Monitoring Autophagy Flux and Activity: Principles and Applications

Macroautophagy is a major degradation mechanism of cell components via the lysosome. Macroautophagy greatly contributes to not only cell homeostasis but also the prevention of various diseases. Because macroautophagy proceeds through multi‐step reactions, researchers often face a persistent question of how macroautophagic activity can be measured correctly. To make a straightforward determination of macroautophagic activity, diverse monitoring assays have been developed. Direct measurement of lysosome‐dependent degradation of radioisotopically labeled cell proteins has long been applied. Meanwhile, indirect monitoring procedures have been developed. In these assays, autophagosome marker proteins, microtubule‐associated proteins 1A/1B light chain 3B‐II (LC3B‐II) and gamma‐aminobutyric acid receptor‐associated protein‐II (GABARAP‐II) have been analyzed and the validity of the assays strongly depends on appropriate assessment of the fluctuation of LC3‐II and/or GABARAP‐II levels in the presence or absence of lysosomal inhibitors. This article describes these monitoring methods, paying special attention to the principles and characteristics of each procedure.


Introduction
Homeostasis of living organisms is maintained by a dynamic flow of continuous synthesis and degradation of body constituents. Ubiquitin-proteasome (UPS)-dependent protein degradation and autophagy, two major degradative systems, make a critical contribution to maintaining cellular homeostasis. Autophagy is a mechanism by which intracellular components are degraded via the lysosome. While the UPS is specialized for selective degradation of ubiquitylated proteins, autophagy is involved in the degradation not only of cell proteins, but also nucleic acids, lipids, and carbohydrates.

DOI: 10.1002/bies.202000122
There are three categories of autophagy: microautophagy, chaperon-mediated autophagy, and macroautophagy. Microautophagy is a process in which a small portion of cytoplasm is directly engulfed by invagination of an endosome or lysosome to be degraded. [1] Chaperon-mediated autophagy is a selective degradation process; cytoplasmic proteins harboring a KFERQ motif are recognized by heat shock cognate 71 kDa protein and directly transported across lysosomal membranes into the lysosomal lumen through the interactions with the lysosome-associated membrane protein type 2A. The polypeptide substrates incorporated in this way are then degraded by lysosomal proteinases. [2] Last, macroautophagy is the main route by which cytoplasmic organelles and proteins sequestered in an autophagosome are transported to the lysosome to be degraded. [3][4][5] Quantitatively, the contribution of macroautophagy to the turnover of cell components is the most significant of the three modes. In addition, studies conducted over a quarter of a century have revealed that macroautophagy plays an essential role in maintaining normal and healthy functions of tissues and cells. Impairment of macroautophagy is associated with diverse diseases, including cancer, neurodegenerative diseases, inflammatory diseases, etc. [6][7][8] Hence interest in estimating macroautophagy activity is increasing in many realms of biological and biomedical research. However, the methods for quantitative determination of autophagy activity are rather complicated, and experimental results frequently tend to be misinterpreted. To quantify autophagy activity, measurement of "autophagy flux" or the rate of autophagic degradation is necessary. Basically, this can be accomplished by direct determination of lysosome-dependent degradation of intracellular proteins. Traditionally, quantification of free amino acids released via autophagy was a sole and most reliable procedure in the early days of the research. However, often this approach is technically unachievable, depending on experimental materials (animals, tissues, or cells) and experimental conditions. Increasingly, alternative approaches are becoming available, such as indirect determination based on imaging, biochemical, and morphological analysis. New approaches, which target autophagosomal membrane marker proteins, microtubule-associated proteins 1A/1B light chain 3B (LC3B), and gamma-aminobutyric acid receptor-associated protein (GABARAP), have been vigorously developed. Technical aspects of these new monitoring methods of autophagy activity were www.advancedsciencenews.com www.bioessays-journal.com extensively discussed in recent expert reviews by Klionsky et al. as well as by Yoshii and Mizushima. [9,10] In this article, we focus on the principle, versatility, and simplicity of monitoring procedures for understanding autophagy flux directly or indirectly.

Macroautophagy Is a Membrane Dynamic Process Supported by Six Functional Units
Macroautophagy (hereafter referred to "autophagy" in a simplistic form) occurs constitutively (basal or constitutive autophagy), but is substantially enhanced under stress conditions, such as nutrient starvation (induced autophagy). Autophagy proceeds in a stepwise fashion. First, a lipid-rich crescentic membrane called isolation membrane appears in the cytoplasm and extends by engulfing organelles and cytosolic proteins to form a doublemembraned autophagosome. Then, the autophagosome fuses with the lysosome to become a single-membraned autolysosome. Finally, sequestered cytoplasmic organelles and proteins are degraded by lysosomal hydrolases and the resulting amino acids, sugars, and fatty acids are secreted out from the autolysosome into the cytoplasm for metabolic reuse. The sequence of these membrane dynamic phenomena is corroborated by the functions of core ATG proteins, which are classified into six functional units: the serine/threonine-protein kinase ULK1 complex; the class III phosphatidylinositol 3-kinase complex; the ATG9 vesicle; the phosphatidylinositol 3-phosphate-binding protein complex; the ubiquitin-like protein ATG12 conjugation system; and the LC3 conjugation system (Figure 1A,B). [11][12][13] The LC3-conjugation system has important roles in both elongation and closure of the isolation membrane as well as the recognition of selective substrate for autophagy. ATG8 family proteins comprise LC3A, LC3B, LC3C, GABARAP, gamma-aminobutyric acid receptor-associated protein-like (GABARAPL)1, and GABARAPL2. Both ATG8 family proteins are reliable markers that localize on isolation membranes and autophagosomes dependent on their lipidation and degraded in the lysosome together with the cellular components ( Figure 1B). Meanwhile, each ATG8 subfamily has distinct function(s) in membrane tethering, fusion activities, and substrate specificities. [14] For example, LC3s are involved in elongation of the phagophore membrane whereas the GABARAPs are essential for a later stage in autophagosome maturation. [15] GABARAPs are the primary drivers of starvation-induced autophagy and PINK1-Parkin mediated mitophagy. [16][17][18] LC3C plays an important role in autophagy on bacterial-invaded cells through the specific interaction of Calcium-binding and coiledcoil domain-containing protein 2 (also known as NDP52). [19]

Autophagic Activity Can Be Determined by Measuring Free Amino Acids Released via Autophagic Proteolysis
Quantification of amino acids released via autophagic degradation of cell proteins has been regarded as a strongly reliable biochemical method, provided that the degradation is enhanced under nutrient-deprived conditions and inhibited by lysosomal proteinase inhibitors. The procedure has been applied to perfused organs as well as cultured cell system.

Determination of Amino Acid Release Using Organ Perfusion
The lysosome was first discovered in rodent livers in 1955 and subsequent electron microscopic observation of autolysosomes in rat hepatocytes became a milestone for establishing autophagy concept. So, during earlier eras of autophagy research, many studies performing liver perfusion experiments were published. To obtain clear-cut experimental data with a high signal/noise ratio, usually radioactive leucine or valine was injected intraperitoneally to rats or mice to label total liver proteins. Then, anesthetized animals were immobilized and kept alive in the perfusion device. Liver was perfused first with nutrient rich medium, for example, an artificially conditioned plasma. Next, to induce autophagy, perfusion medium was switched to nutrientdeprived medium, such as Hanks' balanced salt solution or Krebs-Ringer bicarbonate buffer in the presence of cycloheximide to inhibit liver protein synthesis. Perfusate was collected to determine the rate of amino acid (radioactivity) release.
Some important characteristics of autophagy, including inhibition by insulin and amino acids and the activation by glucagon have been demonstrated using sophisticated perfusion experiments by Mortimore's group. [20] In general, liver perfusion needs skilled techniques and considerable care to obtain reproducible experimental results. Hence, liver perfusion experiments were gradually replaced by more easy-to-handle experiments using cultured cells including primary hepatocytes as described in the next section. Even though, organ perfusion is an advantageous approach on its own to measure in vivo autophagy flux. More recently, the experimental perfusion system has attracted new attention in line with the growing importance of artificial organs. A rapid technical progress in the perfusion devices has enabled small liver slices to be likened to living liver organ. [21]

Degradation Assay of Long-Lived Proteins in Cultured Cell System
Autophagy is thought to be involved in the turnover of long-lived cell proteins. Degradation rate of long-lived proteins (percentage of total per hour) has been used as a convenient parameter for estimating the magnitude of autophagic proteolysis. A simple and convenient experimental protocol can be excerpted in common from published articles. First, adherent cells, such as primary hepatocytes, HepG2, and HeLa cells, are cultured in microplates with the base medium containing a radioactive amino acid to label total cell proteins (>18 h) (Figure 2A). [22] Usually radioactive leucine ( 14 C-leucine) or valine ( 14 C-valine) is used. Since branched chain amino acid-specific aminotransferase is expressed only in skeletal muscles, leucine, valine, and isoleucine released from autophagic protein degradation in most of cultured cells cannot be catabolized and remain unchanged. Thus, radioactivity released into cultured supernatant truly corresponds to that of leucine or valine. After the labeling, the cells are incubated with the base medium containing excess nonradioactive leucine or valine (1-2 mm) for 1 to 2 h (chase medium). The chasing is necessary to prevent reincorporation of released radioactive amino acids to cell proteins. Also, during the chase period, radioactive short-lived proteins become completely degraded. Finally, the cells are incubated with either Figure 1. A) Mammalian autophagosome formation governed by core ATG proteins. Metabolic stresses, including nutrient and growth factor deprivation, elicit inactivation of mechanistic target of rapamycin complex 1 (mTORC1) and activation of 5′-AMP-activated protein kinase (AMPK), which coordinately enhances ULK1 kinase activity. The active ULK1 kinase complex, consisting of ULK1, ATG13, RB1-inducible coiled-coil protein 1 (RB1CC1, also known as FIP200), and ATG101, translocates to sites near the endoplasmic reticulum and assembles. Next, ATG9 vesicles derived from the trans-Golgi network and plasma membrane recruit to super-assembled ULK1 kinase complex and participate in autophagosome precursor formation. The class III phosphatidylinositol 3-kinase (PI3K) complex, which is required for subsequent nucleation of the isolation membrane, consists of vacuolar protein sorting 34 (VPS34), BECN1, ATG14L, p150, and NRBF2. phosphatidylinositol 3-phosphate (PI3P) (blue diamonds) generated by the active class III PI3K complex recruits the PI3P-binding protein zinc finger FYVE domain-containing protein 1 (ZFYVE1, also known as DFCP1) to the pre-autophagosomal structure (PAS). DFCP1 promotes the formation of the omegasome, from which isolation membranes are generated. Another PI3P-binding protein complex WIPI2 recruits ATG2 to the isolation membrane. ATG2 tethers between ER membrane and the isolation membrane and also has a lipid-transfer activity, contributing to the expansion of the isolation membrane. Ubiquitin-like protein ATG12 is activated by the E1 enzyme ATG7, transferred to the E2 enzyme ATG10, and subsequently covalently conjugated with ATG5. The resulting ATG12-ATG5 conjugate forms in a complex with ATG16L1. Another group of ubiquitin-like proteins, ATG8 family proteins including microtubule-associated proteins 1A/1B light chain 3 (LC3), are cleaved by ATG4 in their C-termini to expose a glycine residue. Similar to ATG12, C-terminal cleaved LC3 (LC3-I) is activated by ATG7, and subsequently transferred to ATG3. LC3 is finally conjugated with phosphatidyl ethanolamine (PE) on the isolation membrane with the help of Atg12-Atg5 conjugates in a complex with ATG16L1, since Atg12-Atg5-ATG16L1 complex acts as an E3-ligase for LC3. B) Role of ATG8 family proteins (LC3 and GABARAP family proteins) in autophagy. The LC3-conjugation system has important roles in both elongation and closure of the isolation membrane as well as the recognition of selective substrate for autophagy. ATG8 family proteins including LC3 comprise LC3A, LC3B, LC3C, gamma-aminobutyric acid receptor-associated protein (GABARAP), gamma-aminobutyric acid receptor-associated protein-like (GABARAPL)1, and GABARAPL2. These ATG8 family proteins localizing on inner autophagosomal membranes are degraded in the lysosome together with the cellular components, and thus ATG8 family proteins have been employed to evaluate autophagic activity. The ATG8 family proteins interact with a battery of adaptor proteins, which have binding ability to both selective substrates and LC3 and/or GABARAP proteins. Some adaptor proteins, including p62 and Calcium-binding and coiled-coil domain-containing protein 2 (also known as NDP52), also interact with core ATG proteins such as FIP200, promoting autophagosome formation around the specific substrates. The ATG8 family proteins and adaptors, together with substrates, are degraded by autophagy. As in the case of ATG8 family proteins, the level of adaptors such as p62 has been used to estimate the autophagic activity.  Figure 2. A) Schematic representation of autophagic protein degradation (an example of cultured primary hepatocytes). Isolate hepatocytes from rat or mouse liver using collagenase perfusion procedure. [22] Seed hepatocytes into 24-well microplates at 5 × 10 4 cells per well. Culture the hepatocytes with Williams E medium containing 10% fetal calf serum (Williams E/10%FCS) (0.5 mL per well) overnight. i) Labeling the primary hepatocytes with 14 C-leucine. Replace the medium with fresh Williams E/10% FCS containing 14C-leucine (0.5 µCi Ml −1 , labeling medium). Culture the cells for 18-22 h. Discard the labeling medium. Wash the wells with 20 mm Na-phosphate (pH 7.5)-0.15 m NaCl (1 mL per well). ii) Chasing and incubating the cells under starved or non-starved conditions. Add 0.5 mL of Williams E/10% FCS containing 1 mm leucine to each well and incubate the plate for 90 min. Discard the chasing medium. Wash the wells with 20 mm Na-phosphate (pH 7.5)-0.15 m NaCl (PBS). iii) Culture under starvation medium. Add starvation medium (Krebs-Ringer bicarbonate buffer containing 1 mm leucine) or Williams E/10% FCS to the wells in the presence or absence of inhibitors (0.5 mL per well). For quantification of degradation, at least three wells must be used for a single condition (triplicate). Incubate the microplates at 37°C for 3 h. iv) Determination of protein degradation. After the incubation, retrieve 400 mL of cultured supernatant from each well and mix with 100 mL of 50% trichloroacetic acid (TCA). Centrifuge the mixtures at 10 000 rpm for 5 min. Retrieve 250 µL of the supernatant to determine radioactivity using a liquid scintillation counter. Discard the remaining cultured supernatant with an aspirator and wash the wells with PBS (1 mL per well). Add ice-cold 10% TCA to the wells and immediately remove TCS. Add 1 N NaOH to the wells (0.5 mL per well). Incubate the microplates at 37°C for 10 min in a moist chamber. Retrieve 250 µL of the solution to determine radioactivity using a liquid scintillation counter. From the radioactivity determined at step1 (S), one can calculate the CPM of released leucine as S × 2 × 1.25. From the radioactivity determine at step 2 (C), the CPM remaining in the cells as 2 × C. Protein degradation (percentage of total) is S × 2 × 1.25 / [(S × 2 × 1.25) + 2 × C] × 100. B) Degradation of long-lived nutrient-rich or nutrient-deprived medium for a couple of hours. Care must be taken to add 1 mm leucine or valine to nutrientdeprived medium, because a significant portion of released radioactive amino acid is reincorporated for protein synthesis under nutrient-deprived conditions (absence of extracellular amino acids), causing underestimation of protein degradation. After the incubation, cultured supernatant is collected, mixed with trichloroacetic acid (TCA), and centrifuged to spin down denatured proteins. The radioactivity of released amino acid in the supernatant is determined. Meanwhile, the cells on the microplate are fixed with 10% TCA. The fixed cells are then solubilized with NaOH and the radioactivity remaining in the cells is determined separately. Protein degradation (percentage of total) can be calculated as [(total radioactivity released in cultured supernatant) / (total radioactivity released in cultured supernatant) + (total radioactivity remaining in the cells)] × 100. Maximal rate of protein degradation (percentage of total per hour) is acquired with the cells incubated with nutrient-deprived medium, which is much larger than the degradation under nutrient-rich conditions ( Figure 2B). The rate is markedly diminished in the presence of [(2S,3S)-3-Ethoxycarbonyloxirane-2-carbonyl]-L-leucine (3-methylbutyl)amide (E64d) (lysosomal cysteine proteinase inhibitor), Pepstatin A (an inhibitor for lysosomal aspartic proteinases, including cathepsin D), and 20 mm NH 4 Cl (membrane-permeant salt to neutralize lysosomal pH), the combination that maximally inhibits lysosomal proteolysis. The presence of Bafilomycin A 1 (an inhibitor of lysosomal acidification) by itself is also effective. The difference in the degradation between the presence or absence of inhibitor(s) corresponds to the magnitude of starvation-induced autophagy ( Figure 2B).

Diverse Monitoring Procedures on the Basis of the Fluctuation of Autophagosomal Membrane Marker Proteins
Direct determination of autophagic flux, pursued either by organ perfusion or radioisotope tracer experiment requires specialized equipment or a radiation-controlled area. Instead, more simple procedures feasible at a conventional laboratory to monitor autophagy flux have been developed. In considering these assay procedures, it is important to give a close look at the behavior of LC3-II, a sole, universal autophagosomal membrane marker protein, during autophagy process. First, the precursor form of LC3 (proLC3) is synthesized de novo. proLC3 is then processed to form LC3-I, in which the carboxyl-terminal glycine of proLC3 is cleaved by ATG4, an autophagy specific proteinase. LC3-I, a soluble protein, is then conjugated with proteins under starved and non-starved conditions. The histograms shown herein are basically summarized from the data published in ref. [22]. phosphatidylethanolamine via autophagy-specific conjugation reactions by ATG7 and ATG3 to become LC3-II ( Figure 1A). Behaving as an integral membrane protein, LC3-II is distributed in both inner and outer autophagosomal membranes. Once autophagosome matures into autolysosome, LC3-II in the inner autophagosomal membrane is digested by lysosomal proteinases ( Figure 1B). Thus, LC3-II level increases upon induction of autophagosome formation, but then decreases after fusion of autophagosome with lysosome. Monitoring this LC3-II transition provides a key clue to create indirect methods to measure autophagy flux. As will be described below, autophagy-monitoring by means of biochemical, imaging, and morphological analyses targeting to LC3-II have been invented enthusiastically.
In induced autophagy, the number of autophagosomes begins to increase within 10 min after starvation, accompanied with an increase in autophagosomal LC3-II. On the other hand, prompt transition from autophagosome to autolysosome after fusion of autophagosome with lysosome causes degradation of LC3-II in the inner autophagosomal membrane, resulting in a decrease in LC3-II. Furthermore, it should be noted that formation of individual autophagosomes occurs asynchronously or randomly during the starvation periods. Therefore, the quantity of total LC3-II that can be monitored at the steady state of autophagy reflects the sum of autophagosomal LC3-II and autolysosomal LC3-II on the way to degradation. In other words, LC3-II level itself, when determined by various procedures, is not directly related with autophagy flux. To ascertain whether autophagy flux is truly activated, monitoring lysosomal turnover of LC3-II is most important and, for this purpose, the use of lysosome-specific inhibitors is essential. If autophagy is induced under starvation for a certain period (usually more than 2 h), in the presence of Bafilomycin A 1 or E64d plus Pepstatin A, LC3-II accumulates markedly compared with control (in the absence of inhibitors), due to inhibition of LC3-II degradation in autolysosomes. The increment of LC3-II should correspond to the magnitude of autophagy flux (Figure 3A).

Monitoring Autophagy Flux Using Western Blotting Analysis of LC3-II and p62
Monitoring endogenous LC3-II in vitro and in vivo by western blotting is the most popular assay for assessing autophagy flux ( Figure 3A, lower part). Cultured cells grown with base medium (nutrient-rich) are transferred to fresh base medium or starvation medium in the presence or absence of various lysosomal inhibitors and incubated for 4-12 h. Prolonged treatment should be avoided due to possible secondary effects. Cells are harvested and total cell lysates are prepared and subjected to western blotting. Under starvation conditions, significantly more accumulation of LC3-II is recognized in the presence of Bafilomycin A 1 , concanamycin A, or E64d plus Pepstatin A than that in their absence (control). [9] Autophagy occurs at basal level under nutrient-rich conditions, so the difference in LC3-II accumulation between the presence and absence of these inhibitors is marginal except for thymic epithelial cells [23] and tumor cells. Note that many tumor types demonstrate elevated basal autophagy, and this can be seen in a cell-autonomous manner, even under nutrient-replete conditions. [24,25] In such tumor cells, the accumulation of LC3-II by the treatment of abovementioned reagents should be prominent.
LC3-II western blotting can be applied also to the estimation of in vivo autophagy flux, too. Administration of leupeptin, a lysosomal cysteine proteinase inhibitor, or chloroquine, an inhibitor of lysosomal acidification, to starved rats and mice, has been implemented. [26][27][28][29] Leupeptin is preferentially incorporated into the hepatocytes. In case of western blotting of LC3, the accumulated autolysosomes can be detected as an increase in LC3-II in liver homogenate. [26,27] When administered intraperitoneally, leupeptin elicits abundant accumulation of autolysosomes in hepatocytes. [30,31] Due to cessation of autophagic proteolysis, the density of autolysosomes in leupeptin-administered liver becomes much higher than that of other cytoplasmic organelles. The heavier autolysosomes can be purified by centrifugation in Percoll gradients. [30] Enzyme activity of cytosolic enzyme such as lactate dehydrogenase in isolated autolysosomes can be determined as a sequestration marker of autophagy. [30,31] Based on this observation, LDH sequestration assay was developed. [32] Intraperitoneally administered chloroquine is on the other hand effectively incorporated into cardiac muscle to cause accumulation of autolysosomes, which can be detected as increased level of LC3-II in western blotting analysis. [28] Interestingly, colchicine that is thought to inhibit autophagosome-lysosome fusion, can be also used to assess autophagy flux in mouse skeletal muscle. [29] In parallel with monitoring LC3-II, degradation of p62/ SQSTM1 (hereafter as p62), has been routinely investigated. [33] Although p62 can be degraded by endosomal-related autophagy under certain circumstances, [34] it is primarily degraded by selective autophagy. [35] Accordingly, similarly to LC3, measuring its level has been a common method for monitoring autophagic flux. [36] Fluctuations on p62 can be studied even in in vivo with a p62-GFP knock-in mice (p62-GFP KI/+ ), in which C-terminal GFPtagged p62 is driven by endogenous p62 regulatory elements, and thus the turnover of endogenous p62-GFP in an autophagydependent fashion can be detected by immunoblot analysis. [37] p62 serves as an adaptor protein for selective autophagy of ubiquitinated cargos ( Figure 1B). [38] When its activity as an autophagy adaptor is not required, p62 is held inactive through homodimerization of its UBA domain, [39] which prevents it from interacting with ubiquitin. Phosphorylation events drive the liberation of the UBA domain from dimeric inactivation; in particular, phosphorylation of serine 407 of p62 by ULK1 has been shown to facilitate the transition from dimer to monomer. [40] This modification is followed by phosphorylation of p62 on serine 403, either by ULK1, casein kinase 2, [41] or TANK-binding kinase 1, [42] which enhances its binding to ubiquitin chains. Once p62 binds to ubiquitin chains, it acquires liquid-like properties. [43] Such phase-separated droplets allow the exchange of their components, including ubiquitin and LC3, with the surrounding environment. [43][44][45] Considering the analogy to Ape1 transport into vacuole through selective autophagy in yeast [46] and the cases of parkin-mediated mitophagy [47] and CCPG1-mediated ER-phagy, [48] upstream autophagy factors such as the ULK1-kinase complex are predicted to be translocated onto the droplets. Actually, a recent report revealed that RB1inducible coiled-coil protein 1 (RB1CC1, also known as FIP200), a component of the autophagy-initiation complex interacts with On the other hand, RFP-LC3-DG remains in the cytoplasm, due to the lack of the carboxyl-terminal glycine necessary for the conjugation with PE. During starvation-induced autophagy, GFP-LC3-PE becomes quenched and degraded by lysosomal proteinases during the autophagosome maturation into autolysosome, resulting in diminished green fluorescence. Hence, the ratio of fluorescence intensity between GFP and RFP decreases. D) A classical view of autophagy-related vesicular/vacuolar structures. In transmission electron microscopy, autophagosomes are recognized as an electron-lucent particle with double membranes (at high magnification) with clear cytoplasmic organelles sequestered in the lumen. Meanwhile, autolysosomes are identified as single-membraned particles with partly degraded material in the lumen. A small population of double-membraned vesicles contains autolysosomes immediately after fusion of autophagosomes with lysosomes, but degradation has yet to begin (transition state). Autophagosomes and these transitional autolysosomes are collectively classified as nascent or initial autophagic vacuoles (AVi). Autolysosomes at later stages are called degradative autophagic vacuoles (AVd). Active staining of lysosomal acid phosphatase enables distinction between autophagosomes and autolysosomes.
p62 to initiate autophagy. [49] Once the pathway is activated, p62-positive structures together with the recruited cargos are degraded by autophagy in an LC3 and/or GABARAP-binding dependent manner ( Figure 1B). Considering that master gene regulators Nuclear factor erythroid 2-related factor 2, NF-B and Transcription factor EB upregulate the expression of p62/SQSTM1 in response to various stresses [50] as well as the abovementioned scenario, it is plausible that estimation of phosphorylated p62 level rather than total amount of p62 reflects autophagic activity [51] (manuscript in preparation).

Fluorescence Microscopy/Imaging Analysis
Monitoring induced autophagy in living cells and tissues using fluorescence microscopy was first developed by Mizushima et al. using GFP-LC3 transgenic mouse as well as cultured cells overexpressing GFP-LC3. [52] In liver, heart, and skeletal muscles, induction of autophagosomes under starvation for 24 h was clearly demonstrated. [52] Since then, monitoring autophagosome induction was investigated extensively with cultured cells overexpressing GFP-LC3 using fluorescence cytochemistry (for a review, see ref. [53]). Autophagosomes can be detected as green fluorescent dots. Increase in these dots per cells is a good index for autophagy induction. GFP-LC3 of inner autophagosomal membrane is delivered to autolysosomes, resulting in a decrease in the fluorescence, but the fluorescence increases in the presence of lysosomal acidification inhibitors. Hence, similarly to the monitoring LC3-II in western blotting, fluorescence cytochemical analysis of GFP-LC3 can be used for monitoring autophagy flux in combination with the use of inhibitors. [54] GFP emits strong fluorescence at neutral pH, but the fluorescence is quenched in acidic milieu (pH less than 5.5), such as in autolysosomes. Hence, GFP-LC3 is not a stable reporter for the autophagosomal maturation to autolysosome. To circumvent the weakness of GFP-LC3 fluorescence, a new assay using a mRFP-GFP tandem fluorescent-tagged LC3 (mRFP-GFP-LC3) was invented for monitoring autophagy flux more precisely. [55] mRFP fluorescence is stable at pH 4∼5. In addition, mRFP is more resistant than GFP to lysosomal proteinases, so even after autolysosomal degradation of mRFP-GFP-LC3, red fluorescence of RFP, a degradation product, remains in the lumen. When overexpressed mRFP-GFP-LC3 is recruited onto the autophagosome. mRFP-GFP-LC3 emits both red and green fluorescence, because the pH in the autophagosomal lumen is ≈7. Thus, a merged image of fluorescence microscopy shows yellow signal. Acidification of the lumen after fusion of autophagosome with lysosome causes quenching of GFP fluorescence. As a result, the merged image shows mRFP red fluorescence, confirming autophgosomeautolysosome transition (autophagy flux) ( Figure 3B). Conversely, in the presence of Bafilomycin A 1 , yellow fluorescence in merged image is still observable. One of the main advantages of mRFP-GFP-LC3 is that it can be used for monitoring autophagy flux more clearly than GFP-LC3 without using lysosomal inhibitors. For in vivo analysis, transgenic mice overexpressing tandem RFP-GFP-LC3 were generated. [56,57] In kidney proximal tubules, ischemia-reperfusion injury elicited marked increase in RFP-GFP-LC3 puncta with both green and red fluorescence, indicating upregulation of autophagy. [56] Three days after, the number of GFP puncta decreased, whereas RFP puncta persisted. Thus, autophagy flux elapsed for 3 days after the injury. [56] In another transgenic mouse, in which RFP-GFP-LC3 was expressed selectively in neurons using Thy1 promoter, red fluorescence puncta were present, while yellow puncta are scarce at the basal state. [57] This indicates autophagy-related vesicles mainly consist of autolysosomes. Increase in yellow fluorescence was observed when chloroquine was administered. [57] Based on these works, we expect that RFP-GFP-LC3 transgenic mouse models will bear a great importance for monitoring in vivo autophagy activity in future years.
Monitoring GFP-LC3 and RFP-GFP-LC3 images traces the dynamic membrane transition from autophagosome to autolysosome. Meanwhile, a fluorescent protein, Keima, has been used for monitoring the transition of an autophagy substrate during autophagosomal maturation into autolysosomes. Keima, a fluorescent variant of a protein from the stony coral, has a unique fluorescence characteristic. [58,59] The emission spectrum of Keima peaks at 620 nm, whereas the excitation spectrum exhibits bimodal peaks at 440 and 586 nm. At neutral pH, the fluorescence intensity excited at 440 nm is much higher than that excited at 586 nm. In contrast, at acidic pH (pH ≈5), the fluorescence intensity excited at 586 nm is much higher than that excited at 440 nm. [59] Thus, in the fluorescence microscopy, the excitation peak ratio (586 nm/440 nm) is a good indicator for understanding acidity around Keima. When transfected into mammalian cells, Keima distributed diffusively in the cytosol has a lower excitation peak ratio. Upon autophagy induction, punctate Keima sequestered into autolysosomes via autophagosome has a higher excitation peak ratio (586 nm/440 nm). As Keima is resistant to lysosomal proteinases, the fluorescence dots with a higher excitation peak ratio persist longer, which renders the probe reliable for analyzing autophagosomal maturation into autolysosome. [59] As a potential application, Keima fused with mitochondrial-targeting signal (mt-Keima) has been used for time-lapse imaging analysis of parkin-dependent selective degradation of mitochondria (mitophagy) in cultured cells. [59] For in vivo analysis, transgenic mice expressing mt-Keima were generated and basal mitophagic activity in liver, neural tissues, and heart under various conditions were investigated. [60]

RFP-LC3-GFP-LC3 G, a New Probe for Quantifying Autophagy Flux
GFP-LC3-RFP-LC3ΔG is a fusion protein of GFP-LC3 and RFP-LC3ΔG (RFP-LC3 with deleted carboxyl-terminal glycine). [61] When overexpressed in the cells, GFP-LC3-RFP-LC3ΔG is cleaved by Atg4 to produce GFP-LC3 and RFP-LC3ΔG. Upon induction of autophagy, GFP-LC3 is conjugated with PE and recruited onto autophagosomes, and subsequently quenched and degraded in autolysosomes. Meanwhile, due to deletion of carboxyl-terminal glycine, RFP-LC3ΔG is unable to undergo autophagy-specific conjugation reaction and remains in cytosol. As Atg4-dependent cleavage of GFP-LC3-RFP-LC3ΔG produces equal number of GFP-LC3 and RFP-LC3ΔG, RFP fluorescence can be used as an internal standard. When GFP-LC3-RFP-LC3ΔG overexpressed cells is placed under starvation conditions to induce autophagy, GFP/RFP ratio becomes decreased along with an increase in autophagy flux. Accordingly, the GFP/RFP ratio has been used as a key index for screening autophagy activators and inhibitors. [61]

Electron Microscopy
Electron microscopic analyses of lysosomes in 1950s revealed that some lysosomes contained partially degraded cellular organelles, such as mitochondria. This observation hinted at the role of lysosome in digesting self-components of the cell, that is, autophagy. Indeed, many studies conducted to characterize the structures of autophagosome, autolysosome, and lysosome in these earlier days were dependent on electron microscopy. [62] In these assays, the term "autophagic vacuole" was used in general as the collective designation of autophagosome and autolysosome. [63] Autophagic vacuoles are further classified into nascent or initial autophagic vacuoles (AVi) and degradative or late autophagic vacuoles (AVd) ( Figure 3D). AVi consist of autophagosomes shortly after formation and autolysosomes immediately after the fusion of autophagosomes with lysosomes/endosomes, but degradation of sequestered cytoplasmic components has not yet begun. AVi is characterized by intact structures of sequestered organelles in the lumen. AVd consist of autolysosomes containing partially degraded cytoplasmic components and characterized by the presence of lysosomal acid phosphatase, which can be detected histochemically by Gomori lead procedure. Morphometry of acid phosphatase-negative autophagic vacuoles and acid phosphatase-positive autophagic vacuoles was used for monitoring the transition process from autophagosome to autolysosome. [64] Electron microscopic analysis can make a strong contribution to distinguishing autophagy-related membrane structures under both physiological and pathological states of cells and tissues. Autophagosomes as clear double-membraned vesicles with intact organelle structures sequestered in its lumen, autolysosomes as single-membraned vesicles with dimmed structures of sequestered components, and electron-dense smaller lysosomes have long been characterized. In wild-type cells, autophagosomes increase temporally for a short while after induction of autophagy, then more autolysosomes can be detected. Notably, at the steady state under prolonged starvation, not many autophagosomes/autolysosomes can be observed due to rapid turnover of the vesicles, unless lysosomal proteinase inhibitors are present. [63] More recently, electron microscopy has been enthusiastically utilized to characterize abnormal autophagy-related structures under pathological conditions. Briefly, the structures are categorized in two types. One is typically detected in autophagy-inactive tissues, in which numerous curved membranes and incomplete autophagosome-like membranes accumulate in the cytoplasm. [65][66][67][68] Another example is detected in mouse tissues with deficiency in lysosomal hydrolases, as typically observed in lysosomal storage diseases [69,70] abundant lysosomal vacuoles with amorphous debris, irregular vesicles with electron-dense osmiophilic materials in the lumen, and granular inclusions accumulate in the cytoplasm. As autolysosomal degradation is severely hindered, some of these accumulated vesicles seemingly correspond to malfunctional autolysosomes in state of indigestion. Electron microscopy is a beneficial technique for tracing whether autophagy flux occurs normally or is retarded at some steps.

Novel Autophagy Monitoring Assays Using Atg8-Sensor Probes
There are two modes of autophagy: bulk and selective autophagy. [71] While the former is responsible for metabolic adaptation, the latter one contributes to cellular homeostasis. [71] Both modes are driven by core ATG proteins, but a key difference between them is whether the autophagosome randomly sequesters cytoplasmic components or it forms along specific cargos. [72] Specific autophagic cargos such as liquid-droplets, damaged or excess organelles and aggregated proteins appear in the cytoplasm and are subsequently tagged with ubiquitin, leading to assembly of adaptor proteins such as p62, Next to BRCA1 gene 1 protein, NDP52, and Optineurin that bind to the ubiquitin chain. [72] Alternatively, trans-membrane type of adaptor proteins such as BCL2/adenovirus E1B 19 kDa proteininteracting protein 3-like (also known as NIX), Reticulophagy regulator 1 (also known as FAM134), and Testis-expressed protein 264 directly localize on cargos. In the case of the transmembrane adaptors, molecular markers like ubiquitin are not needed. [5] Core autophagy-related proteins such as FIP200 also recognize the labeled targets through the interaction with adaptor proteins, [49,73,74] beginning the process of autophagosome formation around the targets. Ubiquitin-binding and transmembrane type adaptor proteins, both have LC3-interacting region(s)/GABARAP-interaction motif(s), called LIR/GIM to interact with autophagosome-localizing ATG8 family proteins (including LC3A, LC3B, LC3C, GABARAP, GABARAPL1, and GABARAPL2) around the cargos. [75] The interaction of ATG8 family with selective substrates is indispensable for tethering cargos to elongating isolation membranes/phagophores ( Figure 1B). [75] Both LC3 and GABARAP are reliable markers that localize on isolation membranes and autophagosomes dependent on their lipidation. In addition, the specific function(s) for each ATG8 family protein has been gradually unveiled. For example, LC3C plays an important role in autophagy on bacterial-invaded cells through the specific interaction of NDP52. [19] Recently, two groups independently developed autophagy probes focusing on the LC3 and GABARAP-binding peptides. [76,77] Lee's group noticed that the LIR motif of p62 is insufficient to localize on isolation membranes and autophagosomes. They fused the LIR with a short hydrophobic domain of Aplysia phosphodiesterase 4 short-form and GFP, generating HyD-LIR-GFP constructs. The fusion protein efficiently bound to LC3-II and GABARAP-II localizing on the inner membrane of autophagosomes and did not affect autophagosome biogenesis at least in cell lines (Figure 4). They screened HyD-LIRs that possess each LIR of 34 LC3-interacting proteins and showed that while a sensor with an LIR of TP53INP2 preferentially recognizes GABARAP-positive autophagosomes, a sensor with an LIR of FYCO efficiently localizes on LC3A-and LC3B-positive ones. [76] Using these probes, they demonstrated that both probes recognize autophagosomes surrounding damaged mitochondria (mitophagosomes). [77] Dikic's group conducted a different way to identify the LC3and GABARAP-binders. They initially specified LC3-and GABARAP-binding peptides using an unbiased peptide phage display that has been utilized for a peptide-binding profiling of protein domains. The peptides were modified by multiplication, charge distribution, and fusion with a FYVE, which serves for membrane recruitment through the binding to PI3P, or with a oligomerizing PB1 domain. These optimizations as well as tagging of mCherry allowed them to develop specific probes for each LC3A and LC3B, and GABARAPs and LC3C (Figure 4). [77] They further produced an LC3C-specific probe and showed the specific function of LC3C during xenophagy. [77] These probes binding to each ATG8 family protein become powerful tools not only to trace the autophagosomes and evaluate selective autophagy, but also to investigate the roles of each ATG8 family protein.

Conclusion and Outlook
In many published studies, induction of autophagosomes under stress conditions is often mistakenly described as "autophagy activation." As described earlier, increase in autophagosome/ autolysosome number by itself suggests induction of autophagy, but the data cannot tell us anything about the dynamic flow of autophagosome to autolysosome transition or autolysosomal degradation. To ascertain whether autophagy flux is truly accelerated, careful experiments with or without inhibitors of autophagosome-lysosome fusion or lysosomal degradation must be performed. This complexity is reminiscent of a scene in which a person stands still on a river bank. The flow of the river is quite slow and almost unrecognizable. But once the flow is stemmed by closing the sluice gate downstream to keep the water back, the flow of the river can be sensed or recognized clearly by the increasing pressure on the gate.
There are the methodological difficulties and limitations in both direct and indirect procedures. As described in the main text, the direct measurements mainly rely on amino acids released from lysosomes or on cytosolic enzymes transported in lysosomal lumen. Thus, in addition to their technical complication, we always need to pay attention to consequences of other lysosomal degradation pathways such as endocytosis, microautophagy, and chaperon-mediated autophagy. Meanwhile, the indirect procedures are dependent on the amount of LC3-II and GABARAP-II incorporated in inner membrane of autophagosomes. Under some stress conditions, autophagy proceeds independently of ATG5 and ATG7. [78] Though less compared with that in wild-type cells, fusion between autophagosome and lysosome is observed in ATG3-deficeint cells, even in simple starvation conditions. [79] Since conversion of LC3-I and GABARAP-I to LC3-II and GABARAP-II is totally dependent on ATG5, ATG7, and ATG3, current indirect monitoring assays focusing on LC3 and GABARAP cannot be applied to measure such ATG8-conjugation independent autophagy.
In view of the rapid progress in the field of autophagy research, it is important to establish a simpler and more streamlined procedure for monitoring autophagy activity in future investigations. In the yeast, Saccharomyces cerevisiae, a genetically modified form of vacuolar alkaline phosphatase (Pho8Δ60p) was expressed in the cytosol as an inactive precursor. [80] Under starvation conditions, Pho8Δ60p is sequestered into autophagosomes and transported to vacuoles, in which the precursor Pho8Δ60p is processed to an active mature form with alkaline phosphatase activity. Increase in alkaline phosphatase activity is closely related to starvation-induced autophagy, therefore this experimental system is convenient for measuring autophagic activity by a simple enzymatic assay. Similarly, new assays to quantify mammalian autophagy need to be developed in the near future in order to allow a better understanding of the process.