Oncogenic and oncosuppressive signal transduction at mitochondria-associated endoplasmic reticulum membranes.

The different mechanisms employed by proto-oncogenes and tumor suppressors to regulate cell death pathways are strictly linked to their localization. In addition to the canonical control of apoptosis at a transcriptional/nuclear level, intracellular zones are emerging as pivotal sites for the activities of several proapoptotic and antiapoptotic factors. Here, we review the function of the endoplasmic reticulum-mitochondria interface as a primary platform for decoding danger signals as well as a structural accommodation for several regulator or effector proteins.

endoplasmic reticulum

inner mitochondrial membrane

inositol 1, 4, 5-trisphosphate receptor

membrane-associated membrane

mitochondrial Ca2+ uniporter

mitochondrial permeability transition pore

mechanistic target of rapamycin complex 2

outer mitochondrial membrane

RNA-dependent protein kinase (PKR)-like ER kinase

promyelocytic leukemia

phosphatase and tensin homolog deleted on chromosome 10

reactive oxygen species

voltage-dependent anion channel

Introduction

Over the past 15 years, alternative subcellular districts have been described as pivotal sites of action for several oncogenes and oncosuppressors. Although the core apoptotic machinery is tightly controlled at the transcriptional level, a series of post-translational mechanisms, such as translocation to different intracellular compartments, phosphorylation events, and protein–protein interactions, represent important aspects of regulation of the apoptotic pathway. Although the nucleus has been identified as the main target of different oncogenes, such as c-myc, and tumor suppressors, such as p53, it is now clear that coordination of the apoptotic process occurs at multiple cellular levels, far removed from the biochemical scheme of one protein–one function. These unconventional pathways probably act in synergy with transcription-dependent pathways, and in some pathological contexts they function not only to support the nuclear mechanisms but can also be considered the primary molecular route.

Among the different non-nuclear activities, mitochondrial signaling is one of the most studied and best characterized. Mitochondria are the major site of localization for oncogenes and oncosuppressors because of their central role as integrators and transducers for proapoptotic signals. However, protein targeting to mitochondria generally requires the presence of specific import signals; therefore, a large number of proteins cannot easily enter mitochondria but instead exert their effects in the surrounding zone, especially in the contact areas between the endoplasmic reticulum (ER) and mitochondria.

Membrane-bound organelles exchange metabolic signals and information through the formation of specific membrane contact sites, and the ER-mitochondria interface represents one such connection. The mitochondria and ER join together at multiple contact sites, forming a specified subcellular fraction that is currently termed mitochondria-associated membranes (MAM). The ER and mitochondria not only physically couple but also establish a tight communication that plays a crucial role in several processes, such as lipid trafficking, calcium (Ca2+)-transfer, inflammation, and apoptosis. Nevertheless, several regulatory factors are shared by the 2 organelles, leading to the establishment of a molecular platform to receive and decode a wide range of inputs, including apoptotic signals. During cellular stress, ER-mitochondria connections lead to the prompt activation of caspase-dependent and caspase-independent cell death effector mechanisms based on the capacity of the 2 organelles to sense and react to a multiple array of danger signals. Thus, the MAM can be conceived as a physical and functional scaffold for the primary response to alterations of cellular homeostasis, and it is therefore logical that oncogenes and tumor suppressors can localize at or move to MAM upon cellular stress.

This review will address the structure of MAM, the main apoptotic-related functions of MAM (with a focus on reactive oxygen species (ROS) and Ca2+ exchange) and the proteins that regulate the cell death pathways through their localization to the ER-mitochondria interface ().

Table 1.

Summary of the most important oncogenes and oncosuppressors discussed in this review. MAM, mitochondria-associated ER membranes; Mt, mitochondria; PAM, plasma membrane associated membranes; ER, endoplasmic reticulum; Nu, nucleus; Cyt, cytosol; PM, plasma membrane

Protein	Localization	Functions at MAM	MAM Interactors	References
p66Shc	MAM, PAM, Cyt, Mt	Cytosolic adaptor protein involved in cellular response to oxidative stress	Unknown	24,26
Akt	Cyt, Nu, MAM	Serine/threonine protein kinase. Inhibition of Ca2+ release from ER; antiapoptotic functions	Bad, Bax, HK2, IP3R, PACS2, PTEN, PML, mTORc2	33,37,38,40,41
Bcl-2	Nu, ER, Mt, Cyt, MAM	Induction of Ca2+ leakage from ER; antiapoptotic functions	Bad, Bcl-xL, IP3R	44,46,47,51
Bcl-xL	Nu, Cyt, Mt	Induction of Ca2+ leakage from ER; antiapoptotic functions	Bad, Bcl-2, IP3R	57
Bad	Cyt, Mt, ER	Proapoptotic functions	Bcl-2, Bcl-xL	48,68,69
Bax	Cyt, Mt, ER	Proapoptotic functions	Akt, Bcl-2	48,68,69
HK2	Mt	Glucose phosphorylation; antiapoptotic functions	Akt, VDAC1	33,35,36
PTEN	Cyt, Mt, MAM, Nu	Most commonly lost or mutated tumor suppressor in human cancers; negative regulator of Akt, regulation of Ca2+ release via IP3R3; proapoptotic functions	Akt, IP3R, PP2a	87
PML	MAM, ER, Nu	Implicated in the pathogenesis of leukemia and other cancers; negative regulator of Akt; proapoptotic functions	Akt, IP3R, PP2a	81
mTORc2 (mTOR complex 2)	ER, Mt, MAM	Serine/threonine protein kinase; Akt activator; control of MAM integrity; regulation of Ca2+ uptake; regulation of mitochondrial bioenergetics; antiapoptotic functions.	Akt, PACS-2	41
Sig1-R	MAM/ER	Molecular chaperone stabilizing the conformation of proteins at the MAM; promotes cellular survival; antiapoptotic functions	GRP78/BiP, (IP3R3)	44,52,53
GRP78/BiP	ER, MT, Cyt, MAM, Nu	ER chaperone - folding and assembly of membrane or secreted proteins; stabilizes IP3R3 at MAM	Sig1-R, CLU	52,56
CLU	ER, Mt, Cyt	Stress-induced chaperone; antiapoptotic functions	GRP78 Sig1-R	56
Mcl-1	MT, Nu, ER, MAM	Induction of Ca2+ leakage from ER; antiapoptotic functions	Bok, IP3R	58
K-Ras4B	PM, ER, Mt, Cyt	Regulation of Bcl-xL activity; antiapoptotic functions	IP3R, Bcl-xL	60
H-Ras12v	MAM, PAM, Mt, Cyt,	Regulation of Ca2+ signaling; antiapoptotic functions	Caveolin-1	61
K-Ras	MAM, PAM, Mt, Cyt	Inhibition of Ca2+ release from ER and reduction of ER Ca2+levels; antiapoptotic functions	IP3R	62
vMIA	Mt, MAM	Inhibition of apoptosis	Bax	63,64
HBx	OMM, Cyt, Nu	Induction of mitochondrial fragmentation and mitophagy; induction of dysfunction of permeability transition pore (PTP) complex	VDAC3	65
Enterovirus 2B protein	ER, Golgi-derived vesicles	Regulation of Ca2+ homeostasis; antiapoptotic functions	Unknown	67
Bok	Golgi, ER, MAM, Cyt	Upstream of Bax and Bak in control of the transmission of ER/MAM-derived apoptotic signals toward mitochondria; proapoptotic activity	Mcl-1, IP3R	71-74
Ero1-α	MAM, ER	Key controller of oxidative folding and ER redox homeostasis; enriched at MAM and regulates Ca2+ fluxes	IP3R, PAC-1	76
Fis1	MAM, Mt	Formation of a tripartite protein complex with procaspase-8 and Bap31; induction of apoptosis.	Bap31	78
Bap31,	MAM, ER, OMM	Formation of a tripartite protein complex with procaspase-8 and Fis1; induction of apoptosis.	Fis1, caspase-8	78
PERK	MAM, ER	Involved in folded protein response during ER stress; physically increases contacts between mitochondria and ER		79

MAM structure

It has been demonstrated that, in order to function properly, mitochondria need to communicate with other organelles and intracellular structures. Such communication between the ER and mitochondria can occur at close contact sites between organelles, even without the direct fusion of “interacting” membranes (). Studies on the specific interactions between mitochondria (which form an efficient calcium ion buffer) and the ER (the main intracellular calcium store) were initiated by Copeland and Dalton in the late 1950s. These pioneer studies were performed in cells of the pseudobranch gland of teleost fish and showed that part of the ER exists in association with mitochondria. Approximately 10 years later, other groups visualized these contacts in the rat liver and the onion stem by electron microscopy. Initial evaluation of the extent of the mitochondrial involvement in such interactions gave surprisingly high values, indicating that approximately 80 % of the mitochondrion interacted with the ER. In contrast, later studies showed that only 5–20 % of the mitochondrial surface interacts with the ER. Initially, studies with electron microscopy tomography estimated the distance between the ER and mitochondria as 100 nm, but later studies by Achleitner et al. indicated that this distance varied, and was in the range of only 10–60 nm. More recent data further reduced this distance to 10–25 nm, which allows proteins from the ER to associate directly with proteins and lipids present at the outer surface of the mitochondrial membrane. Protocols describing the isolation of ER-mitochondrial contacts indicated that these interactions are strong enough to be preserved upon subcellular fractionation. The subcellular fraction that was enriched for the contact sites between mitochondria and ER was named the “mitochondria-associated membrane” (MAM) fraction. A detailed protocol to isolate the MAM fraction was first described by Jean Vance in the early 1990s. Since then, the isolation procedure has been improved and adapted to isolate the MAM fraction from different organs, tissues, and various cell lines as well as from yeast. The isolated MAM fraction is composed of membrane shreds from both the ER and outer mitochondrial membrane (OMM) that were in close contact at the time of subcellular fractionation. More recently, the MAM fraction has also been regarded as an intracellular lipid raft of detergent-resistant domains of the ER. The contact sites between mitochondria and the ER are dynamically formed as a direct result of a stochastic apposition of ER with mitochondria and are dependent on intracellular signaling. Thus, MAM composition is transient and can be changing at any given time. The variety of roles played by the MAM fraction, as described in the literature, is related to their unique lipid and protein composition. Studies performed in the last decade revealed the molecular components of the MAM fraction, demonstrating that it contains several proteins (more than 75 according to Raturi and Simmen) and is crucial for many cellular processes, including protein sorting, inflammation, ER stress, Ca+ handling, lipid synthesis, trafficking, and apoptosis. However, the localization of some proteins in the MAM fraction and the extent of their enrichment are still under debate because their connection to the MAM fraction is unclear.

Figure 1.

Oncogenes and oncosuppressors at mitochondria-associated membranes. Oncogenes and tumor suppressors acting at the ER–mitochondria interface are shown as red hexagons and green octagons, respectively. Ca2+ players are shown in yellow. GRP75 interacts with IP3R and VDAC to bridge Ca2+ transfer from ER to mitochondria. Bcl-2 counteracts Bax–Bak activities both at mitochondria and ER sides of the MAM. Bok interacts with IP3R and Mcl-1. Akt phosphorylates IP3R, reducing Ca2+ release from the ER. At the mitochondria, Akt promotes the association between Hk-2 and VDAC. Akt activity is positively/negatively regulated by mTORc2, PML, and Pten. Bap31, caspase 8, and Fis1 form a platform to transduce the cell death signals between the ER and mitochondria. GRP78 might translocate from the ER to MAM upon ER stress induction. See text for further details. Bak, Bcl-2-antagonist killer; Bap31, B-cell receptor-associated protein 31; Bax, Bcl-2-associated X protein; Bcl-2, B-cell CLL/lymphoma 2; Bcl-xL, B-cell lymphoma-extra large protein; Bok, Bcl-2-related ovarian killer; CLU, clusterin; C sub, c subunit of mitochondrial ATP synthase; cyt. c, cytochrome c; ER, endoplasmic reticulum; Ero1-α, ER oxidoreductin-1 α; Fis1, Fission 1 homolog; GRP75, glucose regulated protein 75; GRP78, glucose regulated protein 78; Hk2, hexokinase 2; IP3R, inositol 1,4,5 trisphosphate receptor; Mcl-1, myeloid cell leukemia sequence 1; MCU, mitochondrial calcium uniporter; mPTP, mitochondrial permeability transition pore; mTORc2, mechanistic target of rapamycin complex 2; PERK, RNA-dependent protein kinase (PKR)-like ER kinase; PML, promyelocytic leukemia protein; pro-casp. 8, pro-caspase 8; Pten, phosphatase and tensin homolog deleted on chromosome 10; SERCA, sarco/endoplasmatic reticulum Ca2+ ATPase; Sig1-R, Sigma 1 receptor; VDAC, voltage-dependent anion channel.

Originally, the MAM fraction was described as the location of lipid synthesis and trafficking between the ER and mitochondrial membranes based on the presence of long-chain fatty acid-CoA ligase type 4 (FACL4) and phosphatidylserine synthase-1 (PSS-1) enzymes. The close apposition of mitochondria to the ER also explains the selective transmission of physiological and pathological Ca2+ and ROS signals directly from the ER to the mitochondria. The MAM also contains Ca2+ signaling elements of both organelles, thus supporting the central role of ER/mitochondria crosstalk in signal transduction. Therefore, the ER-mitochondria contact sites can be considered specialized microdomains for the transfer of Ca2+ signals. Ca2+ ions released from the ER by inositol 1,4,5-trisphosphate receptors (IP3Rs) cross the freely permeable OMM through voltage-dependent anion channels (VDACs), reach the inner mitochondrial membrane (IMM), and accumulate in the matrix via the mitochondrial Ca2+ uniporter (MCU) complex. Close apposition between the ER and mitochondria ensures the formation of microdomains at high [Ca2+] that overcome the low apparent Ca2+ affinity of the MCU. Hence, at the "Ca2+ hotspot" stage, the local [Ca2+] is >10 μM, allowing rapid Ca2+ transduction to the matrix despite the low Ca2+ affinity of the uniporter pore (). At the molecular level, the MAM chaperone glucose regulated protein 75 (GRP75) mediates the interaction between the IP3R and VDAC, structurally linking the Ca2+ efflux system at the ER with the channels at the OMM to favor positive regulation of mitochondrial Ca2+ uptake.14 Interestingly, VDAC1, but not VDAC2 and VDAC3, interacts with IP3R, sustaining transmission of the low-amplitude apoptotic Ca2+ signals to mitochondria.

Although Ca2+ exchange between the ER and mitochondria serves as a regulator of cellular bioenergetics, accumulation of Ca2+ can trigger opening of the mitochondrial permeability transition pore (mPTP), leading to release of proapoptotic factors, such as cytochrome c, into the cytosol. The molecular nature of the mPTP is still controversial, but recent evidence suggests the involvement of new structural components in pore formation, in particular the c subunit of mitochondrial ATP synthase. Conversely, the molecular composition of the MCU complex has been determined and its importance in the regulation of cell death pathways has been described in many cellular environments. Notably, silencing of a core component of the MCU complex, the regulatory subunit mitochondrial calcium uptake 1 (MICU1), exposes the mitochondria to drastic Ca2+ accumulation at basal conditions, produces ROS, and triggers the apoptotic process. The ER and mitochondria are 2 of the major sites for ROS production inside the cell. Exchange of ROS takes place at the MAM, and this ROS trafficking has a wide relevance in many pathological contexts, especially during ER stress (as discussed under "MAM as a strategic platform for oncosuppressor-dependent cell death"). Therefore, many regulators of the oxidative state of the cell are located at the MAM. p66Shc, a cytosolic adaptor protein involved in a cellular response to oxidative stress, has recently been identified at mitochondria-ER association sites. Moreover, an increasing body of evidence indicates that p66Shc is involved in tumorigenicity. There is a positive relationship between the level of p66Shc and the rate of cell proliferation in prostate cancer cells. This relationship is particularly relevant to steroid-induced signaling and elevated levels of ROS, which act as secondary messengers in cancer cells. The upregulation of ROS production by androgen or estrogens is accompanied by an increase in p66Shc, thereby promoting cell proliferation in the aforementioned types of cancer cells. Interestingly, androgen treatment decreases p66Shc phosphorylation at Ser36. The association of p66Shc with mitochondrial ROS production has also been repeatedly documented in our previous studies. The level of p66Shc in the MAM fraction increases in an age-dependent manner and correlates with mitochondrial ROS production, which has also been found to increase with age.

Oncogenic function of MAM

As cited in the aforementioned studies, Ca2+ and ROS transfer is the main apoptotic-related function of MAM. By localizing at ER-mitochondria membranes, oncogenes and oncosuppressors can alter this physiological exchange by modifying the cellular response to apoptosis (). This is the case for the serine/threonine kinase Akt. Survival signals involving Akt activation include both the caspase cascade and transcriptional control of apoptosis. Upon growth factor stimulation, phosphorylated Akt translocates to the nucleus from the cytoplasm via an activation process. Through a phosphorylation cascade, Akt promotes nuclear exclusion (cytoplasmic retention) of transcription factors of the forkhead family, thereby preventing transcription of the proapoptotic genes Fas ligand, BIM, TRAIL, and TRADD. Additionally, phosphorylation of Mdm2 by Akt is necessary for localization of Mdm2 to the nucleus, where it associates with p53 to target its ubiquitination. Moreover, Akt-dependent NF-κB nuclear translocation promotes the transcription of antiapoptotic genes, such as BFL1, cIAP1, and cIAP2.

In addition to its transcriptional activity, Akt has also been physically and functionally linked to both the ER and the mitochondria. Akt-dependent phosphorylation of the proapoptotic BH3-only protein Bad causes its dissociation from the Bcl-2/Bcl-XL complex at the outer mitochondrial membrane, inhibiting its cell death functions. Similar to its relationship with Bad, phosphorylation of Bax by Akt results in inhibition of the apoptotic features of Bax, such as oligomerization, insertion into mitochondrial membrane, and the formation of large pores to allow the release of proapoptotic factors. Akt also phosphorylates hexokinase 2 (HK2) to promote its association with the MAM protein VDAC1. This association not only affects the metabolic state of the cell by increasing efficiency and rate of the glycolytic pathway, but also protects cells from apoptosis. Hexokinase–VDAC1 binding seems to promote the closed state of the channel, preventing Ca2+-dependent opening of the mPTP and release of the proapoptotic protein cytochrome c.

Furthermore, at the ER side of MAM, Akt phosphorylates all IP3R isoforms, inhibits Ca2+ release from ER, and protects cells from apoptosis. Our group showed that Akt inhibits Ca2+ fluxes and apoptosis in the ER by preferentially phosphorylating the type 3 IP3R (IP3R3). IP3R3 is mainly localized to the MAM, suggesting a pivotal role of the proto-oncogene Akt at the ER-mitochondria interface. These results provided evidence for the hypothesis of individuation at MAM of both negative regulators of Akt (the tumor suppressors PTEN and PML, see section below) and the Akt activator mechanistic target of rapamycin complex 2 (mTORc2). Previous observations suggest that mTORc2 can interact with both the ER and the mitochondria. Indeed, mTORc2 resides at the MAM, where it regulates the phosphorylation state of IP3R3 and Ca2+ release from the ER. mTORc2 deficiency, as well as Akt downregulation, causes MAM disruption. mTORc2 controls MAM integrity, at least in part, via Akt-dependent phosphorylation of phosphofurin acidic cluster sorting protein 2 (PACS2), as suggested by the observation that PACS2 is a substrate of Akt and is required for MAM integrity. Moreover, mTORc2 controls mitochondrial functions and physiology in an Akt-dependent manner through HK2 phosphorylation. Thus, mTORc2 localization to MAM can be considered the ideal link to the multiple apoptotic-related functions of Akt, underlining the crucial role of the proto-oncogene at ER–mitochondria contact sites.

As in the case of Akt, strategic positioning at the MAM and regulation of Ca2+ fluxes is shared by other oncoproteins, including the Bcl-2 family members. Bcl-2, the “patriarch” of the family, is highly enriched at the MAM, where it functions at both the ER and the mitochondrial side to exert its antiapoptotic function. At the mitochondria, Bcl-2 binds Bax/Bak, preventing their oligomerization and inhibiting Bax/Bak pore formation (for a review, see). At the ER, Bcl-2 modulates Ca2+ transfer by decreasing net influx into the ER through increased Ca2+ leakage rather than reduced Ca2+ release. Increased Ca2+ leakage results in reduced Ca2+ transfer to mitochondria and inhibits apoptosis. Nevertheless, Bax/Bak double knockout cells were shown to have a reduced steady state of ER Ca2+, and hence are protected from a variety of apoptotic challenges. However, an Akt-like function (i.e., inhibition of ER Ca2+ release), rather than augmented Ca2+ leakage, has also been suggested for Bcl-2,49, as the antiapoptotic Ca2+ effect of Bcl-2 might be due to direct interaction of its BH4 domain with IP3R. In either case, Bcl-2 promotes survival in multiple cellular environments by limiting Ca2+ transfer from the ER to the mitochondria.

Interestingly, Bcl-2 expression, but not stabilization, is significantly regulated by the MAM protein sigma-1 receptor (Sig1-R). As a molecular chaperone, Sig1-R interacts with many effectors at MAM, but not with Bcl-2. Conversely, Sig1-R transcriptionally controls the expression of Bcl-2 by regulating the ROS/NF-κB pathway. Of note, the Sig1-R-mediated downregulation of bcl-2 mRNA is abolished by ROS scavengers and by the inhibition of NF-κB. Thus, Sig1-R affects cell survival through the regulation of bcl-2 levels, revealing that increased MAM activities during stress conditions can directly impact the response at nuclear level.

Sig-1R is a molecular chaperone that stabilizes the conformation of proteins at the MAM, such as IP3R3 or the ER stress sensor IRE1. Sig1-R has been implicated in several human diseases, and one of its most important functions is its robust cellular protective effect. Sig1-R agonists have been shown to promote cellular survival by preventing oxidative stress caused by multiple pathological conditions. Moreover, Sig1-R plays a crucial role in the control of ER-mitochondrial interorganelle Ca2+ signaling. Sig1-R at the MAM forms a complex with GRP78 (also known as Bip) to regulate Ca2+ homeostasis between the ER and the mitochondria through IP3R. GRP78 is primarily located in the ER lumen, but under ER stress a significant pool of GRP78 is localized in different subcellular compartments, such as the cytosol and mitochondria. A recent study reported the association between GRP78 and clusterin (CLU), a stress-induced multifunctional secreted and cytoplasmic molecular chaperone, under ER stress conditions. This interaction elicits CLU redistribution on the mitochondria, promoting survival of prostate cancer cells during treatment stress.

At ER-mitochondria contact sites, a Bcl-2-like activity has been described for other antiapoptotic members of the family. Bcl-xL interacts with IP3Rs and sensitizes them to low IP3 concentrations, thus reducing ER Ca2+ concentrations, stimulating mitochondrial energy, and preserving survival. The same molecular mechanism is shared by myeloid cell leukemia sequence 1 (Bcl-2-related) (Mcl-1), providing additional evidence for the crucial role of Ca2+ leakage in the prosurvival functions of the antiapoptotic Bcl-2 family subgroup. The localization of Bcl-xL at ER and mitochondria indicates its distinct roles in cell death and Ca2+ homeostasis. Using cell lines derived from ER- and mitochondria-targeted Bcl-xL chimeras that are deficient for Bcl-xL, Li and co-workers showed that ER-targeted Bcl-xL is required to restore Ca2+ homeostasis in knockout cells, whereas mitochondrial localization alone is sufficient to provide protection. Recently, the Bcl-xL activity at ER has been linked to a specific form of the Ras oncoprotein, K-Ras4B. Phosphorylation of K-Ras4B by protein kinase C promotes its translocation from the plasma membrane to the ER and OMM, which is associated with induction of the cell death pathway. Phospho–K-Ras4B associates with IP3R, limiting the ability of Bcl-xL to sensitize IP3R to the activity of its ligand IP3 and thereby abolishing its typical antiapoptotic role.

Interestingly, our group has shown that oncogenic H-Ras (H-Ras12v) is localized at both MAM and plasma membrane-associated membranes, suggesting a cooperation between the plasma membrane, the ER, and the mitochondria that is essential for Ca2+ signaling and the maintenance of Ca2+ homeostasis in cancer progression. Moreover, oncogenic K-Ras inhibits Ca2+ release from ER, reduces ER Ca2+ levels, and suppresses Ca2+ influx to the mitochondria in colon cancer cell lines. Thus, multiple forms of Ras act at the ER–mitochondria interface to manipulate Ca2+ transfer, which in turn contributes to the prosurvival properties of Ras that are associated with the oncogenic phenotype.

In addition to its interaction with cancer-related proteins, the oncogenic functions of MAM extend to some viruses. Human cytomegalovirus encodes multiple antiapoptotic proteins, including viral mitochondrion-localized inhibitor of apoptosis (vMIA). vMIA resides at the MAM, where it re-targets proapoptotic Bax to the ER/MAM lipid drafts to induce Bax poly-ubiquitination, proteasome degradation, and consequent inhibition of apoptosis.

Human hepatitis B virus (HBV) is associated with chronic liver disease and with the development of hepatocellular carcinoma. HBV encodes the regulatory protein HBx, which localizes at the OMM and interacts with the MAM protein VDAC3. HBV/HBx induces mitochondrial fragmentation and mitophagy (the selective autophagic-removal of damaged mitochondria), leading to apoptosis attenuation and most likely viral persistence.

Lastly, enteroviruses, such as coxsackievirus, poliovirus, and echovirus, confer an antiapoptotic state that not only suppresses the host defense mechanisms but is also protective against cell death induced by pharmacological treatments. The enterovirus 2B protein is localized to the surface of the ER- and Golgi-derived membrane vesicles where viral replication takes place. Expression of 2B protein decreases the steady-state Ca2+ levels of both the ER and Golgi, reducing the mitochondrial Ca2+ content and suppressing caspase activation and apoptotic cell death induced by various stimuli.

MAM as a strategic platform for oncosuppressor-dependent cell death

In recent years it has been demonstrated that many oncosuppressor proteins are localized to the ER and at MAM (), where they modulate different cell death programs.

As described above, different antiapoptotic members of the Bcl-2 family appear to play an important role in modulating Ca2+-dependent apoptotic signals within the mitochondria at the ER and MAM side. In contrast, various proapoptotic members of the same family exert an opposite effect on the ER Ca2+ stores, and thus on the amplitude of Ca2+ signals reaching the key effector during apoptosis, the mitochondria.

In the first phase of Bax protein overexpression, before the catastrophic changes in mitochondrial and ER morphology and other intracellular parameters, there is an increase in ER [Ca2+] levels. This higher ER [Ca2+] content correlates with an increase in mitochondrial Ca2+ loading after activation by stimuli causing the release of Ca2+ from the ER Ca2+ stores. As discussed above, these results agree with findings from Bax/Bak knockout embryonic fibroblasts, in which a dramatic reduction in [Ca2+]er, was observed.48 Bcl-2 and Bax/Bak proteins primarily target the IP3R type 1 to affect [Ca2+]er. Indeed, downregulation of IP3R1 counteracted the reduction of [Ca2+]er in cells from Bax/Bak knockout animals, which is consistent with the correlation between low expression of IP3R and inhibition of apoptosis. These data indicate that Bax/Bak directly counteract the effect of Bcl-2 on Ca2+ signaling.

Recently, another proapoptotic member of the Bcl-2 family, Bcl-2-related ovarian killer (Bok) has recently been described as localized to the ER and the MAM. Bok has been shown to selectively interact with Mcl-1 and BFL-1/A1, but not with Bcl-2 or Bcl-xL. Although Bok shares high sequence similarity with Bax and Bak, Bok is not a surrogate of these proteins and is unable to compensate for the combined loss of BAX and BAK in triggering mPTP opening and cell death via apoptosis. In contrast, Bok-induced apoptosis is almost blunted in the absence of Bax or Bak, and cells lacking Mcl-1 are significantly more sensitive to Bok-induced apoptosis than control cells. These observations suggest that Bok may function upstream of Bax and Bak in the control of the transmission of ER/MAM-derived apoptotic signals toward mitochondria. Furthermore, it has been recently shown that Bok can interact with IP3R channels, affecting IP3R levels by protecting the proteins from proteolytic degradation without modifying their ability to release ER calcium stores. However, the role of Bok in regulation of the apoptotic process remains to be elucidated, especially considering recent results obtained in vivo that describe a minimal impact of loss of Bok in mice. Indeed, Bok deficiency in lymphoid and myeloid cells fails to confer any protection against a wide range of apoptotic stimuli and loss of Bok does not accelerate lymphoma development in c-MYC-overexpressing transgenic mice.

Involvement of the MAM domain is also important in cell death independent of the Bcl-2 family members. Indeed, procaspase-activating compound-1 (PAC-1) (a small molecule that converts procaspase-3 to caspase-3 in cancer cells in vitro and in vivo) does not require Bax and Bak but is dependent on the engagement of MAM through ER oxidoreductin-1 α (Ero1-α). Ero1-α, a key controller of oxidative folding and ER redox homeostasis, is enriched at MAM and regulates Ca2+ fluxes. The efficacy of PAC-1 in activating caspase-3 requires cytochrome c release from the mitochondria, which is induced by mitochondrial Ca2+ overload and an increase in mitochondrial ROS through ER stress that is mediated by p53 upregulated modulator of apoptosis (PUMA) and Ero1-α. ER stress induces ER Ca2+ release that is preferentially transferred into the mitochondria because PAC-1 treatment leads to an increase in the number of MAM via the upregulation of Ero1-α.

Additionally, the MAM is an important site for the recruitment and processing of procaspase-8 to caspase-8. The proteins participating in this complex are B-cell-receptor-associated protein 31 (Bap31) at the ER and Fission 1 homolog (Fis1) at the mitochondria. The physical interaction between Bap31 and Fis1 provides a tethering force between the ER and the mitochondria, thus facilitating the MAM structure. During the apoptotic program Bap31 is cleaved by caspases. This processing occurs at the MAM with the formation of a tripartite protein complex between procaspase-8, Bap31, and Fis1. The key role of the caspase-activated Bap31–Fis1 complex at the MAM is to transduce the cell death signals between the ER and mitochondria. Indeed, Bap31 cleavage (i.e., activation) results in a downstream increase in cytosolic [Ca2+] caused by ER Ca2+ release, leading to mitochondrial Ca2+ uptake. This dyshomeostasis of mitochondrial [Ca2+] will trigger PTP opening, causing release of the mitochondrial cofactors into the cytosol to complete the apoptotic process.

Thus, the MAM appears to be a hotspot domain for decoding different intracellular signals of stress. In light of this role, it is not surprising that the p66shc protein (an important ROS sensor involved in apoptosis) and the RNA-dependent protein kinase (PKR)-like ER kinase (PERK) (the ER stress sensor of the unfolded protein response) are present at the MAM. The PERK protein is particularly enriched at the MAM, where it promotes efficient crosstalk between the ER and the mitochondria based on the transfer of Ca2+ and ROS. PERK appears to be crucial for tethering the ER to the mitochondria and thus for MAM integrity. Indeed, cells deficient in PERK display a fragmented ER structure, causing impairment in the Ca2+ signaling from the ER to the mitochondria that is required for efficient apoptosis. Moreover, destabilization of MAM in the absence of PERK prevents the rapid transfer of ROS from the ER to the mitochondria that is required for the insertion of Bax into the outer mitochondrial membrane and release of cytochrome c from mitochondria via oxidation of the phospholipid cardiolipin.

In recent years, the increased interest in MAMs, in particular their emerging role in controlling cell death, has led to in-depth analysis of other tumor suppressors that share this intracellular localization.

The promyelocytic leukemia (PML) protein, which is encoded by a tumor suppressor gene implicated in the pathogenesis of leukemia and other cancers, displays both nuclear and cytosolic distribution. At the nucleus, PML forms multiprotein nuclear structures called PML-nuclear bodies (PML-NBs). In the cytosol, PML is associated with endosomes and is present at the ER and MAM. Studies in knockout mice and cells revealed an essential pleiotropic role for PML in multiple p53-dependent and -independent apoptotic pathways. As a result, PML-null mice and cells are protected from apoptosis triggered by a number of stimuli. We were able to demonstrate that the ER/MAM localization of PML is essential for the apoptotic pathway and orchestrating Ca2+ homeostasis and, eventually, for cell death. In particular, PML modulates IP3R type 3 activities by promoting the formation of a multiprotein complex containing IP3R type 3, Akt, and the protein phosphatase PP2a. In the absence of PML, PP2a is unable to localize to the MAM and thus unable to prevent the IP3R3 phosphorylation mediated by Akt. As described above, hyperphosphorylation of IP3R3 inhibits Ca2+ transfer from the ER to mitochondria, thereby inhibiting the apoptotic process.

The tumor suppressor phosphatase and tensin homolog deleted on chromosome 10 (PTEN) also localizes to MAM. PTEN is one of the most commonly lost or mutated tumor suppressors in human cancers, and germline mutations of PTEN have been found in cancer predisposition syndromes. PTEN is a phosphatase that has both lipid and dual-specificity protein phosphatase activity. Its growth-attenuating activity has primarily been ascribed to the dephosphorylation of plasma membrane-localized PIP3. However, emerging evidence demonstrates that additional PTEN-dependent mechanisms are implicated in tumor suppression. Indeed, PTEN at the MAM regulates ER Ca2+ release via IP3R3 in a protein phosphatase-dependent manner that counteracts Akt phosphorylation of the IP3R3. The final result of the action of PTEN on IP3Rs is thus the maintenance of sustained activity of the IP3R3 during the apoptotic stimulation and, consequently, enhanced transfer of Ca2+ from the ER to mitochondria.

Lastly, consistent with the importance of sustained Ca2+ transfer from the ER to the mitochondria for the induction of apoptosis, it has been demonstrated that overexpression of the sarco/endoplasmatic reticulum Ca2+ ATPase (SERCA), a protein that is enriched at the MAM, causes ER [Ca+] overload, which increases spontaneous and induced apoptosis, favoring ER-mitochondria Ca2+ transfer and leading to a breakdown of mitochondrial function.

In conclusion, data emerging from recent literature highlight the MAM as an important hub for the control and integration of apoptosis operated by different oncosuppressors via different signal transduction mechanisms that in many cases share the same Ca2+ signal key effectors. The MAM can be conceived as hotspot sites for Ca2+ homeostasis,5 and any perturbation of the fine regulation of Ca2+ signaling can induce tumor suppressors to switch from a survival role to a cell death mechanism.

Conclusion

The large body of knowledge reviewed here depicts MAM as a regulatory scaffold for the control of cell death. Many factors that localize to the nucleus modulate apoptosis at a transcriptional level but can also reside or translocate to the MAM under stress conditions. Evidence for this role is the reciprocal transmission of different signals between the ER and mitochondria through physical contact. Additionally, MAM may be considered the primary platform for the detection of intracellular danger. We have also reported that manipulation of Ca2+ fluxes represents the primary method of action for several oncogenes and tumor suppressors located at the MAM. IP3R channels, including IP3Rs present at MAM, represent the key targets of different oncogenic and oncosuppressive proteins. Indeed, regulation of Ca2+ flux through IP3R is also mediated by ROS; superoxide anions cause oxidation of the IP3R and sensitization of Ca2+ release to promote cytoplasmic Ca2+ oscillations and mitochondrial uptake.

In addition to Ca2+ input, the MAM provides specialized contact sites for transmitting ROS-mediated signals. In this regard, the role of the MAM protein Sig1-R as an interorganelle signaling modulator has recently been reported, providing a new mechanism whereby MAM controls cell fate by conveying the ROS message from the mitochondria to the nucleus.

Characterization of MAM as the main decoder of intracellular danger signals implicates a role for MAM dynamics in several physiopathological scenarios.