Half a century ago, Christian de Duve coined the term “autophagy” (literally, “self-eating” in Greek) to describe a process where the cell digests its cytoplasmic materials within lysosomes1. At least three major types of autophagy have been identified: macroautophagy, characterized by the formation of a unique double-membrane organelle called the autophagosome; microautophagy, where lysosomes engulf cytoplasmic materials by inward invagination of the lysosomal membrane; and chaperone-mediated autophagy, mediated by the chaperone hsc70, co-chaperones, and the lysosomal-associated membrane protein type 2A2,3. This review focuses on the role of macroautophagy (hereafter referred to as autophagy) in human diseases.

In recent years, genetic deletion of the autophagy-related (Atg) genes in various model organisms, including mammals, has revealed that autophagy plays critical roles in adaptive responses to starvation and other forms of stress, homeostasis, and cellular differentiation and development2,4,5,6,7. In addition, analysis of mice with systemic or tissue-specific deletion of Atg genes has revealed the connection between dysregulated autophagy and various kinds of disease-like phenotypes including cancer, neurodegenerative diseases, infectious diseases, and metabolic diseases2,6,7,8,9,10,11. However, these experimental results do not directly demonstrate that defects in autophagy contribute to pathogenesis of human diseases. Thus, it has become particularly important to understand the genetic basis of putative human autophagy-related diseases.

With the completion of the Human Genome Project in 2003 and the International HapMap Project in 2005, researchers now have a powerful set of research tools, including the high-speed DNA sequencing technology that make it possible to identify the genetic contributions to specific diseases, even if they are rare. Indeed, genome-wide studies have identified disease-associated loci and genes in many human diseases. Table 1 summarizes the association between genetic variants of autophagy-related genes and selected human diseases.

Table 1 Human diseases associated with defective autophagy

Static encephalopathy of childhood with neurode-generation in adulthood (SENDA)

Recently, two groups identified de novo mutations in WDR45, an autophagy-related gene located at Xp11.23, in individuals with SENDA by whole-exome sequencing using next-generation sequencing technologies12,13. SENDA is a recently established subtype of neurode-generation with brain iron accumulation14 that begins with early-onset spastic paraplegia and mental retardation, which remain static until adulthood. Patients subsequently develop sudden-onset parkinsonism and dystonia during their late 20s to early 30s. Additional features include eye movement abnormalities, frontal release signs, sleep disorders, and dysautonomia. Brain magnetic resonance imaging has revealed iron accumulation in the globus pallidus and hypointensity in the substantia nigra, as well as white matter changes14,15.

The hit gene WDR45 (also known as WIPI4) is one of the four mammalian homologues of yeast Atg18, which plays an important role in autophagosome formation16,17,18,19. Atg18/WIPIs belong to the PROPPIN family of proteins. They contain seven-bladed β-propellers formed by seven WD40 repeats and bind to phosphatidylinositol 3-phosphate and the lysosomal/vacuolar lipid phosphatidylinositol 3,5-bisphosphate (PtdIns(3,5)P2)17,20. Atg18/WIPIs also interact with Atg220,21,22. The crystal structure of Hsv2, a yeast Atg18 paralogue, shows two phosphoinositide-binding sites at blades five and six, and an Atg2-binding region at blade 223,24,25. Atg18/WIPIs are recruited to the autophagosome formation site through binding to phosphatidylinositol 3-phosphate, which is synthesized by the class III PtdIns 3-kinase complex18,21. Caenorhabditis elegans has two Atg18 homologues, ATG-18 and EPG-619. Interestingly, C. elegans requires both ATG-18 and EPG-6 for autophagy because the two molecules function sequentially, not redundantly. Human WDR45/WIPI4 shows a higher similarity to EPG-6 than to ATG-18, and loss of epg-6/WIPI4 causes the accumulation of premature autophagic structures in both C. elegans and mammalian cells19. In fact, by using lymphoblastoid cell lines derived from SENDA patients, Saitsu et al. confirmed that the protein expression of WIPI4 was severely reduced in affected individuals. Specifically, blocked autophagic flux and accumulation of abnormal ATG9A- and LC3-double-positive structures, which may represent aberrant early autophagic structures, were observed in the lymphoblastoid cell lines of affected individuals13. Since WDR45/WIPI4 is encoded by the X chromosome and one of the X chromosomes is subjected to X inactivation, female patients should possess mosaic loss of function of WDR45/WIPI4. It is unclear, however, whether hemizygous mutations in male patients are lethal. Hayflick's group reported three male SENDA patients with similar phenotypes12,26; all three may have had somatic mosaicism.

These studies provided the first direct evidence that the deficiency of a core autophagy factor is indeed a contributing factor to human neurodegenerative diseases. However, the exact mechanism of brain iron accumulation due to an autophagy defect and why only the brain is affected remain to be clarified. Further investigation of these aspects is needed.

Vici syndrome

A recent study by Cullup et al. showed that recessive mutations in EPG5, a key factor implicated in the maturation of autolysosomes, play a causative role in Vici syndrome27. Vici syndrome is a recessively inherited multisystem disorder characterized by callosal agenesis, cataracts, hypopigmentation, cardiomyopathy, psychomotor retardation, and immunodeficiency with cleft lip and palate28,29,30,31.

EPG5 is a metazoan-specific autophagy gene first identified by genetically screening C. elegans for mutants with defective degradation of autophagy substrates. C. elegans epg-5 mutant and knockdown of mEPG5 in mammalian cells show accumulation of non-degradative autolysosomes, indicating the role of EPG-5/mEPG5 in autolysosome maturation32. It was later shown that knockdown of EPG5 in HeLa cells results in another defect in the endocytic pathway33. By using fibroblasts derived from patients with Vici syndrome, Cullup et al. showed that autophagic flux is blocked and the autophagy adapters NBR1 and SQSTM1/p62 accumulate, confirming the decreased autophagic activity in Vici syndrome27. However, as EPG5 is also involved in the endocytic pathway, it is important to examine whether dysregulated endocytic trafficking also contributes to the pathogenesis of Vici syndrome. Furthermore, the Epg5-deficient mice display only some features of Vici syndrome33,34. For example, although patients with Vici syndrome demonstrate facial dysmorphism and cataracts, these features are not marked in the Epg5-deficient mice. In addition, psychomotor abnormalities appear to be milder in mice than in humans. Further studies are needed to elucidate the reason for phenotypic differences between mice and humans as well as the exact molecular role of EPG5 in the autophagy and endocytic pathways.

Hereditary spastic paraparesis

Oz-Levi et al. reported a recessive mutation in TECPR2, an autophagy-related WD repeat-containing protein, in five individuals with SPG49, a novel form of recessive hereditary spastic paraparesis (HSP)35. HSP is a diverse group of neurodegenerative disorders characterized by axonal degeneration of the corticospinal or pyramidal motor and sensory tracts that control the lower extremities. It leads to progressive spasticity and hyperreflexia of the lower limbs36,37,38. The newly characterized HSP subtype, accompanied by lower-limb spasticity and other neurological symptoms, appears to be an autosomal-recessive form of complicated HSP that is caused by a single base deletion in the TECPR2 gene, resulting in a premature stop codon accompanied by full degradation of its protein product35.

TECPR2, an uncharacterized protein belonging to the tectonin β-propeller repeat-containing protein family, was previously found to interact with ATG8 orthologues, suggesting a possible role in the autophagy pathway39. Skin fibroblasts from an HSP patient showed decreased autophagic flux, but no accumulation of the autophagic substrate SQSTM1/p62, implying that some autophagic activity could be maintained in affected individuals. Knockdown of TECPR2 in HeLa cells also reduced autophagic activity, suggesting that TECPR2 is a bona fide autophagy factor35. However, the exact role of TECPR2 in the autophagy pathway warrants further examination. The fact that TECPR2 shows some similarity to two autophagy-involved proteins — TECPR1 and HPS535,40,41,42,43 — is expected to shed new light on this issue.

Recently, Vantaggiato et al. reported that ZFYVE26/SPG15, the causative gene of another recessive complicated form of HSP (HSP type 15), is also involved in the autophagy process44. ZFYVE26/SPG15 encodes a zinc-finger protein with a FYVE domain and a leucine zipper, termed spastizin45. Spastizin interacts with the Beclin 1-UVRAG-Rubicon complex and mediates autophagosome maturation. Both spastizin-mutated fibroblast cells derived from HSP patients and spastizin knockdown cells showed impaired autophagic flux and accumulation of autophagosomes due to reduced autophagosome–lysosome fusion44. However, as this complex also plays an important role in the endocytic pathway46,47,48,49, and as spastizin is not present on the autophagic membranes44, whether spastizin specifically regulates autophagosome-lysosome fusion needs to be clarified.

Parkinson's disease

Parkinson's disease is the most common form of a group of progressive neurodegenerative disorders characterized clinically by bradykinesia (paucity and slowness of movement), rest tremor, muscular rigidity, shuffling gait, and flexed posture. It can also be accompanied by various non-motor symptoms, including sleep, autonomic, sensory, cognitive, and psychiatric disturbances. Nearly all forms of Parkinson's disease result from reduced dopaminergic transmission in the basal ganglia50,51. Many genes, mutations, and polymorphisms have been implicated in the pathogenesis of the disease. Among them, mutations in the PARK2/Parkin and PARK6/PINK1 have been shown to lead to autosomal recessive or sporadic juvenile-onset Parkinson's disease52,53,54.

PTEN-induced putative kinase protein 1 (PINK1, encoded by PARK6/PINK1) is a mitochondria-associated protein kinase that acts upstream of Parkin (encoded by PARK2/Parkin), an E3 ubiquitin ligase implicated in the selective degradation of damaged mitochondria by autophagy, a process termed “mitophagy”55,56,57. When mitochondria are damaged and lose their membrane potential, mitochondrial PINK1 is stabilized and recruits Parkin, which ubiquitinates a number of mitochondrial membrane proteins, resulting in selective mitophagy. Consistent with this finding, excessive mitochondrial damage has been linked to Parkinson's disease58. Thus, this type of Parkinson's disease can be caused by the accumulation of mitochondrial damage. However, Parkin is also reported to mediate other biological processes, including translocation of some mitochondrial outer membrane proteins to the endoplasmic reticulum to escape autophagic degradation59. Furthermore, other studies have shown that Parkin also mediates proteasome-dependent degradation of outer membrane proteins of depolarized mitochondria, although it is controversial whether this process is required for mitophagy60,61,62, these findings suggest that an autophagic defect may not be the only factor contributing to the pathogenesis of PINK1/Parkin-related Parkinson's disease. It would also be important to know whether PINK1/Parkin-mediated mitophagy occurs under physiological conditions, because most previous studies were performed in cells overexpressing Parkin, and PINK1/Parkin knockout mice failed to faithfully recapitulate Parkinson's disease in humans63,64,65.

Lysosomal storage disorders

Lysosomal storage disorders (LSDs), characterized by progressive accumulation of undigested macromolecules within the cell, are a family of disorders caused by inherited gene mutations that perturb lysosomal homeostasis. As lysosomes also play an important role in the autophagy pathway by fusing with autophagosomes and degrading autophagic cargo, lysosomal dysfunction in LSDs impacts the autophagy pathway. In fact, in most LSDs, the lysosomal dysfunction is accompanied by impaired autophagic flux, resulting in defective autophagosome-lysosome fusion and secondary accumulation of autophagy substrates such as SQSTM1/p62, polyubiquitinated proteins, and damaged mitochondria66. In some sense then, LSDs can be regarded as “autophagy disorders”. Some excellent reviews on the genetic basis of LSDs are available11,67,68.


An association between autophagy and cancer has long been proposed. The role of autophagy likely differs in different stages of cancer development; initially, autophagy probably has a preventive effect against cancer, but once a tumor develops, the cancer cells could utilize autophagy for their own cytoprotection9,69.

Monoallelic deletion of BECN1 has been detected in human breast, ovarian, and prostate tumor specimens70,71,72,73. In particular, the aberrant expression of Beclin 1 (encoded by the human BECN1 gene) in many kinds of tumor tissues correlates with poor prognosis74,75,76,77,78. Beclin 1, the mammalian orthologue of yeast Atg6/vacuolar protein sorting (Vps)-30, plays an essential role in autophagy. It interacts with the class III PtdIns 3-kinase, Vps34 (also known as PIK3C3 in mammals), to form the Beclin 1-Atg14-Vps34-Vps15 complex, which is important for the localization of downstream autophagic proteins to the autophagosome formation site to induce autophagy73,79. Beclin 1 also has other important biological functions including roles in anti-apoptosis80,81 and endocytic trafficking47,82,83.

A recent study in C. elegans identified EI24/PIG8, whose human homolog was reported to be mutated in breast cancers84, as a critical factor of autophagic degradation32. However, it remains to be clarified whether EI24-mutated human breast cancer cells indeed show decreased autophagic activity. Furthermore, since EI24/PIG8 is also known as the proapoptotic factor84,85, this role may contribute to tumor suppression. Besides Beclin 1 and EI24, altered expression of several autophagy proteins such as ATG586,87, and UVRAG88 are reported to be associated with human cancers7,89.

Crohn's disease

Genome-wide association studies of non-synonymous SNPs have linked ATG16L1 variants with susceptibility to Crohn's disease90,91, a major type of inflammatory bowel disease that can affect any part of the digestive tract from the mouth to the anus. The disease causes a wide variety of symptoms including abdominal pain, diarrhea, vomiting, and weight loss, as well as complications outside the gastrointestinal tract such as fatigue, skin rash, inflammation of the eye, anemia, arthritis, and lack of concentration92.

Atg16L1, a core component of the autophagy machinery, forms a complex with Atg12-Atg5 to induce LC3 lipidation and is essential for autophagosome formation93,94. Recent studies have shown that the interaction between Atg16L1 and FIP200 is important for the localization of the Atg12-Atg5-Atg16L1 complex to the autophagosome formation site or isolation membrane95,96. The Atg16L1 protein possesses a C-terminal WD repeat domain, and the Crohn's disease-associated mutation (T300A, also known as Ala197Thr) is within or immediately upstream of this domain. However, it was shown that the Atg16L1 WD repeat domain is not essential for autophagic activity96,97. Thus, it is important to clarify how the ATG16L1 T300A mutation contributes to the pathogenesis of Crohn's disease in humans.

Investigations of mice carrying two distinct mutations that reduce or eliminate the expression of Atg16L1 have suggested potential links between Atg16L1 mutations and Crohn's disease. It was shown that Atg16L1-deficient macrophages produced more of the inflammatory cytokines IL-1β and IL-18 upon stimulation with lipopolysaccharides98. On the other hand, the Atg16L1 hypomorph mice exhibited aberrant granule formation in Paneth cells, which play an important role in the innate immune response of the intestine99. Recently, Marchiando et al. reported that Atg16L1 possesses an immunosuppressive role during intestinal bacterial infection100.

Apart from Atg16L1, other autophagy-related proteins such as IRGM101,102,103 and NOD2104,105,106 are reported to be associated with Crohn's disease in humans107. However, since these proteins also play roles in biological processes other than autophagy, it remains unclear whether they relate to Crohn's disease via autophagy modulation.

Conclusion and future prospects

In this article, we have summarized recent findings on the relationship between autophagy and human diseases. It is expected that new efficient technologies such as exome sequencing will help to identify more autophagy-related diseases over the next few years. Given that autophagy is associated with a plethora of human diseases, there are at least two important issues to address.

First, the development of pharmacological agents that modulate autophagy in these pathological conditions is critical; in fact, it has become a major priority in the field. Pharmacological approaches to activate or inhibit autophagy are also required because autophagy can play either a protective or destructive role in different diseases, even in different stages of the same diseases. Many drugs and compounds that modulate autophagy are currently receiving considerable attention11,89,108. These include, for example, autophagy inducers such as the mTORC1 inhibitor rapamycin109 and its analogues (e.g., CCI-779109, RAD001110,111, and AP23573112), mTOR kinase inhibitors (e.g., Torin 1113, and PP242114), trehalose115,116, carbamazepine117, and the newly identified autophagy-inducing peptide Tat–beclin 1118; autophagy inhibitors such as chloroquine119,120 and hydroxychloroquine121, Lys05122, 3-methyladenine123 and its derivatives124, PIK3C3 inhibitors125, ATG4B inhibitors126,127, and ATG7 inhibitors128,129. Autophagy-modulating drugs that are currently used in clinical trials are summarized in Table 2. An improved understanding of how autophagy defects contribute to the pathogenesis of human diseases and the development of other more specific and less toxic compounds will benefit many more patients.

Table 2 Autophagy-modulating compounds in clinical trials

Second, and perhaps a more challenging issue, is the monitoring of autophagic activity in humans, in tissue samples at the least, but preferably in blood samples. In particular, it is more important to measure autophagic flux than autophagosome number. To date, however, measurement of autophagic flux in paraffin-embedded tissue samples has been unsuccessful, and the simple detection of endogenous LC3-II, a commonly used marker for autophagosomes, has proved problematic in tissue sections. The appearance of more LC3-positive puncta (which may represent autophagosomes) does not necessarily indicate higher autophagic activity in the tissue. Autophagosomes can accumulate due to the induction of autophagy or due to blocking of a late step of the autophagy pathway, including impaired autophagosome-lysosome fusion and compromised lysosomal activity130. This is a frequent occurrence in human diseases and even during the normal aging process. It should also be remembered that LC3 can be incorporated into protein aggregates independently of autophagy131. To help overcome these problems, it may be beneficial to combine immunohistochemical assays of other autophagy-related marker proteins such as ATG5 and Beclin 1 to detect autophagy in clinical tissue samples.