Melanoma targeting with the loco-regional chemotherapeutic, Melphalan: From cell death to immunotherapeutic efficacy

Abstract

All immunoregulatory chemotherapeutics are chiefly applied in a systemic setting for anticancer therapy. However, immune responses following loco-regional application of chemotherapy may differ from those after systemic application. We recently found that Melphalan, a prototypical loco-regionally applied chemotherapeutic agent, exhibits the ability to increase the immunogenicity of dying melanoma cells.

Subversion of immunosurveillance by cancer cells contributes to their ability to grow and metastasize.¹ Despite remarkable therapeutic efficacy of recently approved cancer immunotherapies, complete responses are observed in only ∼30% of cancer patients. Therefore, identification of additional strategies capable of eradicating bulk tumors while concomitantly stimulating host's immune response against residual cancer cells may hold the highest therapeutic value. However, a potential roadblock is that at the doses required to kill the cancer cells, most chemotherapeutics tend to induce poorly immunogenic cancer cell death. Recent evidence, however, indicates that selected types of anticancer treatments (e.g. radiotherapy, anthracyclines, hypericin-based Photodynamic Therapy/Hyp-PDT) elicit an apoptotic cell death sub-routine with a pronounced immunogenic character, called immunogenic cell death (ICD).^1,2 A crucial hallmark distinguishing ICD from physiological/tolerogenic apoptosis is the spatiotemporally-defined and active ‘emission’ of damage-associated molecular patterns (DAMPs) by the dying cancer cells, acting as danger signals.² DAMPs mediate directional recruitment of immune cells to the site of their release followed by the promotion of recognition and phagocytosis of dying cells by antigen presenting cells (e.g., dendritic cells (DCs)). Such DAMP-“rich” uptake of dying cancer cells induces fully mature immunogenic DCs that, in turn, stimulate T cell-mediated immune response against residual (immune-resistant) tumor cells.³

All chemotherapeutics so far characterized as efficient DAMPs or ICD-inducers are chiefly used as systemic agents while physicochemical modalities (like radiotherapy/Hyp-PDT) are principally used as loco-regional therapies.⁴ Notably, imm-une responses following loco-regional application of a therapy may differ from those induced by systemic application.⁵ Therefore, it is crucial to characterize the immunoregulatory effects and DAMPs-inducing capacities of loco-regionally applied chemotherapeutics – a knowledge that is seldom available. Thus, we decided to characterize the immunogenic effects of Melphalan (Mel), a standard-of-care loco-regional treatment for limb-localized melanoma used in so-called isolated limb-perfusion (ILP) or -infusion (ILI). Of note, a previous study had shown that Mel mediates transcriptional-signatures of ER stress (a pre-requisite for ICD) and some inflammatory cytokines in patients' biopsies.⁶ However, no coherent study yet existed on Mel's ICD or danger signaling-inducing potential.

To fill this gap in knowledge we recently carried out a systematic in vitro and in vivo study to evince the immunomodulatory features of Mel-induced melanoma cell death.⁵ We showed that Mel-induced apoptosis in melanoma cells was accompanied by surface exposed (ecto)-heat shock protein 90/HSP90, but not by ecto-calreticulin/CRT or ecto-HSP70 and was not associated with the secretion of ATP (Fig. 1). We further demonstrated that the surface exposure of HSP90 required caspase signaling and both ER stress and reactive oxygen species (ROS) production. Although the exact molecular mechanisms underlying the surface exposure of HSP90 require further studies, it is interesting to note that the chaperone activity of HSP90 has been implicated in the regulation of the unfolded protein response (UPR) and intracellular trafficking and protein sorting and fusion events, 2 processes intimately linked to danger signaling.⁷

Mel-treatment failed to induce ecto-CRT despite the fact that Mel induced significant amounts of IL-8, a cytokine implicated in an anthracycline-mediated autocrine loop leading to the presentation of ecto-CRT.⁸ This observation demonstrates the contextual role of the IL-8-ecto-CRT connection. Interestingly, we found that ecto-CRT in melanoma cells could be induced by combining Mel with thapsigargin (but not with various other bona fide ER stress inducers), thereby (re-)establishing that ER-Ca²⁺ release is vital for proper trafficking of chemotherapy-induced ecto-CRT.²

At the level of the interacting immune cells, we observed that Mel-based treatment favored inflammatory/immune effector mechanisms. This conclusion was supported by several observations: (1) the spectra of different cytokines (namely IL-6, IL-8 and IL-1β) detected in Mel-ILP treated patients' loco-regional sera samples; (2) the induction of semi-mature DCs (CD86^highHLA-DR^highIL-8^highIL-1β^lowIL-6^low) by Mel-treated melanoma cells, coupled to (3) moderate activation of T cells marked by the proliferation of CD3⁺CD4⁺ and CD3⁺CD8⁺ lymphocytes coupled with low but detectable IFNγ release; and finally (4) a moderate anticancer vaccination effect. Interestingly, in vitro DC maturation and the anticancer vaccination effect did not rely on ecto-HSP90. Here, it is noteworthy that Zappasodi et al. had suggested that ecto-CRT is interchangeable with ecto-HSP90 in the paradigm of anti-B-cell lymphoma immunity in patients.⁹ In contrast, our results showed that Mel-induced ecto-HSP90 per se was not sufficient and could not substitute ecto-CRT in boosting anticancer immunity in vivo. Instead, the vaccination potential of Mel-treated melanoma cells was strongly stimulated by the exogenous addition of recombinant CRT (rCRT). Our in vivo intervention analysis also showed that the immunogenic effects of Mel were dependent on CD8⁺ T cells but not on CD4⁺ T cells. Moreover, we also confirmed that tumor necrosis factor (TNF), which is often combined with Mel-based ILP/ILI to allow better Mel-perfusion, did not solicit improved immunogenic effects. In aggregate, our study delineates that treatment of metastatic melanoma cells with Mel elicits a cell death modality that stimulates partial immunogenic effects, which can be elevated to the level of ICD, by the sole addition of exogenous rCRT. These results clearly suggest that Mel is a potential candidate for combinatorial strategies that can re-establish CRT surface exposure, for example G202, the clinically-relevant analog of thapsigargin,¹⁰ currently being tested for patients with glioblastoma and prostate cancer (NCT02067156, NCT02381236).

Due to its confinement to one limb, Mel-ILP/ILI seems to be a perfect setup to test whether combining Melphalan with rCRT could improve patients' outcome (a combinatorial strategy not possible for systemic treatment). While these results should be further tested by using appropriate murine melanoma models and under curative tumor settings, it may be speculated that Mel-loco-regional therapy can be improved by combination with soluble DAMPs/danger signals boosting antitumor responses.

Figure 1.

Antitumor immunity triggered by melphalan is potentiated by melanoma cell surface-associated calreticulin. (A) As soon as 1 h after Melphalan-based isolated limb perfusion (Mel-ILP), loco-regional serum levels of interleukin (IL)-1β and IL-6, as well as expression of IL1β, IL-6 and IL-8 in the tumor bed are increased. (B) Under in vitro settings, Mel induces apoptotic cell death accompanied by surface exposure of (ecto-)HSP90 and release of IL-8. Mel-treated melanoma cells are capable of stimulating semi-mature dendritic cells (DCs; HLA-DR⁺CD86⁺IL-1β⁺IL-6⁺IL-8⁺) as well as T-cell activation in vitro. (C) Mel-treated murine melanoma cells show anticancer vaccine potential, which can be significantly potentiated by coating Mel-treated cells with recombinant calreticulin (rCRT).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.