Bcl-2-Protein Family as Modulators of IP₃ Receptors and Other Organellar Ca²⁺ Channels

Abstract

The pro- and antiapoptotic proteins belonging to the B-cell lymphoma-2 (Bcl-2) family exert a critical control over cell-death processes by enabling or counteracting mitochondrial outer membrane permeabilization. Beyond this mitochondrial function, several Bcl-2 family members have emerged as critical modulators of intracellular Ca²⁺ homeostasis and dynamics, showing proapoptotic and antiapoptotic functions. Bcl-2 family proteins specifically target several intracellular Ca²⁺-transport systems, including organellar Ca²⁺ channels: inositol 1,4,5-trisphosphate receptors (IP₃Rs) and ryanodine receptors (RyRs), Ca²⁺-release channels mediating Ca²⁺ flux from the endoplasmic reticulum, as well as voltage-dependent anion channels (VDACs), which mediate Ca²⁺ flux across the mitochondrial outer membrane into the mitochondria. Although the formation of protein complexes between Bcl-2 proteins and these channels has been extensively studied, a major advance during recent years has been elucidating the complex interaction of Bcl-2 proteins with IP₃Rs. Distinct interaction sites for different Bcl-2 family members were identified in the primary structure of IP₃Rs. The unique molecular profiles of these Bcl-2 proteins may account for their distinct functional outcomes when bound to IP₃Rs. Furthermore, Bcl-2 inhibitors used in cancer therapy may affect IP₃R function as part of their proapoptotic effect and/or as an adverse effect in healthy cells.

B-CELL LYMPHOMA-2 (Bcl-2) FAMILY OF PROTEINS

The Bcl-2 family of proteins consists of pro- and antiapoptotic members, which are characterized by the presence of at least one of the four highly conserved α-helical motifs, termed Bcl-2 homology (BH) domains (Adams and Cory 1998). The antiapoptotic family members, such as Bcl-2, Bcl-Xl, and Mcl-1, contain all four BH domains where the BH1, BH2, and BH3 domains form a hydrophobic cleft (Fig. 1A). The hydrophobic cleft is separated from the amino-terminal BH4 domain by an unstructured loop. The proapoptotic family members include the multidomain Bax, Bak, and Bok, which also contain all four BH domains (Westphal et al. 2011), and the BH3-only family members. On activation, Bax/Bak form oligomeric proteinaceous pores in the outer mitochondrial membrane (OMM), which function as channels. The Bax/Bak pores release apoptotic factors such as cytochrome c and SMAC/Diablo into the cytosol to activate apoptosis, and also enable mitochondrial DNA loss (Kalkavan and Green 2018; McArthur et al. 2018; Riley et al. 2018). BH3-only proteins, such as Bim and Bid (which becomes proaptotic following proteolytic cleavage), are termed activators, owing to their ability to directly activate Bax/Bak. Other BH3-only proteins, like Bad and Noxa, are termed sensitizers because they antagonize the action of the antiapoptotic members without binding and directly activating Bax/Bak (Chipuk et al. 2008). Most of the Bcl-2 family proteins contain a carboxy-terminal transmembrane domain (TMD), which targets the proteins to various intracellular membranes, including the mitochondrial and the endoplasmic reticulum (ER) membranes (Akao et al. 1994; García-Sáez 2012; Popgeorgiev et al. 2018).

Figure 1.

Bcl-2 family of proteins. (A) Representation of the linear structure of the Bcl-2 proteins. The antiapoptotic (green) and the proapoptotic (red) family members are shown. The BH1-4 domains, the transmembrane domain (TMD), and the hydrophobic cleft are indicated. The sequestration of the BH3 domain of the proapoptotic proteins by the hydrophobic cleft of the antiapoptotic members, which mediates apoptosis prevention, is also indicated. (B) Bcl-2 proteins at the mitochondria. The antiapoptotic Bcl-2 proteins bind to the proapoptotic Bax/Bak proteins and BH3-only proteins, thereby neutralizing their proapoptotic activity and facilitating cell survival. Oligomerization of the proapoptotic Bax/Bak proteins in response to activator BH3-only proteins (such as Bim and truncated Bid) results in outer mitochondrial membrane (OMM) permeabilization, enabling cytochrome c (cyt c) release into the cytosol with subsequent caspase 3/7 activation, eventually leading to cell death.

The pro- and antiapoptotic Bcl-2 family members interact with each other, forming a network of protein complexes that tightly control apoptosis. The interactions within the family revolve around the pro-death role of the BH3 domain. Indeed, the antiapoptotic members use their hydrophobic cleft to scaffold and neutralize the BH3 domain of the proapoptotic family members. The BH4 domains of Bcl-2 and Bcl-Xl were also implicated in counteracting Bax activation (Ding et al. 2010, 2014; Barclay et al. 2015). The interactions within the Bcl-2 family of proteins occur mainly at the mitochondria, where antiapoptotic Bcl-2 family members prevent Bax/Bak oligomerization, thereby suppressing apoptosis (Fig. 1B; Adams and Cory 1998; Chipuk et al. 2008, 2010; Shamas-Din et al. 2013).

Many of the Bcl-2 family proteins also localize at the ER (Popgeorgiev et al. 2018), where they may act as modulators of Ca²⁺ signals (Ferrari et al. 2002; Vervliet et al. 2016), which play a central role in the regulation of cell survival and death (Orrenius et al. 2003; Zecchini et al. 2007; Marchi et al. 2008; Zhivotovsky and Orrenius 2011). Here, we summarize the important role of Bcl-2 family members as direct modulators of the stability or function of organellar Ca²⁺ channels: the inositol 1,4,5-trisphosphate receptors (IP₃Rs) and the ryanodine receptors (RyRs) at the ER, and the voltage-dependent anion channels (VDACs) at the OMM. Our main focus is the IP₃R, for which the crystal structure was recently solved in both ligand-free (apo) (Fan et al. 2015, 2018) and ligand-bound states (Fan et al. 2018).

Emerging topics in this field and recent insights are that:

1. Different Bcl-2-family members, including Bok, Bcl-2, Bcl-Xl, Mcl-1, and Bcl-2L10/Nrh directly target IP₃Rs;

2. Each Bcl-2-family member modulates the properties of the IP₃R in a unique way, impacting IP₃R stability (Bok) or function, resulting in IP₃R inhibition (Bcl-2, Bcl-Xl at high concentrations, Bcl-2L10/Nrh) or IP₃R sensitization (Bcl-Xl at low concentrations, Mcl-1);

3. The interaction between Bcl-2 proteins and IP₃Rs often involves multiple binding sites, enabling modulatory Ca²⁺-signaling outputs;

4. The binding of certain Bcl-2 family members to IP₃R impacts properties of the Bcl-2 protein; for instance, Bok is stabilized on binding to IP₃Rs;

5. Bcl-2 family members exert part of their cell death and survival functions by forming complexes with IP₃R: Bcl-2 prevents proapoptotic Ca²⁺ transients, whereas Bcl-Xl promotes prosurvival Ca²⁺ oscillations;

6. Targeting IP₃R/Bcl-2-family protein complexes offers novel means to interfere with Bcl-2 protein function in malignancies that are dependent on these proteins, for instance, peptides targeting IP₃R/Bcl-2 complexes can kill B-cell cancers, whereas peptides targeting IP₃R/Bcl-2L10 complex can kill breast cancer cells; and

7. Other organellar Ca²⁺ channels such as RyRs and VDACs are targeted and modulated by Bcl-2 family members.

These aspects are discussed below in more detail.

STRUCTURE AND FUNCTION OF IP₃Rs

IP₃Rs function as tetrameric Ca²⁺-release channels with a monomeric molecular mass of ∼300 kDa. In vertebrates, IP₃Rs exist as three isoforms (IP₃R1, IP₃R2, and IP₃R3), which are encoded by three different genes (ITPR1, ITPR2, and ITPR3) (Foskett et al. 2007; Mikoshiba 2007; Parys and De Smedt 2012). All three isoforms show strong sequence homology (∼70% sequence identity), especially within their TMDs (∼90% sequence identity), yet they show significant functional differences that result in isoform-specific properties. The diversity of the IP₃Rs is further expanded by alternative gene-splicing variants and by homo- and heterotetrameric assembly of isoforms and splice variants into functional channels (Monkawa et al. 1995; Patel et al. 1999; Onoue et al. 2000).

The IP₃R monomer is organized into three functional regions. The amino-terminal region, consisting of ∼600 amino acids (aa), represents the ligand-binding domain (LBD), which contains the suppressor domain and the IP₃-binding core (IBC). The LBD is followed by the central, modulatory region, which couples the ligand binding to the channel opening. Finally, the carboxyl terminus consists of six TMDs, whereby the channel pore is located between the fifth and the sixth TMD, and a cytosolic tail (Fig. 2A,B; Mignery et al. 1990; Südhof et al. 1991; Michikawa et al. 1994). Structural information on IP₃R channels was previously limited to X-ray structures of IP₃R1's soluble LBD (Lin et al. 2011; Seo et al. 2012), which represent only ∼15% of the protein sequence. Yet, the detailed mechanism by which the channel agonists activate the passage of Ca²⁺ through the channel remained unresolved. With major advances in the fields of cryo-electron microscopy (cryo-EM) (Kühlbrandt 2014) and the availability of biochemically optimized cryo-specimen (Murray et al. 2013), the structure of the full-length, tetrameric IP₃R1 channel was determined to a resolution of 4.7 Å (Fan et al. 2015, 2018). The channel assembly shows a unique architecture, whereby each of the subunits is arranged around two four-helix bundles, forming a central core along the fourfold axis (Fan et al. 2015). The transmembrane bundle is formed by the sixth TMD helix from each subunit packed in a right-handed fashion and the cytosolic left-handed packed bundle is formed by the ∼80 Å long helix from the CTD of each subunit.

Figure 2.

Inositol 1,4,5-trisphosphate receptor (IP₃R) structure with indication of the binding sites for the different Bcl-2-family members. (A) Representation of the linear structure of IP₃R1. The structural domains according to Fan et al. (2015) are depicted as colored lines. The functional regions (i.e., ligand-binding domain with its suppressor domain and IP₃-binding core; central, modulatory region, and carboxy-terminal region with the transmembrane domains [TMDs]) are indicated in black, and the five fragments resulting from trypsinization are indicated in blue. The Bcl-2 family members are indicated in different colors and positioned along the structure, according to the targeted region. Wherever possible, the exact binding site is indicated: the BH4-binding site (BH4-BS) for Bcl-2 in ARM2 in the fragment 3 is indicated as a yellow box; the binding site for Bok (Bok-BS) in ARM3 in the fragment 4 is indicated as an orange box; the helix 1 and 4 in the carboxyl terminus of IP₃R, targeted by Bcl-Xl are indicated in green. (B) IP₃R monomer viewed along the membrane plane. IP₃R monomer is depicted in grey with its ligand-binding domain, central, modulatory region, and carboxyl terminus. The IP₃-binding site is indicated in red. (C–G) IP₃R dimers viewed along the membrane plane. The binding sites for Bok (C), Bcl-2 (D), Bcl-Xl (E), Mcl-1 (F), and Nrh/Nrz (G) are indicated in the same color as in A.

At the monomer level, the cytoplasmic part of the channel appears to be rich in α-helices, enabling flexibility and fast changes in the channel conformation on ligand binding. The apical portion of the structure contains the amino-terminal LBD, which is comprised of two β-trefoil domains (β-TF1 and β-TF2, aa 1–436, according to rat IP₃R1 sequence) and a consecutive armadillo solenoid fold (ARM1, aa 436–714). ARM1 is followed by a helical domain, which together with two other α-helical domains of armadillo repeats (ARM2, aa 1030–1494 and ARM3, aa 1594–2192) form the central, modulatory region. The ARM3 expands into two antiparallel β-strands followed by a helix–turn–helix motif. This extension is called the intervening lateral domain (ILD, aa 2193–2272) and together with the ARM2 and ARM3 facilitates the communication between the LBD and the carboxyl terminus, where the ILD connects to the six α-helical TMDs. The sixth TMD helix contains the bulky hydrophobic residues (F2586 and I2590), which constitute a physical gate and can function to either hinder or permit ion translocation across the ER membrane. The sixth TMD helix extends beyond the lipid membrane boundaries into the cytosol where it is connected to a helical linker domain (LNK, aa 2609–2681) followed by the long cytosolic α-helix, designated as a carboxy-terminal domain (CTD, aa 2682–2750) (Fan et al. 2015).

A striking feature of IP₃R1 monomer arrangement is the sandwich formed by the LNK domain surrounded by the β-strands and the helix–turn–helix regions of the ILD. With both domains directly connected to the TMD, the ILD/LNK unit is uniquely positioned to integrate all channel gating signals arising from the cytosolic domains to the gating machinery within the TMD. Further details regarding the molecular mechanism of channel gating revealed that the cytosolic carboxy-terminal helix of one subunit directly interacts with the amino-terminal LBD, in particular the β-TF2 domain, of the adjacent subunit, suggesting an allosteric type of regulation of the channel (Fan et al. 2015).

Recent structures of the IP₃R1 in a ligand-bound (4.1 Å) and apo state (3.9 Å) solved by cryo-EM revealed considerable changes in both the pore and cytoplasmic domains on binding of activating ligands (Fan et al. 2018). Thereby, a network of intra- and interdomain interfaces responsible for conformational coupling between ligand binding and gating activation were identified. Adenophostin A, an IP₃ analog and potent IP₃R agonist of fungal origin (Takahashi et al. 1994), evoked structural changes in the LBD, where the β-TF2 and ARM1 undergo a 5 Å closure of the cleft between them. The changes in the LBD trigger pronounced rearrangements of the cytoplasmic domains, rotation of the CTD helix and lateral movements in the interfacial ILD/LNK region. The ILD/LNK region appears to be mechanically connected to pore opening, whereby changes in the ILD/LNK interface traverse to the TMDs. Ligand gating would require the pore-forming helical bundle comprised of four pairs of the inner sixth TMDs to dilate at the hydrophobic constrictions formed by the side-chains of F2586 and I2590. Mutations within the ILD (Hamada et al. 2017) and LNK (Uchida et al. 2003; Bhanumathy et al. 2012) domains support the mechanism of signal transduction through the ILD/LNK assembly formed at the membrane cytosol interface. Overall, the structures of IP₃R1 provide the basis for understanding the allosteric regulation of channel gating.

Other IP₃-bound IP₃R structures were determined either by X-ray crystallography (in the case of isolated LBD [Lin et al. 2011; Seo et al. 2012] and large, monomeric cytoplasmic portion of IP₃R1 [Hamada et al. 2017]) or by cryo-EM (IP₃R3) (Paknejad and Hite 2018). These structures described conformational changes within some cytoplasmic domains upon ligand binding. However, it remains unclear whether these changes are representative of the ligand-evoked channel gating, because either the TMD was not present (Hamada et al. 2017) or no conformational changes within the TMD were described for the ligand-bound channel (Paknejad and Hite 2018). Structural studies of IP₃R1 in multiple functional states guided by biophysical characterizations and in physiological conditions will be necessary to fully understand the allosteric mechanism of IP₃R channel gating and regulation.

New insights in the mechanism of activation were gained by using concatenated IP₃Rs. This study showed that proper IP₃R activation and initiation of Ca²⁺ release occurs only when all four subunits are occupied by IP₃ (Alzayady et al. 2016). In addition, previous work showed that the controlled trypsinization of each monomeric IP₃R1 resulted in five fragments, as follows according to the mouse IP₃R1 sequence: fragment 1 (aa 1–345), fragment 2 (aa 346–922), fragment 3 (aa 923–1581), fragment 4 (aa 1582–1931), and fragment 5 (aa 1932–2749) (Fig. 2A). Interestingly, when co-expressed, these five fragments were able to self-assemble into a functional IP₃R1 channel that enabled Ca²⁺ release from the ER in response to IP₃ (Yoshikawa et al. 1999; Alzayady et al. 2013).

The ability of IP₃Rs to respond adequately to the cell's requirements and to tightly control versatile Ca²⁺-dependent processes is a result of a precise regulation. The basic regulators of IP₃Rs are IP₃, Ca²⁺, ATP, and various posttranslational modifications like phosphorylation, glycosylation, palmitoylation, thiol modification, and oxidation by reactive oxygen species, such as H₂O₂ (Bezprozvanny 2005; Foskett et al. 2007; Mikoshiba 2007; Vanderheyden et al. 2009; Booth et al. 2016; Joseph et al. 2018). An important part of the IP₃R population resides at specialized domains at which the ER is in close apposition with other organelles, including the mitochondria (Decuypere et al. 2011; Raturi and Simmen 2013; Marchi et al. 2014) and the lysosomes (Kilpatrick et al. 2013; Atakpa et al. 2018). These domains are often referred to as membrane contact sites and represent bidirectional interactions between the ER and these organelles through IP₃R and Ca²⁺ signaling (La Rovere et al. 2016; Roest et al. 2017). Furthermore, there is an increasing number of documented interactions with regulatory proteins, including several proteins directly involved in the regulation of cell-death and -survival processes such as apoptosis, metabolism, unfolded protein responses, and autophagy (Choe and Ehrlich 2006; Ivanova et al. 2014; Prole and Taylor 2016; Parys and Vervliet, in press).

REGULATION OF IP₃R BY Bcl-2 FAMILY PROTEINS

Proapoptotic Bok as a Stabilizer of IP₃R Proteins

The multidomain proapoptotic Bcl-2-family-member Bok, which failed to bind to any of the Bcl-2 family members (Echeverry et al. 2013; Llambi et al. 2016), emerged as a prominent binding partner of IP₃Rs (Fig. 2; Table 1). The interaction, which occurred between the amino terminus of Bok and a small site in the central, modulatory region of IP₃Rs and in particular within the fragment 4 (aa 1895–1903 of mouse IP₃R1, part of the ARM3 domain), appeared to control the proteolytic degradation of IP₃Rs (Schulman et al. 2013). Vice versa, it was shown that Bok is constitutively bound to IP₃Rs, which promotes Bok stabilization. The unbound Bok is ubiquitinated by the AMFR/gp78 E3 ubiquitin ligase complex, targeted to the proteasome through VCP/p97 and degraded through the ER-associated degradation pathway (Schulman et al. 2016). Bok showed greater affinity for IP₃R1 and IP₃R2, while barely binding to IP₃R3. It is well established that cells stimulated with IP₃-generating agonists for prolonged time down-regulate their IP₃Rs via a process that involves ubiquitination and proteasomal degradation (Oberdorf et al. 1999; Wojcikiewicz et al. 2009). Cells exposed to chronic IP₃R activation showed a concomitant decline in both Bok and IP₃R levels, whereas other Bcl-2-family members remained unaltered. Thus, it seems that IP₃R/Bok complexes are degraded as a “combined unit” (Schulman et al. 2013). In in vitro cleavage experiments, IP₃Rs from Bok^−/− cells were more susceptible to cleavage by chymotrypsin. This correlated with in cellulo experiments demonstrating that caspase-3-mediated cleavage of IP₃R1 on exposure to staurosporine was more readily observed in Bok^−/− cells than in wild-type cells. Yet, wild-type mouse embryonic fibroblasts (MEFs) and Bok^−/− MEF cells displayed very similar IP₃R-mediated Ca²⁺ responses when intact cells were stimulated with lysophosphatidic acid, or if IP₃Rs were directly activated by addition of IP₃ to permeabilized cells. It should be noted that Bok^−/− cells derived from Bok^−/− mice displayed increased IP₃R1 levels, but a marked decline in IP₃R2 levels and a modest decline in IP₃R3 levels (Schulman et al. 2013). Thus, it remains possible that changes in IP₃R function caused by loss of Bok were masked by changes in IP₃R expression. To perform a more direct assessment of the impact of Bok on IP₃R function, Bok-deficient MEF cells have recently been generated by CRISPR/Cas9 (Schulman et al. 2019). In these models, the expression levels of the different IP₃R isoforms were unaffected on deletion of Bok, indicating that IP₃R stability in basal conditions is not affected by Bok. However, also these Bok^−/− MEF cells displayed similar levels of IP₃R-mediated Ca²⁺ release and furthermore of mitochondrial Ca²⁺ influx as the wild-type cells. Interestingly, the deletion of Bok led to mitochondrial fragmentation, rather than altered Ca²⁺ signaling. The mitochondrial fragmentation was attributed to a decreased mitochondrial fusion rate. This also increased the mitochondrial spare respiratory capacity in Bok^−/− cells, potentially the result of an increased mitochondrial surface (Schulman et al. 2019). The exact mechanism of how Bok, a protein localized at the ER, has such tremendous effect on mitochondria requires further investigation. The impact of Bok on the mitochondria appeared to be independent of its sequestration by IP₃R, which was elegantly shown by using a Bok mutant defective in IP₃R binding. Therefore, these investigators speculated that IP₃R-bound Bok is not only stabilized, but could also maintain the normal rate of mitochondrial fusion, thereby preventing excessive mitochondrial fusion promoted by unbound Bok (Schulman et al. 2019).

Table 1.

Bcl-2-family members as modulators of IP₃R

Protein	Impact on cell fate	Effect on [Ca2+]	Effect on IP3R	Binding site on IP3R	Binding site on Bcl-2 family members	References
Bok	Proapoptotic	No apparent effect on Ca2+-flux properties	Increased stability	Fragment 4 (ARM3 in the modulatory region)	Amino terminus	Schulman et al. 2013
Bcl-2	Antiapoptotic	Suppresses the Ca2+ release from the endoplasmic reticulum (ER), decreasing the ER–mitochondrial Ca2+ transfer	Inhibition	IP3-binding core (IBC); fragment 3 (ARM2 in the modulatory region) and carboxyl terminus	BH4 and terminal transmembrane domain (TMD)	Rong et al. 2009; Monaco et al. 2012; Ivanova et al. 2016; Ivanova et al. 2019
		Promotes prosurvival Ca2+ oscillations, increasing the Ca2+ flux in the mitochondria	Sensitization	Carboxyl terminus	?	Zhong et al. 2006; Eckenrode et al. 2010
		Reduces the [Ca2+]ER, decreasing the ER–mitochondrial Ca2+ transfer				Pinton et al. 2001; Oakes et al. 2005
Bcl-XI	Antiapoptotic	Suppresses the Ca2+ release from the ER, decreasing the ER–mitochondrial Ca2+ transfer	Inhibition	Fragment 3 (ARM2 in the modulatory region) and helix 1 (carboxyl terminus)	Hydrophobic cleft	Yang et al. 2016
		Reduces the [Ca2+]ER, decreasing the ER–mitochondrial Ca2+ transfer	Sensitization	Helices 1 and 4 (carboxyl terminus)	Hydrophobic cleft	Li et al. 2007; Yang et al 2016
		Promotes prosurvival Ca2+ oscillations, increasing the Ca2+ flux in the mitochondria				White et al. 2005; Yang et al. 2016
Mcl-1	Antiapoptotic	Reduces the [Ca2+]ER, decreasing the ER–mitochondrial Ca2+ transfer	Sensitization	Carboxyl terminus	?	Eckenrode et al. 2010
Nrh	Antiapoptotic	Suppresses the Ca2+ release from the ER, decreasing the ER–mitochondrial Ca2+ transfer	Inhibition	IBC (β-TF2 in the ligand-binding domain, LBD)	BH4	Bonneau et al. 2016
Nrz	Antiapoptotic	Suppresses the Ca2+ release from the ER, decreasing the ER–mitochondrial Ca2+ transfer	Inhibition	LBD	BH4, BH1, and BH3	Bonneau et al. 2014

Summarizing the different Bcl-2 family members with their effect on apoptosis and intracellular Ca²⁺, the impact they exert on IP₃R and the binding determinants underlying the protein–protein complexes.

The role of Bok as a Bcl-2-family member in apoptosis remains controversial. The results vary from reporting Bok as a proapoptotic protein that functions in a Bax/Bak-dependent (Echeverry et al. 2013) or independent manner (Llambi et al. 2016), through suggesting that Bok can exert antiapoptotic effects (D'Orsi et al. 2016) and finally to the recent findings indicating the Bok does not directly affect stimuli-induced apoptosis (Schulman et al. 2019). In any case, it is clear that Bok recruitment to IP₃R with concomitant interference of the channel proteolytic cleavage could represent a novel and intriguing mechanism of IP₃R abundance regulation. Furthermore, differences in Bok expression and/or in association with IP₃Rs might account for the different susceptibilities reported for IP₃R cleavage by caspase-3. These findings might also have implications for human cancers, because high-resolution analyses of somatic copy-number alterations from more than 3000 cancer specimens revealed the loss of Bok as a significant alteration (Beroukhim et al. 2010).

Antiapoptotic Bcl-2 Family Proteins as Modulators of IP₃R Activity

Bcl-2

The ability of Bcl-2 to directly modulate IP₃R activity was first observed in T-lymphocytes. Overexpression of Bcl-2 in an immature T-cell model (WEHI7.2 cells), which displays very low endogenous levels of Bcl-2, dampened T-cell receptor (TCR)-induced IP₃R-mediated Ca²⁺ release (Chen et al. 2004). In the same study, an endogenous Bcl-2/IP₃R-protein complex was identified in S49.A2 lymphocytes, and since then, the Bcl-2/IP₃R interaction was further confirmed in various cell models from different origin (Xu et al. 2007; Hanson et al. 2008; Rong et al. 2008; Akl et al. 2013). Moreover, different cancer cell types, including lymphoma, leukemia, lung, and ovarian cancer cells have been described as dependent on the Bcl-2/IP₃R interaction for their survival, because disruption of this complex led to proapoptotic Ca²⁺ signals and cell death (Zhong et al. 2011; Akl and Bultynck 2013; Akl et al. 2013, 2015; Greenberg et al. 2015; Lavik et al. 2015; Xie et al. 2018). The effect of Bcl-2 on Ca²⁺ signaling was extensively studied in T-lymphocytes, in which the TCR-mediated Ca²⁺ release plays a crucial role for cell-fate decisions (Feske 2007; Fracchia et al. 2013; Joseph et al. 2014). Bcl-2 displayed divergent impacts on proapoptotic versus prosurvival Ca²⁺ signals. Although it suppressed the proapoptotic cytosolic Ca²⁺ transients with high amplitude, triggered by strong TCR activation, Bcl-2 did not affect prosurvival Ca²⁺ oscillatory signals, triggered by weak TCR activation (Zhong et al. 2006).

Of note, Bcl-2 was also proposed as able to sensitize IP₃R. This notion was based on Bcl-2's tendency to increase Ca²⁺ oscillation frequency, although not statistically significantly different (Zhong et al. 2006). This observation relates to another study, in which the open probability (P_o) of single IP₃R channels and the IgM-induced Ca²⁺ oscillations in DT40 cells were monitored. The analyses showed that Bcl-2 was capable of promoting IP₃R activity and enhancing the frequency of Ca²⁺ oscillations (Eckenrode et al. 2010). The proposed Bcl-2-mediated increase of IP₃R activity has also been linked to a lower steady-state ER Ca²⁺ content in cells overexpressing Bcl-2. Effectively, the increase of IP₃R activity acts as constitutive ER Ca²⁺ leak (Oakes et al. 2005). The ability of Bcl-2 to lower steady-state ER Ca²⁺ levels was originally proposed to be the key component of its antiapoptotic effect (Pinton et al. 2001), although now it is clear that Bcl-2 can fulfill its antiapoptotic function in cells while not affecting the Ca²⁺ store content (Hanson et al. 2008; Monaco et al. 2012b; Ivanova et al. 2016). Here, the increased IP₃R activity was explained by more binding of Bcl-2 to the IP₃R and an enhanced protein kinase A–dependent phosphorylation state of the IP₃R1, thereby sensitizing IP₃R-mediated Ca²⁺ release to basal concentrations of IP₃. In that context, it is important to mention that Bcl-2 can regulate the phosphorylation state of the IP₃R by protein kinase A, via its scaffolding of DARPP-32 and calcineurin (Chang et al. 2014a). Thus, Bcl-2 interaction with IP₃Rs appears to be modulated by additional factors and proteins.

Bcl-2-mediated decrease of ER Ca²⁺ stores has also been shown to be dependent on other proteins besides IP₃Rs, including Bax Inhibitor-1 (BI-1) and CISD2. First, Bcl-2 overexpression failed to lower ER Ca²⁺ levels in cells lacking BI-1, the founding member of the transmembrane Bax inhibitor motif (TMBIM)–containing protein family (Xu et al. 2008). BI-1 and the other TMBIM-family members function as ancestral regulators of cell death (Rojas-Rivera and Hetz 2015; Carrara et al. 2017; Liu 2017). This function is tightly connected to Ca²⁺ dynamics, because BI-1 operates as an ER Ca²⁺-leak channel (Bultynck et al. 2012, 2014), a function conserved during evolution (Chang et al. 2014b; Guo et al. 2019). Second, CISD2, a protein involved in longevity (Chen et al. 2009; Wu et al. 2012), breast cancer (Sohn et al. 2013), and Wolfram syndrome type 2 (Rigoli and Di Bella 2012) has been identified to be present in IP₃R/Bcl-2 multiprotein complexes, where it enables Bcl-2 to lower the ER Ca²⁺ content (Chang et al. 2010). Thus, also in cells lacking CISD2, Bcl-2 fails to reduce the ER Ca²⁺ levels.

As we discussed, the modulation of IP₃R activity by Bcl-2 is complex, with alternative outcomes being proposed: inhibition and sensitization with reduced ER Ca²⁺ content. The molecular mechanism of interaction between Bcl-2 and IP₃R has been best studied in the context of channel inhibition. The interaction between these proteins relies on complex multidomain-binding determinants (Fig. 2; Table 1; Parys 2014). IP₃R inhibition (Fig. 3) was initially attributed to binding between Bcl-2's BH4 domain and the fragment 3 of the IP₃R (aa 923–1581) (Rong et al. 2008, 2009a,b; Ivanova et al. 2017). This interaction appeared to be highly conserved during vertebrate evolution (Ivanova et al. 2017). Based on detailed molecular insights in the Bcl-2-binding determinants within the fragment 3, the precise location of the Bcl-2-binding site was identified (aa 1389–1408). This 20-aa region, located in the ARM2 domain in the central, modulatory region of IP₃R1, is highly conserved among all three IP₃R isoforms and during evolution (Monaco et al. 2012b). Consistent with this, Bcl-2 binding to the IP₃R was observed for all three IP₃R isoforms. A peptide corresponding to the Bcl-2-binding site on IP₃R1 (IP₃R-derived peptide [IDP]) was able to disrupt endogenous IP₃R/Bcl-2 complexes, antagonizing the ability of Bcl-2 to inhibit IP₃Rs in a variety of experimental models (Rong et al. 2008; Zhong et al. 2011; Akl et al. 2013). These findings have recently been corroborated by an electrophysiological analysis performed on giant unilamellar vesicles prepared from Bcl-2-overexpressing WEHI7.2 cells (Shapovalov et al. 2017). Addition of IDP-augmented IP₃-induced channel activity caused by a dramatic increase in the P_o of IP₃Rs. A cell-permeable variant of the IDP peptide (TAT-IDP) enhanced Ca²⁺-dependent cell death in T-cell models (cell death was triggered by strong TCR activation) (Rong et al. 2008). A stabilized, proteolysis-resistant, form of TAT-IDP (Bcl-2/IP₃ receptor disruptor-2, BIRD-2) provoked cell death by itself in a variety of cancer cell models, particularly in those that are less sensitive to venetoclax, a BH3-mimetic Bcl-2 inhibitor (Vervloessem et al. 2017b). In diffuse large B-cell lymphoma cells, BIRD-2-induced cell death depended on the combination of high IP₃R2 expression levels (Akl et al. 2013) and of elevated chronic IP₃ signaling, which occurs downstream from the B-cell receptor in these cells (Bittremieux et al. 2019). Furthermore, recent work indicated that both Ca²⁺ release from the ER and influx of extracellular Ca²⁺ contributed to BIRD-2-induced cell death (Bittremieux et al. 2018). A further in-depth discussion on the IP₃R-mediated function of Bcl-2 in cancer (Akl et al. 2014) and its targeting by BIRD-2 has been provided in recent reviews (Distelhorst 2018; Kerkhofs et al. 2019a,b). The investigative path that resulted in IP₃R-derived peptides as novel Bcl-2 inhibitors with anticancer potential and how this inspired the development of novel first-in-class small-molecule Bcl-2 inhibitors that provoke cell death in chronic lymphocytic leukemia through Ca²⁺ overload was reviewed elsewhere in this collection (Distelhorst and Bootman 2019).

Figure 3.

Overview of Bcl-2's and Bcl-Xl's effects on inositol 1,4,5-trisphosphate receptor (IP₃R), ryanodine receptor (RyR), and voltage-dependent anion channel (VDAC) activity. The various regulatory mechanisms are illustrated in four cells with their endoplasmic reticulum and mitochondria. Bcl-2, depicted in yellow is able to inhibit all three of the discussed intracellular Ca²⁺ channels: (A) IP₃R (gray), RyR (blue), and (B) VDAC (purple). Depending on its concentration, Bcl-Xl (green), could sensitize (C) or inhibit IP₃R (D). Similarly to Bcl-2, Bcl-Xl also inhibits RyR (C). Finally, the effect of Bcl-Xl on VDAC is still under debate as both stimulation and inhibition were reported (B).

As stated above, the interaction between Bcl-2 and IP₃R is multimodal, underlying the complex IP₃R channel modulation by Bcl-2. Such interaction profiles enable complex signaling outputs. For instance, Bcl-2 fails to inhibit IP₃R-mediated Ca²⁺ signaling provoked by supramaximal concentrations of (1) IP₃ in ⁴⁵Ca²⁺ fluxes using permeabilized cells and in single IP₃R channel recordings using nuclear patch clamping, or (2) agonist in intact cells using Ca²⁺ measurements at the single-cell level (Hanson et al. 2008; Ivanova et al. 2019). The molecular basis for the ability of Bcl-2 to sensitize or inhibit IP₃R activity depending on the level of IP₃/agonist stimulation were not clear until very recently, when a novel interaction between the BH4 domain of Bcl-2 and the LBD of IP₃R1, in particular the IBC, was identified (Fig. 2; Ivanova et al. 2019). Furthermore, the interaction was antagonized by physiological (IP₃) and pharmacological (adenophostin A) IP₃R agonists that target the LBD. Vice versa, the BH4 domain of Bcl-2 decreased the ligand binding to the LBD in in vitro Förster resonance energy transfer (FRET) measurements. This finding is consistent with a previous observation that Bcl-2 overexpression interfered with the ligand–receptor interaction (Hanson et al. 2008). As such, a mutual antagonism between Bcl-2 and IP₃ for controlling IP₃R channel activity seems to exist. In line with the existence of this second BH4 domain-binding region, Bcl-2 was able to inhibit a mutated IP₃R1 channel that lacked the binding site in the fragment 3 (IP₃R^Δ1389-1408). Although, at first sight, this study challenges the existing model of IP₃R inhibition by Bcl-2, whereby the BH4 domain and its binding site in fragment 3 are the driving force, notably compared with the wild-type channel, the inhibition of IP₃R^{Δ1389–1408} was less prominent. Thus, instead of contradicting, these new findings might represent an additional layer of the complex mechanism of IP₃R modulation by Bcl-2. This is also supported by the 3D cryo-EM structure of IP₃R1 (Fan et al. 2015, 2018), where aa 1389–1408 from each subunit appeared to be located relatively close to the LBD from the neighboring IP₃R1 subunit. However, the importance of both BH4 domain-binding sites, that is, LBD and aa 1389–1408, and their interplay with Bcl-2, require further investigation. A fascinating consequence of Bcl-2 binding to the LBD of IP₃R is that in theory one Bcl-2 molecule per channel could be sufficient to inhibit IP₃R activity, because all four subunits ought to be occupied by IP₃ to open the channel (Alzayady et al. 2016; Taylor and Konieczny 2016). Yet, further work will be needed to determine the stoichiometry of IP₃R/Bcl-2 complexes and the minimal ratio of Bcl-2 proteins per IP₃R to impose IP₃R inhibition.

As the BH4 domain of Bcl-2 is crucial for the modulation of IP₃R, different studies were dedicated to elucidate its unique characteristics. It is clear that Bcl-2's BH4 domain is necessary and sufficient for IP₃R inhibition (Rong et al. 2009b). As such, Bcl-2 lacking its BH4 domain did not bind to the ARM2 in the fragment 3 of IP₃R. A number of critical surface-accessible residues (R6, S7, D10, R12, K17, H20, Y21, Q25, R26, Y28), which contribute to the inhibitory effect of the BH4 domain on IP₃R, were identified (Rong et al. 2008, 2009a; Monaco et al. 2012b). Among them, the K17 is of a particular interest. This positively charged residue represents a critical difference with the BH4 domain of the other major antiapoptotic member, Bcl-Xl, where it corresponds to a negatively charged D11. In contrast to the BH4 domain of Bcl-2, the BH4 domain of Bcl-Xl displayed poor IP₃R-binding properties and was much less effective in inhibiting IP₃Rs. Interestingly, a mutated version of the BH4 domain of Bcl-2 (BH4-Bcl-2 K17D), which mimics the BH4 domain of Bcl-Xl, was severely compromised in binding and inhibiting IP₃Rs. The K17 was important also at the full-length protein level, because the full-length Bcl-2 mutant K17D was impaired in inhibiting agonist-induced Ca²⁺ release within intact cells and preventing apoptosis, triggered by staurosporine (Monaco et al. 2012b). The α-helical properties of the BH4 domain of Bcl-2 have also been found to be essential for IP₃R binding and inhibition. Indeed, replacing I14 and V15, two α-helical backbone residues, by two G residues reduced the propensity of the BH4 domain to form an α-helix. These mutations strongly impaired the interaction of the BH4 domain with fragment 3 of the IP₃R and the ability of the BH4 domain to inhibit IP₃R function and to protect against Ca²⁺-dependent apoptosis (Monaco et al. 2013). The I14G/V15G mutated version of full-length Bcl-2 also abrogated Bcl-2's ability to inhibit IP₃R-mediated Ca²⁺ release. However, the underlying mechanism appeared complex. Indeed, the protein levels of ectopically expressed Bcl-2 I14G/V15G were consistently lower than those of ectopically expressed wild-type Bcl-2. Additional biophysical experiments indicated that I14G/V15G mutations impacted Bcl-2's conformation, caused a destabilization, and shortened half-life of the protein in cells (Monaco et al. 2018). This indicates that the BH4 domain is also important for the overall stability of the Bcl-2 proteins. Of importance, Bcl-2 K17D was not severely destabilized, indicating that the reduced ability of Bcl-2 K17D to inhibit IP₃Rs was not implicitly the result of a decrease in Bcl-2 stability. Furthermore, these data also hold a warning on the interpretation of results obtained with BH4 domain mutants, as their ability to abrogate a particular function of Bcl-2 may not imply a specific prevention of an interaction or a novel function/target for the BH4 domain itself, but rather an effect on overall Bcl-2's structural organization and/or stability. Consequently, Bcl-2 I14G/V15G failed to bind Bax and to inhibit IP₃R, both contributing to its reduced antiapoptotic properties (Monaco et al. 2018).

Besides the central, modulatory region and the LBD, the carboxyl terminus of IP₃R has also been proposed as a Bcl-2-binding site. The sixth TMD appeared as a critical binding determinant, because Bcl-2 displayed efficient binding only to a GST protein, corresponding to the carboxy-terminal tail together with the sixth TMD of IP₃R1 (aa 2512–2749 according to the mIP₃R1 sequence), but not to the carboxy-terminal IP₃R fragment lacking the sixth TMD (aa 2590–2749 according to the mIP₃R1 sequence) (Rong et al. 2008; Eckenrode et al. 2010; Monaco et al. 2012a). Bcl-2 binding to the carboxyl terminus of IP₃R was associated with sensitization of the channel (Eckenrode et al. 2010) and it was proposed that the hydrophobic cleft of Bcl-2 might be involved in the interaction (Yang et al. 2016). Using genetic and pharmacological approaches to neutralize the hydrophobic cleft of Bcl-2, this part of the protein appeared dispensable for the binding and the regulation of IP₃R. Consistent with that, BH3-mimetic drugs that occupy Bcl-2's hydrophobic cleft and abolish the interaction with proapoptotic Bcl-2-family members did not cause Ca²⁺ dysregulation and did not alter agonist-induced Ca²⁺ increases in a variety of cell lines, cancer cells, and primary cells such as pancreatic acinar cells (Vervloessem et al. 2017a; Jakubowska et al. 2018). In contrast to the hydrophobic cleft, the TMD of Bcl-2 was identified as a binding partner of the carboxyl terminus of IP₃R and this interaction was required for the efficient IP₃R inhibition. It is anticipated that the binding between both IP₃R and Bcl-2 carboxyl termini augments the local concentration of Bcl-2 and thus of its BH4 domain in the proximity of the IP₃R, and in particular of the fragment 3 binding site, enabling in cellulo interaction and inhibition of Ca²⁺ release (Ivanova et al. 2016). It is important to note that the isolated BH4 domain displays a rather low-affinity inhibition of the IP₃R (half-maximal inhibitory concentration [IC₅₀] value of ∼30 µm) and a binding affinity in the low µm range for the isolated fragment 3 and IBC of IP₃R1. Thus, inhibition of IP₃Rs by Bcl-2's BH4 domain likely requires a high local concentration of Bcl-2 in a cellular context.

Bcl-Xl

Bcl-Xl has been shown to bind to IP₃R1, IP₃R2, and IP₃R3 (Fig. 2; Table 1; White et al. 2005). The functional relevance of the Bcl-Xl–IP₃R interaction was studied by patch clamp electrophysiological experiments, using Sf9 (White et al. 2005) and DT40 (Li et al. 2007) cells, where the P_o of IP₃Rs was measured. Purified recombinant Bcl-Xl, applied in the low µm range, enhanced the channel activity triggered by submaximal IP₃ concentration in both models (White et al. 2005; Li et al. 2007), but did not display any significant effect when IP₃R was activated by supramaximal (IP₃) in Sf9 cells (White et al. 2005). Furthermore, Bcl-Xl overexpression in DT40 cells increased the frequency of spontaneous and IgM-induced Ca²⁺ oscillations and the number of oscillating cells (White et al. 2005). Similar results were obtained in a study that aimed to decipher the impact of Bcl-Xl on the various IP₃R isoforms (Li et al. 2007). Recombinant Bcl-Xl was able to sensitize each IP₃R isoform to low concentration of IP₃ in patch clamp experiments, causing an increase in the P_o of the IP₃R. Consistent with this, the overexpression of Bcl-Xl in DT40 cells with a triple knockout of all three IP₃R isoforms (DT40-TKO) with reconstituted IP₃R1, IP₃R2, or IP₃R3 resulted in an increased frequency of the spontaneous Ca²⁺ oscillations. Interestingly, although Bcl-Xl appeared to decrease the ER Ca²⁺ content and the amplitude of IgM-induced Ca²⁺ signals in DT40-TKO cells expressing IP₃R3, overexpression of Bcl-Xl resulted in suppressed IgM-induced apoptosis in DT40-TKO cells expressing any of the three IP₃R isoforms (Li et al. 2007). This was in line with the observation that Bcl-Xl protected DT40 cells against IgM-induced apoptosis by promoting mitochondrial bioenergetics leading to cell survival (White et al. 2005). Moreover, it was shown that Bcl-Xl-mediated protection against cellular toxins was strongly dependent on the expression of IP₃R, as Bcl-Xl overexpression particularly protected DT40-TKO cells when IP₃Rs were reintroduced (Li et al. 2007).

More light was shed on the mechanism of IP₃R regulation by Bcl-Xl by a study elucidating the differential effect of Bcl-Xl levels in patch clamp experiments. Low Bcl-Xl concentrations, in a range with a maximal potentiating effect at 1 µm, enhanced IP₃R activity, whereas high concentrations (10 µm or more) inhibited the channels (Fig. 3; Yang et al. 2016). These divergent effects of Bcl-Xl were underscored by its binding to alternative IP₃R regions with different affinities. In contrast to Bcl-2, Bcl-Xl displayed a much higher binding efficiency for the carboxy terminal tail fragment compared with the central fragment 3 (Monaco et al. 2012a; Yang et al. 2016). The carboxy-terminal region of IP₃R contains two adjacent sites targeted by Bcl-Xl, helix 1, and helix 4, which both contain BH3-domain-like features. Bcl-Xl binding to both helixes was required for the activation of IP₃R, whereas Bcl-Xl binding to helix 1 and to ARM2 domain in the fragment 3 resulted in inhibition of the channel, as measured by the decrease of the P_o (Yang et al. 2016). Disturbance of the interactions of Bcl-Xl with the carboxyl terminus of IP₃R affected cell survival. Indeed, DT40-TKO cells expressing IP₃Rs that carry a mutation in the BH3-domain-like sequences of helix 1 (F2585A D2590A) and helix 4 (L2716A D2721A) were more sensitive to proapoptotic stimuli than DT40-TKO cells expressing wild-type IP₃Rs, at least when these IP₃Rs were expressed at low-to-medium levels. In addition, abrogating the high-affinity interaction of Bcl-Xl with IP₃Rs using cell-permeable peptides representing the helices 1 and 4 of IP₃R enhanced the apoptotic effect of staurosporine in DT40 cells. Finally, helix 1 and 4 as peptides were able to induce cell death by themselves in breast cancer cells and in Burkitt's lymphoma cells (Yang et al. 2016). However, it is not clear whether the cell death effects were solely caused by disruption of the IP₃R/Bcl-Xl complexes or may be in part caused by the disturbance of Bcl-Xl complexes with proapoptotic Bcl-2-family members at the mitochondrial membranes.

As stated above, IP₃R inhibition by Bcl-Xl requires the simultaneous binding of Bcl-Xl to fragment 3 and to helix 1 in the carboxyl terminus. Previously, it was shown that the BH4 domain of Bcl-Xl lacked sufficient affinity or selectivity to bind fragment 3 and inhibit IP₃R (Monaco et al. 2012b). Yet, in a recent study, it was speculated that the inhibitory interaction might be executed by Bcl-Xl's BH4 domain, because a synthetic peptide, corresponding to the BH4 domain, could antagonize IP₃R inhibition by high concentrations of Bcl-Xl (Yang et al. 2016). Of note, in a study elucidating the role of Bcl-Xl at ER–mitochondrial contact sites, the BH4 domain and the TMD of Bcl-Xl were proposed to be involved in the interaction with IP₃R, because mutants lacking either of these domains displayed decreased binding to the receptor (Williams et al. 2016). Indeed, the lack of the TMD caused a ∼90% reduction in the binding, suggesting a crucial role for Bcl-Xl localization at the ER–mitochondria contact sites, where it would bind to and regulate IP₃R activity. Nevertheless, deletions of the BH4 domains of Bcl-Xl/Bcl-2 might result in conformational changes and less stable proteins. The binding to the carboxy-terminal tail of the IP₃R occurred via Bcl-Xl's hydrophobic cleft, because mutations that abrogate this functional domain resulted in failure of Bcl-Xl to form complexes with both IP₃R helices (helix 1 and helix 4) identified as its binding partners (Yang et al. 2016). Furthermore, such mutated Bcl-Xl proteins were severely impaired in regulating IP₃R function. The contribution of the hydrophobic cleft of Bcl-Xl was also emphasized by experiments using ABT-737, antimycin A, and a peptide derived from the BH3 domain of Bax, which all allegedly target the hydrophobic cleft. All these compounds prevented the effects of recombinant Bcl-Xl on IP₃R activity (Yang et al. 2016). This is in line with a previous study reporting that recombinant Bax and Bid disrupt the interaction of Bcl-Xl and IP₃R and overcome the sensitizing effect of Bcl-Xl on the channel activity (White et al. 2005).

Mcl-1

Similarly to Bcl-2 and Bcl-Xl, Mcl-1 also binds to and regulates IP₃Rs (Fig. 2; Table 1). Mcl-1 interacts with all three isoforms of the IP₃R, targeting their carboxyl terminus. Mcl-1 overexpression in DT40 cells resulted in a decreased Ca²⁺ level in the ER, suggesting that the sensitivity of the IP₃R is enhanced by the presence of Mcl-1. Furthermore, Mcl-1 displayed a positive effect on spontaneous Ca²⁺ oscillations, as well as those induced by IgM, with increased peak amplitudes and total amount of released Ca²⁺. Finally, Mcl-1 overexpression in DT40 cells resulted in enhanced protection against staurosporine- or etoposide-induced apoptosis compared with DT40-TKO cells, suggesting that IP₃R potentiates the antiapoptotic activity of Mcl-1 (Eckenrode et al. 2010). However, at this point, not much is known about the molecular determinants responsible for IP₃R/Mcl-1 complex formation and more work is required to reveal them.

Bcl-2L10 and Its Orthologs

Nrz, a zebrafish ortholog of the antiapoptotic Bcl-2 homolog Bcl-2L10, has been shown to decrease IP₃ binding to the IP₃R, a process implicated in the regulation of epiboly in zebrafish embryos (Popgeorgiev et al. 2011; Bonneau et al. 2014). Via its BH4 domain, Nrz directly binds to the ligand-binding region of the IP₃R (Fig. 2; Table 1), thereby antagonizing the interaction of IP₃ to its receptor (Bonneau et al. 2014). Because Ca²⁺ release occurs only when the four IP₃R monomers are occupied by IP₃ (Alzayady et al. 2016), such interference with IP₃ binding to the ligand-binding region might be exploited for achieving efficient inhibition of the IP₃R channel activity, as is the case with Nrz. This prevents the phosphorylation of myosin light chain and the premature contraction of the actin–myosin ring, thereby ensuring epiboly (Popgeorgiev et al. 2011; Bonneau et al. 2014). The efficient IP₃R inhibition by Nrz required the presence of the BH3 and BH1 domains in addition to the BH4 domain. Furthermore, only phosphorylated Nrz was able to bind and thereby inhibit IP₃R (Bonneau et al. 2014), confirming the complex mechanism of IP₃R modulation by Bcl-2 proteins.

The human ortholog of Bcl-2L10, Nrh, was also shown to bind through its BH4 domain to the LBD of IP₃R and to decrease Ca²⁺ release (Fig. 2; Table 1; Bonneau et al. 2016). Nrh appeared to participate in a three-protein complex together with IP₃R and IP₃R-binding protein released with IP₃ (IRBIT), resulting in inhibition of IP₃R activity. The investigators speculated that one part of the BH4 helix is involved in the interaction with the LBD of IP₃R, whereas the other part binds to IRBIT. Nrh and IRBIT colocalized at the membrane contact sites between ER and mitochondria. IRBIT appeared to promote the formation of ER–mitochondrial contact sites. Furthermore, IRBIT and Nrh cooperated to suppress the Ca²⁺ transfer between ER and mitochondria, thereby conferring apoptotic resistance. In cells undergoing apoptosis, IRBIT becomes dephosphorylated and promotes apoptosis. First, dephosphorylated IRBIT fails to bind IP₃Rs and thus does not compete with IP₃ anymore, which potentially increases IP₃-induced Ca²⁺ release. Second, dephosphorylated IRBIT antagonized Nrh's ability to bind and inhibit IP₃Rs, thereby counteracting its antiapoptotic effects on IP₃R channels. The investigators proposed a model in which dephosphorylated IRBIT, which still binds to Nrh, but not to IP₃R, displaces Nrh from IP₃Rs. As a consequence, ER–mitochondrial Ca²⁺ transfer will increase and promote apoptotic cell death (Bonneau et al. 2016). In this context, a recent study suggested that targeting Nrh interaction with IP₃R can be a promising strategy in breast cancer treatment (Nougarede et al. 2018).

OTHER INTRACELLULAR Ca²⁺ CHANNELS REGULATED BY Bcl-2 PROTEINS

Structure and Function of RyRs and Their Regulation by Bcl-2 Proteins

RyRs form another protein family of Ca²⁺ release channels, which are present at the ER or sarcoplasmic reticulum (SR) membranes of several, although not all, cell types (Lanner et al. 2010). RyRs are large tetrameric Ca²⁺-release channels (monomers >500 kDa). Similar to the IP₃Rs, three isoforms of RyRs are known to exist in vertebrates. These isoforms have a distinct expression pattern with RyR1 predominantly expressed in skeletal muscle cells (Lai and Meissner 1989), RyR2 in the heart (Imagawa et al. 1989), whereas all three isoforms are expressed in different parts of the brain (Martin et al. 1998). RyRs are also expressed at lower levels in pancreatic acinar cells and β cells, T cells, smooth muscle cells, and liver cells (Lanner et al. 2010). A major activator of RyRs is Ca²⁺ itself (Meissner et al. 1997). As such, RyRs are often seen as amplifiers of smaller Ca²⁺ signals, which may originate at different sites of the cell. In the heart for instance, Ca²⁺ influx over the plasma membrane through voltage-gated L-type Ca²⁺ channels triggers RyR2 activation via Ca²⁺-induced Ca²⁺ release, which then provides the Ca²⁺ increase necessary for contraction. Given the restricted expression pattern of RyRs, these channels contribute to specialized cellular functions, including muscle contraction, memory formation, and secretion of insulin and digestive enzymes (Unni et al. 2004; Lanner et al. 2010; Lanner 2012; Rebbeck et al. 2014; Llanos et al. 2015).

Excessive RyR-mediated Ca²⁺ release is known to contribute to the progression of several diseases (Lanner 2012). Acute pancreatitis, for instance, is characterized by excessive intracellular Ca²⁺ release, which triggers the premature activation of digestive enzymes, resulting in necrosis of the pancreatic acinar cells (Gerasimenko et al. 2014). Inhibition of IP₃R- (Huang et al. 2017) or RyR-mediated Ca²⁺ release (Husain et al. 2012) has been shown to inhibit necrosis, and as such may be beneficial for treating acute pancreatitis. The excessive Ca²⁺ release responsible for the onset of acute pancreatitis is usually triggered by the presence of toxic agents, such as alcohol metabolites, bile acids, and certain chemotherapy (Petersen and Sutton 2006; Peng et al. 2016), resulting in hyperactivation of the RyRs. However, in other diseases, like malignant hyperthermia and catecholaminergic polymorphic ventricular tachycardia, mutations in RyR1 and RyR2 are very well known to be the underlying cause for the diseases (Yano et al. 2006). Three mutational hotspots exist on RyR1 and RyR2 in which more than 100 point mutations have been identified. These mutations lead to aberrant RyR-mediated Ca²⁺ release, which is causal in these diseases. As such, compounds that stabilize/normalize RyR-mediated Ca²⁺ release under disease conditions may hold significant therapeutic value.

Bcl-2

Recently, it was shown that RyR activity can be modulated by antiapoptotic Bcl-2 proteins. Sequence alignment revealed high similarity between the 20 aa of the Bcl-2-binding site identified in the regulatory region of the IP₃R, and a stretch of 22 aa in the central region of the RyR (Vervliet et al. 2014). Interestingly, this 22-residue (aa 2448–2469 for rabbit RyR1) region is located within one of the previously described mutational hotspots contributing to malignant hyperthermia and catecholaminergic polymorphic ventricular tachycardia (Yano et al. 2006). The region also contains a proposed FKBP12.6-binding site, a major regulator of RyR activity (Marx et al. 2000; Vervliet et al. 2015c), further highlighting the importance of this region for channel regulation.

In line with the high degree of similarity between the described amino acid stretches in the central regions of IP₃R and RyR, it was shown that full-length Bcl-2 binds to RyR1 and RyR3 in human embryonic kidney (HEK) cells overexpressing these particular RyR isoforms. Furthermore, the interaction occurs between the BH4 domain of Bcl-2 and the central region of RyRs, as shown using GST-fused fragments corresponding to the central region of each RyR isoform. Consistently, Bcl-2 appeared to be a potent inhibitor of RyR-mediated Ca²⁺ release (Fig. 3) without altering the ER store content (Vervliet et al. 2014). In addition, a completely endogenous interaction between RyRs and Bcl-2 was observed in lysates from rat hippocampus. Introducing the BH4 domain of Bcl-2 into HEK cells overexpressing RyRs or in dissociated hippocampal neurons inhibited RyR activity. More details of the Bcl-2–RyR interaction at the molecular level were revealed and it appeared that the hydrophobic cleft of Bcl-2 does not contribute to the interaction with RyR (Vervliet et al. 2015b), which is another similarity to its binding to IP₃R. Despite the analogy in the mechanisms of interaction of Bcl-2 with the two channels, a critical difference at the level of the BH4 domain was revealed. Strikingly, the above-described mutant Bcl-2 K17D, which was impaired in binding and inhibiting IP₃R (Monaco et al. 2012b), retained its ability to interact with RyR and to inhibit its function (Vervliet et al. 2014).

Bcl-Xl

The other major regulator of IP₃Rs from the Bcl-2 family of proteins, Bcl-Xl also modulates the activity of RyRs (Fig. 3). In contrast to the interaction with IP₃Rs, in which the BH4 domain of Bcl-Xl appeared dispensable for the complex formation, here the BH4 domain binds to full-length RyRs. Moreover, it targets the same region in RyR where Bcl-2 binds (Vervliet et al. 2015a). In addition to the BH4 domain, K87, a residue located in the beginning of the BH3 domain of Bcl-Xl, was required for efficient binding and regulation of RyR (Vervliet et al. 2015a). While Bcl-Xl appeared to modulate IP₃R activity, depending on its concentration (Yang et al. 2016), only an inhibitory effect was observed when studying the impact of full-length Bcl-Xl or of its BH4 domain on RyR activity (Vervliet et al. 2015b). Nevertheless, the relation between the level of Bcl-Xl and its effect on RyR was not addressed.

The above-discussed studies suggest that the RyR-inhibitory abilities of the BH4 domains of Bcl-2 and Bcl-Xl could be good starting points for developing peptidomimetic drugs for treating diseases with excessive RyR-mediated Ca²⁺ release. This idea was recently illustrated in the framework of acute pancreatitis. Using primary isolated pancreatic acinar cells from mice, it was shown that in the acinar cells the BH4 domains of Bcl-2 and Bcl-Xl inhibit physiological RyR-mediated Ca²⁺ release (Vervliet et al. 2018). In addition, both BH4 domains were similarly capable of inhibiting pathological bile acid–induced Ca²⁺ release and subsequent necrosis, two hallmarks of acute pancreatitis. These results show that the BH4 domains, and/or drugs derived from them, may be therapeutically relevant in diseases caused by excessive RyR activity.

Structure and Function of VDACs and Their Regulation by Bcl-2 Proteins

VDACs are members of the protein family of porins, which are mainly located on the OMM. Three isoforms of VDACs, with ∼70% of sequence similarity and a molecular weight of ∼30 kDa (Messina et al. 2012), have been identified. VDACs are best known for acting as large conductance channels that permeate metabolites, nucleotides, and ions, such as Ca²⁺, across the OMM (Magrì et al. 2018). Each isoform is widely expressed in most mammalian tissues (Messina et al. 2012). Exceptionally, VDAC2 and VDAC3 are highly expressed in testis, whereas mouse VDAC1 is poorly expressed in this tissue. Nevertheless, VDAC1 is the most abundant isoform, and is ubiquitously expressed across all other tissue types. This isoform contributes the most to mitochondrial Ca²⁺ fluxes in concert with the mitochondrial Ca²⁺ uniporter that in turn allows for Ca²⁺ transport across the inner mitochondrial membrane (De Stefani et al. 2012; Nemani et al. 2018). This Ca²⁺ transfer to the mitochondria is further facilitated by an IP₃R–VDAC1 physical link provided by the chaperone glucose-regulated protein 75 (Szabadkai et al. 2006). Additionally, in cardiomyocytes, proper Ca²⁺ transfer from SR to the mitochondria is ensured by the interaction between RyR2 and VDAC2 (Min et al. 2012; Naghdi and Hajnóczky 2016).

Close regulation of mitochondrial Ca²⁺ uptake via VDACs is critical for cell life or death by maintaining mitochondrial energy production or driving mitochondrial Ca²⁺ overload, respectively (Shoshan-Barmatz et al. 2018). Studies using various cell models (e.g., HeLa, MEF, skeletal myotubes, endothelial cells) showed that ectopic overexpression of VDAC1 leads to increased levels of mitochondrial Ca²⁺ and apoptosis under both resting and stress conditions (Rapizzi et al. 2002; Yuan et al. 2008; Monaco et al. 2015; Shoshan-Barmatz et al. 2018). Conversely, VDAC1 genetic silencing attenuates mitochondrial Ca²⁺ uptake and cell apoptosis induced by several agents or stressors (e.g., H₂O₂, ceramide) (Yuan et al. 2008; De Stefani et al. 2012; Shoshan-Barmatz et al. 2018). Intriguingly, a variety of pathological conditions associated with aberrant intracellular Ca²⁺ signaling are characterized by increased VDAC1 expression levels in the affected tissues. Pathological overexpression of VDAC1 is, for instance, observed in cancer (Shoshan-Barmatz and Ben-Hail 2012; Shoshan-Barmatz and Golan 2012), cardiovasculopathies (Branco et al. 2011; Liao et al. 2015), Alzheimer disease (Manczak et al. 2006; Cuadrado-Tejedor et al. 2011), and type 2 diabetes (Ahmed et al. 2010; Zhang et al. 2019).

VDAC1 is composed of 19 antiparallel β-strands that form a transmembrane β-barrel pore (Fig. 4A; Hiller et al. 2008; Zeth and Zachariae 2018). The amino-terminal domain of this channel, encompassing the first 25 residues, is a short α-helical stretch that flips in and out of the pore lumen (Geula et al. 2012; Shuvo et al. 2016). A glycine-rich sequence (²¹GYGFG²⁵) was proposed to provide the flexibility required for such properties. It remains under debate whether the amino terminus of VDAC1, and/or part of its pore wall, would serve as a voltage sensor (Song et al. 1998; Teijido et al. 2012). However, it is now generally accepted that the high flexibility of both the amino-terminal α-helix (Mertins et al. 2012; Shuvo et al. 2016) and the VDAC1 β-barrel (Grosse et al. 2014; Ge et al. 2016; Zeth and Zachariae 2018) underlies the ion permeation and gating mechanisms of VDAC1. A narrower, partially “closed state” of the VDAC1 pore, was described as favorable for Ca²⁺ and cations’ conductance while obstructing the passage of metabolites and anions (Tan and Colombini 2007; Shoshan-Barmatz et al. 2018). Further experiments, exploiting reagents known to specifically interact with several Ca²⁺-binding proteins (Gincel et al. 2002; Israelson et al. 2005), revealed that VDAC1 is equipped with two low-affinity Ca²⁺-binding sites shaped by the pore β-strands 4, 13, and 14 (residues E66, E73, E189, and E203) (Israelson et al. 2007, 2008). Pertinently, cellular events or agents (e.g., apoptosis-inducing agents) that promote an increase in cytosolic Ca²⁺ have been shown to increase VDAC1 permeability to Ca²⁺ (Bathori et al. 2006), decrease channel plasticity (Ge et al. 2016), and in many cases stimulate VDAC1 protein expression and its proapoptotic oligomerization (Weisthal et al. 2014; Shoshan-Barmatz et al. 2018). Finally, some of the VDAC1 regions previously assumed to be loop regions (namely, LP1, LP2, LP3, and LP4 and including β-strands 4, 7, 11, and 14, respectively), as well as the amino-terminal α-helix itself (Fig. 4B), might act as a docking site for several cytosolic-modulatory proteins, including anti- and proapoptotic Bcl-2 family proteins (Hiller et al. 2008; Geula et al. 2012).

Figure 4.

Voltage-dependent anion channel 1 (VDAC1) structure. (A) Linear sequence of hVDAC1 (UniProtKB, P21796). The amino acids encompassing for the 19 b-stands (b1–b19) are highlighted in light blue. The amino-terminal (N-term) and the loop peptide (LP) regions (LP1–4) are shown above in red bars. The four key residues allegedly encompassing the Ca²⁺-binding properties of VDAC1 (E66, E73, E189, and E203) are boxed in red. (B,C) Color-coded views of the structure model of hVDAC1 (PDB:2K4T) with the key regions involved in its interaction with Bcl-Xl highlighted in color: N-term (green), LP4 (red), β17 (light blue), and β18 (orange). A top view is presented in B and a side view in C. The latter structure is membrane-embedded to facilitate the visualization of the possible interaction interfaces.

Bcl-2, Bcl-Xl, and Mcl-1

Beyond their role as modulators of ER-resident Ca²⁺ channels, Bcl-2, Bcl-Xl, and Mcl-1 have the ability to interact with VDAC1. In experimental settings precluding the contribution of IP₃Rs, some studies found that the antiapoptotic Bcl-2 members inhibit Ca²⁺ uptake into the mitochondria and protect cells from OMM permeabilization induced by Ca²⁺ overload (Tornero et al. 2011; Arbel et al. 2012; Huang et al. 2014; Monaco et al. 2015). Alternatively, other reports indicated that Bcl-Xl and Mcl-1 could promote VDAC1-mediated Ca²⁺ transfer into the mitochondria (Fig. 3; Huang et al. 2013, 2014). Indeed, Bcl-Xl or Mcl-1 overexpression enhanced mitochondrial Ca²⁺ uptake in line with observations in knockout cells for either of the proteins. Accordingly, VDAC1-deficient cells failed to show any Bcl-Xl or Mcl-1-mediated regulation of mitochondrial Ca²⁺ uptake (Huang et al. 2013, 2014). A more recent study (Morciano et al. 2016) revealed that the exogenous overexpression of the short proapoptotic Mcl-1 isoform was also accompanied by increased mitochondrial Ca²⁺ uptake, thereby increasing susceptibility to apoptotic stimuli. Although apparently conflicting and certainly puzzling, these evidences on VDAC1 modulation presumably underscore the activities of the channel as both driver of basal mitochondrial bioenergetics and gateway to cell death (Michels et al. 2013; Shoshan-Barmatz et al. 2017).

The molecular basis for Bcl-2, Bcl-X, and Mcl-1's modulatory effect on VDAC1 remains poorly characterized, especially with regard to its Ca²⁺-transporting activity. Bcl-2, Bcl-Xl, and Mcl-1 target the amino terminus of VDAC1, as shown by using VDAC1-amino-terminal derived (N-Ter) peptides, which also interfered with the antiapoptotic function of all three proteins when intracellularly delivered in multiple cell models (Abu-Hamad et al. 2009; Arbel and Shoshan-Barmatz 2010; Arbel et al. 2012; Huang et al. 2013, 2014; Morciano et al. 2016). The BH4 domain of Bcl-Xl, but not the one of Bcl-2, appears as the sole molecular determinant sufficient for binding and suppressing ATP-induced and VDAC1-mediated Ca²⁺ uptake. Consistently, N-Ter peptides would only counteract the channel inhibitory action of BH4-Bcl-Xl but not of BH4-Bcl-2 (Monaco et al. 2015). Among the VDAC1–cytosol accessible regions, the LP4 stands out as a conserved and validated binding site for Bcl-2, Bcl-Xl, and Mcl-1 (Abu-Hamad et al. 2009; Arbel and Shoshan-Barmatz 2010; Arbel et al. 2012; Huang et al. 2013, 2014). Importantly, stable and cell-permeable versions of VDAC1-amino-terminal and -LP4-derived peptides are under scrutiny as future cancer therapeutics because of their ability to induce cell death by, at least in part, modulating the Ca²⁺-transport activity of VDAC1 (Pittala et al. 2018; Shteinfer-Kuzmine et al. 2018). Finally, nuclear magnetic resonance (NMR) mapping experiments and detailed biochemical analyses revealed that, in a membrane-mimicking environment, the primary interacting interface is shaped by the carboxy-terminal portion of Bcl-Xl (encompassing BH1, BH2, and TMD) and by the β-strands 17 and 18 of VDAC1 (Losonczi et al. 2000; Malia and Wagner 2007; Hiller et al. 2008). The latter β-strands are relatively proximal to β-strand 14 (Fig. 4C), which contains the above-mentioned LP4 region. Therefore, in this “membrane-embedded” interaction model, the BH4 region would only partially contribute to the binding by reaching over the top of the VDAC pore (Malia and Wagner 2007; Hiller et al. 2008), in agreement with the data suggesting a putative pore contact between the mobile VDAC1-amino-terminal and BH4-Bcl-Xl (Monaco et al. 2015). Analogous analyses are still missing to conclusively determine whether or not the VDAC1–Bcl-2 and VDAC1–Mcl-1 interactions retain similar binding interfaces.

Proapoptotic Bcl-2 Proteins

Finally, the proapoptotic Bcl-2 relatives (i.e., Bak, Bax, tBid, and BNIP3) have been shown to interact with VDAC1 (Shoshan-Barmatz et al. 2018). However, there are (1) no direct and validated evidences for their regulation of VDAC1-mediated Ca²⁺ transport, and (2) insufficient information about the molecular determinants responsible for their complex formation with VDAC1.

CONCLUSIONS

Various Bcl-2 family members, including anti- and proapoptotic proteins, were identified as direct binding partners of intracellular Ca²⁺ channels. Despite the similar structure and function of antiapoptotic Bcl-2-family members as Bax/Bak inhibitors, they display divergent binding profiles for IP₃Rs, RyRs, and VDACs, translating to distinct impacts on Ca²⁺ homeostasis and dynamics and subsequently cell death and survival. Gaining more insights in the molecular determinants underlying these interactions will be instrumental for the successful development of selective strategies and tools for modulating the function of these channels, including in pathological conditions associated with altered IP₃R, RyR, or VDAC activity.