Apoptosis, autophagy, necroptosis, and cancer metastasis

Cancer metastasis is a complex process that can be divided into five major steps: the first step, invasion, is characterized by increased cell motility caused by alterations in cell-cell and cell-ECM interactions [2]. The second step is intravasation, in which tumor cells escape from the primary site and migrate into circulation systems. The third step, dissemination, is the process in which malignant cells travel through the circulation systems to reach a capillary bed, where the cancer cells adhere to the vessel walls or are detained at these sites because of size constraints. The fourth step is extravasation, in which cancer cells permeate the vessels to enter their destination organs. Colonization is the final step, in which metastatic cells proliferate and form micrometastases or macrometastases [2]. Alternatively, metastasis can be considered as a two-phase process according to a new perspective [3]: the first phase involves the physical translocation of a cancer cell to a distant organ, whereas the second phase encompasses the process of the development of the cancer cells into a metastatic lesion at the distant site. Typically, the initial steps of metastasis (invasion, intravasation, dissemination, and extravasation) proceed at a very high efficiency, but the final step, colonization, is less efficient. It has been estimated that only ~0.01% of circulating tumor cells ultimately produce macrometastases [4]. This inefficiency may be closely related to the activation of cell death machinery by various stresses before or after the cells reach a new environment. Such stresses include the loss of cell-cell contacts, the recognition and destruction of the cancer cells by the immune system, and the lack of necessary growth factors, all of which may trigger programmed cell death, including apoptosis, autophagy and necroptosis [4].

Apoptosis is a type of programmed cell death that is characterized by cell membrane blebbing, cell shrinkage, nuclear fragmentation, chromatin condensation, and chromosomal DNA fragmentation [5,6]. There are two basic apoptotic signaling pathways: the extrinsic and the intrinsic pathways [7]. The intrinsic apoptotic pathway is activated by various intracellular stimuli, including DNA damage, growth factor deprivation, and oxidative stress. It relies on the formation of a complex termed the apoptosome, composed of procaspase-9, apoptotic protease-activating factor (Apaf-1), and cytochrome c. A series of Bcl-2 family members, such as Bax, Bak, Bcl-2, and Bcl-x_L, control the release of cytochrome c by regulating mitochondrial membrane permeabilization. The extrinsic pathway of apoptosis is initiated by the binding of death ligands [e.g., Fas ligand (FasL), TNF-related apoptosis inducing ligand (TRAIL), and TNF-α] to death receptors of the TNF receptor superfamily. This interaction is followed by the assembly of the death-inducing signaling complex (DISC), which consists of the Fas-associated death domain (FADD) protein and procaspase-8/10. DISC then either activates downstream effector caspases (caspase-3, 6 and 7) to directly induce cell death or cleaves the Bcl-2 family member Bid into tBid to activate the mitochondria-mediated intrinsic apoptotic pathway [7]. Numerous factors, such as p53, cellular inhibitor of apoptosis proteins (cIAPs), and NF-κB, have been reported to be involved in the regulation of apoptotic pathways [2,8]. Many small molecules targeting apoptotic pathways have been developed for cancer therapy. For example, ABT-737, ABT-263, and GX15-070 have been reported to act on Bcl-2 family members; GDC-0152, birinapant, AT-406, and HGS-1029 have been designed to antagonize cIAPs, among the most promising targets for anti-cancer agent development; and Nutlins, MI-219 and MI-77301 have been shown to antagonize murine double-minute 2 (MDM2), a critical negative regulator of p53 that promotes p53 ubiquitination and degradation [9,10].

Apoptosis may block metastatic dissemination by killing misplaced cells. Thus, apoptosis serves as an important process for inhibiting metastasis. The success of the metastatic process relies on the ability of malignant cells to escape apoptosis. Apoptotic resistance is indispensable for all steps of metastatic progression, but the most critical step may be the resistance to cell death induced by the loss of cell-cell and cell-ECM contacts [8]. The detachment of cells from the ECM induces a type of apoptosis termed anoikis. Numerous reports have demonstrated that anoikis resistance is frequently observed in metastatic cells [11-14]. For instance, TrkB, a neurotrophic tyrosine kinase receptor, was found to act as a specific suppressor of caspase-associated anoikis in non-malignant epithelial cells. TrkB activates the phosphatidylinositol-3-OH kinase (PI3K)/protein kinase B (PKB) pathway to promote the formation of large cellular aggregates that survive and proliferate in suspension, and these cellular aggregates develop into rapidly growing tumors that infiltrate lymphatics and blood vessels to colonize a distant organ in mice [13]. In addition, signal transducer and activator of transcription 3 (STAT3) was found to play a role in conferring anoikis resistance to pancreatic cancer cells and in promoting metastasis. Enhanced STAT3 expression and phosphorylation at Tyr 705 have been associated with the anoikis resistance and the metastatic capacity of pancreatic cells [14].

In addition, metastatic cells must develop a mechanism to evade cell death resulting from recognition and destruction by cytotoxic lymphocytes such as natural killer (NK) cells. Furthermore, tumor cells must survive in the environment of reactive oxygen species (ROS) produced by endothelial cells when crossing the vessel or tissue barrier during the extravasation step [15,16]. Finally, malignant cells must tolerate hypoxic conditions and proliferate in an environment lacking the necessary cytokines for growth to achieve successful colonization at their destination sites [8,17].

Even after the successful formation of micrometastases, macroscopic tumors may not develop because of dormancy [18]. The nature of dormancy remains to be elucidated. One hypothesis states that the rate of proliferation is balanced by the rate of apoptosis such that no net tumor growth occurs [19], whereas another hypothesis states that dormant cells neither proliferate nor undergo apoptosis [20]. However, genetic variation and/or environmental stimulation may drive cells out of dormancy and into an anti-apoptotic, highly proliferative state [21].

In the Table 1, we summarize the roles of the known major apoptotic factors that participate in cancer metastasis.

Among these modulators of apoptosis in Table 1, the c-Jun N-terminal kinases (JNK), TGF-β, and matrix metalloproteinase (MMP) pathways play dual roles in apoptosis and metastasis. JNKs, a subgroup of the MAP kinase superfamily, are indispensable for both cell proliferation and apoptosis. Whether the activation of JNKs leads to cell proliferation or apoptosis is dependent on the cell type, the nature of the death stimulus, the duration of its activation and the activities of other signaling pathways [66]. In the absence of NF-κB activation, enhanced JNK activation contributes to TNF-α induced apoptosis. JNK is also a critical mediator of UV radiation-induced apoptosis. JNK promotes apoptosis via different mechanisms. Activated JNK translocates to the nucleus and transactivates c-Jun and other transcription factors (e.g., p53), which further transactivate various pro-apoptotic genes, such as Fas-L, Bak, and p53-upregulated modulator of apoptosis (PUMA) [67]. In addition, JNKs contribute to apoptosis by modulating the activities of mitochondrial pro- and antiapoptotic proteins via distinct phosphorylation events. However, JNK inhibits apoptosis in IL-3-dependent hematopoietic cells via the phosphorylation and antagonism of the proapoptotic Bcl-2 family protein BAD [66].

It was reported that JNK signaling prevented the progression of invasive adenocarcinoma in PTEN^−/− prostate cancer. Mice exhibiting JNK deficiency in the prostate epithelium (JNK and PTEN double-deficient mice) develop androgen-independent metastatic prostate cancer more rapidly than control (PTEN-deficient) mice. In addition, JNK-deficient progenitor cells exhibited increased proliferation and tumorigenic capacity compared with progenitor cells from control prostate tumors [45].

A group reported that Notch and myocyte enhancer factor 2 (Mef2) cooperated to promote proliferation and metastasis via JNK signal activation and the consequent induction of the invasion marker MMP1 in a Drosophila model [46]. Another study showed that receptor for advanced glycation end products (RAGE) splice variant 1 inhibited tumor formation, cell invasion, and angiogenesis induced by RAGE ligand signaling, which was closely related to the strong suppression of JNK by this splice variant protein [47].

The transforming growth factor-β (TGF-β) family proteins bind to cell surface type I and type II serine/threonine kinase receptors (TGFβRI and TGFβRII, respectively) and SMAD mediators to regulate many biological processes. Numerous studies have reported roles of the TGF-β pathway in apoptosis. Many pro-apoptotic genes are under the control of the SMAD transcription-regulating complexes. For example, TGF-β-inducible early response gene (TIEG1) [56], death-associated protein kinase (DAPK) [57], and SH2 domain-containing inositol-5-phosphatase (SHIP) [58] have been shown to be essential for the suppression of cell proliferation and the induction of apoptosis in many cell types. Another mechanism of TGF-β-related apoptosis is the induction of the mislocalization of a mitochondrial septin family member ARTS. Upon translocation, ARTS binds to and inactivates XIAP, a key inhibitor of apoptosis, which leads to the activation of caspase-3 and apoptosis [59]. In addition, TGF-β signals induce the death associated protein (Daxx)-JNK pathway to induce apoptosis under certain circumstances [60].

The roles of the TGF-β signaling pathway in cancer development are uncertain [61]. Prior to tumor initiation and during the early stage of progression, TGF-β acts as a tumor suppressor via cell cycle arrest and the induction of apoptosis; however, at advanced stages, TGF-β often acts as a tumor promoter via the acceleration of the EMT, invasion, and angiogenesis, the maintenance of tumor stem cells, and the alteration of the tumor microenvironment [61]. Interestingly, TGF-β is expressed at high levels during the late stages of tumor progression in many human cancers, but paradoxically, the TGF-β pathway is frequently mutationally inactivated in cancer cells. Further studies revealed that elevated TGF-β expression played an important role in promoting stromal cells to secret cytokines that favor cancer cell metastasis [68].

Matrix metalloproteinases (MMPs) constitute one of the most prominent families of proteinases associated with tumor metastasis. MMPs interfere with the induction of apoptosis in malignant cells via the cleavage of ligands or receptors in apoptotic pathways [62]. For instance, MMP-7 was reported to cleave membrane-bound FasL on doxorubicin-treated cancer cells, thereby attenuating apoptosis and increasing the resistance of these cells to chemotherapy [63,64]. MMP-13 was shown to be involved in the shedding of nerve/glial antigen 2 (NG2), a novel anoikis receptor, thereby contributing to the attenuation of anoikis [65].

Autophagy is an evolutionarily conserved catabolic process in which intracellular membrane structures package protein complexes and organelles to degrade and renew these cytoplasmic components. It is thus critical for cell growth regulation and internal homeostasis [69]. Autophagy is physiologically a cellular strategy and mechanism for survival under stress conditions. When over-activated under certain circumstances, excess autophagy results in cell death. To date, three types of autophagy have been identified: macroautophagy, microautophagy, and chaperone-mediated autophagy. We only discuss macroautophagy in this review; therefore, henceforth, "autophagy" specifically refers to macroautophagy. Autophagy is a multi-step process that includes nucleation, elongation, and autophagosome and autolysosome formation and that is executed by a series of highly conserved genes termed autophagy-related genes (ATGs) [70]. Autophagy is often triggered by nutrient deprivation, ROS, hypoxia, drug stimuli, and endoplasmic reticulum (ER) stress via complex signal transduction pathways. Alterations in the autophagy machinery may lead to diverse pathological conditions, such as neurodegeneration, ageing, and cancer [71]. Mammalian target of rapamycin complex 1 (mTORC1), class I PI3K, AKT, class III PI3K, Beclin-1 and p53 are critical components of the autophagic pathway that have become major targets of autophagy-related drug design. Numerous small molecules have been found to target these components and to play a role in tumor treatment. For example, rapamycin and its derivatives (i.e., rottlerin, PP242 and AZD8055) target the PI3K/AKT/mTOR signaling pathway to induce autophagy; spautin-1 and tamoxifen regulate Beclin-1 activity to inhibit and promote autophagy, respectively; and oridonin and metformin trigger p53-mediated autophagy and cell death [72].

The role of autophagy in cancer metastasis is complex, as reports have indicated both pro-metastatic and anti-metastatic roles of autophagy. Stage-specificity may affect the cellular response to autophagy during cancer metastasis [73]. During the early stage of cancer metastasis, autophagy may act as a suppressor of metastasis by restricting tumor necrosis and inflammatory cell infiltration and by alleviating oncogene-induced senescence. These processes may help to reduce the invasion and dissemination of cancer cells from the primary site. During the advanced stages of metastasis, autophagy tends to act as a promoter of metastasis by promoting ECM-detached metastatic cell survival and colonization in a distant site and by inducing metastatic cells that fail to establish contact with the ECM in the new environment to enter dormancy.

Necrosis was originally considered to be an accidental and unregulated cell death. Accumulating evidence has shown that necrosis can be induced and proceed in a regular manner like apoptosis, although in a caspase-independent fashion. Regulated necrosis is termed “programmed necrosis” or “necroptosis” to distinguish it from necrosis caused by physical trauma [110]. Necroptosis can be induced by the activation of the TNF receptor superfamily [111], T cell receptors [112], interferon receptors [113], Toll-like receptors (TLRs) [114], cellular metabolic and genotoxic stresses, or various anti-cancer agents. It can be pharmacologically inhibited by chemical compounds such as necrostatin-1 (Nec-1) [115]. The formation of the “necrosome” by receptor-interacting protein kinase 1 (RIP1) and RIP3 is one of the most critical characteristics of necroptosis. It is a multi-step process that contains three key checkpoints. For example, in TNF-α mediated necroptosis [110], at the first checkpoint, the E3 ligases cellular inhibitor of apoptosis 1 (cIAP1) and cIAP2 induce RIP1 ubiquitination [116], which blocks necroptosis via NF-κB-dependent or -independent mechanisms. The removal of ubiquitin chains from RIP1 by the deubiquitinase cylindromatosis (CYLD) is critical for the packaging of Complex IIa (including caspase-8, FADD, and RIP1) and Complex IIb [(including caspase-8, FADD, RIP1, RIP3, and mixed lineage kinase domain-like (MLKL)] [117]. At the second checkpoint, activated caspase-8 cleaves and abolishes the activities of RIP1, RIP3, and CYLD [118-120]. Cleaved RIP1 and RIP3 lose their capabilities of trans-phosphorylation and downstream substrate phosphorylation. At the third checkpoint, when the disruption of RIP1 and RIP3 is prevented by caspase-8 inhibitors (i.e., zVAD) or by the genetic inhibition of caspase-8 or FADD, the trans-phosphorylation of RIP1 and RIP3 promotes their aggregation into the filamentous-like necrosome [110]. MLKL is further phosphorylated by RIP3 and is recruited to the necrosome by its interaction with RIP3 [121]. MLKL forms a homotrimer via its amino-terminal coiled-coil domain and translocates to the plasma membrane during TNF-induced necroptosis, which leads to necrotic plasma membrane permeabilization [122]. In addition to MLKL, phosphoglycerate mutase 5 (PGAM5) is a downstream substrate of RIP3 [123].

Several molecules and pathways contribute to the execution of TNF receptor-mediated necroptosis. Some of these effectors are also involved in other receptor-mediated necroptosis pathways [124]. During the progression of necroptosis, ROS are generated [125,126], resulting in lipid peroxidation and increased mitochondrial membrane permeability. The cytosolic ATP levels are sharply reduced due to the attenuation of ATP transport from the mitochondria to the cytosol and ATP consumption by over-active poly(ADP-ribose) polymerase 1 (PARP1). Apoptosis-inducing factor (AIF) is released from mitochondria due to the increase in the mitochondrial membrane permeability and enters the nucleus to cleave DNA. Lysosomal membrane permeabilization (LMP) also occurs during necroptosis, resulting in the leakage of cytotoxic hydrolases into the cytosol [124]. In addition, dynamin-related protein l (Drp1), which acts downstream of PGAM5, is thought to regulate mitochondrial fission to execute necroptosis [123].

Necroptosis plays an indispensable role during normal development. Moreover, it has been implicated in the pathogenesis of a variety of human diseases, including cancer [127]. The necroptosis machinery is often impaired during tumorigenesis and tumor progression. For example, chronic lymphocytic leukemia (CLL) cells failed to undergo necroptosis upon stimulation using TNFα combined with the pan-caspase inhibitor zVAD. Two key components of necroptotic machinery, RIP3 and CYLD, were markedly downregulated in CLL [128]. In non-Hodgkin lymphoma, single nucleotide polymorphisms (SNPs) in the RIP3 gene were detected in 458 patients and correlated with increased risk of non-Hodgkin lymphoma, which indicates that genetic variations in the RIP3 gene may contribute to the onset of this disease [129]. There is a growing list of compounds and anticancer agents that have been shown to induce necroptosis in cancer cells. Shikonin was the first reported small molecule that induced necroptosis [130]. Numerous studies have shown that shikonin and its analogs not only are highly tumoricidal but also exhibit the ability to bypass drug resistance machineries [130-133]. In addition, compounds such as 5-benzylglycinyl-amiloride, obatoclax, and D-galactose employ necroptosis to kill malignant cells [134-136]. Notably, some traditionally pro-apoptotic anti-cancer agents have recently been demonstrated to share the ability to induce the necroptosis of tumor cells under certain circumstances. For example, interferon-β-armed oncolytic adenovirus (ZD55-IFN-β) induced both apoptosis and necroptosis in cancer cells. However, Nec-1 treatment converted ZD55-IFN-β-induced necroptosis to apoptosis [137]. In addition, although TRAIL is a well-known apoptosis inducer, one study showed that acidic extracellular pH converted TRAIL-induced apoptosis to necroptosis in human HT29 colon and HepG2 liver cancer cells via a process that involved RIPK1/RIPK3-dependent PARP-1 activation [138].

To date, few studies have associated necroptosis with metastasis. Fu et al. reported that shikonin greatly reduced the lung metastasis of osteosarcoma by inducing RIP1- and RIP3-dependent necroptosis [132]. The induction of a high level of ROS via necroptosis may represent one factor that restricts cancer cell metastasis [139]. As we have described above, metastatic cells in the circulation or at new sites must survive in an environment without interacting with the ECM; under these conditions, tumor cells face major difficulties in maintaining nutrient and energy equilibration and in antagonizing metabolic stresses, especially ROS. Disseminated tumor cells have evolved different strategies to restore their ATP levels and to restrict cellular ROS production, including the over-activation of pro-survival signals (such as PI3K, Ras-ERK, and NF-κB), the enhancement of antioxidant activity, the altered activation of metabolic pathways (preferentially the glycolysis and pentose phosphate pathways), and the initiation of autophagy. However, necroptosis represents another mechanism to eliminate metastatic cancer cells by triggering ROS bursts. RIP3 has been observed to be critical for regulating ROS production during necroptosis [126], and this finding is in agreement with another study showing that RIP3 activates several metabolic enzymes [including glycogen phosphorylase (PYGL), glutamate-ammonia ligase (GLUL), and glutamate dehydrogenase 1 (GLUD1)] to regulate TNF-induced ROS production. Silencing each of these enzymes results in decreased levels of ROS accumulation and cell death [125]. Therefore, it appears to be reasonable that necroptosis is an important mechanism that restricts tumor metastasis. In this case, tumor cells must overcome both anoikis and necroptosis to successfully metastasize.

Programmed cell death in vivo involves the complex interaction between apoptosis, autophagy, and necroptosis [140]. In some cases, a specific stimulus triggers only one type of programmed cell death, but in other situations, the same stimulus may initiate multiple cell death processes. Different types of mechanisms may co-exist and interact with each other within a cell, but ultimately, one mechanism dominates the others. The decision taken by a cell to undergo apoptosis, autophagy, or necroptosis is regulated by various factors, including the energy/ATP levels, the extent of damage or stress, and the presence of inhibitors of specific pathways (e.g., caspase inhibitors). ATP depletion activates autophagy. However, if autophagy fails to maintain the energy levels, necroptosis occurs [141]. Slight/moderate damage and low levels of death signaling typically induce apoptosis, whereas severe damage and high levels of the death signaling often result in necroptosis [142]. Although apoptosis is often the first mode of cell death and although necroptosis is triggered only as a backup mechanism to ensure that cell death occurs, emerging evidence has shown that the necroptotic pathway may predominate under certain pathological conditions [142]. The complex relationships between different types of cell death and cancer metastasis are depicted in Figure 2.

In the course of cancer metastasis, malignant cells must overcome a series of unfavorable conditions, including detachment from the ECM, attack by immune cells, hypoxia and a growth factor-lacking environment, which cause increased cellular ROS production and DNA damage and an insufficient energy status. Therefore, most metastatic cells from the primary tumor are unable to successfully macrometastasize and are killed via apoptosis or necroptosis. On one hand, autophagy greatly improves the fitness of cancer cells under stressful conditions and, thus, attenuates apoptosis and necroptosis, but on the other hand, autophagy antagonizes metastasis by restricting tumor necrosis and subsequent immune cell infiltration. Additionally, excess autophagy induces the death of metastasizing cells. Therefore, the interaction between different types of cell death and cancer metastasis is highly complex. In addition, cancer cells have evolved sophisticated mechanisms to antagonize apoptosis and necroptosis. However, defects in the machinery of one type of cell death may not affect that of another. Thus, triggering a single type of programmed cell death may not be sufficient for the treatment of cancer metastasis. The selection of different cell death inducers or the combined use of different cell death pathway inducers will help to overcome drug resistance to kill metastatic cells [140,143,144].