Autophagy of the ER: the secretome finds the lysosome

Lysosomal degradation of the endoplasmic reticulum (ER) and its components through the autophagy pathway has emerged as a major regulator of ER proteostasis. Commonly referred to as ER‐phagy and ER‐to‐lysosome‐associated degradation (ERLAD), how the ER is targeted to the lysosome has been recently clarified by a growing number of studies. Here, we summarize the discoveries of the molecular components required for lysosomal degradation of the ER and their proposed mechanisms of action. Additionally, we discuss how cells employ these machineries to create the different routes of ER‐lysosome‐associated degradation. Further, we review the role of ER‐phagy in viral infection pathways, as well as the implication of ER‐phagy in human disease. In sum, we provide a comprehensive overview of the current field of ER‐phagy.


Introduction
Lysosomal recycling of cellular contents such as proteins and lipids is essential for general cellular metabolism as well as survival during stressful conditions [1,2].To deliver cytosolic or organellar components across the lysosomal membrane and into its lumen, cells use the process of macro-autophagy.Highly conserved from yeast to humans, the macro-autophagy pathway is a major cellular recycling pathway, and its dysregulation has been linked to numerous pathologies including cancer, neurodegeneration, metabolic disorders, and aging [2,3].
Macro-autophagy has been studied extensively at the molecular level, with approximately 30 "core" autophagy family proteins and many other peripheral actors identified for involvement [4,5].A simplified view of the macro-autophagy pathway can be boiled down to a handful of distinct steps, which include the following: (a) formation of the omegasome on the endoplasmic reticulum (ER) membrane, (b) maturation of the omegasome into the "phagophore" membrane (also called the isolation membrane), (c) phagophore elongation and closure to form the double-membrane organelle called the autophagosome, which contains the cargo destined to be degraded, and (d) fusion of the autophagosome with the lysosome to deliver its contents to the lytic lysosomal lumen (Fig. 1A) [4].
Numerous core autophagy proteins are required to generate the omegasome, phagophore, and autophagosome [4].The Atg8 family of proteins is of high interest to autophagy researchers due to their essential functions within this pathway.Named for the single yeast gene Atg8, there are 6 Atg8 homologs in the mammalian genome.These include MAP1LC3A, MAP1LC3B, MAP1LC3C (commonly referred to collectively as LC3), and GABARAP, GABARAPL1, and GABAR-APL2 [6].Atg8s such as LC3 and GABARAP are recruited to the growing omegasome where the protein is conjugated directly onto the headgroup of the lipid phosphatidylethanolamine (PE) that is positioned within the autophagy membrane [7].Once membrane-bound, Atg8s serve to further recruit other core autophagy proteins for membrane expansion, autophagosome closure, and assist in autophagosomelysosome association/fusion [6,8].
Atg8 family proteins are also important in directing the phagophore membrane to a select subcellular destination.This form of autophagy is referred to as "selective autophagy" [6].The specificity of selective autophagy is controlled by proteins harboring subdomains that contain LC3 interaction regions (LIRs) or GABARAP interaction motifs (GIMs), which bind to LC3 or GABARAP and therefore recruit the growing autophagosome to a specific location within the cell [6].In this review, we will focus on the selective autophagy pathway known as ER-coupled autophagy (ER-phagy), which has emerged as a major regulator of ER protein homeostasis (proteostasis), quality control, and secretion.
The ER is a major site of protein biogenesis, folding, and modification.For example, within the ER lumen of a professional secretory cell such as a pancreatic beta cell, as many as 6000 molecules of the secretory protein proinsulin can be synthesized per second [9].Because of this rich source of protein, in addition to the lipids that comprise its membrane, the ER can be engulfed through ER-phagy during metabolic stress to provide necessary nutrients.Additionally, because of the extreme folding burden that-when compromised-can lead to the buildup of misfolded and toxic proteins, ER-phagy can be used to selectively degrade the ER to restore ER proteostasis.Not surprisingly, either compromised or hyperactive ER-phagy can lead to disease outcomes [10][11][12][13].
Further relevant to human health, the ER is often exploited by viruses trafficking into cells through the retrograde secretory pathway, where the ER is the final destination and is used as a hub for viral protein biogenesis and replication [14].Interestingly, ER-phagy can be either anti-viral to defend the cells against viral infection, or pro-viral to be exploited by viruses to promote infection [15,16].In this review, we will also summarize recent findings that have uncovered how each of these different ER-phagy outcomes are regulated.

ER-to-lysosome-associated degradation (ERLAD)
ER-to-lysosome-associated degradation is a general term referring to all degradative pathways that route ER components to the lysosome for degradation, regardless of the intermediate components that are required.[17,18].Hence macro-ER-phagy, micro-ERphagy, and ER-associated autophagy (ERAA) are all sub-categories of ERLAD, and their distinctions are discussed below (Fig. 1B-D).

Macro-ER-phagy
Macro-ER-phagy refers to any ER-to-lysosome degradation pathway that requires the double membranous autophagosome for trafficking (Fig. 1B).In this way, macro-ER-phagy exemplifies the classical view of autophagic degradation and requires any core autophagy component that is required for generation of autophagosomes.Both starvation-induced ER-phagy for recycling and selective autophagy can employ macro-ER-phagy.

Micro-ER-phagy
Micro-ER-phagy is an ER-to-lysosome pathway that does not require the autophagosome and therefore does not require many core autophagy factors (Fig. 1C).In this pathway, the ER is sectioned off and directly consumed by the lysosome through ESCRTmediated engulfment [19].
One possible reason for the divergence of macro-ER-phagy vs micro-ER-phagy could be the energy cost associated with each-micro-ER-phagy could be predicted to be more energy-efficient as it does not involve the creation of intermediate autophagy structures such as the phagophore or autophagosomal membranes.However, this would come at the cost of effectiveness, as micro-ER-phagy may be a less targeted and inducible process.

ER-associated autophagy
ER-associated autophagy is a chaperone-mediated process that co-opts the omegasome to route misfolded ER membrane proteins to the lysosome (Fig. 1D) [20][21][22].The omegasome uses ER membrane as a donor source for the growth of the phagophore, which will mature into the autophagosome and eventually fuse with the lysosome to degrade cellular cargos.During ERAA, misfolded membrane proteins can be incorporated directly into the omegasomal membrane itself, which positions these membrane proteins as cargo for degradation as the autophagy process continues.ERAA is unique in that it does not require any ER-phagy membrane receptor.

ER-phagy receptors and their associated adaptors
Lysosome-dependent degradation of the ER (and its contents) is controlled by critical ER membrane proteins, termed ER-phagy receptors.These receptorsacting in conjunction with their associated adaptorsfunction by recruiting the degradative cargo on the luminal side of the ER and binding to the autophagy machinery on the cytosolic side of the ER.In this manner, the receptors can deliver the cargo for lysosomal degradation.Although ER components have been found in the lysosomes since the earliest description of the autophagy process [23], how the ER is targeted to the lysosome during autophagy was not understood until the discovery of the first autophagy receptor FAM134B (and its yeast homolog Atg39) in 2015 [24].Since then, several other ER-phagy receptors (along with their adaptors) have been identified.Here we briefly describe the individual receptors.

FAM134B
As the first to be identified, Family with sequence similarity, Member b (FAM134B) is the most well-studied ER-phagy receptor.Structurally, it contains a reticulon homology domain (RHD) that promotes its localization to the curved edges of ER sheets, as well as LIRs harbored within the cytosolic C-terminal region (Fig. 2A) [24].As discussed below, FAM134B is regulated by post-translational modifications, promoting its oligomerization which enables its ER-phagy receptor function [12,25,26].FAM134B paralogs and FAM134A and FAM134C also function as ER-phagy receptors, and FAM134C is also subject to regulation through post-translational modification [27,28].

Sec62
Sec62 serves a unique function in returning the ER to its normal size after ER stress has resolved.When misfolded proteins accumulate in the ER, the organelle accommodates by increasing its membrane lipid content.Classically known as a component of the ER translocon complex, Sec62 becomes orphaned during ER stress (Fig. 2A).Under this condition, the orphaned Sec62 can link the ER to LC3 and the lysosome after ER stress has been resolved [29].This process-termed "recovER-phagy"-is also distinct in that it does not use the double membrane autophagosome to route the ER to the lysosome.Instead, pieces of ER that contain the orphaned Sec62 are engulfed directly by the lysosome through micro-autophagy [19].Comparable to the ER membrane, the outer nuclear membrane also swells during ER stress, and Sec62 serves a similar function in resolution of nuclear membrane size in a process termed "micro-ONMphagy" [30].

RTN3
Reticulon3 (RTN3) was discovered through bioinformatic scanning for the presence of LIR regions, six of which are present in its large N-terminal soluble region (Fig. 2A) [31].Similar to FAM134B, RTN3 contains an RHD.However, unlike FAM134B, RTN3 is localized primarily to ER tubules, as opposed to ER sheets.Consequently, RTN3 is required only for the turnover of ER tubules during non-selective macro-ER-phagy (see "Bulk flow recycling of the ER").

CCPG1
Cell cycle progression 1 (CCPG1) was identified as an ER-phagy receptor through proteomic screening for GABARAP-interacting proteins (Fig. 2A) [13].This receptor is required for clearance of the ER mass during ER stress, and studies in mice have demonstrated that loss of CCPG1 leads to impaired pancreatic ER proteostasis.Further supporting its role in the ER stress response, CCPG1 expression is controlled by the unfolded protein response.

ATL3
Atlastin3 (ATL3)-a transmembrane protein with intrinsic GTPase activity known for its role in ER tubule formation [32]-also functions as an ER-phagy receptor in degrading ER tubules [33].Like FAM134B and RTN3, ATL3 contains a double hairpin structure that mimics a partial RHD which promotes its localization to ER tubules (Fig. 2A).Here, ATL3 can interact with GABARAP through its N-terminal cytosolic domain that harbors the GIM.

Tex264
Tex264 was identified as an ER-phagy receptor through two separate proteomic screens [34,35].It has a long cytosolic region that contains an extended intrinsically disordered region (IDR) as well as an LIR (Fig. 2A).Intriguingly, because of its physical extension, the IDR is thought to be important for reaching across large ER-docked protein complexessuch as the ribosomes-in order to recruit LC3 to the ER.

Associated adaptors of ER-phagy receptors
In addition to ER membrane receptors that have been identified, numerous associated ER-phagy factors have been uncovered.

Trim13 and p62
Trim13 is an ER membrane protein that mediates basal and ER-stress-induced macro-ER-autophagy [36].Trim13-mediated ER-phagy is unique in that it does not directly bind to LC3 to recruit the phagophore membrane.Instead, Trim13 is ubiquitinated, and ubiquitinated Trim13 is bound by the cytosolic macro-autophagy adapter p62 (Fig. 2B).p62 is an LIR-containing protein that recruits the phagophore membrane through direct interaction with LC3.

VAPA/B and CALCOCO1
CALCOCO1 is a soluble protein that was identified to be enriched in autophagosomes through proteomic screening [37].Further proteomic screening revealed that CALCOCO1 binds to the ER membrane proteins VAPA and VAPB (Fig. 2B).CALCOCO1 also binds to Atg8 family proteins.In an analogous manner to Trim13 and p62, VAPA/B and CALCOCO1 form a complex that mediates ER-phagy, with a preference for degradation of ER tubules.

Lunapark and Atlastins
Lunapark and Atlastins (Atlastin1, 2, and 3) are ER morphogenic proteins required for formation of ER tubules [32,38].Lunapark and Atlastins are proposed to use their ER shaping capability to control ER-phagy [39][40][41].Lunapark, in conjunction with RTNs, can limit the size of ER tubules to restrict the size of misfolded ER protein condensates, maintaining their ability to be engulfed by autophagosomes [41].Atlastins are proposed to act downstream of FAM134B to section off pieces of ER degradation, in addition to Atlastin3's role as an ER-phagy receptor (Fig. 2B).

SEC24C
SEC24C is a component of the COPII vesicular coat complex during ER-to-Golgi trafficking.SEC24C is required for RTN3 and FAM134B-mediated ERphagy [41,42].Just as SEC24C/COPII marks the ER exit site of anterograde secretory trafficking, SEC24C also marks the ER-phagy exit site for substrates of RTN3 and FAM134B and promotes cargo incorporation into these structures (Fig. 2B).

DDRGK1 and Ufm1
CRISPRi screening revealed the requirement for the ufmylation pathway in bulk flow macro-ER-phagy.
Hits from this screen included the ER membrane protein DDRGK1, which functions as the Ufm1 E3-like enzyme with its cytosolic partner UFL1 to ligate Ufm1 to its substrates [43].
Further studies have indicated that DDRGK1 has a direct role in formation of an ER-phagy receptor complex.During ER stress, a cytosolic Atg-8-binding protein called C53 is recruited to the ER membrane to complex with DDRGK1 and UFL1 [44].C53 is constitutively ufmylated, but during ER stress, the Ufm1 from C53 is transferred from C53 to stalled ribosomes by DDRGK1/ UFL1, thereby activating the complex to allow for C53mediated phagophore recruitment to the ER (Fig. 2B).Interestingly, DDRGK1-UFL1-C53-mediated ER-phagy only functions during ER stress and is not induced by starvation.Recent studies have also started to clarify the role for ufmylation during starvation-induced macro-ERphagy.To facilitate macro-ER-phagy, the ufmylated ER membrane protein NADH-cytochrome b5 reductase 3 (CYB5R3) binds to C53 to recruit Atg8-bound phagophore membrane to the ER [45].

Different tasks of ER-phagy: bulk flow recycling vs select cargo degradation
Lysosomal degradation of ER components can be either non-selective for mass ER turnover, or selective for specific ER subdomains containing distinct misfolded cargos.In non-selective ER-phagy, the ER is engulfed and degraded in the lysosome so that its contents can be recycled for reuse, often needed during times of stress.This feature of ER-phagy is a facet of general macroautophagic targeting of cellular components.
Selective ER-phagy, however, allows for specific sections of ER that contain misfolded, aggregated/condensated, or otherwise unwanted proteins to be sequestered and specifically targeted to lysosomes irrespective of cellular nutrition status.In this section, we will discuss these overlapping pathways that use many of the same ER-phagy factors, but for different purposes.

Bulk flow recycling of the ER
Initial studies of ER-phagy focused on "non-selective" degradation of the ER that occurs when general macro-autophagy is initiated [24,47].Accordingly, ERphagy was first observed during conditions such as nutrient deprivation or pharmacological mTOR inactivation.During such conditions of stress, ER turnover occurs en masse.This is in line with the classical objective of autophagy, which routes cellular components to the lysosome for the recycling of macromolecules (Fig. 3, top panel).
Given that the ER can constitute as much as 30% of total cellular protein content and 60% of cellular lipid content, the ER represents a rich source of material for repurposing [48].Such recycling potential of the ER may explain the continually growing number of known ER-phagy receptors that have been discovered.Accordingly, individual loss-of-function studies have shown that each ER-phagy receptor is required for the proper delivery of ER contents to the lysosome during macro-autophagy induction.This would allow for more efficient capture of the expansive ER membrane, leading to an increase in the recycling of ER materials.
The earliest studies of ER-phagy, which characterized FAM134B in mammalian cells and the yeast homolog Atg39, described ER-phagy as a process by which these receptors control the routing of ER contents to the lysosome during nutrient deprivation [24,47].For example, the ER sheet marker CLIMP63, and the general ER tubule marker RTN4 were used as general ER markers and are normally degraded through the lysosome when cells are cultured in minimal media, but this turnover is blocked in the absence of FAM134B.Additionally, FAM134B-depleted cells display excessively expanded ER, indicating that overall ER turnover is impaired [24].Similar data have been shown for cells depleted of RTN3, ATL3, Tex264, and CCPG1 [13,31,[33][34][35].Sec62 is unique in that it is the only ER-phagy receptor that does not participate in general non-selective ER-phagy.Instead, Sec62 is required for restoration of the ER and homeostasis after ER stress, when it functions to route excess ER mass to the lysosome after ER stressinduced ER volume expansion, thereby restoring the ER to its normal size [29].Additionally, Sec62 is required for resolution of nuclear envelope size following ER stress-induced outer nuclear membrane swelling [30].

Selective cargo degradation
In contrast to stress-induced ER-phagy that is intended for recycling, recent studies have also described how ER-phagy pathways are specifically dedicated to the removal of misfolded ER proteins [17,46,49].In this way, cells use ER-phagy more selectively as a means for protein quality control (Fig. 3, bottom panel).How misfolded proteins can be selected for degradation has become a point of emphasis, as multiple cargo receptors have been described which recruit problematic misfolded ER proteins to specific ER-phagy receptors.
FAM134B is responsible for the selective degradation of numerous misfolded proteins, including procollagen and procollagen mutants [17], mutant NPC-I1061T [49], and mutant alpha-1-antitrypsin.In cells expressing the mutant human a-1-antitrypsin variant Z (ATZ), FAM134B acts to constitutively target ATZ polymers to the lysosome to maintain proper ER function [18].Topologically, FAM134B is restricted to the outer leaflet of the ER membrane, precluding its ability to directly recruit a luminal cargo such as misfolded procollagen (Fig. 2A).To achieve selective recruitment of ER luminal proteins, FAM134B physically complexes with the ER membrane protein CNX, which is critical for its role in glycoprotein folding [50].CNX binds directly to the misfolded procollagens, thereby bridging FAM134B to its substrate [17].
RTN3 has also been found to have specific misfolded ER clients that it recruits for degradation.Thus far, these clients are all mutant prohormones, including mutant proinsulins, mutant pro-opiomelanocortin (POMC), and pro-arginine vasopressin (pro-AVP) [51].Similar to FAM134B, RTN3 is also topologically restricted to the outer ER leaflet and hence unable to bind to its substrates directly (Fig. 2A).To achieve selective recruitment of mutant proinsulins and POMC, RTN3 uses the ER membrane protein PGRMC1 as a dedicated cargo receptor.PGRMC1 is a single-pass transmembrane protein that binds directly to misfolded prohormones in the ER lumen and binds to RTN3 through transmembrane domain interactions [46].
CCPG1 is the only known ER-phagy receptor that can directly bind to its ER luminal cargo, as well as Atg8 family proteins [52].This is possible because this receptor harbors a large ER luminal region that contains multiple conserved cargo binding regions (Fig. 2A).Cargos for CCPG1 include ectopically expressed islet amyloid polypeptide and endogenous prolyl 3-hydroxylase family member 4 (P3H4) [52].

Mechanisms of regulation of ER-phagy receptors
ER turnover benefits starving cells where the ability to recycle cellular macromolecules is essential.For instance, in vivo studies have shown that loss of ER-phagy function can lead to the buildup of detergent-insoluble proteins in the mouse pancreas that undermines cellular-and consequently-tissue integrity.However, ER turnover must be restricted only to times of necessity in order to ensure that proper ER functions (e.g.secretion) is not compromised.Accordingly, cells have developed multiple regulatory strategies to ensure that ER recycling is tightly controlled.Recent findings revealed that one key strategy to control the function of ER-phagy receptors is through post-translational modifications of the receptor itself, including phosphorylation and ubiquitination [12,25,26,28,53,54].

Phosphorylation
FAM134B binds directly to LC3 to facilitate ER incorporation into autophagosomes and is itself degraded within this pathway, but how FAM134B function is regulated has only recently been uncovered.Within the small cytosolic region connecting the two membrane-inserted hairpin segments, FAM134B contains a consensus phosphorylation site for the calcium/ calmodulin-dependent protein kinase (CAMK) which is essential for proper FAM134B-mediated ER-phagy [12].Phosphorylation by CAMK provides a level of ER-mediated signaling to induce ER-phagy given that calcium efflux from the ER lumen into the cytosol during times of stress actives CAMK proximal to FAM134B-this locally promotes the bulk degradation of ER contents.
FAM134C is a paralog of FAM134B that is also involved in starvation-induced ER-phagy [27].Unlike FAM134B, which is activated by phosphorylation, FAM134C is kept inactive under fed conditions through phosphorylation by casein kinase 2 (CK2).CK2 phosphorylation of FAM134C blocks the interaction between FAM134C and LC3.During starvationinduced macro-ER-phagy, phosphorylation is prevented in an mTOR-dependent manner, which allows FAM134C to bind LC3 and facilitate macro-ER-phagy.
Like FAM134B, Tex264 is also activated by phosphorylation [53].Tex264 phosphorylation by CK2 increases the affinity between Tex264 and Atg8 family proteins, and phosphorylation is necessary for Tex264 to localize to autophagosomes [53].
Similar to the FAM134 family proteins and Tex264, CCPG1 activity is controlled by phosphorylation [54].Unlike phosphorylation events on FAM134s and Tex264, CCPG1 phosphorylation promotes its association with the core autophagy protein FIP200, which is part of the upstream autophagy ULK kinase complex that is required for the early autophagy initiation and autophagosome biogenesis [13,55].This interaction lends to an intriguing model in which CCPG1 promotes autophagosome formation directly at the site of ER-phagy exit, which could provide an efficient mechanism for packaging of CCPG1 substrates into autophagosomes.

Ubiquitination
In addition to phosphorylation, ubiquitination of FAM134B has also been shown to be essential for its function [25,26].Structural modeling experiments showed that ubiquitination of FAM134B, mediated by the E3 ubiquitin ligase AMFR, drives oligomerization, which is essential for this receptor to function.

ER-phagy and viral infections
Because ER-phagy recognizes and degrades potentially toxic proteins, it is an ideal cellular defense mechanism against viral infection.Many viruses rely heavily on the ER during their replication cycle: some use the ER for simple protein translation, while others remodel the ER membrane into full-fledged genome replication factories [56,57].Unsurprisingly, many viral infections trigger an ER-phagic response, which can degrade viral components and limit replication [56].
A few viruses, however, appear to use ER-phagy receptors to promote infection.In some cases, these viruses co-opt the ER-phagy receptors for their morphogenic functions; in others, the virus may have evolved to hijack autophagy itself as part of productive infection [58,59].To illustrate the complex interplay between ERphagy and viral infection, we focus here on advances in flavivirus research, as well as a few studies of filovirus, picornavirus, and most recently, coronavirus.

Flaviviruses
A family of enveloped, positive-sense RNA viruses, Flaviviridae encompasses many notorious arthropod-borne viruses that cause widespread human disease, such as Dengue virus (DENV), Zika virus (ZIKV), West Nile virus (WNV), yellow fever virus (YFV), and Japanese encephalitis virus (JEV) [14].The basic infection cycle of flaviviruses relies heavily on the ER.Following receptor-mediated endocytosis, the viral genetic material escapes into the cytosol, then is uncoated and targeted to the ER for translation [60].All flaviviruses induce significant remodeling of ER-derived membranes to form organelle-like structures where the viral genome is replicated, though the precise morphology of these structures varies between viruses [61].Following replication, the amplified viral genome is transferred to assembly sites thought to be at the ER membrane [62].Progeny nucleocapsids bud into the ER lumen, traffics to the Golgi apparatus for maturation, and then finally secreted for the next round of infection [14,62].
Since the flavivirus replication cycle requires ER membrane remodeling, it is no surprise that ER-phagy may act as an antiviral defense mechanism for the host cell.For example, knockdown of the ER-phagy receptor FAM134B was demonstrated to significantly increase viral replication for both DENV and ZIKV [16].This suggests that FAM134B normally serves to restrict flavivirus infection.Furthermore, the same group showed that the cytoplasmic protein BPIFB3 regulates FAM134B-mediated ER-phagy, such that depletion of BPIFB3 enhanced ER-phagy and blocked DENV and ZIKV infection [63].These findings further support the role of ER-phagy as a cellular defense against flavivirus infection.
In response to cells deploying ER-phagy to slow viral replication, flaviviruses have evolved countermeasures to block, evade, or subvert ER-phagy.One key example is the viral response to the FAM134B-mediated autophagy described above: several flaviviruses generate proteases that cleave and inactivate FAM134B.Expression of the nonstructural viral protein NS2B3 from DENV, ZIKV, and WNV was shown to cleave FAM134B (Fig. 4A).Not surprisingly, expression of DENV NS2B3 significantly decreased the amount of viral protein sequestered in autophagosomes [16].Through the activity of these proteases, flaviviruses can effectively limit ER-phagy and avoid being targeted for turnover.
While ER-phagy appears to serve as a defense mechanism against flaviviruses, surprisingly, other ERphagy receptors may in fact promote flavivirus infection.Knockdown of RTN3, for instance, was observed to significantly inhibit DENV, ZIKV, and WNV replication, possibly by preventing effective remodeling of the ER membrane [64].In fact, RTN3 directly interacted with the non-structural protein NS4A of WNV, inducing ER remodeling [64,65].Similarly, the ATL3 ER-phagy receptor, along with ATL2, play key roles in DENV and ZIKV infection: ATL2 depletion distorted the size and location of viral replication organelles, while ATL3 depletion was postulated to block virion assembly [66].For all of these studies, it is important to note that flaviviruses are proposed to hijack the ERmorphogenic properties of RTNs and ATLs at various steps of the replication cycle.Further work is needed to clarify whether the degradative function associated with these ER-phagy receptors plays any role in infection.

Coronaviruses
The basic biology of coronavirus infection has captured global attention since the beginning of the COVID-19 pandemic.Like flaviviruses, coronaviruses are a family of enveloped, positive-sense RNA viruses that rely heavily on the ER during their replication cycle [67].In brief, virions undergo receptor-mediated endocytosis, fusion with the endomembrane, and uncoating in the cytoplasm that allows the viral RNA genome to be released into the cytosol [67].Viral non-structural proteins are translated from the viral RNA, resulting in formation of organelle-like double-membraned vesicles (DMVs) derived from ER membrane.These DMVs serve as replication factories for the viral genome [68].Next, the daughter genomes transit back to the ER, where they are translated and packaged into immature viral protein complexes; these complexes are finally transported through the ER-Golgi intermediate compartment (ERGIC) for maturation, assembly, and secretion [67,69].The parallels between the flavivirus and coronavirus infection cycles are clear: both require extensive remodeling of ER-derived membranes in order to generate replication organelles.
The importance of general macro-autophagy in severe acute respiratory coronavirus 2 (SARS-CoV-2) infection has been scrutinized early in the COVID-19 pandemic, when pilot studies indicated that autophagy inhibitors like chloroquine significantly suppressed viral replication [70,71].Since those pilot studies, further evidence has emerged that autophagy is beneficial for SARS-CoV-2 replication.Several recent papers suggest that the viral protein ORF3a induces incomplete autophagy, causing accumulation of autophagosomes which the virus hijacks for replication and egress [72][73][74][75][76].The specific role of ER-phagy, however, is more mysterious.One study showed that ORF3a expression was sufficient to induce FAM134B-mediated ER-phagy, in turn inducing ER stress and turnover (Fig. 4B) [77].The authors posit that the ER-phagy machinery may play a role in forming DMV replication structures [76].Intriguingly, this induction of ER-phagy is the exact opposite of the strategy taken by flaviviruses, which cleave FAM134B to suppress ER-phagy.Another study suggested that SARS-CoV-2 utilizes the morphogenic ER-phagy receptor RTN3, as well as RTN4, to reshape the ER membrane into DMVs [78].For these receptors, more work is needed to clarify the mechanism by which ER-phagy supports SARS-CoV-2 replication.

Other RNA viruses
The studies described above offer two options for the relationship between ER-phagy and RNA viruses: does ER-phagy suppress replication (as with DENV and ZIKV) or enhance it (as with SARS-CoV-2)?This fascinating question has been investigated for only a few other viruses, with much territory still unexplored.
One virus that may be negatively regulated by ERphagy is Ebola virus (EBV).A member of the Filoviridae family, EBV has caused many outbreaks of severe hemorrhagic fever that spread from primates to humans and often have high mortality [56,79].Similar to flaviviruses, EBV exhibited increased replication in FAM143b-depleted cells [80].The authors thus posit that FAM134-mediated ER-phagy limits EBV replication, possibly by triggering turnover while the viral protein is processed in the ER.The precise mechanism by which ER-phagy restricts EBV replication remains unknown.
Similarly, ER-phagy may suppress viral replication for the Picornaviridae member foot-and-mouth disease virus (FMDV), a highly contagious livestock virus with significant and widespread economic impact [56].A small, nonenveloped RNA virus, FMDV enters cells via macropinocytosis, uncoats in the endosome, then carries out its replication in the cytosol [81].The rapid production of viral proteins by the ER induces ER stress and autophagy [82].A recent study identified the Sec62 ER-phagy receptor as a key modulator of autophagic response during FMDV replication [83].Depletion of Sec62 led to a significant increase in FMDV protein levels and viral titer.This effect was blocked by autophagy inhibitors, demonstrating that Sec62 facilitates ER-phagic degradation of viral proteins.FMDV appears to have evolved its own defense against this degradation: the investigators observed that levels of Sec62 in FMDV-infected cells decrease as infection continues, suggesting that the virus downregulates expression of Sec62 to preserve its replication process [83].

ER-phagy and disease
Given the key role of ER-phagy in maintaining cellular proteostasis, it is no surprise that ER-phagy dysregulation has been linked to many diseases, from metabolic disease to neurological disease to cancers.Hence, elucidating the basic biology of ER-phagy is crucial for both understanding the pathogenesis of these disorders and discovering new therapeutic approaches.

Metabolic disorders
Several classic ER-phagy substrates are misfolded, mutant proteins associated with a range of metabolic diseases [51,84].Studying these molecules-including mutant proinsulin, procollagen, and ATZ-has illuminated our understanding not just of metabolic disease, but also of basic ER biology [85].
Several studies of diabetes mellitis (a complex metabolic disease) have implicated autophagic dysregulation in pancreatic b-cells [86,87].While impaired macro-autophagy in b-cells has been shown in type 1 and type 2 diabetes [88,89], the role of ER-phagy specifically has been most clearly defined for the diabetic syndrome mutant INS-gene-induced diabetes of youth (MIDY).MIDY patients express a misfolded mutant proinsulin that exerts a dominant negative effect, blocking export of functional wildtype (WT) proinsulin from b-cells [90].Akita proinsulin (responsible for MIDY in the Akita mouse), forms high-molecularweight, detergent-insoluble aggregates that can trap WT proinsulin in the ER lumen [51].To remove these aggregates, cells employ RTN3-dependent ER-phagy (Fig. 5A).By clearing the insoluble species, cells not only dispose of the Akita aggregates, but also partially restore their ability to secrete WT proinsulin [51].ERphagy thus has the mechanistic potential to ameliorate the severity of MIDY.
In addition to Akita, RTN3-dependent ER-phagy has been shown to be capable of clearing several other mutant prohormone aggregates.Both the C28F mutation of POMC (responsible for early-onset diabetes) and the G57S mutation of pro-AVP (causing familial neurohypophyseal diabetes insipidus) form detergentinsoluble aggregates under RTN3 depletion, indicating they are normally cleared by ER-phagy [51].The details of this clearance mechanism were further revealed in a later study, showing that the RTN3interacting ER transmembrane protein PGRMC1 binds to lower-molecular-weight mutant prohormone aggregates and delivers them for RTN3-mediated ERphagy (Fig. 5B) [46].
Beyond mutant prohormones, two other ER-phagy substrates capable of forming detergent-insoluble aggregates are misfolded ATZ and procollagen.ATZ is also prone to misfolding and aggregation in the ER lumen; accumulation of mutant ATZ in liver cells can lead to liver cirrhosis or hepatocellular carcinoma [ 91,92].Similar to Akita proinsulin, ATZ may be degraded via ERAD (in its monomeric form) or ERphagy (in its aggregated form) [91,93,94].In its aggregated form, ATZ ER-phagic degradation is dependent on the p62 adaptor and polyubiquitinated TRIM13 receptor (Fig. 5C) [36].Alternatively, ATZ may be delivered to lysosomes via micro-ER-phagy [18].Notably, this alternate pathway relies on the ER-phagy receptor FAM134B and its binding partner, CNX, to package ATZ in vesicles and deliver those vesicles to the endolysosome (Fig. 5C) [18].As a large and abundant protein, collagen requires precise folding: 20% of newly synthesized type I procollagen is misfolded and degraded [95].Misfolded procollagen aggregates in the ER lumen, where it has been shown to be cleared by FAM134B-mediated ERphagy [17,96].As with ATZ, misfolded procollagen uses CNX as a coreceptor with FAM134B in order to be targeted to the autophagosome (Fig. 5D) [17].Another study reported that procollagen also accumulates at ER exit sites (ERES) before being directly engulfed by lysosomes (i.e.micro-ER-phagy) (Fig. 5D) [97].Without efficient ER-phagic turnover via macroor micro-ER-phagy, toxic procollagen aggregates accumulate and ultimately lead to cell death.

Neurodegenerative diseases
Because many neurodegenerative diseases are characterized by the presence of pathogenic protein aggregates, the role of autophagy in these diseases is an area of active research investigation [98,99].Induction of macro-autophagy has been suggested as a therapeutic approach to protect against accumulation of pathogenic proteins [99].Furthermore, mutations in ER-phagy receptors have been linked to a range of neurodegenerative diseases, including Alzheimer's disease, Parkinson's disease, and hereditary spastic parapalegia [100].Research to uncover possible mechanisms linking ER-phagy with pathogenesis is ongoing.
One neurodegenerative disease strongly linked to mutations in ER-phagy receptors is HSAN II, a rare inherited disorder that causes peripheral neuropathy [101].Multiple case reports of HSAN II patients have identified loss-of-function mutations in FAM134B [10,102].These data were echoed by studies of aged FAM134B-null mice, which demonstrated significant degeneration of sensory axons [24].Further work is needed to uncover the mechanism by which ER-phagy supports axon maintenance.
Another metabolic and neurodegenerative disorder connected to ER-phagy is Niemann-Pick type C disease, an inherited progressive disease leading to ataxia, loss of vision and hearing, and death.Niemann-Pick type C patients express defective, misfolded mutants of the lipid trafficking protein NPC1, the most common of which is the I1061T mutant [103].A substantial fraction of I1061T is degraded by FAM134Bdependent ER-phagy: depletion of FAM134B or Beclin-1 significantly increased I1061T levels [49].Additionally, preliminary analysis of Niemann-Pick patient cerebellum samples showed increased levels of FAM134B, p62, and LC3-II [49], suggesting an alteration in ER-phagy in Niemann-Pick disease.

Cancers
Cancer has long been linked to increased ER stress, caused by both intrinsic and extrinsic factors.The uncontrolled proliferation of cancer cells drives increased protein translation and therefore more ER stress; this stress is exacerbated by a tumor microenvironment low in oxygen and nutrients and high in certain fatty acids and other metabolites [104].This strong link between cancer and ER stress begs the question: what role do the major ER protein quality pathways-including the unfolded protein response (UPR), ER-associated degradation (ERAD) and ERphagy-play in tumorigenesis and metastasis?Multiple studies have shown that UPR sensors are important in promoting cancer growth, particularly in angiogenesis (formation of blood vessels supplying the tumor) and metastasis [104,105].However, the role of ER-phagy in cancer remains more mysterious, with researchers disputing whether it plays a tumor-promoting or tumor-suppressing role.
Several studies have demonstrated that specific ERphagy receptors promote cancer cell survival and migration for certain cancers [104].This has been most clearly documented with Sec62, CCPG1, and FAM134B.Higher Sec62 expression has been observed in a wide range of cancer cells, including prostate cancer, thyroid cancer, gastric cancer, and breast cancer, among others [106][107][108][109][110]. In non-small cell lung cancer (NSCLC), metastatic tumors displayed significantly higher Sec62 protein levels compared to nonmetastatic tumors [111].Functional cell migration assays revealed that cells overexpressing Sec62 display greater migration than cells that do not, underscoring the possibility that Sec62 promotes cancer cell survival and metastasis [111].Sec62 overexpression also promotes cancer cell migration and invasion in gastric cancer, and researchers were able to block this increased metastatic ability by inhibiting autophagy [109].This suggests that Sec62-mediated autophagy may be a viable target for gastric and other cancers.
From its first identification as an ER-phagy receptor, CCPG1 has been linked to proteostasis of pancreatic cancer cells [13].Pancreata from CCPG1-deficient mice were shown to harbor detergent-insoluble masses of various ER enzymes and chaperones; these pancreata also exhibited elevated levels of ER stress markers and, in older mice, inflammatory infiltrates in the tissue itself [13].These findings led the authors to speculate that CCPG1 might serve to protect the pancreatic cells from ER stress and potentially eliminate cancerous cells.
Finally, FAM134B has long been observed to be overexpressed in esophageal squamous cell carcinoma (ESCC) [112], and ESCC tissue samples showed high rates of FAM134B mutations, particularly for metastatic ESCC samples [113].This suggests that FAM134B mutations may promote cancer cell survival in new tumor microenvironments, though further work is needed to determine mechanism for this.This hypothesis was further bolstered by a study of breast cancer tumors, in which researchers knocked down FAM134B and observed significantly decreased cell proliferation under hypoxic conditions [114].Furthermore, pharmacological inhibition of FAM134Bmediated ER-phagy significantly reduced tumor volume in mice with breast cancer cell xenografts [114].This work points to the possibility of targeting ERphagy as an anticancer therapeutic strategy.
While FAM134B overexpression and mutations seem to promote cancer growth and metastasis in some contexts, this effect is highly dependent on the specific cancer.In colorectal cancer, by contrast, FAM134B is frequently deleted or downregulated in tissue samples, with low FAM134B protein levels associated with more advanced disease [115,116].Moreover, depletion of FAM134B significantly increased colon cancer cell migration and proliferation capacity, suggesting that FAM134B normally acts as a tumor suppressor [116].It seems likely that this cancerspecific pattern will hold true for other ER-phagy receptors as well.Therefore, any attempt to pharmacologically target ER-phagy in cancer patients must be carefully tailored to the specific cancer, and much further research is needed to determine the viability of such a strategy.

Conclusion and future directions
In this review, we have summarized recent studies of how ER-phagy is controlled at the molecular level, and the different roles that ER-phagy can serve within cells.Further, we have highlighted the importance of this pathway under human pathological conditions such as viral infection, metabolic disorders, neurodegeneration, and cancer.
Recent studies have shown that ER-phagy is regulated both by ER membrane receptors as well as soluble cytosolic adapters that can be recruited to ER for its incorporation into autophagosomes during macro-ER-phagy.The ER can also be directly fed into lysosomes without the use of the autophagosome during micro-ER-phagy and ERAA.The mechanisms by which ER-phagy receptors and adaptors are regulated is only beginning to be uncovered.Thus far, FAM134B has been found to be controlled by both phosphorylation and ubiquitination.Future studies of how ufmylation regulates ER-phagy will be necessary, as there is a clear requirement for Ufm1 and the machinery required for this post-translational modification.
Current findings also detail a difference between non-selective ER-phagy for bulk recycling and selective ER-phagy for protein quality control.Interestingly, each known ER-phagy receptor is involved in nonselective ER-phagy, either during nutrient or proteotoxic stress, but only FAM134B, RTN3, and CCPG1 have been shown to have specific selective ER-phagy substrates.To achieve specificity, FAM134B and RTN3 use cargo receptors that bind directly to misfolded proteins, while CCPG1 can directly engage to its ER luminal clients to promote their lysosomal degradation.Whether other ER-phagy receptors are responsible for the degradation of specific cargos is a key remaining question.Additionally, more detailed study of why the litany of ER-phagy receptors serve redundant functions during ER recycling, or if they participate within complexes (or subcomplexes) will clarify the question of why so many different ERphagy receptors exist.
Amidst the many and complex interactions between ER-phagy and human disease, much territory remains unexplored.Beyond studies of filoviruses, limited work has been done to determine the mechanism by which specific ER-phagy receptors act to limit (or in some cases, promote) viral entry and replication.Research surrounding ER-phagy in neurodegeneration and cancer is still largely in its infancy: while mutations in various ER-phagy receptors have been observed in many patient tissue samples from a range of diseases and cancers, much work is needed to understand the role of such mutations in disease pathogenesis and progression.Finally, an overarching topic of interest is the viability of using drugs targeting ER-phagy (or autophagy more generally) in the treatment of any of these diseases.Given the centrality of autophagy in maintaining cellular homeostasis, such drugs are a challenging but potentially powerful therapeutic approach [117].As described previously, inhibition of ER-receptormediated autophagy has already shown some promise in limiting the growth of certain cancers [109,114].Studies of ER-phagy thus represent the opportunity to make major strides in combating disease.

Fig. 1 .
Fig. 1.Schematic of the macro-autophagy pathway.(A) The omegasome grows at the ER membrane to form the phagophore membrane, which encapsulates cargo and closes to form the double membrane autophagosome.The autophagosome fuses with the lysosome to facilitate degradation and recycling of cellular components.(B) Macro-ER-phagy involves the use of autophagosome to route ER components to the lysosome.(C) Micro-ER-phagy does not require the autophagosome; ER-derived vesicles are engulfed directly by the lysosome.(D) ERAA incorporates ER membrane proteins directly into the ER-derived omegasome.Created with BioRender.com.

Fig. 2 .
Fig. 2. Topologies of ER-phagy receptors and adaptors.(A) Topologies of the six known ER-phagy receptors on the ER membrane, with RHDs and LIRs indicated.(B) Topologies of select ER-phagy adaptors, with LIRs and ubiquitination indicated.Created with BioRender.com.

Fig. 3 .
Fig.3.Degradation of the ER for recycling vs. degradation of the ER for proteostasis.Top panel: bulk sections of ER are incorporated into autophagosomes and degraded to replenish and repurpose macromolecules during times of metabolic stress.Bottom panel: specific subdomains of the ER that contain misfolded or unwanted proteins are selectively degraded through ERLAD.Created with BioRender.com.