Targeting TNF alpha as a novel strategy to enhance CD8⁺ T cell-dependent immune response in melanoma?

ABSTRACT

Tumor Necrosis Factor α (TNF) is a pleiotropic cytokine exhibiting a dual activity in oncoimmunology, either acting as a cytotoxic effector produced by leukocytes or behaving as an immunosuppressive molecule. We have just discovered that TNF signaling impairs the accumulation of tumor-infiltrating CD8⁺ T lymphocytes in experimental melanoma.

TNF has been initially shown to be part of the cytotoxic effectors of the immune cells such as natural killers and CD8+ T lymphocytes. More recently, TNF has emerged as a putative immune suppressor cytokine, facilitating the accumulation and/or biological activity of regulatory T lymphocytes (Tregs)¹ as well as myeloid-derived suppressor cells (MDSC).² In addition, TNF triggered melanoma dedifferentiation, promoting melanoma relapse in an adoptive CD8+ T cell transfer protocol.³

We have recently demonstrated that TNF signaling facilitates the tumor growth of mouse melanoma exhibiting a high-expression level of major histocompatibility class I (MHC-I^high), while having no or minimal effects toward mouse melanoma expressing MHC-I at low levels (MHC-I^low). As a matter of fact, we provided evidence that the tumor growth of MHC-I^high, but not that of MHC-I^low melanoma, was impaired in TNF-deficient mice. Moreover, the administration of Etanercept, a soluble form of human TNF-R2, which efficiently neutralizes mouse TNF, significantly reduced the MHC-I^high melanoma growth.⁴

In terms of molecular mechanisms, we showed that TNF receptor 1 (TNF-R1), rather than TNF-R2, was required to transduce the pro-tumorigenic effect of TNF in melanoma. Indeed, the MHC-I^high melanoma growth was impaired in mice lacking TNF-R1, but not TNF-R2. The TNF-R1-dependent TNF signaling inhibited the accumulation of tumor-infiltrating CD8+ T lymphocytes (CD8+ TILs), including specific CD8+ T cells. The latter phenomenon likely accounted for the pro-tumorigenic role of TNF since immunodepletion of CD8+ T cells fully restored the MHC-I^high melanoma growth in TNF-deficient mice. Conversely, the antitumor effect of Etanercept was observed only in immunocompetent mice, TNF blockade having no effect in mice lacking IFNγ or CD8 as well as in nude mice.⁴

We demonstrated that TNF exerts a direct negative effect toward CD8+ T cells by inducing cell death of activated CD8+ T lymphocytes (Fig. 1). Although this finding is in agreement with a previous study,⁵ we provided genetic evidence that TNF-R1, rather than TNF-R2, is the key receptor that triggered activated CD8+ T apoptosis in response to TNF. To definitely demonstrate that TNF-R1 signaling directly impaired CD8+ TIL homeostasis, we performed an adoptive transfer protocol by injecting activated TNF-R1-deficient or proficient CD8+ T cells in melanoma tumors developed in CD8⁺-deficient mice. Three days later, the proportion of live CD8+ TILs was enhanced by TNF-R1 deficiency.⁴

Figure 1.

TNF impairs CD8⁺ TIL accumulation in melanoma. TNF-R1-dependent TNF signaling (i) limits the intra-tumor content of HEV, which are involved in lymphocyte extravasation, and (ii) induces activated CD8⁺ T cell death, thus facilitating melanoma escape from the immune system.

Because the pro-tumorigenic role of TNF was unlikely restricted to a direct effect on CD8+ T cells, we also analyzed the tumor microenvironment in TNF-deficient mice, focusing on tumor vasculature and immunosuppressor cells. Under our experimental conditions and in sharp contrast with recent studies highlighting the critical function of TNF-R2 in the accumulation of immunosuppressor cells,^1,2 TNF deficiency or blockade was not associated with a reduction of the tumor content of both Tregs and MDSC. Moreover, MDSC immunodepletion by injecting an anti-Gr1 antibody did not result in an increased CD8+ TIL content in wild-type and TNF-deficient mice (Bertrand, Colacios and Ségui, unpublished data). Thus, the effect of TNF on the limitation of CD8+ T cell-dependent immune response was unlikely a consequence of the alteration of MDSC homeostasis and biological activity, at least under our experimental conditions.

The tumor vasculature was not drastically altered by TNF deficiency as evaluated by the quantification of CD31 staining (Bertrand and Ségui, unpublished data). However, the abundance of high endothelial venules (HEV), which are critical blood vessels for lymphocyte extravasation into lymph nodes and tumors,^6,7 was increased in melanoma tumors from mice lacking TNF or TNF-R1. Of note, these vessels were surrounded by a large number of CD8+ TILs. Similar data were obtained in wild-type mice injected with Etanercept. Although it remains unclear how TNF signaling impairs HEV in melanoma, their increased frequency likely contributed to the accumulation of CD8+ TILs upon TNF blockade (Fig. 1).

Anti-TNF antibodies and Etanercept have been tested in patients affected with diverse advanced cancers, with no or limited clinical responses. To the best of our knowledge, only one melanoma patient was reported as a good responder to anti-TNF for several weeks.⁸ Our preclinical findings may be of great interest in the clinic since we have identified high MHC-I expression by melanoma cells as a critical point to condition the antitumor effects triggered by the TNF blockade strategy. In addition, we document that TNF blockade was efficient only in immunocompetent animals. MHC-I expression by melanoma may serve as a biomarker, and its assessment in tumor biopsies may help select patients for anti-TNF therapies in melanoma and possibly in other cancer types. In addition, a pre-existing infiltration of CD8⁺ TILs in tumor biopsies, that may reflect the immunogenicity of the tumor cells, is most likely a prerequisite to observe a clinical benefit of TNF blockade strategy. Clinical trials are obviously necessary to evaluate this tenet and to determine eligibility criteria for such a treatment. Because monotherapy based on anti-TNF may have limited effects, a combination of TNF blockade with emerging therapies (such as anti-CTLA4 or BRAFV600E inhibitors), which are associated with TNF production and tumor lymphocyte infiltration,^9,10 may lead to a positive clinical outcome by enhancing the accumulation of CD8+ TILs in melanoma.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by RITC (Recherche et Innovation Thérapeutique en Cancérologie), Cancéropôle Grand Sud-Ouest, INSERM and Paul Sabatier University (Toulouse III).