Pharmacological modulation of mitochondrial ion channels

Abstract

The field of mitochondrial ion channels has undergone a rapid development during the last three decades, due to the molecular identification of some of the channels residing in the outer and inner membranes. Relevant information about the function of these channels in physiological and pathological settings was gained thanks to genetic models for a few, mitochondria‐specific channels. However, many ion channels have multiple localizations within the cell, hampering a clear‐cut determination of their function by pharmacological means. The present review summarizes our current knowledge about the ins and outs of mitochondrial ion channels, with special focus on the channels that have received much attention in recent years, namely, the voltage‐dependent anion channels, the permeability transition pore (also called mitochondrial megachannel), the mitochondrial calcium uniporter and some of the inner membrane‐located potassium channels. In addition, possible strategies to overcome the difficulties of specifically targeting mitochondrial channels versus their counterparts active in other membranes are discussed, as well as the possibilities of modulating channel function by small peptides that compete for binding with protein interacting partners. Altogether, these promising tools along with large‐scale chemical screenings set up to identify new, specific channel modulators will hopefully allow us to pinpoint the actual function of most mitochondrial ion channels in the near future and to pharmacologically affect important pathologies in which they are involved, such as neurodegeneration, ischaemic damage and cancer.

Linked Articles

This article is part of a themed section on Mitochondrial Pharmacology: Featured Mechanisms and Approaches for Therapy Translation. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v176.22/issuetoc

Abbreviations

CGS7184

Ethyl1‐[[(4‐chlorophenyl)amino]oxo]‐2‐hydroxy‐6‐trifluoromethyl‐1H‐indole‐3‐carboxy‐late

CIC

intracellular chloride channel

CPP

cell penetrating peptide

CSA

cyclosporin A

DPC

diphenyl carbonate

EMRE

essential MCU regulator

GSK‐3

glycogen synthase kinase 3

G3139

phosphorothioate oligodeoxyribonucleotide

Hsp

heat shock protein

IMAC

inner membrane anion channel

IMM

inner mitochondrial membrane

HK

hexokinase

MCU

mitochondrial calcium uniporter

MMC

mitochondrial mega‐channel

NS1619

(1‐[2‐hydroxy‐5‐(trifluoromethyl)phenyl]‐5‐(trifluoromethyl)‐1,3‐dihydro‐2H‐benzimidazol‐2‐one

OMM

outer mitochondrial membrane

OSCP

oligomycin sensitivity conferral protein

PAP‐1

5‐(4‐phenoxybutoxy)psoralen

PDAC

pancreatic ductal adenocarcinoma

PM

plasma membrane

PTP

permeability transition pore

ROMK

renal outer medullary potassium channel

RR

ruthenium red

RyR1

ryanodine receptor

SITS

stilbene isothiocyanate sulfonic acid

TPP

triphenylphosphonium

TRPC

transient receptor potential cation channel

UCP

uncoupling protein

VDAC

voltage‐dependent anion channel

Introduction

A relatively recent and certainly noteworthy development in the health sciences is the emergence of ‘mitochondrial medicine', based on the recognition of the key role mitochondria play in many physiological processes whose alteration contributes to disease, such as metabolism, Ca²⁺ handling, redox signalling and cell death (see recent, excellent reviews on this topic, e.g. Heller et al., 2012; Smith et al., 2012; Edeas and Weissig, 2013; Szeto et al., 2014; Milane et al., 2015; Guzman‐Villanueva and Weissig, 2017; Masgras et al., 2017).

An important opportunity for intervention to modulate these processes is offered by mitochondrial ion channels (reviews on mitochondrial channels; Brini et al., 2017; Citi et al., 2017; Krabbendam et al., 2018; Mammucari et al., 2018; O'Rourke, 2007; Ponnalagu and Singh, 2017; Szabo and Zoratti, 2014; Szewczyk et al., 2009). These channels are especially important, at least in the contexts of cancer (Figure 1) (for reviews, see Leanza et al., 2013a, 2014b; Madamba et al., 2015; Peruzzo et al., 2016; Mazure, 2017; Szabo and Leanza, 2017; Shoshan‐Barmatz et al., 2017b), neurodegeneration (Rodriguez et al., 2013; Rao et al., 2014; Kalani et al., 2018) and ischaemia (Balderas et al., 2015; Testai et al., 2015b; Smith et al., 2017). In the present review, we discuss the recent developments in the field of mitochondrial ion channels, especially from the point of view of their pharmacological targeting. Specifically modulating members of this varied population may well be a rewarding approach, in particular in the case of cancer due to the major roles mitochondria play in cancer cell metabolism and in cell death. Pharmacological manipulation of mitochondrial channels may lead to cell death independently of upstream elements of the apoptotic cascade, such as http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=910 expression or p53 status, and alterations in signalling pathways (Fulda et al., 2010; Leanza et al., 2013a, 2014a, 2017, 2014b). Expression levels of ion channels in general and of mitochondrial channels in particular may vary from one cell type to another, and between healthy and diseased or activated cells. Examples of drastic variations of this type are provided by the http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=520 (Mu et al., 2003) and intracellular anion‐selective http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=128 channels, which also change localization and function in malignant cells (e.g. Peretti et al., 2015; Tasiopoulou et al., 2015). Silencing K_2P9.1 channels affects mitochondrial parameters and induces apoptosis in melanoma cells (Nagy et al., 2014). In the case of the potassium‐selective http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=540 channel, a higher expression correlated with greater sensitivity to chemotherapeutic treatment (Leanza et al., 2014b).

Figure 1.

A cartoon illustrating the main mitochondrial ion channels discussed in this review and their impact on cancer.

Multiple localizations of some of the ion channels found in mitochondria

When taking into account the possible pharmacological modulation of mitochondrial ion channels, one important distinction must be immediately made between mitochondria‐specific and multiple‐location channels. In fact, a few channels, such as the mitochondrial calcium uniporter (MCU), the inner membrane anion channel (IMAC), the magnesium‐transporting Mrs2 and the uncoupling protein (UCP), are specific to mitochondria. However, most of the other known ones reside in the plasma membrane as well, and possibly also elsewhere, as detailed in Table 1. These channels include the voltage‐dependent anion channel (VDAC), the big‐, intermediate‐ and small‐conductance calcium‐dependent potassium channels (http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=380 (BK_Ca), http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=384(http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=384) and http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=381 (SKCa)] as well as the voltage‐dependent K⁺ channels Kv1.3, http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=542 and http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=563, the two‐pore channel K_2P9.1, the ATP‐dependent K⁺ channel (KATP; http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=437) and the intracellular chloride channels http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=128 and http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=128 (for summarizing review, e.g. Szabo and Zoratti, 2014). The mechanisms underlying multiple targeting are being investigated (Singh et al., 2013; Kilisch et al., 2015; von Charpuis et al., 2015; Sole et al., 2016) but are still unclear in most cases. The mechanisms accounting for different localizations of proteins in general vary and may include differences among isoforms (e.g. due to alternative splicing), ‘piggybacking' other proteins, the formation of different heteromers [e.g. in the case of K_Ca1.1 (BK_Ca ) channels; Chen et al., 2009], post‐translational modifications such as, for example, phosphorylation or myristoylation, interactions with structures such as, for example, ‘scaffolds', the cytoskeleton or, in membranes, ‘lipid rafts'.

Table 1.

Mitochondrial channels in non‐mitochondrial locations

Channel	Locationa	Proposed functionsb	Referencesb
VDAC	Plasma membrane	Cell volume regulation Apoptosis ATP release	Thinnes (1992, 2014, 2015) Jakob et al. (1995) Bathori et al. (2000, 1999)
VDAC	Endosomes	–	Reymann et al. (1998)
VDAC	Endo/sarcoplasmic reticulum	ATP transport SR/ER‐Mito Ca2+ transport/signalling	Jurgens et al. (1995) Shoshan‐Barmatz et al. (1996); Shoshan‐Barmatz and Israelson (2005)
BKCa (SLO1, KCa1.1)	Plasma membrane	Integration of cell signalling	Singh et al. (2012); Yu et al. (2016); Latorre et al. (2017)
BKCa (SLO1, KCa1.1)	ER membrane	–	Ma et al. (2007); Cox et al. (2014)
BKCa (SLO1, KCa1.1)	Nuclear envelope	Control of nuclear Ca2+ signalling	Singh et al. (2012); Li et al. (2014)
BKCa (SLO1, KCa1.1)	Lysosomal membrane	Control of lysosomal Ca2+	Wang et al. (2017a, 2017b); Sterea et al. (2018)
IKCa (KCa3.1, KCNN4)	Plasma membrane	Cell proliferation, differentiation, migration Control of NO synthesis (in vascular epithelia)	Sforna et al. (2018); Zhan et al. (2018) Grgic et al. (2005); Sheng and Braun (2007); Ruggieri et al. (2012); Ohya and Kito (2018)
SKCa (KCa2.x)	Plasma membrane	Control of electrical excitability	Yang et al. (2015); Kshatri et al. (2018)
SKCa (KCa2.x)	ER membrane	Control of ER Ca2+ homeostasis	Kuum et al. (2012, 2015); Richter et al. (2016); Honrath et al. (2017a)
Kv7.4	Plasma membrane	Control of electrical excitability Smooth muscle function Regulation of vascular tone	Chang et al. (2018) Miceli et al. (2008) Martelli et al. (2013)
SLO2.1 (KNa1.2)	Plasma membrane	Control of electrical excitability and of synaptic transmission Anaesthetic preconditioning	Kameyama et al. (1984) Wojtovich et al. (2016)
Kv1.3	Plasma membrane	Control of voltage and Ca2+ fluxes	Cahalan and Chandy (2009); Zhao et al. (2015); Perez‐Verdaguer et al. (2016); Chandy and Norton (2017); Serrano‐Albarras et al. (2018)
Kv1.3	ER and Golgi	–	Sole et al. (2009); Zhu et al. (2014)
Kv1.3	Nuclear envelope	Regulation of nuclear transmembrane potential	Jang et al. (2015)
Kv1.5	Plasma membrane	O2 sensing	Comes et al. (2013)
TASK‐3 (K2P9.1)	Plasma membrane	Cell excitability (background current)	Zuzarte et al. (2007); Kilisch et al. (2015)
ROMK2 (Kir1.1b)	Plasma membrane (renal epithelia)	Ion homeostasis	Zhou et al. (1994); Yoo et al. (2005); Angsanakul and Sitprija (2013); Frindt et al. (2013); Welling (2016)
KATPc	Plasma membrane	Control of transmembrane potential	Noma (1983)
KATPc	Nuclear envelope	Control of nuclear Ca2+ ‘waves'	Quesada et al. (2002)
KATPc	ER	Control of Ca2+ load and dynamics	Ashrafpour et al. (2008); Salari et al. (2015)
TRPC3	Plasma membrane	Mechanosensitive Ca2+ signalling Cellular Ca2+ homeostasis	Mizoguchi and Monji (2017); Yamaguchi et al. (2017); Tiapko and Groschner (2018)
RyR	Sarcoplasmic reticulum	Ca2+ dynamics Excitation/contraction coupling	Allard (2018) Treves et al. (2017)
RyR	Nuclear envelope	–	Gerasimenko and Gerasimenko (2004); Marius et al. (2006); Zheng et al. (2012)
IP3R	Endoplasmic reticulum	Ca2+ signalling Signal integration ER‐mitochondria communication	Foskett et al. (2007); Bustos et al. (2017); Kania et al. (2017)
IP3R	Nuclear envelope	Ca2+ signalling	Mak and Foskett (1994); Stehno‐Bittel et al. (1995); Gerasimenko and Gerasimenko (2004); Mak et al. (2013)
IP3R	Plasma membrane (low levels)	Ca2+ entry	Mayrleitner et al. (1995); Dellis et al. (2006)
CIC‐4	Plasma membrane	–	Peretti et al. (2015)
CIC‐4	Nuclear envelope	–	Domingo‐Fernandez et al. (2017)
CIC‐1	Plasma membrane	Volume regulation Cell migration Metastasis	Valenzuela et al. (2000) Novarino et al. (2004) Wang et al. (2012); Setti et al. (2013); Peretti et al. (2015)
CIC‐1	Nuclear envelope	–	Valenzuela et al. (1997)

^aRefers to subcellular compartments. Organ‐, tissue‐ or cell type‐specific or signalling‐dependent expression is not addressed. Due to cellular membrane traffic and protein flux, integral PM proteins are expected to be found also in the ER, Golgi (anterograde transport) and endosomes, lysosomes (degradative retrograde transport). Thus, for example, IK_Ca, which has a rapid turnover, is abundant in lysosomes (Balut et al., 2010). The presence in these compartments is therefore not highlighted here, except if evidence exists of a functional presence in the organellar membrane and in the case of VDAC, a largely mitochondrial channel for which it constitutes confirmation of its presence also in the PM.

^bExamples are given.

^cK_ir/SUR assemblies.

The channel populations residing in different compartments in general have different tasks, roles and/or substrates, and it may therefore be useful to pharmacologically target them specifically in mitochondria or elsewhere. Unfortunately, the art of subcellular targeting is still in its infancy – mitochondrial targeting is by far the most developed – so that little information is available comparing the effects of inhibiting a given channel in one subcellular compartment or another. A partial exception is provided by the plasma membrane, since some of the ion channels in it are inhibited by membrane‐impermeant toxins. The clearest example may be that of potassium‐selective channel K_v1.3: selective block of the plasma membrane population by peptide toxins impairs cell proliferation and migration (e.g. Chhabra et al., 2014; Aissaoui et al., 2018), while selective inhibition of the mitochondrial population by permeant, mitochondriotropic psoralenic compounds precipitates cell death (Leanza et al., 2017). It ought also to be mentioned that mitochondria themselves may differ in structure, properties, functions and proteic composition depending on their location within the cell (e.g. see Diaz et al., 1999, 2000; Riva et al., 2005; Boengler et al., 2009; Baseler et al., 2011; Asemu et al., 2013; Kasumov et al., 2013; Hollander et al., 2014; Fedorovich et al., 2017), not to mention cell type and age. It is also important to mention that not all mitochondrial ion channels have been described/found in all tissues, but they are certainly observable in healthy tissues, not only in pathological cells (cell lines). In the present review, we illustrate our current knowledge on the drugs identified so far that act on mitochondrial ion channels and comment on some of the strategies that can be useful to target specifically the mitochondria‐located channels.

Mitochondrial outer membrane (OMM) channels

VDAC

The VDAC, or ‘mitochondrial porin', is the most extensively studied mitochondrial channel (rev.s: Szabo and Zoratti, 2014; Magri et al., 2018; Shoshan‐Barmatz et al., 2018). The most abundant and most investigated of the three mammalian isoforms (VDAC1, VDAC2 and VDAC3) (for recent review, see, e.g. Messina et al., 2012) is VDAC1, which is present both in the OMM and in the PM (Bathori et al., 2000; De Pinto et al., 2010), in endosomes (Reymann et al., 1998) and in the sarco/endoplasmic reticulum (Jurgens et al., 1995; Shoshan‐Barmatz et al., 1996) (Table 1). The effects of the various modulators have nearly always been automatically attributed to their action on mitochondrial porin(s). The possibility of a contribution by populations located elsewhere has rarely been adequately explored.

VDAC1 is a β‐barrel composed of 19 β‐strands and of an N‐terminal mobile α‐helical segment (for recent review, see, e.g. Zeth and Zachariae, 2018). The latter contains the major binding sites for ions and other proteins (Caterino et al., 2017; Magri and Messina, 2017). As the functions of the N‐terminal peptide and the network of interactions were progressively characterized, VDAC1 has progressed from a more or less inert ‘molecular sieve' in the OMM to a dynamic participant in a host of cellular processes. One is that of allowing and regulating the passage of Ca²⁺ ions through the OMM (Gincel et al., 2001; Bathori et al., 2006) and at ER/mitochondria contact sites (Hajnoczky et al., 2002; Rapizzi et al., 2002). The binding of Bcl‐xL to the N‐terminal segment may modulate this flux. Mitochondrial Ca²⁺ overload may induce cell death, in which VDAC1 is implicated in various other ways as well. In particular, VDAC can oligomerize in a Ca²⁺‐driven process, with the involvement of the N‐terminus, and oligomers are proposed to be implicated in the release of pro‐apoptotic factors such as cytochrome c and apoptosis‐inducing factor (AIF) (for recent review, see Shoshan‐Barmatz et al., 2017a). Furthermore, VDAC1 regulates cell death due to its interactions with anti‐apoptotic members of the Bcl‐2 family (i.e. Bcl2 and Bcl‐xL) and with http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=890 (HK) (Shimizu et al., 2000; Arbel and Shoshan‐Barmatz, 2010).

Various chemical compounds interact with and alter the activity of VDAC1 and may eventually be exploited for therapy in the context of VDAC‐related diseases (Figure 2). The first VDAC blocker to be identified was Konig's polyanion, a 1:2:3 copolymer of methacrylate, maleate and styrene that acts at rather low concentrations (27 μg·mL⁻¹ polyanion) (Colombini et al., 1987). The general anion channel inhibitors http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4177, http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4314, H(2) DIDS, 4,4′‐dinitrostilbene‐2,2′‐disulfonic acid (DNDS) and diphenyl carbonate (DPC) were also shown to interact with VDAC1 and to decrease its conductance in electrophysiological experiments with bilayer‐reconstituted VDAC1 (Thinnes et al., 1994; Ben‐Hail and Shoshan‐Barmatz, 2016). Furthermore, ubiquinone specifically binds to VDAC1 through its quinone‐head ring (Murai et al., 2017).

Figure 2.

Chemical structure of some VDAC modulators. For further information, see text.

The phosphorothioate oligodeoxyribonucleotide G3139 (http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=8269) was shown to induce rapid flickering of the VDAC1 channel activity and even its closure at 1 μM concentration without affecting the respiration complexes, the adenine nucleotide translocator or the ATP synthase (Tan et al., 2007; Stein and Colombini, 2008). G3139 interferes with metabolite flow through VDAC1 (for review, see, e.g. Klasa et al., 2002). G3139 targets the first six codons of Bcl‐2 mRNA and is successfully used in anticancer treatments in combination with other therapeutic agents (including chemotherapy and radiotherapy), in tumour types ranging from chronic lymphocytic leukaemia beta cell lymphoma to breast cancer and many solid tumours (Kamal et al., 2014). The relevance of the block of VDAC1 activity for the anticancer effect of G3139 is not clear. In general, closure of VDAC would lead to an overall lowering of the cell energy metabolism, subsequently triggering apoptosis.

Other compounds that decrease the conductance of VDAC are avicins, a class of plant stress‐induced metabolites of triterpenoids displaying selective pro‐apoptotic and cytotoxic activity in cancer cells, as well as antioxidant and anti‐inflammatory properties (see, e.g. Haridas et al., 2007). Avicins however stimulate tumour cell death via multiple mechanisms, as they target various cellular components in addition to VDAC, such as tubulin and topoisomerase. They increase the OMM permeability to cytochrome c by inhibiting respiration and by reducing the efficiency of antioxidant systems, leading to a hypersensitivity of mitochondria to oxidative stress. In particular, mitochondria are the main targets of avicin D and avicin G, which can induce apoptotic cell death via intrinsic and extrinsic pathways but also autophagic programmed cell death due to depletion of cell energy. Avicins could thus be of value as chemical agents for the treatment and/or prevention of malignancy (Wang et al., 2010).

The activity of VDAC can also be manipulated by other plant‐derived molecules. http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4150, from the Cannabis species, has strong in vitro anti‐tumour effects in numerous cancer cells types (Massi et al., 2013), and was shown to act on VDAC (Rimmerman et al., 2013). http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=7000, a component of Curcuma longa with anti‐inflammatory and anti‐tumour activity, has also been shown to affect VDAC function by interacting with its N‐terminal domain, at least in vitro (Tewari et al., 2015). Chrysophanol, emodin, rhein, aloe‐emodin and catechin, identified as the bioactive components of the herb Rheum officinale Baill and able to induce apoptosis in many human cancer cell lines, were shown to target VDAC‐1 through Thr²⁰⁷ and the N‐terminal region of the protein (Li et al., 2017). However, a direct effect on channel conductance and behaviour has still not been proven.

The list of chemicals that directly modify VDAC1 activity by decreasing channel conductance also includes http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=203 (Prozac) (Nahon et al., 2005; Thinnes, 2005), a clinically used antidepressant drug, that also inhibits the opening of the mitochondrial permeability transition pore, the release of cytochrome c, and protects against http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=346‐induced apoptotic cell death. Lastly, cyathin‐R, a cyathane diterpenoid, was shown to decrease both channel conductance and cancer growth, by promoting apoptosis even in Bax/Bak‐deficient cells (Huang et al., 2015). Cyathin‐R also induced VDAC1 oligomerization, and apoptosis was found to be inhibited by VDAC1‐interacting molecules, such as DIDS and DIDS analogues (H₂DIDS, SITS and DPC). These drugs decrease the channel conductance (Thinnes et al., 1994) and counteract apoptosis at the early stage, putatively by diminishing VDAC1 oligomerization and thus preventing cytochrome c release to the cytosol (Keinan et al., 2010).

The above examples point to an important role for VDAC in the regulation of apoptosis; however in some cases, inhibition/decrease of its activity by pharmacological means leads to apoptosis (e.g. G3139), while in other cases, the same modulation of VDAC activity results in protection against cell death (e.g. Prozac). In fact, the role of VDAC activity/oligomerization in OMM permeabilization (MOMP)‐induced apoptosis seems to be rather complex. Bax and t‐Bid have been proposed to take part in the formation of OMM pores in apoptotic cells by assembling heteromeric complexes with VDAC (Tajeddine et al., 2008). Other studies however indicate a cooperative role of Bak and Bax and VDAC2, rather than of VDAC1, in MOMP (see, e.g. Cheng et al., 2003 and for review, Lauterwasser et al., 2016), and also the involvement of other outer membrane ‘receptors' for pro‐apoptotic proteins, including cardiolipin (for recent review, see, e.g. Veresov and Davidovskii, 2014). This may explain why cells lacking all VDAC isoforms were found to be still capable of undergoing apoptosis (Baines et al., 2007). In addition, the interaction of VDAC with other partners such as with the glycolytic enzyme hexokinase is also important, as it gives a proliferative advantage to cancer cells (see below).

Unfortunately, one drawback in this field is the absence of VDAC isoform‐specific pharmacological agents. VDAC2 (Gattin et al., 2015) and VDAC3 (Checchetto et al., 2014) also form ion channels, although the latter one with small conductance. It is worthwhile to mention that a recent work reported the synthesis of an arsenate‐based fluorescent probe, which can preferentially target the mitochondrial VDAC2, this channel being a member of the membrane‐bound vicinal dithiol protein class (Yang et al., 2018). Even if isoform‐specific small molecule inhibitors are not available, as an alternative strategy for specific targeting of VDAC isoforms, peptides copying specific VDAC sequences that can modulate the interaction with various pro‐ and anti‐apoptotic proteins (see below) can be designed and can be therefore useful to modulate a specific VDAC‐related function. Of course this approach would require an extension of our knowledge on the isoform‐specific interaction partners (if there are specific ones) for all three isoforms.

Other mitochondrial outer membrane channels

Beside VDAC isoforms, the OMM also harbours a few other channels such as an unidentified inward rectifying potassium channel (http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=74) (Fieni et al., 2010) and CIC‐4, a member of the intracellular chloride channel family (Ponnalagu et al., 2016a; Ponnalagu and Singh, 2017). CICs are http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4217 (IAA‐94)‐sensitive, but several mitochondrial CIC family members are present also in the endoplasmic reticulum (Ponnalagu et al., 2016b) making it difficult to delineate the function of mitochondrial CICs by pharmacological means. The existence of CICs and K_ir channels, both small‐conductance channels, in the OMM challenges the traditional view that the outer membrane forms an unspecific size‐exclusion filter for the flux of small hydrophilic molecules (Becker and Wagner, 2018). In yeast mitochondria, the recent identification of three novel ion channels in the OMM with unknown substrate specificity (Kruger et al., 2017) points to the same conclusion (Checchetto and Szabo, 2018). Whether homologues of the yeast OMM channels also operate in mammalian OMM is still an open question as is the pharmacology of these channels.

Inner membrane ion channels

The permeability transition pore

The permeability transition is a well‐known process that occurs in mitochondria: an increased permeability to solutes of the inner mitochondrial membrane (IMM) caused by the opening of the permeability transition pore (PTP) (Zoratti and Szabo, 1995; Bernardi et al., 2015). PTP is a 1.3 nS conductance unspecific channel located in the IMM whose opening induces depolarization, loss of ionic homeostasis and block of respiratory chain activity and ATP synthesis. The expression ‘permeability transition' to describe these phenomena (considered as an aspecific, lipid‐perturbation effect) was used for the first time in the late seventies by Haworth and Hunter (1979). The discovery of the ability of http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=1024 (CsA) to inhibit PTP in classical bioenergetics experiments (Crompton et al., 1988) and the observation of a CsA‐sensitive 1.3 nS ion channel in patch clamp experiments on mitoplasts, that is, mitochondria devoid of OMM (Szabo and Zoratti, 1991), completely changed the perspectives concerning the permeability transition (PT). The observed channel, the so‐called mitochondrial mega‐channel (MMC), reproduced all known pharmacological features of the PTP: inhibition by CsA, adenine nucleotides, Mg²⁺, acidic pH and reducing agents, induction by Ca²⁺ (Szabo and Zoratti, 2014). PT, that thanks to these studies could be ascribed to a pore (PTP), changed its ‘status' from being a supposed in vitro artefact or lipid perturbation to an important physio‐pathological process involved in cell death, apoptosis resistance in cancer cells, oxidative stress, anoxia and ischaemia followed by reperfusion, just to name a few (for reviews, see, e.g. Di Lisa and Bernardi, 2006; Brenner and Moulin, 2012; Rasola and Bernardi, 2014; Leanza et al., 2014b).

Despite the fact that the role of PTP in mitochondrial physiology was intensively studied in the last decades in different pathophysiological contexts, the structure and molecular identity of PTP is still highly debated (Biasutto et al., 2016; Giorgio et al., 2018), making difficult the ‘smart design' of a structure‐based specific channel modulator. As to date, the scientific community is converging towards the idea that PTP is formed by dimers of the F_OF₁ ATP synthase (Alavian et al., 2014; Beutner et al., 2017; Bonora et al., 2013; Carraro et al., 2014; Giorgio et al., 2013; He et al., 2017), even though the exact mechanism and the site and mode of the actual pore formation is still unclear. In this respect, as well as regarding the possibility of designing specific drugs, the strategies undertaken, involving site‐directed mutagenesis combined with functional assays (Giorgio et al., 2017; Antoniel et al., 2018), might turn out to be extremely useful. Indeed, strong support in favour of the identification of the ATP synthase with PTP/MMC is the finding that modulation of MMC activity directly recorded by patch clamp in the native mitochondrial inner membrane by protons was altered in a single point mutant of the oligomycin sensitivity conferral protein (OSCP) subunit of the ATP synthase complex (Antoniel et al., 2018). In addition, subunits e, g and the first transmembrane segment of subunit b have recently been shown to contribute to channel formation in yeast, by combining multiple gene deletion with electrophysiology (Carraro et al., 2018).

Given the central role of the PTP in pathophysiological contexts, considerable effort was made in the field to discover new, direct modulators to treat disorders, even in the absence of a clear molecular identity. Pharmacological targeting of PTP is not straightforward, since no bona fide pore blockers are available. In fact, even the action of the relatively specific CsA, which is effective at sub‐μM concentrations, is indirect, as it is mediated by cyclophilin D, whose expression level varies among different cells. Furthermore, not only is cyclophilin D expression important but also its ratio with F‐ATPase OSCP subunits: considering that normally there are more F‐ATP synthase complexes than cyclophylin D molecules in the cells, most of the PTP complexes present are expected to be insensitive to CsA treatment (Bernardi et al., 2015). CsA inhibits calcineurin and thus acts as an immunosuppressant (Griffiths and Halestrap, 1995). Therefore, several new, calcineurin‐indifferent cyclophilin D inhibitors have been developed over the past 20 years. The most promising are sanglifehrin A (Clarke et al., 2002), 9‐(N‐methyl‐L‐isoleucine)‐cyclosporin A (Zulian et al., 2014) and MeAla(3)EtVal(4)‐cyclosporin (Tiepolo et al., 2009). Recently, other factors have also been shown to modulate PTP by targeting cyclophilin D, such as http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=621 (Burstein et al., 2018), http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=7527 (Roy et al., 2015) and n‐phenylbenzamides (Roy et al., 2016). A virtual chemical screening identified several compounds that were able to bind to cyclophilin D with a K _D of <100 nM (Park et al., 2017). Synthetic compounds from the diarylquinoline and organotin families instead target the c‐ring of the F‐ATPase and strongly inhibit the enzyme (Pagliarani et al., 2016). A member of these families, tributyltin, induces mitochondrial swelling, which has long been associated with the PT, but is CsA‐insensitive, at least according to some authors (see, e.g. Gogvadze et al., 2002).

Several different classes of compounds acting at sites that do not involve the PTP pore itself have been characterized for their ability to antagonize the mitochondrial permeability transition in an indirect way (Figure 3). Among these substances are gasotransmitters (NO‐donors, H₂S‐donors and Co‐donors) (for review, see Andreadou et al., 2015) and antioxidants (Javadov and Karmazyn, 2007), just to name a few classes. Altogether, all the above‐mentioned inhibitors, even though by an indirect effect, were proven to affect PTP activity in various systems. Some of them might represent good starting points for the design of more specific pharmacological agents able to counteract those degenerative diseases where PTP opening plays a crucial role.

Figure 3.

Chemical structure of some modulators of the permeability transition pore. For further information, see text.

Conversely, several PTP inducers with different chemical structures and displaying distinct mechanisms of PTP activation are promising agents for the treatment of cancer. Indeed, some of these compounds are already in clinical trials; others have been tested in vitro in tumour cell lines or in vivo in preclinical models for different neoplasms. Some of these molecules are mastoparan‐like peptides (Jones et al., 2008), 4‐(N‐(Sglutathionylacetyl)amino)phenylarsonous acid (Elliott et al., 2012), ursolic acid (Lu et al., 2014), http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=3945 (Fulda et al., 1999; Potze et al., 2015), B4G2 (an hydroxybetulinic acid derivative) (Yao et al., 2015), triterpenoid derivatives (Laszczyk, 2009; Serafim et al., 2014), curcumin (Qiu et al., 2014), lonidamine (Lena et al., 2009), http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=8320 (a BH3 mimetic) (Yu et al., 2015), the K⁺/H⁺ antiporter salinomycin (Manago et al., 2015; Qin et al., 2015), berberine (Shen et al., 2014), ebselen (Pavon et al., 2015), the gold complex AUL‐12 (Chiara et al., 2012), icaritin (Zhou et al., 2015), EM20‐25 (Milanesi et al., 2006) and ferutinin (Ilyich et al., 2018). It has to be stressed that in most cases, possible side effects can be envisioned as opening of the PTP might lead to loss of mitochondrial function in healthy cells as well. Specific targeting of these molecules exclusively to tumour tissues employing various strategies (for a recent review, see Ruoslahti, 2017) is a challenge but might be a worthwhile goal to pursue.

As in the case of VDAC, one might envision not only the direct modulation of channel activity by small molecules but also the possibility to fine‐tune the activity of the PTP (proposed to correspond to the ATP synthase) or of its partners/modulators (for review, see Miura and Tanno, 2012), by post‐translational modifications by kinases, such as hexokinase II (HK II) (Rasola et al., 2010a) or the mitochondrial http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=2030 (mGSK‐3) (Rasola et al., 2010a). PTP can be indirectly and negatively modulated in cancer cells by mGSK‐3 phosphorylation of cyclophilin D, and mGSK‐3 can in turn be inhibited by a mitochondria‐located http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=1494 fraction, which is normally under the control of the often dysregulated http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=897. An active ERK in mitochondria of cancer cells was shown to lead to mGSK‐3 phosphorylation, thus preventing cyclophilin D phosphorylation and in turn ‘desensitizing' PTP (Rasola et al., 2010a). Similarly, the PTP is inhibited when cyclophilin D is nitrosylated on specific cysteine residues (Nguyen et al., 2011) or interacts with the transcription factor STAT‐3 (Boengler et al., 2010). Furthermore, chaperones such as heat shock protein (Hsp) 60, Hsp90 and tumour necrosis factor receptor‐associated protein‐1 have been demonstrated to modulate PTP activity, and these chaperones themselves can be regulated by post‐translational modification (Masgras et al., 2017; Rasola, 2017). Opening of the PTP can be induced also by HK II detachment from the OMM: this event can be controlled by signalling pathways that involve either the Ser/Thr kinase Akt or GSK‐3, even if the mechanism accounting for this regulation is still debated (Miyamoto et al., 2008; Rasola et al., 2010b). Finally, some direct sites of post‐translational modifications, for example, phosphorylation, acetylation and S‐nitrosylation, have also been identified in the F‐ATPase subunits (Bernardi et al., 2015), although they may hardly offer a more specific way of intervention.

Mitochondrial calcium uniporter and other calcium channels

Mitochondria play a crucial role in intracellular Ca²⁺ regulation by shaping, remodelling, relaying and decoding Ca²⁺ signals, due to their ability to rapidly and transiently accumulate Ca²⁺ (Drago et al., 2011). Ca²⁺ uptake into mammalian mitochondria is efficiently blocked by low concentrations of ruthenium red (RR) and Ru360, thanks to the fact that they lead to a direct inhibition of the uniporter (Moore, 1971; Vasington et al., 1972; Reed and Bygrave, 1974). The molecular identity of the protein(s) giving rise to uniporter activity was not known until the last decade. The finding that a highly Ca²⁺ selective ion channel replicated all key characteristics of the mammalian mitochondrial uniporter represented a milestone towards its molecular identification (Kirichok et al., 2004). Another important step was the publication of the MitoCarta database (Pagliarini et al., 2008), which then provided the basis for the identification of several mitochondrial calcium uniporter complex (MCUC) components in mammals, including the central pore forming protein MCU (mitochondrial calcium uniporter) (Baughman, 2011; De Stefani et al., 2011). At the current stage, in mammals, the MCUC appears to include at least the pore‐forming protein MCU, an MCU paralogue (MCUb), the essential MCU regulator (EMRE), the regulatory MICU proteins and, possibly, the mitochondrial calcium uniport regulator 1 (MCUR1). In other systems, such as plants, MCUC seems simpler and consists primarily of MCU and MICU (Wagner et al., 2016). Mammalian MCU activity can be regulated through its paralogue MCUb (Raffaello et al., 2013), through interaction with MICU1, the first uniporter component identified (Perocchi et al., 2010), by two other MICU1 isoforms, namely, MICU2 and MICU3 (Plovanich et al., 2013) and by post‐translational modifications (Jhun et al., 2016). MICU2 forms a heterodimer with MICU1 through an intermolecular disulphide bond and closes the channel at low extramitochondrial Ca²⁺ concentrations (Patron et al., 2014; Petrungaro et al., 2015). Currently, two models coexist that find MICU1 to act exclusively as a uniporter activator at high cytosolic Ca²⁺ concentrations (Patron et al., 2014) or to gradually disinhibit the uniporter with increasing Ca²⁺ concentrations in the cytosol (Csordas et al., 2013). Studies resolving for the first time the 3D structure of proposed pentameric MCU by a combination of NMR and electron microscopy allowed the proposal of the hypothesis to explain the requirement of the presence of EMRE in the native membrane to observe MCU activity: the outer and inner juxtamembrane helices as well as a loop region in MCU seem to be unstable regions, which may undergo conformational changes upon activation by EMRE in order to create the lateral exit path for Ca²⁺ in the MCU channel (Oxenoid et al., 2016). Recently, a more in‐depth insight into the structure of the tetrameric MCU has been obtained, thanks to the resolution of the structure of MCU by X‐ray and cryo‐EM simultaneously by different groups and from different organisms (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018). EMRE is a metazoan‐specific protein able to mediate MCU‐MICU1/MICU2 dimer interaction and, as shown recently, able to regulate MCU channel activity depending on the matrix Ca²⁺ concentration (Sancak et al., 2013; Vais et al., 2016). Furthermore, the activity of MCU can be also indirectly regulated, for example, by the uncoupling protein UCP2, as mitochondria isolated from UCP2^−/− mice showed decreased RR‐sensitive calcium uptake (Motloch et al., 2016).

The importance of the correct regulation of MCU activity is underlined by the finding that patients carrying a loss‐of‐function mutation of MICU1 are characterized by myopathy, cognitive impairment and an extrapyramidal movement disorder (Logan et al., 2014). MICU1 has recently been shown to play a role in Ca²⁺ overload‐induced mitochondrial PTP opening and also to be a decisive factor for adaptation to postnatal life and for tissue repair after injury of liver (Antony et al., 2016). Other studies underline the importance of MCU expression/activity in the context of the regulation of metabolism in various organisms ranging from Trypanosoma to zebrafish and mice (e.g. Duchen, 2000; Huang et al., 2013; Prudent et al., 2013; Gorlach et al., 2015; Mammucari et al., 2018). The manipulation of MCU expression changed muscle function, through a novel Ca²⁺‐dependent mitochondria‐to‐nucleus signalling pathway (Mammucari et al., 2015). MCU function was also shown to affect oxidative phosphorylation and to ensure correct pacemaker cell function (Wu et al., 2015). MCU is overexpressed in several type of cancer cells (e.g. Davis et al., 2013; Peruzzo et al., 2016; Bustos et al., 2017) and has been implicated in cancer‐related processes, in particular in the control of migration and metastasis (Hall et al., 2014; Tang et al., 2015; Tosatto et al., 2016; Szabo and Leanza, 2017).

Given the vast literature demonstrating the fundamental impact of calcium uptake into mitochondria for many physiological and pathological processes, it became clear that either activation (to induce calcium overload and PTP opening) or inhibition (to decrease ATP production and/or to prevent calcium overload) of MCU might be useful in pathological settings. However, regarding pharmacological modulation of MCU, the direct inhibitors identified to date are only RR and Ru‐360 (Kirichok et al., 2004; De Stefani et al., 2011; Wu et al., 2018), which also affect other cationic channels. No specific activators are known to our knowledge. It has been proposed that a small heterocyclic molecule, 5‐[(4‐methylphenyl)thio]‐9H‐pyrimido[4,5‐b]indole‐2,4‐diamine (AG311), retards tumour growth by activating MCU, but a direct proof is still missing (Bastian et al., 2015). Likewise, the protective effect of Necrox‐5 ([5‐(1,1‐dioxo‐thiomorpholin‐4‐ylmethyl)‐2‐phenyl‐1H‐indol‐7‐yl]‐(tetrahydro‐pyran‐4‐ylmethyl)‐amine) on myocardial damage in ischaemic heart disease has been ascribed to its ability to inhibit MCU (Thu et al., 2012) but also in this case, a direct electrophysiological proof is missing. A recent work identified http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=7242 as an MCU modulator by systematic chemical screening (Arduino et al., 2017). Likewise, a high‐throughput screening of 120 000 small‐molecule compounds identified the membrane‐permeant drug DS1657, an indolebutyric acid derivative (Sawada et al., 2001), that dose‐dependently inhibited serum‐induced mitochondrial Ca²⁺ influx in HEK293A cells with an IC₅₀ of 7 μM (Kon et al., 2017). Formal proof and information about the mechanism of inhibition is lacking in the latter case. In this respect, electrophysiological lipid bilayer experiments using recombinant proteins might be useful. MCU produced in a cell‐free system as well as that purified from Escherichia coli using His‐tag, either added to the recording chamber as the purified protein or following purification and incorporation into liposomes, showed channel activity in a reproducible manner in more than 200 experiments performed in different settings (De Stefani et al., 2011; Raffaello et al., 2013; Patron et al., 2014; Teardo et al., 2017). However, a recent work using human MCU expressed in mammalian cells with FLAG‐tag concluded, on the basis of only three experiments, that this protein does not form channels by itself (Wu et al., 2018). Further work will be required to clarify the reasons for this discrepancy. It may be due to different conditions used for protein solubilization and purification and/or the presence of cholesterol in the latter system that is known to affect the activity of several channels (but is largely absent from the IMM) (Brini et al., 2017).

Other mitochondrial Ca²⁺ channels might contribute to the mediation of Ca²⁺ fluxes into/out of this organelle, particularly in specialized tissues where the MCU may play a secondary role. Potential candidates mediating calcium flux include the transient receptor potential cation http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=488 (Feng et al., 2013; Wang et al., 2017a, 2015, 2017b) and the mitochondrial ryanodine receptor (mhttp://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=747). A low level of RyR1 is detectable in heart mitochondria and provides rapid transport of Ca²⁺ that is insensitive to ruthenium red (Beutner et al., 2001, 2005). The TRPC3 channel, whose structure has been recently resolved (Tang et al., 2018), is specifically inhibited by ethyl‐1‐(4‐(2,3,3‐trichloroacrylamide)phenyl)‐5‐(trifluoromethyl)‐1H‐pyrazole‐4‐carboxylate (Pyr3) that was shown to alter mitochondrial membrane potential and ROS production when applied together with dexamethasone (Abdoul‐Azize et al., 2016). However, TRPC3 and RyR1 show multiple localizations within the cell, making it difficult to ascribe the observed physiological effects to the regulation of these channels only in mitochondria.

Mitochondrial inner membrane potassium channels

IMM K⁺ channels studied by electrophysiology are the calcium‐dependent channels – big‐conductance potassium channel (mtBKCa) (Singh et al., 2012), intermediate‐conductance K⁺ channel (mtIKCa) (De Marchi et al., 2009) and small‐conductance K⁺ channel (mtSKCa) (Dolga et al., 2013), voltage‐gated shaker type K⁺ channel K_v1.3 (Szabo et al., 2005) and the ATP‐dependent potassium channel (mtKATP) (Inoue et al., 1991). An interesting new development in the field is the discovery that mtBKCa behaves as a mechanosensitive channel in electrophysiological experiments (Walewska et al., 2018). Biochemical/genetic evidence indicates the presence of the pH‐sensitive two‐pore potassium channel K_2P9.1 (TASK‐3) (Pocsai et al., 2006; Toczylowska‐Maminska et al., 2014), of K_v1.5 (Leanza et al., 2012b), of http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=429 (Foster et al., 2012), of the http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=385 (Wojtovich et al., 2011; Smith et al., 2018) and of K_v7.4 (Testai et al., 2015a, 2015b) and another pH sensitive K⁺ channel (Kajma and Szewczyk, 2012) (for recent reviews, see, e.g. Szabo and Zoratti, 2014; Laskowski et al., 2016; Augustynek et al., 2017). Almost all of the above‐mentioned IMM K⁺ channels identified thus far are considered to be the mitochondrial counterparts of well‐known plasma membrane (PM) channels and many of them even display multiple subcellular localizations in the ER, nucleus, Golgi and PM membranes in addition to mitochondria (Table 1) (for review, see, e.g. Laskowski et al., 2016; Brini et al., 2017). Nevertheless, mtKATP seems to have a different composition from its PM counterpart: a splicing variant of the renal outer medullary potassium channel (ROMK) has been proposed to form the ion‐conducting subunit (Foster et al., 2012) and indeed, the ROMK protein is present in the IMM (Bednarczyk et al., 2018). Since ROMK apparently does not have a wide tissue‐expression while mtKATP activity has been observed in various tissues, a contribution of other pore‐forming proteins to mtKATP channel activity cannot, as yet, be fully excluded.

IMM K⁺ channels have been proposed to contribute to the regulation of matrix volume, in addition to influencing the mitochondrial transmembrane potential (ΔΨm) and ΔpH, calcium transport, production of ROS and mitochondrial dynamics. Activation of a mitochondrial calcium‐dependent K⁺ channel modulates K⁺ uptake and matrix volume and has been proposed to increase bioenergetic efficiency (Aon et al., 2010). Alternatively, a protective mechanism involves activation of different IMM K⁺ channels that induces a slight uncoupling effect that would in turn decrease energetic efficiency (Cardoso et al., 2010). Activity of the evolutionarily conserved ATP‐dependent K⁺ channel mitoKATP (Garlid et al., 2009; Lefer et al., 2009), of BKCa (Frankenreiter et al., 2017), of the voltage‐gated channel K_v7.4 (Testai et al., 2015a, 2015b) as well as of the SK channels (Dolga et al., 2013) has been linked to ischaemic preconditioning, ischaemic postconditioning and cytoprotection in general. For this latter channel, a recent work reveals that a neuroprotective mechanism involves attenuation of [Ca²⁺]_m uptake upon SK channel activation and that respiration and complex I activity upon pharmacological activation or overexpression of mitochondrial K_Ca2.2 (SK2) channels resulted in reduced mitochondrial ROS formation (Honrath et al., 2017a, 2017b). However, in general, the exact basis of K⁺ channel openers' cytoprotective effects still remains to be elucidated (for recent reviews, see, e.g. Honrath et al., 2017a; Krabbendam et al., 2018). Importantly, in the case of mtBKCa, a cGMP‐dependent protective effect in the cases of acute cardiac damage and adverse long‐term events that occur after myocardial infarction has been demonstrated also in vivo, thanks to tissue‐dependent knockout of BKCa in the cardiac tissue in mice (Frankenreiter et al., 2017).

Interestingly, many of the K⁺ channels found in the IMM (in addition to the PM and other membranes) are highly overexpressed in cancer cells/tissues. For example, the two‐pore leak channel K_2P9.1(TASK‐3) is overexpressed five‐ to 100‐fold in 44% of breast tumours and is also highly expressed in lung, colon and ovarian cancers (Peruzzo et al., 2016). Its overexpression promotes tumour formation and confers resistance to hypoxia in vitro, suggesting that its up‐regulation plays a pathologically important role in human breast cancer (Mu et al., 2003). IKCa (also called K_Ca3.1) is expressed in almost all migrating cells and various studies point to its important role in the control of proliferation in human breast cancer, in hepatocellular carcinoma, in chronic lymphocytic leukaemia (B‐CLL) and in lung cancer (for review, see Schwab et al., 2012). mtIKCa is functional in HCT116 colon carcinoma and HeLa cells (Sassi et al., 2010), is inhibited by http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2336 and http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2330 (De Marchi et al., 2009) and regulates oxidative phosphorylation in pancreatic ductal adenocarcinoma cells (Kovalenko et al., 2016). In vitro data suggest that inhibition of IKCa and likely of mtIKCa by membrane‐permeant inhibitors sensitizes melanoma cells to B‐Raf inhibitors, such as http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=5893, and induces release of mitochondrial ROS (Bauer et al., 2017).

More information is available about mtK_v1.3: it has been demonstrated to have a crucial role in apoptosis in various cancer cells (Leanza et al., 2012b; Szabo et al., 2008). The K_v1.3 channel is overexpressed in various cancer tissues/cells (Pardo and Stuhmer, 2014) and is important for anti‐tumour immunity (Eil et al., 2016). Membrane‐permeant inhibitors of the K_v1.3 channel such as http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?tab=summary&ligandId=9184 and psoralen derivatives act by inducing intrinsic apoptosis via the following chain of events: stopping the depolarizing K⁺ influx causes IMM hyperpolarization, with ensuing increased ROS level, PTP activation, swelling, loss of Δψm, loss of cytochrome c and further ROS release (Leanza et al., 2015; Checchetto et al., 2018). These drugs also killed pathological cells in the absence of Bax and Bak, independently of p53 status and of Bcl‐2 overexpression thanks to the direct targeting of mt function and permeability at the level of the inner membrane. In vivo experiments in melanoma and pancreatic ductal adenocarcinoma orthotopic models corroborated the effectiveness of Kv1.3 inhibitors and showed their selective action only on pathological cells. Specificity depends on the synergy between different factors such as high K_v1.3 expression and altered basal redox state in cancer cells (Leanza et al., 2017).

Most of the evidence supporting the involvement of the above K⁺ channels in protection against ischaemic/reperfusion damage (Szewczyk et al., 2010; Laskowski et al., 2016) or induction of apoptosis (Szabo et al., 2012) is pharmacological (for a review summarizing the pharmacology of these channels, see, e.g. Szabo and Zoratti, 2014) (Figure 4). Unfortunately, other cellular targets possibly accounting for the observed effects have been identified (Augustynek et al., 2017). Activators of BKCa, such as Ca²⁺, diCl‐DHAA (12,14‐dichlorodehydroabietic acid), http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4272, http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=8802 and hypoxia, or inhibitors of this channel like http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2328, http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4218 and http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2309 do not only seem to act on the mitochondrial mtBKCa channel (Szewczyk et al., 2009). CGS7184 (ethyl1‐[[(4‐chlorophenyl)amino]oxo]‐2‐hydroxy‐6‐trifluoromethyl‐1H‐indole‐3‐carboxy‐late), a synthetic BKCa channel opener, directly activates the ryanodine receptor calcium release (http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=748) channel in the sarcoplasmic reticulum (Wrzosek et al., 2012). CGS7184 and CGS7181, in contrast to NS1619 and http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4271, were also able to induce cell death in neurons (Augustynek et al., 2018). Furthermore, SERCA, complex I of the respiratory chain and ATP‐synthase inhibition are involved in pleiotropic effects of the BKCa channel activator NS1619 on endothelial cells (Lukasiak et al., 2016). One of the most specific BKCa activators identified so far is NS11021 (N′‐[3,5‐bis(trifluoromethyl)phenyl]‐N‐[4‐bromo‐2‐(2H‐tetrazol‐5‐yl‐phenyl]thiourea; however, possible intracellular off‐target effects of this drug have not been studied in detail to our knowledge. KATP channel activators instead might act as uncouplers (Szewczyk et al., 2010) and K_Ca3.1 (IKCa) as well as K_Ca2.1 (SKCa) modulators may well have off‐target effects, similarly to those affecting BKCa channels. In the case of mtK_2P9.1 (TASK‐3) channels, no highly specific modulators are available but dihydropyrrolo[2,1‐α]isoquinolines compounds that are able to inhibit these K_2P (TASK) channels could become possible candidates for developing new, selective inhibitors (Noriega‐Navarro et al., 2014).

Figure 4.

Chemical structure of some modulators of other mitochondrial ion channels: (A) MCU; (B) K_v1.3; (C) K_Ca1.1 (BK_Ca); (D) K_Ca3.1(IK_Ca); (E) K_2P9.1 (TASK‐3). For further information see text and, for example, Augustynek et al. (2017).

One possible solution to the issue of pleotropic effects of the drugs due to multiple localization of the channels within the cells is chemical targeting of the most specific modulators of these channels to mitochondria by virtue of addition of a functional lipophilic cation (e.g. triphenylphosphonium group) that drives the accumulation of a certain drug into mitochondria (Murphy and Smith, 2007). This strategy has been successfully applied also in vivo in the case of mtK_v1.3 (see below). In summary, the relationship between mitochondrial potassium transport and diseases linked to altered mitochondrial function is still only partially explored, but the data available so far point to IMM K⁺ channels as possible targets for therapeutic applications against various pathologies, including ischaemia and cancer, and possibly neurodegenerative diseases (see, e.g. Matschke et al., 2018).

Inner mitochondrial membrane chloride channels

In addition to CIC‐4 in the OMM, CIC‐1 (Yang et al., 2009) and CIC‐5 (Ponnalagu et al., 2016a, 2016b) have been located to the IMM where the latter plays a role in regulating the release of mitochondrial ROS from respiratory chain complexes. Indeed, the emerging view is that mitochondrial chloride channels also have a role in cardiac function and cardioprotection from ischaemia–reperfusion (IR) injury. Chloride intracellular channels (CIC) are non‐classical ion channels lacking a signal sequence for membrane targeting. Interestingly, CIC‐like proteins exist even in bacteria (Gururaja Rao et al., 2017; Ponnalagu and Singh, 2017). Among CICs in mammals, CIC‐1, CIC‐4 and CIC‐5 are abundant in the heart. In contrast to CIC‐1 and CIC‐4 (found in the OMM, see above), CIC‐5 does not show co‐localization to the endoplasmic reticulum (Ponnalagu et al., 2016b). Again, as yet, no isoform‐specific modulators of CICs are available (for recent review, see Ponnalagu and Singh, 2017). As to the long‐studied IMAC, due to the uncertainties regarding its molecular identity, only general modulators have been identified (for review, see Ponnalagu and Singh, 2017).

Successfully applied strategies for in vivo subcellular targeting of small molecule modulators of mitochondrial ion channels

Thanks to the highly negative matrix side membrane potential across the IMM, mitochondria can be massively ‘loaded' with positively charged, membrane‐permeant compounds, achieving concentration gradients of up to 1000‐fold. This concept has been used in many studies since its inception 60 years ago (Liberman et al., 1969) (for reviews, see, e.g. Ross et al., 2005; Horobin et al., 2007; Hoye et al., 2008; Murphy, 2008; Smith et al., 2008, 2011; Zielonka et al., 2017). Molecules directed to mitochondria must obviously be capable of entering the cell and be at least moderately lipophilic. If mitochondrial targeting is associated with the presence of a positive charge, as is generally the case, the charge is therefore distributed by resonance over a lipophilic structure. Since the seminal study of Liberman, Skulachev and colleagues (1969), this has most often been the triphenylphosphonium group, connected to the rest of the molecule by a linker, commonly an alkyl chain. Other moderately lipophilic positively charged groups that have been used for this purpose have been mainly heteroaromatic systems, including rhodamines (e.g. Chernyak et al., 2013), berberine and palmatine (Chernyak et al., 2013), pyridinium (Beckham et al., 2013; Hou et al., 2014), indolium (Huang et al., 2017) and cyanines. Constructs comprising a number of positively charged guanidinium groups have also been found to be effective mitochondria‐targeting devices (Fernandez‐Carneado et al., 2005; Maiti et al., 2007). These constructs may be considered as non‐peptidic penetrating agents.

The dication dequalinium – more generally a family of compounds comprising two quinolinium moieties linked by an aliphatic chain spacer –was used early on to obtain mitochondria‐targeted nanocarriers (DQAsomes), intended principally as a tool for mitochondrial transfection (e.g. Weissig, 2015). Interaction with the mitochondria was proposed to be mediated by cardiolipin (Weissig and Torchilin, 2000; Weissig et al., 2001). Dequaliniums are known inhibitors of K⁺ channels (Castle et al., 1993; Galanakis et al., 1996; Malik‐Hall et al., 2000), and – unsurprisingly at this point – they are toxic for leukaemia cells (Galeano et al., 2005; Sancho et al., 2007; Garcia‐Perez et al., 2011). Toxicity has been linked to redox unbalance (Sancho et al., 2007; Garcia‐Perez et al., 2011). DQAsomes are at any rate only one example of the several nanostructured devices that have been devised to selectively transport cargo to mitochondria (reviewed by: Marrache et al., 2015; Ma et al., 2016; Paleos et al., 2016; Sato et al., 2016). The underlying principle generally is the use of positive charges to promote delivery to mitochondria.

To minimize the possibility of interference with the pharmacological activity, one may resort to prodrugs in which the mitochondria‐targeting portion is shed, hopefully after performing its task. In some such studies, the strategy has relied on the incorporation, into the linker portion of the molecule, of a hydrolyzable bond system such as a carbamate group ( Leanza et al., 2017; Peruzzo et al., 2017; Teixeira et al., 2017; Venturini et al., 2017; Mattarei et al., 2018). Insurance that the regeneration of the active principle takes place only, or mostly, at mitochondria may be a desirable feature. This may be achieved by exploiting a mitochondrial enzyme, such as nitroreductase (Chevalier et al., 2016) or aldehyde dehydrogenase ALDH‐2 (Ripcke et al., 2009), or a reactive compound plentiful in mitochondria, such as H₂O₂ (McQuaker et al., 2013).

The lipophilic or amphipathic character of mitochondria‐targeted constructs, necessary to allow permeation of cell membranes, also confers affinity for the membranes themselves, leading to extensive binding, in particular of triphenylphosphonium (TPP)‐comprising compounds (James et al., 2007; Ross et al., 2008). Association with membranes may pose difficulties but also be a bonus if the targeted proteins, like ion channels, reside in mitochondrial membranes. This strategy has been successfully applied recently in the case of the mitochondrial potassium channel K_v1.3. MtK_v1.3 mediates an inward potassium flux to the mitochondrial matrix and likely modulates coupling between ATP synthesis and mitochondrial respiration. In vivo evidence has been obtained suggesting that inhibition of mtK_v1.3 by pharmacological means represents an unconventional but promising strategy to selectively eliminate cancer cells. The K_v1.3 (and mtKv.3) channel is overexpressed in various cancer tissues/cells, and mtK_v1.3 was identified as a novel target of Bax (Szabo et al., 2008, 2011). http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=9583, PAP‐1 (5‐(4‐phenoxybutoxy)psoralen) and clofazimine, three distinct membrane‐permeant inhibitors of K_v1.3 channels (Cahalan and Chandy, 2009), induced death in multiple human and mouse cancer cell lines by triggering a series of events involving hyperpolarization, ROS release, activation of PTP and apoptosis (Szabo and Zoratti, 2014). In contrast, membrane‐impermeant, selective and high‐affinity K_v1.3 inhibitors http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2549 or http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2547 did not trigger cell death, suggesting a crucial role for the mitochondrial K_v1.3 versus PM K_v1.3 channel. In preclinical mouse models for melanoma (Leanza et al., 2012a) and for pancreatic ductal adenocarcinoma (PDAC) (Zaccagnino et al., 2016), an i.p. injection of clofazimine significantly reduced tumour size while no adverse effects were observed in several healthy tissues. Clofazimine is a clinically used drug with anti‐inflammatory and immunosuppressive activities (Cholo et al., 2012). This drug also inhibits the transporter http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=768 (Van Rensburg et al., 1994) and may activate the tumour suppressor p53 (Casey et al., 2009; Patil, 2013). Therefore, a more specific drug, acting exclusively on the mitochondrial counterpart of K_v1.3, was required to more efficiently and specifically target this channel. Two mitochondriotropic derivatives of PAP‐1, the most specific membrane‐permeant inhibitor, were designed and synthesized and shown to drastically reduce tumour growth at a low concentration in animal models for melanoma and PDAC, without causing in vivo side effects such as cardiotoxicity and immune depletion (Leanza et al., 2017). These mitochondria‐targeted mtK_v1.3 inhibitors, therefore, can be considered especially promising molecules for the treatment of several types of cancer, provided the tumour cells express high levels of mtK_v1.3. A recent study identified some TPP‐related compounds that might even eradicate cancer stem cells (Ozsvari et al., 2018).

An alternative strategy to target mitochondria may still exploit transmembrane potential differences but make use of peptides rather than of lipophilic charged groups like TPP (reviewed by: Yousif et al., 2009b; Frantz and Wipf, 2010; Zielonka et al., 2017). This approach may be considered to be related to the use of so‐called ‘cell‐penetrating peptides' (CPPs) to overcome the barrier posed by the plasma membrane (reviewed by: Borrelli et al., 2018; Guidotti et al., 2017; Kalafatovic and Giralt, 2017). CPPs are short sequences of amino acids, typically comprising basic residues along with hydrophobic ones, which are taken up by cells mostly via endocytotic pathways (clathrin‐ or caveolin‐mediated, macropinocytosis) – entailing a subsequent escape from the endosome – or via processes involving a degree of membrane disorganization, for example, the formation of transient pores, which can result in toxicity (reviewed by: Cleal et al., 2013; LeCher et al., 2017; Walrant et al., 2017).

Reaching the cytosol in principle allows a second‐stage delivery to other subcellular compartments (Cerrato et al., 2017; Ngwa et al., 2017). This has been achieved in a mitochondria‐specific manner with the so‐called Szeto–Schiller peptides, in particular SS‐31 (e.g. Szeto, 2006, 2014) (for other examples, see Cerrato et al., 2017; Zielonka et al., 2017). These devices, which interact with cardiolipin, have antioxidant effects. Peptides can in principle serve to transport small‐molecule cargos linked via peptide bonds (Yousif et al., 2009a; Jean et al., 2016). A variation on the theme involves gramicidin S, a natural peptidic antibiotic with a marked affinity for mitochondrial membranes (Kanai et al., 2007; Frantz and Wipf, 2010).

Mitochondrial leader peptide sequences in principle offer a specific approach, as they are recognized and accepted by the mitochondrial protein import system (MPTS), which, incidentally, needs a transmembrane potential to operate (Wiedemann and Pfanner, 2017). The attempts to deliver cargo to mitochondria in this manner have been few (Vestweber and Schatz, 1988, 1989; Flierl et al., 2003), but the idea has been adapted to favour interactions of multifunctional nanocarriers with mitochondria (Kawamura et al., 2013).

With regard to mitochondrial ion channels, peptides have been used to functionally modulate VDAC. The most relevant case may be that of its interaction with hexokinase, which results in inhibition of apoptosis, promotion of glycolysis and cancer cell survival (Nakashima, 1989; Zaid et al., 2005; Galluzzi et al., 2008; Krasnov et al., 2013). The interaction was mapped to the N‐terminal portion of VDAC, and detachment of hexokinase was accomplished by methyljasmonate, a plant hormone (Goldin et al., 2008). More efficient are specific peptides based on the relevant sequences and made into membrane‐penetrating peptide that antagonized the interaction and had pro‐apoptotic effects in vitro (Arzoine et al., 2009; Prezma et al., 2013). A recent paper reports that two VDAC1 N‐terminal sequences fused to a cell‐penetrating sequence from the Drosophila antennapedia‐homeodomain or to a human transferrin receptor (hTfR)‐recognition sequence, respectively, were able to attack glioblastoma, a notoriously elusive target, also in an in vivo murine xenograft model. At least the transferrin‐based construct was able to cross the blood brain barrier and was thus effective in an orthotopic model upon i.v. administration as the free compound or as an inclusion in PLGA nanoparticles (Shteinfer‐Kuzmine et al., 2017). The same group provided evidence that a cell‐penetrating peptide derived from VDAC1 is also highly effective against hepatocellular carcinoma in a preclinical model (Pittala et al., 2018).

It should also be mentioned that the pH difference between the mitochondrial matrix and cytoplasm (more alkaline in the matrix) may be exploited to drive the accumulation of permeant weak acids into mitochondria. Normally, however, this difference is in the order of only 0.5 units, permitting a concentration difference of about three. This approach therefore has not been much used, but it should be kept in mind.

Conclusions

The field of mitochondrial ion channels can be likened to a vast stretch of territory comprising some freshly ploughed, fertile, already productive land and a dense thicket waiting to be cleared to put the ground to further good use. To do so will involve imaginative research to design specific drugs or specifically targeted drugs and to reach a more detailed understanding of their distinct localization‐dependent functions, of the partnerships – with proteins and lipids – involved in the execution of these functions, of the signalling pathways affected, of the mechanisms regulating ion channel localization and of their possible modulation, of possible species and/or tissue specificities for each of these aspects. Subcellular targeting of drugs is clearly a new frontier of pharmacological science and the practical exploitation of mitochondrial channels for the improvement of public health is one of the goals the metaphorical pioneers/researchers have set for themselves. As in many other cases, the effectiveness and specificity of drug delivery is a major hurdle.

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017a,b,c,d,e,f).

Conflict of interest

The authors declare no conflicts of interest.

Leanza, L. , Checchetto, V. , Biasutto, L. , Rossa, A. , Costa, R. , Bachmann, M. , Zoratti, M. , and Szabo, I. (2019) Pharmacological modulation of mitochondrial ion channels. British Journal of Pharmacology, 176, 4258–4283. 10.1111/bph.14544.