and IFN-γ were detected in tumors isolated from mice that had been immunized previously with the tumor-specific E7 peptide with or without administration of CD137 mAb (Fig. 1 A, Lanes 3 and 4). Although RT-PCR assay is not quantitative, our results suggest that immunization with the E7 peptide results in the infiltration of CD8⁺ T cells and IFN-γ secretion at the site of tumor growth.

We next sought to determine whether IFN-γ was required for tumor eradication after the E7 peptide and CD137 mAb treatment. Consequently, both wt and IFN-γ-deficient B6 mice were inoculated with C3 cells and were treated with both the E7 peptide and CD137 mAb 7 days later as described previously (15). Whereas tumors in the wt B6 mice regressed after treatment, all of the IFN-γ-deficient mice developed progressively growing tumors (Fig. 1 B, top panel). Similarly, tumor regression was only observed in 20% of the treated wt mice that received a neutralizing anti-IFN-γ mAb. In contrast, tumors regressed in all of the treated mice that received a control mAb. Therefore, tumor eradication after E7 peptide and CD137 mAb treatment is dependent on IFN-γ.

We next sought to determine the mechanism by which IFN-γ promotes tumor regression after anti-CD137 mAb treatment. The IFN-γ receptor is ubiquitously expressed on most cells, including tumor cells. Previous studies have shown that responsiveness of a tumor to IFN-γ may be required for effective tumor immunity (16). To test this possibility, C3 cells were transfected with a dominant-negative IFN-γ receptor (mγRΔ1C). Unlike the mock transfectant, the mγRΔ1C-transfectant was unresponsive to IFN-γ as demonstrated by its inability to up-regulate the expression of H-2D^b (Fig. 2,A) and H-2K^b (data not shown) after exposure to IFN-γ. Both the mock- and mγRΔ1C-transfectants were lysed equally well by E7-specific CTL in vitro (Fig. 2,B), and both transfectants had similar growth kinetics in wt B6 mice (Fig. 2 C), suggesting no major differences in tumorigenicity and immunogenicity between these cell lines.

To determine whether or not the tumors unresponsive to IFN-γ would regress after CD137 mAb administration, both mock- and mγRΔ1C-transfectants were inoculated into the flanks of B6 mice. Seven days later, the mice were treated with both the E7 peptide and anti-CD137. Whereas 100% of the mock-transfectants regressed after treatment, tumor regression was also observed in 90% of the mice given the mγRΔ1C-transfectant (Fig. 2 D). Similar results were obtained with a second mγRΔ1C-transfectant (data not shown). We conclude that binding of IFN-γ to tumor cells is not required for the therapeutic effect of CD137 mAb.

Previous reports have suggested that IFN-γ may be important for the differentiation of fully competent effector CTL. Therefore, the lack of IFN-γ could affect the generation of tumor-specific CD8⁺ CTL. To test this possibility, we examined both the frequency and cytolytic activity of the E7-specific CD8⁺ CTL in IFN-γ deficient mice. Both wt and IFN-γ-deficient mice were immunized with the E7 peptide and given CD137 mAb. Seven days later, the DLNs were harvested and an H-2D^b-E7 tetramer was used to determine the frequency of E7-specific CTL in the DLN of mice. As shown in Fig. 3 A, ∼4% of the total CD8⁺ cells in the DLN of mice were E7-specific in wt and IFN-γ-deficient mice, indicating no significant difference between these mice.

Cytolytic activity of the E7-specific CTL was also examined. DLN cells from the immunized mice were restimulated in vitro with irradiated C3 cells. Four days later, these cells were used as effector cells in a standard 4-hour ⁵¹Cr release assay against a C3 target. CTL derived from the IFN-γ-deficient mice were as effective as the wt T cells in killing C3 cells (Fig. 3 B). Our results suggest that IFN-γ does not affect the induction of a fully competent CTL response after CD137 mAb administration.

Although the frequency and activity of the E7-specific CTL in lymphoid organs are not impaired in the absence of IFN-γ, it is unknown whether these CTL actually infiltrate the tumor. To address this issue, mice were inoculated with C3 cells suspended in a collagen (Matrigel) matrix. After treatment with the E7 peptide or a combination of the E7 peptide and CD137 mAb, tumor masses were removed and TILs were isolated. TILs were stained with the H-2D^b-E7 tetramer and anti-CD8. In initial experiments, we observed very few E7-specific TILs in both groups of mice 7 days after treatment (data not shown). However, a significant increase in both the total numbers of CD8⁺ T cells and E7-specific T cells was observed in the TILs isolated from the CD137 mAb-treated mice 14 days after treatment (Fig. 4,A). There were no detectable TILs in the untreated mice (data not shown). Therefore, tetramer analysis was performed 12–14 days after treatment in all of the subsequent experiments. In a similarly performed experiment, both wt and IFN-γ-deficient mice were inoculated with C3 cells in Matrigel and subsequently treated with the E7 peptide and CD137 mAb. Whereas ∼10% of the TILs were CD8⁺ in the wt mice given the E7 peptide and control mAb, ∼60% of TILs were CD8⁺ in the mice given the E7 peptide and CD137 mAb (Fig. 4,B). Similarly, an almost 3-fold increase was observed in the frequency of E7-specific T cells in those mice treated with anti-CD137 mAb compared with the mice that received the control antibody (Fig. 4,B). Therefore, an ∼10-fold net increase in the number of E7-specific T cells was observed in the wt mice treated with anti-CD137 mAb. In contrast, the total number and frequency of E7-specific T cells were greatly reduced in TILs isolated from IFN-γ-deficient mice, despite receiving CD137 mAb (Fig. 4 B). Our results demonstrate that IFN-γ is required for the accumulation of tumor-infiltrating T cells at the tumor site after anti-CD137 administration.

The data presented here demonstrate that C3 tumor eradication after administration of the E7 peptide and agonistic CD137 mAb, a method effectively breaking immunological ignorance in tumor-specific CTL (15), is entirely dependent on IFN-γ. This was shown both in mice given an IFN-γ neutralizing mAb and in IFN-γ deficient mice. On closer examination, we found that in the absence of IFN-γ, the total number of tumor-specific CD8⁺ CTLs infiltrating the tumor was significantly reduced, whereas the generation of these CTLs in lymphoid organs was unaffected. Because IFN-γ was not required to bind its receptor expressed on tumor cells, our results reveal a selective role for IFN-γ in promoting the infiltration of tumor-specific CTLs within the tumor.

Previous in vitro and in vivo studies have demonstrated that CD137 costimulates the production of IFN-γ by CD8⁺ T cells (4, 5, 6). CD8⁺ T cells stimulated by anti-CD137 mAb produced significantly higher levels of IFN-γ than those that were stimulated with anti-CD28, although CD28 costimulation enhanced T-cell proliferation and IL-2 secretion (3). The ability of CD8⁺ T cells to secrete IFN-γ on CD137 stimulation was demonstrated for several CD137 mAbs tested, and the CD137 mAb we have generated and used in this study is no exception (data not shown). After CD137 mAb treatment, we were able to detect IFN-γ transcripts in tumor tissue of the mice immunized with a tumor-specific peptide. This supports that IFN-γ, released by tumor-specific CTL, may be an important effector cytokine in vivo. This was subsequently confirmed in experiments using IFN-γ-deficient mice and IFN-γ neutralizing mAbs. Although tumor-specific CTLs are the most likely source of IFN-γ in vivo, other cells, including NK cells, cannot be excluded.

There is ample evidence implicating IFN-γ as a critical effector cytokine in tumor immunity (17). IFN-γ induces the expression of many molecules involved in antigen processing and presentation, thus promoting the presentation of antigenic peptides (17). In addition, IFN-γ activates macrophages (25) and inhibits angiogenesis (19), mechanisms believed to contribute to its antitumor effect. Expression of a dominant-negative IFN-γ receptor α chain in mouse fibrosarcoma cell lines decreased their immunogenicity (23), suggesting a direct interaction between IFN-γ and the tumor. However, in a model of CD4⁺ T cell-mediated tumor immunity, the action of IFN-γ on tumor cells is not required, and host cells, but not tumor cells, are the targets of IFN-γ (19). C3 cells express low levels of MHC class I, but expression could be greatly up-regulated by IFN-γ, leading to its increased antigenicity in cytotoxicity assays performed in vitro (data not shown). These observations support the notion that IFN-γ stimulation of the tumor cell directly after CD137 mAb treatment may be beneficial. We engineered tumor cells unresponsive to IFN-γ by transfecting a dominant-negative IFN-γ receptor α chain. To our surprise, the transfected C3 tumor cells regressed after treatment with CD137 mAb, a result similar to that of the mock-transfected C3 tumor. Importantly, these cells were similar to the mock-transfected cells in both the in vitro and in vivo tests performed thus far. Furthermore, expression of the dominant-negative IFN-γ receptor was retained in vivo, as cells isolated from tumor-bearing mice 2 weeks after tumor inoculation remained unresponsive to IFN-γ (data not shown). Our results support the notion that an indirect role of IFN-γ on host cells rather than tumor cells is required for the therapeutic effect of CD137 mAb.

To additionally explore this possibility, we examined the importance of IFN-γ in two processes critical for effective tumor immunity: the generation of tumor-specific CD8⁺ effector CTLs in lymphoid organs and infiltration of these cells into the tumor site. It has been shown that IFN-γ is involved in the differentiation and maturation of fully competent CTLs (18). CTLs isolated from IFN-unresponsive, STAT-1-deficient mice lacked cytolytic activity and failed to up-regulate serine esterase, a component of the cytolytic granule. Thus, it is possible that increased release of IFN-γ after CD137 mAb treatment promotes functional maturation of tumor-specific CTLs. In sharp contrast, we have found that the frequency and cytolytic activity of E7-specific CTLs were similar in both the wt and IFN-γ-deficient mice in DLNs. In fact, in some experiments, the cytolytic activity of the IFN-γ-deficient CTL was slightly higher than that observed for the wt CTL (data not shown), an observation that may be explained by the antiproliferative effects of IFN-γ in vitro. Our results, obtained in a tumor model, are consistent with the findings that IFN-γ-deficient mice developed a fully competent CTL response after lymphocytic choriomeningitis virus infection (26). Thus, our results exclude the effect of IFN-γ in the generation of effector CTL in lymphoid organs after anti-CD137 treatment. Using a collagen matrix, we enumerated tumor-infiltrating T cells after CD137 mAb therapy. We demonstrate that the frequency of total and E7-specific CD8⁺ CTL within the C3 tumor was significantly reduced in the IFN-γ deficient mice 2 weeks after treatment. On the basis of our calculation, an ∼10-fold reduction in the total number of E7-specific T cells in IFN-γ-deficient mice was observed compared with wt mice. Our results indicate that the accumulation of specific CTL within the tumor site is, at least in part, an IFN-γ-dependent process. IFN-γ has been shown to induce the secretion of chemokines including IP-10, Mig, and I-TAC, which play important roles in regulating the migration of tumor-specific T cells to the tumor site (17, 27, 28). The importance of adhesion molecules like LFA-1 and VLA-4 in T-cell migration is also well documented (29). Chemokines secreted at the site of tumor growth may not only chemoattract activated T cells, but they may also enhance the affinity of these integrins for their ligands, thus permitting T-cell extravasation (30, 31). Alternatively, IFN-γ may up-regulate the expression of costimulatory molecules or other cytokines by cells, including macrophages or dendritic cells, within the tumor site, which indirectly promote the expansion or survival of tumor-specific CTL (17, 32).

Our findings suggest that even though a vigorous CTL response is generated in secondary lymphoid organs, the ability of these T cells to accumulate at the site of tumor growth is crucial, and this process can be modulated by IFN-γ. It is tempting to speculate that an increased secretion of IFN-γ within the tumor site could facilitate the accumulation of antitumor CTL. In contrast, the down-regulation of IFN-γ or the lack of responsiveness on T cells by factors from either host or tumor-derived mechanisms may selectively inhibit tumor infiltration of CTL. Our observations unveil an unexpected in vivo function of IFN-γ in the generation of antitumor immunity.

We thank Dr. Robert Schreiber (Washington University, St. Louis, MO) for the constructs expressing dominant-negative IFN-γ receptor; Dr. W. M. Kast (Loyola University, Chicago, Maywood, IL) for C3 tumor line; and Julie S. Lau and Kathy Jensen (Mayo Clinic, Rochester, MN) for editing the manuscript.