Stress-induced self-cannibalism: on the regulation of autophagy by endoplasmic reticulum stress

Abstract

Macroautophagy (autophagy) is a cellular catabolic process which can be described as a self-cannibalism. It serves as an essential protective response during conditions of endoplasmic reticulum (ER) stress through the bulk removal and degradation of unfolded proteins and damaged organelles; in particular, mitochondria (mitophagy) and ER (reticulophagy). Autophagy is genetically regulated and the autophagic machinery facilitates removal of damaged cell components and proteins; however, if the cell stress is acute or irreversible, cell death ensues. Despite these advances in the field, very little is known about how autophagy is initiated and how the autophagy machinery is transcriptionally regulated in response to ER stress. Some three dozen autophagy genes have been shown to be required for the correct assembly and function of the autophagic machinery; however; very little is known about how these genes are regulated by cellular stress. Here, we will review current knowledge regarding how ER stress and the unfolded protein response (UPR) induce autophagy, including description of the different autophagy-related genes which are regulated by the UPR.

Introduction

The endoplasmic reticulum (ER) is a very complex and elaborate cellular organelle. It is composed of a single continuous phospholipid membrane that is comprised of the outer nuclear envelope, flattened peripheral sheets with ribosomes (rough ER), and a complex network of smooth tubules (smooth ER) that extend throughout the cell. The ER has many different cellular functions which are accommodated by its heterogeneous structures. While detoxification of drugs, fatty acid and steroid biosynthesis, and Ca²⁺ storage occurs in the smooth ER, most of the folding and post-translational processing of membrane bound and secreted proteins takes place in the rough ER. It contains an array of chaperone systems such as glycosidases, Ca²⁺-dependent chaperones, and members of the protein disulfide isomerase (PDI) family. These chaperones are responsible for the correct folding of proteins under normal physiological conditions [1]. This process is highly sensitive and is dependent on ER luminal factors such as Ca²⁺ concentration, redox homeostasis, and oxygen supply [2]. The processing of nascent proteins in the ER lumen requires an array of chaperones and folding enzymes that depend on the ER’s rich oxidizing environment and Ca²⁺ pools to function optimally. Physiological or pathological conditions that disrupt this finely balanced, unique environment cripples the ER’s protein folding machinery and results in a condition referred to as ER stress.

The cellular responses to ER stress are multifaceted and include the activation of a set of signaling pathways termed the unfolded protein response (UPR), a catabolic process termed autophagy, and cell death [3]. These processes are not mutually exclusive, and there is significant cross-talk between these cellular stress responses.

The UPR’s primary aim is to sustain cell survival by attenuating protein synthesis and restoring cellular homeostasis via the activation of a cascade of transcription factors, which regulate expression of genes encoding for chaperones, components of the ER-associated degradation (ERAD) system, and components of the autophagy machinery [4]. ER stress-induced increases in autophagic flux also act to help the cell survive the adverse conditions. There is accumulating evidence to help delineate the mechanism by which ER stress/UPR can transcriptionally activate or regulate the components of the autophagy machinery. In this review, we will summarize what is known about this aspect of ER stress and the UPR, in order to highlight the importance of this process in re-establishing cellular homeostasis during ER stress.

Unfolded protein response signaling

Disturbances in the ER’s homeostatic environment disrupts the protein folding machinery and results in an accumulation of unfolded proteins in the ER lumen, thus activating the UPR.

The UPR is orchestrated by three ER transmembrane receptors—pancreatic ER kinase (PKR)-like ER kinase (PERK), activating transcription factor 6 (ATF6), and inositol-requiring enzyme 1 (IRE1) (Fig. 1). In resting cells, all three UPR receptors are kept inactive through their association with the ER chaperone, glucose-regulated protein 78 kDa (Grp78; also known as BiP or HSPA5). Upon ER stress, unfolded proteins accumulate in the ER lumen resulting in the dissociation of Grp78 from PERK, IRE1, and ATF6, subsequently activating the UPR [5, 6].

Fig. 1.

The accumulation of unfolded protein in the ER lumen results in the dissociation of Grp78 from the three UPR sensors PERK, ATF6, and IRE1. Following Grp78 dissociation, PERK dimerizes and autophosphorylates, activating its cytosolic kinase domain. PERK phosphorylates EIF2α inhibiting general protein synthesis and facilitating/permitting non-canonical translation of ATF4 mRNA. Active PERK also phosphorylates NRF2 resulting in its dissociation from KEAP1, allowing NRF2 to translocate to the nucleus. Activation of ATF6 leads to its translocation to the Golgi where it is processed by site 1 and site 2 proteases (S1P and S2P) into an active transcription factor which results in the transcription of XBP1 mRNA. Activation of IRE1 results from its dimerization and autophosphorylation in a manner similar to PERK. IRE1 contains an endoribonuclease domain which processes unspliced XBP1 mRNA. Spliced XBP1 (XBP1s) mRNA is translated into an active transcription factor. IRE1 also possesses a kinase domain that recruits TRAF2 and ASK1 leading to the activation of JNK

PERK-(EIF2AK3)

PERK-(EIF2AK3) is a type I ER transmembrane protein with serine/threonine kinase activity. Its N-terminus is in the ER lumen, involved in the regulation of its dimerization, and is kept inactive through interaction with Grp78, while the C-terminus is cytosolic and harbors its autophosphorylation sites and the kinase domain. Upon release of Grp78, in response to accumulated proteins in the ER lumen, PERK homodimerizes and subsequently trans-autophosphorylates for activation.

PERK mediates the phosphorylation of the α sub-unit of eukaryotic initiation factor 2 (eIF2α) resulting in translation attenuation of mRNAs with a 5′ cap. The proteins encoded by these mRNAs are generally destined to be involved in cell growth and proliferation [7], thus eIF2α phosphorylation reduces the protein load in the ER and attenuates cell growth and proliferation [7]. eIF2α phosphorylation results in non-canonical translation of ATF4 mRNA via an open reading frame in its 5′-untranslated region that is bypassed only when eIF2α is inactivated [8]. ATF4 mRNA encodes for a cAMP response element binding transcription factor which activates a number of genes which play roles in amino acid metabolism, redox balance, protein folding, autophagy, and apoptosis [5, 9, 10]. Although ATF4 is an essential player in the pro-survival response of the UPR, it also plays a key role in the pro-death response via the transcriptional upregulation of C/EBP-homologous protein (CHOP), which is also called growth arrest and DNA damage-inducible gene 153 (GADD153) [11]. CHOP is reported to downregulate Bcl-2 [12] and upregulate transcription of certain BH3-only proteins [13, 14]. This event favors Bax/Bak activation which leads to mitochondrial outer-membrane permeabilization (MOMP) and initiation of the intrinsic apoptotic cascade [15]. Furthermore, CHOP knockout mice show lower rates of apoptosis in response to ER stress [16]. Although CHOP is thought to be a major factor in determining cell fate in response to ER stress, it is clear that other factors are also involved (for review, see [17]).

Another PERK substrate is the transcription factor, nuclear factor erythroid 2-related factor 2 (NRF2), required for free radical scavenging, detoxication of xenobiotics, and maintenance of redox potential [17]. PERK phosphorylates NRF2 and causes its nuclear translocation upon dissociation from KEAP1 [18]. Known targets of NRF2 are closely associated with redox homeostasis, and include glutamate cysteine ligase (GCL), both catalytic and modulatory subunits, heme oxygenase-1, and glutathione S-transferase (GST) isoforms [19]. These genes contain AU-rich elements (AREs) in their promoter region which is recognized by NRF2, and are also believed to be activated by ATF4 in response to ER stress, suggesting that these two transcription factors can act in synergy [20].

ATF6

ATF6 is a type II transmembrane receptor and a member of the leucine zipper protein family, that is synthesized as an ER membrane-tethered precursor, with its C terminal domain located in the ER lumen and its N-terminal DNA-binding domain facing the cytosol [21]. There are two isoforms of ATF6, ATF6α and β. Upon ER stress, Grp78 dissociates from ATF6, unmasking its two Golgi localization signals, allowing ATF6 to interact with the protein trafficking complex COPII, which causes translocation of ATF6 to the Golgi for processing [22]. At the Golgi, the 90-kDa ATF6 protein is cleaved by site 1 protease (S1P) and site 2 protease (S2P) into its active 50-kDa fragment which translocates to the nucleus where it acts as a transcription factor [23, 24].

Activated ATF6 is responsible for the transcriptional upregulation of XBP1 mRNA, which subsequently undergoes processing by IRE1 (see below) to produce a spliced XBP1s mRNA which encodes an active transcription factor [21]. ATF6, together with XBP1, are capable of binding to the cis acting response elements, ER stress response element (ERSE), and UPR element (UPRE), activating the expression of ER-localized chaperones [25]. Although the expression of ATF6 alone is enough to fully activate transcription from ERSE, in contrast to the ability of XBP1s to fully activate the UPRE [26]. Moreover, activation of ATF6 has also been described to regulate an array of miRNAs to alleviate ER stress [27].

IRE1

IRE1 (ERN1) is a type I ER transmembrane protein containing a serine/threonine kinase domain and an endoribonuclease. There are two IRE1 isomers in humans, IRE1α and IRE1β. IRE1α is ubiquitously expressed, whereas IRE1β expression is restricted to the epithelial cells of the intestine and the lungs. Most of our understanding of IRE function is based on studies of IRE1α.

IRE1 is the most conserved branch of the UPR, and has been suggested to play a role in processes such as development, metabolism, immunity, inflammation, and neurodegeneration [28]. Upon activation, IRE1 is known to oligomerize through self-assembly of the cytosolic region, leading to RNase activation. It has been shown that IRE1 oligomers consist of more than four molecules and, upon attenuation of its signaling during unmitigated ER stress, IRE1 clusters dissociate, the kinase is dephosphorylated, and its endoribonuclease activity is decreased [29]. Auto-phosphorylation at serine724 and ADP binding are other events that can be observed upon initiation of the signaling cascade [30]. Activation of IRE1 is closely associated with pro-survival pathway, providing cells an opportunity to readjust to unfavorable conditions, that cause increases in the amount of unfolded proteins [31, 32]. IRE1 transmits the UPR signal through excision of a 26-base nucleotide intron from X-box-binding protein 1 (XBP1) mRNA, which is then ligated by an uncharacterized RNA ligase and translated to produce XBP1s [30]. XBP1s is widely recognized as an important pro-survival gene in the UPR artillery. XBP1s transcriptional activity leads to the translation of stable transcription factors involved in the activation of ER regulatory proteins [33]. XBP1s has been shown to be crucially important for the activation of unfolded protein response element (UPRE), controlling the expression of the ER-associated degradation system, thus helping to degrade unfolded proteins in the ER [26]. Interestingly, acetylation of XBP1 by p300 and deacetylation by sirtuin 1 (SIRT1) have recently been shown to provide posttranslational modifications that can enhance or inhibit XBP1 transcriptional activity [34].

In addition to its ribonuclease activity, the cytoplasmic part of IRE1 is known to bind tumor necrosis factor-α (TNF-α) receptor-associated factor 2 (TRAF2), resulting in activation of c-JUN NH₂-terminal kinases (JNK) [35]. This is one of the mechanisms necessary for the activation of nuclear factor-κB (NF-κB) upon ER stress [36].

ER stress and cell death

More recently, it has been shown that the UPR also leads to the transcriptional upregulation of a number of autophagy-related genes essential for both the induction and construction of the autophagy machinery during ER stress [37, 38]. Autophagy upregulation during ER stress is an essential pro-survival response and is required for the removal of unfolded proteins, protein aggregates, and damaged organelles. These events are activated in order to relieve the stress and to reinstate homeostasis of the ER [4, 39]. However, prolonged or unresolved ER stress results in the activation of the apoptotic program [40].

The mechanism by which ER stress induces apoptosis is not fully delineated, but what is clear is that the intrinsic apoptosis pathway leading to mitochondrial damage and mitochondrial outer membrane permeabilization (MOMP), release of the mitochondrial factor, and activation of apoptosome and caspase-9 is central to the process [41]. This pathway is generally thought to be regulated by the balance between anti-apoptotic and pro-apoptotic Bcl-2 family proteins. If this balance favors the pro-apoptotic Bcl-2 family proteins, mitochondrial permeabilization will occur and lead to the activation of the intrinsic pathway. It has been well established that caspase-9 is the apical caspase required to execute apoptosis in response to ER stress; however, studies have also suggested that caspase-2 may play a role in inducing MOMP in certain cell models upstream of caspase-9 activation [41]. Because of the complexity of the UPR and the array of conditions that can result in ER stress, it is likely that some other cell- or tissue-specific cell death subroutines may also exist.

Autophagy

Macroautophagy (hereafter referred to as autophagy) is an evolutionarily-conserved, lysosomal-mediated system for bulk degradation of proteins, organelles, and cellular components. Autophagy was first coined in 1966 by Christian de Deuve, who identified double-membraned structures during his studies of mammalian cells using electron microscopy [42]. However, the molecular machinery of autophagy was extensively characterized for the first time in yeast by Yoshinori Ohsumi [43], and was later found to be evolutionarily conserved following the identification of the mammalian orthologues of yeast autophagy genes [44].

Autophagy is characterized by the induction of a small isolation membrane which elongates into a vacuole with a double membrane, capable of engulfing large amounts of cytosolic components such as unfolded protein aggregates, damaged organelles, and invading pathogens such as bacteria [45]. Autophagy is ongoing at basal levels in eukaryotic cells allowing the cell to function optimally by removing unwanted substrates which may otherwise lead to cellular toxicity [46–48]. Eukaryotic cells are continuously exposed to environmental changes which inflict minor stresses on the cell, disrupting its homeostatic environment. These constant fluctuations in the cell’s environment can result in the accumulation of misfolded protein aggregates, reactive oxygen species (ROS), and damaged organelles.

While basal autophagy activity is important for general maintenance of cellular homeostasis, defective autophagy may lead to DNA transformation and subsequent tumorigenesis. The exact mechanism by which defective autophagy results in tumorigenesis is still unclear; however, mounting evidence suggest a tumor-suppressor role for autophagy as a tumor-suppressing mechanism required by cross-talking with such events as cell death, senescence, and metabolism, all of which are overcome for tumorigenesis to progress (for review, see [49]). Autophagy is also very important during various, more acute, cellular stress responses [44]. Autophagy is dramatically increased in response to cellular stress, such as starvation, hypoxia, heat shock, microbial infection, and ER stress [47]. During cellular stress, large quantities of proteins are damaged, resulting in their unfolding/misfolding and aggregation that accumulate, and if they are not rapidly dealt with, they can ultimately induce apoptosis. Autophagy’s robust and efficient removal of these toxic factors can help relieve the cell of the stress and reinstate homeostasis [48].

The autophagic process requires the induction of a double membrane which is subsequently elongated by two specialized ubiquitin-like conjugation systems. The expanding double membrane is capable of engulfing large amounts of cytoplasmic components such as unfolded proteins, protein aggregates, and organelles. This elongated double membrane encloses to form a cytosolic dense, double-membraned vacuole termed an autophagosome. The mature autophagosome binds to a lysosome forming an autolysosome, where the autophagosome’s contents are released into the lysosomal lumen and degraded by resident cathepsins [46, 47]. The autophagic pathway is a very complex process, involving over 34 known proteins to assemble the machinery [50]. Here, we will discuss the different stages of the autophagy pathway (autophagy induction, vesicle nucleation, origin of the phagophore, autophagosome elongation, and maturation of the autophagosome), and bring to light the complexity of this unique process.

Autophagy induction

ATG1/ULK induction complex

The induction of autophagy requires the activation of the ATG1/ULK induction complex, a complex which consists of four known proteins, ULK1/2, mATG13, FIP200, and ATG101 [50]. This complex is essential for the induction of a small double membrane known as a phagophore or an isolation membrane. The phagophore eventually matures into a double-membraned vacuole termed an autophagosome via an elongation step involving two conjugation systems [51–53]. The induction complex is regulated by two kinases, mammalian target of rapamycin (mTOR) complex (mTORC) 1 and adenosine monophosphate-activated protein kinase (AMPK), via a series of phosphorylation events [54, 55]. It has long been established that mTOR is a key kinase in the regulation of autophagy [51]. It exists in two different complex forms, mTORC1 and mTORC2. mTORC1 is involved in autophagy regulation, and the complex is made up of mTOR, regulatory associated protein of mTOR (Raptor), mammalian LST8/G-protein β-subunit-like protein (mLST8/GβL), and the recently identified partners, PRAS40 and DEPTOR [56]. mTORC1 is incorporated into the ATG1/ULK induction complex and phosphorylates mATG13 and ULK1/2, maintaining the complex in an inactive state during normal resting conditions [57]. The phosphorylation of ULK1 by mTORC1 on serine757 has been shown to destabilize AMPK binding. AMPK is the main sensor of intracellular energy under conditions of starvation or environmental stress [55]. AMPK has been recently shown to play a crucial role in the positive regulation of the induction complex. Six AMPK phosphorylation sites have been identified on ULK1 (S467, S555, T574, S637, S777, S317) which all result in the activation of ULK1. AMPK can also negatively regulate mTORC1 via the tuberous sclerosis complex (TSC) to relieve mTORC1 inhibitory effects on ULK1 [54, 55]. Taken together, it is believed that ULK1 activation occurs in a stepwise series of phosphorylation events. First, mTORC1 is inactivated, resulting in the dephosphorylation at serine757 which facilitates AMPK binding. AMPK then activates ULK1 via a series of phosphorylation events. To add further complexity to this process, active ULK1 is capable of relaying feedback messages to both mTORC1 and AMPK. It phosphorylates mTORC1 resulting in its inactivation and thus amplifying the positive regulation of ULK1 [58]. In contrast, ULK1 has also been shown to phosphorylate AMPK’s three subunits resulting in its inactivation, and thus resulting in a negative feedback loop to ULK1 [59]. It is clear that there is great complexity in the regulation of the induction complex, and different stress responses may result in different phosphorylation events to activate ULK1 (for review, see [60]).

Vesicle nucleation

PI3K complex

The induction of the isolation membrane via the ATG1/ULK induction complex requires the activation of the PI3K complex (also known as the beclin1 complex) for vesicle nucleation, expansion, and curvature of the membrane. Mammalian cells have two forms of the PI3K complex, PI3K complexes I and II. The PI3K complex I consists of the class III PI 3-kinase Vps34, p150, Beclin 1, and ATG14L. ATG14L has been shown to increase stability of Beclin 1 and Vps34, and functions as the mediator which recruits the PI3K complex I to the isolation membrane. PI3K complex II consists of Vps34, p150, Beclin 1, and UV radiation resistance-associated genes (UVRAG); ATG14L does not associate with this complex. UVRAG interacts with BIF1 and localizes to the isolation membrane. BIF1 has an N-BAR domain which has been shown to bind membranes and cause them to undergo curvature. PI3K complex I regulates nucleation and PI3K complex II is involved in the expansion and curvature of the membrane [61].

Inhibitors of PI3K complex such as 3-methyladenine (3MA), wortmannin, and LY294002 result in complete inhibition of autophagosome formation, thus emphasizing the importance of the PI3K complex in the autophagy process. A lot is still unknown about how this complex regulates autophagy; however, it is clear that phosphatidylinositol 3-phosphates play an important role in the signaling process. WD repeat proteins interacting with phosphoinositides (WIPI1 and WIPI2), autophagy-linked FYVE protein (Alfy), and double FYVE domain-containing protein (DFCP1) have all been reported to play important roles in the autophagy process, and all require phosphatidylinositol 3-phosphate signaling for their recruitment to the phagophore. The exact function of these proteins is still to be fully elucidated. However, it is likely that they play a role in scaffolding and signaling for other autophagy machinery proteins [62–65].

Origin of the phagophore

The origin of the autophagic membrane has been a subject of debate for many years. Many hypotheses have been formed to explain the origin of the phagophore (also known as the isolation membrane), including suggestions that it originates from already formed membrane structures, such as the ER, the Golgi, and the mitochondria, or that it is formed through de novo synthesis.

Early publications investigated the origin of the phagophore by fractionating the autophagosomes from rat hepatocytes and carrying out biochemical assays such as immunoblotting for protein markers of various membrane structures in the cell. These studies did not identify any positive markers and thus it was hypothesized that the autophagosome was in itself a unique organelle and thus was generated via de novo synthesis [66, 67]. However, major advances in microscopy techniques and identification of new autophagy markers led to new insights into the origin of the phagophore which refuted earlier studies. Recent publications have convincingly described the phagophore originating from structures in the ER membrane termed ‘omegasomes’[64], as well as from the mitochondria [68] and the plasma membrane [69] (reviewed in [70]). It is likely that the phagophore’s origin is not from a unique location in the cell, and that, depending on the stress, cell type, or the extent of autophagy required, all these structures may contribute to the formation of the phagophore.

Autophagosome elongation

The elongation of the phagophore requires two ubiquitin-like conjugation systems, the ATG12–ATG5 conjugation system and the ATG8 conjugation system.

Atg12–Atg5 conjugation system

The UBL proteins, ATG12 and ATG5, are essential players in the elongation of the pre-autophagsomal structure (PAS). The covalent conjugation of ATG12 to ATG5 is mediated by the E1 enzyme ATG7 and the E2 enzyme ATG10 [50]. This complex further interacts with ATG16L through ATG5. ATG16L function is not entirely known; however, because its C-terminal contains seven WD repeats, it is believed to serve as a platform for protein–protein interaction at the autophagosomal membrane [71].

Recruitment of the ATG12–ATG5–ATG16L complex to the autophagsomal membrane requires the formation of phosphatidylinositol 3-phosphate by the PI3K complex; however, the exact mechanism for its recruitment is unknown.

ATG8 conjugation system

The UBL protein ATG8 is another major player in the elongation of the PAS. In contrast to yeast which contains only one ATG8 protein, mammals express a family of mATG8 proteins which is subdivided into LC3s and γ-aminobutyric acid receptor-associated proteins (GABARAPS). The mATG8 family proteins are translated as pro-forms which are subsequently cleaved at the C-terminal region by the protease ATG4, exposing a glycine residue. The E1 enzyme, ATG7, and the E2 enzyme, ATG3, facilitate the binding of a phosphatidylethanolamine (PE) to mATG8s exposed glycine residue via the PE amino group.

The lipidated form of ATG8 is recruited to the autophagosomal membrane, and is thought to require the ATG12–ATG5–ATG16 complex as a platform (Figs. 2, 3)

Fig. 2.

The autophagy pathway is divided into different phases; induction, vesicle nucleation, elongation, maturation, lysosomal fusion and degradation. Activation of the ULK1/2 complex requires mTORC1 inhibition and AMPK-mediated phosphorylation of ULK1. This complex is essential for the initial induction of the phagophore. The PI3K complex (see text and Fig. 3) is activated upon Bcl-2/Bcl-X_L dissociation from Beclin’s BH3 domain. PI3K complex I is required for the induction and nucleation of the phagophore whereas PI3K complex II is involved in the expansion and curvature of the autophagosomal membrane (see text for details). The elongation phase of the autophagsome requires the conversion of LC3I to LC3II and the formation of the ATG12–ATG5–ATG16 complex. LC3II and ATG12–5–16 complex are required for substrate specificity and scaffolding roles on the autophagosome. Upon maturation of the autophagosome, ATG12–5–16 and the outer membrane bound LC3II are recycled back in the cytosol. The mature autophagosome fuses with a lysosome where it is degraded by resident cathepsins

Fig. 3.

The UPR can regulate autophagy at different stages in the process, induction, vesicle nucleation, and elongation and maturation. Left hand panel induction of autophagy by the UPR can occur through multiple pathways. Ca²⁺ release from the ER lumen via the inositol 1,4,5-trisphosphate receptor (IP3R) can activate calcium calmodulin kinase II (CaMKII). CaMKII can subsequently phosphorylate and activate AMP kinase (AMPK) which in turn phosphorylates and activates the tuberous sclerosis complex (TSC). TSC inhibits mTORC1 and subsequently relieves mTORC1 inhibition on the ULK1/2 complex. PERK activation results in the non-canonical translation of the transcription factor ATF4. ATF4 can transcriptionally upregulate REDD1 which results in the activation of TSC and subsequent inhibition of mTORC1. ATF4 can also transcriptionally upregulate another transcription factor known as CHOP. CHOP targets expression tribbles-related protein 3 (TRB3). TRB3 can directly inhibit Akt which relieves Akt’s inhibitory effects on TSC, resulting in mTORC1 inhibition and subsequent activation of ULK1/2 complex. CHOP also transcriptionally upregulates ERO1-α. ERO1-α has been shown to stimulate IP3R-mediated Ca²⁺ release from the ER lumen resulting in activation of CaMKII–AMPK–TSC arm leading to mTORC1 inhibition and subsequent activation of ULK1/2 complex. Middle panel the activation of the PI3K complex is an essential step for the induction, nucleation and curvature of the phagophore. Multiple players are involved in the activation of the PI3K complex in response to ER stress. DAPK1 remains in a phosphorylated inactive state under resting conditions. In response to ER stress DAPK1 is dephosphorylated resulting in the activation of its kinase domain. DAPK1 can phosphorylate Beclin’s BH3 domain preventing the inhibitory association of Bcl-2/Bcl-X_L. The PERK–ATF4–CHOP arm can also promote the activation of the PI3K complex. CHOP has been reported to transcriptionally upregulate BH3-only proteins. BH3-only proteins can bind to Bcl-2/Bcl-X_L and displace them from Beclin’s BH3 domain. IRE1-mediated activation of JNK can also result in activation of the PI3K complex. JNK has been shown to phosphorylate Bcl-2/Bcl-X_L and inhibit their association with Beclin1. Right hand panel elongation and maturation of the phagophore requires two important processes to occur, the conversion of LC3I to LC3II and the formation of the ATG12–5–16 complex (see text for details). ATF4 can transcriptionally upregulate LC3 and ATG12, while CHOP can transcriptionally upregulate ATG5. The transcriptional upregulation of these three proteins is essential for the formation of the autophagosome

Maturation of the autophagosome

The final stage in the autophagy pathway is the transport and fusion of the mature autophagosome to the lysosome. The trafficking of the autophagosome to the lysosome is facilitated by the cytoskeleton, specifically the microtubule network. The FYVE protein, FYCO, functions as an adaptor protein between autophagosomes and the microtubule network to promote the trafficking of autophagosomes on the lysosome. Multiple binding partners of FYCO have been identified at the autophagosomal membrane. A complex between FYCO, phosphatidylinositol 3-phosphate, Rab7, and LC3 is believed to be formed on the autophagosomal membrane at the maturation stage. This adaptor complex is believed to bind to kinesins through FYCO and facilitate microtubule plus end-directed transport of autophagic vesicles [50, 72].

It has been shown that fully formed autophagosomes can bind to early endosomes forming structures known as amphisomes, before binding to lysosomes. However, this stage of the autophagic pathway is not clearly understood, and it remains unclear whether the endocytic pathway is required for autophagosomal degradation [50].

The binding of the autophagosome to the lysosome is facilitated by Rab7, which binds to LAMP1/2 on the lysosomal membrane. ATG9 is believed to be involved in the transport of the SNARE machinery, VAM9, VAM7, and Vti1b to the autophagosome to facilitate the fusion of the autophagosomal and lysosomal membranes [50].

Selective autophagy

Until recently, autophagy of organelles and protein aggregates was considered a non-selective process. However, identification of the autophagy receptor proteins provides compelling evidence for the selective targeting of cargo for autophagy [73]. Publications reporting selective autophagy include evidence for selective degradation of mitochondria (mitophagy), ER (ER-phagy/reticulophagy), ribosomes (ribophagy), peroxisomes (pexophagy), Golgi (crinophagy), endosomes (heterophagy), pathogens (xenophagy), aggresomes (aggrephagy), and lipids (lipophagy) [74–82]. It is becoming clear that autophagy is not as random as first perceived, and can in fact be a relatively selective and regulated process. Despite this rising evidence, autophagy is still often described as a non-selective process, primarily in response to starvation. However, a study recently showed that autophagy due to starvation led to the degradation of proteins and organelles in a systematic, selective way and not in a non-selective bulk degradation manner [83].

There are several proteins which have been identified to be required for the selective removal of specific substrates, including the autophagy receptors p62, Nbr1, Nix, NDP52, Smurf1/optineurin, and c-Cbl. What differentiates these from other proteins involved in the selective removal of substrates is the presence of the LC3-interacting region (LIR) which mediates the interaction with the autophagosome membrane bound LC3 family members LC3/GABARAP/GATE-16, as well as a domain which recognises the specific substrate to be targeted to the autophagosome [73, 84, 85].

The selective autophagosomal degradation of most substrates has been shown to require p62 and Nbr1 [86]. However, for the selective removal of invasive pathogens, NDP52 and Smurf1/optineurin have been shown to be the important mediators for their targeting to the autophagosome [85, 87]. Nix has only been shown to be involved in the selective removal of mitochondria, most prominently shown during reticulocyte differentiation [88, 89]. Damaged or depolarized mitochondria have also been shown to be selectively removed by autophagy (mitophagy). This process also relies on Nix for their removal; however, it also requires the kinase PINK1 and the E3 ligase Parkin for the ‘priming’ of the mitochondria for their selective removal [90, 91]. More recently c-Cbl has been described as an autophagy receptor protein following the identification of an LIR. C-Cbl has been shown to selectively target Src to the autophagosome in conditions where FAK is compromised, in turn preventing Src toxicity and promoting cancer cell survival [84].

Autophagy receptor proteins are quite well characterized both structurally and functionally; however, very little is known about their regulation and the functional consequence of their absence during different stress responses. Further discovery of autophagy receptor proteins and new insights into the regulation of these proteins in response to cellular stress responses will advance the field of selective autophagy.

Autophagy regulation by ER stress

ER stress and autophagy are individually very elaborate and complex systems. It is well established that autophagy is upregulated in response to ER stress; however, very little emphasis is put on its importance as a mediator in relieving ER stress. Here, we will describe what is known about how ER stress can affect various stages of autophagy, including autophagy induction, vesicle nucleation, and elongation of the phagophore, and we will discuss/describe known autophagy machinery genes which are transcriptionally upregulated by UPR signaling.

Autophagy induction

Calcium release

As discussed earlier, the ER is the site of the cells Ca²⁺ stores and is required for the folding of nascent proteins by Ca²⁺-requiring molecular chaperones in the ER lumen. The release of Ca²⁺ from the ER lumen to the cytosol can be both an inducer of ER stress or/and a result of ER stress. A commonly used pharmacological inducer of ER stress is thapsigargin, an inhibitor of the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) pump. Thapsigargin inhibits the reuptake of Ca²⁺ into the ER lumen and thus results in depletion of ER Ca²⁺, resulting in malfunctioning ER chaperones and accumulation of unfolded proteins in the ER lumen.

An increase in cytosolic Ca²⁺ has been shown to lead to initiation of autophagy. Studies have also demonstrated that even Ca₂(PO₄)₃ precipitates, that are introduced to cells during transfections, are capable of specifically inducing autophagy in cells [92]. This process is mediated by Ca²⁺/calmodulin-dependent kinase kinase-β which is activated in response to increased cytosolic Ca²⁺ and subsequent activation of AMPK [93]. AMPK in turn is involved in autophagy activation through inhibition of mTORC1 and direct phosphorylation of ULK1 [8, 94].

ER stress has been shown to be important in the modulation of this particular Ca²⁺ flux through inhibition of ER-resident Bcl-2. Under resting conditions, ER-localized Bcl-2 is important for the maintenance of ER Ca²⁺ stores [95]. JNK-mediated phosphorylation of Bcl-2 affects the latter’s ability to control the ER Ca²⁺ stores [96].

Another factor which contributes to ER stress-induced Ca²⁺ release is CHOP-mediated transcriptional upregulation of ER oxidoreductin 1 alpha (ERO1α). ERO1α plays an essential role in the ER during resting conditions to provide an oxidative environment to facilitate disulfide bond formation by enzymes such as PDI during the folding of nascent proteins. Under these conditions, ERO1α plays an essential protective role at the ER; however, at high levels, ERO1α stimulates activity of the inositol 1,4,5-trisphosphate receptor (IP3R), resulting in the release of Ca²⁺ into the cytosol. Tabas’ group have reported that CHOP-mediated transcriptional upregulation of ERO1α is required for Ca²⁺ release in response to ER stress, and that knockdown of either CHOP or ERO1α prevents Ca²⁺ release and delays ER stress-induced cell death [97]. ERO1α activation of IP3R is thought to be mediated independently of its oxidative ability. It has been hypothesized that ERO1α activates IP3R via the sequestration of Erp44, a negative regulator of IP3R [98]. These findings may support the evidence that autophagy induction in response to fluctuation in the cytosolic Ca²⁺ is directly regulated by ER stress in those conditions [93].

It is possible that a feedback mechanism between these two processes also exists, allowing autophagy to regulate the extent of ER stress-mediated autophagy through Ca²⁺ signaling. It has been shown that, in T lymphocytes defective in autophagy, ER Ca²⁺ stores are increased due to a defect in redistribution of stromal interaction molecule-1, resulting in ER expansion upon defective Ca²⁺ flux to those cells [99].

REDD1

The expression of regulated in development and DNA damage responses 1 (REDD1; also known as DDIT4) mRNA is upregulated in response to an array of stress stimuli, including ER stress [100]. During ER stress, REDD1 expression is regulated by the PERK–ATF4 arm of the UPR. Experiments with PERK^−/− and ATF4^−/− MEFs demonstrated that ER stress failed to induce REDD1 mRNA in these cells [100]. In support of a role for PERK signaling in this process, overexpression of ATF4 in HEK293T cells was sufficient to induce upregulation of REDD1 [100]. Moreover, further studies identified that activation of REDD1 during ER stress depends on ATF4 and its downstream effector CCAAT/enhancer-binding protein-β (C/EBP-β) [101]. REDD1 transactivation leads to inhibition of mTOR in a TSC1/TSC2-dependent manner, that will consequently activate the autophagic pathway upon the release of ULK1 from inactive mTOR [8, 101, 102].

Akt

The PI3K–Akt signaling pathway is a positive regulator of mTORC1 and has been well described in a number of cell models and organisms [103, 104]. The PI3K–Akt pathway is a pro-survival pathway involved in survival, cell growth, and proliferation through the positive regulation of mTORC1. As described earlier, the PI3K–Akt pathway regulates mTORC1 activation [105].

ER stress results in the inactivation of the Akt pathway, contributing to the decrease of mTOR activity and subsequent autophagy induction [106, 107]. The mechanism of how ER stress and the UPR can inhibit the Akt pathway is still unclear; however, a few insights have been made.

The ER chaperone, Grp78 which is transcriptionally upregulated by ATF6, has been demonstrated to prevent phosphorylation of Akt at serine473, thus preventing Akt regulation of downstream kinases [108]. Grp78 was shown to interact with Akt at the plasma membrane in response to ER stress; however, it is still unclear whether there is a direct interaction or if other factors are required [108, 109].

Tribbles homolog 3 (TRB3) is a negative regulator of the Akt signaling pathway. TRB3 is transcriptionally upregulated in response to ER stress through CHOP and ATF4 working in concert [110]. TRB3 has been shown to directly bind to Akt and inhibit its downstream signaling [111]. Knockdown studies have demonstrated that both ATF4 and CHOP are required for the transcriptional upregulation of TRB3. However, high protein levels of TRB3 result in a negative feedback loop by binding to ATF4 and CHOP and targeting them for degradation [110, 111].

Vesicle nucleation

CHOP

The PI3K complex is required for PAS induction and vesicle nucleation. As previously discussed, Beclin 1 is a core component of the PI3K complex and can be tightly regulated by anti-apoptotic Bcl-2 family members [112, 113]. CHOP expression during ER stress is tightly correlated with the inhibition of Bcl-2 expression, both at the protein and transcript level providing a direct link between ER stress and Beclin 1 activation [114, 115].

JNK

In response to ER stress, IRE1 kinase domain recruits the adaptor molecule TRAF2. Apoptosis-signal-regulating kinase (ASK1) is recruited by the IRE1–TRAF2 arm where it mediates signaling by MAP kinases, JNK and p38 [40]. JNK activation results in phosphorylation of pro-apoptotic proteins enhancing their activity, and also phosphorylates anti-apoptotic Bcl-2 inhibiting its activity [116, 117]. Under cellular resting conditions, PI3K remains in an inactive state due to the association of the anti-apoptotic proteins Bcl-2 and Bcl-X_L with Beclin’s BH3 domain. JNK-dependent phosphorylation of Bcl-2 and Bcl-X_L results in their dissociation from Beclin’s BH3 domain and the activation of the PI3K complex [118].

Study of cells deficient in IRE1, ATF6, or PERK showed that IRE1 plays a significant role in induction of autophagy, as measured by the intensity of LC3 puncta formation and LC3-I conversion to LC3-II [119]. In this report, the importance of transient JNK activation was highlighted in the promotion of pro-survival autophagy against prolonged autophagy that leads to initiation of apoptosis [119]. Independent studies, using dihydrocapsaicin treatments, confirmed that ER stress-induced autophagy relies heavily on transient activation of JNK signaling [120].

DAPK1

Death-associated protein kinase 1 (DAPK1) is a calcium/calmodulin (CaM)-regulated serine/threonine kinase. DAPK1 is activated in response to ER stress mediated via its dephosphorylation, allowing CaMK to bind and positively regulate it. It is believed that protein phosphatase 2A (PP2A) is involved in the dephosphorylation of DAPK1 in response to ER stress; however, the involvement of other phosphatases also plays a role in this process. DAPK1 is a positive regulator of autophagy and it exerts its effects via the phosphorylation of Beclin 1[121]. DAPK1-mediated phosphorylation of Beclin 1 reduces its affinity for Bcl-2 and thus causes dissociation of Beclin 1, relieving its inhibitory effects and allowing the formation of the Vps34 complex [122].

Elongation of the phagophore

PERK–ATF4–CHOP

As discussed above, the elongation of the phagophore requires two events to occur: the conversion of LC3I to LC3II and the covalent binding of ATG12 to ATG5. During prolonged stress-induced autophagy, ATG5, ATG12, and LC3I are quickly engaged in autophagosome formation, and thus these genes must be transcriptionally upregulated in order to maintain flux through the pathway. PERK activation results in the transcriptional upregulation of ATG5, ATG12, and LC3 [37, 123]. ATG12 is transcriptionally upregulated in response to ER stress in a PERK-eIF2α-dependent manner; however, the transcription factor involved in its upregulation has yet to be identified [123]. LC3 and ATG5 are also transcriptionally upregulated through the PERK–eIF2α arm; however, LC3 is upregulated by ATF4 whereas ATG5 is upregulated by CHOP [37]. Thus, during ER stress-induced autophagy, PERK replenishes cellular supplies of ATG5, ATG12, and LC3I allowing for sustained autophagy flux.

XBP1

FoxO1 has been described as a major regulator of autophagy in various cell lines, both as a cytosolic protein and a transcription factor [124, 125]. For example, acting as a transcription factor, FoxO1 is involved in BNip3 expression and neuronal survival [126]. On the other hand, acetylated cytosolic FoxO1 binds to ATG7, and can promote autophagy in response to stress, leading to cell death [127]. Studies in Hungtington’s disease mouse models have demonstrated that XBP1 deficiency leads to increased levels of macroautophagy in cells, and that this was correlated with high expression of FoxO1[128]. This suggests that ER stress can also act as a negative regulator of autophagy, being unfavorable for the physiological outcome in this particular disease model.

ER stress and autophagy in disease

As previously discussed, both autophagy and ER stress are important cellular adaptation programs initiated in response to various stress stimuli. Increased levels of autophagy and ER stress markers are observed in many chronic respiratory diseases, such as asthma and chronic obstructive pulmonary diseases (COPD) [129–131], as well as cardiovascular diseases, such as ischemia/reperfusion injury, cardio-hypertrophy, and heart failure [132, 133]. Some recent exciting research describes the role of ER stress and autophagy in diabetes, neurodegeneration, and cancer. With this in mind, development of novel drugs which target and manipulate these pathways holds great therapeutic promise in the treatment of many diseases [2, 134], some of which are discussed below.

Diabetes

Type 2 diabetes mellitus is a disorder characterized by insulin resistance and low insulin production in β-cells. The highly secretory pancreatic β-cells rely on an organised and active ER for the folding of large amounts of proinsulin. Defects in UPR signaling or excessive ER stress is implicated in the development of experimental and clinical diabetes [135]. PERK is highly expressed in pancreatic β-cells and is essential for their development. A mutation in the EIF2AK3 gene, which encodes PERK, has been shown to lead to Wolcott–Rallison syndrome, a disorder that results in infantile diabetes. Nevertheless, chronic activation of PERK is implicated in both type 1 and 2 diabetes [136]. These studies suggest that a fine tuning of the UPR is essential for β-cell function.

On the other hand, autophagy is a beneficial pro-survival mechanism in β-cells, required for the removal of malfolded proteins and protein aggregates, a general occurrence in these highly secretory cells. Defects in autophagy contribute to the increased incidence of type 2 diabetes with age due to reduced β-cell function [137]. Moreover, studies on autophagy-deficient mice indicate that autophagy is required for appropriate physiological UPR signaling, resulting in increased ER stress. As such, defects in autophagy predisposed mice to type 2 diabetes when cross-bred with obese mice due to lipid injury, which exerted unmanageable levels of ER stress on the β-cells [138, 139]. Another study reported marked autophagy induction during treatment of human pancreatic islets with palmitic acid, but not with glucose [140]. Thus far, relatively little is known about the exact interplay between autophagy and ER stress, but most of the evidence is in favor of targeting autophagy as a protective mechanism of β-cells to ameliorate ER stress [141].

Neurodegenerative diseases

The dysregulation of autophagy has been implicated in numerous neurodegenerative diseases, such as Alzheimer disease, Parkinson disease, Lewy body disease, Huntington disease, multiple sclerosis, and amyotrophic lateral sclerosis (ALS) [142–146]. A common feature of neurodegenerative diseases is the accumulation of aggregated proteins in the cytosol of neuronal cells, which eventually leads to neuronal toxicity, due to the inability of the neurons to clear misfolded and aggregated proteins (reviewed in [147]). Since autophagy is capable of degrading large structures, including protein aggregates, it thus represents a potential therapeutic target in these diseases [148].

A recent study demonstrated that Alzheimer disease (AD) patients show reduced expression of Beclin 1 in the entorhinal cortex resulting in autophagy attenuation. Further studies in mouse models of AD also showed reduced levels of Beclin 1 mRNA and subsequent overexpression of Beclin 1 using lentivirus, resulted in clearance of protein aggregates and reduced amyloid pathology. However, a small subclass of mice showed no change in Beclin 1 expression, suggesting that Beclin 1 may not be required to initiate the pathology; however, it may be influential in the progression of the disease [145].

Parkinson disease (PD) is commonly associated with mutations most commonly in the Parkin gene and second most frequently in the PINK1 gene [149]. Parkin is an E3 ligase required for the selective removal of dysfunctional mitochondria, signaling them for autophagic degradation, which may otherwise result in apoptosis. PD is often accompanied by the accumulation of dysfunctional mitochondria in the neuronal cells which inevitably leads to their demise. PINK1 is a kinase which is frequently mutated in PD. In vitro studies have demonstrated that PINK1 can directly target Beclin 1 as a positive regulator of autophagy, and subsequent inhibition of PINK1 resulted in defective autophagy. This function of PINK1 again demonstrates the importance of defective autophagy for the progression of PD [89, 146].

Lewy body disease is a disorder which closely relates to an overlap in the pathology of PD and AD. The disease results in accumulation of Lewy bodies, aggregates which are mainly composed of aggregated α-synuclein. Beclin 1 expression is strongly reduced in these cells and results in compromised autophagy induction. Both in vitro and in mouse models, the introduction of beclin1 led to the upregulation of autophagy and the clearance of Lewy bodies [142, 144].

In Huntington disease. the autophagy pathway is impaired and is once again due to low levels of beclin1 expression. Beclin1 expression decreases with age and is a contributing factor to why older people are more susceptible to these diseases [150]. The upregulation of autophagy using rapamycin, the chemical inhibitor of mTOR, in mouse models of Huntington disease led to the increased clearance of the mutant huntingtin protein and significantly decreased neurodegeneration [151].

In a recent study in cells from ALS patients, the inhibition of IRE1/XBP-1 axis led to the upregulation of autophagy and cytoprotection. This for the first time shows the UPR having a negative regulatory effect on autophagy [152].

It’s clear that autophagy dysregulation plays a contributory factor to many neurodegenerative diseases and is an appealing therapeutic target in clinical research. ER stress clearly plays an integral role in autophagy regulation; however, the decreased expression of key regulators required for autophagy disrupts this synchronicity. Without autophagy, cytotoxic protein aggregates accumulate in the cell. The proteasome is unable to degrade these large aggregates, resulting in chronic state of stress and apoptosis.

Cancer

Since prolonged irreversible ER stress leads to apoptosis, induction of acute ER stress has emerged as a therapeutic approach for developing anti-cancer drugs, although concomitant induction of autophagy with the UPR provides cancer cells a possible survival mechanism. Therefore, a number of laboratories have been developing combination therapies, using ER stress inducers with autophagy inhibitors for a number of different cancer types. For example, combination therapy with autophagy inhibitor chloroquine and histone deacetylase or cyclooxygenase-2 inhibitors, which induce ER stress, suppress the growth of triple-negative breast cancer cells both in vitro and in vivo [153, 154]. Similar synergy has been shown between bortezomib (a proteasome inhibitor that leads to induction of both ER stress and autophagy) and chloroquine for the treatment of lymphoma [155]. An investigation of precise molecular mechanisms are necessary for the combination of each of the particular compounds, as the cross-talk between the two processes is very tight, not to mention the dual role of autophagy in cancer progression [156, 157].

The nuclear hormone receptor PPARγ agonist, 4-O-carboxymethyl ascochlorin, induces both ER stress and autophagy in hepatocarcinoma cell lines. Interestingly, in this case, inhibition of autophagy by 3-methyladenine also inhibited ER stress response, blocking the induction of cell death [158].

On the other hand, some types of cancer like human high-grade gliomas are much more prone to die through induction of autophagic cell death, making all the ER stress-inducing agents very efficient, as they would induce in this case a mixture of apoptosis and autophagic cell death. Overexpression of CHOP alone is sufficient to decrease cell viability in human high grade glioma cell lines though the induction of both cell death modalities, making those two events a promising target for therapy through induction of just the ER stress pathway [159].

Another clinical approach that could be developed from the current knowledge of autophagy and ER stress interplay is specific targeting of the IRE1 branch during tumor development. Tumors develop under conditions of hypoxia, and inhibiting the IRE1 branch would be suppressive for tumor angiogenesis as well as decreasing the levels of autophagy, and therefore substantially decreasing the prosurvival mechanisms of the whole tumor.

Conclusion

Autophagy, a highly regulated and evolutionarily conserved process of self-cannibalism, plays an important role in basic cellular homeostasis and is activated during cellular stress response to help the cell deal with the otherwise harsh stress conditions. During ER stress, the components of the autophagic machinery are tightly regulated by the key mediators of the UPR (Table 1). Both autophagy and UPR signaling are cytoprotective strategies; however, if the intensity or duration of cellular stress is increased, these pathways converge on the cell death machinery. The synchronicity of autophagy and ER stress is also displayed in several human pathologies, which emphasizes the importance of better understanding their relationship. Autophagy and UPR markers are observed in numerous pathologies including chronic respiratory diseases such as asthma and COPD, cardiovascular diseases, carcinomas, diabetes, and numerous forms of neurodegenerative disorders. Intensive research is underway to form a better understanding of the role which autophagy and ER stress play in these disease models. The highly regulated nature of autophagy and the UPR, and in particular the fact that both of these processes are regulated by transcription factors and multiple kinases, make them druggable and amenable to both genetic and pharmacological manipulation.

Table 1.

A list of key ER stress sensors and how they regulate autophagy

ER stress component	Targets	Autophagy component/phase	Regulation mechanism
IRE1	JNK	ULK1/induction	IRE1 leads to JNK activation, that in turn mediates Bcl-2 inhibition and disturbances in calcium flux, leading to induction complex assembly in response to AMPK signaling
		Vesicle nucleation	JNK-mediated Bcl-2 phosphorylation results in its dissociation from Beclin1
ATF6	Akt	ULK1/induction	Grp78 is transcriptionally upregulated by ATF6 Grp78 localizes with Akt, inactivating it and leading to subsequent mTOR inhibition
PERK	ATF4	ULK1/induction Elongation of the phagophore	PERK activation induces REDD1 expression through activation of ATF4, that in turn is involved in mTOR inhibition and subsequent activation of ULK1 induction complex Transcriptional upregulation of Atg12 and LC3
	ERO1β	ULK1/induction	CHOP-mediated transcriptional upregulation of ERO1α stimulates activity of IP3R, thus leading to calcium release into a cytosol and induction complex activation in response to AMPK
	Akt	ULK1/induction	Grp78, mainly regulated by ATF6, localizes with Akt, inactivating it and leading to subsequent mTOR downregulation. In this case, TRB3 upregulated by CHOP and ATF4 plays a role in Akt inhibition
	CHOP	Vesicle nucleation Elongation of the phagophore	CHOP expression causes Bcl-2 inhibition and therefore Beclin1 activation Transcriptional upregulation of Atg5
Unknown	DAPK1	Vesicle nucleation	DAPK1 phosphorylates the BH3 domain of Beclin1 preventing the association with Bcl-2

Paradoxically, autophagy can also lead to cell death. However, the cell death process does not seem to be a unique or new cell death modality, rather it appears to take the route of already elucidated death pathways, such as caspase-mediated apoptosis or caspase-independent necrosis. Ongoing studies in our own group and others should shed further light on these processes in the near future.