Arrestins in Apoptosis

Abstract

Programmed cell death (apoptosis) is a coordinated set of events eventually leading to the massive activation of specialized proteases (caspases) that cleave numerous substrates, orchestrating fairly uniform biochemical changes than culminate in cellular suicide. Apoptosis can be triggered by a variety of stimuli, from external signals or growth factor withdrawal to intracellular conditions, such as DNA damage or ER stress. Arrestins regulate many signaling cascades involved in life-or-death decisions in the cell, so it is hardly surprising that numerous reports document the effects of ubiquitous nonvisual arrestins on apoptosis under various conditions. Although these findings hardly constitute a coherent picture, with the same arrestin subtypes, sometimes via the same signaling pathways, reported to promote or inhibit cell death, this might reflect real differences in pro- and antiapoptotic signaling in different cells under a variety of conditions. Recent finding suggests that one of the nonvisual subtypes, arrestin-2, is specifically cleaved by caspases. Generated fragment actively participates in the core mechanism of apoptosis: it assists another product of caspase activity, tBID, in releasing cytochrome C from mitochondria. This is the point of no return in committing vertebrate cells to death, and the aspartate where caspases cleave arrestin-2 is evolutionary conserved in vertebrate, but not in invertebrate arrestins. In contrast to wild-type arrestin-2, its caspase-resistant mutant does not facilitate cell death.

1 Apoptotic Pathways

Apoptosis is a form of programmed cell death (Vaux et al. 1994; Steller 1995) involving the activation of caspases (Thornberry and Lazebnik 1998; Crawford and Wells 2011). Caspases concentrate on key pathways, producing stereotypic morphological and biochemical changes in apoptotic cells (Dix et al. 2008; Mahrus et al. 2008; Chipuk et al. 2010). Apoptosis can be triggered by a number of factors including signaling via death receptors (DR); DNA damage by UV, γ-irradiation, or genotoxic drugs; load on endoplasmic reticulum (ER); withdrawal of growth or trophic factors; oxidative stress; and a large number of other factors.

The apoptotic pathway initiated by DR activation is known as extrinsic pathway (Fig. 1). Eight members of the DR family have been described, including the best-studied tumor necrosis factor receptor alpha 1 (TNFR1), Fas (also known as CD95), TNF-related apoptosis-inducing ligand receptor 1 (TRAILR1), and TRAILR2 (Lavrik et al. 2005). The activation of DR by corresponding ligands leads to the formation of signaling complexes assembled at the intracellular surface of the receptor. Fas and TRAILR1/2 recruit what is known as death-inducing signaling complexes (DISC) containing FADD (Fas-Associated Death Domain), pro-caspase-8, and the long and short forms of the cellular FLICE inhibitory protein (FLIP_L/S) as main components (Lavrik et al. 2005; Guicciardi and Gores 2009). The formation and internalization of DISC result in the activation of initiator caspase-8 that cleaves and activates effector caspase-3, caspase-6, and caspase-7, triggering apoptotic cell death (Lavrik et al. 2005; Guicciardi and Gores 2009). In cell type I, DISC is effectively internalized, resulting in massive caspase-8 activation sufficient to activate downstream caspases (Scaffidi et al. 1998, 1999; Fulda et al. 2001). In cell type II, lower level of DISC formation results in weak caspase-8 activation that requires amplification to trigger apoptosis. The amplification cascade includes capspase-8-mediated cleavage of the member of the BCL-2 family BID to yield truncated BID (tBID) that translocates to the mitochondria inducing the cytochrome C release (Li et al. 1998; Luo et al. 1998) via a still poorly understood mechanism (Chipuk and Green 2008). The cytochrome C release results in the formation of the apoptosome and the activation of caspase-9, which in turn cleaves and activates the effector caspases (Danial and Korsmeyer 2004; Bratton and Salvesen 2010). In type II cells, Fas-induced apoptosis could be blocked by antiapoptotic BCL family members such as BCL-2 and BCL-XL (Scaffidi et al. 1998, 1999; Fulda et al. 2001). Therefore, tBID serves as a mediator of the positive feedback, or amplification, loop involving the mitochondria-dependent apoptotic signaling.

Fig. 1.

Numerous arrestin functions play a role in apoptosis and cell survival. The activation of the death receptor TNFR1 by TNFα results in the assembly of the multi-protein complex I that activates the NFκB pathway along with JNK and p38 pathways. The NFκB signaling is antiapoptotic, mostly via transcriptional activation of FLIP that inhibits caspase-8 activation. Activation of JNK and p38 promotes apoptosis via transcriptional as well as posttranscriptional mechanisms (see text). Signaling complex II formed following internalization and reshuffling of signaling proteins directly activates caspase-8 that cleaves and activates effector caspase-3, caspase-6, and caspase-7 initiating apoptosis. Apoptotic pathway induced via death receptor is known as the extrinsic pathway. Caspase-8 also cleaves BID, generating tBID that translocates to the mitochondria inducing BAX-BAK oligomerization and cytochrome C release. Cytochrome C organizes apoptosome, activating caspase-9, which then activates massive amounts of caspase-3. Apoptosis can also be initiated by a variety of stress stimuli that engage the mitochondria-based apoptotic pathway termed the intrinsic pathway. Specifically, genotoxic drugs and other stimuli causing DNA damage initiate the intrinsic pathway via transcription factor p53 that upregulates pro-apoptotic genes BAX, PUMA, and NOXA. Arrestins have been shown to promote activation of MAP kinases by scaffolding (Chapters 12–14). GPCR-dependent activation of ERK1/2 by both nonvisual arrestins has been shown to provide protection against apoptosis induced by various agents in many cell types. Arrestin-3, but not arrestin-2, is able to activate neuron-specific JNK3 isoform. Although JNK3 has been shown to play an important role in neuronal apoptosis, the evidence of the role of arrestin-3-dependent JNK3 activation in neuronal death is so far lacking. Both arrestin isoforms interact with IκBα, an inhibitory protein that binds NFκB and holds it in the cytosol, preventing NFκB-dependent antiapoptotic transcription. Arrestins have been demonstrated to regulate the activity of the pro-survival Akt pathway. Arrestin-3 is able to reduce the Akt activity via scaffolding Akt together with PP2A that dephosphorylates Akt in response to D2 dopamine receptor stimulation. Arrestins have been shown to stabilize Mdm2 and promote its activation, as well as affect p53 degradation and level (see text for details). Arrestin-2 also interacts with p53 in the nucleus, acting as important adaptor for Mdm2 required for Mdm2-dependent p53 degradation. Interestingly, caspase-8 that cleaves BID also cleaves arrestin-2 at the C-terminus generating Arr2-(1-380) fragment. Arr2-(1-380) translocates to the mitochondria, directly binds tBID, and greatly enhances its ability to induce cytochrome C release from mitochondria, thereby promoting apoptosis. Black arrows indicate direct or indirect posttranslational activation; black bar—inhibitory modification; white arrows—transcriptional upregulation; dotted line—translocation. Abbreviations: TNFR1 TNFα receptor 1, RIPK1 receptor-interacting serine/threonine-protein kinase 1, FADD Fas-associated death domain protein, TRADD TNF receptor-associated death domain (TRADD), TRAF TNF receptor-associated factor, FLIP FLICE-like inhibitory protein (a.k.a. CFLAR, CASP8, and FADD-like apoptosis regulator), DD death domain, DED death effector domain, RTK receptor tyrosine kinase, GPCR G protein-coupled receptor

Stimulation of TNFR1 and similar DR results in the formation of two signaling complexes. Complex I assembled at the membrane includes TRADD (TNFR-associated death domain protein), RIPK1 (receptor-interacting serine/threonine-protein kinase 1), and TRAF2/5 (TNFR-associated factor) as main components (Micheau and Tschopp 2003; Lavrik et al. 2005). Complex I mediates TNFR1-induced activation of the NFκB and JNK pathways (Dempsey et al. 2003; Lavrik et al. 2005). The NFκB pathway is activated via recruitment of the IKK complex in the TRADD-dependent manner with participation of RIPK1 and TRAF2/5/6 through a series of K63 “nondestructive” ubiquitination events (Micheau and Tschopp 2003; Ea et al. 2006; O'Donnell and Ting 2010; Pobezinskaya and Liu 2012). Recruitment of the IKK complex leads to the phosphorylation of the NFκB inhibitory protein IκBα, with its subsequent degradation, and activation of NFκB-dependent transcription of antiapoptotic genes such as cFLIP, cIAP1, cIAP2, BCL-XL, and XIAP (Kreuz et al. 2001; Micheau et al. 2001; Dempsey et al. 2003; Chipuk et al. 2010). JNK activation by TNFR1 is TRAF2 dependent (Natoli et al. 1997; Reinhard et al. 1997; Yuasa et al. 1998; Habelhah et al. 2004). TNFR1 also activates the p38 pathway in a TRAF2- and RIPK1-dependent manner (Yuasa et al. 1998; Lee et al. 2003). The MAP kinase pathways are activated via recruitment and activation of upstream kinases MEKK1, ASK1, or TPL2 to TRAF2 (Nishitoh et al. 1998; Yuasa et al. 1998; Das et al. 2005) (see also chapters “Arrestin-Dependent Activation of ERK and Src Family Kinases,” “Arrestin-Dependent Activation of JNK Family Kinases,” and “Arrestin-Mediated Activation of p38 MAPK: Molecular Mechanisms and Behavioral Consequences”).

Complex I is internalized and transformed in the cytosol into complex II by exchange of signaling proteins associated with TNFR1. FADD and pro-caspase-8 are recruited, leading to caspase-8 activation and initiation of apoptosis (Micheau and Tschopp 2003; Schneider-Brachert et al. 2004). Unlike Fas and TRAIL receptors, TNFR1 is mostly involved in mediating inflammation and not cell death, and the outcome of the TNFR1 stimulation is cell type dependent. Inhibition of RNA or protein synthesis resulting in the blockade of complex I-mediated pro-survival NFκB-mediated signaling is required to induce apoptosis via TNFR1 stimulation in most cell types. Blockade of NFκB signaling promotes TNFR1-induced apoptosis mostly by blocking the synthesis of cFLIP that inhibits caspase-8 activation (Kreuz et al. 2001; Micheau et al. 2001). Alternatively, TNFR1 signaling could be switched from pro-survival to pro-apoptotic mode by Smac, also known as Diablo (or its mimetics). Smac is a protein released from the mitochondria together with cytochrome C that interacts with and inhibits apoptotic inhibitors XIAP, cIAP1, and cIAP2 (Chai et al. 2000; Du et al. 2000). Smac can also trigger RIPK1-dependent mode of capsase-8 activation by promoting degeneration of IAPs (Wang et al. 2008). The positive regulation of TNFR1 apoptotic signaling by Smac/Diablo released from the mitochondria is another mitochondria-based amplification pro-apoptotic mechanism.

The apoptotic pathway mediated by the release of pro-apoptotic factors from the mitochondria followed by the formation of apoptosome, activation of initiator caspase-9, and subsequent activation of effector caspases is referred to as the intrinsic pathway (Danial and Korsmeyer 2004) (Fig. 1). The intrinsic apoptotic pathway is triggered by a large variety of stimuli including DNA damage, withdrawal of growth factors, hypoxia, or endoplasmic reticulum stress. The signaling converges on the mitochondria where the interplay of pro- and anti-apoptotic BCL family members regulates cytochrome C release, although the exact biochemical mechanism of this process has not been elucidated (Danial and Korsmeyer 2004; Youle and Strasser 2008). Effectors BAK and BAX oligomerize and form pores in the outer mitochondrial membrane (Wei et al. 2000), allowing cytochrome C (and other mitochondrial proteins such as Smac/Diablo) to escape to the cytoplasm (Lindsten et al. 2000; Wei et al. 2001). The biochemical nature of this pore and the number of BAK or BAX proteins necessary to create it remain unknown (Youle and Strasser 2008). BCL-2 homology 3 (BH3)-only members of the BCL-2 family either directly activate BAX and BAK and induce cytochrome C release or do so indirectly via antagonistic interaction with antiapoptotic members of the same family. It is still unclear how exactly the interactions of pro- or anti-apoptotic BCL-2 proteins with BAK and BAX affect the pore formation process (Chipuk and Green 2008; Youle and Strasser 2008). Truncated BID (tBID) cleaved by caspase-8 activated in the extrinsic apoptotic pathway is the most potent cytochrome C releaser among BH3 proteins (Korsmeyer et al. 2000; Wei et al. 2000; Lovell et al. 2008), providing a strong amplification signal for apoptosis induced by DR activation. The release of cytochrome C from mitochondria promotes the formation of a structure known as apoptosome composed of cytochrome C, pro-caspase-9, and apoptotic protease activation factor 1 (Apaf 1) that results in capsape-9 activation (Danial and Korsmeyer 2004; Bratton and Salvesen 2010). Active caspase-9 cleaves and activates effector caspases. Thus, the apoptosome serves a function analogous to that of DISC, i.e., activation of an initiator caspase, albeit achieved via a different molecular mechanism. The apoptosome formation that promotes massive activation of executioner caspase-3/caspase-6/caspase-7 is the key checkpoint in cell commitment to death (Youle and Strasser 2008). Stressful stimuli capable of engaging the intrinsic apoptotic pathway do so by increasing the expression of pro-apoptotic proteins or via direct inactivation of antiapoptotic proteins. Thus, DNA damage caused by UV, γ-irradiation, or genotoxic drugs results in stabilization of the transcription factor p53 that translocates to the nucleus promoting the expression of multiple pro-apoptotic genes and subsequent cell death (Oda et al. 2000; Nakano and Vousden 2001; Yu et al. 2001; Villunger et al. 2003; Naik et al. 2007; Michalak et al. 2008). Growth factors such as NGF and BDNF can induce neuronal death acting via p75 receptor, which is a member of the death receptor superfamily that includes Fas and TNFR1 (Chao 1994; Sessler et al. 2013). The mechanisms of this apoptosis are poorly understood but appear to involve JNK activation and downstream engagement of the mitochondrial pathway (Harrington et al. 2002; Salehi et al. 2002; Nykjaer et al. 2005; Ichim et al. 2012; Sessler et al. 2013). Some trophic factor receptors seem to function as so-called dependence receptors that induce positive apoptotic signaling in the absence of ligands. The survival of certain types of cells in culture is strictly dependent on the presence of specific trophic factors, and apoptosis of many cultured cell lines can be induced by serum withdrawal. Apoptosis mediated by dependence receptors involves caspase interaction and caspase-mediated cleavage of the receptor cytoplasmic domain yielding a pro-apoptotic peptide that mediates downstream signaling including transcriptional activation of pro-apoptotic genes (Rabizadeh and Bredesen 2003; Bredesen et al. 2005; Goldschneider and Mehlen 2010; Ichim et al. 2012). Thus, multiple cellular signaling pathways impact the core apoptotic machinery, thereby affecting cell death and survival.

2 Signaling Mechanisms in Apoptosis

Given the irreversible nature of apoptosis, it is not surprising that large number of checks and balances is incorporated into the core apoptotic mechanisms. Furthermore, multiple cellular signaling pathways impact the function of most proteins involved in apoptosis. A good example is TNFR1: its activation triggers anti- and pro-apoptotic mechanisms, such as activation of the NFκB and JNK pathways. The NFκB pathway suppresses apoptosis via transcriptional upregulation of antiapoptotic genes (Dempsey et al. 2003; Lavrik et al. 2005; Chipuk et al. 2010). The activity of the NFκB pathway is regulated by inhibitory protein IκBα that keeps NFκB inactive in the cytoplasm. IκBα is phosphorylated by the IKK complex, which induces its polyubiquitination and proteasomal degradation (Chen et al. 1996; Roff et al. 1996; Napetschnig and Wu 2013). NFκB, thus released, translocates to the nucleus and activates transcription (Napetschnig and Wu 2013). The IKK complex is composed of two related catalytic subunits, IKKα and IKKβ, and an important although catalytically inactive component NEMO/IKKγ (Chen et al. 1996; DiDonato et al. 1997; Mercurio et al. 1997; Yamaoka et al. 1998). IKKβ phosphorylation is required for the NFκB activation via so-called canonical pathway (turned on by TNFα), and it appears that TAK1, which also serves as an MAPKKK in the JNK pathway, can phosphorylate IKKβ in the activation loop (Ninomiya-Tsuji et al. 1999; Wang et al. 2001; Sato et al. 2005). MEKK3, another upstream MAP kinase, has also been implicated in TNFα-induced IKK activation (Yang et al. 2001). NEMO specifically binds to linear and K63 polyubiquitin chains, which is critical for the activation of the TNFα-induced IKK recruitment and NFκB activation (see Napetschnig and Wu (2013) and references therein). The IKK complex could be activated by receptor belonging to Toll-like-interleukin-1 receptor superfamily involved in the innate immunity responses via recruitment of TRAF6 (Bradley and Pober 2001).

2.1 The JNK Pathway

Active NFκB leads to a rapid quenching of TNFR1-induced JNK activation. The proposed mechanisms of NFκB-induced suppression of JNK activity include upregulation of Gadd45 beta factor that inhibits MKK7, an upstream kinase activating JNK (De Smaele et al. 2001; Papa et al. 2004), and upregulation of XIAP (Tang et al. 2001). However, cells lacking Gadd45 beta or XIAP showed TNFR1-induced JNK activation similar to that of wild type (Amanullah et al. 2003; Kucharczak et al. 2003). An alternative mechanism involves TNFα-generated reactive oxygen species that inhibit JNK phosphatases (Kamata et al. 2005), which normally ensure low level of JNK activity (Cavigelli et al. 1996). Several studies reported that sustained JNK activation augments TNFR1-induced death in cells with deficient NFκB pathway (De Smaele et al. 2001; Tang et al. 2001). Therefore, JNK activity could play a decisive role in the outcome of the TNFR1 activation if the function of the NFκB pathway is compromised due to genetic defects or drug action. JNK activation plays the key role in apoptosis induced by UV irradiation and genotoxic drugs mediated by the intrinsic apoptotic pathway (Zanke et al. 1996; Tournier et al. 2000). JNK3, a JNK isoform selectively expressed in neurons, has been shown to be involved in apoptosis caused by excitotoxic (Yang et al. 1997) or other toxic (Namgung and Xia 2000) agents and by growth factor deprivation (Bruckner et al. 2001; Eilers et al. 2001; Coffey et al. 2002; Barone et al. 2008; Ambacher et al. 2012). JNK activation plays an important role in neuronal apoptosis following focal ischemia (Okuno et al. 2004; Gao et al. 2005) and in beta-amyloid-induced neuronal apoptosis (Morishima et al. 2001; Yao et al. 2005).

It appears that sustained JNK activation is required to promote apoptosis, whereas transient JNK activity is involved in cell proliferation and survival (Sánchez-Perez et al. 1998; Chen and Tan 2000; Dhanasekaran and Reddy 2008). The pro-apoptotic action of JNK is in many cases transcriptional, mediated by JNK-dependent phosphorylation and transactivation of the transcription factor c-jun and subsequent expression of pro-apoptotic genes (Behrens et al. 1999; Coffey et al. 2002; Barone et al. 2008; Dhanasekaran and Reddy 2008). Ironically, the nature of genes induced by JNK activation has never been extensively defined. One gene proposed to be transcriptionally activated by JNK and involved in apoptosis was Fas ligand (Le-Niculescu et al. 1999; Mansouri et al. 2003; Wang et al. 2004). Sustained JNK activation may promote TNFα-induced apoptosis via JNK-mediated activation of E3 ubiquitin ligase Itch that ubiquitinates cFLIP, leading to its proteasomal degradation and, subsequently, enhanced caspase-8 activation (Chang et al. 2006). JNK activation can also lead to caspase-8-independent cleavage of BID at a different site, and the cleaved product, jBId, translocated to the mitochondria, inducing preferential release of Smac/Diablo; this, in its turn, promotes TNFα-dependent apoptosis by disrupting the TRAF2-cIAP1 interaction inhibitory for caspase-8 activation (Deng et al. 2003). The JNK activity has been shown to affect the p53-dependent apoptosis in different cell types via p53 phosphorylation that alters the activity or stability (Fogarty et al. 2003; Oleinik et al. 2007). JNK is also known to phosphorylate members of the BCL-2 family, thus directly affecting their function (Yamamoto et al. 1999; Donovan et al. 2002; Lei and Davis 2003; Putcha et al. 2003; Okuno et al. 2004). JNK can also alter their functions indirectly by phosphorylating interacting proteins. The best known such effect is translocation of BAX to the mitochondria promoted by JNK-dependent phosphorylation of BAX cytoplasmic anchoring protein 14-3-3 (Tsuruta et al. 2004; Gao et al. 2005). The function of neuron-specific JNK3 isoform in neuronal apoptosis caused by ischemia/hypoxia is believed to be mediated by induction of BIM and other pro-apoptotic genes (Kuan et al. 2003; Zhang et al. 2006; Zhao et al. 2007).

2.2 The p53 Pathway

The tumor suppressor protein p53 is a transcription factor that mediates apoptosis caused by multiple stressors, including DNA damaging agents such as UV, γ-irradiation, or genotoxic drugs (e.g., topoisomerase inhibitor etoposide) (Vousden and Lane 2007; Delbridge et al. 2012). As mentioned above, p53 promotes cell death by increasing the expression of pro-apoptotic genes such as BAX, PUMA, and NOXA (Oda et al. 2000; Nakano and Vousden 2001; Yu et al. 2001), with PUMA and, to a lesser degree, NOXA being the main culprits (Villunger et al. 2003; Naik et al. 2007; Michalak et al. 2008). The level of p53 in cells is tightly controlled to keep the balance between cell death and tumor development that occurs when p53 function is compromised (Delbridge et al. 2012). Oncoprotein RING finger E3 ubiquitin ligase Mdm2 is the main negative regulator of p53 that ubiquitinates p53, promoting its proteasomal degradation (Fang et al. 2000; Honda and Yasuda 2000). Apparently, Mdm2 requires collaboration with MdmX, a related protein without intrinsic E3 ligase activity, to polyubiquitinate p53 (Parant et al. 2001; Wang et al. 2011). In its turn, p53 stimulates Mdm2 transcription. Thus, Mdm2 and p53 form a regulatory feedback loop that is strongly impacted by cellular stress, resulting in inactivation of Mdm2 and activation of p53 (Stommel and Wahl 2005). The key importance of p53 for survival and Mdm2 for its regulation is strongly supported by the fact that mice lacking Mdm2 die in early embryogenesis, whereas mice lacking both Mdm2 and p53 are grossly normal (Jones et al. 1995; Montes de Oca Luna et al. 1995). Recent data demonstrate that, in addition to its transcriptional role, p53 regulates the mitochondrial apoptotic pathway in a transcription-independent manner via direct interaction with BCL-2 proteins at the mitochondria (Chipuk et al. 2003, 2004, 2005).

3 Arrestins Regulate Apoptosis via Signaling Mechanisms

The canonical mode of arrestin function in homologous desensitization of GPCRs involves arrestin binding to phosphorylated activated receptors that terminates G protein activation by blocking its access to the receptor cytoplasmic surface (Wilden 1995; Krupnick et al. 1997). Ubiquitous arrestin-2 and arrestin-3¹ regulate most GPCRs, suppressing G protein activation (Attramadal et al. 1992; Lohse et al. 1992). Arrestins also bind numerous non-receptor partners, thus regulating multiple cellular signaling pathways (Lefkowitz and Shenoy 2005; Gurevich and Gurevich 2006). Since many of these pathways are involved in “life-or-death” decisions in the cell, arrestins have been reported to influence cell death and survival via signaling mechanisms (Fig. 1). Indeed, considering a wide variety of signaling pathways regulated by arrestins (Gurevich and Gurevich 2006; Luttrell and Miller 2013), it would have been surprising if arrestins did not affect apoptosis. However, the data on the exact mechanisms involved are remarkably fragmentary.

GPCR stimulation followed by G protein activation can induce pro-apoptotic signaling, and arrestins would counteract that signaling simply by virtue of desensitizing the offending receptors. One example of such situation is apoptosis induced by stimulation of various GPCRs in arrestin-2/arrestin-3 DKO MEFs, with expression of either arrestin protecting these cells (Revankar et al. 2004). The molecular mechanism of GPCR-induced apoptosis, which occurs via the intrinsic pathway, includes the activation of p38, JNK, phosphatidylinositol 3-kinase (PI3K), and Gi/o-dependent signaling. Interestingly, excessive signaling due to defective receptor desensitization in the absence of arrestins does not seem to be the culprit. At least in case of the N-formyl peptide receptor, which is internalized in arrestin-independent manner but requires arrestins for recycling (Vines et al. 2003), receptor phosphorylation and internalization were required to induce apoptosis. Arrestin interaction with adaptor protein-2 (AP-2) participating in post-endocytotic receptor trafficking (see chapter “β-Arrestins and G Protein-Coupled Receptor Trafficking”) was involved in the protection from GPCR-induced apoptosis by arrestins (Wagener et al. 2009).

In an alternative case scenario, arrestins could be involved in apoptosis-related signaling by various receptors, and as recent studies indicate, their role is not limited to GPCRs. Whether the outcome of arrestin-dependent signaling is pro-survival or pro-apoptotic depends on the specific configuration of the signaling system in which they act. When arrestin-mediated signaling results in the activation of pro-survival pathways such as ERK or Akt, arrestins provide cytoprotection. In the opposite case, when the arrestin action leads to the suppression of pro-survival or induction of pro-apoptotic signaling, arrestins serve to facilitate cell death. Thus, arrestin-2 protected cells from apoptosis caused by serum deprivation via NK1 receptor and arrestin-dependent ERK1/2 activation (DeFea et al. 2000). Similarly, arrestin-2-dependent ERK activation mediated protective effect of glutamate acting via metabotropic glutamate receptor 1 (mGluR1) against serum-deprivation-induced apoptosis (Emery et al. 2010). Both arrestin-2 and arrestin-3 mediate transactivation of the epidermal growth factor (EGF) receptor by the Gq-coupled receptor of neuropeptide urotensin II (Esposito et al. 2011). Urotensin II, which is expressed in the nervous, cardiovascular, and urogenital systems, and its receptor are upregulated in the pathological heart (Zhu et al. 2006), and this increase seems to be protective, since treatment with urotensin II antagonist exacerbates heart pathology and promotes apoptosis of cardiomyocytes, the effect linked to reduced EGF receptor transactivation and resulting ERK activity (Esposito et al. 2011). Arrestin-3 mediated the protection conferred by the angiotensin II receptor 1A to primary rat vascular smooth muscle or to HEK293 cells against hydrogen peroxide (H₂O₂)- or etoposide-induced apoptosis (Ahn et al. 2009). Arrestin-3-dependent ERK activation followed by activation of the P90 ribosomal S6 kinase (P90RSK) and activation of Akt was required for its antiapoptotic activity. P90RSK and Akt in their turn phosphorylated pro-apoptotic BCL-2 protein Bad at Ser¹¹² and Ser¹³⁶, respectively, thus inhibiting its pro-apoptotic activity. A similar mechanism was described for the arrestin-2 role in the protection against glucose deprivation-induced apoptosis afforded by stimulation of the glucagon-like peptide-1 Gs-coupled receptor (GLP-1) to pancreatic beta cells (Quoyer et al. 2010). Arrestin-2-mediated ERK activation resulted in the activation of the P90RSK, leading to phosphorylation of Bad at Ser¹¹². Apparently, in some cases arrestin-dependent ERK activation could be harmful to cells. Thus, dopamine at high concentration acting at the D1 dopamine receptor has been reported to cause apoptotic death of primary and cultured neuronal cells via sustained arrestin-2-dependent ERK activation in the cytosol (Chen et al. 2004, 2009).

Arrestins are known to be involved in the regulation of the pro-survival Akt pathway. Previously, arrestin-3 (and to a lesser extent arrestin-2) has been shown to reduce the activity of the Akt pathway by scaffolding Akt with protein phosphatase 2 at the D2 dopamine receptor, which resulted in dephosphorylation of Akt at its main activating residue Thr⁴⁰⁸ (Manning and Cantley 2007). It has not been examined whether this regulatory effect of arrestin-3 on the Akt pathway activity plays any role in apoptosis, which appears likely. Arrestin-2 is protecting from serum deprivation-induced apoptosis by coupling the insulin-like growth factor 1 receptor to the activation of PI3K and subsequent activation of the Akt pathway (Povsic et al. 2003). This signaling process occurs independently of the tyrosine kinase activity of the receptor, Gi, or ERK activity. Platelet-activating factor acting at its receptor induces apoptosis in colon cancer cells by promoting dephosphorylation of Akt at Ser⁴⁷³ via assembly at the receptor of arrestin-2 and PH domain and leucine-rich repeat protein phosphatase 2 that dephosphorylates Akt at this residue (Crotty et al. 2013; Xu et al. 2013). This is a novel mode of arrestin-dependent inhibition of the Akt pathway that may play a role in apoptosis via suppression of the pro-survival Akt signaling. The glycogen synthase kinase-3 (GSK3) is the main substrate of Akt. This kinase is constitutively active, and it is inhibited by Akt phosphorylation (Manning and Cantley 2007). GSK3β isoform is known to potentiate mitochondrial apoptotic signaling (Hetman et al. 2000; Beurel and Jope 2006; Eom et al. 2007; Mishra et al. 2007; Watcharasit et al. 2008). The pro-survival effect of the PI3K/Akt activation is largely mediated by inhibition of GSK3β. Arrestins, particularly arrestin-3, by modulating the activity of PI3K and/or Akt, could impact the GSK3β-dependent apoptosis. This notion is supported by the fact that MEFs lacking arrestin-3 demonstrate higher level of GSK3β activity (lower phosphorylation) coupled with increased apoptotic death (Li et al. 2010).

Arrestins turned out to be involved in the regulation of the pro-survival NFκB pathway. Arrestin-3 has been reported to inhibit NFκB activation induced by TNFα (Gao et al. 2004). The mechanism of the inhibition involves direct interaction of arrestin-3 with the inhibitor of NFκB IκBα, which prevents phosphorylation and degradation of the latter, thus precluding the activation of NFκB. In this study, arrestin-3 was shown to significantly inhibit TNFα-induced translocation of NFκB p65 subunit into the nucleus and transcription of NFκB-dependent genes. Importantly, arrestin-3 association with IκBα, as well as its effect on the NFκB activation, was significantly increased by stimulation of β2-adrenergic receptor (b2AR). Thus, arrestin-3 mediates the effect of b2AR on the NFκB activity, which may play a role in the sympathetic regulation of TNFα immune responses. In this study, the TNF-α-induced apoptosis was not directly examined, although arrestin-3-dependent inhibition of the pro-survival signaling induced by TNFR1 should be expected to favor cell death. Interestingly, arrestin-2 was reported to be unable to stabilize IκBα and affect the p65 translocation or the expression of NFκB target genes. Another group, however, demonstrated that both arrestin isoforms interacted with IκBα, significantly inhibiting the NFκB activity induced by various stimuli (Witherow et al. 2004). Moreover, knockdown of arrestin-2 and not arrestin-3 resulted in significant increase in the TNFα-induced activation of NFκB, suggesting that arrestin-2 isoform is the prime regulator of the NFκB activation in response to TNFα. The NFκB is also activated by UV irradiation, and arrestin-3 was shown to suppress that activation via interaction with IκBα facilitating apoptotic cell death (Luan et al. 2005). The ability of arrestin-3 to interact with IκBα was blocked by its phosphorylation by casein kinase II, and stimulation of b2AR in epidermal cells promoted arrestin-3 dephosphorylation together with arrestin-3-dependent suppression of NFκB activity. Therefore, arrestin-3 facilitated UV-induced apoptosis in the b2AR-dependent manner via inhibition of the NFκB pathway. An alternative mechanism of arrestin-dependent regulation of the NFκB activity was demonstrated in the immune system. Arrestin-3 was shown to interact with TRAF6 and inhibit TRAF6 autoubiquitination and oligomerization after stimulation of interleukin receptors, leading to suppression of the NFκB activation and immunological response to endotoxin challenge (Wang et al. 2006). Since TRAF6 is also involved in the TNFR1-induced NFκB activation, suppression of this effect may also favor apoptosis instead of cell survival.

An alternative mechanism of arrestin contribution to apoptosis caused by DNA damage is through its modulation of the p53 pathway. Arrestin-3 has been shown to bind E3 ubiquitin ligase, Mdm2, but not MdmX (Wang et al. 2003b), the key regulator of p53-dependent apoptosis mediated by the intrinsic pathway. Arrestin-3 binding to Mdm2 inhibited Mdm2 self-ubiquitination and Mdm2-dependent p53 ubiquitination, thus suppressing p53 degradation and promoting apoptosis. It appears somewhat inconsistent that arrestin-3 stabilizes both Mdm2 and p53, although normally high level of Mdm2 leads to a reduction in p53, and Mdm2 needs to be destabilized to allow the p53 level to rise (Stommel and Wahl 2005; Vousden and Lane 2007). Importantly, arrestin-3 binding to Mdm2 was reported to be strongly promoted by the activation of GPCRs such as δ-opioid, bradykinin, or b2AR. In this situation, arrestin-3 acted as a pro-apoptotic agent facilitating DNA damage-induced apoptosis. Conversely, arrestin-2, but not arrestin-3, recruited to active b2AR has been reported to facilitate Akt-mediated activation of Mdm2 promoting Mdm2-dependent degradation of p53 (Hara et al. 2011). Reduced level of p53 leads to the accumulation of stress-induced DNA damage in cultured cells and in the thymus in vivo, presumably due to defective p53-dependent apoptosis of damaged cells. Similarly, behavioral restrained stress leads to a reduction in the p53 level and accumulation of DNA damage in the mouse frontal cortex in the b2AR-and arrestin-2-dependent manner (Hara et al. 2013). Thus, arrestin-2 appears to play the antiapoptotic role via its activation of Mdm2. It remains unclear whether the opposite functions ascribed to arrestins in these studies could be explained away by the difference in arrestin isoforms, arrestin-3 versus arrestin-2, acting via different mechanism, direct interaction with Mdm2 versus indirect activation by Akt-dependent phosphorylation. Importantly, in these studies the mode of apoptosis, DNA damage induced, was the same, and the stimulating factor such as b2AR activation was also similar.

Although the role of JNK pathway in apoptosis is reasonably well established, and arrestin-3 is known to activate JNK3 (McDonald et al. 2000; Miller et al. 2001; Song et al. 2009a; Seo et al. 2011; Zhan et al. 2011, 2013; Breitman et al. 2012), this function of arrestins in apoptosis received surprisingly little attention. This is possibly because JNK3 expression is largely limited to the nervous system, with lower levels in the heart and testes (Gupta et al. 1996; Martin et al. 1996). Both arrestin isoforms attenuated H₂O₂-induced apoptosis by suppressing the JNK activation via direct interaction with apoptosis signal-regulated kinase-1 (ASK1), the upstream kinase (MAPKKK) in the JNK pathway (Zhang et al. 2009). Arrestin-ASK1 interaction, which was increased by H₂O₂, promoted ASK1 ubiquitination, via recruitment of E3 ligase CHIP, and subsequent proteasomal degradation, resulting in reduced JNK activation and increased cell survival without apparent contribution from receptors. Ironically, this is the opposite paradigm to the classic arrestin-mediated JNK3 activation based on scaffolding by arrestin-3 of JNK3 upstream kinases, leading to enhanced JNK3 activation (McDonald et al. 2000). In the study by Zhang et al., the authors claim that arrestin-3-dependent JNK3 activation facilitated neuronal apoptosis following ischemia (Zhang et al. 2012). However, no evidence of role of arrestin-3-dependent JNK3 activation was presented. Instead, the experiments demonstrated reduced JNK3 activation by angiotensin II type 1 receptor antagonist losartan accompanied by protection against ischemia/reperfusion-induced neuronal death.

Arrestin-3 has been shown to mediate endocytosis of the type III transforming in the serum starvation condition growth factor-beta (TGFβ) receptor (TGFβRIII) in complex with TGFβRII and to reduce TGF-beta signaling (Chen et al. 2003). The loss of arrestin-3 has been shown to increase the rate of TGFβ-induced apoptosis, but the effect was mediated by enhanced TGFβ-dependent activation of p38 MAP kinase and not by Smad activation (McLean et al. 2013). The arrestin-3 effect on the p38 activity is likely mediated by its effect on the trafficking of TGFβ receptors, since the surface expression of the receptors increases upon the loss of arrestin-3. Arrestin function in apoptosis could be mediated by interactions with proteins outside of canonical pro- or antiapoptotic pathways. For example, arrestin-2 has been shown to protect human urothelial cells from staurosporine-induced apoptosis in b2AR-dependent manner via interaction with 27-kDa heat shock protein (Rojanathammanee et al. 2009).

Arrestins mostly regulate apoptosis via signaling mechanisms in the cytosol, although both are known to shuttle between the cytosol in the nucleus (Scott et al. 2002; Wang et al. 2003a; Song et al. 2006). Arrestin-3, but not arrestin-2, possesses a strong nuclear export signal and is able to relocalize nuclear proteins such as JNK and Mdm2 from the nucleus to the cytosol (Scott et al. 2002; Wang et al. 2003b; Song et al. 2006). Arrestin-2, on the other hand, has a single amino acid difference with arrestin-3 in the corresponding region, and unless the nuclear export signal is engineered, it is unable to relocalize its binding partners from the nucleus (Wang et al. 2003a; Song et al. 2006). Arrestin-2 possesses nuclear localization signal (Hoeppner et al. 2012) and has been reported to localize to the nucleus in some cell types (Hoeppner et al. 2012) and perform nuclear functions (Kang et al. 2005). Therefore, arrestins, arrestin-2 in particular, could interfere with apoptosis via signaling in the nucleus. Indeed, arrestin-2 has been shown to confer cytoprotection by stimulating transcription of antiapoptotic BCL-2 and thus promoting the survival of CD4+ native and activated T cells (Shi et al. 2007). Nuclear arrestin-2 also interacts with p53, acting in a somewhat poorly defined role of E3 ligase “adaptor” required for Mdm2 to ubiquitinate p53 and promote its degradation, although cytoplasmic arrestin-2 is sufficient to activate Mdm2 via the Akt pathway upon b2AR stimulation (Hara et al. 2011).

The ability of arrestins to engage the survival mechanisms via arrestin-dependent signaling could be taken advantage of via so-called biased ligands that are able to stimulate arrestin recruitment upon binding to GPCRs without inducing G protein activation (see chapter “Quantifying Biased β-Arrestin Signaling”). In some pathological conditions, the activity of select GPCRs is harmful, and in such cases antagonists are used as therapeutic agents. However, arrestin-dependent signaling, which might be beneficial, is also abolished by such drugs. The use of biased ligands achieves both ends: suppresses G protein-mediated and engages arrestin-mediated signaling. Thus, arrestin-biased ligand of angiotensin II type 1 receptor confers protection against cardiac injury induced by ischemia reperfusion injury or mechanical stretch, which is superior to that provided by angiotensin II antagonist losartan, a commonly used therapeutic agent. The protection was arrestin dependent, since it was absent in mice lacking arrestin-3 (Kim et al. 2012a).

4 Arrestins Regulate Apoptosis via Direct Interference in the Core Apoptotic Machinery

Thus, arrestins can affect cell survival in many ways via signaling, but direct pro-apoptotic action of arrestins at the core of cell death machinery was only recently reported (Kook et al. 2013). The signaling in the intrinsic apoptotic pathway involves cytochrome C release from the mitochondria that is orchestrated by the complex interplay of pro- and antiapoptotic members of the BCL-2 family of proteins (Chipuk and Green 2008, 2009). However, the exact mechanism of the process remains elusive. The involvement of additional players that do not belong to BCL family is one of the emerging ideas (Chipuk and Green 2008). Proteomic surveys suggest that caspase cleavage might supply regulators of apoptosis (Dix et al. 2008; Mahrus et al. 2008), but specific functional roles of cleavage products are rarely established. Active caspases are a notable example of caspase cleavage products playing critical role in cell death (Wolan et al. 2009). Another well-known example is BID: the product of its cleavage by caspases tBID translocates to mitochondria and promotes cytochrome C release. Possibly, some of the so-called dependence receptors require caspase-mediated cleavage for their death domains to be revealed or released (Bredesen et al. 2005). However, the full signaling potential of caspase cleavage products to affect this crucial step in the apoptotic pathway has not yet been explored.

As it turned out, arrestin-2 is cleaved by caspases at evolutionarily conserved Asp380 yielding an Arr2-(1-380) fragment (Kook et al. 2013). Apoptosis initiated via extrinsic (stimulation with TNFα combined with inhibition of protein synthesis by cycloheximide) or intrinsic (genotoxic drug etoposide) pathway resulted in the appearance of the same arrestin-2 fragment. A secondary cleavage site in arrestin-2, at Asp406 that is conserved only in mammals, was identified. When both aspartates were mutated to glutamates, the mutant (DblE) was resistant to caspases in all cell types (Kook et al. 2013). The presence of Asp380 in homologous positions in arrestin-2 from multiple species indicates that this mechanism is conserved in vertebrate evolution. Unlike many substrates, arrestin-2 is not just an “innocent victim” of caspases. 1-380 translocated to the mitochondria and enhanced cytochrome C release by “assisting” another product of caspase-8 activity, tBID. Since virtually every mammalian cell expresses both arrestin-2 and arrestin-3 (Gurevich and Gurevich 2006), specific functions of individual subtypes can only be dissected in cells lacking one or the other. Arrestin-2 (A2KO) and arrestin-3 (A3KO) knockout mouse embryonic fibroblasts (MEFs), as well as arrestin-2/arrestin-3 double-knockout (DKO) MEFs, established more than a decade ago (Kohout et al. 2001), proved to be extremely useful tools in this regard. Increased cytochrome C release due to 1-380 significantly accelerated the progression of apoptosis. The rate of caspase activation and cell death in A3KO MEFs expressing only arrestin-2 was two- to threefold higher as compared to DKO MEFs lacking both arrestins. The ectopic expression of 1-380 in DKO MEFs facilitated TNF-α-induced apoptosis to the level observed in A3KO MEFs. WT arrestin-2 but not its uncleavable mutant also rescued vulnerability of DKO MEFs to cell death (Kook et al. 2013). Arrestin-2 does not have an identifiable mitochondrial localization signal, and mitochondria contain very little full-length arrestin-2, but large proportion of 1-380 localizes to mitochondria. Direct binding of purified 1-380 to isolated mitochondria and mitochondrial localization of expressed 1-380 even in non-apoptotic cells shows that, in contrast to full-length arrestin-2, it has an increased affinity for protein(s) residing in this compartment. 1-380 did not induce cytochrome C release by itself in cells or isolated mitochondria. Instead, it directly interacted with tBID and specifically facilitated cytochrome C release induced by tBID. The absence of BID completely abrogated pro-apoptotic effect of 1-380. Thus, caspase cleavage of arrestin-2 is a gain-of-function event resulting in a stronger interaction with tBID and the ability to enhance tBID-induced cytochrome C release that uncleaved arrestin-2 does not possess (Kook et al. 2013) (Fig. 1).

Both 1-380 and tBID are effectively generated by caspase-8, suggesting that their convergence at mitochondria plays crucial role in the extrinsic apoptotic pathway. However, apoptosis, like most cellular processes, has multiple backup mechanisms (Slee et al. 2000; Crawford and Wells 2011). Although the canonical way for caspase-8 activation is via death receptors, caspase-8 can also be activated in death receptor-independent manner, as seen, for example, in genotoxic drug-induced apoptosis (von Haefen et al. 2003; de Vries et al. 2007). Such activation occurs downstream of the mitochondria, cytochrome C release, and activation of effector caspases. Furthermore, in the absence of caspase-8, 1-380 could be generated by other caspases such as caspase-9 or caspase-6 (Kook et al. 2013).

Caspase activity in the cell is greatly increased by released cytochrome C via the apoptosome (Riedl and Salvesen 2007). Thus, cooperation of 1-380 and tBID in cytochrome C release creates a potent positive feedback loop, tipping the balance towards cell commitment to apoptotic death. This mechanism also sets a threshold for an irreversible cell “decision” to die: simultaneous generation of both fragments is necessary to maximize the death signal. The arrestin-2-dependent positive feedback loop greatly contributed to the mitochondrial apoptotic pathway, with magnitude of 1-380 effect on isolated mitochondria and intact cells comparable to that of tBID (Kook et al. 2013). The permeabilization of the outer mitochondrial membrane and the resulting cytochrome C release is usually the point of no return, committing the cell to death (Danial and Korsmeyer 2004). Extensive studies of this step focusing on the interactions of pro- and antiapoptotic BCL family members with each other and pore-forming effectors BAK and BAX suggest that BID, BIM, and PUMA act as direct activators (Wei et al. 2000; Kim et al. 2009a; Ren et al. 2010). However, many molecular details necessary for mechanistic understanding of this process are missing (Chipuk and Green 2008; Youle and Strasser 2008). Our recent finding of the role of arrestin-2 cleavage product (Kook et al. 2013) supports the idea that direct involvement of additional players may explain inconsistencies between in vitro studies with BCL proteins and in vivo apoptosis (Chipuk and Green 2008). Our results suggest that tBID in complex with 1-380, rather than tBID alone, is the biologically relevant inducer of cytochrome C release (Kook et al. 2013). It is tempting to speculate that in cytochrome C release BIM and PUMA might also have their specific “helpers,” possibly generated by caspases.

Caspases often produce discrete stable cleavage products likely serving as functional effectors in apoptosis (Dix et al. 2008; Mahrus et al. 2008). However, the functions of caspase-generated fragments are rarely established. The functional consequences of the cleavage of most of the 777 caspase substrates in CASBAH database (Lüthi and Martin 2007) remain unknown. Caspase cleavage of several kinases unleashes or abrogates their pro-apoptotic or pro-survival functions, respectively, via changes in activity, subcellular localization, or substrate preferences (Kurokawa and Kornbluth 2009). Caspase cleavage products of diverse proteins contribute to the progression of apoptosis due to loss or gain of function or via dominant-negative action (Kim et al. 2009b; Crawford and Wells 2011; Oliver et al. 2011). Our experiments revealed a direct role of 1-380 in cytochrome C release, identifying it as an earlier unappreciated active participant in the core mechanism of apoptosis (Kook et al. 2013). This is the first example of direct cooperation of two caspase products, 1-380 and tBID, at the point where the cell makes a fateful decision to live or die. This cooperation likely contributes to making this decision irreversible and also effectively sets a threshold for cell commitment to apoptotic death.

5 Visual Arrestins in Apoptosis

Of the four vertebrate arrestin isoforms, two, arrestin-1 (a.k.a. visual or rod arrestin) and arrestin-4 (a.k.a. cone arrestin), are expressed in retinal photoreceptors. Arrestin-1 is expressed in both rods and cones, whereas arrestin-4 is found in cones (Gurevich and Gurevich 2009, 2010; Gurevich et al. 2011). Photoreceptors are highly specialized neurons adapted for their function. Both rods and cones have specialized compartment, the outer segment, where photopigment and proteins of the signaling cascade reside, largely separated from the rest of the cell (Pugh and Lamb 2000). Rods function in dim light and are exquisitely light sensitive, being capable of detecting one photon of light (Baylor et al. 1979). Such sensitivity is achieved, among other things, by high levels of expression of main signaling proteins such as photosensitive pigment rhodopsin (~3 mM) and arrestin-1 (>2 mM), an important component of the potent shutdown system ensuring almost zero background signaling (Pugh and Lamb 2000; Gurevich and Gurevich 2009; Gurevich et al. 2011). For comparison, the concentrations of higher expressed nonvisual arrestin isoform arrestin-2 in the adult rat central nervous system are ~200 nM and that of arrestin-3—almost 20-fold lower—~12 nM (Gurevich et al. 2004). Because of such high load of signaling proteins, the balance in rods is very precarious, and changes in the expression levels of signaling proteins often lead to rod death. A well-known example is rhodopsin: an excellent correlation between the level of overexpression of this perfectly normal protein and the rate of photoreceptor degeneration was established (Tan et al. 2001). Transgenic overexpression of wild-type arrestin-1 did not undermine photoreceptor survival, although it somewhat compromised the health of their outer segments in older mice (Song et al. 2011). Conversely, the loss of arrestin-1 induced defect in signaling shutoff, excessive signaling, and light-dependent degeneration of rod outer segments and rod death by apoptosis (Xu et al. 1997; Song et al. 2009b, 2013). Even hemizygous mice with ~50 % level of arretin-1 showed somewhat lower level of rod survival (Song et al. 2009b, 2013). Interestingly, the expression of arrestin-1 as low as 5 % of wild-type (WT) level was sufficient to maintain photoreceptor health and support their functional performance (Cleghorn et al. 2011; Song et al. 2011). Thus, the total loss of arrestin-1 function of quenching rhodopsin signaling is detrimental for rod survival and leads to rod death by apoptosis.

Arrestin interaction with phosphorylated rhodopsin that quenches phototransduction is required for rod survival. However, a very tight arrestinrhodopsin interaction could be detrimental for rods, resulting in rod death. Such tight interaction is believed to cause retinal degeneration, a group of retinal degenerative diseases known as retinitis pigmentosa characterized by variable loss of rod photoreceptors across the retina followed by the death of cone photoreceptors (Mendes et al. 2005). Most of the cases are autosomal dominant and are caused by mutations in rhodopsin, leading to it being constitutively active or constitutively phosphorylated by rhodopsin kinase, with both conditions resulting in persistent arrestin-1 binding (Rim and Oprian 1995). Arrestin-1 mislocalizes rhodopsin from the outer segments to endosomes in inner segments and cell bodies, leading to rod death (Chuang et al. 2004; Chen et al. 2006). Recruitment of endocytic adapter protein-2 (AP-2) via arrestin-1 plays a role in rod death induced by arrestin-1 complex with constitutively active rhodopsin mutant (Moaven et al. 2013). A naturally occurring splice variant of arrestin-1 p44 lacking a part of the arrestin-1 C-tail and thus incapable of interacting with AP-2 but competent to quench phototransduction prevents the death of photoreceptors expressing constitutively active rhodopsin (Moaven et al. 2013). This pathway is evolutionarily conserved, since the same tight association of arrestin with activated rhodopsin induces apoptotic death of Drosophila photoreceptors (Alloway et al. 2000; Kiselev et al. 2000; Kristaponyte et al. 2012).

Arrestin-1 of several species has recently been shown to cooperatively form dimers and tetramers, with only a small fraction of it existing as monomer at physiological concentrations (Schubert et al. 1999; Imamoto et al. 2003; Hanson et al. 2007b, 2008b; Kim et al. 2011; Chen et al. 2013) (see chapter “Self-Association of Arrestin Family Members”). The physiological function of this phenomenon remained unclear. However, the mouse line expressing arrestin-1 mutant with reduced ability to form oligomers at ~240 % of WT arrestin-1 (which resulted in the monomer concentration exceeding that in WT mouse almost threefold) demonstrated rapid age-related apoptotic death of rod photoreceptors (Song et al. 2013). The mouse line expressing the same mutant at a much lower level (50 % of wild type, which yielded only ~20 % increase in monomer concentration) showed a very slow age-dependent degeneration. Importantly, mice overexpressing WT arrestin-1, which robustly oligomerizes, so that overexpression leads to minimal increase in the monomer concentration, did not show photoreceptor death. Furthermore, co-expression of WT arrestin-1 with the mutant protected rods from the mutant-induced apoptosis, suggesting that previously demonstrated ability of WT to recruit mutant arrestin-1 into mixed oligomers (Hanson et al. 2007b, 2008b) may be responsible. These data suggest that monomeric arrestin-1 is toxic to rods and provides a functional explanation for the ability of arrestin-1 to oligomerize. Only monomeric arrestin-1 interacts with rhodopsin (Hanson et al. 2007b, 2008b), and it binds rhodopsin monomer with high affinity (Hanson et al. 2007a; Bayburt et al. 2011; Kim et al. 2012b; Singhal et al. 2013; Vishnivetskiy et al. 2013; Zhuang et al. 2013). Thus, arrestin-1 oligomers likely represent a nontoxic storage form not only in rods but also in cones. Arrestin-4, or cone arrestin, in spite of its name, is outnumbered in cone photoreceptors by arrestin-1 by ~50:1 (Nikonov et al. 2008). Since cone arrestin is unable to self-associate (Hanson et al. 2008a), and as monomer could be toxic to photoreceptors, cones simply cannot afford to express arrestin-4 at high level necessary to rapidly quench phototransduction in bright light in which they operate. Therefore, a certain amount of arrestin-4 might be produced for immediate use, but the main stock is kept as arrestin-1 oligomers to be employed when needed.

In retinal photoreceptors, rhodopsin is the main binding partner of visual arrestins, and the main function of arrestins in photoreceptors is to quench rhodopsin signaling (Gurevich et al. 2011). However, arrestin-1 is found not only in close proximity to its key binding partner rhodopsin, i.e., in the rod outer segments, but also in other rod compartments, specifically in synaptic terminals (Nair et al. 2005; Hanson et al. 2007a; Huang et al. 2010). Very little attention so far has been paid to possible rhodopsin-independent functions of arrestin-1 and their potential role in retinal photoreceptor death and survival. It has recently been shown that arrestin-1 interacts with N-ethylmaleimide-sensitive factor (NSF) (Huang et al. 2010). NSF is localized to photoreceptor synapses and functions to sustain high rate of neurotransmitter exocytosis. Arretin-1 interaction with NSF was enhanced in the dark when rods were depolarized and neurotransmitter release was elevated. Mice lacking arrestin-1 displayed reduced levels of NSF and of other synaptic proteins, as well as reduced exocytosis, suggesting that arrestin-1 interaction with NSF was required for normal synaptic function in rods. Therefore, it is conceivable that arrestin-1 modulates rod survival via its interactions with proteins other than rhodopsin, such as synaptic proteins. It has been shown that arrestin-1 mutant with reduced ability to self-associate caused damage to synaptic terminals, which was detectable earlier than the loss of photoreceptors, suggesting that synapses might be the site of toxicity of arrestin-1 monomer (Song et al. 2013). Co-expression of WT arrestin-1 that protected photoreceptors from apoptosis conferred even more significant protection to synapses, again supporting the notion of synapses being the primary site of damage. These data indicate that the role of arrestin-1 in the photoreceptor death and survival could involve arrestin-1 interaction with proteins other than rhodopsin. Specifically, arrestin-1 function in photoreceptor synapses could be necessary to maintain photoreceptor health. Interestingly, the binding of arrestin-2 to NSF was described more than a decade ago (McDonald et al. 1999), suggesting that nonvisual subtypes might also be involved in synaptic functions, including neurotransmitter release.

6 Conclusions and Future Directions

Apoptotic cell death plays an important role in embryonic development, in homeostasis of multicell organisms, as well as in numerous pathological processes. Arrestins appear to be intimately involved in the regulation of a variety of signaling pathways involved in cell death and survival. Thus, reengineered signaling-biased arrestins with enhanced pro-apoptotic or pro-survival functions can be used as molecular tools to influence cell decision to live or die in desired direction (Gurevich and Gurevich 2012). Arrestin-3 mutants lacking the ability to activate JNK family kinases (chapter “Arrestin-Dependent Activation of JNK Family Kinases”) are obvious candidates to be tested in this regard. In addition, caspase cleavage product of arrestin-2 directly participates in the critical step in vertebrate apoptosis, assisting tBID in cytochrome C release (Kook et al. 2013). Importantly, caspase-resistant arrestin-2 mutant is lacking pro-apoptotic function (Kook et al. 2013), suggesting that it might be useful for cytoprotection. This finding also raises a question whether other arrestin subtypes, such as arrestin-1 and arrestin-3, are targeted by caspases and whether generated cleavage products acquire new functions, similar to that of arrestin-2. Visual arrestin-1, the functions of which were long believed to be limited to the shutoff of rhodopsin signaling (Gurevich et al. 2011), turned out to be an important regulator of photoreceptor health and survival acting at synaptic terminals (Song et al. 2013). Since the integrity of synapses is necessary for survival of all neurons, nonvisual arrestins might play a role in neuronal survival similar to that described for arrestin-1. Three out four vertebrate arrestin subtypes self-associate forming distinct oligomers (Chen et al. 2013). Since arrestin-1 monomers appear to be cytotoxic, whereas oligomers are perfectly harmless (Song et al. 2013), self-association of nonvisual arrestins might also affect their role in cell survival (see chapter “Self-Association of Arrestin Family Members”). Considering that arrestins can impact cell death and survival via numerous mechanisms, every redesigned arrestin mutant, including those with enhanced ability to bind unphosphorylated receptors (chapter “Enhanced Phosphorylation-Independent Arrestins and Gene Therapy”), targeting individual GPCRs (chapter “Targeting Individual GPCRs with Redesigned Non-visual Arrestins”), and with modified trafficking properties (chapter “β-Arrestins and G Protein-Coupled Receptor Trafficking”), changing interactions with ubiquitinating and deubiquitinating enzymes (chapter “Arrestin Interaction with E3 Ubiquitin Ligases and deubiquitinases: Functional and Therapeutic Implications”), must be specifically tested for its effect on apoptosis and cell survival.