Autophagy and Organelle Homeostasis in Cancer

Abstract

Beginning with the earliest studies of autophagy in cancer, there have been indications that autophagy can both promote and inhibit cancer growth and progression. Autophagy regulation of organelle homeostasis is similarly complicated. In this review, we discuss pro- and anti-tumor effects of organelle-targeted autophagy and how this contributes to several hallmarks of cancer, such as evading cell death, genomic instability, and altered metabolism. Typically, the removal of damaged or dysfunctional organelles prevents tumor development but can also aid in proliferation or drug resistance in established tumors. By better understanding how organelle-specific autophagy takes place and can be manipulated, it may be possible to go beyond the brute force approach of trying to manipulate all autophagy in order to improve therapeutic targeting of this process in cancer.

Thorburn ETOC

Autophagy can both promote and inhibit cancer growth and progression. Miller and Thorburn review the pro- and anti-tumor effects of organelle-targeted autophagy and their relationship to cancer hallmarks, such as evading cell death, genomic instability, and altered metabolism. These insights may help improve therapeutic targeting of autophagy in cancer.

Introduction

The recycling processes that deliver excess or damaged cytoplasmic material to lysosomes for degradation are one of three pathways: micro-autophagy, chaperone-mediated autophagy, and macro-autophagy. Macro-autophagy (hereafter referred to as autophagy) is a complex process that promotes both bulk and selective degradation of damaged proteins and organelles or spare molecules to generate macromolecular building blocks and fuel metabolic pathways (Dikic and Elazar, 2018). Autophagy involves over 20 core autophagy proteins, encoded by ATG genes, that envelop cytoplasmic cargo within a double-membrane vesicle structure called the autophagosome (Suzuki et al., 2001). After engulfment of the cargo, autophagosomes fuse with lysosomes, where pH-sensitive hydrolases mediate degradation of the engulfed material (Dikic and Elazar, 2018). Selective autophagy occurs through targeting of specific cargos, including organelles, to autophagosomes. Micro-autophagy and chaperone-mediated autophagy are considered to be selective forms of autophagy that involve direct delivery mechanisms of targeted material to the lysosome (Anding and Baehrecke, 2017, Mijaljica et al., 2011, Sahu et al., 2011, Kaushik and Cuervo, 2018). Selective autophagy has three distinct steps: designation, targeting and sequestration, and degradation. Designation occurs when a cargo receives a molecular tag for its recognition by the autophagy machinery. Ubiquitination often provides this signal in mammalian cells (Kim et al., 2008, Rogov et al., 2014). Next, targeting and sequestration occur when designated cargoes are delivered to autophagosomes by specific receptors, allowing the autophagosome to engulf and sequester cargo. Numerous Autophagy Cargo Receptors (ACRs) have been identified in mammals, including SQSTM1/p62, NBR1, optineurin (OPTN), NDP52, TAX1BP, BNIP3-like (BNIP3L)/NIX, and TRIM proteins (Rogov et al., 2014, Mandell et al., 2014). A key step in the formation of autophagosomes is the conjugation of the LC3/GABARAP family of proteins to the lipid phosphatidylethanolamine (PE), which is mediated by the autophagy conjugation machinery including ATG3, ATG5, ATG7, and ATG12. Importantly, these conjugation events are frequently used as a method to monitor autophagy levels (the most common method to quantitate the amount of autophagy is to measure the rate of conjugation of LC3B to PE). A common way to block autophagy involves the inactivation by knockout or knockdown of one or other component of the conjugation machinery. These conjugation events are important for proper development of autophagosomes and the conjugated LC3 molecules on the inner autophagosome membrane serve as targets that are recognized by the ACRs. Then, fusion of the autophagosome with the lysosome takes place, followed by degradation of the inner autophagosome membrane, allowing lysosomal hydrolases access to degrade the cargo. Lastly, macromolecular precursors including amino acids, lipids etc. resulting from degradation of the cargo are released to the cytoplasm to fuel metabolic pathways and serve as substrates for the creation of new macromolecules.

Autophagy can both promote and inhibit tumor growth in different contexts (White, 2015, Amaravadi et al., 2019). For example, genetic inhibition of Atg5 or Atg7 showed that deletion of autophagy in tumor-prone mouse models following RAS pathway activation caused an increase in preneoplastic lesions and tumor incidence, indicating that autophagy protects against cancer development (Strohecker et al., 2013, Rao et al., 2014, Rosenfeldt et al., 2013, Yang et al., 2014). In these mice, it is likely that the defective autophagy process causes chronic tissue damage, leading to tumor-initiating inflammation (White et al., 2010). Various other tumor suppressive functions of autophagy have also been proposed, as autophagy plays an important role in tissue homeostasis. Most of these mechanisms likely affect tumor initiation and early steps in cancer progression, including removal of damaging, reactive oxygen species (ROS)–inducing mitochondria, degradation of oncogenic viruses, maintenance of genomic stability, and a role in oncogene-induced senescence via degradation of the nuclear lamina (Levine and Kroemer, 2019, Wang et al., 2016, Dou et al., 2015). However, it is important to note that it is possible to have mutations or expression of alternate splice variants of autophagy genes that negatively regulate autophagy (Wible et al., 2019), this seldom occurs in cancer (Lebovitz et al., 2015). Later in disease progression, it is more common for oncogenic events to activate autophagy or enhance lysosomal biogenesis to withstand nutrient deprivation as well as prevent excessive accumulation of the toxic molecules discussed above (Amaravadi et al., 2019).

Therefore, once a tumor becomes established, autophagy inhibition often results in less aggressive cancers. For example, pancreatic cancers have increased autophagy flux that contributes to their growth (Yang et al., 2011) through multiple tumor cell intrinsic and nontumor-intrinsic mechanisms (Piffoux et al., 2020, Yang et al., 2018). As such, mutations in autophagy genes would thwart the tumor’s rapid growth and metastasis and instead result in death or senescence. This increase in autophagy flux makes pancreatic cancer a strong candidate for therapeutic targeting of autophagy. Roles of autophagy in cancer and the potential of targeting autophagy as a therapeutic strategy for various cancers have been extensively reviewed (Mulcahy Levy et al., 2017, Galluzzi et al., 2015, Amaravadi et al., 2019, Mulcahy Levy and Thorburn, 2020, White, 2015). Because tumors are established at the time of detection and treatment, autophagy inhibition would be effective for this stage of the disease. However, autophagy also has pro- and anti-metastatic effects. For example, autophagy can aid in the ability to endure environmental stress as well as promote an anti-tumor immune response or cell dormancy for migrating cancer cells {Kenific, 2010 #387}. Given these competing roles of autophagy in general, targeting organelle homeostasis rather than general autophagy processes could potentially be a more effective way to treat cancer by allowing more precision in targeting pro-tumor functions while avoiding anti-tumor promoting functions of autophagy.

In this Review, we discuss how the role of autophagy in organelle homeostasis may affect tumor behavior. However, it is important to keep in mind that, although many studies propose critical roles for specific types of organelle-phagy in cancer cell behavior, such conclusions may be over simplified. In many of these studies, a component of the autophagy machinery, often a component of the conjugation machinery, is inhibited as a way to block organelle-specific autophagy. But such studies do not exclude the possibility that the biological effect depends not (only) on the loss of degradation of the specific organelle in question, but also on the absence of degradation of other molecules or organelles. In addition, in many cases it remains possible that the effect is due to autophagy-independent functions autophagy components (Galluzzi and Green, 2019). Therefore, specific manipulation of organelle-specific autophagy pathways is required to confirm that a biological effect arises due to autophagy of that specific organelle. However, this may be difficult due to the redundancy of autophagy adaptors utilized across different types of organelle-targeted autophagy.

Mitophagy and Cancer

Mitochondria are well-known for their role in ATP generation (Green and Kroemer, 2004), but they have several other functions that are intertwined with hallmarks of cancer (Giampazolias and Tait, 2016). Mitochondria are vulnerable to damage from high levels of ROS, a byproduct of aerobic respiration that can also cause DNA mutations and protein damage/misfolding. Mitochondria are also involved in fatty acid synthesis, amino acid production, heme synthesis and iron–sulfur cluster biogenesis, and play an essential role in the execution of apoptosis by releasing cytochrome c (Wang and Youle, 2009). As such, the clearance of mitochondria can have diverse effects on tumor development, growth, and progression.

The autophagy-targeted degradation of mitochondria is accomplished by a selective form of autophagy, named mitophagy (Lemasters, 2005) (Figure 1). Mitophagy can be initiated by stimuli that cause mitochondrial damage, such as hypoxia (Bellot et al., 2009; Zhang et al., 2008), chemical uncouplers (Narendra et al., 2008), or ROS (Frank et al., 2012; Zhang et al., 2008). The best understood mechanism of mitophagy involves PTEN-induced putative kinase 1 (PINK1), a mitochondrially localized kinase, and Parkin, a cytosolic E3 ubiquitin ligase (Narendra et al., 2010), both of which are mutated in early-onset recessive Parkinson’s disease. A small portion of PINK1 is imported into mitochondria through two complexes: the translocase of the outer membrane (TOM) and the translocase of the inner membrane (TIM23). While in the inner mitochondrial membrane, matrix processing peptidase (MPP) clips the intra-membrane section of PINK1 and the inner membrane protease PINK1/PGAM5- associated rhomboid-like protease (PARL) (Jin et al., 2010). After cleavage and activation, PINK1 phosphorylates Parkin to increase its ubiquitin E3 ligase activity (Shiba-Fukushima et al., 2012, Koyano et al., 2014) resulting in activation of Parkin and the addition of ubiquitin to various mitochondrial substrates, eventually forming ubiquitin chains on a variety of these proteins (Okatsu et al., 2015). Several ACRs mediate Parkin-dependent mitophagy in mammalian cells: p62/SQSTM1, NDP52, OPTN, NIX, FUNDC1, BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (BNIP3), and TAX1BP1 (Lazarou et al., 2015, Heo et al., 2015, Chen et al., 2016). These ARCs direct ubiquitinated mitochondria to phagophores by binding the ubiquitin molecules on the mitochondrial surface and LC3 family members on the inner autophagosome membrane via a LC3-interacting region (LIR) domain (Figure 1).

Figure 1. Dysfunctional Mitochondria can both Promote and Inhibit Cancer Growth and Survival.

A. Healthy mitochondria produce ATP for cellular energy use. Overworking mitochondria or treatment with anticancer agents such as cisplatin can promote mitochondrial damage. Damaged mitochondria can promote tumor survival and progression through ROS and mitochondrial DNA release as well as promoting the Warburg effect. However, the release of cytochrome C from mitochondria in this state could promote apoptosis. B. Clearance of mitochondria through mitophagy typically occurs through PINK1-mediated phosphorylation of Parkin, which promotes protein ubiquitination. However, mitophagy can also occur in the absence of Parkin, for example via AMBRA1–1 and HUWE1-mediated ubiquitination of MFN2. Also, BNIP3 and NIX can recruit LC3 to the mitochondria in the absence of ubiquitinated mitochondrial membrane proteins.

Although cancer cells often do not express Parkin (Chang et al., 2017, Zhang et al., 2020), they adopt similar mechanisms for removal of mitochondria, usually through alternate ubiquitin ligases (Villa et al., 2018). ARIH1 is a ubiquitin ligase that has the potential to promote mitophagy in the absence of Parkin in various types of cancer cells (Villa et al., 2017). Similarly, Autophagy and Beclin1 Regulator 1 (AMBRA1) localizes to damaged mitochondria through LIR motif-dependent interactions with LC3, promoting both Parkin-dependent and –independent mitochondrial degradation (Strappazzon et al., 2015). This interaction enhances the mitochondrial localization of HECT, UBA And WWE Domain Containing E3 Ubiquitin Protein Ligase 1 (HUWE1), which ubiquitinates mitofusin-2 (MFN2) to promote mitophagy after membrane depolarization (Di Rita et al., 2018). Mitophagy receptors BNIP3 and NIX can also trigger mitophagy in the absence of protein ubiquitination (Novak et al., 2010, Hanna et al., 2012). All these mechanisms rely upon LC3 conjugation to selectively sequester mitochondria in the autophagosome. One major unanswered question is whether mitophagy can still occur in cancer cells when the autophagy apparatus is disabled as occurs when, as discussed below, we design experiments where we inactivate a component of the conjugation machinery in tumors.

Mitophagy as a Pro-Survival Mechanism

Because mitochondrial permeabilization is the key rate limiting step in apoptosis, one obvious mechanism by which mitophagy might allow cancer cells to evade death is by removing mitochondria that could potentially be permeabilized to induce cell death. Degradation of dysfunctional mitochondria has been reported to promote cell survival by preventing the production or release of toxic byproducts such as ROS and cytochrome c to cause apoptosis (Galluzzi et al., 2006, Colell et al., 2007). Consistent with this, mitophagy has been reported to cause cancer cell resistance to radiation (Wang et al., 2016), microtubule inhibitors (Wei et al., 2018), doxorubicin (Zhang et al., 2012, Yan et al., 2017), cisplatin, and etoposide (Villa et al., 2017, Yao et al., 2019), supporting a pro-survival role of mitophagy in cancer cells in response to chemo/radiotherapy (Abdrakhmanov et al., 2019). However, the mechanisms at play in this survival response may be complicated. Cancer cells are primed to undergo apoptosis because they more easily undergo mitochondrial permeabilization compared to normal cells. Indeed, this is the basis for the therapeutic window that allows many cancer therapies to work (Sarosiek et al., 2013). Autophagic degradation of specific proteins can regulate the efficacy of and sensitivity to anti-cancer drugs by altering the ease with which mitochondrial permeabilization can take place (Fitzwalter et al., 2018), potentially setting the apoptotic threshold and thus the therapeutic window for cancer drugs (Tompkins and Thorburn, 2019). This mechanism, though resulting in a similar effect of increased drug sensitivity when autophagy is blocked, is independent of mitophagy itself. This emphasizes the importance of the caveat described above– it is critical to carry out detailed mechanistic studies to determine if the degradation of the organelle itself is responsible for the biological effect. Indeed in the examples cited above where mitophagy has been linked to cancer cell drug resistance, Villa et al. (2017) provided evidence that their effects were mediated by a specific mitophagy ACR, but the others involved more general inhibition of autophagy. Blocking mitophagy could nevertheless be a promising anti-cancer therapeutic strategy to reduce resistance to other anti-cancer treatments (Drake et al., 2017). Moreover, autophagy and mitophagy are involved in dampening the anti-tumor immune response through the clearance of damage-associated molecular patterns (DAMPs), such as mitochondrial DNA (mtDNA) (Nakahira et al., 2011, Saitoh and Akira, 2016). Upon release from a damaged cell, pattern recognition receptors (PRRs) recognize the mtDNA, or other DAMPS, and initiate signaling that will upregulate pro-inflammatory interferons or cytokines or trigger the NLRP3 inflammasome to activate the innate immune system (Riley and Tait, 2020). More recent studies suggest that blocking mitophagy may enhance immunotherapy by allowing accumulation of cytosolic mitochondrial DNA after radiation, which promotes an enhanced Interferon-mediated tumor immune response that can make immune checkpoint inhibitors more effective (Yamazaki et al., 2020).

Accumulating evidence indicates that autophagy supports the metabolic plasticity of cancer cells by providing virtually all essential components of carbon metabolism through the degradation of carbohydrates, proteins, lipids, and nucleotides (Poillet-Perez and White, 2019). KRAS mutations, which frequently occur in several types of cancers, upregulate glycolysis and overwork the mitochondria (Weinberg et al., 2010), creating a need for increased mitochondrial turnover. In such tumors, increased autophagy or mitophagy has been suggested to support aggressive tumor growth (Kim et al., 2011, Rao et al., 2014). It has been well-documented that autophagy-deficient tumors have dysfunctional mitochondria that exhibit altered morphology, ineffective fatty acid oxidation, reduced carbon flux through the Krebs cycle, and lipid accumulation, all of which can act as tumor suppressors (Guo et al., 2011, Strohecker et al., 2013). Given that many tumors exhibit the Warburg effect (Warburg, 1956, Warburg et al., 1927), hyperactivation of mitophagy could solidify their glycolytic addiction by diverting the flux of metabolites away from the mitochondria (Narendra et al., 2008). Further evidence of a pro-tumorigenic role of mitophagy can be seen in oncocytomas, which in mouse models can arise from adenomas with defective autophagy (Guo et al., 2013). In this context, a reduction in autophagy flux may cause accumulation of dysfunctional mitochondria, resulting in switching to the less aggressive oncocytoma from more aggressive adenomas. In human oncocytomas, which also tend to be less aggressive, autophagy is also compromised (Joshi et al., 2015). Therefore, the role of mitophagy in the context of metabolism appears to be primarily pro-tumorigenic through maintaining healthy mitochondria (Figure 1).

Tumor Suppressive Roles of Mitophagy

Mitophagy is also associated with tumor suppression due to reduced ROS levels or energy production, which negatively impacts tumor metastasis and prevents genomic instability (Figure 1). Dysfunctional mitochondria and increased mitochondrial ROS can promote tumorigenesis, cancer progression, metastasis, and drug resistance through DNA, lipid, and protein oxidation (Weinberg et al., 2010, Chourasia et al., 2015, Yan et al., 2017). In a mouse model of breast cancer, depletion of the mitophagy ACR BNIP3 reduced mitophagy leading to increased mitochondrial ROS levels. This contributed to increased Hypoxia-Inducible Factor 1 alpha (HIF1α) stabilization, eventually promoting the Warburg effect and, subsequently, tumor progression (Chourasia et al., 2015). The loss of Parkin in many cancers has been interpreted as suggesting a tumor suppressive function for mitophagy (Chang et al., 2017, Zhang et al., 2020). Indeed, Parkin or PINK1 deletion in mice leads to the spontaneous development of hepatocellular carcinoma (Fujiwara et al., 2008) and increases KRAS-driven pancreatic tumorigenesis (Li et al., 2018). The accumulation of defective mitochondria induced by Parkin deficiency decreases mitochondrial oxidative phosphorylation (OXPHOS), increases ROS production, and increases glycolysis, therefore possibly contributing to the Warburg effect and consequently increasing tumorigenesis (Zhang et al., 2011). Additionally, both Parkin- and PINK1-null mice display an inflammatory phenotype due to Simulator of Interferon Genes (STING)-mediated interferon response (Sliter et al., 2018), suggesting that mitophagy may play a role in tumor suppression through prevention of tumor-promoting inflammation. However, as noted above, because numerous alternative mechanisms can lead to mitophagy in the absence of Parkin or PINK1, it is presently unclear whether the tumor suppressive functions of these molecules are due solely to reduced mitophagy.

Lysophagy

The lysosome is an acidic, membrane-bound organelle containing hydrolytic enzymes such as cathepsins, which are the main executioners of protein degradation in autophagy. During transformation, cancer cells alter their lysosomal compartment, including its size, localization, cathepsin expression, and activity (Gocheva et al., 2006, Nishimura et al., 1998). These changes in lysosomes often correlates with invasive growth, angiogenesis, and drug resistance (Fehrenbacher et al., 2004). Macropinocytosis, which relies on the lysosome for degradation of extracellular material, is also elevated in nutrient-deprived cancers, including RAS-driven bladder or pancreatic cancers (Commisso et al., 2013, Kamphorst et al., 2015, Sukhai et al., 2013) and tumor cells with WNT pathway activation (Tejeda-Munoz et al., 2019). Lysosomes also play a vital role in cancer drug resistance by sequestering drugs, which can prevent them from reaching their intended target molecule and provide a mechanism for removal of drugs (Zhitomirsky and Assaraf, 2016). Conversely, the lysosome can become a dangerous organelle when the lysosomal membrane is permeabilized (LMP), allowing leaking of cathepsins and other molecules into the cytoplasm (Aits and Jaattela, 2013) (Figure 2). This lysosomal cell death is of interest in cancer therapy because it may help avoid some of the oncogenic effects that can arise when apoptosis is the main death mechanism for tumor cells (Cao and Tait, 2018).

Figure 2. Lysophagy Promotes Cell Survival by Removal of Toxic Permeabilized Lysosomes.

A. Functional lysosomes act as the cell’s recycling system to degrade proteins and other macromolecules. When the lysosome becomes permeabilized by lysosome membrane permeabilization agents or anticancer agents such as paclitaxel, cathepsins and ROS may be released from this organelle, which often promotes cell death. B. Permeabilization of the lysosome is first detected by Galectin proteins. The cell can either repair the membrane damage through increased HSP70 expression to prevent large scale lysosome membrane permeabilization and cathepsin release or target the lysosome for lysophagy. UBE2QL1 promotes K48 ubiquitin chain elongation on lysosomal membrane proteins, which then attracts p97. Part of the p97 deubiquitation complex is the deubiquitinase YOD1, which removes the K48 ubiquitin chains. This promotes K63 ubiquitination through an unknown mechanism, allowing for p62 interaction and engulfment of the lysosome.

Cells degrade damaged or permeabilized lysosomes through lysophagy (Hung et al., 2013, Maejima et al., 2013). Once the lysosome has become permeabilized, several sensors, galectin (Gal)-1, −3, −8 and −9, bind exposed β-galactosides at the inner lysosomal membrane and elicit downstream signaling events (Thurston et al., 2012). Coordinated actions of galectin binding proteins are then thought to elicit lysophagy. Once galectins sense the damaged organelle, Ubiquitin Conjugating Enzyme E2 Q Family Like 1 (UBE2QL1) tags lysosomal membrane proteins with ubiquitin on K48 residues (Koerver et al., 2019, Kravic et al., 2020). After UBE2QL1-mediated protein ubiquitination, the ubiquitin-directed AAA-ATPase p97 is recruited to damaged lysosomes (Papadopoulos et al., 2017, Seczynska and Dikic, 2017, Tanaka et al., 2010). The K48-linked ubiquitin chains are subsequently removed by p97 cofactor and deubiquitinating enzyme YOD1, which then, through an unknown mechanism, promotes K63 ubiquitination, recruitment of p62, and association of the LC3-decorated phagophore (Papadopoulos et al., 2017, Seczynska and Dikic, 2017). It is important to note that different galectins have different roles within this process. For example, Gal3 recruits the tripartite motif (TRIM) protein TRIM16, which serves as a platform for autophagic initiation factors that induce phagophore formation (Chauhan et al., 2016). Meanwhile, Gal8 binds the ARC NDP52, which recruits LC3-positive phagophores that mediate lysophagy (Thurston et al., 2012) (Figure 2).

Because functional lysosomes are critical for cancer cell survival and play a vital role in drug resistance, there is interest in killing cancer cells through the induction of LMP. Several agents that can promote LMP and can also induce cell death include ROS, sphingosine, lysomotrophic toxins, or vacuolar H⁺-ATPase inhibitors (Aits and Jäättelä, 2013; Maejima et al., 2013). It has also been noted that many anti-cancer agents whose mechanism of action is usually considered to be through different cellular targets, such as TNFα, paclitaxel, vincristine, or siramesine, can also induce LMP (Werneburg et al., 2007, Broker et al., 2004, Groth-Pedersen et al., 2007, Erdal et al., 2005, Ostenfeld et al., 2005, Groth-Pedersen and Jaattela, 2013, Settembre et al., 2013). For such agents, lysophagy may prevent the cell from dying by removing these organelles therefore leading to drug resistance. There are currently no known specific inhibitors of lysophagy that might be used to test feasibility of this strategy. It is important to note that autophagy inhibitors that are used clinically, such as hydroxychloroquine and related molecules such as Lys05 and DQ661, operate through inhibition of lysosome function, specifically through preventing lysosomal palmitoyl protein thioesterase 1 (PPT1)-mediated removal of fatty acids on proteins (Rebecca et al., 2019). Combination therapy of LMP-inducing agents with autophagy/lysosomal inhibition agents such as these may also provide a strategy for cancer therapy.

ER-phagy (Reticulophagy)

The endoplasmic reticulum (ER) has multiple functions, including protein folding, processing, and transport, lipid and steroid synthesis, calcium storage and homeostasis, and detoxification. When the cell demands a high amount of protein synthesis and accumulates misfolded proteins, as often occurs in cancer, the ER becomes stressed, which can trigger cell death (Friedman and Voeltz, 2011, Nixon-Abell et al., 2016) (Figure 3). In addition to constitutive recycling, the ER must undergo more active turnover under certain stress conditions to achieve optimal quality of individual proteins and the entire organelle. A well-known mechanism to combat this event is the unfolded protein response (UPR), which halts protein translation and upregulates various chaperones to aid in correctly organizing misfolded proteins. Another mechanism to relieve ER stress is by the autophagic targeting and degradation of the ER, termed ER-phagy. ER-phagy is thought to aid in ER remodeling to minimize ER stress. However, while toxicities are often attributed to the loss of autophagy following ER stress, it is often unclear whether such effects are due to loss of ER-phagy itself or if other functions of autophagy contribute to the survival of the ER stressed cell. For example, tissue-specific deletion of general autophagy proteins, such as Atg5 in terminally differentiated B-lymphocytes (Pengo et al., 2013) or Atg5 or Atg7 in pancreatic acinar cells, drives excessive ER expansion, which may contribute to cell death (Antonucci et al., 2015), but in such studies, general autophagy is also compromised. To more specifically test if ER-phagy is responsible, one could try to selectively block only this kind of autophagy. However, this is difficult to do because, as with mitophagy, multiple ACRs mediate selective targeting of the ER, including Family With Sequence Similarity 134 Member B (FAM134B), Reticulons (RTN1–4), Alastin-3 (ATL3), SEC62, CCPG1, TEX264, and NDP52/CALCOCO2 (Nthiga et al., 2020). Each receptor has a specific target and function in proteostasis and/or ER fragmentation. FAM134B and RTN1–4 drive ER tubulation, sheet edge curvature, and sheet fenestration to promote the remodeling and scission of ER tubules through its reticulon domain (Schroeder et al., 2019, Voeltz et al., 2006). SEC62 mediates macroER-phagy during cellular recovery from an acute UPR response and delivers portions of ER into autophagosomes following ER stress in a process termed ‘recovER-phagy’ (Fumagalli et al., 2016, Loi et al., 2019). CCPG1 is a type II, single-pass transmembrane protein that delivers portions of the ER to autophagosomes upon ER-stress (Kostenko et al., 2006), typically in response to UPR, thus helping rid the cell of misfolded proteins or protein aggregates (Smith et al., 2018). TEX264 is a single pass, transmembrane receptor for nutrient starvation-induced ER-phagy that targets three-way ER junctions upon nutrient starvation (Chino et al., 2019, An et al., 2019). All adaptors contain an LIR motif, which can bind to the autophagosome to target the ER for degradation (Chen et al., 2019, Smith et al., 2018, Grumati et al., 2017, Khaminets et al., 2015, An et al., 2019, Fumagalli et al., 2016). Also, all adaptors besides ATL3 have long intrinsically disordered regions between the ER transmembrane domain and the LIR that cannot fold into a compact structure, which allows for ER-targeting when ribosomes are on the ER (Chino et al., 2019, Meszaros et al., 2007) (Figure 3).

Figure 3. ER Stress in Cancer Requires Increased ER Turnover.

A. A healthy ER is crucial for protein translation. However, ER stress occurs in cancer cells upon nutrient deprivation, hypoxia, and anticancer treatments, often due to excessive amounts of protein translation, protein aggregation, and UPR. B. ER-phagy has many known adaptors that promote the degradation of stressed or damaged ER to prevent cell death. SEC62, CCPG1, and FAM134B all function in promoting the degradation of ER sheets, while RTN, ALT3, and TEX364 are utilized for ER tubule degradation. TEX264 is unique in its targeting of three-way junctions. Receptors all have LIR interacting domains allowing for binding to LC3, and can be activated upon nutrient deprivation or UPR.

There is evidence suggesting specific pro-tumor roles of ER-phagy in cancer. Overexpression of FAM134B can transform noncancerous cells (Tang et al., 2007) and mutations and alterations in FAM134B expression have been observed in various malignancies (Islam et al., 2018, Kasem et al., 2014b, Kasem et al., 2014a, Kasem et al., 2014c). Glioma cells require FAM134B-driven ER-phagy to survive proteotoxic stress (Peng et al., 2018), suggesting the potential for this molecule as a therapeutic target in cancer. Additionally, a number of cancers have amplifications in the SEC62 gene, including lung adenocarcinomas, prostate, thyroid, and head and neck squamous cell carcinoma (Greiner et al., 2011, Jung et al., 2006, Linxweiler et al., 2016, Linxweiler et al., 2012, Wemmert et al., 2016, Weng et al., 2012). These studies suggest that SEC62 plays a role in tumor cell invasion and migration in addition to aiding in resistance to anti-cancer ER stress (Peng et al., 2018).

Due to the critical role of ER stress in cancer, promoting even greater amounts of ER stress could be a potential therapeutic strategy for killing tumors (Wang and Kaufman, 2014). The proteasome inhibitor bortezomib was the first FDA-approved drug for cancer therapy that functions as an inducer of ER stress (Takenokuchi et al., 2015). One could also target specific components of the protein-folding or UPR pathways, such as PERK, IRE1α, HSP90, GRP94, and BiP, thereby augmenting ER stress and sensitize cancer cells to cell death (Wang and Kaufman, 2014). Similarly, inhibition of HSP molecular chaperones leads to a disruption in protein processing and induces ER stress. Increased expression of pro-survival HSP90 and HSP70 has been observed in many types of cancers, and their inhibition has emerged as a promising anti-cancer strategy (Calderwood et al., 2006). Stress-inducible HSP70 was considerably upregulated as a pro-survival response after HSP90 inhibition in vitro, allowing for greater capacity for myeloma cells to fold damaged proteins (Davenport et al., 2007).

The sarcoendoplasmic reticulum Ca²⁺-ATPase (SERCA) is a major source of Ca²⁺ transport out of the ER. This release of stored Ca²⁺ is critical for many cellular processes such as proliferation, metabolism, muscle contraction, gene transcription, and cell death (Berridge et al., 2000). Another method of inducing ER stress to kill cancer cells is by depleting ER Ca²⁺ through SERCA inhibitor thapsigargin and its prodrugs G202 and 12ADT-Asp (Denmeade et al., 2012, Mahalingam et al., 2016). Similarly, because SEC62 relies on Ca²⁺ for its function, attempts have been made to treat cancers with Ca²⁺ sensor and effector protein Calmodulin (CaM) antagonists to prevent Ca²⁺ efflux into the ER lumen. Indeed, CaM antagonists trifluoperazine (TFP) and ophiobolin A inhibited cell migration and proliferation in cancerous cells with high SEC62 expression and sensitized cancer cells to thapsigargin-induced ER-stress (Greiner et al., 2011, Linxweiler et al., 2016, Linxweiler et al., 2012). Other studies have suggested that ER stress is an indirect response to anti-cancer agents such as cisplatin (Mandic et al., 2002), plant alkaloid ellipticine (Hagg et al., 2004), or doxorubicin (Jang et al., 2004). Furthermore, the glycosylation inhibitors tunicamycin and 2-deoxyglucose can induce ER stress through accumulation of misfolded proteins (Wu et al., 2018, Yu and Kim, 2010). However, because ER stress, either endogenous or treatment-induced, can upregulate ER-phagy (Zhang et al., 2019b, Stephani et al., 2020, Smith et al., 2018, Jiang et al., 2020, Loi et al., 2019), these molecules might need subsequent treatment with ER-phagy or autophagy inhibitors to prevent cancer cells from rescuing themselves from the toxicities of protein misfolding and aggregation. One suggested ER-phagy inhibitor, berberine, enhances expression of various ER stress markers as well as reduces ER localization to LC3 suggesting inhibition of ER-phagy (Zhang et al., 2019a). Interestingly, cancer cells that are dependent on autophagy, become vulnerable to proteasome inhibitors upon loss of autophagy, consistent with the idea that blockade of autophagy (and therefore ER-phagy) can increase sensitivity to agents that cause ER stress through different mechanisms (Towers et al., 2019, Towers and Thorburn, 2019).

Other Potentially Important Types of Selective Autophagy of Organelles in Cancer

Pexophagy

Another important player in cancer cell metabolism is the peroxisome, which plays a vital role in the oxidation of many types of molecules, notably fatty acids and amino acids, that can promote cell growth and survival (De Duve and Baudhuin, 1966, Poirier et al., 2006). These oxidation reactions generate hydrogen peroxide that can damage DNA and proteins, which is then detoxified by antioxidant enzyme catalase (Till et al., 2012, Fransen et al., 2012). There are two stimuli that promote pexophagy: nutrient starvation and ROS (Sargent et al., 2016, Zhang et al., 2015, Lee et al., 2018). Although few studies have shown the specific effects of peroxisomes in cancer, their increased activity could contribute to tumor growth through lipid oxidation (Carracedo et al., 2013, Santos and Schulze, 2012, Liu et al., 2017). Peroxisome proliferator-activated receptors (PPARs) are involved in regulating the expression of genes involved in peroxisome metabolism, such as Acetyl-CoA Oxidase 1 (ACOX1) (Hostetler et al., 2006). ACOX1 is upregulated to aid in the oxidation of hepatic fatty acids and results in an increase of H₂O₂ and subsequent tumor development (Misra and Reddy, 2014, Yeldandi et al., 2000). As such, high ACOX1 levels are associated with poor survival rate in HER2 positive breast cancers (Kim et al., 2015). Alpha-methylacyl-CoA Racemase (AMACR) is another possible pro-tumorigenic PPAR-regulated protein involved in alpha oxidation of molecules to prepare them for beta-oxidation. Elevation of AMACR is associated with poor prognosis in a variety of cancers, including prostate (Box et al., 2016, Lloyd et al., 2008), colon (Jiang et al., 2003, Chen et al., 2005), gastric (Jindal et al., 2016), breast (Witkiewicz et al., 2005), renal and hepatocellular (Went et al., 2006), and in myxofibrosarcomas (Li et al., 2014).

Because peroxisomes provide cancer cells with vital molecules for survival and their activity is increased in response to nutrient or oxidative stress, promoting pexophagy is a potential therapeutic strategy for cancer. While there are no current therapeutics, or even genetic or other experimental strategies, that can directly promote pexophagy in a specific manner, oral treatment with lipolytic agent 3,5-dimethylpyrazole or 5-carboxy-2-methylpyrazine 1-oxide (Acipimox) can reduce both mitochondrial and peroxisome beta-oxidation (Locci Cubeddu et al., 1985), and therefore could be a useful agent to treat cancer. Hypolipidemic drugs, phthalate esters (Yokota and Dariush Fahimi, 2009), or non-classical peroxisome proliferators, such as 4-phenylbutyrate (Sexton et al., 2010) might also indirectly induce pexophagy.

Lipophagy

Lipids are the major component of biological membranes and are also used for energy storage, production, and cellular signaling. Intracellularly, lipids are stored in lipid droplets when energy supplies are high. Degradation of lipid droplets can occur in response to starvation to help meet energy needs of the cell. Lipolysis and lipophagy (autophagic degradation of lipids by acidic lipases) are two pathways the cell uses to metabolize these stored fats (Singh et al., 2009, Zechner et al., 2017). Once lipid droplets are catabolized to free fatty acids, they are broken down through mitochondrial fatty acid β-oxidation pathway to meet the energy requirements of rapidly proliferating cells (Carracedo et al., 2013).

Lipophagy, like mitophagy and the other forms of selective autophagy or organelles, could potentially be both tumor promoting and tumor suppressive. Cancer cells often upregulate fatty acid synthesis and uptake, leading to increased accumulation of lipid droplets (Menard et al., 2016, Tirinato et al., 2015). Lipophagy provides cancer cells with lipid metabolites for synthesis of biomolecules under stressed or nutrient-deprived conditions and this may aid in survival (Gomez de Cedron and Ramirez de Molina, 2016). For example, lipophagy-mediated production of free fatty acids aids and subsequent ROS production promotes senescence in prostate cancer. Senescence is reversed through inhibition of Lysosomal acid lipase (LAL) (Panda et al., 2020). Studies have also linked reduced lipophagy, i.e. lipid droplet accumulation, with increased cancer aggressiveness (Nieva et al., 2012, Du et al., 2015, Kaini et al., 2012) and chemotherapy resistance (Rak et al., 2014). LAL deficiency results in an abnormal hematopoiesis leading to an abundance of immature myeloid-derived suppressor cells, which mediate suppression of immune surveillance and promote tumor evasion of host immunity to stimulate tumorigenesis and metastasis (Qu et al., 2009, Zhao et al., 2015). Conversely, expression of LAL and subsequent enhanced lipid metabolism has also been reported to reduce metastasis in lung and liver cancer (Zhao et al., 2016, Du et al., 2015). Another recent study reported that increased lipophagy mediates ER stress through free fatty acid accumulation to sensitize cancer cells to apoptosis (Mukhopadhyay et al., 2017). Involvement of lipophagy in lipid turnover makes it a crucial player in cancer survival and metastasis. Therefore, targeting this autophagic pathway could be another therapeutic strategy for a variety of cancers.

Concluding Remarks

As is the case with general autophagy, there is a complex relationship between organelle-targeted autophagy and cancer cell growth, survival, or progression. Although we have focused on selective autophagy of the mitochondria, ER, lysosomes, peroxisomes, and lipid droplets, other organelles and parts of organelles can also be targeted for autophagic turnover and may contribute to cancer-related phenotypes. While we typically think that autophagy of organelles is important in cancer through the removal of toxic and/or damaged molecules, there are situations, such as in mitophagy, where increased degradation of these organelles could be both advantageous and detrimental to the cell. However, it is difficult to be certain of many conclusions in the literature that have suggested a particular role for a certain type of organelle-specific autophagy because we lack effective methods to positively or negatively manipulate these processes without also affecting general autophagy. However, despite these limitations and complexities, it is very likely that selective turnover of organelles is critical in cancer cells.

If we were able to manipulate these forms of selective autophagy, it might be feasible to use that approach as a way to better treat or prevent cancer. However, many questions remain regarding organelle homeostasis in cancer. Can we understand the context-dependent functions of organelle homeostasis and design reliable and effective ways to selectively manipulate these processes in experimental models of cancer? An even bigger challenge is to find ways to design interventions that could realistically be applied therapeutically since this would require the development of drugs that can achieve these goals. These problems are difficult because of the nature of evolved biological systems. First, cancer cells adapt in response to selective pressure. Thus, if a cancer cell adapts to perform cellular functions in a different way, this may stymie efforts to interfere with the process that was thought to be critically important. For example, although we believe that mitophagy is critical in cancer growth, if cancer cells can maintain healthy mitochondria through other methods that do not involve autophagy at all, they may be able to circumvent even highly effective methods of mitophagy inhibition. Second, biology often uses different molecules to achieve the same purpose while conversely using the same machinery for different purposes. Thus, if we target one ACR that is critical for a particular type of organelle-phagy, another may take its place. Beyond their role in organelle homeostasis, ACRs are also vital to promoting cancer cell growth and survival. For example, p62 accumulation results in alteration in NFᴋB and NRF2 signaling, which promotes tumor growth through increased proteosome activation as well as enhanced oxidative stress and DNA damage response (Towers et al., 2019, Mathew et al., 2009, Wei et al., 2014). Another example is that accumulation of the autophagy receptorNBR1 enhances breast cancer metastatic growth through an incompletely understood autophagy-independent mechanism (Marsh et al., 2020). Disrupting organelle homeostasis via autophagy inhibition may therefore have unintended consequence, such as accumulation of ACRs, which may then promote tumor signaling via alternate methods. It is also unclear how organelle-specific autophagy pathways intersect with other organelle functions, such as serving as signaling hubs. Such problems cause practical problems in experimental interpretation. However, this complexity also opens up potential opportunities to refine and improve interventions. For example, by interfering with autophagy to alter organelle turnover, it may be possible to manipulate signaling events that take place at these organelles. By better understanding such mechanisms as they relate to cancer cell behaviors, it may be feasible to develop more rational approaches for therapy.