Small-molecule autophagy modulators: Mechanisms, therapeutic potential, and development challenges in pathological settings

Abstract

Autophagy is an evolutionarily conserved process in eukaryotic cells that delivers intracellular components to lysosomes for degradation and recycling. Increasing evidence has elucidated the regulation of autophagy, highlighting its involvement in cellular metabolism, survival, and development, as well as its association with diverse physiological and pathological processes. There are often mutations in autophagy-regulating genes or abnormal autophagy function in multiple diseases, such as cancer, immune system diseases, and neurodegenerative diseases. Additionally, the regulation of the autophagy process shows potential therapeutic effects for these diseases. Several small molecules have been developed as autophagy regulators based on traditional drug discovery strategies, such as high-throughput screening, structure-activity relationship (SAR) optimization, and computer-aided drug design. Mechanistically, these compounds that bind specifically to such autophagy-related proteins or kinases can act as agonists or antagonists, with downstream consequences on the autophagy process. Several pharmacologic agents that regulate the autophagy process with extraordinary potential in disease treatment have come into clinical use. But most of these molecules still suffer from many obstacles, including low efficacy, low selectivity, poor pharmacokinetic profile, drug resistance, and toxicity. Moreover, some inappropriate and undruggable autophagy-related targets, as well as ubiquitous protein aggregates in neurodegenerative diseases, also bring serious challenges to the identification of small-molecule drugs. In this review, we briefly introduce autophagy and summarize its function and regulatory role in various diseases and disorders, and discuss the possibility of autophagy-targeted therapy in these diseases. The present review highlights current developments regarding the fundamental molecular mechanisms and signaling cascades of autophagy, while also addressing strategies for small-molecule–based therapeutic intervention.

1. Introduction

In 1974, Prof. Christian won the Nobel Prize in Physiology or Medicine for the earliest discovery of lysosomes. In several follow-up studies, the research team of Prof. Christian documented an important phenomenon that intracellular components are encapsulated in membranes to form double-membrane vesicles, which are eventually transported to lysosomes for degradation and recycling. Therefore, Prof. Christian named this special life process “autophagy”, and the double-membrane vesicle structure was named “autophagosome”. Due to its specific role in mammalian cells, autophagy has gained the attention of the research community. In the 1990s, Yoshinori Ohsumi and colleagues identified several autophagy-related genes (ATG) using the Saccharomyces Cerevisiae yeast model. Additionally, homologous genes related to nearly all yeast ATGs have been identified in higher eukaryotes, including humans, significantly advancing autophagy-based research. In 2003, Klionsky et al. designated these genes as ATGs and investigated the interactions between the encoded proteins and their functions in autophagy. Subsequently, Yoshinori Ohsumi and his team used yeast mutants to identify key ATGs and elucidated the mechanisms and significance of autophagy. In 2016, Professor Ohsumi was awarded the Nobel Prize for his groundbreaking work on elucidating the mechanisms of autophagy, highlighting the growing recognition of this biological process [1]. Autophagy is an evolutionarily conserved pathway essential for maintaining cellular homeostasis. This process includes lysosomal degradation and the recycling of misfolded proteins, damaged organelles, and cytoplasmic components in eukaryotes.

Apoptosis is known as type I programmed cell death (PCD), and its morphological changes include nuclear shrinkage, nuclear fragmentation, and nuclear lysis. Recently, it has been confirmed that there is a complex interactive regulation between autophagy and apoptosis. Autophagy is also a PCD that is different from apoptosis and termed type II PCD. Notably, autophagy engages in complex interplay with apoptosis, and many stress-related signaling pathways sequentially induce autophagy and apoptosis within the same cell. Specifically, autophagy leads to a stress adaptation that inhibits apoptosis and ensures cell survival. However, in some cases, excessive autophagy can induce cell death, referred to as autophagic cell death [2]. Recent evidence indicates that autophagy deficiency or dysfunction is closely linked to the occurrence and progression of various diseases, including tumors, neurodegenerative disorders, cardiovascular diseases, metabolic disorders, and infectious and immune diseases. Meanwhile, ATG genes fulfill diverse physiological functions across various cellular pathways and protect mammals from aging and numerous diseases. Recent studies confirm that mutations in ATG genes are associated with the pathogenesis of human diseases [3,4]. Consequently, investigating the complex regulatory mechanism of autophagy helps understand its role in diseases. Furthermore, it also provides the theoretical foundation for the development of novel autophagy modulators. With the development of biology, molecular biology, and biomedicine, the exploration of novel autophagy modulators with structural and functional diversity has moved into a new epoch. The research directions in the field of medicinal chemistry in the past decade have shown that designing and synthesizing small-molecule compounds that selectively target the autophagy mechanism and evaluating the potential therapeutic effects of these compounds in human diseases are considered to be new, clinically practical therapeutic strategies. In this review, we first summarize the functional and regulatory roles of autophagy in various diseases and disorders and explore the potential for autophagy-targeted therapies in these conditions. Specifically, this review offers an overview of recent advancements in the core molecular machinery and signaling pathways involved in autophagy and discusses the development of small molecule-based targeted therapy. Given the key role of the autophagy process in the development and progression of diseases, the identification of autophagy modulators may provide efficacious treatment strategies for preclinical and clinical applications. From this perspective, the research progress of therapy strategies for a variety of diseases by targeting autophagy is summarized, focusing on the structure-activity relationships (SAR) and biological activities of autophagy modulators. In addition, we discussed challenges and future directions in the identification of more potent and highly selective autophagy modulators.

2. Types, mechanisms, and physiological functions of autophagy

2.1. Types of autophagy

Autophagy represents a phylogenetically conserved intracellular catabolic pathway indispensable for the preservation of cellular homeostasis; within mammalian systems, three mechanistically distinct variants—macroautophagy, microautophagy, and chaperone-mediated autophagy (CMA)—have been definitively characterized [5] (Table 1).

Table 1.

Three distinct types of autophagy and their core mechanisms.

Type	Description	Core mechanism	Key features
Macroautophagy	Sequestration of cytoplasmic material within double-membrane vesicles (autophagosomes), which fuse with lysosomes for degradation.	1. Autophagosome formation: Cytoplasmic contents enclosed by double membranes. 2. Lysosome fusion: Autophagosomes fuse with lysosomes. 3. Recycling: Degraded products are utilized for biosynthesis or ATP generation.	1. Double-membrane vesicles: Autophagosomes are double-membrane structures. 2. Involvement of lysosomes: Fusion with lysosomes to degrade content. 3. Multifunctional: Involved in the degradation and recycling of a variety of cellular components.
Microautophagy	Direct invagination or tubulation of the lysosomal membrane to sequester and degrade cytoplasmic content inside the lysosome.	1. Membrane invagination: Lysosomal membrane directly sequesters cytoplasm. 2. Non-selective degradation: Degradation occurs within the lysosome.	1. No autophagosome formation: Direct degradation through lysosomal membrane invagination. 2. Non-selective: No specific selection of degradation substrates.
Chaperone-mediated autophagy	Selective degradation of proteins containing a KFERQ-like motif, mediated by chaperones that facilitate substrate translocation into the lysosome.	1. Substrate recognition: Chaperones bind to KFERQ-like motifs. 2. LAMP2A binding: The complex binds to LAMP2A on the lysosomal membrane, facilitating translocation. 3. Lysosomal degradation: Substrate degradation in the lysosome.	1. Selective degradation: Only proteins with the specific KFERQ-like motif are degraded. 2. No autophagosome formation: Does not require the formation of autophagosomes, relying solely on lysosomal degradation.

Macroautophagy, commonly referred to as autophagy, is the main and most widely studied type of autophagy. The general process behind macroautophagy is that the Golgi apparatus or cytoplasmic membrane encapsulates cytoplasmic contents to form a double-membrane vesicle, namely an autophagosome. Autophagosomes thereafter converge with lysosomes to generate degradative autolysosomes whose luminal acid hydrolases disassemble sequestered substrates into their constituent metabolites; these catabolites are subsequently effluxed to the cytosol and re-routed either to anabolic pathways for de novo macromolecular synthesis or to mitochondrial oxidative phosphorylation for ATP generation [6].

In microautophagy, cytoplasmic constituents are sequestered via direct invagination or tubulation of the lysosomal limiting membrane and are degraded inside. Microautophagy constitutes a non-selective lysosomal degradative pathway integral to biosynthetic trafficking, metabolic adaptation, organellar remodeling, and proteostatic quality control. In contrast, CMA is a selective autophagic process in which all substrate proteins share a common feature: a KFERQ or KFERQ-like pentapeptide motif that is recognized by heat shock cognate 70 (Hsc70, also known as HSPA8) and other accessory chaperones. After the chaperone-substrate complex binds to the receptor, lysosome-associated membrane protein type 2A (LAMP2A), on the lysosomal membrane, the substrate unfolds and translocates into the lysosomal lumen, where it is finally degraded [7]. It is worth noting that only macroautophagy can form autophagosomes of double-membrane vesicles; however, neither microautophagy nor CMA requires the participation of autophagosomes, but relies on the degradation function of lysosomes (Fig. 1).

Fig. 1.

Three types of autophagy. a) Macroautophagy; b) Chaperone-mediated autophagy (CMA); c) Microautophagy.

2.2. Molecular mechanism of autophagy

Autophagy is a complicated intracellular process, and its molecular mechanisms were first discovered in yeast, which is highly conserved in mammals. Over 30 ATG proteins encoded by ATG genes, along with their associated regulators and proteins, are involved in this process [4]. The classic autophagy process mainly includes five steps: (1) autophagy induction, (2) nucleation and expansion of the phagophore, (3) autophagosome formation, (4) autophagosome trafficking and fusion with lysosomes, and (5) degradation of the autophagosome.

2.2.1. Autophagy induction

Autophagy is elicited by diverse stimuli encompassing extracellular cues—ischemia, hypoxia, growth-factor deprivation, nutrient scarcity, and microbial invasion—as well as intracellular perturbations such as metabolic stress, senescence-associated organelle deterioration, and the accretion of misfolded or aggregated proteins. Notably, different triggers can induce autophagy through different pathways to maintain cellular homeostasis. Certain proteins play important roles in the autophagy process by forming multiple subunit complexes, among which the ATG1/ULK1 protein kinase complex is essential for autophagic initiation [8]. The ATG1 protein is encoded by the ATG1 gene, and its homologous protein in mammals is Unc-51-like kinase 1 (ULK1), whose activity is critical to the recruitment of autophagy-related proteins. The ULK1 complex consists of the kinases ULK1, ATG13, FIP200 (FAK family kinase-interacting protein of 200 kDa), and ATG101 [9]. The ATG1/ULK1 complex is directly regulated by the autophagy center regulator mammalian target of rapamycin (mTOR). Under nutrient-replete conditions, active mTORC1 phosphorylates and suppresses the ATG1/ULK1 kinase complex; upon nutrient withdrawal, mTORC1 is inactivated, permitting ULK1 dephosphorylation and ensuing kinase activation. Activated ULK1/2 then phosphorylates ATG13 and FIP200, thereby nucleating autophagosome formation [10]. Autophagy nucleates at the pre-autophagosomal structure (PAS) in yeast; in mammals, the process originates within an endoplasmic reticulum (ER)–derived phosphatidylinositol-3-phosphate (PI3P)-enriched subdomain designated the omegasome. The activated ATG1/ULK1 complex is recruited to PAS as a platform to recruit other ATG proteins under the action of proteins such as ATG13, ATG17, ATG29, ATG31, Yeast Rab1 (Ypt1), and transport protein particle III (TRAPPIII), and can phosphorylate many downstream proteins at the same time [11].

2.2.2. Nucleation and expansion of the phagophore

Once autophagy is activated, the phagophore (also known as the isolation membrane) becomes nucleated and expands to form a cup-shaped double-membrane structure. The class III phosphatidylinositol 3-kinase (PI3KC3) complex constitutes a pivotal autophagy initiation node and a primary ULK1 substrate. Upon ULK1-mediated phosphorylation, PI3KC3 converts phosphatidylinositol to PI3P, thereby generating the PI3P-enriched membrane platform indispensable for autophagosome nucleation [12]. The vacuolar protein sorting 34 (VPS34), VPS15, and Beclin1/ATG6 catalytic subunits are found in both the PI3KC3-C1 and PI3KC3-C2 forms of the PI3KC3 complex. ATG14L/ATG14, which is present in PI3KC3-C1, leads the complex to PAS and promotes phagophore nucleation and elongation, while UVRAG-containing PI3KC3-C2 promotes autophagosome maturation and trafficking [13].

ATG9, a transmembrane protein with six transmembrane domains, is connected to both phagophore nucleation and elongation. Additionally, the transport vesicle between the Golgi and the endosome is home to the autophagy-related protein ATG9, which controls autophagy by affecting vesicle transport. ATG9 vesicles are recruited to PAS by TRAPPIII and the ATG1/ULK1 complex and merge with COPII vesicles for nucleation [12]. Notably, Phagophore nucleation is contingent upon the bidirectional trafficking of ATG9 between the nascent phagophore and the peripheral ATG9 reservoir [14]. Following nucleation, the phagophore undergoes progressive expansion and elongation to generate an intact autophagosome; ATG9, ATG2, and ATG18 converge at the extending rim of the phagophore, where they cooperatively mediate membrane accretion and play crucial roles [15]. ATG2 is essential for the growth of the phagophore due to its lipid transfer activity that can move lipid molecules from the ER to the autophagic membrane. The ATG18 protein is a PI3P-binding protein that assists in the binding of the ATG9 complex to the autophagic membrane. The ER was previously assumed to be the source of membranes during autophagy since the PAS structure and growth of the autophagic membrane were usually observed close to it. However, new research has demonstrated that membranes may also come from mitochondria, the Golgi, and circulating endosomes [16].

One of the distinguishing features of autophagy is the phagophore, a lipid double membrane structure that resembles a flat pocket and can be seen under an electron microscope. Phagophore elongation further requires the sequential operation of two ubiquitin-like conjugation cascades: the ATG12–ATG5–ATG16 complex assembly and the lipidation of ATG8/LC3 with phosphatidylethanolamine (PE) [17]. In both systems, ATG7 serves as a ubiquitin-activating enzyme (E1-like enzyme), whereas ATG10 and ATG3 serve as ubiquitin-conjugating enzymes (E2-like enzymes), respectively, for ATG12 and ATG8. The ATG12-ATG5-ATG16 ubiquitin linkage system is an alternative name for the ATG12 conjugation process. Once synthesized, ATG12 binds with ATG5 and ATG16 to form the ATG12-ATG5-ATG16 complex, which can attach to the autophagic membrane and function as a platform for interactions with proteins involved in phagophore elongation. In addition to promoting the elongation of the phagophore, which eventually develops from a tiny two-layer membrane structure to a pocket, the binding of the ATG12-ATG5-ATG16 complex is in charge of attracting LC3-II to the autophagic membrane by combining LC3-I with PE to produce LC3-II [18].

The ATG8/LC3-PE conjugation cascade, also termed LC3 lipidation, encompasses two paralogous subfamilies: the microtubule-associated protein 1 light chain 3 (LC3) isoforms (LC3A, LC3B, and LC3C) and the γ-aminobutyric acid receptor–associated protein (GABARAP) subfamily (GABARAP, GABARAPL1, and GABARAPL2/GATE-16) [19,20]. Specifically, LC3 isoforms predominantly orchestrate phagophore elongation, whereas GABARAP/GATE-16 members are indispensable for autophagosomal maturation. Upon autophagy induction, nascent ATG8 orthologs are translated as C-terminally extended pro-forms (pro-LC3) that are proteolytically processed by the cysteine protease ATG4 to yield cytosolic LC3-I bearing an exposed C-terminal glycine residue. Then, when combined with PE under the action of ATG7, ATG3, and ATG12-ATG5-ATG6, LC3-II is produced and located on the phagophore and autophagosome as a hallmark of autophagy. Additionally, the removal of ATG8 from the phagophore membrane is also mediated by ATG4. According to research, the ATG1/ULK1 complex binds to ATG8 family proteins through the ATG8 family protein interaction motif (AIM) or LC3 interaction region (LIR) and phosphorylates ATG4 to inhibit the activity of ATG4 on ATG8 family protein delipidation and promote phagophore expansion.

2.2.3. Autophagosome formation

After the phagophore initiation, it continues to elongate and wrap the parts of the cytoplasm that need to be degraded into pockets, such as damaged organelles and misfolded or aggregated proteins. Finally, the autophagosome, a double‐membrane‐bound vesicle, is formed by closing the phagophore membrane through the machinery of the endosomal sorting complex required for transport (ESCRT). The ATG protein bound to the autophagic membrane dissociates before the membrane closes, but LC3 and its family members remain on the interior surface of the autophagosome and are degraded by the hydrolase in the lysosome after the fusion of the autophagosome with the lysosome. Autophagic activation elicits a marked accumulation of total LC3 and a pronounced shift from cytosolic LC3-I to lipidated LC3-II; quantification of this LC3-II increment therefore serves as a reliable biochemical indicator of autophagy induction or suppression. Functionally, ATG8 molecules situated on the inner autophagosomal membrane are delivered to lysosomes for degradation, whereas those on the outer membrane are recycled before lysosomal fusion. ATG8 on the inner membrane binds to the LIR motif of autophagy receptors located on the cargo or bound via the ubiquitin chain through the LIR docking site (LDS) so that selective autophagy can be achieved, and this process also requires the participation of P62/SQSTM1 [21,22]. ATG8 at the outer membrane is involved in autophagosome trafficking and fusion with lysosomes.

2.2.4. Autophagosome trafficking and fusion with lysosomes

Following maturation, autophagosomes undergo microtubule-based transport to perinuclear lysosomal microdomains, where their outer membranes fuse with lysosomal limiting membranes to generate degradative autolysosomes [23]. To ferry nascent autophagosomes through the cytoplasm, the cell leans on its two favorite highways: actin filaments and microtubules. Among the cargo couriers, LC3-tethered receptors such as nuclear dot protein 52 (NDP52), optineurin (OPTN), and TAX1BP1 hitch a ride on actin tracks, sliding along myosin motors as if catching an express subway to the lysosome. The microtubule and its motor proteins can interact with LC3-II localized on autophagosomes through the LIR motif of FYVE and the coiled-coil domain containing 1 (FYCO1), the effector of RAB7, thereby driving the transportation of autophagosomes to the cell periphery [24].

Autophagosome–lysosome merger is no solo act. Soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs) zip the two bilayers together, small GTPases give the green light, and assorted tethering proteins act like bouncers, making sure only compatible partners meet [25]. In the case of SNARE-mediated autophagosome-lysosome fusion, four types of SNARE motifs, Qa, Qb, Qc, and R, provided by each of the SNAREs form a parallel four alpha-helix bundle. Specifically, the Q-SNAREs, N-ethylmaleimide-sensitive factor syntaxin 17 (STX17) and synaptosomal-associated protein 29 (SNAP29), at first form the Qabc bundle on the autophagosome, which then complexes with lysosome-localized R-SNAREs, such as vesicle-associated membrane protein 7/8 (VAMP7/8). Finally, SNARE complexes promote the fusion of the autophagosome and lysosome with the assistance of various regulators, including the homotypic fusion and protein sorting (HOPS) complex, ATG14, divergent protein kinase domain 2A (DIPK2A), and ULK1 [25]. RAB7A deploys two effectors in quick succession. First, pleckstrin homology and RUN domain-containing M1 (PLEKHM1) drags HOPS to the lysosome, letting STX17 slip into the SNARE bundle; then ectopic P-granule autophagy protein 5 homolog (EPG5)—also under RAB7A control—locks onto LC3 on the autophagosome, steadying the whole complex so the membranes can fuse [26]. Autophagosomes can also fuse with endocytic vesicles such as phagosomes or endosomes to form an amphisome; however, amphisomes are not involved in all autophagy pathways.

2.2.5. Degradation

The outer membrane of the autophagosome fuses with the lysosomal membrane to form an autolysosome. At the same time, lysosomal/vacuolar hydrolases degrade the inner membrane and contents of the autophagosome to produce amino acids, fatty acids, and other materials that are recycled back into the cytosol where they can be reused by the cell as energy sources or as building blocks for the synthesis of macromolecular substances such as proteins and lipids, while the residues that cannot be recycled may be excreted out of cells or retained in the cytoplasm [27].

2.3. The function of autophagy

Autophagy is a process that exists widely in eukaryotic cells. As a “double-edged sword”, it is not only a self-protection mechanism of cells but also a PCD that is parallel to apoptosis and necrosis, known as type II PCD. The importance of autophagy in physiological functions of cells, including material metabolism balance, cell homeostasis, cellular waste removal, structural reconstruction, growth, and development, is amply supported by research [28].

2.3.1. Maintain cellular homeostasis

Meeting shifting metabolic demands is autophagy’s oldest job, so cells keep the pathway ticking at just the right rate even when life is easy. Push them with starvation or hypoxia, and the gears turn faster: proteins, lipids, and glycogen are shredded into amino acids, fatty acids, and sugars, a quick-and-dirty refuel that buys time until the next meal arrives. These metabolites can subsequently enter metabolic pathways to generate ATP and serve as building blocks for the synthesis of proteins and lipids, thereby supporting both energy production and macromolecular biosynthesis. Autophagy quietly balances the cellular ledger, yet when its gears slip, the fallout spreads far beyond the single cell. Misfolded proteins linger, lipid droplets overstay, and metabolic signals blur, setting the stage for diabetes, obesity, and atherosclerosis [29,30].

In post-mitotic neurons, the absence of cell division leaves no easy exit route for misfolded proteins; they pile up, turn toxic, and quietly set the scene for Parkinson’s, Huntington’s, or amyotrophic lateral sclerosis (ALS). Studies now show that misfolded, clumped proteins steadily poison motor neurons, inching them toward cell death. Moreover, changes in the autophagy activity of cardiomyocytes accompany the pathological processes of cardiovascular diseases [31]. Desmin-related cardiomyopathies, such as late-onset cardiomyopathy with pathological remodeling of the heart, can result from the accumulation of misfolded proteins in cells, as in the case of the mutation of the major intermediate filament protein desmin α-B-Crystallin (CryAB), which is present in the myocardium, to CryAB^R120G. Additionally, different types of myopathies require autophagy to function in different ways. On the one hand, autophagosome and lysosome fusion obstacles result in autophagosome accumulation, which may cause vacuolar myopathy. On the other hand, induction of autophagy acts as a defense against certain types of diseases. As an illustration, the deficiency of α-glucosidase in lysosomes contributes significantly to the etiology of type II glycogen storage disorder (Pompe disease), which results in the accumulation of glycogen and the subsequent malfunction of the heart and skeletal muscles. However, autophagy can reduce intracellular glycogen buildup and avoid causing muscle injury [32].

Degrading damaged organelles and preventing apoptosis are two further crucial tasks performed by autophagy. To maintain cellular homeostasis, autophagy specifically eliminates malfunctioning cellular components without affecting normal organelles. Mitochondria run the cell’s power grid, churning out ATP that fuels virtually every pathway inside [33]. Damage to the mitochondria can cause the generation of reactive oxygen species (ROS) or apoptotic factors, which can harm cells or cause apoptosis. Fortunately, selective autophagy may eliminate damaged mitochondria (known as mitophagy), thereby preserving cellular homeostasis and preventing apoptosis.

The generation of ROS via respiratory chain leakage is a type of single-electron oxygen reduction product in organisms. The impacts of ROS on DNA, proteins, and biological membranes include DNA breakage and mutation, amino acid modification, peptide chain breakage, protein cross-linked polymer formation, protein structure and immunogenicity changes, lipid peroxidation, and membrane integrity destruction [34]. When ROS strike mitochondria head-on, they nick mitochondrial DNA (mtDNA), loosen mitochondrial membrane permeability (MMP), and uncouple the electron transport chain. Each insult spawns more radicals, locking the organelle into a self-feeding loop of oxidative injury. Autophagy lowers oxidative stress by selectively digesting protein clumps and battered mitochondria, trimming the very sources of ROS rather than merely mopping up the aftermath. Beyond curbing DNA lesions by throttling ROS, autophagy recycles the repair crew itself and, when the cell needs to slam on the brakes, turns its own cargo into fuel that keeps the cycle parked [35].

2.3.2. Immunity and inflammation

Autophagy and immunity now look less like distant cousins and more like roommates sharing the same playbook. When innate sentinels spot intruders, the pathway swings into action, corralling bacteria, viruses, and assorted pathogens straight to the lysosome for disposal. On the other hand, autophagy quietly shapes adaptive immunity, tuning lymphocyte numbers, fine-tuning effector programs, guiding lineage decisions, and keeping cell survival. Furthermore, autophagy may contribute to immune responses by controlling inflammation, engaging in antigen presentation, and boosting immune mediator release [36].

Inflammation is the body’s protective response activated by microbial pathogen infection or tissue damage [37]. Inflammatory disorders are frequently associated with defects in autophagy or mutations in proteins involved in autophagy. Inflammatory cues and autophagy talk back to each other. The moment stress signals rise, inflammasomes assemble, primed to cleave pro-interleukin-1β (IL-1β) and IL-18 into their secreted forms. Yet the same aggregates, ROS, misfolded proteins, or battered organelles that flip the inflammasome switch can also nudge autophagy awake, setting up a quiet brake that reins the inflammatory burst back in. However, autophagy can eliminate endogenous inflammasome agonists, thereby inhibiting inflammasome activation. Additionally, autophagy may reduce inflammation by directly degrading inflammasomes, reducing pro-IL-1, and contributing to the removal of apoptotic cells [38].

2.3.3. Autophagy and cell death

Numerous investigations have shown that ATG genes contribute to cell survival during stress-induced cell death. Furthermore, medicine or genetic intervention that inhibits autophagy activity might hasten cell death. When cells are starved, autophagy can promote the destruction of cytoplasmic components by lysosomes and provide substrates for redox processes and ATP generation to sustain cell life activities. Deletion of ATG genes directly led to the death of yeast under starvation, demonstrating that autophagy is essential for yeast survival [28,39]. The development of aging and aging-related diseases is further supported by mounting evidence that autophagy is a critical factor. When autophagy stalls, neurons and other long-lived cells turn into cluttered attics, brimming with misfolded and mutant proteins that quietly seed neurodegeneration and the broader spectrum of age-linked disorders. Data from multiple species now show that turning autophagy up, especially when nutrients run low, sweeps out senescent cells and battered organelles, steadies the intracellular milieu, and, in doing so, stretches lifespan.

The link between autophagy, apoptosis, and necrosis is exceedingly complicated and contentious, although, at normal levels, autophagy can protect cells and inhibit cell death [40]. Yet ramped-up autophagy can kill. Because the field now counts this as bona fide PCD, the pathway’s lethal credentials must be weighed head-on. The rub is telling when autophagy is the executioner versus a bystander that merely shows up as the cell dies, a distinction that calls for meticulous, context-by-context analysis. In the first scenario, the cell dies because autophagy pushes too hard; in the second, autophagy simply lingers at the scene while another lethal program does the actual killing [2,39]. The former possibility is that autophagy is induced when cells face an unfavorable environment. However, when autophagy is no longer able to aid cells in maintaining a stable internal environment, but instead consumes intracellular components at an excessive rate, it leads to PCD. Autophagy-related cell death may be the process of cell apoptosis and necrosis, accompanied by the occurrence of autophagy. There is a view that autophagy-dependent cell death and autophagy-related cell death are both present throughout the process of cell death and do not vary morphologically. However, it is feasible to differentiate between the two conditions by observing what happens to cells when autophagy is blocked: in the former, cells survive, but in the latter, cells perish.

Emerging evidence now links ferroptosis—an iron-driven, non-apoptotic route to programmed death—with autophagy, hinting that the two pathways share more than a casual acquaintance [41]. Ferroptosis is an iron-catalyzed death program in which rising intracellular Fe²⁺ drives lipid peroxide buildup and ROS amplification, pushing the cell past its redox tipping point. Increased autophagic flux in response to canonical activators of ferroptosis, such as erastin and RSL3, has been observed in a variety of cells. Nuclear receptor coactivator 4 (NCOA4) is a cargo receptor for autophagy-dependent ferritin degradation that enables ferritin to release free iron through the autophagy pathway (ferritinophagy), which is critical for regulating cellular iron levels [42]. According to one study, lipid droplet levels were negatively correlated with ferroptosis induced by oxidative stress. Ramping up lipophagy strips lipid droplets to their core, flooding the cytosol with substrates that fuel peroxidative chain reactions and push the cell toward ferroptosis [43]. Besides, Autophagy can also promote ferroptosis by degrading aryl hydrocarbon receptor nuclear translocator-like protein 1 (ARNTL) and blocking hypoxia-inducible factor 1 subunit alpha (HIF-1α)-dependent fatty acid uptake and lipid storage. The recently characterized autophagy receptor hippocalcin-like 1 (HPCAL1) preferentially directs cadherin 2 (CDH2) for lysosomal degradation under ferroptotic stress. By stripping CDH2 from the plasma membrane, HPCAL1 relaxes tension and removes a physical barrier that normally limits lipid peroxide propagation, tipping the balance toward ferroptotic cell death [44]. More importantly, there is overlap between the signaling pathways that regulate autophagy and ferroptosis. Beclin1, p53, adenosine monophosphate-activated protein kinase (AMPK), and p62, long known for tuning autophagic flux, now also modulate ferroptosis, hinting at a shared regulatory nexus [45]. Overall, the connection between ferroptosis and autophagy is quite intricate. A balanced autophagic tone shields the cell, yet when the rheostat drifts too high, cargo selection falters, or lysosomes lose their bite, the same pathway can quietly hand the cell over to ferroptosis.

2.3.4. Other functions

By degrading certain cell cycle proteins, regulating cell division, and boosting DNA damage repair, autophagy preserves genomic stability and cellular homeostasis. In turn, several cell cycle regulators, including cyclin-dependent kinases (CDKs), cyclins, Aurora kinases, and Polo-like kinases (PLKs), are involved in the regulation of autophagy. Additionally, the midbody ring (MR), a crucial ring structure in cytokinesis, may be eliminated by autophagy, indicating that autophagy may be crucial to cell division [46]. Several candidate antineoplastic agents drive autophagy past its comfort zone, unsettling cyclin pools and nudging cells into arrest, senescence, or outright autophagy-dependent death [46,47]. Pairing autophagy modulators with genotoxic agents, DNA repair blockers, or cell cycle poisons is fast becoming a workable game plan for cancer therapy.

Autophagy also has a significant impact on cell development, differentiation, and the maintenance of stem cell function. According to reports, autophagic activity remains at modest levels in unfertilized eggs, but dramatically rises 4 h after fertilization. The possible reason is that the types of proteins in the fertilized egg will change significantly in a short time. A certain level of autophagy can provide amino acids that are beneficial to protein synthesis, as well as the development and differentiation of fertilized eggs and embryos [48]. Furthermore, autophagy rapidly and effectively degrades intracellular enzymes, transcription factors, and other substances in response to the requirement to strictly control the number of intracellular proteins and organelles during stem cell self-renewal and differentiation. Work across several stem cell niches now says the same thing: autophagy quietly pulls the strings behind self-renewal, lineage choices, and the slow slide into senescence in hematopoietic, mesenchymal, and neural stem cells (NSCs) [49] (Table 2).

Table 2.

Physiological function of autophagy.

Major function	Sub-function	Key mechanism
Cellular homeostasis	Responds to metabolic demands such as hunger	Degrading macromolecules into simple small molecules such as monosaccharides, amino acids, and fatty acids to maintain protein synthesis and energy production
	Maintains protein quality	Degrading misfolded and abnormally aggregated proteins
	Maintains organelle quality	Eliminating excess, damaged, harmful, or ruptured organelles
	Responds to oxidative stress	Degrading oxidized or damaged proteins, lipids, DNA, and mitochondria
	Maintains genomic stability	1. Reducing oxidative damage and ROS level 2. Recycling key protein for DNA repair 3. Acting as a source of energy to maintain cell cycle arrest and to sustain DNA repair activities.
Immunity and inflammation	Participates in innate immune responses	Selectively eliminating pathogens and activating the immune response
	Regulates adaptive immunity	Regulating lymphocyte function, development, and differentiation, participates in antigen presentation, and promotes cytokine secretion
	Regulates inflammation	Suppressing excessive inflammatory responses by inhibiting inflammasome activation and reducing inflammatory responses
Cell survival and death	Promotes cell survival	Maintaining the balance of material and energy metabolism and maintaining cellular homeostasis
	Mediates cell death (apoptosis, autophagic cell death, necrosis, and ferroptosis)	Whether antagonize or promote each other, the mechanism is very complex
Other functions	Regulates cell development and differentiation	Degrading related substances and maintaining homeostasis
	Delays aging	Degrading related substances and maintaining homeostasis
	Regulates cell cycle	Selectively degrading specific cell cycle proteins

3. Signaling pathway of autophagy

Precise regulation of autophagy is critical for cells to respond to different external stimuli, such as nutrient deficiency, growth factor deficiency, and hypoxia. Cranking autophagy up wipes out the damage these stresses leave behind, then quietly dials itself back once the threat passes [50]. Pinning down how autophagy is switched on, dialed up, or reined in is the only way to see why some cells survive stress while others collapse. Mapping the wiring has turned up a tangle of signaling routes, each tugging at different levers (Fig. 2.) The next pages walk through these pathways, one node at a time.

Fig. 2.

Mechanisms and major signaling pathways of autophagy. Autophagic processes can be divided into five stages: autophagy induction, the nucleation and elongation of the phagophore, autophagosome formation, autophagosome trafficking and fusion with lysosomes, and degradation of the contents of the package.

3.1. Signaling pathways that induce autophagy

3.1.1. TOR signaling pathway

Target of rapamycin (TOR), as a highly conserved serine/threonine protein kinase, is a member of the phosphoinositide 3 kinase-related kinase (PKK)-related protein kinase family. As the central molecule in autophagy regulation, TOR may either increase or decrease autophagy in response to changes in internal cell conditions such as ATP concentration, hypoxia, and nutrients. MTOR is a homolog of TOR in mammalian cells that can be regulated by a variety of stimulatory signals and is involved in the regulation of cell growth, proliferation, apoptosis, autophagy, protein translation, and immunosuppression. mTOR splits its duties across two assemblies. Rapamycin tames mTORC1, the hub that balances growth, survival, energy, and autophagy. Rapamycin leaves mTORC2 untouched, leaving this complex free to sculpt the cytoskeleton and keep cells alive [51]. Notably, the current research on autophagy mainly focuses on mTORC1.

The mTORC1 complex consists of the core subunit mTOR, regulatory-associated protein of mTOR (Raptor), mammalian lethal with SEC13 protein 8 (mLST8), 40kDa proline-rich Protein kinase B (AKT) substrate (PRAS40), and DEP domain-containing mTOR-interacting protein (DEPTOR). Raptor sits on mTORC1 as the gatekeeper, tuning kinase output, while mLST8 quietly braces the whole assembly. Shift to mTORC2 and the scene changes: mTOR partners with rapamycin-insensitive mTOR chaperone (Rictor), alongside mammalian stress-activated protein kinase interacting protein 1 (mSIN1) and the mLST8/GbL, forging a signaling core that ignores rapamycin yet reorganizes cytoskeletal tracks and survival circuits [52]. Under nutritionally sufficient conditions, the ULK1 complex directly binds to the Raptor subunit of mTORC1, leading to phosphorylation of ULK1 and ATG13, which inactivates the ULK1 complex as well as inhibits autophagy. But following inactivation of mTORC1 under starvation conditions, ULK1 and ATG13 are dephosphorylated and dissociated from mTORC1, resulting in recovery of ULK1 activity. Then the phosphorylation of ATG13, FIP200, and ATG101 by activated ULK1 leads to the induction of autophagy. In general, phosphorylation of ATG13 and ULK1 increased in nutrient-enriched conditions, while in starvation conditions, phosphorylation of FIP200 increased [51]. Beyond nutrients, mTOR listens to a chorus of cues. Rapamycin clamps the catalytic site, PI3K/AKT and MAPK/ERK1/2 modulate upstream relays, and p53 adjusts the volume when stress or oncogenic signals demand a reply.

3.1.2. AMPK signaling pathway

AMPK, an ancient serine/threonine kinase, keeps cellular books balanced. When glucose dips, oxygen thins, or redox swings push the AMP/ATP ratio upward, the enzyme flips from accountant to alarm bell, dialing catabolism up and anabolism down while nudging autophagy to come online [53,54]. AMPK relies on more than falling ATP. Liver kinase B1 (LKB1) shoulders most of the workload when energy stalls, while calmodulin-dependent protein kinase kinase 2 (CAMKK2) listens to calcium chatter and transforming growth factor-β-activation kinase 1 (TAK1) adds its own twist under cytokine fire, together flipping the kinase switch that leads cells into autophagy. Upstream extracellular signals such as insulin, lipopolysaccharide (LPS), amino acids, hormones, and glucose can bind to the corresponding receptors and then increase intracellular Ca²⁺ levels, thereby activating CAMKK2. And TAK1 is now considered to play a role in a specific environment or stress signals [54]. Although there are several signaling cascades and regulatory mechanisms that might affect autophagy activity, the AMPK signaling pathway may be the most evolutionarily conserved.

AMPK steers metabolism two ways: it clamps or boosts key enzymes on the spot, and it tugs at transcriptional levers to reshape the metabolic map. AMPK nudges autophagy along two tracks. It grabs mTORC1 by the collar, gives ULK1 a quick phosphate kick, and taps PIK3C3/VPS34 to get the membrane ball rolling. When that is not enough, it leans on forkhead box O3 (FOXO3), transcription factor EB (TFEB), and even the chromatin reader bromodomain-containing protein 4 (BRD4), coaxing them to crank out the ATG transcripts, which the cell needs, for a full-scale clean up [54]. Additionally, the tuberous sclerosis complex 1/2 (TSC1/2) complex has GTPase activity that hydrolyzes GTP and inactivates the small GTPase RHEB (Ras Homolog Enriched in Brain), resulting in mTOR inhibition. When AMPK is activated by autophagy-inducing signals such as energy deficiency, on the one hand, the activated AMPK can phosphorylate and activate the TSC1/2 complex to indirectly inhibit mTORC1; on the other hand, AMPK can also directly phosphorylate Ser772 and Ser792 of Raptor to inhibit mTORC1. AMPK first quiets mTORC1, stripping the brake from ULK1 at Ser757. The freed kinase then partners with AMPK itself, picks up activating marks at Ser317 and Ser777, and autophagy rolls [10,54,55]. The PIK3C3/VPS34 complex plays an important role in the stage of phagophore nucleation, and increasing evidence suggests that AMPK phosphorylates the PIK3C3/VPS34 complex to promote autophagy. Specifically, AMPK phosphorylates Becin1 at Thr388 and promotes the binding of Beclin1 to VPS34 and ATG14L to induce autophagy under glucose starvation conditions. Furthermore, AMPK can also promote the stability of the PIK3C3 complex by phosphorylating progestin and adipo-Q receptor member 3 (PAQR3) and receptor for activated C kinase 1 (RACK1), which are related to the formation or stability of the PIK3C3/VPS34 complex, and subsequently induce autophagy. Another research demonstrates that AMPK may also trigger mitophagy and induce mitochondrial fission, both of which are essential for maintaining mitochondrial homeostasis (Fig. 3).

Fig. 3.

AMPK signaling pathway regulates autophagy.

3.1.3. MAPK signaling pathway

Mitogen-activated protein kinase (MAPK) cascade constitutes a highly evolutionarily conserved signaling hub that integrates diverse extracellular stimuli—including growth factors, cytokines, genotoxic stress, and microenvironmental changes—and transduces them into precise intracellular biological responses. This system operates through a classic three-tiered kinase phosphorylation amplification cascade, whereby MAP3K activates MAP2K, which in turn activates MAPK, ultimately leading to the specific phosphorylation of a broad array of downstream substrates. These substrates encompass transcription factors, cytoskeletal proteins, metabolic enzymes, and key regulators of apoptosis and autophagy, enabling precise control over core cellular processes such as proliferation, differentiation, stress adaptation, and programmed cell death. Within the highly dynamic catabolic process of autophagy, the three major MAPK branches—c-Jun NH2-terminal kinase (JNK), p38 MAPK, and extracellular signal-regulated kinase (ERK)—have been demonstrated to play particularly critical and complex roles [56]. Each exhibits a distinct activation profile: JNK primarily responds to genotoxic stress, ultraviolet radiation, and pro-inflammatory cytokines; p38 MAPK is highly sensitive to osmotic changes, inflammatory signals, and endoplasmic reticulum stress; while ERK is typically driven by growth factors through the canonical Ras-Raf-MEK axis. However, their downstream effector networks are extensively interconnected, collectively forming a dynamic and highly context-dependent regulatory network through direct or indirect modulation of transcriptional programs, metabolic states, and core autophagy machinery, such as the B-cell lymphoma 2 (Bcl-2)/Beclin1 complex, the ULK1 initiation complex, and the membrane trafficking system of mATG9 [57].

Concretely, the JNK pathway functions as a central stress sensor. Its activation not only disrupts the inhibitory interaction between Bcl-2 and Beclin1 by directly phosphorylating Bcl-2 at residues Thr69, Ser70, and Ser87—thereby releasing Beclin1 to initiate VPS34 complex-mediated autophagosome nucleation—but also systematically upregulates the expression of numerous autophagy-related genes, including Beclin1, at the transcriptional level [58]. This is achieved through the activation of transcription factors such as AP-1 (e.g., c-Jun/c-Fos) and the promotion of nuclear translocation and transactivation activity of FOXO family proteins, establishing a sustained positive feedback loop. In contrast, the regulatory logic of the p38 MAPK pathway exhibits more pronounced duality and microenvironment dependency. It can exert diametrically opposed functions depending on the cell type and stimulus context. For instance, in certain tumor cells facing proteasome inhibition, p38 MAPK induces protective autophagy by phosphorylating and inhibiting Glycogen synthase kinase 3β (GSK-3β), thereby promoting cell survival [59]. In nutrient sensing, the p38α isoform, in coordination with the p38-interacting protein, negatively regulates the vesicular localization of mATG9 to suppress basal autophagy under nutrient-replete conditions, while relieving this suppression during starvation to promote autophagosome biogenesis, underscoring its refined role as an integrator of metabolic and stress signals [57]. Although traditionally associated with pro-growth signaling, the ERK pathway also plays a profound and complex role in autophagy regulation. On one hand, under certain metabolic or pharmacological stress conditions, ERK activation can phosphorylate and activate TSC2, thereby inhibiting mTORC1 activity, releasing its suppression of the ULK1 complex, and indirectly inducing autophagy. On the other hand, ERK can directly phosphorylate the Gα-interacting protein, accelerating the GTP hydrolysis cycle of the Gα_i3 protein. Since the Gα_i3-GDP state is known to activate autophagy, this provides an alternative direct regulatory route [60]. Furthermore, sustained ERK activation may also exert inhibitory effects on autophagic flux through mechanisms involving modulation of LKB1 activity or other, not yet fully elucidated, feedback pathways [61].

3.1.4. Immune and inflammatory signaling

Immune and inflammatory signaling pathways constitute critical bridges linking extracellular environmental threats to intracellular autophagy activation. Cells utilize a repertoire of pattern recognition receptors (PRRs), such as Toll-like receptors (TLRs) located on the plasma or endosomal membranes and cytosolic NOD-like receptors (NLRs), to precisely detect pathogen-associated molecular patterns (PAMPs) from pathogens or damage-associated molecular patterns (DAMPs) released by tissue injury [62]. This recognition event can rapidly initiate the autophagy program, positioning it as a core component of innate immune defense and homeostasis maintenance. For instance, upon recognizing lipopolysaccharide (LPS), TLR4 transmits signals via its adaptor proteins myeloid differentiation primary-response protein 88 (MYD88) or TIR-domain-containing adaptor-inducing interferon-β (TRIF). The MYD88-dependent pathway recruits, activates and phosphorylates IL-1R-associated kinase (IRAK)-1/4 and tumor necrosis factor (TNF) receptor-associated factor 6 (TRAF6); the latter catalyzes K63-linked ubiquitination of Beclin1. This modification disrupts the inhibitory interaction between Beclin1 and Bcl-2, enabling Beclin1 to assemble the VPS34 complex essential for autophagy induction [63]. Concurrently, the TLR4-MYD88 axis can upregulate the expression of the selective autophagy receptor p62 by activating the transcription factor Nrf2, thereby enhancing clearance capacity. On the other hand, the TLR4-TRIF axis can recruit TRAF3 and activate TANK-binding kinase 1 (TBK1), which in turn phosphorylates the autophagy receptor OPTN, promoting its binding to LC3 and guiding intracellular pathogens into the autophagic degradation pathway. Unlike the membrane localization of TLRs, cytosolic NLRs, such as Nucleotide­binding oligomerization domain 1 (NOD1) and NOD2, can directly interact with the key autophagy protein ATG16L1 to initiate antibacterial autophagy (xenophagy). Furthermore, certain NLRs (e.g., NLRP3 and AIM2), in the context of inflammasome activation, can also trigger the activation of the small GTPase RALB, thereby promoting the assembly of the ULK1 and Beclin1-VPS34 complexes and inducing autophagy in response to danger signals like uric acid, nigericin, or cytosolic DNA [64].

Cytokines, released by immune cells or produced within the tissue microenvironment, play vital roles in amplifying and modulating signals within this network, forming complex bidirectional regulatory circuits with autophagy. Interferon-gamma (IFN-γ), a potent pro-autophagic cytokine, can exert its effects independently of the canonical JAK-STAT pathway, instead acting through mechanisms such as MAPK14 (p38α) activation, ultimately enhancing antigen presentation and promoting pathogen clearance [65]. Autophagy induction, in turn, can promote IFN-γ production and limit ROS accumulation, thereby positively modulating signaling strength. Tumor necrosis factor-alpha (TNF-α) can positively regulate autophagy by activating multiple downstream pathways, including ERK, JNK, p38 MAPK, and NF-κB [66]. The autophagic process can then feed back to modulate TNF-α levels, jointly influencing the progression of inflammatory diseases. Interleukin family members, such as IL-17, can induce autophagy via the ERK-Beclin1-p62 pathway, yet may also upregulate Bcl-2 expression through the PI3K-GSK-3β pathway, thereby inhibiting autophagy. Its net effect is thus highly context-dependent. IL-1α and IL-1β can induce autophagy by triggering endoplasmic reticulum stress (ERS) and intracellular calcium release [67].

In the specific context of antiviral defense, the coupling between immune recognition and autophagy is particularly sophisticated. For RNA virus invasion, cytosolic RIG-I-like receptors (RLRs, such as RIG-I and MDA5), upon recognizing viral RNA, interact with mitochondrial antiviral-signaling protein (MAVS). This interaction not only activates IRF3/7 and NF-κB to induce type I interferon responses, but its overexpression can also directly trigger both autophagy and apoptosis. The mitochondria-anchored NLR family member NLRX1 cooperates with the mitochondrial Tu translation elongation factor (TUFM) and the autophagy protein complex ATG5-ATG12 to divert RIG-I signaling towards the autophagy pathway [68]. This mechanism activates autophagy while appropriately suppressing excessive type I interferon production, preventing detrimental immunopathology. For cytosolic DNA, activation of the cGAS-cGAMP-STING pathway is central to initiating type I interferon responses. Intriguingly, activated STING protein can colocalize with autophagy proteins like LC3 and ATG9a, while Beclin1 can directly bind cGAS and recruit the VPS34 complex to degrade cytosolic DNA [69]. This process clears potentially immunogenic material while also providing negative feedback to inhibit cGAS-mediated interferon production, preventing excessive inflammation. In summary, immune and inflammatory signaling pathways control autophagy through a multi-layered, highly interconnected network incorporating fine-tuned feedback regulation. This enables autophagy activity to appropriately respond to infection, injury, or homeostatic imbalance, playing indispensable physiological and pathological roles in host defense, inflammation regulation, and tissue repair [70].

3.1.5. ER stress signaling pathway

ER, as a pivotal organelle responsible for protein synthesis, folding, modification, and calcium ion storage, is essential for maintaining cellular function [71]. The excessive accumulation of unfolded or misfolded proteins within the ER lumen, disturbances in calcium ion homeostasis, or alterations in membrane potential can lead to ERS [67]. To counter this stress, cells have evolved a sophisticated response system known as the unfolded protein response (UPR), which aims to restore ER homeostasis. This response is primarily initiated by three sensor proteins anchored in the ER membrane: inositol-requiring enzyme 1α (IRE1α), protein kinase R-like ER kinase (PERK), and activating transcription factor 6 (ATF6). Under homeostatic conditions, these sensors are bound to the chaperone protein Bip/GRP78 and remain inactive. Upon stress, Bip/GRP78 dissociates to bind aberrant proteins, leading to the activation of these three sensors and the initiation of downstream signaling. Within this context, the induction of autophagy emerges as a key pathway to clear misfolded proteins and alleviate ER stress [72].

IRE1α is the most evolutionarily conserved ERS sensor. Its activation involves dimerization and autophosphorylation, which subsequently activate its intrinsic endoribonuclease activity. This enzyme catalyzes the unconventional splicing of X-box binding protein 1 (XBP1) mRNA, generating the active transcription factor sXBP1. Upon nuclear translocation, sXBP1 comprehensively upregulates the expression of genes encoding ER-associated degradation (ERAD) components and various molecular chaperones, thereby enhancing the protein folding and degradation capacity of the ER. Furthermore, activated IRE1α can recruit TRAF2 and apoptosis signal-regulating kinase 1 (ASK1), activating the JNK signaling pathway. Activated JNK, in turn, phosphorylates Bcl-2, prompting its dissociation from Beclin1. The released Beclin1 then initiates autophagosome formation [73]. The activation of PERK similarly relies on dimerization and autophosphorylation. Activated PERK phosphorylates eukaryotic translation initiation factor 2α (eIF2α) at Ser51. This event globally suppresses protein translation to reduce ER load while selectively promoting the translation of ATF4 and its downstream target, C/EBP homologous protein (CHOP) [73]. As transcription factors, ATF4 and CHOP synergistically upregulate the expression of multiple core autophagy genes, including ATG5, ATG12, LC3, Beclin1, and p62, thereby systemically inducing the autophagy program at the transcriptional level. The activation mechanism of ATF6 differs: following its dissociation from Bip/GRP78 upon stress, ATF6 translocates to the Golgi apparatus, where it is sequentially cleaved by site-1 protease (S1P) and S2P, releasing its cytosolic domain. This fragment enters the nucleus and functions as an active transcription factor [74]. It not only directly upregulates genes for ER chaperone proteins but also indirectly promotes the transcription of XBP1, thereby engaging in crosstalk with the IRE1α pathway to collectively amplify the UPR and promote autophagy.

3.1.6. p53 signaling pathway

The tumor suppressor p53 encodes a transcription factor that gates cell cycle entry. p53 responds to DNA lesions, metabolic imbalance, and oxidative insults, deploying both transcriptional programs and direct protein interactions to impose cycle arrest, trigger apoptosis, or clear senescent cells. Autophagy may adversely regulate p53, which in turn can positively or negatively regulate autophagy, illustrating the complexity of the interaction between autophagy and p53. Additionally, activated p53 can bind to the promoter regions of multiple genes encoding pro-autophagy regulators, including AMPK, death-associated protein kinase 1 (DAPK1), damage-regulated autophagy modulator (DRAM), sestrin, and TSC2 [75].

A large share of p53 target genes converge on mTOR, collectively throttling the core autophagy checkpoint. Specifically, p53 can directly activate AMPK by upregulating the β1 and β2 subunits or upregulating the expression of TSC2 to inhibit mTOR, thereby promoting autophagy. Insulin-like growth factor-1 (IGF-1) can bind to the IGF-1 receptor (IGFR) to activate the mTOR pathway. It is worth noting that when intracellular stress, such as hypoxia and DNA damage, occurs, p53 is activated and upregulates the expression of insulin growth factor binding protein 3 (IGF-BP3) that can bind with IGF-1 with high affinity and competitively blocks the binding of IGF-1 to IGFR, resulting in blocking IGF-1-mediated mTOR signaling activation and inducing autophagy. Furthermore, hypoxia and DNA damage are known to stimulate the stress response gene known as regulated in development and DNA damage responses 1 (REDD1), which is hypothesized to govern cell growth in response to stress. REDD1 sits upstream of TSC1/TSC2 and damps mTOR signaling. Hypoxia lifts p53-driven hypoxia-inducible factor 1 (HIF-1), which boosts REDD1 transcription; the surge of REDD1 then silences mTOR and tips the balance toward autophagy [76,77].

DAPK is a tumor suppressor that mediates multiple cell death pathways. Emerging evidence positions DAPK family kinases as stage-specific regulators that gate multiple steps along the autophagy continuum. Concretely speaking, DAPK1 may phosphorylate protein kinase D (PKD) to further activate VPS34 or directly phosphorylate Beclin1 to induce autophagy. Microtubule-associated protein 1B (MAP1B) docks onto LC3, and DAPK1 locks onto MAP1B; together they seed autophagic membranes and drive autophagosome buildup [78]. Another study demonstrates that the lysosomal protein DRAM can promote autophagy by regulating the fusion of autophagosomes and lysosomes. p53 also lifts transcription of BH3-only members Bad, Bax, Bnip3 and Puma; these peptides pry Beclin1 from Bcl-2/Bcl-XL, unleashing autophagy. p53 also drives sestrin1 and sestrin2, which in turn switch on AMPK and phosphorylate TSC2, thereby igniting autophagy [79].

3.1.7. Acetylation signaling pathway

Reversible protein acetylation is a crucial post-translational mechanism (PTM) regulating the autophagy process [80]. This dynamic modification is coordinately controlled by lysine acetyltransferases (such as p300/CBP) and deacetylases (including SIRT1 and HDACs) [80]. By directly modifying core autophagy proteins or regulating relevant transcription factors, acetylation broadly influences various stages of autophagy, including initiation, membrane formation, and degradation. Under nutrient-replete conditions, the acetyltransferase p300 can catalyze the acetylation of the key autophagy protein LC3. This modification inhibits its binding to phosphatidylethanolamine, thereby blocking the elongation of the autophagosome membrane. Conversely, during energy stress, the deacetylase SIRT1 is activated and can deacetylate core components such as ATG5, ATG7, and LC3, promoting autophagosome formation [80]. Furthermore, p300 can acetylate Raptor, a regulatory subunit of mTORC1, under amino acid deprivation. This acetylation inhibits mTORC1 activity, subsequently relieving its suppression of the ULK1 complex and initiating the autophagy program. Acetylation also modulates key transcription factors, such as FOXO family proteins. When the PI3K-AKT pathway is active, acetylation of FOXO proteins represses their transcriptional activity. During energy stress, however, SIRT1-mediated deacetylation enhances the nuclear localization and transcriptional function of FOXO proteins. This leads to the upregulation of autophagy-related gene expression, establishing a comprehensive regulatory axis that integrates signal perception with gene expression programming [81].

3.1.8. Nuclear receptor signaling pathway

Nuclear receptors (NR) constitute a superfamily of ligand-dependent transcription factors that extensively participate in development, metabolism, and stress responses by regulating gene expression programs [82]. Within the autophagy regulatory network, members of the nuclear receptor 4A (NR4A) subfamily—including Nur77 (NR4A1), Nurr1 (NR4A2), and Nor1 (NR4A3)—function as critical stress sensors, playing a distinctive role in bridging external stimuli and intracellular autophagy activation [83]. Under basal conditions, these receptors are primarily localized within the nucleus, where they modulate the transcription of specific target genes. However, under stimuli such as mitochondrial damage, oxidative stress, or specific agonists (e.g., cytochalasin B), they can undergo nucleocytoplasmic translocation, relocating to cytoplasmic sites such as mitochondria, and subsequently regulate the autophagy process directly through non-genomic mechanisms. Their core mechanism involves interaction with key signaling adaptor proteins. For instance, at the mitochondria, Nur77 can bind to TRAF2, triggering its own ubiquitination and recruiting the autophagy adapter protein p62/SQSTM1, thereby specifically initiating the autophagic clearance of damaged mitochondria, known as mitophagy. Similarly, Nurr1 and Nor1 have also been reported to participate in regulating autophagy-related signaling pathways under various stress conditions. Thus, as a class of stress-responsive proteins capable of shuttling between the nucleus and cytoplasm, NR4A family members constitute an important regulatory layer that rapidly responds to damage signals and directly intervenes in the execution of autophagy, particularly selective autophagy [84]. This highlights an emerging function of nuclear receptors beyond transcriptional regulation—direct involvement in cellular quality control.

3.1.9. Calcium signaling pathway (dual regulation)

Calcium ion (Ca²⁺), as a ubiquitous second messenger, plays a crucial role in autophagy regulation through precise control of its cytosolic concentration. Notably, Ca²⁺ signaling exerts a dual and context-dependent effect on autophagy, capable of acting as either an inducer or an inhibitor. The ultimate outcome is profoundly influenced by the source, amplitude, and spatiotemporal dynamics of the Ca²⁺ signal, as well as the overall metabolic state of the cell [85].

Under various stress conditions, an increase in cytosolic Ca²⁺ serves as a key signal to induce autophagy. This Ca²⁺ can be released from the endoplasmic reticulum (ER) via inositol 1,4,5-trisphosphate receptors (IP3Rs) or from lysosomes through transient receptor potential mucolipin 1 (TRPML1) channel [86]. Elevated Ca²⁺ promotes autophagy via multiple parallel pathways. First, Ca²⁺ binding to calmodulin (CaM) activates Ca²⁺/calmodulin-dependent protein kinase kinase β (CaMKKβ), which in turn phosphorylates and activates AMPK. Activated AMPK initiates autophagy by inhibiting mTORC1 and directly phosphorylating ULK1 [87]. Second, Ca²⁺ signaling can activate death-associated protein kinase (DAPK). Activated DAPK both phosphorylates and activates protein kinase D (PKD), leading to VPS34 complex activation, and directly phosphorylates Beclin1, promoting its dissociation from the Bcl-2 complex, thereby driving autophagosome formation. Third, localized Ca²⁺ bursts from lysosomes can activate calcineurin, leading to the dephosphorylation and subsequent nuclear translocation of transcription factor EB (TFEB). This results in the large-scale upregulation of autophagy and lysosomal genes, systemically enhancing autophagic capacity. Additionally, the fusion of autophagosomes with lysosomes itself has been shown to require a moderate increase in cytosolic Ca²⁺ concentration [88].

Conversely, under basal or non-stressed conditions, constitutive (or tonic) Ca²⁺ signaling primarily functions to inhibit autophagy, maintaining cellular energy homeostasis and preventing unnecessary catabolism. This inhibitory effect is achieved through several mechanisms. The sustained, low-level Ca²⁺ release mediated by IP3Rs is taken up by adjacent mitochondria [89]. The increased intramitochondrial Ca²⁺ promotes oxidative phosphorylation and ATP generation by activating Ca²⁺-dependent dehydrogenases. The resulting high ATP/AMP ratio maintains AMPK in an inactive state, thereby relieving its inhibition of mTORC1 and keeping the ULK1 complex inactive, ultimately suppressing basal autophagy. Furthermore, beyond its role as a Ca²⁺ channel, IP3R acts as a molecular scaffold that directly binds both Beclin1 and Bcl-2, stabilizing their inhibitory interaction and spatially preventing autophagy initiation. Under specific pathological conditions like oxidative stress, Ca²⁺ influx through TRPM2 channels can activate CaMKK2, which phosphorylates Beclin1. This phosphorylation, conversely, enhances the binding between Beclin1 and Bcl-2, further reinforcing autophagy inhibition and potentially steering the cell towards death [90].

3.1.10. Epigenetics

Epigenetic mechanisms, which alter gene expression states without changing the DNA sequence, provide a crucial foundation for the long-term and adaptive regulation of autophagy [91]. This multi-layered regulatory network primarily involves DNA methylation, histone modifications, and the actions of non-coding RNAs, collectively shaping the autophagic capacity of cells under various physiological and pathological conditions. DNA methylation typically represses gene expression. Studies have shown that in various tumor cells, the promoter regions of several core autophagy genes, including ULK2, Beclin1, and LC3A, exhibit a hypermethylated state. Their demethylation can restore autophagic activity, thereby influencing tumor initiation and progression [92]. Histone modifications are more dynamic and complex, with acetylation being particularly important. For example, the histone acetyltransferases P300/CBP can not only directly regulate the autophagy process by acetylating autophagy-related proteins (such as ATG5 and ATG7) but also influence chromatin openness by acetylating lysine 56 on histone H3 (H3K56ac), thereby modulating the transcription of autophagy-related genes. Furthermore, acetylation of lysine 16 on histone H4 (H4K16ac), catalyzed by the acetyltransferase hMOF/KAT8, promotes the transcription of autophagy genes. Notably, the progression of autophagy can, in turn, degrade hMOF, forming a negative feedback regulatory loop that prevents excessive autophagy activation. Beyond acetylation, histone methylation modifications such as H3K4me2/3, H3K9me2, and H4K20me3 have also been confirmed to participate in the transcriptional regulation of autophagy-related genes [93]. Non-coding RNAs, particularly microRNAs and long non-coding RNAs, exert precise post-transcriptional control over autophagy by targeting ATG gene mRNAs or upstream regulatory factors. In summary, DNA methylation, histone modifications, and non-coding RNAs constitute an interconnected epigenetic network. This network enables cells to integrate sustained internal and external environmental signals, facilitating stable yet plastic long-term programming of autophagic activity. This process holds significant importance in development, aging, and the progression of chronic diseases [94].

3.2. Signaling pathways that inhibit autophagy

3.2.1. PI3K-AKT-mTOR signaling pathway

The PI3K-AKT-mTOR axis operates as a central growth and survival network under both physiological and pathological contexts. Phosphatidyl-inositol-3-kinases (PI3Ks) are lipid kinases that transfer phosphate to the 3’ hydroxyl of inositol lipids, generating second messengers that steer downstream signaling. AKT, the principal serine/threonine kinase that lies immediately downstream of PI3K, exists as three isoforms: AKT1, AKT2, and AKT3. mTOR is considered a central factor in the regulation of autophagy [52].

Receptor tyrosine kinases (RTKs) are high-affinity surface receptors for polypeptide growth factors, cytokines, and hormones. In their unliganded state, they remain monomeric and quiescent; ligand binding drives dimerization, triggering trans-phosphorylation of tyrosine residues within their cytoplasmic tail. Intracellularly phosphorylated residues of RTK receptors provide a docking site for the regulatory subunit p85 of PI3K, but in some cases, it will recruit an adaptor protein. For example, when insulin activates its receptor, an insulin receptor substrate (IRS) protein must be recruited to facilitate binding to PI3K. The regulatory subunit p85 of PI3K is recruited to sites adjacent to the plasma membrane upon receipt of a signal from RTK. Upon docking with the regulatory p85 subunit, the catalytic p110 subunit of PI3K phosphorylates phosphatidylinositol 4,5-bisphosphate (PIP2) to generate PIP3, a reaction counteracted by the lipid phosphatase phosphatase and tensin homolog deleted on chromosome 10 (PTEN). The generated PIP3 then recruits AKT, which contains the pleckstrin homology (PH) domain and 3-phosphoinositide-dependent protein kinase 1 (PDK1) on the cell membrane. The AKT protein is subsequently activated by phosphorylation of Thr308 and Ser473 with the assistance of PDK1 and mTORC2 [52,95,96]. There are two direct and indirect pathways by which activated AKT activates its substrate mTORC1: one is to directly phosphorylate the PRAS40 subunit of mTORC1, and the other is to inactivate TSC2, thereby preventing the negative regulation of RHEB by the TSC1/2 complex, resulting in the maintenance of the binding state of RHEB and GTP. RHEB, a small Ras-related GTPase, flips mTORC1 on by docking to its catalytic pocket when loaded with GTP. However, RHEB activation terminates upon the hydrolysis of GTP to GDP [51].

Active mTORC1 clamps down on autophagy in several ways. First, it phosphorylates ULK1, severs its contact with AMPK, and thereby blocks assembly of the ULK1 complex. Moreover, mTORC1 may phosphorylate ATG14, AMBRA1, and NRBF2 to directly inhibit the activity of the PI3KC3 complex [10]. Additionally, the acetyltransferase p300 has been shown to participate in many processes in the control of autophagosome formation. Specifically, phosphorylation of the p300 protein by mTORC1 promotes acetylation of LC3, which in turn prevents the lipidation of LC3 and hence blocks autophagy [97]. WIPI proteins (WIPI1-4) are a mammalian precursor protein family of phosphoinositide effectors that are essential for autophagosome biogenesis. WIPI2 induces autophagy by binding to ATG16 and promoting the ATG12-ATG5-ATG16 complex’s binding to autophagosomes. However, mTORC1 phosphorylates WIPI2 at Ser395, resulting in WIPI2 degradation by the ubiquitin-proteasome and suppressing autophagy [98]. In addition, mTORC1 also dampens autophagy by repressing the transcriptional program. In nutrient-rich settings, mTORC1, along with ERK2 and AKT, heavily phosphorylates TFEB, locking the factor onto cytoplasmic YWHA and thereby silencing its nuclear function. Once mTORC1 is shut down, TFEB loses its phosphates and races into the nucleus, driving transcription of autophagy and lysosomal genes. Elevated TFEB further boosts UVRAG, WIPI, MAPLC3B, SQSTM1, VPS11, VPS18, and ATG9B, proteins that act across the autophagy pathway. Finally, mTORC1 can inhibit the activity of ATG14-containing PIK3C3/VPS34 complex by phosphorylating ATG14, resulting in autophagy inhibition, but it does not affect the ATG14-free PIK3C3 complex [99].

MTORC2 influences autophagy chiefly through AKT. Rictor recruits AKT, allowing mTORC2 to phosphorylate AKT at Ser473. Phospho-AKT then tags FOXO3 on Thr32, triggering its nuclear exit and shutting down autophagy gene transcription. More importantly, this effect cannot be prevented by rapamycin (Fig. 4).

Fig. 4.

PI3K-AKT-mTOR signaling pathway regulates autophagy.

3.2.2. MAPK signaling pathway

MAPK pathway converts extracellular cues into nuclear programs that drive proliferation, lineage commitment, and senescence. Previously, three pathways that regulate autophagy through JNK, p38 MAPK, and ERK have been described. However, the function of the JNK signaling pathway is mainly to induce autophagy to protect the organism from acute oxidative and exogenous damage. Therefore, only p38 MAPK and ERK typically have negative effects on the regulation of autophagy. A previous study showed that inhibition of p38 MAPK signaling increased necroptosis and autophagic cell death in TNF-α-treated L929 cells, suggesting that p38 MAPK has an inhibitory effect on autophagy [100]. Furthermore, autophagy inhibition was observed to alleviate immunosuppression and promote inflammatory responses in microglia. The potential mechanism is that p38 MAPK directly phosphorylates ULK1 in microglia, resulting in inhibition of ULK1 activity. TFEB, which is regulated by ERK, is generally believed to modulate the expression of ATG genes related to lysosomal function. For example, in BRAF^V600E melanoma cells, autophagy is suppressed when ERK phosphorylates TFEB, which leads to the retention of TFEB in the cytoplasm and damage to autophagolysosomal target gene expression [101].

3.2.3. cAMP signaling pathway

The cAMP/PKA axis tunes autophagy in a context-dependent manner. Elevated cAMP blocks flux in HT-29 colon adenocarcinoma cells yet leaves HeLa cervical cancer cells unaffected. This divergence likely reflects restricted PKA pools that act autonomously of bulk cAMP levels [102], as PKA has been shown to regulate autophagy via the phosphorylation of many ATG proteins (including ATG1, ATG13, LC3, and ATG16L1). In neurons, autophagy curtails protein aggregation yet can simultaneously drive neurite degeneration. And cAMP can phosphorylate LC3 through PKA, thereby inhibiting autophagy to maintain neuronal function [103].

3.2.4. p53 signaling pathway

In addition to its role as a transcription factor in the nucleus, p53 is also present in the cytoplasm and mediates a variety of intracellular processes. In general, the regulation of autophagy by p53 may be related to its localization. Autophagy is induced when the nuclear export signal sequence deletion results in the accumulation of p53 in the nucleus. Conversely, when the nuclear localization signal sequence is absent, p53 accumulates in the cytoplasm and inhibits autophagy [76]. RB1-inducible coiled-coil protein 1 (RB1CC1), also known as FIP200, is a mammalian counterpart of yeast ATG17. By increasing the stability and phosphorylation of ULK1, RB1CC1 may promote autophagy. However, it can also bind to p53 and prevent the formation of the ULK1 complex, thereby exerting a role in inhibiting autophagy [104]. Additionally, RNA-activated protein kinase (PKR) induces autophagy by phosphorylating eIF2α in response to the accumulation of intracellular double-stranded RNA (dsRNA). And p53 dampens autophagy by destabilizing dsRNA, thereby curbing PKR activation. Moreover, miR34a, a class of highly conserved miRNAs in evolution, is one of the target genes of p53, which can be regulated by p53 to inhibit autophagy in two ways. The first is to inhibit the expression of ATG9 and STRT1, and the second is to upregulate Bcl-2, resulting in the interaction between Bcl-2 and Beclin1 [105,106]. In cardiomyocytes, p53 governs autophagy and ischemia/reperfusion injury through myocardin. The autophagy repressor cardiac autophagy inhibitory factor (CAIF) binds p53, blunting its transactivation of myocardin; the drop in myocardin suppresses autophagy and alleviates myocardial infarction.

3.2.5. Hedgehog signaling pathway

Inhibitory effects of Hh signaling on autophagy have been identified in some investigations. In particular, PERK induces autophagy by upregulating various autophagy genes through the eIF2α-ATF4 pathway, while GLI2, a downstream signal of Hh, exerts its autophagy-inhibiting effect by inhibiting PERK and reducing the phosphorylation of eIF2α [107]. Another study showed that metformin could downregulate hyperglycemia-induced BNIP3 expression via GLI1, which inhibited the binding between BNIP3 and BCL-2 while enhancing the binding of Beclin1 and BCL-2. What counts is that this autophagy-inhibiting effect of metformin attenuates hyperglycemia-induced endothelial damage. Finally, arachidonate lipoxygenase (ALOX) can inhibit the autophagy induced by IFN-γ through JAK-STAT. Pathogens such as human immunodeficiency virus, influenza virus, and mycobacteria may produce ALOX5/ALOX15 through the SHH signaling pathways to inhibit autophagy, which leads to immune escape [108].

4. Autophagy and disease

Autophagy is an evolutionarily conserved catabolic route in which cytoplasmic constituents are engulfed and delivered to lysosomes for degradation and recycling. By orchestrating organelle quality control and metabolic adaptation, this pathway governs cell growth, maturation, lineage commitment, and death, and is therefore inseparable from physiological homeostasis and disease pathogenesis. Normal autophagy safeguards intracellular homeostasis and metabolic equilibrium; its failure accompanies neoplasia, diabetes, hepatic injury, leukemia, and also intersects with aging, infection, and immunity [109]. In conclusion, autophagy plays a complicated function in illness progression, but our understanding of the link between autophagy and disease will improve with further research. Autophagy has emerged as a breakthrough in disease treatment that has sparked renewed interest in previously incurable conditions [110] (Fig. 5).

Fig. 5.

Autophagy-related diseases.

4.1. Cancer

Autophagy plays a complex and dynamic role in tumorigenesis and cancer progression, with its function being highly dependent on the stage of malignancy, tumor type, and the specific microenvironmental context [111]. During the initial phases of tumor development, autophagy primarily exerts tumor-suppressive effects. It helps maintain genomic stability and intracellular homeostasis through the selective clearance of damaged proteins, dysfunctional organelles—particularly mitochondria—and by preventing the abnormal aggregation of oncogenic proteins such as p62. Multiple studies have demonstrated reduced autophagic flux and downregulated expression of autophagy-related genes in tumor tissues from liver, breast, ovarian, and prostate cancers. Furthermore, frameshift mutations in genes such as ATG2B, ATG5, and ATG9B have been identified in gastrointestinal tumors, providing genetic evidence that defective autophagy can promote tumor initiation and early development [112].

Conversely, in established and progressing tumors, autophagy often shifts to function as a pro-survival mechanism. Many cancer cells, including those expressing oncogenic Ras, exhibit elevated basal autophagy levels [113]. Within hypoxic and nutrient-deprived regions of the tumor, autophagy supports cancer cell survival under metabolic stress and limited vascularization by recycling intracellular components to supply metabolic intermediates and ATP. Notably, a complex crosstalk exists between autophagy and programmed cell death pathways. On one hand, autophagy can induce autophagic cell death in apoptosis-resistant cancer cells, serving as an alternative mechanism for cell elimination—exemplified by Bax/PUMA-deficient colon carcinoma cells. On the other hand, autophagy can inhibit anoikis, thereby promoting the survival of circulating tumor cells that have detached from the primary site and facilitating hematogenous dissemination and metastasis [114].

In the context of therapeutic response, autophagy similarly exhibits a “double-edged sword” characteristic. Various treatment modalities, including radiotherapy and chemotherapy, can activate autophagy and lead to cancer cell death [115]. However, autophagy also functions as a critical cellular defense mechanism by clearing therapy-induced damage and maintaining metabolic equilibrium, which can promote tumor cell tolerance and survival, ultimately contributing to therapy resistance. For instance, enhanced autophagic flux is associated with increased expression of ATP-binding cassette (ABC) transporter proteins, accelerating the efflux of chemotherapeutic agents. Autophagy-related microRNAs, such as the downregulation of miR-199a-5p or altered expression of miR-181a, can finely regulate the expression of key genes like ATG5 to influence cancer cell sensitivity to drugs such as cisplatin [116]. Additionally, autophagy aids in the clearance of therapy-damaged mitochondria, alleviates oxidative stress, and may help maintain cancer stem cell properties, collectively fostering a multidrug-resistant (MDR) phenotype. More intricately, autophagy may also induce therapy-treated cancer cells to enter a senescent state. While this represents a temporary growth arrest rather than cell death, it may establish a dormant reservoir of cells with the potential to drive tumor recurrence.

4.2. Neurodegenerative disease

The autophagy-lysosomal pathway serves as a central mechanism for neuronal clearance of abnormal protein aggregates and damaged organelles. Its dysfunction is closely linked to the pathogenesis and progression of various neurodegenerative diseases. These disorders—including Parkinson’s disease (PD), Alzheimer’s disease (AD), Huntington’s disease (HD), and ALS—share common pathological hallmarks: the accumulation of misfolded proteins and neuronal death. A decline in autophagic flux is widely regarded as a key factor driving this collapse of proteostasis [117].

In PD, disrupted autophagy directly contributes to core pathological processes. The primary pathological hallmark, Lewy bodies, is predominantly composed of abnormally aggregated α-synuclein. Under physiological conditions, α-synuclein is degraded via chaperone-mediated autophagy. However, under pathological conditions, mutations or overexpression of this protein can overwhelm the degradation capacity of both the ubiquitin-proteasome system and macroautophagy, leading to the formation of toxic aggregates [118]. Concurrently, impaired mitophagy mediated by the PD-associated PINK1/Parkin pathway results in the accumulation of dysfunctional mitochondria and consequent reactive oxygen species, further exacerbating oxidative damage and death of dopaminergic neurons.

In AD, autophagy is involved in regulating two key pathogenic proteins: Amyloid-β (Aβ) and Tau. Autophagic vacuoles accumulate abnormally in the brains of AD patients, and their acidic interior may conversely promote the formation of Aβ oligomers and fibrils [119]. On the other hand, hyperphosphorylated Tau protein not only loses its normal function but also impairs microtubule-dependent transport of autophagic vesicles, creating a vicious cycle of “autophagic flux blockade”. Furthermore, mitochondrial damage and calcium overload induced by Aβ deposition can activate downstream signaling pathways, such as CaMKK2 and AMPK, which exacerbate Tau pathology and induce massive generation of autophagic vacuoles, collectively driving neuronal dysfunction.

HD is caused by the aggregation of mutant huntingtin protein fragments containing expanded polyglutamine repeats. When the ubiquitin-proteasome system is compromised due to substrate misfolding, autophagy becomes the primary route for clearing these toxic aggregates. Studies have found that although the number of autophagosomes increases in HD model brains, their efficiency in recognizing and degrading cargo—specifically mutant Huntingtin (mHTT) aggregates—is insufficient. This leads to incomplete degradation and a functional defect in autophagic flux. Selective autophagy adaptor proteins, such as Alfy/Wdfy3, and mitophagy also play important compensatory roles in the clearance process [120].

In ALS, the role of autophagy exhibits stage-specific characteristics [120]. During the early stages of the disease, activation of autophagy helps clear toxic aggregates, such as mutant superoxide dismutase 1 (mSOD1) protein, exerting a protective effect. As the disease progresses, autophagic function may become dysregulated, shifting to participate in synaptic modification and activating pro-apoptotic pathways, thereby accelerating motor neuron degeneration. Additionally, ALS-associated gene mutations (e.g., in OPTN and TBK1) can directly impair mitophagy, leading to the accumulation of damaged mitochondria and triggering neuroinflammation, which further propels disease progression [121].

4.3. Cardiovascular disease

Autophagy exerts dynamic and highly context-dependent regulatory functions within the cardiovascular system. The specific state of autophagy, whether it is deficient, moderate, or excessive, profoundly influences the progression and outcomes of major cardiovascular diseases, including myocardial ischemia-reperfusion injury (MI/RI), atherosclerosis (AS), and heart failure (HF).

In MI/RI, the role of autophagy exhibits distinct temporal specificity. During the ischemic phase, cells activate autophagy primarily through the AMPK-mTOR pathway. This response degrades intracellular components to sustain ATP supply, thereby playing a crucial protective role for cell survival [122]. However, upon reperfusion, the persistence or overactivation of autophagy can become detrimental. This shift is closely associated with upregulated Beclin1 expression, JNK-mediated phosphorylation of Bcl-2, and subsequently enhanced oxidative stress and mitochondrial damage. These events can ultimately work in concert to promote cardiomyocyte apoptosis and death. Concurrently, mitophagy operates throughout this process to clear dysfunctional mitochondria, which serves as a vital mechanism for maintaining cellular homeostasis [123].

Regarding AS, the autophagic activity of macrophages is a pivotal factor determining plaque phenotype and stability. In the early stages of the disease, basal levels of autophagy can delay plaque progression and enhance stability by promoting intracellular lipid degradation, facilitating cholesterol efflux, and suppressing inflammatory responses. Autophagy also contributes to the polarization of macrophages from a pro-inflammatory M1 phenotype toward an anti-inflammatory M2 phenotype, which helps mitigate local inflammation [124]. Conversely, in advanced lesions, insufficient autophagy leads to impaired clearance of apoptotic cells, exacerbated inflammation, and degradation of the extracellular matrix. These effects promote the expansion of the necrotic core and thinning of the fibrous cap, thereby increasing plaque vulnerability and the risk of rupture [125].

In the development and progression of HF, autophagy similarly plays a dualistic role. Moderate autophagy exerts a protective effect on cardiac function by promptly degrading misfolded proteins and damaged organelles. This process counteracts pressure overload-induced myocardial hypertrophy and pathological remodeling [126]. Research indicates that specific knockout of ATG5 in cardiomyocytes results in an almost complete absence of autophagy, accompanied by aberrant protein accumulation and the development of cardiac hypertrophy. In contrast, excessive activation of autophagy, as induced by Beclin1 overexpression, accelerates the exaggerated degradation of proteins. This can lead to cardiomyocyte atrophy and death, worsening cardiac dysfunction and ventricular remodeling, and ultimately driving the progression of HF [127].

4.4. Metabolic disease

The capacity of autophagy to eliminate damaged organelles and degrade biomacromolecules such as glycogen, lipids, and proteins for reuse by cells is crucial for maintaining intracellular metabolic balance. Abnormal autophagy is often discovered to be associated with metabolic diseases. This section primarily introduces the role of autophagy in lipid and glucose metabolism.

4.4.1. Autophagy and lipid metabolism

Autophagy, particularly the selective degradation of lipid droplets known as lipophagy, serves as a core mechanism for cellular regulation of lipid homeostasis. Under conditions of nutrient deprivation, lipophagy is activated to break down lipid droplets, providing cells with free fatty acids and energy. Conversely, in states of energy excess, dysfunction of autophagy directly contributes to diseases such as obesity, non-alcoholic fatty liver disease, and atherosclerosis [128]. In obesity, altered autophagic activity, typically manifested as functional impairment, in the hypothalamus and adipose tissue disrupts energy sensing and balance, exacerbates inflammation, and promotes abnormal fat accumulation [129]. During the course of non-alcoholic fatty liver disease (NAFLD), autophagy plays a dual role. On one hand, it counteracts hepatic steatosis and injury, often described as opposing the “first hit” and “second hit”, by clearing excess lipids from hepatocytes, reducing oxidative stress, and inhibiting inflammatory responses. On the other hand, excessive or inappropriate autophagy may promote the activation of hepatic stellate cells, thereby aggravating the process of liver fibrosis. In AS, the autophagic function of macrophages is crucial. It inhibits foam cell formation by degrading ingested oxidized low-density lipoprotein and promoting cholesterol efflux, thereby retarding plaque development. Conversely, defective autophagy leads to impaired lipid clearance and accelerates atherosclerotic lesion progression [130].

4.4.2. Autophagy and glucose metabolism

An intricate and complex interplay exists between autophagy and glucose metabolism, which is essential for maintaining blood glucose stability. During the neonatal period, when the gluconeogenic pathway is not fully established, autophagy-mediated glycogen degradation, termed glycophagy, becomes a critical alternative pathway for sustaining blood glucose. Newborn animals with a knockout of ATG5 or ATG7 genes die due to severe hypoglycemia [131]. In Pompe disease, also known as type II glycogen storage disease, a deficiency in the lysosomal enzyme acid alpha-glucosidase leads to glycogen accumulation within lysosomes, accompanied by abnormal autophagosome accumulation. Modulation of the autophagic pathway is thus regarded as a potential therapeutic strategy. The role of autophagy in diabetes is more multifaceted and cell-type specific. In pancreatic beta cells, autophagy is a central protective mechanism crucial for maintaining their normal function, differentiation, survival, and response to various stresses such as endoplasmic reticulum stress and oxidative stress. Deficiency in autophagy increases susceptibility to diabetes. In insulin target tissues like the liver and muscle, insulin itself suppresses autophagy via the PI3K-AKT-mTORC1 pathway. Therefore, insulin resistance or deficiency leads to elevated basal autophagy levels [132]. Chronic hyperglycemia and metabolic disturbances can further disrupt the normal regulation of autophagy, creating a vicious cycle that exacerbates insulin resistance and cellular dysfunction. Consequently, autophagy represents a significant regulatory node within the pathophysiological network of diabetes and its complications.

4.5. Inflammation

Inflammation is an immediate defensive response to infectious pathogens, harmful chemicals, or physical tissue damage [38]. The interaction between autophagy and inflammation serves as a pivotal hub linking cellular homeostasis regulation and disease progression, with its complexity and dynamism profoundly manifested across various pathological conditions. Autophagy not only responds to inflammatory signals for activation but also finely modulates the intensity, duration, and biological outcomes of inflammatory responses through multiple feedback mechanisms [133]. At the fundamental level, autophagy suppresses excessive activation of inflammasomes like NLRP3 by clearing damaged mitochondria, ubiquitinated protein aggregates, and reactive oxygen species, thereby limiting the release of damage-associated molecular patterns and maintaining immune homeostasis. For instance, in chronic low-grade inflammation associated with metabolic syndrome, intact autophagy function is crucial for preventing the pro-inflammatory polarization of adipose tissue macrophages and alleviating insulin resistance. Conversely, autophagy defects or dysregulation can drive inflammation toward chronicity and fibrosis. This occurs either through persistent activation of the NF-κB pathway due to abnormal accumulation of p62/SQSTM1, or via activation of the mtDNA-mediated cGAS-STING pathway resulting from failed mitochondrial quality control [134].

In infectious and immune-mediated diseases, autophagy exhibits a pronounced “double-edged sword” effect. On one hand, autophagy directly eliminates intracellular pathogens through “heterophagy” and enhances host defense by promoting antigen presentation. On the other hand, numerous pathogens have evolved mechanisms to hijack or suppress autophagy to facilitate their survival and replication [135]. For instance, in inflammatory liver diseases, autophagy inhibition prevents timely clearance of damaged mitochondria, leading to excessive reactive oxygen species accumulation that activates NLRP3 inflammasomes and triggers excessive inflammatory responses. Conversely, simply enhancing autophagy can mitigate acute hepatocyte injury, primarily by suppressing NLRP3 inflammasome signaling [136]. Furthermore, autophagy is crucial for the development and progression of viral hepatitis. Hepatitis C virus (HCV) may exploit autophagy mechanisms to enhance its replication capacity. Autophagy plays a dual role in HCV transmission: core autophagy factors (including Atg5, Atg7, Atg12, and Beclin1) are essential for initiating the viral replication cycle; conversely, viral RNA polymerase components NS5A and NS5B, along with HCV RNA aggregated on phagosome membranes, provide critical substrates for genomic amplification and viral particle assembly. More critically, HCV exploits autophagy to promote viral particle maturation and release, while negatively regulating the host innate immune response, leading to chronic infection in the vast majority of individuals. In autoimmune diseases such as systemic lupus erythematosus (SLE), mutations in autophagy-related genes and impaired autophagy function result in impaired clearance of apoptotic cells and increased exposure of nuclear antigens. This disrupts immune tolerance, triggering autoantibody production and inflammatory damage [137].

Recent studies further reveal that autophagy profoundly engages in inflammation resolution and tissue repair. Through LC3-associated phagocytosis, autophagy enhances macrophage clearance of apoptotic cells. Simultaneously, autophagy actively guides the inflammatory microenvironment from a pro-inflammatory state toward repair by regulating arachidonic acid metabolism and generating specific anti-inflammatory mediators [138]. In models such as neuroinflammation and myocardial ischemia-reperfusion injury, moderate autophagy induction has been shown to mitigate inflammatory damage and improve outcomes. However, under conditions of aging, chronic stress, or persistent metabolic dysfunction, autophagic function often declines or becomes disrupted. This impedes inflammatory resolution and may induce “inflammatory memory” in immune cells through epigenetic remodeling, thereby increasing recurrence risk and the probability of chronic disease progression [139].

Thus, autophagy’s role in inflammation is highly context-dependent, with its ultimate effects determined by cell type, disease stage, metabolic state, and the integrity of the autophagy flux itself [140]. Intervention strategies targeting the autophagy-inflammation axis must be grounded in a deep understanding of specific pathological contexts, aiming to restore its physiological regulatory function. For instance, enhancing autophagy in chronic inflammation to promote resolution or modulating autophagy in autoimmunity to rebuild tolerance are preferred approaches over simple activation or inhibition.

4.6. Immune system disease

Recent work has consolidated the view that autophagy is integral to immunity. In innate responses, xenophagy removes intracellular microbes while priming antimicrobial signaling; in adaptive responses, controlled autophagy sustains T-cell differentiation and supports the viability and maturation of B cells. In addition, autophagy participates in immune responses by regulating inflammation, participating in antigen presentation, and promoting the secretion of immune mediators [36].

4.6.1. Microbial infection

As a primitive form of innate immunity, autophagy plays a crucial role in resisting pathogen invasion [141]. Some pathogens, such as Listeria, Shigella flexneri, Salmonella, Toxoplasma, and Sindbis virus, can be ubiquitinated after invading cells and degraded by the autophagy pathway (so-called “xenoautophagy”). However, autophagy also plays a dual role in pathogen defense. An example is the negative factor (Nef), a human immunodeficiency virus type-1 (HIV-1) protein with immunosuppressive functions, which inhibits autophagy by binding to Beclin1, resulting in immune escape [142]. Some parasites can also avoid host cell autophagic clearance and instead use autophagy of the host cell to facilitate reproduction [143]. Autophagy also shapes antigen presentation. It captures viral or self-antigens within autophagosomes, enabling their loading onto major histocompatibility complex class II (MHC II) for recognition by CD⁴⁺ T cells. During HSV-1 infection, the same pathway channels viral peptides onto MHC I, thereby alerting CD⁸⁺ T cells [144]. In recent years, infectious diseases have become an increasingly serious public health problem, and numerous studies have shown that autophagy also plays an important role in infectious diseases such as superbugs, SARS, Ebola virus, avian influenza virus, Middle East respiratory syndrome (MERS), and malaria. In particular, SARS-CoV-2 has been shown to regulate cellular metabolism and reduce autophagy, while autophagy-inducing agents can limit the proliferation of SARS-CoV-2 in primary human lung cells. Thus, modulating intracellular autophagy holds promise as a practical strategy against intracellular pathogens.

4.6.2. Autoimmune diseases

Autophagy plays a pivotal role in maintaining immune tolerance and regulating immune homeostasis. Its dysfunction is closely associated with the pathogenesis of multiple autoimmune diseases [145]. In SLE, defective autophagy leads to impaired clearance of apoptotic cells. The resulting accumulation of uncleared apoptotic debris serves as a source of autoantigens, triggering the production of autoantibodies and perpetuating inflammatory responses. Concurrently, impaired mitophagy within T cells causes an accumulation of reactive oxygen species, further exacerbating abnormal lymphocyte death and immune dysregulation [146]. In rheumatoid arthritis (RA), autophagy participates in the presentation of self-antigens, such as citrullinated proteins, and the establishment of central tolerance within the thymus. Insufficient autophagic function may contribute to the escape of autoreactive T cells. Locally within the joints, rheumatoid arthritis synovial fibroblasts utilize autophagy to adapt to hypoxic and inflammatory microenvironments. This adaptation sustains their abnormal proliferation and invasive properties while also promoting the release of pro-inflammatory cytokines, collectively driving joint destruction [147]. In systemic sclerosis (SSC), the dysregulated control of collagen metabolism by autophagy constitutes a core pathological link. Under normal conditions, TGF-β signaling can activate autophagy to promote the degradation of excessively deposited collagen. However, in diseased fibroblasts, despite enhanced TGF-β signaling, autophagic function remains relatively insufficient. This imbalance between collagen synthesis and degradation accelerates the process of tissue fibrosis. In inflammatory bowel disease (IBD), polymorphisms in autophagy-related genes are linked to disease susceptibility. Functional defects in autophagy can compromise Paneth cell secretion of antimicrobial peptides, impair macrophage clearance of intracellular bacteria, and hinder the endoplasmic reticulum stress response in intestinal epithelial cells. These impairments collectively disrupt intestinal barrier integrity, exacerbate mucosal inflammation, and contribute to microbial dysbiosis [148].

4.7. Bone disease

Autophagy, functioning as a core intracellular mechanism for quality control and homeostasis maintenance, plays a crucial role in both the physiological and pathological processes of the skeletal system. Consistent with its context-dependent nature observed in other diseases, autophagy exhibits a “double-edged sword” characteristic within bone tissue. Whether it is moderately activated, deficient, or excessively activated profoundly influences the onset and progression of major bone disorders, including osteoporosis (OP), osteoarthritis (OA), and fracture healing [149].

Autophagy deeply participates in the bone remodeling process by precisely regulating the fate of various bone lineage cells, such as bone marrow mesenchymal stem cells, osteoblasts, and osteoclasts [150]. In osteoblasts, normal autophagic activity helps them resist stress and maintain function, thereby supporting bone formation. In osteoclasts, autophagy is also involved in their differentiation process. The overall function of autophagy is to maintain a dynamic balance between these two cell types. Research has found that alterations in the expression of autophagy-related molecules, such as Beclin-1 and P62, are closely associated with OP. Multiple signaling pathways, including the classic PI3K/AKT/mTOR pathway, participate in disease development precisely by regulating autophagy to influence the balance between bone formation and resorption [151]. Therefore, a decline in autophagic function is considered one of the key mechanisms leading to relatively insufficient bone formation and the onset of OP.

In OA, autophagy primarily plays a protective role for articular cartilage. It promptly clears organelles, especially mitochondria, and misfolded proteins damaged within chondrocytes due to factors like inflammation and oxidative stress, thereby maintaining cellular homeostasis [152]. For instance, recent research on the SIRT3-PINK1-PKM2 regulatory axis demonstrates that activating mitophagy and driving metabolic reprogramming in chondrocytes can effectively prevent osteoarthritis progression. However, the protective effect of autophagy has a threshold. When joint degeneration is severe and stress signals persist, autophagy may become overactivated. This can lead to “crosstalk” with apoptosis pathways or even induce chondrocyte death, accelerating joint degeneration. Furthermore, a complex relationship exists between autophagy and chondrocyte senescence. Consequently, the role of autophagy in OA is highly dependent on the disease stage and the level of autophagic activity itself [153].

Following a fracture, a local stress microenvironment characterized by hypoxia, acidity, and nutrient deficiency forms [154]. Research confirms that under such conditions, osteoblasts initiate a protective autophagic response. For example, under acidic culture conditions, the expression of the autophagy marker protein LC3 in osteoblasts increases, while inhibiting autophagy significantly elevates apoptosis. This indicates that autophagy is a key adaptive mechanism for osteoblast survival in unfavorable environments. Animal experiments further demonstrate that in the early stages of fracture healing, the expression levels of activated AMPK (p-AMPK) and LC3B-II in callus tissues with successful healing are significantly higher than in non-union groups. Concurrently, the degradation of the autophagic substrate P62 is more active, and a higher density of autophagosomes is observed under electron microscopy [155]. This collective evidence strongly suggests that autophagy activated via the AMPK pathway is crucial for promoting fracture healing, while insufficient autophagic activity may be associated with the occurrence of non-union.

4.8. Aging

Aging is defined as the irreversible degradation of body structure and function accompanied by decreasing fitness and resistance due to increasing age, which is associated with a variety of causes, including genomic instability, telomere loss, epigenetic changes, loss of protein stability, impairment of nutrient sensitivity, mitochondrial dysfunction, cell senescence, stem cell exhaustion, and changes in intercellular communication [156]. Among these, protein stability and mitochondrial function are closely related to autophagy, and growing evidence suggests that autophagy is a critical factor in aging and aging-related diseases.

Ageing is accompanied by a progressive build-up of defective organelles and misfolded proteins within cells. By eliminating damaged organelles and accumulated proteins in cells, normal levels of autophagy support cellular homeostasis and are essential for increasing cell viability and prolonging lifespan. However, aging dampens autophagy, core regulators—ATG5, ATG7, and Beclin1—are markedly less abundant in aged tissues, and this decline is evident in the human brain. Therefore, enhancing autophagy may be beneficial for delaying aging and prolonging lifespan. Calorie restriction appears to improve human health by boosting autophagy through AMPK and SIRT1. This activation lowers the risk of diabetes, cardiovascular disease, cancer, and brain atrophy, while also extending lifespan and delaying aging [157]. With the same food intake, the mice with moderate overexpression of ATG5 exhibited increased autophagy activity, lower weight, increased median life span by 17.2%, and improved characteristics of aging, including increased insulin sensitivity and improved motor abilities. Furthermore, in mice with Beclin1 point mutations, enhanced autophagy was also accompanied by increased lifespan due to the disruption of the interaction between Beclin1 and Bcl-2.

Beyond its role in ageing itself, autophagy influences the onset and course of age-related disease [158]. Cancer is a typical progressive disease of aging. On the one hand, the cumulative mutation rate of genes rises with the number of cell divisions due to increasing age; on the other hand, the degree of protective mechanisms, such as immune response and autophagy, declines throughout aging, which makes cells more vulnerable to stress damage [159]. Therefore, the vast majority of cancer patients are over 65 years old, and the risk of cancer increases with age. Autophagy deficiency is also the basis for the occurrence and development of degenerative aging-related diseases, including neurodegenerative diseases, age-related macular degeneration (AMD), type 2 diabetes, osteoporosis, and AS (Table 3).

Table 3.

The role of autophagy in diseases.

Disease category	Disease	Protective effect	Disease-promoting effect
Cancer	Cancer (general)	1. Eliminates misfolded proteins and damaged organelles 2. Maintains genomic stability 3. Suppresses cellular damage and inflammation	1. Supports tumor cell survival under nutrient stress 2. Confers resistance to radiotherapy and chemotherapy
Neurodegenerative diseases	Parkinson’s disease	Clears α-synuclein, ROS, and damaged mitochondria	Lysosomal dysfunction causes autophagosome accumulation and neuronal death
	Alzheimer’s disease	Degrades amyloid-β and Tau proteins	Defective lysosomal or axonal trafficking enhances autophagosome accumulation and Aβ production
	Huntington’s disease	Removes mutant huntingtin (mHtt)	Impaired cargo recognition reduces mHtt clearance and accelerates disease progression
	Amyotrophic lateral sclerosis	Degrades mutant SOD1	Late-stage autophagy promotes synaptic dysfunction and apoptosis
Cardiovascular diseases	Myocardial ischemia-reperfusion injury	Maintains energy supply and promotes cardiomyocyte survival during ischemia	Excessive autophagy during reperfusion induces cardiomyocyte death
	Atherosclerosis	1. Degrades lipids 2. Regulates macrophage polarization 3.Reduces oxidative stress and apoptosis 4. Stabilizes plaques	Hyperactivated autophagy induces death of smooth muscle and endothelial cells, increasing plaque instability
	Heart failure	Enhances protein turnover and limits cardiac hypertrophy	Excessive autophagic activity exacerbates hypertrophy and cardiomyocyte loss
Metabolic diseases	Obesity	Maintains lipid metabolic homeostasis	Promotes adipocyte differentiation and adipogenesis
	Non-alcoholic fatty liver disease	1. Clears lipids 2. Eliminates ROS 3. Suppresses inflammation	Activation of hepatic stellate cells promotes liver fibrosis
	Diabetes	Preserves pancreatic β-cell survival and function	Autophagy impairment contributes to insulin resistance
Inflammatory & immune diseases	Viral hepatitis	1. Clears pathogens 2. Suppresses inflammasome activation and inflammatory cytokines	Facilitates viral replication
	Microbial infection	Eliminates pathogens via xenophagy	Certain pathogens exploit autophagy for immune evasion and replication
	Systemic lupus erythematosus	Clears apoptotic bodies	Excessive autophagy induces immune cell death
	Rheumatoid arthritis	Maintains immune tolerance	Inhibition of synovial fibroblast apoptosis and promotion of hyperproliferation
	Systemic sclerosis	Regulates collagen degradation and deposition	NA
	Inflammatory bowel disease	Suppresses inflammation and maintenance of epithelial integrity	NA
Aging	Aging	1. Removes damaged organelles and protein aggregates 2. Maintains cellular homeostasis 3. Extends cellular lifespan	NA
Bone disease	Osteoporosis	Maintains osteoblast function and supports bone formation by clearing damaged components	Impairs bone formation
	Osteoarthritis	Maintains chondrocyte homeostasis by clearing damaged mitochondria and protein aggregates	Induces chondrocyte death
	Fracture Healing	Promotes repair process	Its insufficient activity weakens cellular repair capacity, leading to delayed or non-union healing

*NA means not applicable.

5. Small-molecule compounds targeting autophagy

Understanding the function of autophagy helps to understand the significance of autophagy in diseases. As previously indicated, autophagy may play either a protective or destructive function in illness. Therefore, identifying and developing small-molecule drugs that may promote or inhibit autophagy is of enormous therapeutic significance. Fortunately, multiple autophagy signaling pathways and related proteins provide prospective targets for disease intervention therapy [160]. The current clinical drugs for autophagy regulation mainly include mTOR inhibitors, AMPK agonists, and chloroquine (CQ) or hydroxychloroquine (HCQ) that inhibit lysosomal function. Researchers have uncovered numerous small-molecule compounds that either boost or block autophagy by acting on key signaling pathways. Agents directed at ULK1, ATG4B, or VPS34 already display promising therapeutic activity [161]. Here, we mainly focus on these autophagy-targeting small-molecule compounds, which directly target ATG proteins or autophagy processes to play a role in disease treatment (Table 4).

Table 4.

Small-molecule compounds regulating autophagy.

Name	Mechanism	Condition or disease	Clinical status	NCT number
PI-103 (1)	PI3K/mTOR inhibition	NA	NA	NA
PI-540 (2)	PI3K/mTOR inhibition	NA	NA	NA
PI-620 (3)	PI3K/mTOR inhibition	NA	NA	NA
GDC-0941 (4)	PI3K/mTOR inhibition	Solid tumors Non-squamous non-small cell lung cancer	Phase 1 Phase 1	NCT00996892 NCT00974584
NVP-BEZ235 (5)	PI3K/mTOR inhibition	Malignant solid tumor Advanced solid tumors	Phase 1 Phase 1	NCT01343498 NCT01482156
GSK2126458 (6)	PI3K/mTOR inhibition	Idiopathic pulmonary fibrosis Solid tumors	Phase 1 Phase 1	NCT01725139 NCT00972686
PF-04691502 (7)	PI3K/mTOR inhibition	Cancer Hepatitis C	Phase 1 Phase 1	NCT00927823 NCT00445315
Voxtalisib (8)	PI3K/mTOR inhibition	Cancer Glioblastoma Lymphoma	Phase 1 Phase 1 Phase 2	NCT00485719 NCT01240460 NCT01403636
Gedatolisib (9)	PI3K/mTOR inhibition	Triple negative breast cancer Neoplasms	Phase 1/Phase 2 Phase 1	NCT03911973 NCT00940498
GDC-0980 (10)	PI3K/mTOR inhibition	Renal cell carcinoma Endometrial carcinoma Breast cancer	Phase 2 Phase 2 Phase 2	NCT01442090 NCT01455493 NCT01437566
NVP-BGT226 (11)	PI3K/mTOR inhibition	Solid tumors, breast cancer, Cowden syndrome	Phase 1/Phase 2	NCT00600275
Taselisib (18)	PI3K inhibition	Breast cancer Advanced solid tumors/Breast cancer	Phase 2 Phase 1	NCT02273973 NCT02389842
Perifosine (19)	AKT inhibition	Multiple myeloma Cancer Pediatric solid tumors	Phase 3 Phase 1 Phase 1	NCT01002248 NCT01224730 NCT01049841
MK-2206 (20)	AKT inhibition	Colorectal neoplasms Pancreas, neuroendocrine Recurrent nasopharyngeal carcinoma	Phase 2 Phase 2 Phase 2	NCT01333475 NCT01169649 NCT01370070
BI-69A11 (21)	AKT inhibition	NA	NA	NA
PP242 (23)	mTOR inhibition	NA	NA	NA
PP30 (24)	mTOR inhibition	Melanosis	Phase 2	NCT05119413
Ku-0063794 (25)	mTOR inhibition	NA	NA	NA
AZD8055 (26)	mTOR inhibition	Recurrent gliomas Solid tumors cancer	Phase 1 Phase 1 Phase 1	NCT01316809 NCT00731263 NCT00973076
AZD2014 (27)	mTOR inhibition	Advanced cancer Glioblastoma multiforme Non-small cell lung cancer metastatic Meningioma	Phase 1 Phase 1 Phase 2 Phase 2	NCT02193633 NCT02619864 NCT02117167 NCT03071874
Torin 1 (28)	mTOR inhibition	NA	NA	NA
Torin 2 (30)	mTOR inhibition	NA	NA	NA
Sapanisertib (31)	mTOR inhibition	Advanced solid tumors Recurrent lung non-small cell carcinoma Relapsed and refractory multiple myeloma	Phase 1 Phase 1 Phase 1	NCT02197572 NCT04479306 NCT01118689
OSI-027 (32)	mTOR inhibition	Solid tumor or lymphoma	Phase 1	NCT00698243
Palomid 529 (33)	mTOR inhibition	Age-related macular degeneration	Phase 1	NCT01033721
WAY-600 (34)	mTOR inhibition	NA	NA	NA
WYE-687 (35)	mTOR inhibition	NA	NA	NA
WYE-354 (36)	mTOR inhibition	NA	NA	NA
Rapamycin (37)	mTORC1 inhibition	Diabetes mellitus, Type 1 Facial angiofibroma Coronary heart disease Kidney transplant Tuberous sclerosis, lymphangioleiomyomatosis	Phase 3 Phase 3 Phase 4 Phase 4 Phase 2	NCT01060605 NCT03140449 NCT00402636 NCT00223678 NCT00457808
Temsirolimus (38)	mTORC1 inhibition	Non-Hodgkin’s lymphoma Mantle cell lymphoma Ovarian cancer	Phase 4 Phase 3 Phase 2	NCT01180049 NCT01646021 NCT00926107
Everolimus (39)	mTORC1 inhibition	Kidney transplant infection Neuroendocrine tumors Metastatic renal cell carcinoma	Phase 4 Phase 3 Phase 2	NCT03468478 NCT02246127 NCT01491672
Amiodarone (40)	Inhibiting mTORC1 signaling	Ventricular tachycardia, Ventricular fibrillation Atrial fibrillation COVID-19	Phase 4 Phase 3 Phase 2/Phase 3	NCT03855826 NCT01173809 NCT04351763
Niclosamide (41)	Inhibiting mTORC1 signaling	Familial adenomatous polyposis COVID-19 Colorectal cancer	Phase 2 Phase 3 Phase 2	NCT04296851 NCT04558021 NCT02519582
Perhexiline (42)	Inhibiting mTORC1 signaling	Hypertrophic cardiomyopathy Diastolic heart failure Myocardial reperfusion injury, cardiac output, low	Phase 2 Phase 2 Phase 2/Phase 3	NCT04426578 NCT00839228 NCT00845364
Rottlerin (43)	Inhibiting mTORC1 signaling	NA	NA	NA
Metformin (44)	AMPK activation	Diabetes mellitus, type 2 Pancreatic cancer Breast cancer	Phase 4 Phase 2 Phase 2	NCT00463502 NCT01210911 NCT01266486
A-769662 (45)	AMPK activation	NA	NA	NA
Gefitinib (46)	AMPK activation	Non-small cell lung cancer Stage IV lung cancer Lung neoplasms	Phase 3 Phase 2 Phase 3	NCT01544179 NCT01556191 NCT02518802
Bortezomib (47)	AMPK activation	Acute lymphoblastic leukemia/lymphoma Multiple myeloma Lymphoma	Phase 2/Phase 3 Phase 4 Phase 2	NCT03117751 NCT02268890 NCT01216683
GSK621 (48)	AMPK activation	NA	NA	NA
PT1 (49)	AMPK activation	NA	NA	NA
YLF-466D (50)	AMPK activation	NA	NA	NA
AICAR (51)	AMPK activation	Leukemia, B-Cell, chronic SMD Coronary artery bypass/myocardial infarction/ventricular dysfunction, left/stroke/Cardiopulmonary bypass	Phase 1/Phase 2 Phase 1/Phase 2 Phase 3	NCT00559624 NCT01813838 NCT00872001
OSU-53 (52)	AMPK activation	NA	NA	NA
Hernandezine (53)	AMPK activation	NA	NA	NA
Simvastatin (54)	AMPK activation	Myocardial infarction/inflammation/Endothelial dysfunction Hypercholesterolemia	Phase 4 Phase 3	NCT00906451 NCT03510715
M3A (55)	AMPK activation	NA	NA	NA
RSVA405 (56)	AMPK activation	NA	NA	NA
Buformin (57)	AMPK activation	NA	NA	NA
Dorsomorphin (58)	AMPK inhibition	NA	NA	NA
Oridonin (59)	ERK/p53 activation	Percutaneous coronary intervention	Phase 4	NCT05130892
SU11274 (60)	ERK/p53 activation	NA	NA	NA
Galangin (61)	p53 activation	NA	NA	NA
Fangchinoline (62)	p53 activation	NA	NA	NA
Prazosin (63)	p53 activation	Post-traumatic headache Alcohol abuse/Posttraumatic stress disorder Alcoholism	Phase 4 Phase 2 Phase 4	NCT02965027 NCT01518972 NCT00167687
DHA (64)	p53 activation	NA	NA	NA
JNK-IN-8 (66)	JNK inhibition	NA	NA	NA
SB203580 (68)	p38MAPK inhibition	NA	NA	NA
SB202190 (69)	p38MAPK inhibition	NA	NA	NA
SCH772984 (70)	ERK inhibition	NA	NA	NA
BVD-523 (71)	ERK inhibition	Advanced malignant solid neoplasm Uveal melanoma	Phase 2 Phase 2	NCT03155620 NCT03417739
LYN-1604 (82)	ULK1 activation	NA	NA	NA
BL-918 (83)	ULK1 activation	NA	NA	NA
Flubendazole (94)	ATG4B activation	NA	NA	NA
AS (95)	ATG4B activation	NA	NA	NA
BBP (96)	ATG4B activation	NA	NA	NA
xestospongin B (98)	IP3R inhibition	NA	NA	NA
Gossypol (99)	Up-regulate Beclin1	Non-small cell lung cancer	Phase 3	NCT01977209
ABT737 (100)	Up-regulate Beclin1	NA	NA	NA
EB1089 (101)	Up-regulate Beclin1	Liver neoplasms	Phase 3	NCT00051545
Tamoxifen (102)	Up-regulating Beclin1	Breast carcinoma Breast neoplasms Infertility Breast cancer Vaginal bleeding Endometrium	Phase 2 Phase 2 Phase 4 Phase 4 Phase 4 Phase 4	NCT00183963 NCT00919399 NCT02690870 NCT00537771 NCT04933240 NCT03060304
SGI-1776 (103)	PIM-2 inhibition	Relapsed and refractory leukemias Prostate cancer/Non-Hodgkins lymphoma	Phase 1 Phase 1	NCT01239108 NCT00848601
AZD1208 (104)	PIM-2 inhibition	Acute myeloid leukemia Advanced solid malignancies/Malignant lymphoma	Phase 1 Phase 1	NCT01489722 NCT01588548
HJ-PI01 (105)	PIM-2 inhibition	NA	NA	NA
9-ING-41 (106)	GSK3 inhibition	Myelofibrosis pancreatic adenocarcinoma/Pancreatic adenocarcinoma metastatic	Phase 2 Phase 2	NCT04218071 NCT05077800
SB216763 (107)	GSK3 inhibition	NA	NA	NA
TDZD8 (108)	GSK3 inhibition	NA	NA	NA
Laduviglusib (109)	GSK3 inhibition	NA	NA	NA
LRRK2-IN-1 (110)	LRRK2 inhibition	NA	NA	NA
GSK2578215A (111)	LRRK2 inhibition	NA	NA	NA
3-MA (12)	PI3K inhibition	NA	NA	NA
Wortmannin (16)	PI3K inhibition	NA	NA	NA
LY294002 (17)	PI3K inhibition	NA	NA	NA
SP600125 (67)	JNK inhibition	NA	NA	NA
SBI-0206965 (72)	ULK1/2 inhibition	NA	NA	NA
MRT67307 (78)	ULK1/2 inhibition	NA	NA	NA
MRT68921 (79)	ULK1/2 inhibition	NA	NA	NA
BX-795 (75)	ULK1 inhibition	NA	NA	NA
SR-17398 (80)	ULK1 inhibition	NA	NA	NA
SR-20295 (81)	ULK1 inhibition	NA	NA	NA
SAR405 (84)	VPS34 inhibition	NA	NA	NA
Vps34-in-1 (85)	VPS34 inhibition	NA	NA	NA
PIK-III (87)	VPS34 inhibition	NA	NA	NA
NSC185058 (89)	ATG4B inhibition	NA	NA	NA
UAMC-2526 (90)	ATG4B inhibition	NA	NA	NA
Z-FA-FMK (91)	ATG4B inhibition	NA	NA	NA
Z-FG-FMK (92)	ATG4B inhibition	NA	NA	NA
spautin-1 (97)	USP10/USP13 inhibition	NA	NA	NA
CMI (112)	Blocking autophagic flux	Obsessive compulsive disorder Depression Premature ejaculation	Phase 4 Phase 1 Phase 3	NCT00564564 NCT00913783 NCT01439984
DCMI (113)	Blocking autophagic flux	NA	NA	NA
ARN5187 (114)	Lysosomotropic	NA	NA	NA
CQ (116)	Lysosomotropic	SARS-CoV-2/Symptomatic Condition/COVID-19 HIV infections Malaria, falciparum Plasmodium vivax malaria Hepatitis C virus Malignant neoplasm/Solid tumors	Phase 4 Phase 2 Phase 2/Phase 3 Phase 4 Phase 4 Phase 1	NCT04351191 NCT00819390 NCT00677833 NCT04704999 NCT02058173 NCT02071537
HCQ (117)	Lysosomotropic	COVID-19 Rheumatoid arthritis Graft versus host disease	Phase 2 Phase 4 Phase 3	NCT04395768 NCT02466581 NCT00031824
MFQ (118)	Lysosomotropic	COVID-19 Urinary schistosomiasis	Phase 2/Phase 3 Phase 2	NCT04847661 NCT01132248
QN (119)	Lysosomotropic	Prostatic cancer	Phase 2	NCT00417274
VATG-032 (120)	Lysosomotropic	NA	NA	NA
VATG-027 (121)	Lysosomotropic	NA	NA	NA
Celecoxib (122)	Lysosomotropic	Prostate cancer Pain, postoperative Osteoarthritis/Pain Metastatic cancer	Phase 2 Phase 4 Phase 4 Phase 2	NCT01220973 NCT04790812 NCT00359151 NCT03864575
Lys01 (123)	Lysosomotropic	NA	NA	NA
Lys05 (124)	Lysosomotropic	NA	NA	NA
Bafilomycin A1 (125)	v-ATPase inhibition	NA	NA	NA
BRD1240 (126)	v-ATPase inhibition	NA	NA	NA
Diphyllin (127)	v-ATPase inhibition	NA	NA	NA
Matrine (128)	Impairing the function of lysosomal protease	NA	NA	NA
Lucathone (129)	Inducing lysosomal membrane permeabilization	Brain metastases/Non-small cell lung cancer Glioblastoma multiforme	Phase 2 Phase 2	NCT02014545 NCT01587144

*NA means not applicable.

5.1. Nonspecific autophagy modulators

Autophagy does not function as an isolated cellular process but is deeply integrated into the core regulatory networks that sense intracellular and extracellular cues, including nutrient status, energy levels, and stress signals [162]. Consequently, precise pharmacological intervention in autophagy is not limited to targeting the so-called “core machinery” proteins—such as ULK1, VPS34, or ATG4B—that directly participate in autophagosome formation. An equally valid and mechanistically well-defined strategy involves modulating key upstream signaling nodes. These nodes, which include the PI3K/AKT/mTOR, AMPK, p53, and MAPK pathways, act as signaling hubs [163]. They receive inputs from growth factors, energy status, genomic damage, and oxidative stress, and subsequently relay precise instructions to the core autophagic machinery via direct phosphorylation, transcriptional regulation, or allosteric protein–protein interactions. This upstream regulation ultimately determines whether autophagy is initiated, how it proceeds, and what functional outcome it produces. For instance, nutrient-sufficiency signals activate mTORC1, which directly phosphorylates and inhibits the ULK1/ATG13 complex, thereby physically blocking autophagy initiation. In contrast, energy depletion activates AMPK, which both suppresses mTORC1 to release this brake and directly phosphorylates and activates ULK1, thereby dually driving autophagic flux. Similarly, activated p53 transcriptionally upregulates a set of pro-autophagic genes, such as Sestrin2 and DRAM1, to programmatically induce autophagy in response to genotoxic stress [164]. Compounds targeting these upstream nodes exert their autophagic effects by interfering with defined signaling steps; these effects are both specific and functionally necessary. The resulting induction or suppression of autophagy can be abolished by genetic manipulation of the corresponding node, and the change in autophagic activity often critically contributes to the ultimate cellular fate decision, such as survival, death, or metabolic adaptation. Therefore, including such compounds in the category of autophagy modulators does not reflect an indirect or secondary mechanism, but rather highlights how autophagy is precisely controlled by higher-order cellular signaling networks. From a translational perspective, many of these upstream nodes are themselves validated drug targets; intervening in autophagy through these nodes often offers better druggability and greater potential for combination therapies. The following sections systematically review small-molecule compounds that modulate autophagy by acting on these upstream signaling nodes. Together with compounds that directly target the core autophagic machinery, they constitute a multi-layered and comprehensive toolbox for drug discovery and mechanistic investigation.

5.1.1. Compounds targeting PI3K/AKT/mTOR signaling pathway

Across a wide range of human disorders, the PI3K/AKT/mTOR circuit is among the most commonly altered signaling pathways. Once overactive, it fuels cell growth and survival, propelling disease and eroding the efficacy of many drugs. Drug design targeting the key proteins of this pathway for disease treatment has demonstrated promising application prospects in preclinical and clinical research.

5.1.1.1. Dual PI3K/mTOR inhibitors

Both PI3K and mTOR belong to the PI3KC3-related kinase (PIKK) superfamily of kinases; therefore, mTOR has high sequence homology with PI3K in the hinge region [165]. Dual PI3K/mTOR inhibitors appear more effective than selective agents at restraining the pathway and thereby tuning autophagy. The first described dual PI3K/mTOR inhibitor, PI-103 (1), is a morpholinoquinazoline derivative that has an inhibitory effect on the four catalytic subunits of PI3K (p110α, p110β, p110δ, p110γ) as well as mTOR. Compound 1 further blocks the phosphorylation of AKT and ULK1 to regulate autophagy by inhibiting PI3K and mTOR activities. Specifically, compound 1 increases the level of LC3-II in HeLa cells and causes autophagosome accumulation while it also impairs p62 clearance to block autophagic flux [166]. The outcome of autophagy inhibition is the induction of cell death and the inhibition of cancer cell proliferation and invasion. Unfortunately, compound 1 has limited solubility due to its tricyclic core structure. Additionally, the morpholine ring is critical for its activity; however, this ring is also an unstable moiety susceptible to hydroxylation and oxidation. Consequently, any modifications to the morpholine ring lead to a substantial loss of potency. On the other hand, the phenolic structure is prone to glucuronidation, which results in compound 1 with high plasma and tissue clearance [167]. The poor in vivo pharmacokinetic properties of compound 1 prevented its further clinical study. Subsequent structural optimization based on compound 1 facilitated the development of bicyclic thienopyrimidine derivatives PI-540 (2) and PI-620 (3) and the resulting clinical development candidate GDC-0941 (4). Compounds 2 and 3 retain the phenol ring and are substituted with a solubilizing group at position 6 [168]. Replacing phenol with indazole in compound 4 abolished glucuronidation and thus showed low plasma clearance and better oral bioavailability [169]. All three compounds had better pharmacokinetic properties than compound 1, and compound 4 exhibited excellent dose-dependent oral antitumor activity in the U87MG glioblastoma (GBM) xenograft model. In addition, compound 4 has entered a phase I clinical trial for the treatment of solid tumors, and a phase II clinical trial for the treatment of breast cancer [168] (Fig. 6).

Fig. 6.

Chemical structures of compounds 1–4.

Another widely reported dual ATP-competitive PI3K/mTOR inhibitor, Dactolisib (NVP-BEZ235, 5), is orally active and acts on the four catalytic subunits of class I PI3Ks (p110α/β/γ/δ) and mTOR [170]. The mechanism of compound 5 in regulating autophagy is similar to that of compound 1. Specifically, it blocks the activity of PI3K and mTOR kinases by interacting with their ATP-binding sites, which also inhibits autophagic flow and the fusion of autophagosomes and lysosomes [171]. Preclinical experiments have revealed that compound 5 can induce autophagy and exhibit powerful antitumor activity in GBM [172], chronic myeloid leukemia (CML) [173], and colorectal cancer [174], whether used alone or in combination with other anticancer drugs (vincristine, doxorubicin, or melphalan). In addition, the research of compound 5 on bladder cancer, pancreatic cancer, breast cancer, and other cancers has entered phase II clinical trials.

Omipalisib (GSK2126458, 6) is a highly selective and potent inhibitor of p110α/β/γ/δ and mTORC1/2 with inhibition constants (Ki) of 0.019/0.13/0.024/0.06 nM and 0.18/0.3 nM, respectively. In breast cancer cells, compound 6 markedly lowers p-AKT abundance, halting the cycle at G1 and provoking autophagy that suppresses growth [175]. At present, Compound 6 has completed the Phase I clinical trial for the treatment of advanced solid tumors [176]. In a randomized, double-blind trial, oral dosing of compound 6 suppressed the PI3K/mTOR axis in blood and lung tissue of idiopathic pulmonary fibrosis (IPF) patients in both dose- and time-dependent fashion. Of recent note, the combination of compound 6 with Remdesivir or Tipifarnib yielded some promising results in the cell-based SARS-CoV-2 inhibition tests.

PF-04691502 (7) was identified as an ATP-competitive dual PI3K/mTOR inhibitor with good antitumor activity and potently inhibits both PI3K and mTOR in kinase-selective assays. In addition to effectively inhibiting p-AKT levels in non-small cell lung cancer (NSCLC) cell lines A549 and H1299, compound 7 also induced apoptosis, autophagy, and DNA damage with a dose-dependent cellular toxicity. Interestingly, inhibition of autophagy promotes apoptosis and DNA damage induced by compound 7 in NSCLC cell lines [177]. Other dual PI3K/mTOR inhibitors undergoing clinical trials include Voxtalisib (XL765, 8) [178], Gedatolisib (PKI-587, 9) [179], Apitolisib (GDC-0980, 10) [180] and NVP-BGT226 (11) [181], which have encouraging performance in disease treatment (Fig. 7).

Fig. 7.

Chemical structures of compounds 5–11.

5.1.1.2. Pan-PI3K inhibitors

The PI3K family comprises class I, II, and III isoforms with distinct architectures and roles. Class I PI3K modulates autophagy chiefly through mTOR signaling, whereas the sole class III member, VPS34, operates as a core autophagy regulator. However, the contribution of class II PI3K activity to autophagy remains unclear. 3-methyladenine (3-MA, 12) was the first discovered and widely used autophagy inhibitor targeting VPS34, but this compound also inhibits the activity of class I PI3K, hence exerting dual effects on autophagy. And its off-target activity on p38MAPK or JNK may affect other cellular processes, such as glycogen metabolism, lysosomal acidification, endocytosis, and mitochondrial permeability transition. Due to the unfavorable characteristics of poor solubility and only being effective at high concentrations of around 10 mM, some researchers chemically modified the C6 position of compound 12 to generate a small-molecule compound library, resulting in three new autophagy inhibitors (compounds 13–15) with better solubility and efficacy than compound 12. However, the impact on cell function of these compounds is more severe, necessitating further optimization [182].

Wortmannin (16) is a natural product isolated from the fungus Penicillium wortmanni in 1957. It was initially found to have strong inhibitory cell proliferation and anti-inflammatory effects, but it was not until 1993 that it was also discovered to be able to inhibit PI3K activity. Compound 16 is a potent pan-PI3K inhibitor that inhibits kinase activity by irreversibly covalently binding to the ATP pocket of PI3K with an IC₅₀ of 10 to 50 nM for class I, II, and III PI3K, which can interfere with or block autophagosome formation [183]. Unfortunately, the poor selectivity of compound 16 enables it to inhibit other kinases.

LY294002 (17) was synthesized by structural modification of the natural product quercetin. Compound 17 is an ATP-competitive pan-PI3K inhibitor that suppresses its targets in the same manner as structurally related kinases. Specifically, compound 17 slips its morpholino loop into the ATP pocket, occupying the adenine site and forming a hydrogen bond with Val882 [184]. Compound 17 inhibited autophagy in cellular experiments with satisfactory solubility and bioavailability, and its selectivity for PI3K was greatly enhanced compared to the precursor quercetin. However, the IC₅₀ values of compound 17 for inhibiting PI3Kα/δ/β were all at the micromolar level, which limited its further therapeutic application due to the relatively poor inhibitory potency [185].

Taselisib (GDC-0032, 18), as an isoform-selective inhibitor of PI3Kα, is a compound with an imidazobenzoxazepin skeleton obtained after high-throughput screening and structure optimization [186]. Specifically, the low lipophilicity of the imidazobenzoxazepin structure contributes to reducing unbound clearance and improving solubility. And the α-quaternary amide substituent on the pyrazole ring makes compound 18 have a favorable balance in the properties of enzymatic activity, cellular activity, microsomal clearance, and kinetic solubility [186]. Studies indicated that compound 18 can induce autophagy in different breast cancer models, especially in the most resistant MDAMB231 cell line of triple-negative breast cancer (TNBC). In addition, a significant accumulation of p62-positive aggregates and LC3-II was observed when combined with CQ, along with a significant increase in apoptosis. Since autophagy is associated with tumor cell resistance, the combination of PI3K inhibitors and CQ is a promising strategy for the treatment of TNBC (Fig. 8).

Fig. 8.

Chemical structures of pan-PI3K inhibitors.

5.1.1.3. AKT inhibitors

Perifosine (19) is a member of the alkylphospholipid class of AKT inhibitors with IC₅₀ values of 0.6–8.9 µM for inhibiting the proliferation of different tumor cell lines. Studies have demonstrated that compound 19 not only inhibits AKT phosphorylation but also reduces the level of AKT in Perifosine-sensitive cell lines. Compound 19 may ultimately induce autophagy in multiple ways by blocking the assembly of mTORC2 and mTORC1, as well as promoting the degradation of mTOR, Raptor, and Rictor [187]. Furthermore, inhibition of compound 19-induced autophagy increased CML cell death [188]. Since compound 19 has various activities such as inducing autophagy, inducing apoptosis, and anti-tumor, numerous clinical studies are in progress for the treatment of pancreatic cancer, melanoma, leukemia, and other cancers.

MK-2206 (20) shows high selectivity, oral activity, and antitumor activity, which inhibits phosphorylation at Thr308 and Ser 473 of AKT with IC₅₀ values of 8, 12, and 65 nM for AKT1, AKT2, and AKT3, respectively. Compound 20 can be used alone or paired with cytotoxic or targeted agents. In a Phase I study of advanced solid tumors, it was well tolerated when given together with the autophagy inhibitor HCQ [189].

According to the virtual docking results, BI-69A11 (21) is identified as a novel AKT/NF-κB inhibitor. Compound 21 is currently attracting attention due to its interesting role in inducing apoptosis in melanoma, breast, prostate, and colon cancers. Similar to previous AKT inhibitors, compound 21-induced autophagy was also protective, and the combination of the autophagy inhibitor CQ increased apoptosis in colon cancer cells [190]. Substitution of benzimidazoles with different aryl groups, or N-methylation on the quinolinone ring, can lead to reduced activity or even complete inactivation. However, simply modest modifications to the structure of compound 21, such as introducing fluorine, chlorine, bromine atoms, or methyl and methoxy groups at different positions, may yield compounds with better activity. Compound 22, which is only substituted with fluorine on the benzimidazole ring, exhibits high activity, plasma and microsomal stability, and is anticipated to become a clinical candidate drug [191] (Fig. 9).

Fig. 9.

Chemical structures of AKT inhibitors.

5.1.1.4. Pan-mTORC inhibitors

Based on a protein-based screening assay and rational design, the first pan-mTOR inhibitor PP242 (23) and its analog PP30 (24) have been generated, selectively inhibiting mTOR with IC₅₀ values of 8 nM and 80 nM, respectively [192]. Importantly, compound 23 is a more effective mTORC1 inhibitor than rapamycin. And in a mouse model of leukemia, compound 23 inhibits mTOR signal transduction in CML cells and shows a stronger anti-tumor effect than rapamycin [193]. In addition, compound 23 can not only effectively inhibit tumor growth, but also significantly enhance the therapeutic effect of CML in combination with other anti-tumor drugs such as imatinib or dasatinib [194].

Ku-0063794 (25), a highly selective inhibitor of other PI3K family members and a powerful mTOR inhibitor (IC₅₀ = 2.5 nM), was discovered by high-throughput screening and structural optimization [195]. To further optimize the activity, water solubility, and selectivity of compound 25, substituting both C2 and C4 positions with 3S-methylmorpholine yielded AZD8055 (26), which has potent activity (IC₅₀ = 0.13 nM) and remains highly selective for every class I PI3K isoform as well as for the broader PI3K-like kinase family [196]. Compound 26 significantly inhibits mTORC2-dependent phosphorylation of AKT at the molecular level, and studies have shown that it reduces the phosphorylation of mTORC1 substrates, p70S6K and 4E-BP1, in rapamycin-resistant cells, which induces autophagy and the death of cancer cells [197]. Interestingly, compound 26 successfully combined with the MEK inhibitor AZD6244 to significantly promote tumor regression in an NSCLC xenograft model, indicating that the use of rational combination therapy to inhibit mTORC1/mTORC2 and MEK/ERK pathways is a clinically effective strategy [198]. As ATP-competitive inhibitors, both compound 23 and 26 block the phosphorylation of mTORC1 substrates and AKT, as well as prevent cancer cell proliferation and inhibit tumor growth by inducing autophagy or apoptosis in malignant tumors such as head and neck cancer, squamous cell carcinoma, and acute myeloid leukemia (AML). Although compound 26 showed good promise in preclinical disease models, it exhibited low bioavailability and poor pharmacokinetics in rats due to its benzyl alcohol and aryl methyl ether functional groups. Subsequently, removal of the methoxy group and replacement of benzyl alcohol with a primary amide yielded Vistusertib (AZD2014, 27, IC₅₀ = 2.8 nM) [196], which exhibited good cellular potency, water solubility, and low hepatic turnover. In MCF7 xenograft mice, compound 27 curbed tumor growth in a dose-related manner and is now being evaluated clinically for small-cell lung cancer, solid tumors, meningioma, and other malignancies.

Torin1 (28) emerged from a biochemical screen of heterocyclic compounds as a potent mTOR inhibitor, blocking both mTORC1 and mTORC2 with IC₅₀ values in the low nanomolar range [199]. Compound 28 is now widely used as an autophagy inducer that can induce autophagy to reduce neuronal death in 6-OHDA/ascorbic acid (AA)-injured PD models [200]. However, some unfavorable properties of compound 28, including low synthetic yield, short half-life (T_1/2 = 4 min), poor water solubility, and low oral bioavailability (bioavailability = 5.49%), need further optimization [201]. The scientists discovered that by lowering the molecular weight of compound 28, it might be possible to remove easily metabolized sites and increase water solubility. Based on the original skeleton of compound 28, the propionyl-piperazine moiety was first replaced with a fluorine atom to obtain compound 29, whose half-life was significantly improved (T_1/2 = 17.7 min), but the activity decreased. Further substitution of the quinoline moiety with aminopyridine and removal of the fluorine substitution resulted in Torin2 (30, bioavailability = 54%), which effectively inhibits mTOR with 800-fold selectivity for inhibition of PI3K and other protein kinases. More importantly, compound 30 has a roughly 10-fold higher bioavailability than compound 28 [202].

Sapanisertib (MLN0128, 31) is an oral pan-mTORC inhibitor developed by rational drug design with a level of nanomolar inhibitory effect. Compound 31 has entered clinical trials because of its strong anticancer activity in preclinical in vitro and in vivo models. Specifically, a Phase 1 dose-escalation study of compound 31 in patients with relapsed or refractory multiple myeloma and non-Hodgkin’s lymphoma, as well as a combination study with Alisertib in advanced solid tumors, has been completed [203,204]. Up to now, researchers have developed a variety of mTORC inhibitors, such as OSI-027 (32) [205], Palomid 529 (P529, 33) [206], WAY-600 (34), WYE-687 (35), and WYE-354 (36) [207] and have performed a series of optimization procedures to improve the solubility and pharmacokinetic properties of these compounds. Many of these compounds, which have been reported to exhibit promising inhibitory activity and disease therapeutic properties, have entered clinical trials (Fig. 10).

Fig. 10.

Chemical structures of pan-mTOR inhibitors.

5.1.1.5. mTORC1 inhibitors

mTORC1 plays a critical role in the early steps of autophagy. Under conditions of adequate nutrition and energy, actively expressed mTORC1 not only engaged in the activation of protein, lipid, and ribosome biosynthesis, but also in activating other processes responding to nutrients, growth factors, and cellular energy [208]. Conversely, loss of mTORC1 signaling results in cell growth arrest and induction of autophagy to restore or maintain energy and nutrient levels when nutrients and energy are deficient. mTORC1 is extremely sensitive to inhibition by rapamycin (37) with an IC₅₀ in the nanomolar range. Specifically, compound 37 interacts with the immunoaffinity protein FKBP12 to form a complex in mammalian cells, and then the FKBP12-rapamycin complex binds to the FRB domain in mTOR, which inhibits mTORC1 kinase activity. However, mTORC2 is generally insensitive to inhibition by compound 37, and its regulation remains unclear [208]. Although compound 37 is the most extensively researched mTORC1 complex inhibitor, its unfavorable pharmacokinetic profile is a serious concern. Fortunately, several rapamycin analogs have been identified, of which temsirolimus (CCI-779, 38, IC₅₀ = 1.76 µM) and everolimus (RAD001, 39, IC₅₀ = 5–6 nM) are two typical compounds that have shown striking antitumor effects against a variety of tumors in preclinical models. Furthermore, the U.S. Food and Drug Administration (FDA) authorized compound 38 in 2007 for the first-line treatment of advanced renal cell carcinoma (RCC), and the FDA has cleared compound 39 for patients with advanced renal cell carcinoma whose disease progressed on sunitinib or sorafenib [209].

Balgi et al. screened 3500 drugs and compounds with known pharmacological activity and identified three drugs approved for human use, amiodarone (40), niclosamide (41), perhexiline (42), and a pharmacological agent, rottlerin (43). Although these four compounds are not direct inhibitors of mTORC1, they can also induce autophagy by inhibiting mTORC1 signaling, which has significant application value [210] (Fig. 11).

Fig. 11.

Chemical structures of mTORC1 inhibitors.

5.1.2. Compounds targeting the AMPK signaling pathway

AMPK sits at the core of energy control, adjusting metabolism and growth to keep cellular fuel in balance. Dysregulated AMPK activity may contribute to metabolic disturbances and lead to diseases, including obesity, type 2 diabetes, and cancer. According to some studies, AMPK plays a crucial role in the early process of autophagy by phosphorylating mTOR, ULK1, Beclin1, and TSC2 [211]. Therefore, compounds targeting AMPK are of great significance for regulating autophagy.

5.1.2.1. Compounds that activate AMPK signaling

Metformin (44), a well-known antidiabetic drug, also has additional pharmacological effects, including antitumor, antiaging, neuroprotective, and the prevention of polycystic ovary syndrome. Increasing evidence indicates that autophagy is an important mediator of metformin’s pharmacological effects. Compound 44 is one of the AMPK activators widely used as autophagy inducers, and it may trigger autophagy through distinct AMPK downstream molecules depending on the context [212]. For example, in aged mice, compound 44 induces autophagy via the AMPK-ULK1 pathway, thereby attenuating sevoflurane-induced neurocognitive impairment. This finding suggests its potential for preventing cognitive impairment following anesthesia surgery in elderly patients [213]. In addition, compound 44 promotes autophagic cell death by activating the AMPK/CEBPD pathway, thereby increasing the sensitivity of epidermal growth factor receptor (EGFR)-overexpressing HCC patients to sorafenib [214]. Already approved as first-line therapy for type 2 diabetes, compound 44 is now being tested in trials for pancreatic, breast, and other cancers.

The researchers first screened and identified a compound with a non-nucleoside thienopyridone structure as a direct activator of AMPK, then optimized its structure to obtain A-769662 (45), a potent and reversible AMPK activator (EC₅₀ = 0.8 µM) [215]. Compound 45 activates AMPK through catalyzing Thr172 phosphorylation on the α-subunit and stabilizing the active conformation of AMPK by allosterically binding to the γ-subunit [215]. One study revealed that compound 45 treatment could restore LPS-induced autophagy inhibition, with specific evidence that compound 45 activated AMPK and inhibited LPS-induced elevation of IL6, which attenuated lung histological abnormalities and improved survival in LPS-induced model mice [216]. In addition, in osteoblasts from human and mouse, compound 45 boosted H₂O₂-triggered AMPK-dependent autophagy and thereby protected against oxidative death and apoptosis. But compound 45-induced cytoprotection was attenuated when AMPK was inactivated or depleted [217].

Gefitinib (ZD1839, 46) is a potent, selective, and orally active EGFR tyrosine kinase inhibitor with an IC₅₀ of 33 nM. Compound 46 competitively blocks the binding of EGFR to ATP, which inhibits EGFR phosphorylation and blocks downstream signal transduction. Therefore, it exhibits excellent antitumor activity and tolerance in the treatment of non-small cell carcinoma [218]. Recent research has demonstrated that compound 46 can activate AMPK through LKB1, and then AMPK directly activates ATG1 through phosphorylation of Ser317 and Ser777 to induce high levels of autophagy [219]. Everolimus (an autophagy agonist) strengthens compound 46′s antitumor action, whereas autophagy inhibitor 3-MA or ATG5 knockdown weakens it, showing that the drug’s effect depends on autophagy [220].

Bortezomib (47), as a reversible and selective proteasome inhibitor, induces protective autophagy by triggering endoplasmic reticulum stress and energy disruption, thereby activating the AMPK/ULK1 pathway. This autophagy aims to clear ubiquitinated protein aggregates accumulated due to suppressed proteasome function, serving as a crucial mechanism for cellular drug resistance development [221]. Notably, inhibition of AMPK activity using inhibitors or RNAi significantly enhanced the cytotoxicity of compound 47, which may have important implications for enhancing the therapeutic efficacy of compound 47 [222]. The FDA has approved compound 47 for the treatment of refractory or relapsed multiple myeloma (MM).

GSK621 (48) is a novel AMPK activator with IC₅₀ values of 13–30 µM against AML cell lines. Different from the effect of compound 44 on the AMP/ATP ratio, compound 48 activates AMPK by directly binding to the β subunit of AMPK. And this activation of AMPK induces mTORC1-independent autophagy and eventually leads to autophagic death, which is an important reason for the cytotoxicity of compound 48 to tumor cells [223].

Another novel AMPK activator, PT1 (49) (EC₅₀ = 8.7 µM), activates AMPK by antagonizing the autoinhibition of the α subunit [224]. In vivo administration of compound 49 and 3HOI-BA-01 in a murine myocardial (I/R injury model) remarkably reduced myocardial infarct size and induced autophagy, which could protect cardiomyocytes from I/R injury [225]. However, the 2-imino-4-thiazolidinone structure of compound 49 exists in the form of amino or imino tautomers in solution. To eliminate this tautomerization, the 2-imino-4-thiazolidinone of compound 49 was replaced with 3-alkylideneoxindole, and further substitution with various electron-withdrawing or electron-donating groups yielded another AMPK agonist, YLF-466D (C24, 50, EC₅₀ = 1.21 µM) [226]. Compound 50 activates AMPK by a similar mechanism to compound 49, but has the advantage of higher oral bioavailability and closer potency to AMP, and performs better in the ADMET assay [226,227].

AICAR (51), as an AMP analog, is one of the most commonly used AMPK agonists, but its capacity to activate AMPK is 40–50 times less than that of AMP. Compound 51 activates AMPK by binding directly to the γ subunit of AMPK to allosterically activate AMPK, and by phosphorylating Thr172 of LKB1 to activate AMPK indirectly [228]. According to research, the combined usage of compound 51 and tetrandrine not only enhanced the autophagic flux but also enhanced the growth inhibition and apoptosis-inducing effects of tetrandrine in bladder cancer cells [229]. However, compound 51 accumulates in cells at millimolar concentrations and exerts AMPK-independent or off-target effects, thus limiting its further applications.

Orally bioavailable OSU-53 (52) can activate AMPK and inhibit mTOR simultaneously, which has been demonstrated to play an inhibitory role in various cancers, including thyroid cancer and breast cancer. Compound 52 directly activates AMPK by antagonizing the autoinhibition of the α subunit with an EC₅₀ of 2–5 µM. In contrast to compound 51, which requires intracellular phosphorylation to exert its activity, compound 52 functions directly without further modification [230,231]. Meanwhile, compound 52 directly inhibits mTOR independent of AMPK activation [231]. Additionally, compound 52 exhibited enhanced activation of AMPK, inhibition of mTOR signaling, and stimulation of autophagy in RAS and BRAF^V600E mutant cells. Due to its dual ability to activate AMPK and inhibit mTOR, compound 52 has the potential to become a potent new targeted medication for the treatment of cancer.

Hernandezine (53) and simvastatin (54) are two other AMPK agonists. Compound 53, an isoquinoline alkaloid, induces autophagic cell death in apoptosis-deficient or anti-apoptotic cancer cells, which has shown effective cytotoxicity on various cancer cells, such as liver cancer (HepG2 cells, IC₅₀ = 7.42 µM) and lung cancer (H1299 cells, IC₅₀ = 6.74 µM) [232]. Compound 54, a classic HMG-CoA reductase inhibitor, is widely used in the treatment of cardiovascular diseases. Recent work shows that compound 54 turns on AMPK and, via the AMPK–SKP2–CARM1 pathway, boosts cytoprotective autophagy [233]. In addition, compound 54 also induces autophagy through the LKB1-AMPK-mTOR pathway, thereby inducing astrocytes to secrete insulin-degrading enzyme (IDE) through an autophagy-based unconventional secretory pathway. This role of compound 54 ultimately enhances the extracellular clearance of Aβ and also provides a potential approach for AD therapy [234]. Besides, malvidin-3-O-arabinoside chloride (M3A, 55) [235], RSVA405 (56) [236] and buformin (57) [237] may also induce autophagy by activating the AMPK pathway (Fig. 12).

Fig. 12.

Chemical structures of compounds 44–58.

5.1.2.2. Compounds that inhibit AMPK signaling

Although AMPK is an extremely important kinase that is involved in cell growth, proliferation, survival, and metabolic regulation, there is currently no AMPK-specific inhibitor. A chemical called dorsomorphin or compound C (58) is occasionally used as an AMPK inhibitor with a K_i of 109 nM in the absence of AMP. Unfortunately, when compound 58 is used as an AMPK inhibitor, caution should be exercised in interpreting the results due to its poor specificity and the fact that it even inhibits other kinases more effectively than AMPK [238]. Numerous pathways, including JNK, MAPK, and p53, have been proven in studies to be engaged by compound 58 to induce autophagy [239]. In addition to suppressing AMPK/TET2-mediated FOXP3 expression and promoting benzene-induced autophagy-dependent apoptosis in AML cells [240], compound 58 induces autophagy in cancer cells through inhibiting AMPK to independently downregulate AKT/mTOR signaling [241]. In conclusion, variables such as dosage, cell type, and environment may influence the impact of compound 58 on autophagy (Fig. 12).

5.1.3. Compounds targeting p53 signaling pathway

p53 is a tumor suppressor that controls a broad gene network. DNA damage, oxidative stress, oncogene activation, and other stressors can activate p53, which can bind to the promoter regions of multiple genes encoding autophagy regulators, such as AMPK, DAPK1, DRAM, Bcl-2, Sestrin, and TSC2 [76]. Notably, p53 has dual roles of inducing and inhibiting autophagy depending on the situation. p53 has long been viewed as undruggable, largely because it lacks a deep hydrophobic pocket that readily accommodates high-affinity small molecules. Therefore, most of the small-molecule compounds modulate p53 signaling by targeting key proteins in the pathway.

Oridonin (59), an active diterpenoid derived from Rabdosia rubescens, acts as an antioxidant, antimicrobial, anti-inflammatory, pro-apoptotic, and anti-tumor agent. Compound 59 was originally identified as an AKT inhibitor, but recent studies indicate that it also affects the p53 pathway. Specifically, the mechanism by which compound 59 induces autophagy involves a distinct signal amplifier that promotes nitric oxide (NO) production by activating the ERK-p53 pathway. This NO further enhances the pathway, forming a positive feedback loop. A key downstream event in this cycle is the reduction of Bcl-2 protein levels and direct activation of Beclin1, thereby specifically initiating autophagosome formation [242].

SU11274 (60) was initially identified as a typical cellular-mesenchymal-epithelial transition factor (c-Met) inhibitor (IC₅₀ = 10 nM) that inhibits the growth of non-small cell lung cancer (NSCLC) cells and even promotes apoptosis. Recent research indicates that compound 60 may also induce autophagy. Specifically, after the inhibition of c-Met by compound 60, both ERK and p53 were activated, resulting in the upregulation of autophagy. The possible mechanism is that activated ERK disrupts the association of Beclin1 and Bcl-2 to induce autophagy by promoting the phosphorylation of Bcl-2. Interestingly, upregulation of autophagy levels also enhanced compound 60-induced cell death [243].

Galangin (61) is a flavonoid polyphenolic compound that exhibits excellent activity in the treatment of liver cancer. This compound highlights the central role of p53 in determining the functional output of autophagy. At low concentrations, it strictly relies on p53 to induce protective autophagy (an effect blocked by p53 inhibitors); at high concentrations, it shifts toward apoptosis. This provides a clear model for studying p53 as a molecular switch regulating cellular fate (autophagy versus apoptosis) [244].

Recently, fangchinoline (62) has been reported as a new type of anti-tumor drug with anti-inflammatory and analgesic effects. Compound 62 was shown to activate Sestrin2 by promoting nuclear translocation of p53, and subsequently activating AMPK signaling to induce autophagic cell death (IC₅₀ ≈ 5 mM) in hepatoma cells HepG2 and PLC/PRF/5, suggesting that compound 62 has the potential to be a chemotherapeutic agent [245].

Prazosin (63), a classic α1-adrenergic receptor antagonist, is routinely used to treat mild or moderate hypertension and congestive HF. This clinical antihypertensive agent was found to trigger autophagy by affecting energy homeostasis. The mechanism begins with inducing p53 activation, which subsequently inhibits glucose transporter type 1 (GLUTI) and GLUT4, leading to decreased intracellular ATP levels. The resulting energy crisis directly activates AMPK, thereby inducing autophagy through the canonical pathway [246].

The macrocyclic bisbibenzyl compound dihydroptychantol A (DHA, 64) was also demonstrated to induce autophagy, which subsequently leads to apoptosis with cell cycle arrest in the G₂/M phase of U2OS cells in human osteosarcoma. The possible mechanism is that compound 64-induced p53 phosphorylation increases the nuclear expression of p53, thereby upregulating the expression of p53 target gene p21 (Waf1/Cip1), which mediates p53-related autophagy [247] (Fig. 13).

Fig. 13.

Chemical structures of compounds 59–64.

5.1.4. Compounds targeting the MAPK signaling pathway

MAPK signaling is transmitted via a phosphorylation cascade involving MAP3K, MAP2K, and MAPK, a mechanism that has been well characterized in earlier studies. This pathway exerts critical control over diverse cellular events, such as inflammatory responses, stress adaptation, proliferation, developmental programs, differentiation, and cell death. Among the six identified MAPK subgroups, JNK, p38 MAPK, and ERK have been specifically implicated in the modulation of autophagy [58].

5.1.4.1. JNK inhibitors

Based on structure-based drug design, JNK-IN-1 (65) was identified as a JNK inhibitor that inhibits JNK1, JNK2, and JNK3 with an IC₅₀ greater than 4 µM. Subsequently, compound 65 was further optimized. The linking fragments of pyrimidine and acrylamide were replaced by 1,4-diphenylamine and 1,3-benzamide, resulting in the generation of JNK-IN-8 (66) with significantly improved activity, and the IC_50s for JNK1, JNK2, and JNK3 are 4.7 nM, 18.7 nM, and 1 nM, respectively. This improved activity may be attributed to the more optimized position of the acrylamide of compound 66 relative to Cys154, resulting in a more efficient formation of a covalent bond [248]. Additionally, compound 66, as an ATP-competitive JNK kinase covalent inhibitor, may target cysteines conserved among all JNK kinases. Notably, the regulation of autophagy by JNK is more nuanced than anticipated. JNK1/2 do not respond to starvation or hypoxia-induced autophagy but inhibit autophagy in hepatocytes, while compound 66 administration results in lysosomal inhibition and increased autophagic flux [249]. Another JNK inhibitor, SP600125 (67), inhibits hypoxia- and oxaliplatin-induced autophagy in HT29 colon adenocarcinoma cells, which enhances the sensitivity of cancer cells to oxaliplatin [250]. These JNK inhibitors exert their effects by intervening in a specific step of autophagy regulation: activated JNK phosphorylates Bcl-2, thereby weakening its interaction with Beclin1. Consequently, inhibiting JNK activity helps stabilize the Bcl-2/Beclin1 complex, indirectly suppressing autophagy initiation. This provides a pharmacological tool for investigating the fine-tuned regulation of autophagy flux by JNK signaling under specific stress conditions (Fig. 14).

Fig. 14.

Chemical structures of compounds 65–71.

5.1.4.2. p38 MAPK inhibitors

Adezmapimod (SB203580, 68) was discovered to be the first p38 MAPK inhibitor that prevents substrate phosphorylation in an ATP-competitive manner without inhibiting JNK activity. A study indicates that intra-articular injection of compound 68 in rabbits can prevent early osteoarthritis by activating autophagy [251]. SB202190 (69) is another widely used p38 MAPK inhibitor with various pharmacological effects, including the induction of autophagy, apoptosis, and antitumor activity [[252], [253], [254]] (Fig. 14).

5.1.4.3. ERK inhibitors

SCH772984 (70) emerged from affinity-mass spectrometry screening followed by structure-guided refinement and blocks ERK1 and ERK2 in an ATP-competitive manner [255]. As an ERK1/2 inhibitor, it induces autophagy closely linked to energy-sensing pathways. Its mechanism involves AMPK activation and mTORC1 signaling inhibition, suggesting that ERK signaling suppression may trigger an energy stress-like state, thereby shifting toward autophagy-dependent metabolic adaptation [255]. Another ERK inhibitor, Ulixertinib (BVD-523, 71), exhibited acceptable safety and favorable pharmacokinetics in phase I clinical trials for the treatment of advanced solid tumors [256] (Fig. 14).

5.2. Specific autophagy modulators

5.2.1. ULK1/2 modulators

ULK1 to ULK4 and serine/threonine kinase 36 (STK36) have been identified as five ATG1 homologs in mammals. Among them, only ULK1 and ULK2 have been confirmed to regulate autophagy. Upon starvation, the inhibition of mTORC1 activates the ULK1/2 complex to initiate autophagy [211]. From a functional point of view, the ULK1/2 complex has a critical role in the initiation of autophagy and is, therefore, an interesting target for drug development.

5.2.1.1. ULK1/2 inhibitors

SBI-0206965 (72), a potent and selective ULK1 inhibitor, was identified through a reverse-pharmacology screen. Compound 72 inhibits ULK1-mediated intracellular phosphorylation events such as the phosphorylation of VPS34 and Beclin1. Dose-response analysis of compound 72 in vitro revealed the IC₅₀ value of 108 nM and 711 nM for ULK1 and ULK2, respectively [257]. Compound 72 inhibits autophagy that is induced by mTOR inhibition, while also inhibiting cell survival after nutrient deprivation. Additionally, the combined treatment of compound 72 and mTOR inhibitors can produce cytotoxicity and significantly induce apoptosis, which provides a novel approach for lowering mTOR dosage in cancer treatment [257].

To stabilize the ULK1 kinase domain for structural work, Lazarus et al. screened small molecules in vitro and isolated compound 73 (IC₅₀ = 160 nM), which not only inhibits ULK1 but also stabilizes its crystal structure. The molecule anchors through its aminopyrazole core, donating a hydrogen bond to the ATP-site hinge; the aniline reaches the DFG motif to engage Asp165, while the cyclopropyl group tucks into a pocket beside the gatekeeper methionine [258]. Subsequently, under the guidance of the crystal structure, a series of pyrazole aminoquinazoline derivatives were synthesized, among which compound 74 exhibited significantly enhanced activity (IC₅₀ = 8 nM). According to SAR, cyclopentyl derivatives are inactive, while cyclobutyl can maximize the use of steric volume. Replacing the aniline with benzimidazole triggers a conformational shift: Asp165 swings inward, creating space for the larger ring while simultaneously donating hydrogen bonds to Gln142 and Lys46 within the kinase domain [258]. However, the poor selectivity limits the further application of compound 74. Subsequently, compounds 76 and 77 were developed with the aminopyrimidine structure compound BX-795 (75) as the lead compound, which are two more selective inhibitors of ULK1. The iodine atom occupies a similar space to the cyclobutyl substituent of compound 74, and there may be a dipole-dipole interaction with the gatekeeper methionine. Compound 77 binds in the same pose as the earlier lead 74, its diaminophenyl ring mapping onto the former quinazoline site. A pyrrolidine-urea arm now projects from this ring into a pocket region that had remained vacant, conferring added selectivity to the scaffold [259]. Importantly, these compounds have inhibitory effects on autophagy and are expected to be pharmacological tools for selectively inhibiting autophagy (Figs. 15,16).

Fig. 15.

Chemical structures of compounds 72–77.

Fig. 16.

Structures of 73 (a, PDB code 4WNO), 74 (b, PDB code 4WNP), and 77 (c, PDB code 5CI7) with the ULK1 kinase domain. The ligands are shown in a stick model colored by C in cyan, N in indigo, O in red, and I in purple. The black dashed line indicates a conventional hydrogen bond, and the red dashed line indicates an alkyl interaction.

MRT67307 (78) was initially identified as a TBK1 inhibitor. After screening of known kinase inhibitors in vitro, it was found that compound 78 also efficiently inhibited ULK1/2 with IC₅₀ values of 45 and 38 nM for ULK1 and ULK2, respectively. MRT68921 (79), as an analog of compound 78, is a more potent inhibitor with an affinity 10-fold higher for ULK1 and ULK2 (IC₅₀ of 2.9 and 1.1 nM, respectively) [260]. Although both compounds inhibit ULK1 and autophagy in vitro or in cells, their low specificity for ULK1/2 also enables them to inhibit AMPK and other kinases [261].

Based on the crystal structure of ULK1, Spencer D. Wood et al. identified SR-17398 (80, IC₅₀ = 22.4 µM) as a modest ULK1 inhibitor through an in silico screen. In the docking model, the indazole nitrogens hydrogen bond to Glu93 and Cys95 in the hinge, while the 5-amide carbonyl contacts Lys46, and the cyclohexyl amine engages Asn143 [262]. Compound 80 was subsequently further modified by structure-based rational drug design. Adding an amino substitution at the 3-position of the indazole ring generates a new H-bond interaction with the amide carbonyl of Cys95 in the ATP-binding domain. Further introducing the naphthyl group resulted in a more potent ULK1 inhibitor SR-20295 (81, IC₅₀ = 45 nM), which is expected to be further developed as a selective ULK1 molecular probe to evaluate whether selectively targeting autophagy is an effective anticancer strategy [262] (Fig. 17).

Fig. 17.

Chemical structures of compounds 78–83.

5.2.1.2. ULK1/2 agonists

ULK1 is markedly reduced in breast tumors, with the lowest levels in TNBC, and restoring ULK1-driven autophagy shows therapeutic promise. Through in silico screening, synthesis, kinase assays, and mutagenesis, LYN-1604 (82) was obtained as the first structure-guided ULK1 agonist (EC₅₀ = 18.94 nM). Molecular docking indicated that the piperazinyl group of compound 82 forms a hydrogen bond with Lys50, the naphthyl and benzene ring moieties form hydrophobic interactions with Leu53 and Tyr89, and the presence of methylene groups near the carbonyl group and bulky N substitution play key roles in biological potency. Mechanistic studies have found that compound 82, with significant antitumor activity, induces autophagy or autophagy-related cell death through the ULK1 complex. And compound 82 is expected to become a new small-molecule drug for the treatment of TNBC [263,264].

BL-918 (83) is another ULK1 agonist discovered by structure-based drug design with an EC₅₀ of 24.14 nM. The SAR indicated that π-cation between compound 83 and key residues Arg18 and Lys50 could guarantee high affinity, and the interaction between compound 83 and Ala85, Asn86, as well as Tyr89, enhanced potency and selectivity [265]. Experiments confirmed that compound 83 induced cytoprotective autophagy in SH-SY5Y cells through the ULK1 complex and exerted neuroprotective effects in an MPTP mouse model of PD by targeting ULK1-regulated autophagy. Therefore, compound 83 may become a drug candidate for the treatment of PD [265]. A new study suggests that compound 83 displays therapeutic potential for ALS through inducing cytoprotective autophagy [266] (Fig. 17).

5.2.2. VPS34 inhibitors

VPS34, the sole class III PI3K, catalyzes lipid phosphorylation that initiates autophagosome membrane assembly. VPS34 exists in multiple complexes composed of several subunits, including VPS34/PIK3C3, VPS15, Beclin1/ATG6, AMBRA1, ATG14, and UVRAG, which are involved in various cellular functions such as phagocytosis, endocytic traffic, and autophagy [267]. Therefore, compounds targeting the PIK3C3/Vps34 complex are important for the regulation of autophagy.

SAR405 (84), a potent VPS34 inhibitor (IC₅₀ = 1.2 nM) with high specificity, was identified through a novel image-based high-throughput cell phenotype screening and chemical optimization. Compound 84 inhibits starvation or mTOR inhibitor-induced autophagy in HeLa and H1299 cells [268]. Furthermore, a recent study showed that compound 84 exhibited synergy with the mTOR allosteric inhibitor everolimus. Co-treatment of compound 84 with everolimus inhibited the proliferation of kidney and lung tumor cells in vitro [269]. Therefore, the synergistic effect of compound 84 and chemotherapeutic drugs in multiple cancer types is increasingly worth investigating.

Vps34-in-1 (85) is the first reported bis-aminopyrimidine VPS34 inhibitor with high selectivity and cell permeability (IC₅₀ = 25 nM), which not only inhibits the production of PI3P by the VPS34 complex but also serves as a pharmacological tool for exploring VPS34 function [270]. A study reported that in addition to inhibiting autophagy through VPS34, compound 84 blocks vesicular traffic and mTORC1 signaling, triggering apoptosis in AML cells while sparing normal CD³⁴⁺ progenitors [271]. This effect of compound 84 on the pathogenesis of AML is novel and might lead to a new treatment approach.

A novel bis-aminopyrimidine compound 86 was identified as a potent inhibitor of VPS34 in high-throughput screening [272]. To further optimize its potency, selectivity, and pharmacokinetics (PK) in vivo, researchers discovered that the cyclohexanol moiety results in a high clearance in vivo through glucuronidation after SAR analysis. Subsequent molecular docking and X-ray crystallography revealed that the cyclohexanol moiety is solvent-exposed, and the introduction of a pyridine ring provided an incremental enhancement in potency. In addition, N-demethylation and cyclopropyl oxidation are the main metabolic pathways of compound 86. The cyclopropyl group nestles into a narrow pocket lined by Phe612, Pro618, and Phe684, providing the main basis for selectivity. Removing the methyl group gives PIK-III (87, IC₅₀ = 18 nM), which shows greater metabolic stability and at least a 100-fold preference for VPS34 over PI3Kα or mTOR [272]. Compound 87 not only inhibits LC3 lipidation and prevents the degradation of autophagic substrates through VPS34 in cells, but also inhibits autophagy induced after tyrosine kinase inhibitor (TKI) treatment in CML, which reduces leukemia stem cell (LSC) survival [273,274]. Compound 87 was further substituted with an alkyl group on the amino group and added a hydroxyl group to yield compound 88 with excellent activity (IC₅₀ = 15 nM), selectivity, and metabolic stability against VPS34 inhibition [272] (Figs. 18,19)

Fig. 18.

Chemical structures of compounds 84–88.

Fig. 19.

Binding modes of compound 88 (PDB code 5ENN) in complex with Vps34. The ligands are shown in a stick model, colored by C in cyan, N in indigo, and O in red. The black dashed line indicates a conventional hydrogen bond, and the purple dashed line indicates the π-π interaction.

5.2.3. ATG4B modulators

During autophagosome formation, ATG8 is initially cleaved by ATG4 to expose the C-terminal glycine and then coupled to PE with the participation of ATG7 and ATG3. Lipidated ATG8 subsequently promotes membrane expansion and autophagosome formation. On the other hand, ATG4 also has a delipidase activity that delipidates Atg8-PE to recycle ATG8 from the membrane. ATG4A, ATG4B, ATG4C, and ATG4D have been identified as four ATG4 homologs in mammals. Notably, ATG4B is the only enzyme that has been demonstrated to efficiently convert pro-LC3 to LC3-I in the autophagy cascade, and ATG4B upregulation or downregulation may lead to abnormal progression of autophagy [275]. Thus, ATG4B is a viable target for the pharmacologic autophagy modulation.

5.2.3.1. ATG4B inhibitors

NSC185058 (89) emerged from in silico docking of NCI compounds to ATG4B. It lodges in the active site, engaging both polar and non-polar contacts around His280 and Asp278, residues essential for the enzyme’s proteolytic cleavage. It has been reported that compound 89 effectively inhibited starvation-induced autophagy (IC₅₀ = 51 µM) by suppressing the cleavage of ATG4B and the lipidation of LC3B, which reduced protein degradation promoted by starvation by >50% [276]. Furthermore, the inhibitory effect of compound 89 on Saos-2 osteosarcoma in vivo may be related to the inhibition of autophagy. Another study revealed that the inhibition of ATG4B activity by shRNA knockdown or ATG4B inhibitor suppressed autophagy and tumorigenicity of GBM cells, suggesting that compound 89 has potential value as an ATG4B inhibitor for cancer therapy [277].

Another potent ATG4B inhibitor, UAMC-2526 (90), is a benzotropolone derivative. The ratio of LC3-II/LC3-I, the LC3 immunoreactive spots, and GFP fluorescent spots were decreased in hepatocytes treated with compound 90, while the scaffold protein SQSTM1 was significantly increased, indicating that compound 90 inhibits autophagosome formation and starvation-induced autophagy in vivo [278]. Importantly, compound 90 proved stable in plasma with a half-life of about 2 h. Interestingly, compound 90 can largely prevent autophagy induced by the standard chemotherapy drug oxaliplatin in tumor tissues. Although both compound 90 and oxaliplatin inhibited tumor growth, only the combination produced a striking shrinkage of tumor volume [278].

Z-FA-FMK (91), identified by a high-throughput screening based on time-resolved fluorescence resonance energy transfer (TR-FRET) assay, exhibits high inhibitory activity against ATG4B with an IC₅₀ value of 14.80 µM [279]. Compound 91 mimics the polypeptide sequence of LC3, the natural substrate of ATG4B, and the Fmk warhead forms a stable covalent bond with the S atom of Cys74 of ATG4B. The chemically modified Z-FG-FMK (92) differs from compound 91 by only one methyl group, but it is 10 times more potent than compound 91 (IC₅₀ = 1.16 µM) [279]. The molecular docking model suggested that it might be due to the potential steric conflict between the methyl group in compound 91 and Trp142. In conclusion, the interaction between FMK-based antagonists and ATG4B can be divided into six important segments based on predicted binding mode and SAR. Crystal-guided optimization revealed the critical step: the FMK warhead forms a covalent bond with Cys74. The P1 glycine stacks onto Trp142’s indole, and the P2 benzyl mimics LC3’s phenylalanine. Between these fragments, the amide carbonyls tether buried waters, while the NH between P2 and P3 hydrogen bonds to Tyr143’s carbonyl. In addition, the benzene ring is stacked with the amide backbone of Tyr143-Gly144 [280]. Among the series of modified compounds, FMK-9a (93) exhibited good activity with an IC₅₀ value of <100 nM, which is probably a result of the more efficient stacking of β-naphthyl with the amide backbone [280]. According to one study, compound 91 functions as an autophagy inhibitor, suppressing ischemic injury-induced autophagy to exert a protective effect on neurons [281]. However, compound 93 induces autophagy independently of ATG4B inhibition, and the mechanism needs to be further elucidated [282] (Fig. 20).

Fig. 20.

Chemical structures of ATG4B modulators.

5.2.3.2. ATG4B agonists

Through computer-aided drug design methods, including phylogenetic analysis, network construction, molecular modeling, molecular docking, and molecular dynamics (MD) simulations, Researchers discovered a stable interaction between flubendazole (94) and ATG4B. Further experiments revealed that compound 94 not only induced autophagic death and ROS production in MDA-MB-231 cells but also inhibited the proliferation of MDA-MB-231 cells in a dose-dependent manner (IC₅₀ = 0.75 µM) [283].

Asperphenamate (AS, 95) suppresses the proliferation of human tumor cells, although only at elevated concentrations. To improve its solubility and biological activity, the derivative N-Benzoyl-O-(N’-(1-benzyloxycarbonyl-4-piperidiylcarbonyl)-D-phenylalanyl)-D-phenylalaninol (BBP, 96) was obtained by replacing the phenyl group in the carbamate of AS with NCBZ-4-piperidine. Notably, compound 96 exhibited better solubility and stronger growth inhibitory activity on MCF-7 cells when compared to compound 95. Further studies indicate that this tumor suppressor effect was achieved by activating ATG4B to induce autophagic cell death [284] (Fig. 20).

5.2.4. Small-molecule compounds that target Beclin1 regulation

Beclin1, the mammalian homolog of ATG6/VPS30, is an important component of the VPS34/PI3K complex. Specifically, the Beclin1-VPS34-VPS15 core complex regulates and initiates autophagy by forming distinct complexes with ATG14L, UVRAG, and Rubcon. Beclin1 acts as a tumor suppressor that curbs mammalian cell growth and neoplastic expansion. Its binding to the anti-apoptotic protein Bcl-2 has emerged as a focal point for dissecting the interplay between autophagy and apoptosis [285]. Small molecules that tune Beclin1, therefore hold therapeutic promise for cancer and neurodegeneration.

A ubiquitin-specific peptidase inhibitor, Spautin-1 (97), is considered to be a specific and potent autophagy inhibitor. The mechanism by which compound 97 regulates Beclin1 is to inhibit the two deubiquitinases USP10 and USP13, resulting in the ubiquitination of Beclin1, which promotes the subsequent degradation of Beclin1 and the Vps34/PI3K complex through the proteasomal pathway to inhibit autophagy. Notably, USP10 also mediates the deubiquitination of p53, so targeting Beclin1 also provides a new approach for regulating the level of p53 [286]. In addition, compound 97 inhibited autophagy and promoted cell death under nutrient-deprived conditions, and importantly, cancer cells exhibit higher sensitivity to compound 97 [286]. In conclusion, compound 97 blocks autophagy and simultaneously engages both p53 and Beclin1, positioning it as a promising anticancer lead.

Xestospongin B (98) is an IP3R antagonist that induces autophagy by disrupting the molecular complex formed by IP3R, Beclin1, and Bcl-2 [287]. Gossypol (99), a natural BH3 mimetic that blocks Bcl-2, is in Phase II/III trials for prostate cancer. It triggers autophagy by disrupting Beclin1–Bcl-2/Bcl-xL complexes at the ER [288]. Similar to compound 99, compounds including ABT-737 (100) [289] and seocalcitol (EB 1089, 101) [290], which upregulate autophagy also by disrupting the interaction of Beclin1 and Bcl-2, may be useful in the treatment of cancers associated with mutations in the signaling pathway of apoptosis. Recent studies have reported that tamoxifen (TAM, 102), an antiestrogen drug frequently used in the treatment of endocrine and breast cancer, also upregulates Beclin1 to induce autophagy. On the other hand, autophagy is the primary reason for the resistance of breast cancer patients to TAM, and inhibition of autophagy can achieve resensitization of acquired drug-resistant breast cancer cells [291] (Fig. 21).

Fig. 21.

Chemical structures of compounds 97–102.

5.2.5. PIM-2 inhibitors

Previous studies have demonstrated that the Proviral Integration site for Moloney murine leukemia virus 2 (PIM-2) can regulate autophagy through mTORC1 [292]. SGI-1776 (103), an imidazopyridazine compound, is a potent ATP-competitive inhibitor of PIM-1, PIM-2, and PIM-3 kinases [293]. A recent study indicates that this pan-PIM inhibitor induces autophagy in tumor cells while exhibiting significant cytotoxicity. PIM-2 directly phosphorylates and regulates mTORC1 downstream effector molecules such as 4E-BP1, thereby influencing protein synthesis and co-regulating the autophagy flux [294]. However, compound 103 was considered an unviable clinical candidate due to its metabolism and toxicity issues. Another pan-PIM kinase inhibitor, AZD1208 (104), was developed by AstraZeneca [295]. Compound 104 may be cytotoxic in chronic lymphocytic leukemia cells (CLL) by inhibiting protein translation and inducing autophagy [296]. Since compound 104 shows excellent activity in inhibiting PIM, phase I clinical trials for the treatment of advanced solid tumors and malignant lymphomas have been completed (NCT01588548). In addition, HJ-PI01 (105) is a PIM-2 selective inhibitor obtained by molecular docking and selective modification of PAN-PIM inhibitor, which induced apoptosis and autophagy in MDA-MB-231 cells and inhibited tumor proliferation [297] (Fig. 22).

Fig. 22.

Chemical structures of compounds 103–111.

5.2.6. GSK inhibitors

GSK-3, a widely expressed serine/threonine kinase once linked only to glycogen synthesis, now emerges as a modulator of autophagy that acts on mTORC1, AKT, ULK1, and TFEB [298]. Specifically, GSK-3 inhibition induced protective autophagy in pancreatic cancer cells [299]. 9-ING-41 (106), a maleimide ATP-competitive inhibitor of GSK-3β, its mechanism involves releasing GSK-3β inhibition on key autophagy regulators (such as TFEB), thereby promoting lysosomal biogenesis and completion of the autophagy flux. When combined with the autophagy inhibitor CQ, it increases apoptosis rates in bladder cancer cells [300]. Currently, compound 106 has entered clinical trials for patients with different types of advanced cancer (NCT03678883). Other GSK-3 inhibitors including SB216763 (107) [301], TDZD-8 (108) [302] and Laduviglusib (CHIR99021, 109) [303] may similarly induce autophagy in different cells (Fig. 22).

5.2.7. LRRK2 inhibitors

The leucine-rich repeat kinase 2 (LRRK2) has a complicated association with autophagy, and the mechanism by which LRRK2 regulates autophagy and its possible impact on disease are still poorly understood. Studies have indicated that LRRK2 may induce autophagy in a mTORC1-independent way [304]. Membrane-associated LRRK2 inhibited autophagy by inactivating Beclin1, suggesting that inhibiting LRRK2 is a feasible scheme to induce autophagy [305]. Through a high-throughput kinase test and screen, LRRK2-IN-1 (110) was identified, which inhibits wild-type and G2019S mutant LRRK2 with IC₅₀ values of 13 nM and 6 nM, respectively [306]. In addition to lowering the expression of inflammatory cytokines, inhibition of LRRK2 by compound 110 also restores autophagy in microglia [307]. Another more selective LRRK2 inhibitor, GSK2578215A (111), induced protective autophagy and cleared damaged mitochondria in SH-SY5Y cells [308] (Fig. 22).

5.3. Modulators of lysosomal function

Lysosomes are acidified organelles rich in hydrolases that serve as the terminal degradation stations for all forms of autophagy—macro, micro, and chaperone-mediated. Beyond disposal, these vesicles regulate signaling and metabolic balance. Additionally, lysosomal dysfunction during autophagy is closely related to certain chronic infectious diseases, cardiovascular diseases, neurodegenerative diseases, lysosomal storage diseases, tumor growth, and aging. Therefore, modulators of lysosomal function have potential value in certain diseases.

Clomipramine (CMI, 112) is a widely used tricyclic antidepressant that potently blocks the reuptake of norepinephrine and serotonin in the central nervous system. A recent study reported that the administration of compound 112 significantly increased LC3-II and p62 in the hepatocytes of mice, which may be related to disrupted autophagosome formation and cargo degradation in primary isolated neurons. Subsequent incubation of compound 112-treated mice with a lysosomal inhibitor found that neither LC3-II nor p62 accumulation increased, suggesting that compound 112 can block autophagic flux [309]. The biologically active metabolite of CMI, desmethylclomipramine (DCMI, 113), may block autophagic flux by obstructing autophagosome-lysosome fusion and preventing cargo degradation in tumor cells, which increases the cytotoxicity of chemotherapeutic medicines such as doxorubicin [310].

B Grimaldi et al. demonstrated that the circadian NR reverses the orientation c-erbA gene, variant β (REV-ERBβ) and shields cancer cells when autophagy is impaired, so hitting both REV-ERBβ and autophagy could effectively kill tumor cells. Screening yielded ARN5187 (114), a lysosomotropic ligand that hits REV-ERBβ and autophagy at once. It stalls autophagy late by impairing lysosomes and simultaneously blocks REV-ERB-dependent control of circadian and metabolic genes. Furthermore, the EC₅₀ values of the cytotoxic responses of compound 114 and CQ were 23.5 ± 7.3 and 4100 µM, respectively, indicating that compound 114 is more cytotoxic than the clinically relevant lysosomal autophagy inhibitor CQ and also induces apoptosis more effectively [311]. Subsequent structure-activity optimization of compound 114 revealed that the introduction of a fluorine atom at the meta or para position of the benzene ring seemed to be advantageous for the anticancer activity in vitro. And then substituted N-methylpiperazine with the N-methylpiperidine group to obtain compound 115, which exhibited significantly enhanced REV-ERBβ inhibitory and cytotoxic activity as well as dual inhibition of REV-ERB and autophagy in SK-BR-3, HEP-G2, and LNCaP cells [312]. In conclusion, this multifunctional inhibitor of REV-ERB and autophagy has contributed to the discovery of promising novel cancer treatments.

Both CQ (116) and HCQ (117) are 4-aminoquinoline-based medicines that are primarily used for the prevention or treatment of malaria infection and autoimmune diseases. Both compounds are weakly basic and lipophilic, so they accumulate in acidic compartments and readily exceed 20 mM in lysosomes, resulting in lysosomal alkalization and inhibition of activity for degrading enzymes and autophagy [313]. However, recent research demonstrated that compound 116 leads to a severe disorder of the Golgi and lysosomal systems. Therefore, compound 116 blocks autophagy by preventing autophagosome–lysosome fusion while leaving lysosomal pH intact [314]. Preclinical studies have proven the advantages of using compounds 116 and 117 to effectively suppress autophagy in the treatment of cancer. When combined with chemotherapy drugs such as sunitinib, bevacizumab, and oxaliplatin, the antiproliferative effects and cytotoxicity of these chemotherapeutics were significantly increased, while the development of tumor resistance was suppressed [315]. In addition, the quinoline antimalarial drugs mefloquine (MFQ, 118) [316] and quinacrine (QN, 119) [317] have been shown to evoke ER stress and cytotoxicity in tumor cells by preventing autophagosome degradation.

To develop more potent inhibitors of autophagy, VATG-032 (120, EC = 5 µM, IC₅₀ = 27 µM) and VATG-027 (121, EC = 0.1 µM, IC₅₀ = 0.7 µM) were discovered through high-throughput screening of known antimalarial compounds, including lysosomal inhibitors, followed by chemical optimization [317]. Both compounds 120 and 121 are more potent autophagy inhibitors than compound 116 and similarly suppress autophagy flux by alkalizing the lysosome. Notably, while compound 121 inhibits autophagy more effectively than compound 120, it is also more cytotoxic [318]. Since the planar conformation of the polycyclic aromatic hydrocarbons system is required for DNA binding, the saturated C ring in the 1,2,3,4-tetrahydroacridine structure of compound 120 adopts a twist chair conformation, which may account for its decreased toxicity. Furthermore, the lack of significant change in cytotoxicity to ATG gene knockdown indicates that the cytotoxicity of compounds is independent of autophagy [318]. These results validate compounds 120 and 121 as viable adjuvants for cancer therapy.

The selective COX-2 inhibitor, celecoxib (122, IC₅₀ = 40 nM), is a non-steroidal anti-inflammatory drug (NSAID) regularly used for the treatment of osteoarthritis and rheumatoid arthritis. Recent research has discovered a connection between cell cycle arrest, apoptosis, and autophagy to the anticancer activity of celecoxib in solid tumors that may boost the effectiveness of chemotherapy and radiation. Specifically, compound 122 may affect lysosomal function by changing the pH of lysosomes and thereby inhibiting autophagy. Additionally, similar to other autophagy inhibitors, compound 122 enhanced the cytotoxicity of the chemotherapeutic drug imatinib against CML cells KBM5-T315I and promoted apoptosis [319]. However, contrary to solid tumors, the effect of compound 122 on autophagy in tumor cells such as CML and osteosarcoma was surprisingly the reverse. In addition to inhibiting tumor cell proliferation by inducing apoptosis and cell cycle arrest, compound 122 also induces cytoprotective autophagy, and the addition of other autophagy inhibitors can further boost its anticancer effectiveness [320]. The FDA has authorized compound 122 for the treatment of familial adenomatous polyposis based on evidence that it reduces the risk of colon and rectal cancer [321].

Lys01 (123), a dimeric chloroquine derivative obtained by rational drug design, was identified as a novel lysozyme denaturant. And Lys05 (124), the trihydrochloride of compound 123 with enhanced water solubility, may more effectively accumulate in lysosomes and deacidify lysosomes to inhibit autophagy and tumor growth [322]. In cancer cell lines, compound 124 inhibits autophagy 3 to 10-fold more strongly than compound 117. In melanoma and colon cancer xenografts, intermittent high-dose or chronic low-dose treatment with compound 124 raises lysosomal pH, halts autophagy, and triggers tumor cell death [322,323]. Since autophagy inhibitors may be incorporated into the combination regimen of one or more anticancer therapies, compound 124 and its derivatives with antitumor activity observed at low and intermittent doses are a class of promising candidate compounds.

The macrolide antibiotic bafilomycin A1 (125) affects both autophagosome-lysosome fusion and lysosomal acidification. Specifically, compound 125 acts as a lysosomal proton pump V-ATPase inhibitor to inhibit lysosomal lumen acidification and lysosomal enzyme activation. On the other hand, compound 125 inhibits the ER-calcium ATPase Ca-P60A/dSERCA, thereby affecting the autophagosome-lysosomal fusion [324]. Additionally, compound 125 has the potential to enhance the 5-FU-induced apoptosis of gastric cancer SGC-7901 cells by inhibiting autophagy [325]. Another study demonstrated that compound 125 inhibits the invasion of prostate cancer cells by suppressing autophagy, which exhibited a favorable prospect in cancer treatment [326]. BRD1240 (126) [327] and diphyllin (127) [328] are two other V-ATPase inhibitors that also inhibit autophagic flux to varying degrees.

In recent years, matrine (128), a natural compound extracted from the traditional Chinese medicine Sophora flavescens, has been reported to exhibit anti-tumor effects on a variety of cancers without noticeable side effects, but the potential molecular mechanism of antiproliferative function remains unknown. According to one study, compound 128 blocked autophagy by impairing the function of lysosomal proteolytic enzymes, which resulted in autophagy-mediated abnormal mitochondrial metabolism and inhibited the proliferation of pancreatic cancer cells [329]. Furthermore, compound 128 also inhibited breast cancer cell proliferation by inducing apoptosis and autophagy through the AKT/mTOR pathway [330]. Due to its promising efficacy, compound 128 is currently used in China as an adjunct to boost the immunity of breast cancer patients for improved 5-year survival rates [331].

Lucathone (129) is a thioxanthone compound that was originally synthesized as an anti-schistosomiasis drug. Further investigation into compound 129 revealed that it is a novel autophagy inhibitor that blocks autophagic degradation by increasing lysosomal membrane permeability [332]. In addition, compound 129 also inhibits DNA repair and induces apoptosis. These effects give compound 129 the potential to be used as a sensitizer in combination with chemotherapeutic or radiotherapeutic treatments (Fig. 23).

Fig. 23.

Chemical structures of compounds 112–129.

As a core degradative pathway essential for cellular homeostasis, autophagy is governed by a highly complex and context-dependent regulatory network [162]. This review systematically surveys over a hundred small-molecule modulators, encompassing not only precision-targeted agents that act directly on core autophagic machinery components such as ULK1, VPS34, and ATG4B, but also a substantial number of compounds that modulate upstream signaling hubs like mTOR, AMPK, p53, and MAPK, or affect organelle functions such as lysosomal activity. This intentionally broad inclusion criterion reflects a pivotal paradigm shift in the field of autophagy-directed drug discovery: effective intervention in autophagy has evolved from solely targeting core proteins toward the systemic modulation of multi-layered regulatory networks.

Molecules that directly engage core machinery components, such as compound 72, compound 84, and compound 89, offer the advantages of well-defined mechanisms and direct effects, serving as indispensable tools for dissecting autophagic molecular mechanisms. However, the initiation and progression of autophagy are intimately coupled to the cellular integration of cues regarding nutrients, energy, growth factors, and stress—information that is funneled through upstream signaling nodes such as PI3K/AKT/mTOR, AMPK, and p53. Therefore, compounds targeting these nodes, including the various inhibitors and agonists discussed herein, can modulate autophagic flux by intervening at more proximal signaling checkpoints. For instance, mTOR inhibitors induce autophagy by relieving their suppression of the ULK1 complex, whereas AMPK agonists initiate the process by directly activating ULK1 and concurrently inhibiting mTOR. Although these interventions might be considered indirect, they are nevertheless specific and functionally essential, as they act at critical control points within the autophagic regulatory circuitry. Furthermore, modulators of lysosomal function, including CQ, compound 125, and compound 124, do not target autophagosome formation but critically determine the terminal efficiency of autophagic flux by alkalizing the lysosome or inhibiting its enzymatic activity. These agents represent key tools for controlling the final step of autophagic degradation.

Thus, incorporating modulators acting at these distinct levels into a unified perspective does not conflate core mechanisms with secondary effects; rather, it aims to construct a comprehensive map of pharmacological autophagy regulation. This map delineates the multiple tiers at which autophagy can be therapeutically targeted and underscores the rationale of selecting the appropriate level of intervention based on the specific pathogenic dysfunction in a given disease context—whether it involves aberrant signal activation, impaired core protein function, or defective lysosomal degradation. Looking forward, a major challenge lies in employing systems biology approaches and dynamic monitoring technologies to precisely define the autophagic sub-networks modulated by different compounds in relevant models. Such insights will be crucial for designing intelligent combination therapies that are temporally and spatially controlled, or that coordinately disrupt multiple critical nodes. The toolbox of compounds assembled in this review provides a foundational resource for advancing toward these goals.

5.4. Challenges and limitations

Despite the identification of a large and growing repertoire of small-molecule autophagy modulators, the successful translation of these agents from bench to bedside has proven remarkably challenging. The limited clinical impact of autophagy-targeted monotherapies reflects not a single bottleneck, but rather the convergence of multiple, interdependent constraints spanning chemical, biological, and translational dimensions.

At the chemical level, many early autophagy modulators have been hindered by intrinsic drug-like property liabilities. Representative examples such as compound 1 illustrate the classical trade-off between in vitro potency and pharmacokinetic feasibility: although this pioneering dual PI3K/mTOR inhibitor exhibited strong cellular activity, its development was terminated at the preclinical stage due to poor aqueous solubility associated with its tricyclic core and rapid phenolic glucuronidation [166]. Even optimized derivatives, including compound 4, which achieved improved pharmacokinetic profiles and entered clinical trials, continued to face challenges arising from broad pathway inhibition. The extensive suppression of the PI3K/mTOR network raised concerns regarding metabolic and immune-related toxicities, underscoring the intrinsic selectivity dilemma of multi-target inhibitors [169]. Similarly, highly potent tool compounds such as Torin1, despite robust mTORC1/2 inhibition and reliable autophagy induction in vitro, failed to demonstrate therapeutic value in vivo owing to extremely low oral bioavailability and a half-life measured in minutes [199]. Collectively, these cases highlight how neglecting early drug-likeness parameters risks confining promising molecules to the status of “paper compounds”.

Beyond compound-centric limitations, deeper challenges arise from the networked architecture of the autophagy pathway itself and the extreme context dependency of its functional outcomes. Autophagy is not a linear signaling cascade but a highly interconnected adaptive system, making the effects of pharmacological intervention difficult to attribute and predict. CQ, widely employed as a prototypical autophagy inhibitor, exemplifies this complexity. Rather than acting solely through lysosomal alkalinization, CQ has been shown to broadly disrupt the Golgi–lysosome system, directly impair autophagosome–lysosome fusion, and exert autophagy-independent immunomodulatory effects [311]. This pronounced pleiotropy complicates the interpretation of clinical outcomes observed in combination trials, as the extent to which therapeutic sensitization can be attributed specifically to autophagy inhibition remains unclear, while also contributing to an unpredictable toxicity profile. Conversely, strategies aimed at precisely activating the core autophagy machinery face a distinct “double-edged sword” dilemma. Compound 82, the first structure-guided ULK1 agonist, demonstrated the ability to suppress tumor growth in TNBC cells by inducing autophagic cell death [263]. However, under nutrient-replete conditions or mild stress, ULK1 activation more commonly initiates cytoprotective autophagy. Whether ULK1-driven autophagy promotes survival or death is therefore dictated by the tumor’s microenvironmental pressures—such as hypoxia or nutrient deprivation—and its underlying genetic context. This profound context dependency makes it exceptionally difficult to define a safe and efficacious therapeutic window, substantially elevating translational risk.

Adding further complexity, tumor cells possess robust adaptive feedback mechanisms that actively undermine sustained autophagy modulation. Chronic inhibition of mTORC1 by compound 37 or its analogues relieves negative feedback on the upstream PI3K/AKT pathway, paradoxically leading to aberrant AKT activation and reinforcement of pro-survival signaling, thereby attenuating efficacy and promoting resistance [208]. This realization directly motivated the development of pan-mTOR inhibitors targeting both mTORC1 and mTORC2, such as compound 27. Nevertheless, cancer cells can further adapt by enhancing lysosomal biogenesis, activating alternative protein degradation pathways (e.g., the proteasome), or rewiring autophagic cargo selectivity, enabling escape from autophagy flux inhibition [196]. These compensatory responses fundamentally limit the durability of single-agent strategies targeting the autophagy pathway.

Finally, failures in clinical translation expose a critical disconnect between mechanistic rationale and clinical validation. Although compound 117 and related agents have been evaluated in dozens of clinical trials in combination with chemotherapy, targeted therapy, or immunotherapy, therapeutic outcomes have been highly variable [315]. A central limitation lies in the absence of biomarkers capable of dynamically and specifically monitoring autophagic flux in patients in vivo. Static measurements such as p62 accumulation or LC3-II levels in tumor biopsies are heavily influenced by sampling timing, tumor heterogeneity, and interpatient variability, rendering them unreliable for patient stratification, dose optimization, or efficacy prediction. Consequently, many clinical trials function as “black-box” interventions, unable to confirm whether autophagy modulation is occurring as intended within target tissues or whether it meaningfully contributes to therapeutic benefit.

In summary, the more than one hundred small-molecule compounds systematically reviewed herein, while collectively highlighting the rich therapeutic potential of targeting the autophagy pathway, also converge in exposing the deep-seated challenges that define this field. These challenges span from intrinsic drug-like property liabilities of the compounds themselves, to fundamental scientific issues related to target selectivity and pathway pleiotropy, and further to the biological complexity imposed by disease heterogeneity and adaptive cellular feedback mechanisms, ultimately culminating in the lack of precise tools for monitoring and evaluating autophagy in clinical translation. Importantly, this recognition does not imply a bleak outlook; rather, it provides a clear roadmap for future breakthroughs. Progress will require moving beyond the traditional “inhibitor/activator” paradigm toward the development of intervention modalities with fundamentally new mechanisms of action, such as allosteric modulators and targeted protein degradation technologies. In parallel, rational combination strategies informed by systems biology and tumor microenvironmental features must be designed, alongside concerted efforts to establish multidimensional biomarker platforms capable of dynamically mapping autophagy function in vivo in humans. Only by tightly integrating innovations in medicinal chemistry with a deep understanding of autophagy biology and a pragmatic engagement with translational challenges can autophagy-targeted strategies truly bridge the gap between “active compounds” and “clinically beneficial drugs”.

6. Discussion and perspective

Autophagy is a widely present and highly conserved recycling pathway in eukaryotic cells that can act either as a cytoprotective response or as a distinct form of PCD separate from apoptosis and necrosis. On the one hand, autophagy inhibits cell death and promotes cell survival under most stress conditions; on the other hand, autophagy may promote cell apoptosis or autophagy-dependent cell death under certain conditions. Autophagy-related cell death plays an important role in the occurrence, development, and treatment of diseases. Generally, autophagy is involved in material metabolism balance, intracellular homeostasis, intracellular waste removal, structural reconstruction, as well as cell growth, proliferation, division, and differentiation, while dysfunctional autophagy is associated with a variety of pathological and physiological processes such as aging, infection and immunity, tumors, neurodegenerative disease, diabetes, liver damage, and leukemia. Undeniably, autophagy plays a complicated and diversified function in various disease processes. Specifically, the dual role of autophagy in disease exhibits significant heterogeneity across different pathological contexts, with its functional shift from cytoprotection to cell death demonstrating clear disease-type dependence and stage specificity. During tumor progression, the function of autophagy undergoes a dynamic transition: in early tumorigenesis, autophagy exerts a tumor-suppressive effect by clearing damaged proteins and organelles, maintaining genomic stability, and inhibiting oxidative stress; whereas in established solid tumors, autophagy supports tumor cell survival under stressful microenvironments such as hypoxia and nutrient deprivation, and promotes tumor progression and therapy resistance through mechanisms including metabolic reprogramming, epithelial–mesenchymal transition, and pre-metastatic niche formation. In neurodegenerative diseases, autophagy primarily serves a protective role by clearing misfolded protein aggregates (e.g., β-amyloid, α-synuclein) to maintain neuronal homeostasis; however, when autophagic function is impaired due to aging, genetic defects, or pathological protein accumulation, the collapse of proteostasis leads to neuronal death. In acute pathological processes such as ischemia–reperfusion injury, moderate autophagy activation clears damaged mitochondria and maintains energy balance, thereby providing protection; however, excessive or sustained autophagic flux can directly trigger autophagic cell death and exacerbate tissue damage. This context-dependent functional transition suggests that the ultimate effect of autophagy in a given disease is determined by the metabolic and microenvironmental features at specific disease stages, the autophagic capacity limits of the cell type involved, and the crosstalk between autophagy and other signaling pathways such as apoptosis, inflammation, and metabolic stress. Therefore, systematically elucidating the dynamic network positioning of autophagy in distinct disease settings—rather than isolating its pro-survival or pro-death attributes—represents a theoretical prerequisite for achieving stage-specific autophagy modulation in disease intervention.

Since autophagy was first discovered approximately half a century ago, the emphasis of study has traditionally been on the signaling networks controlling autophagy. Autophagy is finely regulated at multiple levels to fulfill its related physiological function. The various signals, including a shortage of nutrients (amino acids or glucose, etc.), changes in pH, or osmotic pressure, are engaged in signaling pathways that regulate autophagy. In addition to about 40 autophagy-related genes, many other transcription factors, enzymes, proteins, and RNAs are also involved in autophagy regulation. It is of great importance to consider that the downstream of many signaling pathways acts on the central molecule mTOR that regulates autophagy. Under normal conditions, active mTOR negatively regulates autophagy by inhibiting the phosphorylation of the ULK1 complex. However, the activity of mTOR is inhibited when circumstances are unfavorable, such as nutrient deficiency or oxidative stress, which leads to the upregulation of autophagy. Apart from this classic pathway, autophagy can also be mediated by several mTOR-independent pathways such as AMPK, MAPK, and PI3K/AKT/mTOR. Specifically, these signaling pathways may either induce or inhibit autophagy depending on the situation, and in turn, autophagy can regulate proteins associated with these pathways. In conclusion, further investigation is required into the signaling pathway that regulates autophagy, and treating specific disorders by manipulating the autophagy pathway is a promising avenue of investigation.

Medicinal chemists now focus on discovering potent, selective autophagy modulators and clarifying their therapeutic value across human diseases. Fortunately, numerous compounds with different structures, targets, and mechanisms of action have been reported to exert biological effects on autophagy modulation, such as mTOR inhibitor rapamycin, dual PI3K/mTOR inhibitor PI-103, pan-PI3K inhibitor LY294002, pan-mTORC inhibitor PP242, and the ULK1 agonist LYN-1604. Notably, some compounds, including GDC-0941, voxtalisib, and gedatolisib, have shown promising results in clinical trials. However, further development of these compounds may be limited due to poor activity, target selectivity, drug resistance, or other adverse properties. And no small molecule that directly targets the machinery of autophagy has entered the clinical stage. Numerous autophagy modulators are initially identified based on their bioactivity toward other targets or phenotypes. Many current autophagy modulators also disturb apoptosis, necrosis, immunity, and inflammation pathways. Medicinal chemists, therefore, seek new, truly autophagy-specific targets and the ligands that fit them.

Great efforts have recently been made to identify novel selective modulators targeting autophagy. At present, some promising drug design strategies, such as high-throughput screening and structure-based drug design, have also emerged and been applied. Most of the autophagy-targeting compounds are ATG kinase inhibitors/activators. Therefore, targeting the allosteric site of the kinase offers the distinct advantage of enhancing specificity, reducing side effects and toxicity, while covalent inhibitors have superior activity and pharmacokinetic properties. However, there are still certain restrictions on the traditional drug discovery strategy. On the one hand, some proteins, such as the cancer-related Ras protein, lack suitable and druggable targets, making it very difficult to develop compounds that directly target these proteins. Merely turning a single protein on or off rarely corrects diseases rooted in abnormal protein aggregation. Targeted protein degradation (TPD) has therefore emerged as an attractive alternative, and the proteolysis targeting chimera (PROTAC) strategy exemplifies this advance. PROTAC molecules recruit the ubiquitin–proteasome apparatus to tag and eliminate specific pathogenic proteins, thereby extending the scope of drug discovery beyond traditional occupancy-based approaches. Other strategies targeting the autophagy-lysosomal system, including autophagosome-tethering compound (ATTEC), AUTOphagy-TArgeting Chimera (AUTOTAC), autophagy-targeting chimera (AUTAC), and lysosome-targeting chimera (LYTAC), have attracted significant attention from researchers. Compared with PROTAC, which usually degrades short-lived proteins, targeting the autophagy-lysosome system may degrade a wider range of substrates, including protein aggregates, DNA/RNA molecules, peroxisomes, ribosomes, damaged mitochondria and even microbial pathogens. In conclusion, TPD provides a promising drug development strategy based on medicinal chemistry.

Autophagy is a fundamental eukaryotic process that intersects with physiology and pathology; modulating it through inhibition or induction meaningfully influences disease outcomes. Nowadays, potent and selective autophagy modulators have been identified and shown promise in preclinical and clinical evaluations. We believe that with the continuous progress of scientific research, small-molecule compounds targeting autophagy present a fascinating direction of autophagy research and are a promising, beneficial therapeutic approach.

CRediT authorship contribution statement

Yongya Wu: Writing – review & editing, Writing – original draft. Jiaxiang Luo: Writing – original draft. Haolin Tang: Writing – review & editing, Data curation. Aoxue Wang: Writing – review & editing. Maolin Duan: Data curation. Jie Liu: Supervision, Investigation. Guan Wang: Supervision, Funding acquisition. Liang Ouyang: Supervision, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Yongya Wu (BRID: 03510.00.11079) is an assistant researcher at West China Hospital, Sichuan University. She received her Ph.D. in Chemical Biology from Sichuan University in 2024. Wu’s research focuses on building and optimizing small-molecule compound libraries and screening technologies to advance drug target identification and lead compound discovery. In addition, she is dedicated to exploring novel mechanisms of drug action and identifying new therapeutic targets of bioactive molecules. Her work aims to broaden the understanding of drug–target interactions and contribute to innovative drug discovery strategies for the treatment of human diseases.

Liang Ouyang (BRID: 09885.00.29023) is a principal investigator at the National Key Laboratory of Biotherapy, West China Hospital, Sichuan University, and serves as a doctoral supervisor. He is a recipient of the National Science Fund for Excellent Young Scientists and has been selected as a Tianfu Science and Technology Elite under the “Tianfu Ten-Thousand Talents Program” of Sichuan Province. Ouyang’s team focuses on medicinal chemistry and pharmacology. His long-term research interests include novel drug target identification, small-molecule drug development, and pharmacological mechanism studies in major human diseases such as cancer and neurodegenerative disorders. His research seeks to drive original innovation in therapeutic strategies and foster translational applications in precision medicine.