Principles and Current Strategies for Targeting Autophagy for Cancer Treatment

Autophagy is a lysosomal degradative pathway characterized by the formation of double-membrane autophagic vesicles (AV), also known as autophagosomes, which engulf portions of the cytosol, damaged organelles, protein aggregates, and bacteria. AVs are typically transported along microtubule tracks to a perinuclear location. The outer membrane of the AV subsequently fuses with the lysosome, resulting in degradation of the AV contents and inner membrane (Fig. 1; refs. 1, 2). Autophagy occurs at basal levels in virtually all cells, carrying out homeostatic functions such as protein and organelle turnover. Autophagy is upregulated when cells require intracellular nutrients and energy, such as during starvation and growth factor withdrawal or in the context of high bioenergetic demand. Additionally, autophagy is upregulated under other stress conditions, such as when there is a need to clear aggregated proteins, damaged organelles, or intracellular pathogens. A number of signaling pathways intersect with the autophagy system. This intersection allows a tightly regulated and dynamic autophagic response to environmental perturbations.

The anatomy, physiology, and molecular machinery of autophagy are highly conserved among eukaryotic cells. They include distinct steps for AV production and turnover, including (1) initiation, (2) nucleation of, and (3) maturation of AVs, and (4) fusion and degradation of AV contents in lysosomes (Fig. 1; refs. 3, 4). The ULK1 (ATG1) kinase complex consisting of ULK1 (and/or possibly ULK2), Atg13, and Atg17 integrates stress signals from mTOR complex 1 (mTORC1) and controls the initiation of autophagy (5, 6). Once mTORC1 kinase activity is inhibited, the cytoplasmic autophagy machinery described below is recruited onto phospholipid membranes derived from the endoplasmic reticulum (ER; ref. 7) and trans-golgi network (8). More recently, the mitochondrial outer membrane (9) and plasma membrane (10) were identified as additional important sources of phosphatidylethenolamine (PE)–rich membranes, which are characteristic of AVs.

AV formation begins with the generation of phosphoinositide signals on the surface of source membranes by multiprotein complexes that include the class III phosphoinositide 3-kinase (PI3K) Vps34 and Beclin1 (11). The cytoplasmic ubiqituin-like protein Atg8 (LC3) is conjugated to PE on these membranes, which identifies them as incipient AVs. Lipidation of LC3 occurs by a ubiquitin-like protein (UBL) conjugation cascade involving an E1-like enzyme (ATG7) and E2-like enzyme (ATG3), following cleavage by a cysteine protease (ATG4; ref. 12). Once LC3 is integrated into the bilayer, it recruits cargo adaptor proteins (also known as autophagy receptors), such as p62, Nbr1, or NIX. These proteins, in turn, recruit cargo from the cytoplasm [i.e., ubiquinated protein aggregates in the case of p62 (13) and damaged organelles in the case of NIX (14)] to promote AV closure (15). AVs are then delivered to lysosomes in which their luminal and inner membrane constituents are broken down by lysosomal hydrolases. Lysosomal permeases then release the degradation products into the cytosol for reuse (16). AV components not exposed to lysosomal hydrolases are recycled via a system involving multiple components of the outer membrane ATG9, ATG2, ATG18, and ATG21 (17). Alternatively, autophagosomes may also fuse with the plasma membrane and release their contents (18).

Autophagy initiation is associated with downregulation of mTORC1 activity. Activated mTORC1 inhibits autophagy by causing hyperphosphorylation of ATG13, reducing its interaction with ATG1/ULK1, and by controlling phosphorylation of autophagy effectors such as the Vps34-Beclin1 complex. Proteomic studies investigating how inhibition of the mTORC1 pathway affects the global features of autophagy control showed no large-scale changes in core conjugation, lipid kinase, and recycling complexes. This finding implies that post-translational modifications may be involved in AV accumulation when the autophagy pathway is activated (4) and may be a potential means to control autophagy.

AV nucleation represents a second major autophagy control point, involving Vps34 and interacting partners Beclin1 and p150 (19). Drugs that interfere with recruitment of Vps34 to membranes, including wortmannin and 3-methyladenine, are powerful (although nonspecific) proximal inhibitors of autophagy. Direct inhibitors of Vps34 and drugs that sequester or free up Beclin1 may also be deployed for autophagy inhibition (20). Multiple PI3K/Beclin1 complexes may be involved in mammalian autophagy (11). For example, PI3K/Beclin1 complexed with UVRAG and Bif-1 (21) can activate autophagy on membranes, whereas PI3K/Beclin1 complexed to Rubicon (22) plays an inhibitory role in membrane trafficking of AV to lysosomes. Therefore, care must be taken in interpreting results when Vps34 or Beclin1 are pharmacologically or genetically suppressed in autophagy studies.

UBL-containing ATG8 family proteins are central coordinators of AV maturation (4) and represent a third autophagy control point. LC3, the most widely studied ATG8 family member, is cleaved by ATG4 and conjugated to PE by an ATG7- and ATG3-dependent activation and transfer cascade. In this manner, LC3 is incorporated into the membrane where it orchestrates AV growth and cargo recruitment. Cargo recruitment involves a conserved surface on LC3 (23), interacting with motifs in cargo-binding proteins. Mutations in these motifs reduce the binding of cargo adaptor proteins, such as p62, Nbr1, and Nix, to ATG8 proteins and disrupt transfer of AV cargo to lysosomes (13, 24, 25). Nbr1 and p62 contain ubiquitin-binding domains in addition to the motif that interacts with LC3. This characteristic allows these adaptor proteins to bind both ubiquinated cargo and LC3, enabling tight sequestration of ubiquinated cargo by surrounding LC3-containing membranes, with little cytosolic content included (26). The cargo adaptor protein NIX similarly recruits mitochondria to LC3-containing membranes (25). ATG8 family members, such as LC3, dictate cargo binding through cargo adaptor interaction, thereby determining the type of cargo sequestered during autophagy.

Delivery and degradation of AV contents represents a fourth autophagy control point. Because AVs and lysosomes move along microtubules, drugs that disrupt microtubules, such as nocodazole, colchicines, taxanes, and vinca alkaloids, inhibit AV fusion with lysosomes, resulting in AV accumulation. Rab GTPases likely play a role in vesicle maturation and fusion with lysosomes (27). Lysosomes are acidic organelles, with their digesting hydrolases dependent on low pH. Consequently, agents such as bafilomycin or chloroquine derivatives, which disrupt the vacuolar H⁺ ATPase responsible for acidifying lysosomes, block autophagy in its final step, resulting in the accumulation of AVs (28–30).

Types of autophagy vary depending on the stimulus and requirement for substrate degradation (4, 31, 32). Inhibition of mTOR, for example, decreases association of p62 with LC3-containing membranes (4). This type of autophagy, occurring when food supply is limiting, is likely different from autophagy activated when cells are stressed from buildup of damaged organelles and protein aggregates (which uses p62). Differences in starvation- versus stress-induced autophagy are also manifested by the site of AV origin and by the type of cargo sequestered. For instance, starvation-induced autophagy is characterized by AV membrane budding off of the mitochondrial outer membrane, and once formed, starvation-induced AVs are more likely to contain free, soluble cytosol (9).

The breadth of autophagy's crucial roles in survival, adaptability, and overall physiology suggests multiple subtypes of autophagy that are location and cargo specific within the cell, and tissue specific within the organism. Thus, therapeutic strategies for inhibiting or inducing autophagy need to be tailored toward stress- versus starvation-induced autophagy. Further analysis of the physiologic conditions under which different subtypes of autophagy are used, and further clarification of which autophagy pathway is targeted by specific inducers or inhibitors will guide development of autophagy modulators in cancer therapeutics.

Comparison of normal and autophagy-defective mice and cells has illuminated the role of autophagy in suppression of tumor development. Mice with autophagy defects accumulate ubiquitinated keratins, the autophagy cargo adaptor p62, and abnormal mitochondria (33–35). High levels of p62 in many tissues and tumors and phospho-keratin 8 in mammary tissues and tumors are potential biomarkers for autophagy defects (34, 36). These damaged cellular components accumulate, often in large aggregates or inclusions, and are linked to reactive oxygen species (ROS) production, activation of the DNA damage response, cell damage, and death that can lead to a chronic inflammatory state (35, 37). Progressive cell and tissue damage due to failure of autophagy-mediated cellular garbage disposal provokes degenerative and inflammatory diseases (38, 39) and may contribute to cancer. Chronic tissue damage and inflammation is associated with DNA-damaging ROS production, contributing to mutations that can initiate cancer and promote tumor progression (40). Mice with allelic loss of the essential autophagy gene beclin1 display defective autophagy, altered protein homeostasis (accumulation of ubiquitinated proteins and p62), and gross morphologic tissue damage that is particularly striking in liver where there is also an accelerated incidence of hepatocellular carcinoma (Fig. 2A; refs. 33–35). These findings suggest that autophagy stimulators may prevent both degenerative diseases and cancers arising from chronic tissue damage and inflammation, such as hepatocellular carcinomas (41, 42).

Although autophagy can suppress tumor development, it clearly plays a role in promoting the survival of tumor cells within the tumor microenvironment. Although autophagy induction can be associated with cancer cell death, this may be due to a futile attempt of the cancer cells to survive through autophagy, also known as cell death with autophagic features (43). This finding underscores the importance of interrogating the functional role of autophagy when autophagosomes are present. In some cases, knockdown of essential autophagy genes by RNA interference (RNAi) enhances survival. Whether this increased survival is due to the absence of autophagic cell death and prevention of overactivation of autophagy and cell death by fatal self-consumption or another unknown mechanism is not yet known. In other settings, autophagic cell death was limited to in vitro conditions and not manifested in vivo. The most prevailing and convincing evidence, however, is that in vivo, autophagy is induced by cellular stress, including nutrient, growth factor, and oxygen deprivation, and functions to maintain survival of normal cells, mice, and also tumor cells (37, 44, 45). When in vitro models incorporate stresses commonly encountered in vivo, autophagy's contribution to cell survival becomes clearer. For example, autophagy-defective tumor cells undergoing metabolic stress (ischemia) showed impaired survival in comparison with autophagy-proficient cells (Fig. 2B). Furthermore, autophagy localizes to hypoxic regions within tumors, and genetic ablation of autophagy promotes the selective death of those metabolically stressed cells (41). The mechanism by which autophagy enables survival of normal or tumor cells in stress is not known. In oxidative stress, the clearance of damaged proteins and organelles, particularly mitochondria, may limit cellular damage and death through ROS production. When nutrients are limiting, autophagy may promote viability by maintaining cellular metabolism through intracellular recycling (46).

Regardless of how autophagy increases survival in stress, concurrent inhibition of autophagy may improve outcomes in cancer therapy. Cytotoxic cancer therapeutics induce autophagy, most likely by causing damage to DNA, cellular proteins, and organelles. Inhibition of autophagy in preclinical models improves the response of tumors to alkylating agents, suggesting that autophagy promotes survival (47). Targeted cancer therapies also stimulate autophagy, often by mimicking signaling of starvation or factor deprivation. Inhibitors of mTOR, in particular, are potent activators of autophagy, yet the functional consequences of this activation in cancer therapy are not fully understood (48). An important future direction is to establish the functional consequence of autophagy stimulation by cancer therapeutics. Three additional areas of intense focus critical to understanding the role of autophagy in cancer are: (1) the role of commonly activated oncogenes and inactivated tumor suppressor genes in determining autophagy levels and function within the tumor cell; (2) the role of autophagy activation by targeted therapies; (3) network interactions among the proteasome, the ER stress response, and autophagy; and (4) extracellular control of autophagy by the immune system, tumor stroma, and vasculature.

As outlined in the sections below, knowledge gained from these studies will guide more sophisticated approaches to the therapeutic manipulation of autophagy, with the common aim of limiting the development of cancer and reducing mortality in patients presenting with overt malignancies. Furthermore, identifying the “autophagic switch” that mediates the transition from suppressed autophagy, important early in neoplasia, to enhanced autophagy, contributing to malignant progression, is critical to understanding this complicated process and to developing rational therapeutic strategies.

In a recently fed organism, growth factors bind to their cognate receptors [typically a receptor tyrosine kinase (RTK)] and signal through class I PI3Ks, leading to phosphorylation of the pleckstrin homology domain kinase Akt, subsequently activating mTOR. The serine-threonine kinase mTOR plays a prominent role in regulating growth and proliferation in both normal and tumor cells (Fig. 3). It integrates signals from key environmental sensors such as the AMP-activated protein kinase (AMPK; cellular energy status), Rag GTPases (amino acid availability), regulated in development and DNA damage 1 (REDD1; oxygen availability), and p53 (DNA damage; ref. 49); and it modulates the rate of translation of proteins required for growth, proliferation, and metastases. Activated mTOR thus engages anabolic pathways, while in parallel, downregulating catabolic pathways including autophagy. The axis linking growth factor receptors, PI3K, Akt, and mTOR is tightly regulated in cells.

As central integrators of nutrient and growth factor signaling, PI3K, Akt, and mTOR represent critical nodes regulating cell growth, proliferation, and survival. It is, therefore, not surprising that inappropriate activation of this signaling cascade is commonly found in cancer. Mutational activation of PI3K occurs commonly in cancers, whereas mutational activation of Akt is relatively infrequent. Amplification or epigenetic activation of RTKs such as epidermal growth factor receptor (EGFR) and mutational or epigenetic inactivation of negative regulators of this pathway, including the phosphatase and tensin homolog (PTEN), are other common aberrations across malignancies. Thus, the vast majority of cancers have some degree of aberrant activation of the PI3K-Akt-mTOR signaling axis (50).

Activation of mTOR represents the most downstream target in this signaling pathway, suggesting mTOR inhibition as a critical strategy for cancer therapy. When these findings were translated into clinical trials, however, mTOR inhibitors failed to produce clinical benefit in many cancers. Activation of growth factor receptors ultimately leads to activation of mTOR and increased anabolic functions. This increase is short-lived, however, as the target S6K in turn feeds back to phosphorylate and drive degradation of insulin receptor substrate 1, repressing upstream signaling through PI3K (49). Thus, inhibition of mTOR leads to activation of PI3K and Akt. Because Akt has more than 150 downstream targets, the net effect is to inhibit a single target, at the expense of activating a multitude of additional targets.

One way to circumvent this problem is to use a dual inhibitor of PI3K and mTOR, a single molecule that affects mTOR inhibition, while simultaneously blocking feedback activation of Akt. In preclinical models of glioblastoma, treatment with dual PI3K/mTOR inhibitors is associated with a cytostatic rather than cytotoxic response (51). How can a drug that blocks signaling through three key survival kinases, PI3K, Akt, and mTOR, fail to affect survival in cancer?

Recent studies have shown that dual inhibitors of PI3K/mTOR activate autophagy, that this activation is regulated by both mTOR complexes (mTORC1 and mTORC2), and that blockade of autophagy in early or late stages can cooperate with dual inhibitors of PI3K/mTOR to promote cell death (52). This cell death is prominent even in glioma cells mutant for PTEN. In contrast, in the glioma models, inhibitors of mTORC1 did not cooperate with inhibitors of autophagy to induce cell death, possibly because of the S6K–insulin receptor substrate 1 (IRS1)–PI3K feedback loop described above. Whereas dual inhibitors of PI3K/mTOR (including mTORC1 and mTORC2) induce autophagy as a central survival signal, selective inhibitors of mTORC1 activate both autophagy and Akt as separate survival signals. Effecting cell death in preclinical models of PTEN mutant glioma (in vitro and in vivo) thus requires blockade of three targets: Akt, mTOR, and autophagy (51). Further studies are underway to determine if this finding is specific to glioma or can be generalized to all cancers with activated PI3K/AKt/mTOR signaling.

Future directions include elucidating mechanisms through which autophagy blockade cooperates with dual inhibitors of PI3K/mTOR to induce cell death, identifying key Akt targets that block this effect when Akt is activated, and identifying new and more selective autophagy inhibitors that circumvent toxicities associated with chloroquine derivatives. Although PTEN mutation is generally associated with therapeutic resistance in glioma and other cancers (53), dual inhibitors of PI3K/mTOR when combined with chloroquine readily induce apoptosis of PTEN mutant glioma in vivo. This combination of agents could be tested in the near future in patients with this generally lethal tumor.

The autophagy-lysosome system and the ubiquitin-proteasome system (UPS) constitute the two major intracellular degradation systems. Although UPS mainly targets short-lived proteins and soluble misfolded proteins, autophagy is particularly important for the turnover of long-lived proteins, aggregated and misfolded proteins, and organelles (39, 54–56).

A number of studies indicate functional connections between these two degradation systems (57, 58). Inhibition of UPS compensatively activates autophagy. Notably, ER stress plays a critical role in the cross-talk between the two systems. The ER is the major site for processing protein conformation. Misfolded proteins are normally exported out of the ER lumen and degraded by the proteasome via the ER-associated degradation pathway (ERAD; ref. 59). Autophagy is another important mitigating mechanism that clears misfolded proteins in response to ER stress (57, 60). Drug-induced ER stress can also induce autophagy (61–65). Interestingly, ER-associated autophagy (ERAA), like ERAD, can also be regulated by the unfolded protein response, which is orchestrated by transmembrane-cytoplasmic kinase pathways such as protein kinase-like ER kinase (PERK)–eukaryotic initiation factor 2α (eIF2α) and inositol-requiring enzyme-1 (IRE-1)–Jun-N-terminal kinase (JNK; Fig. 3; refs. 57, 58, 60–63, 65, 66). ERAA is particularly important if ER stress is caused by proteasome inhibition, resulting in loss of ERAD's critical degradation machinery.

Because uncompensated ER stress can lead to cell death (59, 67), the compensatory activation of autophagy provides a prosurvival mechanism (64). This notion is particularly relevant for cancer therapy with proteasome inhibitors, such as bortezomib. Indeed, genetic ablation of autophagy sensitizes tumor cells to proteasome inhibitors (35), with combined use of bortezomib and chloroquine increasing tumor cell death in vitro and in vivo (68). Tumor cells are more sensitive to this combination than normal cells (68), highlighting autophagy's role in the survival of cancer cells, perhaps reflecting a higher metabolic rate and stress status in cancer relative to normal cells. Bortezomib is approved by the U.S. Food and Drug Administration (FDA) for the treatment of multiple myeloma, a tumor type likely prone to elevated ER stress because of abundant synthesis of immunoglobulin, and autophagy inhibitors may be similarly effective in this setting. Thus, understanding of the relationship of the proteasome and autophagy via ER stress is not only important for the development of novel cancer therapies, but also for understanding the nature and causes of increased cellular stress in cancer cells and resulting adaptive responses.

Cancer in adults (69), but not in children (70), arises in the setting of chronic inflammation. The tumor microenvironment is characterized by a disordered state associated with hypoxia, glycolysis, perpetual autophagy, and resultant necrosis under conditions of heightened stress. A good example of how intimately linked tumor cell metabolism, autophagy, and immune tolerance can be is highlighted by recent studies focused on the pleiomorphic functions of the highly conserved nuclear protein high mobility group B1 (HMGB1). Autophagic stimuli promote cytosolic and mitochondrial translocation and extracellular release of HMGB1 (Fig. 4; ref. 71). As a cytosolic factor, HMGB1 itself promotes autophagy, enhances ATP production, and limits apoptosis (72). Extracellular HMGB1 serves as a damage-associated molecular pattern molecule (DAMP), which interacts with the receptor for advanced glycation end products (RAGE; ref. 73) and toll-like receptors (74) to recruit inflammatory cells to the site of damage. Thus HMGB1 represents one of likely many molecules that critically link cellular metabolism, cell death decisions, and immunity.

Recently, 3 randomized studies have shown survival benefits for immunotherapy in refractory cancers (75–77). These limited successes come on the heels of decades of failures. Studies of autophagy in tumor tissue and immune cells suggest that cancer patients are suffering from a systemic autophagic syndrome in which autophagy is pathologically increased within the cancer cell and suppressed in the immune cells. Adoptive transfer of T cells, dendritic cell (DC) vaccines, administration of antibodies, or administration of human recombinant cytokines, such as interleukin 2 (IL-2), only hold promise for immunotherapy if the imbalance between host and tumor autophagic response can be ameliorated. DC vaccines involve isolation of the patient's antigen presenting cells (APC), followed by a procedure of ex vivo gene therapy or incubation with targeted tumor associated and specific antigen (TAA, TSA), and subsequent reintroduction of the matured DCs so that they may mediate a highly specific antitumor immune response facilitated by DC-activated CD8+ T cells (78–81). Within professional APCs (Fig. 3), antigen processing and delivery to MHC class I and class II molecules is directed by the proteasome and autophagy. Autophagic cargo, which can be extruded into the extracellular matrix from tumor cells, should be superior sources from which DCs can derive antigen for T-cell priming (78, 82). Thus, systemic induction of autophagy early in the course of adaptive immunity may prevent the emergence of immune tolerance, and ex vivo induction of autophagy in the presence of antigen may improve the efficacy of cellular immunotherapies.

Autophagy inhibition may augment the cytotoxicity of effector T cells and natural killer (NK) cells once they have been activated to lyse the tumor, similar to the notion that autophagy limits the effectiveness of chemo- and radiation therapy. Currently, autophagy inhibition with hydroxychloroquine (HCQ) in combination with IL-2 is being tested in a murine tumor model and is poised to be rapidly translated into a multiinstitution clinical trial. Future approaches may include combining ex vivo induction of autophagy in DCs and systemic autophagy inhibition, delivered at the time of adjunctive treatment and designed to stimulate cytotoxic effectors. This approach may facilitate improved antitumor immunity with DC delivery and enhanced antitumor efficacy of the activated immune system (Fig. 4).

Autophagy inhibition augments the efficacy of anticancer agents in a variety of tumor histologies in multiple preclinical models (37, 47, 68, 83–85). On the basis of reports that effective autophagy inhibition can be achieved in vivo with the antimalarial drug chloroquine (CQ; ref. 47), clinical trials for patients with refractory malignancies were undertaken. For the past 60 years, CQ derivatives have been prescribed for malaria (86), rheumatoid arthritis (87), and HIV (88). They are inexpensive oral drugs that cross the blood-brain barrier. Case reports of infant deaths associated with single tablet ingestions suggest high peak concentrations of CQ may result in significant toxicity. In contrast, suicide attempts involving HCQ did not result in fatalities (89), suggesting HCQ can be safely dose escalated in cancer patients (90). In vitro studies indicate these two drugs are equipotent at autophagy inhibition.

A phase III trial in glioblastoma patients treated with radiation and carmustine with or without daily CQ found a median overall survival of 24 and 11 months in CQ- and placebo-treated patients, respectively (91). This single-institution study was not adequately powered to detect a significant difference in survival, but, established the safety of adding low dose CQ to DNA damaging therapy. Key issues remain that the pharmacology of HCQ (characterized by a long half-life resulting in weeks to achieve peak concentration) and the low potency of the drug (micromolar concentrations are required to inhibit autophagy) may limit its efficacy as an autophagy inhibitor in patients (92). To address these concerns, a phase I-II trial of HCQ with temozolomide and radiation for glioblastoma patients was launched through the American Brain Tumor Consortium and included pharmacodynamic (PD) and pharmacokinetic (PK) analyses. PD evidence of HCQ dose-dependent autophagy inhibition was observed using a novel electron microscopy assay on serial blood mononuclear cells (Fig. 5; ref. 93). Overall survival is the primary endpoint for this phase I-II trial, so information about the antitumor activity of this combination should be forthcoming.

Currently, more than 20 trials involving HCQ are accruing cancer patients nationwide, and many of them have evidence of preliminary antitumor activity (Table 1). The knowledge gained from the PD, PK, and predictive biomarkers in these studies will guide the development of more potent and specific autophagy inhibitors that are being developed by academic and industry discovery programs.

In the era of targeted drug development, efforts to understand, modulate, and develop biomarkers of autophagy as a survival mechanism used by tumor cells to tolerate stress are critically important. As an addition to the hallmarks of cancer originally proposed by Weinberg and colleagues (94), new basic hallmarks of cancer cells were recently highlighted and included the ability to tolerate metabolic, oxidative, DNA damage, mitotic, and proteotoxic stresses (95). Given that autophagy can allow tumor cells to tolerate these multiple stresses, and many novel agents under development in clinical trials have been found to modulate autophagy, the assessment of autophagy and its relevance to a particular agent will likely help improve effectiveness. In fact, multiple agents under development within pharmaceutical companies or the Cancer Therapy Evaluation Program (CTEP; http://ctep.cancer.gov/branches/idb/default.htm) have been shown to modulate autophagy, including histone deacetylase inhibitors, antiangiogenic agents, mTOR inhibitors, BH3 domain mimetics, and glycolytic inhibitors (83, 96–98). In a phase I clinical trial, 2-deoxyglucose, a prototypical agent that inhibits glycolysis, was well tolerated and reduced p62 in peripheral blood mononuclear cells consistent with induction of autophagy (99). Preclinical studies with 2-deoxyglucose showed induction of autophagy, and modulation of autophagy increased cytotoxicity, supporting the hypothesis that further studies of agents such as 2-deoxyglucose that induce autophagy should be tested in combination with autophagy inhibition (100–102). To date, a comprehensive study has not compared multiple classes of inhibitors for their ability to induce autophagy in the same model system.

In addition to focusing research efforts on identifying which anticancer therapeutics are most limited by therapy-induced autophagy, interest is growing in developing more potent and specific autophagy inhibitors. Academic and industry efforts are underway to develop tools that will enable high-throughput screening of chemical libraries to identify novel candidate compounds that inhibit autophagy at various points of control described above. Compounds have been found that unexpectedly inhibit autophagy such as 2-phenylethynesulfonamide (PES), a small molecule heat shock protein 70 (HSP70) inhibitor that results in misfolding of a number of lysosomal proteins (103). A critical component to drug development is the development of assays that can be translated into PD and predictive biomarkers of response to autophagy induction and inhibition. Although studies of biomarkers of autophagy are early in development with additional markers emerging, preliminary data support the ability to measure Beclin1 by immunohistochemistry as a measure of autophagy competence, and the measurement of AV number directly by electron microscopy, LC3, and p62 levels as markers of autophagy modulation (35, 104). Given the basic biological importance of autophagy as a cellular mechanism of survival during multiple forms of cancer and therapeutic-induced stress, an ongoing dialogue between emerging laboratory and clinical research will be imperative to address autophagy as a targetable resistance mechanism in advanced disease and the induction of autophagy as chemoprevention strategy in early phase disease.

No potential conflicts of interest were disclosed.

Grants supporting this work include: R.K. Amaravadi, 1K23CA120862–01A2; W.A. Weiss, NIH P50CA097257, Burroughs Wellcome Fund, American Brain Tumor Association, The Brain Tumor Society, Accelerate Brain Cancer Cure, Pediatric Brain Tumor, and Samuel G. Waxman Foundations; R.S. DiPaola, W81XWH-09–1-0394, W81XWH-09–1-0145, and U01 CA132194; X.M. Yin, CA83817, CA111456; and E. White, R37CA53370, RO1CA130893, and RC1CA147961.