Defects in autophagy have been linked to a wide range of medical illnesses, including cancer as well as infectious, neurodegenerative, inflammatory, and metabolic diseases. These observations have led to the hypothesis that autophagy inducers may prevent or treat certain clinical conditions. Lifestyle and nutritional factors, such as exercise and caloric restriction, may exert their known health benefits through the autophagy pathway. Several currently available FDA-approved drugs have been shown to enhance autophagy, and this autophagy-enhancing action may be repurposed for use in novel clinical indications. The development of new drugs that are designed to be more selective inducers of autophagy function in target organs is expected to maximize clinical benefits while minimizing toxicity. This Review summarizes the rationale and current approaches for developing autophagy inducers in medicine, the factors to be considered in defining disease targets for such therapy, and the potential benefits of such treatment for human health.

The past two decades have witnessed an explosion of research on the molecular mechanisms of autophagy and its roles in physiology and disease. Numerous gene products essential for the induction of autophagy, the formation of autophagosomes, the lysosomal clearance of autophagosomes, and the targeting of specific cargo to the autophagosomes have been identified (1–3). The biochemical and structural mechanisms by which these gene products act in an orchestrated manner to execute autophagy are being defined (4). Through loss-of-function studies in mice and other model organisms, we have learned that autophagy plays crucial roles in differentiation and development (5, 6), cellular and tissue homeostasis (7), protein and organelle quality control (8), metabolism (9), immunity (10), and protection against aging (11) and diverse diseases (refs. 12, 13, and Table 1; supplemental references available online with this article; doi:10.1172/JCI73938DS1). Moreover, an increasing number of human diseases are being linked to polymorphisms or mutations in autophagy genes (Table 2) or deficiencies in autophagy function (14, 15). Based on these advances, considerable interest has emerged in developing new (or exploiting old) strategies to induce autophagy — through either pharmacologic or non-pharmacologic approaches. In this Review we provide an overview of the rationale, potential disease targets, and current efforts and future challenges for the development of autophagy inducers in clinical medicine. Other Reviews in this issue discuss the potential use of autophagy inhibitors in clinical medicine (16–18).

 Second, mutations or polymorphisms in autophagy genes are associated with susceptibility to human diseases (see Table 2), including Parkinson’s disease, inflammatory bowel disease, breast and other malignancies, mycobacterial infections, asthma, chronic obstructive pulmonary disease (COPD), systemic lupus erythematosus, and hereditary neurologic disorders. Third, autophagy gene therapy (via lentiviral or adenovirus-associated viral delivery) in specific target organs results in clinical improvement in rodent models of obesity, α1-antitrypsin deficiency, Parkinson’s disease, Alzheimer’s disease, Pompe disease, muscular dystrophy, cystic fibrosis, and KRAS-driven lung carcinomas, and systemic transgenic expression of an autophagy gene in mice extends lifespan and improves metabolism (Table 3). Fourth, several commonly used drugs and nutritional supplements induce autophagy (Table 4). Although it is generally unknown whether these agents exert their clinical benefits through autophagy or other pathways, there is considerable overlap between diseases that occur in the setting of autophagy deficiency and diseases that respond to drugs that can induce autophagy. Moreover, some of these agents fail to exert beneficial effects in model organisms lacking autophagy genes. For example, spermidine and resveratrol extend life span in wild-type but not autophagy gene–deficient nematodes (22). Similarly, tyrosine kinase inhibitors do not improve amyloid clearance in the brains of mice with Alzheimer’s-like disease when the essential autophagy gene, beclin 1 (Becn1), is depleted by shRNA-mediated knockdown (23).

Factors to consider in defining disease targets for autophagy induction Considerable enthusiasm has emerged for the development of autophagy-inducing agents for the prevention or treatment of diseases in which the upregulation of autophagy is thought to be clinically beneficial. The spectrum of potential disease targets is broad and has been reviewed extensively elsewhere (9, 11, 24, 25); the major focus has been on neurodegenerative disorders, infectious diseases, aging, and metabolic diseases. A common underlying pathophysiologic event in these diseases is the accumulation of harmful contents inside the cell — damaged organelles, protein aggregates, lipid droplets, or pathogens. In these circumstances, the pharmacologic (or non-pharmacologic) enhancement of autophagy-mediated delivery of deleterious structures for lysosomal destruction may be beneficial (Figure 1). For such manipulations to have therapeutic value, the machinery involved in autophagosome formation, cargo recognition and targeting, or autophagic delivery to lysosomes should not be rate limiting. If polymorphisms in autophagy genes associated with autophagosomal formation were to cause disease via loss of autophagy (which has not yet been determined for the mutations listed in Table 2), it is possible that such mutations could also block the successful upregulation of autophagy by agents that would otherwise induce autophagy in wild-type cells with an intact autophagy machinery. 

Similarly, it is not known whether patients with mutations in factors involved in autophagic cargo recognition and targeting (Table 2) (such as PARK2 and PINK1 mutations, which occur in Parkinson’s disease, or SQSTM1/p62 mutation, which occurs in Paget’s disease of the bone) will respond to autophagy inducers. Furthermore, several diseases involve an impairment of the delivery of autophagosomes to lysosomes, including human motor neuron disease associated with mutations in the dynein apparatus (26), lysosomal storage diseases (27), and familial Alzheimer’s disease caused by presenilin 1 mutations (28). In these cases, increasing autophagosomal membrane formation will not necessarily enhance autophagic substrate degradation and may result in a toxic buildup of cellular membranes, polyubiquitinated aggregates, and dysfunctional mitochondria (29). One way to enhance successful autophagic substrate degradation in the setting of lysosomal dysfunction may be to upregulate transcription factor EB (TFEB), a master regulator of both autophagy gene expression and lysosomal biogenesis (30). As a proof of principle, Tfeb gene therapy decreases glycogen storage and excess accumulation of autophagosomes in a murine model of the lysosomal storage disease, Pompe disease (31). However, if the defects in autophagy (at the stage of induction, autophagosome formation, cargo targeting, or autophagolysosomal maturation) are only partial, diseases may still benefit from pharmacologic upregulation of autophagosome formation. 

Indeed, autophagy-enhancing agents show beneficial effects in induced pluripotent stem cells from patients with the lysosomal storage disorder Niemann-Pick type C disease (32) and in NPC1 mutant mouse cells (33). Further studies are warranted to examine the effects of autophagy inducers in other lysosomal storage disorders and diseases associated with impaired cargo delivery to the autophagosome. An interesting question is whether autophagy induction is warranted in clinical conditions without an apparent defect in the autophagic machinery. Drugs that induce autophagy in autophagycompetent animals have favorable effects in diseases characterized by abnormal accumulation of protein substrates (such as Huntingtin’s disease treated with rapamycin [ref. 34] or rilmenidine [ref. 35], and α1-antitryspin deficiency treated with carbamazepine [ref. 36]) or pathogens (such as arboviral infections treated with the autophagy-inducing peptide Tat–beclin 1 [ref. 37] and pulmonary M. tuberculosis infection treated with statins [ref. 38]), suggesting that enhancement of autophagy may be beneficial in the absence of an overt autophagy deficiency. Unfortunately, given our current limitations in measuring autophagic flux in patients, we do not know what constitutes a “normal” range of autophagic activity. However, autophagy function declines with aging in humans and other species, and such a decline likely contributes to aging itself as well as age-related increases in susceptibility to neurodegenerative disorders, infectious diseases, and cancer (11). Preliminary studies also indicate that critically ill patients have an autophagy-deficient phenotype, at least in skeletal muscle and liver (39). 

Thus, many patients, by virtue of advanced age or severe illness, may have a deficiency in autophagy function in the absence of specific mutations in the autophagy pathway. In defining disease targets appropriate for autophagy induction therapy, it is important to consider whether autophagy acts in a cytoprotective or cytotoxic manner in the specific disease context, and whether such cytoprotective or cytotoxic actions contribute to disease progression. Although autophagy is classically regarded as a means of promoting cell survival during nutrient-limited conditions (5, 40), autophagy can also contribute to cell death (41). The pro-survival role of autophagy is commonly believed to promote the progression of cancers driven by RAS mutations (42, 43), which has led to intense efforts to inhibit autophagy in this context (discussed in other Reviews in this issue and in refs. 44, 45). However, this postulated pro-survival action is not necessarily sufficient to override the tumor suppressor effects of autophagy in all cancers. For example, enhanced suppression of autophagy in EGFR-driven non–small cell lung adenocarcinoma xenografts increases tumor cell death but also promotes enhanced proliferation, increased tumor growth and tumor dedifferentiation, as well as resistance to EGFR tyrosine kinase inhibitor therapy (46). 

Even in RAS-driven tumor cells, autophagy inhibition does not have predictable antitumor effects. Specifically, in RAS-driven oncogenesis, autophagy gene knockdown enhances clonogenic survival in human ovarian epithelial cells (47); the presence of a homozygous p53 mutation transforms the actions of autophagy from a pro-tumorigenic to anti-tumorigenic effect in pancreatic carcinoma (48); autophagy suppresses early oncogenesis in lung adenocarcinoma through effects on regulatory T cells (49); and beclin 1 gene transfer prevents the progression from lung adenomas to adenocarcinomas and enhances tumor cell death (50). Moreover, autophagy genes are often required for the cytotoxic effects of chemotherapy (51–53), and the combination of antimetabolite and receptor tyrosine kinase inhibitor therapy can increase autophagy-dependent tumor cell death (54). Autophagy also contributes to radiosensitivity in vivo (through immune-dependent responses) even if it contributes to radioresistance in vitro (55).  In addition, multiple myeloma cells utilize caspase-10 to restrain autophagy and undergo autophagic cell death following caspase-10 inhibition (56). Interestingly, glucose-starved yeast and mammalian cells do not engage in prosurvival autophagy (56, 57), challenging the notion that glucose deprivation (one of the most common forms of metabolic stress in the tumor microenvironment) induces prosurvival autophagy in vivo. 

Thus, because the role of autophagy in cancer progression and response to therapy is complex and context dependent, it is possible that — despite the prosurvival effects of autophagy in some tumor cells — the induction of autophagy may still be useful in certain cancers through autophagy-dependent antitumor immunity, autophagy-dependent cytotoxic effects, or other tumor-suppressor effects. While cytotoxic effects of autophagy or autophagy-dependent anticancer immune responses may be beneficial in certain malignancies, the cytotoxic effects of autophagy may be pathogenic in other diseases. These include mouse models of cigarette smoke–induced COPD (58), acute lung injury caused by avian influenza A H5N1 infection (59), diabetes-induced and pressure overload–induced cardiomyopathies (60–62), pancreatic β-cell death in the setting of Pdx1 deficiency (63), ischemic brain damage in diabetes (64), and traumatic, ischemic, ischemic/reperfusion, and/or hypoxic injury in the brain, heart, or kidney of nondiabetic subjects (65–69). Such “pro-death” effects of autophagy (whether they are direct through autophagic cell death, indirect through enhanced apoptosis as postulated in lung epithelial cells exposed to cigarette smoke, or a combination of both) have yet to be confirmed in patients. 

However, if cigarette smoking or diabetes (or other comorbid conditions) increase susceptibility to autophagy-dependent enhancement of organ pathology in the clinical setting, the coexistence of these common comorbid conditions might affect the safety of utilizing autophagy inducers, particularly if they are not organ specific. Even if present, these unwanted effects could occur at doses higher than or durations longer than those needed to enhance the autophagy-mediated delivery of deleterious structures for lysosomal destruction. This would allow for the identification of a useful therapeutic window and the safe development of low doses or short-term or intermittent treatment strategies. Besides potential unwanted cytoprotective effects or cytotoxic effects, another possible concern regarding autophagy induction relates to the complex roles of autophagy proteins in infectious diseases, immunity, and inflammation. Autophagy plays important roles in protection against several medically important intracellular pathogens through different mechanisms including xenophagy (the selective autophagic degradation of microbes), enhanced adaptive immunity, and the prevention of excessive inflammatory responses (Figure 1), leading to significant optimism that autophagy inducers may represent an important new class of host-direct anti-infective therapy (10, 70). 

However, the autophagy pathway and/or autophagy pathway–independent functions of autophagy proteins may also enhance the replication of certain viruses and intracellular bacteria. For example, pancreatic cell–specific knockout of Atg5 dramatically reduces Coxsackie virus replication and virus-induced pathology (71); similarly, liver-specific knockout of Atg5 reduces HBV DNA replication in mice that transgenically express HBV (72). It is not known whether these phenotypes reflect a role for the autophagy pathway or for autophagy pathway–independent effects of Atg5, which can also function in the negative regulation of inflammasomes, recruitment of immunity-related GTPases, secretion, exocytosis, and formation of membranes that serve as scaffolds for viral replication (10, 70). If the mechanism involves autophagy, autophagy inducers might be contraindicated in the setting of certain infections such as HBV. However, if the mechanism involves autophagy-independent functions of Atg5, it is unknown whether the upstream upregulation of autophagic flux will also enhance autophagy-independent pro-pathogen functions of individual autophagy proteins.

 Suppression of inflammatory responses by autophagy induction might also impair the host capacity to clear infectious organisms. Studies of the loss of autophagy gene function in B cells and dendritic cells suggests a crucial role for autophagy in antigen-specific immune responses. For example, B cell–specific deletion of Atg7 impairs virus-specific B cell memory in mice, leading to lethal influenza virus infection (73). Dendritic cell–specific deletion of Atg5 results in impaired antigen presentation and increased susceptibility to lethal herpes simplex virus (HSV) infection (74). Moreover, autophagy-competent but not autophagy-deficient tumor cells attract dendritic cells and T cells into tumor beds, leading to enhanced antitumor immunity (75). However, it is unknown whether enhancement of autophagy will augment these antigenspecific immune responses; if so, autophagy inducers may have an important clinical role in enhancing vaccine and antitumor immunity. Of note, rapamycin increases the generation of memory CD8+ T cells in mice following lymphocytic choriomeningitis virus infection or vaccination with a modified vaccinia virus (76), although it is not yet known whether this effect of mTOR inhibition is mediated through autophagy.

 Polymorphisms linked to certain genes involved in autophagy regulation or autophagosome formation have been identified in human autoimmune diseases (such as ATG5 and systemic lupus erythematosus) and inflammatory disorders (such as IRGM, NOD2, and ATG16L1 and inflammatory bowel disease) (Table 2), raising the possibility that these disorders may be targets for autophagy induction therapy. However, the effects of the polymorphisms on gene function have not been sufficiently well characterized to allow a prediction of the effects of autophagy induction. For example, the ATG16L1T300A risk allele for Crohn’s disease impairs intestinal Paneth cell function, regulation of proinflammatory IL-1β production, and likely, bacterial autophagy (77–79). However, this region of ATG16L1 is not conserved in yeast Atg16 and may not be required for general autophagy (80); thus, it is unknown whether pharmacologic enhancement of autophagy in Crohn’s disease will correct the pathophysiologic defects imposed by the ATG16L1T300A mutation. Further studies are required to dissect the molecular function of autophagy risk alleles and their specific roles in the pathogenesis of diseases to determine effective strategies for correcting molecular defects imposed by such genetic variations.Non-pharmacologic interventions such as caloric restriction and regular exercise induce autophagy and may improve overall health. Exercise-induced autophagy may be required for exercisemediated protection against high fat diet–induced diabetes in mice (81); this raises the possibility that autophagy enhancement may also underlie other beneficial health effects of exercise, such as delaying the onset or progression of human cancers and neurodegenerative diseases (82).