DICE

Abstract

The conventional view of CD95 (Fas/APO-1) is that it is a dedicated apoptosis-inducing receptor with important functions in immune cell homeostasis and in viral and tumor defense. There is an emerging recognition, however, that CD95 also has multiple non-apoptotic activities. In the context of cancer, CD95 was shown to have tumor-promoting activities, and the concept of this new function of CD95 in cancer is gaining traction. Recently, we showed that not only is CD95 a growth promoter for cancer cells, but, paradoxically, when either CD95 or CD95 ligand (CD95L) is removed, that virtually all cancer cells die through a process we have named DICE (death induced by CD95R/L elimination). In this perspective, I outline a hypothesis regarding the physiological function of DICE, and why it may be possible to use induction of DICE to treat many, if not most, cancers.

Keywords: :

• Fas
• cancer
• cell death
• immune system
• surveillance

The Proapoptotic Function of CD95 and CD95L in Normal Cells

CD95 is an established apoptosis-inducing death receptor, and the details of the signaling pathway mediating apoptosis are well studied.^1,2 In immune cells, the CD95/CD95L system is involved primarily in the downregulation of immune responses.³ This was established through the analysis of 3 mutant mouse strains. Mice with lymphoproliferative disease (lpr mice) express only about 10% of the level of normal CD95 due to a retroviral insertion into the gene. These animals are phenotypically similar to mice with generalized lymphoproliferative disorder (gld), which express a nonfunctional CD95L, and to mice carrying a point mutation in the CD95 death domain (lpr^cg mice). All of these mice exhibit accumulation of nondeleted lymphocytes and suffer from autoimmune manifestations (reviewed in ref. 3). In humans, a similar syndrome has been described.^4-8 Autoimmune lymphoproliferative syndrome (ALPS) type Ia patients carry heterozygous dominant-negative mutations in CD95; ALPS type Ib patients have mutations in CD95L, and ALPS type II patients have mutations in caspase-10. The most well-established pro-apoptotic activity of CD95 is to instruct virus-infected cells or cancer cells to die via apoptosis when engaged by cytotoxic killer cells or other immune cells that use CD95L.⁹ Thus, the fact that most cancer cells express CD95, sometimes in large quantities,¹⁰ is something of a surprise.

Nonapoptotic Signaling Through CD95 in Normal Cells and in Cancer

There is growing evidence that CD95 plays an apoptosis-independent role in nonimmune cells and in solid cancers.^11-13 In fact, CD95 acts in some tissues as a survival factor. For example, CD95 is required for efficient liver regeneration following partial hepatectomy;^14,15 CD95 is important for neurite outgrowth,^16,17 and CD95 is a potent activator of neurogenesis in both healthy and injured brains, where it activates neuronal stem cells.¹⁸ The expression of CD95 on cancer cells suggests that cancer cells maintain CD95 expression for activities other than apoptosis induction. Consistent with this assumption are data demonstrating roles for CD95 in promoting cancer cell migration, growth, epithelial-to-mesenchymal transition (EMT), and development, as shown in lung cancer,^19,20 colon cancer,^21-24 gastrointestinal cancers,^25,26 pancreatic cancer,^27,28 glioblastoma multiforme (GBM),²⁹ and in 22 cell lines representing various cancers.³⁰ Our most recent finding is that cancer fails to develop in mouse models of low-grade or endometrioid ovarian cancer or liver cancer unless CD95 is expressed,^14,31 suggesting that the reason cancer cells usually express CD95 goes beyond a function of CD95 as a tumor promoter.

The Role of CD95L in Elimination Cancer Cells

Apoptosis induced through CD95L is one of the mechanisms used by the immune system to kill virus infected and cancerous cells.³² The role of CD95L used by CD8+ T cells to control viral infections is well established. As one example, adoptive transfer of West Nile virus primed CD8+ T cells from wild-type but not from CD95L-deficient gld mice efficiently limited infection of the virus in the CNS and enhanced survival rates.³³ However, the putative role of CD95L expressing CD8+ T cells to eliminate cancer cells is more controversial.

In the early 1980s it was recognized that CD8-positive cytotoxic lymphocytes (CTL) kill cancer cells in an antigen- and Ca²⁺-dependent fashion.³⁴ The pathway that was identified as mediating Ca²⁺-dependent cytolytic activity was the granule pathway involving the transfer of granzymes into target cells with the help of perforin (Pfp). Indeed, T cells and NK cells from perforin knockout mice were later shown to have impaired ability to kill target cells.³⁵ Even earlier, a Ca²⁺-independent cytolytic pathway was postulated.³⁶ Shortly after CD95 was identified as a major apoptosis-inducing receptor engaged by immune cells, it was reported that CD95 on target cells and CD95L in CTLs was responsible for the long sought-after Ca²⁺-independent mechanism of T cell-mediated cytotoxicity.³⁷ It was shown in a number of systems that these 2 pathways, the Pfp/granzyme pathway and the CD95/CD95L ligand pathway, could account for most cytolytic activity seen in CTLs.^38-40 However, most of these early studies only tested the activity of CTLs in vitro using effector target kill assays. These in vitro findings were generally confirmed, but it was recognized that the situation in vivo might be more complicated. Tumor regression of Renca cells grown in syngeneic BALB/c mice was independent of Pfp, but was enhanced by both interferon (IFN)γ and TNFα.⁴¹ CD95 was found to be critical for mediating tumor regression in a number of models of liver cancer⁴² and models of experimental lung metastasis.^43,44 Upregulation of CD95 induced by irradiation of tumor cells was shown to enhance cytolytic activities of CTLs during adoptive immunotherapy.⁴⁵

In addition to the Pfp/granzyme and the CD95/CD95L systems, evidence of a third cytolytic activity was found, as CTLs deficient in both functional CD95L and Pfp showed residual cytolytic activity.⁴⁶ This finding was supported by other studies suggesting that CTL killing of tumor cells was possible without either of the 2 main systems. T cell-mediated tumor regression and long-term antitumor immunity in mouse models of melanoma and sarcoma were found to be independent of Pfp and CD95L. This was established using adoptive transfer of effector T cells from either perforin-deficient or from gld mice.⁴⁷ In a lung cancer mouse model, Lewis lung cancer cells were rejected in vivo in a manner dependent on CD8+ T cells that killed cancer cells independently of perforin or CD95L.⁴⁸ Depending on the tumor models studied, antitumor activity was found to depend on IFNγ,^49,50 IFNγ and CD95L,⁵¹ first IFNγ then Pfp in a 2-hit model,⁵² IFNγ and TNFα,⁵³ or TNFα.⁵⁴ One study suggested that neither Pfp nor CD95 nor TNFα were involved.⁵⁵ In addition to the role of CD95 as one of a number of mechanisms employed by cytolytic CD8+ T cells, CD95 has been reported to be the primary mechanism utilized by many CD4+ lytic effectors.^56,57 Most recently, Blankenstein and colleagues have revisited the role of CD95 in the elimination of established larger tumors using adoptive transfer of tumor specific T cells.⁵⁸ Often during immune therapy, after an initial tumor regression, the cancer relapses due to the outgrowth of antigen-loss variants (ALVs). Interestingly, in this study not only had the ALVs reduced CD95 expression, supporting a role of CD95 in the elimination of the cancer cells by CTLs, but the cells of the tumor stroma were also found to have downregulated CD95. While further analysis will be required to determine the role of CD95 in the elimination of cancer cells, there is strong evidence that the CD95/CD95L system is an important part of the antitumor defense.

Using CD95 for Cancer Therapy

The discovery in the late 1980s of 2 antibodies that could kill cancer cells^59,60 was the beginning of a new era. For the first time, it became feasible to kill cancer cells by triggering an endogenous protein expressed on most cancer cells. The initial excitement was high regarding the discovery of a natural pathway the body could use to kill neoplastically transformed cells. CD95 was confirmed to be one of the major death receptors used by immune cells to kill cancer cells. However, it soon became clear that apoptosis could not be selectively induced in cancer cells without affecting normal tissues such as the liver.⁶¹ Equally disheartening, most cancer cells are resistant to CD95-mediated apoptosis.¹⁰

Enter DICE

When either CD95 or CD95L were downregulated by siRNAs or shRNAs, growth of all tested tumors cells slowed down,¹⁴ and if the knockdown was sustained for a sufficient length of time, the tumor cells died.³¹ We reported that a substantial fraction of all tested cancer cells underwent death induced by CD95/CD95L elimination (DICE).³¹ DICE is characterized by swelling of the cells, the appearance of large dilated vacuoles, signs of autophagy, generation of ROS, and DNA DSBs. This is followed by a reduction in cell viability, activation of caspase-2, mitochondrial outer membrane permeabilization (MOMP), and, ultimately, cell death.³¹ DICE resembles a necrotic form of mitotic catastrophe. Interestingly, by performing a small-molecule screen and a genome-wide shRNA screen we did not find a single drug or a single gene that would inhibit or promote DICE.³¹ It is interesting that none of the multiple nonapoptotic signaling pathways that have been found to be engaged by CD95 were found, either alone or in combination, to be responsible for keeping cancer cells alive. Based on these data, I hypothesize that CD95 acts as a lynchpin to allow various cell survival pathways to signal at a low level (Fig. 1). The moment the connection between CD95 and any (or all) of these pathways is disrupted cancer cells will turn on the DICE program and die. DICE represents multiple cell death pathways, culminating in mitotic catastrophe, and may also involve induction of senescence based on changes in morphology seen in all cancer cells undergoing DICE.³¹ This would also be consistent with a reported mTOR-regulated connection between mitotic catastrophe and cellular senescence.⁶² This, however, requires further testing. Based on data from both CD95 and CD95L knockout mice, which do not show signs of cell death or growth deficiencies in any tissue outside the immune system,^63-65 DICE may preferentially affect cancer cells and have little effect on normal cells. This prediction is supported by the finding that hTERT immortalized fibroblasts were more susceptible than nonimmortalized fibroblasts to DICE induced by either CD95 or CD95L knockdown.³¹ This suggests that dependence on CD95 and CD95L for survival is a feature of cancer cells that distinguishes them from most normal cells. The few normal cells that seem to be affected in their growth/viability after elimination of CD95 are proliferating T cells⁶⁶ and hepatocytes in the regenerating liver,¹⁴ both tissues prone to neoplastic transformation.

Figure 1. Model of how CD95 protects cancer cells from cell death. CD95 acts as a lynchpin to ensure a low level of signaling through multiple survival pathways. When either CD95 or CD95L is eliminated, cancer cells undergo DICE by activating multiple cell death pathways.

The Physiological Function of DICE, a Hypothesis

As summarized above, it appears that cells of the immune system employ well-established activities of CD95/CD95L in the elimination of neoplastically transformed cells. Infiltrating immune cells kill cancer cells by activating the canonical caspase-8-dependent apoptosis pathway in the cancer cells (Fig. 2, I). On rare occasions, this process fails due to mutations or epigenetic deregulation of genes in the cancer cells, rendering them resistant to CD95 mediated apoptosis. Accumulating data suggest that under these conditions, stimulation of CD95 on the tumor cells by CD95L drives tumor growth, motility, and dedifferentiation of the cancer cells (Fig. 2, II). When a deletion or mutation of CD95 occurs in tumor cells or rapidly proliferating cells (i.e., T cells or regenerating liver cells), DICE is turned on (Fig. 2, III). DICE is, therefore, a fail-safe mechanism that prevents the survival of cells that are devoid of CD95 and, hence, cannot be eliminated by the immune system through CD95L/CD95-mediated apoptosis. The observation that tumor cells rarely if ever mutate or delete both alleles of CD95 (reviewed in ref. 67) is consistent with this interpretation.

Figure 2. Hypothesis on the physiological relevance of DICE. CD95L-induced apoptosis is one of the mechanisms that tumor-infiltrating lymphocytes (TILs) employ to eliminate transformed cells (I). If CD95 engagement fails to induce apoptosis due to cancer cells acquiring mutations of components of the apoptosis pathways, then cancer can develop. These cancer cells now benefit from the CD95L expressed by themselves and, presumably, by TILs (II). Transformed cells and cells in rapidly proliferating tissues under conditions that can lead to neoplastic transformation (e.g., compensatory proliferation of liver tissue) must receive a low level signal through CD95 to survive. If CD95 or its stimulating ligand are completely removed, then cells die by DICE (III). According to this model, CD95 and CD95L are part of a fail-safe mechanism to prevent the occurrence of CD95-deficient cancer cells that can no longer be targeted by immune cells via CD95L. TIL, tumor-infiltrating lymphocyte secreting CD95L. DSB, DNA double-strand breaks.

DICE as Cancer Therapeutics

Developing new cancer therapeutics is often based on the following strategy: A gene or a gene product (RNA or protein) is found to be amplified or upregulated or mutated in primary tumors, metastases, or recurrent cancer; or a gene is identified as a so-called driver mutation by genome-wide analyses. The suspect gene is studied functionally to determine whether it has tumorigenic activities, and then is tested in mouse tumor models and analyzed using patient material for its usefulness as a prognostic marker. If it passes these tests, it is viewed as an oncogene and considered as a potential biomarker or a target for cancer therapy. Despite the wide acceptance of this approach, the fact remains that targeting oncogenes clinically has only been partially successful and only in a few cancers (e.g., BCR-ABL by Imatinib in CML⁶⁸ or B-RAF[V600E] by Vemurafenib in melanoma).⁶⁹ The induction of DICE represents a different concept for cancer therapy. To induce DICE, I propose targeting CD95L and/or CD95, which are widely considered to have tumor-suppressive activities and are often downregulated (albeit only to a basal level) in cancer. Based on our data, I further hypothesize that tumor cells will not readily develop resistance to DICE. DICE is not a mechanism that prevents tumor development; thus, any tumor that develops has not been under selection pressure from DICE. DICE only prevents survival of existing tumor cells that lose CD95 functionality and, therefore, cannot be eliminated by immune cells. Inducing DICE in tumor cells by targeting CD95 or CD95L could be a novel and powerful means to eradicate cancer.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.