Nutritional deprivation is one of the most well-known autophagy inducers, and mTOR inhibition is its best characterized mechanism. In the presence of growth factors such as insulin or insulin-like growth factors, stimulated Akt and extracellular signal-regulated protein kinases 1 and 2 (ERK1/2) can phosphorylate and disrupt the tuberous sclerosis complex 1/2 (TSC1/TSC2), which activates mTOR inhibition and thus inhibiting autophagy [19]. However, when growth factors are depleted, the active TSC1/TSC2 complex suppresses mTOR activity, followed by induction of autophagy. The second most critical molecule that regulates autophagy in response to metabolic stress is via the AMPK pathway. AMPK checks the energy status by sensing the AMP:ATP ratio of the cell. AMPK is also activated by several intracellular stresses including high intracellular calcium levels as well as energy depletion via diverse upstream kinase activation [20]. For example, energy depletion activates liver kinase B1 and high cytosolic Ca⁺⁺ induces calcium/calmodulin kinase kinase-β. This activated kinase stimulates AMPK, which induces autophagy via mTORC1 inhibition (via phosphorylation of TSC2 and the regulatory-associated protein of mTOR, Raptor).

Autophagy is also induced by the unfolded protein response, an important ER stress pathway, and several ER-associated proteins are involved in its signaling pathway. The major ER proteins that regulate autophagy include protein kinase R-like eIF2α kinase (PERK), activating transcription factor-6 (ATF6) and IRE1 (inositol-requiring enzyme 1). PERK induces transcriptional activation of LC3 and Atg5 in response to hypoxia [21]. It was also demonstrated that ATF4 induced by PERK via eIF2α phosphorylation is associated with increased LC3 expression [22]. IRE1, a serine/threonine kinase, can stimulate autophagy via activation of Jun N-terminal kinase 1 (JNK1) or negatively regulate autophagy by action of X-box binding protein 1 [15].

Autophagy induction during infection is mediated by cytokines such as interferon-γ as well as pathogen recognition receptors. Indeed, Toll-like receptors trigger autophagy during innate and adaptive immune responses. For example, lipopolysaccharide induces autophagy via Toll-like receptor-4 in primary human macrophages [23]. Whereas Th1 cytokines such as interferon-γ and tumor necrosis factor (TNF)-α induce autophagy, Th2 cytokines including interleukin (IL)-4 and IL-10 inhibit autophagy [24]. The signaling pathway underlying how inflammatory signals trigger autophagy seems to differ in different cells. For example, in human vascular smooth muscle cells, TNF-α upregulates autophagy via the JNK pathway and inhibition of Akt [25], and in the MCF-7 human breast cancer cell line, the ERK1/2 pathway is involved in TNF-α-induced autophagy [26]. Autophagy also regulates the secretion of numerous cytokines including IL-1, IL-18, and TNF-α [24].

There are several mechanisms by which hypoxia or anoxia induce autophagy. In general, for moderate hypoxia (1% to 3% oxygen), the hypoxia-inducible factor 1-α (HIF1-α) pathway is predominantly used. In brief, stabilization of HIF1-α by hypoxia activates the transcription of Bcl-2/adenovirus E1B 19 kDa-interacting protein (BNIP)3/BNIP3L, which weakens interaction between beclin 1 and Bcl-2 and subsequently elicits autophagy [27]. Furthermore, it was also demonstrated that increased Redd1 by HIF1-α activates TSC1/2 resulting in the inhibition of mTORC1 activity followed by increased autophagy [28]. Severe hypoxic conditions or anoxia (<0.1% oxygen) can induce a HIF1-α-independent autophagic response through activation of AMPK, which inhibits mTOR activity [15] or through the unfolded protein response pathway [29]. In addition, hypoxia enhances the transcription of LC3 and ATG5, which are crucial autophagy genes [21].

Multiple mechanisms are involved in redox stress-induced autophagy. Increased reactive oxygen species levels can induce the activation of PERK and JNK1, which mediate autophagy induction [30]. In addition, exogenous hydrogen peroxide can directly activate Atg4 protease and thus increase the production of mature LC3 [31]. Direct or indirect inhibition of mTOR by oxidative stress was also demonstrated [15].

A dual action on autophagy is exerted by p53, depending on the intracellular location. In the nucleus, p53 activates autophagy via increased transcriptional activity of autophagy-related genes [32], whereas, in the cytoplasm, p53 acts as a negative regulator of autophagy [33].

Conventional electron microscopy is obviously the most standard and reliable method to assess autophagy. Because it is difficult to differentiate by transmission electron microscopy (TEM) between autophagosomes and autophagolysosomes, which are double- and single-membrane structures, respectively, containing cytoplasmic organelles, the term, autophagic vacuoles is commonly used to indicate these structures together. The quantification of autophagy by TEM can be achieved as the ratio of the area of autophagic vacuoles among the total cytoplasmic area [34]. However, TEM has a limitation in that it requires highly skilled experts and is time-consuming. Furthermore, sometimes other organelles such as the ER can be mistaken for autophagic vacuoles by TEM, making it not objectively quantitative.

LC3-positive puncta, a marker of the induction of autophagy, can be observed by immunofluorescence using anti-LC3 antibodies. LC3, fused with green fluorescent protein (GFP) is commonly for in vitro assessment of autophagy. The real-time observation of GFP-LC3 localization is also possible. It was demonstrated that overexpression of GFP-LC3 itself does not modulate endogenous autophagy activity. GFP-LC3 transgenic mice can be used for in vivo analysis of autophagy [35]. Recently, autophagic flux ex vivo in retinal explants from GFP-LC3 transgenic mice using lysosomal inhibitors was also demonstrated [36].

A novel tandem fluorescent tagged LC3 method was proposed to avoid a pitfall in the interpretation of an increase in the number of LC3-positive puncta, which might resulted from the accumulation of autophagosomes rather than the formation of autophagosomes. The limitation of GFP itself for the degradation of lysosomal protease could also be overcome using the tandem fluorescent tagged LC3 vector [37]. In this vector, red fluorescent protein, which is resistant to lysosomal degradation, is also fused with GFP-LC3. According to the stage of autophagy, autophagosomes appear in yellow puncta, whereas mature autophagolysosomes appear red because of the quenching of GFP in the acidic milieu of the lysosome [38].

LC3, a mammalian ortholog of yeast Atg8, is the most frequently used to assess autophagy in cells. Because autophagy is a dynamic process, it is important to understand the basis of the expression of LC3 and the principles of interpretation. In fact, LC3 can localize to any types of autophagic membrane including the phagophore, autophagosome, and autolysosome during the autophagic process [39,40]. Pro-LC3 is cleaved to LC3-I by an Atg4 family protease immediately after synthesis [39]. During autophagy, LC3-I is converted to LC3-II with conjugation to phosphatidylethanolamine by the action of Atg3 and the Atg12-Atg5-Atg16L1 complex, and thus LC3-II is localized in the autophagosomal membranes. Finally, LC3-II is also degraded by autophagy in autophagolysosomes. The amount of LC3-II or LC3 conversion (LC3-I to LC3-II) is often used to monitor autophagy, indicating autophagosome formation. However, it should be noted that increased LC3-II levels can be associated with either increased autophagosome formation or reduced turnover [37]. Moreover, because autophagy is a dynamic process regulated by both the on and off rate, the amount of LC3-II does not necessarily reflect the degree of autophagic activity at a certain time point [41,42]. Thus, autophagic flux, defined as the dynamic process of autophagy, is usually recommended to prove autophagosome formation by comparing the amount of LC3-II in the presence or absence of a lysosomal inhibitor such as bafilomycin A1 (a vacuolar H⁺-ATPase inhibitor) or pepstatin A (lysosomal protein inhibitor). A detailed interpretation of this autophagic flux assay has been well described [43]. Of note, there are some technical issues, which can provide several troubleshooting tips in western blotting for LC3. Although the actual molecular weight of LC3-II is greater than that of LC3-I owing to its binding to phosphatidylethanolamine, the extreme hydrophobicity of LC3-II means it moves faster than LC3-I on sodium dodecyl sulfate polyacrylamide gel electrophoresis. A second point is that certain anti-LC3 antibodies are less sensitive for detecting LC3-I, especially with the semi-dry transfer method [44]. Additionally, it was also reported that the use of a polyvinylidene difluoride membrane rather than a nitrocellulose membrane was associated with successful western blotting for LC3 [37].

Another widely used marker of autophagy is p62, also called as sequestosome 1, which is also an adaptor molecule implicated in the targeting of cargo for autophagosomes. That is, p62 interacts with polyubiquitinated protein aggregates, thereby facilitating these aggregates for degradation at the autolysosome through direct binding with LC3 by a short LC3 interaction region [45]. Because p62 itself is degraded by autophagy, p62 accumulation represents defects in selective autophagy of ubiquitinated aggregates [43]. However, measurement of p62 has some pitfalls. Since p62 is also degraded by ubiquitin-proteasome system as well as autophagy, its level may be increased when the proteasome is inhibited. In addition, the transcription of p62 itself could be increased by oxidative stress. Therefore, an additional method to monitor autophagy is generally needed [46].

Beclin 1 performs a critical role in autophagosome formation by interaction with VPS34. When nutrients are sufficient, this pro-autophagic action of beclin 1 is inhibited by direct binding with Bcl-2, a well-known antiapoptotic protein. During stress conditions, beclin 1 is released from Bcl-2 interaction and induces autophagy. Meanwhile, Bcl-2 inhibits apoptosis by blocking Bax activation and thus preventing mitochondrial outer membrane permeabilization and the subsequent release of cytochrome c. So far, several molecular mechanisms have been reported to be involved in the dissociation of beclin 1 and Bcl-2 binding [51,57]. For example, death-associated protein kinase phosphorylates beclin 1 and interferes with its interaction with Bcl-2, allowing induction of autophagy.

p53 also regulates autophagy in a stimulatory or inhibitory manner. Such dual actions of p53 on autophagy are dependent on the cellular location of p53, which is usually present in the cytosol where it inhibits autophagy as well as apoptosis by PI3K/Akt activation. However, in conditions of cellular stress, p53 moves into the nucleus and transactivates several genes involved in autophagy, such as death-associated protein kinase [58], resulting in enhanced autophagy. It also induces autophagy by increasing phosphorylation of Bcl-2, which weakens its interaction with beclin 1.

P62, a multifunctional protein that binds to ATG/LC3 and exerts proteasomal activity and autophagy, is also interconnected with apoptosis through activation of the caspase 8 complex [59]. Flice inhibitory protein, a known inhibitor of apoptosis, also inhibits autophagy by blocking ATG3 conjugation to LC3 [60]. In addition, it was also suggested that certain Atg proteins may play dual roles in autophagy and apoptosis. For example, Atg5, which regulates autophagy, can be subject to calpain-dependent cleavage to generate a proapoptotic truncation product (tAtg5). This cleavage product promotes apoptosis by binding to and inhibiting antiapoptotic proteins such as Bcl-X_L [61]. Likewise, accumulating evidence has shown that several proteins known to regulate autophagy are involved in apoptosis and vice versa [57].

As neurodegenerative disease is commonly characterized by mitochondrial dysfunction and the accumulation of protein aggregates, autophagy is considered to be a protective mechanism of the pathophysiology of such neurodegenerative disease.

It is well known that Alzheimer's disease is associated with increased accumulation of autophagosomes in human brain tissue resulting from impaired autophagic activity [10]. Direct causal relationship between defective gene function and autophagy was also shown in recessive familial form of human Parkinson's disease, in which the mutation of PTEN-induced putative kinase protein 1 (PINK1) and PARK2 (the genes encodes parkin) proteins are involved in the mobilization of dysfunctional mitochondria to autophagosome, was demonstrated [62]. Animal experiments using mice also suggest that defective autophagy genes promote age-dependent neurodegeneration and pharmacologic treatment with autophagy enhancer can attenuate symptoms from neurodegeneration [63].

Autophagy itself is impaired by the protein aggregates from several neurodegenerative diseases. For example, Alzheimer' disease is associated with excess hyperphosphorylated microtubule associated protein tau. This leads to formation of neurofibrillary tangles and the accumulation of beta amyloid peptide, which impairs lysosomal function and autophagic clearance [10]. Similarly, Parkinson's disease is characterized by the accumulation of α-synuclein aggregate in Lewy bodies—neural cytoplasmic inclusions that impair autophagic clearance [64].

The role of autophagy in infection or inflammation can be categorized into several aspects. First, autophagy can directly degrade intracellular microorganisms through a process called xenophagy. Such process can involve autophagosomes that engulf certain types of microorganisms, which include shigella flexneri, salmonella enteric, group A streptococcus, mycobacterium tuberculosis, listeria monocytogenes, herpes simplex virus, toxoplasma gondii, etc. [10]. Other than that, it can enhance phagosome fusion with lysosome, thereby killing the organisms in toll-like receptor signaling pathway [65].

Autophagy also plays an important role in adaptive immunity by enhancing antigen presentation and lymphocyte development. For example, it was shown that deletion of Atg5, a key autophagy gene, in dendritic cells manifested impaired CD4⁺ T cell priming after herpes simplex virus infection [66]. Similarly, macrophage-specific Atg5 deletion was found to be associated with increased susceptibility to mycobacterium tuberculosis infection [67]. In this context, pharmacologic upregulation of autophagy is considered as an important strategy in immune response to the invasion of microorganisms and furthermore, invoked to the field of vaccine development. Indeed, it was demonstrated that rapamycin-induced autophagy enhanced mycobacterial antigen Ag85B presentation and promotes the efficacy of BCG vaccine in mouse dendritic cells [68]. Autophagy can suppress immune response by downregulation of proinflammatory cytokine response to invading pathogen.

Recently, association between derangement in autophagy genes and susceptibility to certain infections or inflammatory diseases are also identified and among them, mutations in autophagy regulators and Crohn's disease are best characterized. Single nucleotide polymorphisms in ATG16L1, nucleotide-binding oligomerization domain containing protein 2, and immunity-related p47 guanosine triphosphate M protein were identified to increase the risks Crohn's disease by genomewide association studies [69].

In contrast to the most beneficial effect of autophagy in neurodegenerative disease and infection or inflammation conditions, the role of autophagy in malignancy is more complex and can interact as a double-edged sword and multifactorial influences depending on tumor type or context. Earlier study by Liang et al. [9] showed that beclin1 was mono-allelically deleted in 40% to 75% of human breast and ovarian cancers and increased beclin 1 activity inhibited MCF7 cellular proliferation, indicating decreased expression of autophagy proteins may contribute to the pathogenesis of breast and other human malignancies. Mice experiments also proved that monoallelic loss of beclin1 lead to tumoriogenesis [70]. Given the fact that defective autophagy induces mitochondrial dysfunction, making normal cells subject to DNA damage and instability, autophagy may accomplish its role as a safeguard to prevent the occurrence of malignancy.

However, from the view point of cancer cells already established, autophagy can be used as survival strategies to enhance tumor proliferation and invasion and to acquire resistance against chemotherapy or radiation. For example, in primary pancreatic cancer and its cell lines, which showed elevated autophagy under basal conditions, autophagy inhibition by silencing essential autophagy gene ATG5 or pharmacologic agent resulted in significant growth suppression [71]. Of note, autophagy confers the anticancer drug resistance and therefore inhibition of autophagy may sensitize the tumor cells to anticancer therapy. Indeed, several autophagy inhibitors, such as 3-methyladenine, chloroquine and bafilomycin A1, sensitize the tumor cells to therapeutic agents in colorectal, lung, esophageal cancer and leukemia and are currently under study in many clinical trials [72].

Autophagy is involved in metabolism of carbohydrate, protein, and lipid. Recent researches have suggested dysregulation of autophagy is implicated in several metabolic diseases such as diabetes, obesity, and atherosclerosis [73].

Dyregulation of autophagic activity in pancreatic β-cells was associated with insulin resistance and the pathogenesis of type 2 diabetes. Indeed, increased autophagosome formation was observed in mice β-cells in an insulin-resistant state caused by a high-fat diet and also ob/ob mice, a several diabetic rodents model [74,75]. Since pancreatic β-cells are responsible for extensive proinsulin biosynthesis with higher burden of protein-folding, an increase of ER stress due to protein misfolding is also commonly occurred, inducing autophagy. It was suggested that activation of autophagy by increased ER stress in pancreatic β cells plays a protective role against cell dysfunction and death. This may be similarly applied to syncytiotrophoblast from placenta (will be discussed in Part II), which possess high burden of placental hormone biosynthesis, which can lead to excessive ER stress. It can be in this context that metformin, an antidiabetic drug, activates autophagy by inhibition of mTORC1 [76].

Intracellular lipids are stored in lipid droplets were known to be degraded by autophagy (lipophagy). Autophagy inhibition in cultured hepatocytes increased cellular triglycerides content and lipid droplet number and size in response to a lipid challenge [77]. It was demonstrated that excessive tissue lipid accumulation such as hepatic steatosis is caused by impaired lipophagy [78].