Emerging Roles in the Biogenesis of Cytochrome c Oxidase for Members of the Mitochondrial Carrier Family

The mitochondrial carrier family (MCF) is a group of transport proteins that are mostly localized to the inner mitochondrial membrane where they facilitate the movement of various solutes across the membrane. Although these carriers represent potential targets for therapeutic application and are repeatedly associated with human disease, research on the MCF has not progressed commensurate to their physiologic and pathophysiologic importance. Many of the 53 MCF members in humans are orphans and lack known transport substrates. Even for the relatively well-studied members of this family, such as the ADP/ATP carrier and the uncoupling protein, there exist fundamental gaps in our understanding of their biological roles including a clear rationale for the existence of multiple isoforms. Here, we briefly review this important family of mitochondrial carriers, provide a few salient examples of their diverse metabolic roles and disease associations, and then focus on an emerging link between several distinct MCF members, including the ADP/ATP carrier, and cytochrome c oxidase biogenesis. As the ADP/ATP carrier is regarded as the paradigm of the entire MCF, its newly established role in regulating translation of the mitochondrial genome highlights that we still have a lot to learn about these metabolite transporters.

The mitochondrial carrier family (MCF) is a group of transport proteins that are mostly localized to the inner mitochondrial membrane where they facilitate the movement of various solutes across the membrane. Although these carriers represent potential targets for therapeutic application and are repeatedly associated with human disease, research on the MCF has not progressed commensurate to their physiologic and pathophysiologic importance. Many of the 53 MCF members in humans are orphans and lack known transport substrates. Even for the relatively well-studied members of this family, such as the ADP/ATP carrier and the uncoupling protein, there exist fundamental gaps in our understanding of their biological roles including a clear rationale for the existence of multiple isoforms. Here, we briefly review this important family of mitochondrial carriers, provide a few salient examples of their diverse metabolic roles and disease associations, and then focus on an emerging link between several distinct MCF members, including the ADP/ATP carrier, and cytochrome c oxidase biogenesis. As the ADP/ATP carrier is regarded as the paradigm of the entire MCF, its newly established role in regulating translation of the mitochondrial genome highlights that we still have a lot to learn about these metabolite transporters.
Keywords: ADP/ATP carrier, cytochrome c oxidase, mitochondrial carrier family, mitochondrial translation, respiratory supercomplexes, solute carrier family THE SOLUTE CARRIER (SLC) FAMILY Transport of substrates across biological membranes between and among organelles is an important feature of eukaryotic cells. The SLC family, the second largest family of membrane proteins, is a large group of membrane transport proteins; in humans, there are 456 known members that are grouped into 65 subfamilies (Höglund et al., 2011;Perland and Fredriksson, 2017). SLCs facilitate the movement of otherwise membrane-impermeable solutes-such as amino acids, ions, nucleotides, sugars and drugs-across biological membranes. The family includes functionally related proteins that mediate the transport and exchange of solutes across cell membranes. Transport can be facilitative by simply allowing solutes to equilibrate across a membrane according to their relative distribution on either side. Additionally, SLCs can mediate secondary active transport by coupling the downhill flow of one substrate, often an ion, to the uphill movement of another substrate against its relative gradient across a membrane. Primary active transporters, ion channels and aquaporins are not included in the SLC family. The criterion for membership in the SLC family is being an integral membrane protein that transports a solute. Not surprisingly, the SLC family is structurally quite diverse. However, within an individual sub-family, members often share more than 20% sequence homology . Table 1 describes the current list of SLC family members based on http://slc.bioparadigms.org and provides references that review each subfamily. Families SLC53-65 are newly registered, and are based on a work presented at the BioMedical Transporters 2017 conference in Lausanne, Switzerland.

MITOCHONDRIAL CARRIER FAMILY (SLC25)
The Solute Carrier 25 (SLC25) family transports solutes across the inner mitochondrial membrane (IMM), although several members of this family are localized to other cellular organelles such as chloroplasts and peroxisomes (Visser et al., 2002;Bedhomme et al., 2005). The MCF is the largest SLC subfamily and all members are encoded by the nuclear genome. As such, they are synthesized by cytoplasmic ribosomes and need to be imported from the cytosol to their final location. 35 members have been identified in yeast, 58 members in Arabidopsis thaliana, and 53 members have been identified in humans. As summarized in Table 2 (Palmieri and Pierri, 2010), substrates for approximately one third of the MCF have yet to be identified. Tissue distribution can vary from ubiquitous [e.g., SLC25A6 (Stepien et al., 1992)] to tissue-specific [e.g., SLC25A31 (Dolce et al., 2005;Rodić et al., 2005)].
SLC25 members are generally characterized by the presence of a tripartite structure of approximately 300 amino acids, six conserved transmembrane regions, and the three-fold repeated MCF signature motif, P-X-[DE]-X-X-[RK] (Figure 1). Although their substrates vary in size and nature, most members catalyze the exchange of one solute for another (antiport), couple the transport of one solute with another (symport), or facilitate the transport of a solute (uniport). Because of their sequence similarity, it is assumed that the transport mechanism is similar for the extended family.

Physiology of MCF
Collectively, the MCF transports a wide range of solutes across the IMM. In this capacity, they act as important bridges that link many biochemical pathways that are otherwise compartmentalized in either the cytosol or the mitochondrial matrix (Palmieri, 2014). Solutes transported include protons, nucleotides, amino acids, carboxylic acids, inorganic ions, and cofactors. Their fundamental role in enabling metabolic compartmentalization cannot be overemphasized. SLC25 family members are involved in metabolic pathways such as heme synthesis and metal homeostasis (SLC25A28, SLC25A37, and SLC25A38), fatty acid metabolism (SLC23A1 and SLC25A20), amino acid metabolism (SLC25A12, SLC25A13, SLC25A18, and SLC25A22), nucleic acid metabolism (SLC25A26, SLC25A33, and SLC25A36), urea production (SLC25A2, SLC25A13, and SLC25A15) (Shayakul and Hediger, 2004;Shayakul et al., 2013;LeMoine and Walsh, 2015), OXPHOS (SLC25A3, SLC25A4, SLC25A5, SLC25A6, and SLC25A31) and heat generation (SLC25A7 and SLC25A9). As a thorough discussion of the entire SLC family is not the goal of this review, in the next four sections, the physiology of select SLC25 members is briefly discussed to illustrate the diversity of cellular functions in which its members participate.
The ADP/ATP Carrier Protein Is Essential for OXPHOS Eukaryotic cells make energy in the form of ATP in the mitochondrial matrix and the ATP is translocated through the impermeable IMM to power many processes in the cell. The ADP/ATP carriers (AACs) provide the means of transport of ATP and its precursor ADP, across the IMM. Under physiological conditions, 1 molecule of ADP from the cytosol is exchanged for 1 molecule of matrix-localized ATP by the activity of the ADP/ATP carrier. AAC, referred to as ANT in humans, is a notable MCF member as it was the first to have its amino acid sequenced (Aquila et al., 1982) and its 3D structure solved (Pebay-Peyroula et al., 2003). Similar to all members of the MCF and regarded as a paradigm for this family, AACs are nuclear-encoded, integral membrane proteins with approximately 300 amino acids arranged into three repeats linked by two loops on the cytosolic side. There are two transmembrane α-helices in each repeat connected together by a long loop on the matrix side, giving the carrier a threefold pseudosymmetry (Palmieri, 2013).
One of the most abundant proteins in the IMM, Aacs are encoded by multiple different genes in both unicellular and multicellular eukaryotes. There are three yeast AAC isoforms and four human ANT isoforms. The human ANT isoforms overlap in their expression pattern but exhibit tissue-specificity. ANT1 (SLC25A4) is the most equivalent to yeast Aac2p and the predominant isoform in the heart and skeletal muscle (Stepien et al., 1992). ANT2 (SLC25A5) is mostly expressed in regenerative tissues such as the kidney and liver, ANT3 (SLC25A6) is ubiquitously expressed at low baseline levels, and ANT4 (SLC25A31) is selectively expressed in the testis (Stepien et al., 1992;Doerner et al., 1997;Dolce et al., 2005;Rodić et al., 2005;Kim et al., 2007;Dupont and Stepien, 2011). Aac2p is the most abundant of all three isoforms in yeast and the only one absolutely required for OXPHOS and growth on respiratory carbon sources (Lawson et al., 1990). Aac1p and Aac3p are minor isoforms in yeast that are undetectable at the protein level under normal growth conditions. Aac1p expression is repressed in hypoxic conditions (Gavurníková et al., 1996) and Aac3p expression is induced in anaerobic situations (Sabová et al., 1993).
Recently it was demonstrated that RSCs are functional entities (Barrientos and Ugalde, 2013;Lapuente-Brun et al., 2013) whose structures have since provided novel insight into the potential benefits that they may confer (Schäfer et al., 2007;Althoff et al., 2011;Dudkina et al., 2011;Genova and Lenaz, 2014;Gu et al., 2016;Letts et al., 2016;Wu et al., 2016). Functional benefits of RSCs that have been suggested but not yet proven include: improved electron transfer efficiency and reduced ROS generation, each stemming from a substrate channeling based mechanism; increased metabolic flexibility resulting from changes in RSC composition; and finally, enhanced stability and functionality of all participating complexes in the specific context of the protein-dense IMM (Barrientos and Ugalde, 2013;Milenkovic et al., 2017). Nevertheless, there is still some controversy as to the functional and physiological relevance of RSCs. About 10 years ago, a new functional entity was shown to interact with yeast RSCs: Aac2p (Claypool et al., 2008;Dienhart and Stuart, 2008). More recently, this association was shown to be evolutionarily conserved as two distinct human ANT isoforms also form complexes with RSCs . Functionally, this conserved interaction could benefit both RSCs and the AACs. Specifically, the electrogenic exchange of ATP in /ADP out by AAC/ANT is positively influenced by the membrane potential ( ) across the IMM (Krämer and Klingenberg, 1980) which of course is established by the electron transport chain. Similarly, by dissipating the electrical gradient, productive AAC/ANT transport makes it easier for RSCs to pump protons. As such, it is reasonable to hypothesize that this known functional synergy is further enhanced by being physically associated.
ADP/ATP carriers interact with RSCs in the presence of cardiolipin, a unique phospholipid found exclusively in the mitochondrion (Claypool et al., 2008). Available structures of AAC/ANTs depict three tightly bound cardiolipin molecules per monomer (Beyer and Klingenberg, 1985;Pebay-Peyroula et al., 2003;Ruprecht et al., 2014). In the absence of cardiolipin, Aac2p function is impaired and Aac2p assembly is drastically altered (Jiang et al., 2000;Claypool et al., 2008). The absence of cardiolipin also destabilizes the RSCs (Zhang et al., 2002;Pfeiffer et al., 2003), including its association with Aac2p (Claypool et al., 2008). That Aac2p assembly and function is cardiolipin-dependent has led to the hypothesis that the assembly and function of AAC/ANTs may be the "Achilles heel" of a multitude of cardiolipin-based diseases (Klingenberg, 2008;Claypool, 2009). The structural changes in RSCs and Aac2p that occur in the absence of cardiolipin have clear functional consequences (Claypool et al., 2008). However, the relative contribution of each structural change-impaired assembly of RSCs, Aac2p, or RSC-Aac2p-that occurs in the absence of cardiolipin to the associated mitochondrial dysfunction has not been established.
It has also been hotly debated whether the protein exists and/or functions as a monomer or dimer. Mitochondrial carriers were originally accepted to exist and function as homo-dimers (Lin et al., 1980;Klingenberg, 1981;Palmisano et al., 1998;Schroers et al., 1998;Kotaria et al., 1999;Trézéguet et al., 2000;Capobianco et al., 2002;Dyall et al., 2003;Nury et al., 2005;Postis et al., 2005). A number of studies, motivated by the crystal structures (Kunji and Harding, 2003;Pebay-Peyroula et al., 2003;Ruprecht et al., 2014), have challenged this initial view and instead provided evidence that AACs function, and in fact exist in the IMM, as monomers (Kunji and Harding, 2003;Bamber et al., 2006Bamber et al., , 2007aKunji and Crichton, 2010). Obviously, AACs cannot be monomeric in the IMM and interact with RSCs in a cardiolipin-dependent manner. Additional work is needed to reconcile these very different models of AAC connectivity.

Mitoferrins Are Fundamental to Iron Transport
Iron is essential for mitochondrial function (Levi and Rovida, 2009). Mitochondria themselves are intimately involved in the regulation of cellular iron. Iron is important in the heme biosynthetic pathway in the reaction step of ferrous iron incorporation into protoporphyrin IX catalyzed by ferrochelatase (Figure 2) (Ponka, 1997). Heme is needed for synthesis of the mitochondrial cytochromes which are electron carriers critical for OXPHOS. In addition, iron-sulfur cluster biogenesis occurs in the mitochondrial matrix and is tightly linked to many other cellular processes such as heme biosynthesis, ribosome assembly, DNA synthesis, and translation initiation (Lill and Mühlenhoff, 2008;Lill, 2009). SLC25A28 and SLC25A37 encode Mitoferrin 2 (MFRN2) and Mitoferrin 1 (MFRN1), respectively, which are involved in iron import into the mitochondrion. In zebrafish and mammals, MFRN1 is expressed predominantly in hematopoietic tissues whereas MFRN2, with 65% amino acid identity to its paralog, is widely expressed (Shaw et al., 2006;Amigo et al., 2011). MFRN2 has about 38% identity to Mrs3p and Mrs4p (Shaw et al., 2006), two yeast transporters originally identified as suppressors of an intron splicing defect (Waldherr et al., 1993) that have since been associated with iron transport (Foury and Roganti, 2002). Yeast lacking Mrs3p and Mrs4p exhibit poor growth in iron-depleted conditions (Foury and Roganti, 2002). MFRN1 loss-of-function in mice and zebrafish results in reduced iron uptake into mitochondria and defective hemoglobin synthesis (Shaw et al., 2006). In non-erythroid cells, MFRN2 and MFRN1 are both involved in mitochondrial iron uptake (Paradkar et al., 2009). When both transporters are silenced in non-erythroid cells, heme synthesis is severely compromised; further overexpression of one can functionally compensate for the loss of the other (Paradkar et al., 2009). These results establish the fundamental importance of these proteins in mitochondrial iron metabolism in erythroid and non-erythroid cells.

Uncoupling Proteins Provide a Pathway for Proton Leakage
The UCPs are regulated mitochondrial proteins known to transport protons, anions or other mitochondrial substrates (Jezek et al., 2010;Fedorenko et al., 2012;Porter, 2012;Monné et al., 2018). Six UCP homologs have been discovered in humans-UCP1 or thermogenin , UCP2 (Fleury et al., 1997), UCP3 (Boss et al., 1997), UCP4 (Mao et al., 1999), UCP5 or BMCP1 for brain mitochondrial carrier protein 1 (Sanchis et al., 1998), and UCP6 or KMCP1 for FIGURE 2 | Overview of the heme biosynthetic pathway. Three known MCF members are involved in heme biosynthesis. Following its transport into the matrix by Hem25p/SLC25A38, glycine is condensed with succinyl-CoA by ALA synthase to form δ-aminolevulinic acid. The next four steps of the heme biosynthetic pathway occur in the cytosol. The identity of the protein, which may be a MCF member, that mediates the transport of δ-aminolevulinic acid across the inner membrane has not been determined. The active sites of coproporphyrinogen III oxidase and Protoporphyrinogen oxidase face the intermembrane space. In contrast, the final step in the heme biosynthetic pathway occurs in the matrix and is catalyzed by ferrochelatase. The identity of the protein, which may be a MCF member, that transports protoporphyrin IX to the matrix has not been determined. Ferrochelatase incorporates iron (Fe), transported into the matrix by the mitoferrins SLC25A28 and SLC25A37, into protoporphyrin IX to produce heme. kidney mitochondrial carrier protein 1 (Haguenauer et al., 2005). UCPs uncouple OXPHOS from ATP synthesis; they dissipate proton gradients by allowing protons that have been pumped into the intermembrane space by respiratory complexes to flow back into the mitochondrial matrix without being utilized for ATP synthesis (Figure 3). The translocation of hydrogen ion by UCPs requires fatty acids and this activity is inhibited by purine nucleotides such as GDP. Until recently, the role of fatty acids in activating the uncoupling process was hotly debated in relation to the putative UCP transport mechanism. Proton and anion transport by UCP involves fatty acids, as extensively reviewed (Jezek et al., 2018). According to the fatty acid cycling mechanism, protonated free fatty acids spontaneously flip to the matrix IMM leaflet where they release protons based on the proton gradient across the IMM. In turn, UCPs provide a conduit consisting of basic amino acids that facilitates the FIGURE 3 | Uncoupling protein (UCP) provides a pathway for proton leak. (A) Proton pumping across the inner membrane into the intermembrane space by complexes I, III, and IV occurs as electrons (e -) flow down the transport chain. This generates a proton-based electrochemical gradient ( ψ) that is consumed by the ATP synthase to produce ATP, thus coupling ATP synthesis to proton pumping. The proton gradient also provides the driving force for the exchange of ADP and ATP by AAC and the uptake of P i by PIC, transport processes that are essential for OXPHOS. (B) UCPs provide an alternative means to equilibrate the proton gradient which effectively uncouples proton pumping from ATP synthesis. This results in increased flux of electrons through the electron transport chain which generates heat (indicated by red intermembrane space). In (A,B), line thickness reflects relative flux/activity. movement of the FA anion back to the IMS leaflet, a process that is also driven by the electrochemical gradient. In the net, these processes dissipate the proton gradient (Kamp and Hamilton, 1992;Kamp et al., 1993Kamp et al., , 1995Jezek et al., 1997a,b;Jabùrek et al., 2001;Jezek et al., 2010Jezek et al., , 2018. This model is validated by nuclear magnetic resonance and functional mutagenesis studies on UCP2 which provided molecular and structural support for this protonophoretic model (Berardi and Chou, 2014).
The UCP1 is thought to be exclusively found in brown adipose tissue and is the only UCP responsible for adaptive adrenergic non-shivering thermogenesis Matthias et al., 2000;Golozoubova et al., 2001;Porter, 2008). As such, it is firmly established that UCP1 functions as a true UCP that utilizes the electrochemical gradient generated by the respiratory chain to produce heat instead of ATP. Also, there seems to be a close relationship between mitochondrial ROS and UCP1-dependent thermogenesis although whether or not superoxide modulates UCP1 function is debated (Echtay et al., 2002;Silva et al., 2005;Chouchani et al., 2017). Nonetheless, work with ucp1 knockout (ucp1 −/− ) mice has established that UCP1 is a target of redox modification in vivo (Chouchani et al., 2016). A recent study showed that brown adipose tissue (BAT) from ucp1 −/− mice have reduced respiratory chain proteins and increased host defense signaling following exposure to cold . Intriguingly, BAT-derived mitochondria from ucp1 −/− mice are more sensitive to calcium overload in a ROS-dependent manner . Thus, though UCP1 is traditionally linked to thermogenesis, it is becoming clear that UCP1 function extends beyond thermogenesis.
UCP2-5 are not involved in thermogenesis even though they provide mild uncoupling which may be protective against oxidative stress (Jezek et al., 2018). UCP2 transports C4 metabolites such as oxaloacetate, a function that matches well with its close phylogenetic relationship with the dicarboxylic acid carrier, SLC25A10 (Vozza et al., 2014). It is widely expressed (Fleury et al., 1997;Gimeno et al., 1997) and has numerous pathophysiological roles. For instance, due to its ability to reduce ROS generation, UCP2 participates in both host immunity and the inflammatory response (Nègre-Salvayre et al., 1997;Arsenijevic et al., 2000;Mattiasson and Sullivan, 2006). In addition, UCP2 has been implicated in body mass regulation, glucose metabolism, and carcinogenesis (Zhang et al., 2001;Horimoto et al., 2004;Mattiasson and Sullivan, 2006;Derdak et al., 2008;Li et al., 2013Li et al., , 2015Vozza et al., 2014;Sreedhar et al., 2017). UCP3 is expressed mainly in the skeletal muscle and brown adipose tissue, and minimally in the heart (Boss et al., 1997;Vidal-Puig et al., 1997) where it is important for ROS attenuation but not body mass regulation or fatty acid metabolism (Vidal-Puig et al., 2000). Indeed, mitochondria from mice with lower levels of UCP3 have increased ROS production and oxidative damage further suggesting that UCP3 protects against ROS and oxidative damage (Brand et al., 2002). UCP4 is predominantly expressed in the nervous system including different regions of the brain, the spinal cord, hair cells of the inner ear, and Merkel cells in the skin Smorodchenko et al., 2009Smorodchenko et al., , 2011. UCP4 overexpressing neuronal cell lines have reduced OXPHOS with a corresponding increase in glucose uptake and glycolysis . These metabolic changes correlate with a drop in ROS production, a reduced tendency for calcium overload and an overall increased resistance to apoptosis . Overexpression of UCP4 in pre-adipocyte cell lines stimulates proliferation, inhibits differentiation into adipocytes and protects against apoptosis . Furthermore, impaired insulin sensitivity and mitochondrial biogenesis, decreased mtDNA level and increased ROS production occurs in adipocyte cell lines overexpressing UCP4 suggesting a global negative impact of UCP4 on mitochondrial function (Gao et al., 2010). However, in UCP4 overexpressing L6 myocytes, insulin sensitivity is improved with no change in intracellular ROS production, mtDNA levels or mitochondrial biogenesis (Gao et al., 2011). Regulated UCP4 expression, therefore, seems to be critical for optimal mitochondrial and cellular function.
UCP5 is expressed principally in the central nervous system and has three different forms (long form, UCP5L with 325 amino acids; short form, UCP5S with 322 amino acids; and short insert form, UCP5S1 with 353 amino acids) (Ramsden et al., 2012). UCP5 overexpression in human SH-SY5Y cells increases proton leak, reduces mitochondrial membrane potential and ATP production, and increases oxygen consumption (Kwok et al., 2010). UCP6 has not been well studied. To date, it is associated with carcinogenesis (Nohara et al., 2012) and its expression in the kidney cortex is increased following pro-oxidant states (Haguenauer et al., 2005).

The Citrate Carrier Has a Pervasive Role in Cellular Metabolism
The ubiquitously expressed mitochondrial CiC, also known as the tricarboxylate carrier, citrate transport protein or dicarboxylic acid transporter, is encoded by SLC25A1 in mammals. It catalyzes the obligate electroneutral exchange of a tricarboxylate such as citrate for another tricarboxylate, a dicarboxylate such as malate, or phosphoenolpyruvate (Klingenberg, 1972;Bisaccia et al., 1989;Gnoni et al., 2009). CiC exports matricial citrate, the first product of the citric acid (Kreb; tricarboxylic acid) cycle, in exchange for cytosolic malate. Citrate is an important regulatory substrate for many metabolic reactions. As such, CiC provides substrates for both fatty acid and cholesterol biosynthesis (Figure 4) and helps in the transfer of reducing equivalents across the IMM (Palmieri et al., 2015). For fatty acid synthesis, citrate is source of carbons and also allosterically regulates the rate-limiting enzyme of this biosynthetic pathway. Cytoplasmic citrate is first cleaved into acetyl-CoA and oxaloacetate by ATP citrate lyase. Acetyl-CoA is then carboxylated into malonyl CoA by acetyl CoA carboxylase, the rate-limiting step in fatty acid synthesis. For cholesterol synthesis, two molecules of acetyl-CoA (also from Citrate by ATP citrate lyase) are condensed to yield acetoacetyl-CoA which is then condensed with another acetyl-CoA to yield 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA). The committed step in cholesterol synthesis is the reduction of cytosolic HMG-CoA to mevalonate. Fatty acid and cholesterol synthesis also requires NADPH and some of these are generated from cytosolic oxaloacetate via malate dehydrogenase and malic enzyme (Palmieri, 2004). Citrate transport from the mitochondrion increases NADPH synthesis and the activity of FIGURE 4 | Citrate provides substrates for anabolic pathways. Citrate is a substrate for many biological reactions. Here, the role of citrate in fatty acid and cholesterol synthesis is depicted. Citrate is transported from the mitochondrion into the cytoplasm by CiC/SLC25A1. Once in the cytoplasm, it is broken down into oxaloacetate and acetyl-CoA by the action of ATP Citrate Lyase. For fatty acid synthesis, acetyl-CoA is carboxylated by Acetyl-CoA Carboxylase (ACC) to malonyl-CoA. Alternatively, acetyl-CoA can be condensed with itself to generate acetoacetyl-CoA, which then reacts with another acetyl-CoA and water to form 3-hydroxy-3-methyl-glutaryl-CoA (HMG-CoA). Conversion of HMG-CoA to mevalonate, the committed step in cholesterol biosynthesis, is performed by 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase (HMGCR). human CiC is modulated by acetylation in activated immune cells (Palmieri et al., 2015). The orthologous yeast citrate transport protein Ctp1p, preferentially transports tricarboxylates versus other substrates, in contrast to the mammalian transporter which transports tricarboxylates, malate, and phosphoenolpyruvate to similar extents (Kaplan et al., 1995). By supplying citrate-derived acetyl-CoA that is required for normal nucleosome histone acetylation, SLC25A1 and its Drosophila ortholog, scheggia (sea) are also important for the maintenance of genome integrity (Morciano et al., 2009). Recently, it was shown that SLC25A1 is critical for tumor cell metabolism; it enables stem-like properties in cells, contributes to treatment resistance in cancer cells (Fernandez et al., 2018), and has been proposed as a metabolic oncogene (Kolukula et al., 2014). In another recent work, SLC25A1 and its genetic and functional interactions with SLC25A4 is required for optimal neuronal development, function and behavior (Gokhale et al., 2018). Therefore, CiC is critical for intermediary metabolism and has emerging roles in human development and disease including cancer.

Pathology of MCF
The MCF provides substrates for various biochemical processes in the cell. Consistent with their diverse and fundamental roles in metabolism, the absence or dysfunction of assorted MCF members causes a wide variety of disorders including hematologic, neurologic, and cardiac diseases. Underlying many of these disorders is a defect in OXPHOS leading to disturbed mitochondrial energy metabolism that manifests in a wide variety of clinical signs and symptoms. A number of systemic diseases caused by mutations in genes encoding SLC25 members are discussed briefly next and Table 3 displays a summary of MCF-linked diseases clustered by presentation.
Mitochondrial carriers encoded by SLC25A28, SLC25A37, and SLC25A38 are important for heme synthesis which requires cellular iron, glycine and succinyl-CoA (Figure 2). Since red blood cells are very sensitive to defects in heme synthesis, dysfunction in any of these mitochondrial carriers causes anemia (Xu et al., 2013). The erythroid specific SLC25A38, which based on its requirement for erythropoiesis was initially predicted to encode an amino acid carrier capable of transporting glycine, is now regarded as a bonafide glycine transporter (Guernsey et al., 2009;Fernández-Murray et al., 2016;Lunetti et al., 2016). Pathogenic mutations in SLC25A38 and knockdown experiments in zebrafish implicate the carrier in the etiology of congenital sideroblastic anemia (Guernsey et al., 2009) (OMIM 610819). Mutations in the carrier are the second most common cause of inherited sideroblastic anemia and may account for about a fifth of all cases (Harigae and Furuyama, 2010;Horvathova et al., 2010). As previously discussed, SLC25A28 and SLC25A37, which encode the mitoferrins, are critical for iron homeostasis. Although no human mutations have been described for these carriers, MFRN1 has been shown to be important for heme synthesis in erythroid cells (Shaw et al., 2006).
SLC25A46 encodes a novel outer mitochondrial membrane protein that is widely expressed in the nervous system (Haitina et al., 2006), and mutated in numerous neurological diseases including optic atrophy spectrum disorder,

System
Clinical presentation/disease and MCF member associated

Challenges to MCF-Focused Research
Mitochondrial carriers, and SLCs in general, perform a central role in metabolism and their association with a myriad of diseases makes them attractive candidates for basic and translational research. However, it has been noted that this area of research has not grown in commensurate proportion to its size or the gold mine its study could potentially reap (Cesar-Razquin et al., 2015). Most of these carriers are yet to be fully characterized and many of them remain totally uncharacterized. A number of technical factors have hampered growth of research focused on the extended membership of the SLC family that of course includes the MCF. A huge technical hurdle is the systemic absence of validated antibodies specific to most of these proteins. Further, many of the available antibodies are too weak to detect endogenous proteins whose expression is likely low and in general, many of the available reagents have not been rigorously characterized and validated (e.g., absence of signal with appropriate negative controls such as knockout cells). Compounding issues is the fact that many SLC members appear to have low immunogenicity which likely stems from the fact that they are polytopic membrane proteins that often display high interspecies conservation. Finally, transport assays are tedious to perform and limited by the volume of substrates available and/or required to de-orphanize a SLC protein.
Interestingly, a number of carriers have been shown to display substrate promiscuity by transporting more than one type of solute . Two members in Arabidopsis thaliana, AtUCP1 and AtUCP2, previously thought to be UCPs and therefore named as such were recently assigned the function of transporting amino acids, dicarboxylates, phosphate, sulfate, and thiosulfate (Monné et al., 2018). The Pi carrier in mammals, SLC25A3, was originally described as a phosphate symporter (Wohlrab and Flowers, 1982;Seifert et al., 2015) whose mutation is associated with fatal childhood diseases (Mayr et al., 2007;Bhoj et al., 2015). Pic2p, originally thought to be a second albeit minor Pi carrier in yeast, actually functions as a Cu ++ transporter responsible for the import of copper, required for COX assembly, into the mitochondrial matrix (Vest et al., 2013). Similarly, SLC25A3, which shares 65% similarity with Pic2p, was recently shown to have a conserved role for copper transport in vivo and in vitro (Vest et al., 2013;Boulet et al., 2017). Together, these results indicate that even well-established MCF members may have the capacity to transport additional presently unappreciated substrates.
Apparent functional redundancy is another recurring feature of mitochondrial carriers (Taylor, 2017). For example, mitochondrial nucleotide homeostasis not only involves the ADP/ATP carriers (SLC25A4, SLC25A5, SLC25A6, and SLC25A31) but also is influenced by the ATP-Mg/Pi carriers (SLC25A23, SLC25A24, SLC25A25, and SLC25A41). Indeed, in yeast, the calcium-dependent ATP-Mg/Pi carrier, Sal1p, can compensate for the absence of the major isoform of the ADP/ATP carrier, Aac2p, and provide a pool of mitochondrial ATP that is required for yeast viability (Chen, 2004;Cavero et al., 2005;Traba et al., 2008;Laco et al., 2010). Still, there are a lot of unknowns regarding how all of these different transporters maintain the nucleotide pool across the IMM. It has been suggested that different mammalian ADP/ATP carrier isoforms, which do not transport AMP, may have different preferred transport modes, i.e., ATP vs. ADP. For instance, based on its high expression in cancer cells, SLC25A5 (ANT2) was postulated to preferentially import ATP made by glycolysis, an activity that maintains the mitochondrial membrane potential and by extension, other essential mitochondrial functions (Stepien et al., 1992;Giraud et al., 1998;Chevrollier et al., 2005). However, it was recently demonstrated in a range of cancer cells that the uptake of ATP is in fact completely independent of the activity of any ADP/ATP carrier isoform (Maldonado et al., 2016). As such, it remains unresolved whether the different ANT isoforms do or do not have distinct transport activities and/or substrate preferences. Also unclear is how the calcium-dependent ATP-Mg/Pi carriers, which were originally thought to preferentially transport ATP but since demonstrated to also transport other adenine nucleotides (Fiermonte et al., 2004), interface with the ADP/ATP carriers to modulate mitochondrial nucleotide homeostasis. The requirement for calcium provides an opportunity for the flux of adenine nucleotides across the IMM to be regulated. Indeed, the calcium-dependent mitochondrial uptake of adenine nucleotides by SLC25A23 is required for the glucagon-stimulated increase in OXPHOS in liver mitochondria (Amigo et al., 2013). As research progresses in this area, identification of the substrate(s) transported by many of these proteins will provide foundational information as to their physiological roles.

MITOCHONDRIAL TRANSLATION
The vast majority of mitochondrial proteins are translated in the cytosol and imported thereafter into the mitochondrion. For those proteins encoded by the mitochondrial genome (mtDNA), mitochondria have retained a dedicated, dual-origin translational machinery whose architecture is similar to that of bacteria. Consistent with this bacterial origin, mitochondrial translation is pharmacologically unaffected by cycloheximide, an inhibitor of cytosolic translation, and is instead sensitive to antibiotics such as puromycin and chloramphenicol. Eight polypeptides in yeast and thirteen polypeptides in humans are encoded by mtDNA and thus produced via mitochondrial translation. Apart from genes for these polypeptides, the mtDNA also encodes a set of transfer (t)RNAs, ribosomal (r) RNAs and in yeast, the RNA component of the mitochondrial RNAse P (Towpik, 2005;Walker and Engelke, 2008). The mitochondrial translation cycle is subdivided into four steps-initiation, elongation, termination and recycling-and nuclear encoded polypeptide factors are required at different steps for optimal mitochondrial translation (Towpik, 2005;Smits et al., 2010;Kehrein et al., 2013). An example of a nuclear-encoded factor in yeast is the mitochondrial translation initiation factor 3 (mIF3p), the Saccharomyces cerevisiae homolog of the bacterial translation initiation factor 3 (IF3). Its function is conserved and overlaps with human mIF3p (Atkinson et al., 2012;Kuzmenko et al., 2014), and its absence in yeast disrupts mitochondrial translation (Kuzmenko et al., 2016). In addition, a number of translational activators directly interacting with mRNAs of mitochondrial encoded polypeptides are necessary to optimize the translation of a specific mtDNA-encoded polypeptide, e.g., synthesis of Cox1p is affected when Mss51p is absent or limiting (Siep et al., 2000;Perez-Martinez et al., 2003Barrientos et al., 2004;Fontanesi et al., 2011).
Translation of some mitochondrial proteins is tightly coupled to their assembly into respiratory complexes in a manner similar to a mechanism described as "controlled by epistasy of synthesis" (CES) that exists in the biogenesis of photosynthetic protein complexes (Choquet et al., 2001;Towpik, 2005). For example, in yeast, Cox1p synthesis is tightly coupled to the assembly of respiratory complex IV which helps to balance the production of subunits with their assembly into the holoenzyme which in the net preserves mitochondrial proteostasis (Perez-Martinez et al., 2003;Barrientos et al., 2004;Towpik, 2005;Soto et al., 2012).

EMERGING ROLES IN THE BIOGENESIS OF CYTOCHROME C OXIDASE
In yeast and in humans, a link between the function of MCF members and COX biogenesis is emerging (Figure 5). For instance, a destabilizing pathogenic mutation in SLC25A46 impairs OXPHOS and ascorbate/TMPD-dependent respiration and reduces steady state levels of complex IV subunits (Janer et al., 2016). These findings are consistent with a complex IVspecific assembly defect in SLC25A46 mutant fibroblasts. At present, the mechanistic basis for the reduced steady state levels of complex IV subunits has not been determined. A role in the assembly of the OXPHOS machinery is perhaps not unexpected given that many SLC members provide substrates that serve as building blocks needed for processes such as mitochondrial DNA replication, transcription, translation, and/or post-translational assembly of protein complexes. However, it is surprising that these defects seem to specifically impact complex IV without significantly affecting the other OXPHOS complexes that are also built from subunits expressed from both the nuclear and mitochondrial genomes.
Recently, a role for the yeast ortholog of human SLC25A38, Hem25p, important for heme synthesis as a mitochondrial glycine importer, in the stability of respiratory complex proteins was tested (Dufay et al., 2017). While deletion of Hem25p compromises the steady state level of subunits of each respiratory complex, its absence is most detrimental to complex IV (Dufay et al., 2017). Intriguingly, the combined absence of Hem25p and Flx1p, the mitochondrial flavin adenine dinucleotide transporter (ortholog of human SLC25A32), further reduces the steady state level of subunits of the hem25 -affected respiratory complexes except complex IV (Dufay et al., 2017). These results are consistent with a model that Hem25p and Flx1p provide heme FIGURE 5 | An emerging link between mitochondrial carriers and complex IV biogenesis. Cytochrome c oxidase is assembled from protein subunits that originate from the mitochondrial and nuclear genomes. Mitochondrial carriers support these processes at multiple levels. Mutations in the metazoan-specific SLC25A46, an unusual family member that resides in the outer membrane, decreases complex IV levels by an unclear mechanism. The absence of Aac2p function impairs translation of subunits encoded by mitochondrial DNA. Pic2p and Hem25p help provide co-factors that are essential for complex IV assembly and function. Hem25p transports glycine that is used to make Heme and Pic2p mediates uptake of Cu ++ . TOM, translocase of the outer membrane; TIM23, translocase of the inner membrane 23; Oxa1p, mitochondrial inner membrane insertase. and FAD, respectively, which are required for the assembly of the respiratory chain complexes.
Knockdown of the two isoforms of SLC25A3, SLC25A3-A, and SLC25A3-B, in many different cell types results in reduced COX holoenzyme levels and activity (Boulet et al., 2017). When SLC25A3 is limiting, the steady state levels of COX4, a nuclear encoded subunit of COX, and mitochondrial copper are reduced (Boulet et al., 2017). Interestingly, COX4 abundance is rescued by copper supplementation (Boulet et al., 2017). Since copper is critical for the assembly of COX (Diaz, 2010;Baile and Claypool, 2013), defects in copper import are likely to impair COX biogenesis. Nevertheless, SLC25A3 function may be modulated and/or linked to the assembly of COX via a pathway that is presently unidentified (Boulet et al., 2017).
The absence of Aac2p in yeast leads to a specific reduction of COX activity while the activity of complex III is unaffected (Dienhart and Stuart, 2008). The reduced complex IV activity likely stems from lower steady state levels of its subunits in the absence of Aac2p. These findings are consistent with the impaired OXPHOS activity that occurs when Aac2p expression is decreased or ablated (Heidkämper et al., 1996;Muller et al., 1996;Müller et al., 1997;Claypool et al., 2008). Intuitively, the mechanistic basis for the reduced steady state levels of complex IV subunits could derive from a defect in any step in its biogenesis that is regulated/modulated by and/or dependent on the nucleotide transport function of Aac2p. Alternatively, the conserved AAC/ANT-RSC interaction may itself be critical for robust COX expression, assembly, and/or function Since the Aac2p and ANT1/ANT2 interactomes all included other SLC25 family members (Claypool et al., 2008;Lu et al., 2017), it is possible that these MCF-MCF interactions are critical for maintaining the abundance of metabolites that are needed for optimal mitochondrial translation and assembly of respiratory complexes.
Our group recently showed that in yeast regulation of complex IV activity by the Aac2p relies solely on the activity of the protein in a way that is mechanistically dependent on mitochondrial translation (Ogunbona et al., 2018). Indeed, similar to the complete absence of Aac2p (Dienhart and Stuart, 2008), an interaction-competent transport-dead mutant of Aac2p exhibits reduced complex IV activity, reduced levels of the complex IV holoenzyme, and reduced steady states levels of complex IV subunits, especially those encoded by mtDNA (Ogunbona et al., 2018). Interestingly, translation of complex IV subunits encoded by mtDNA is specifically decreased in the absence of Aac2p activity, either genetically imposed or due to the presence of AAC-specific inhibitors. In addition, when Aac2p is expressed but non-functional, the turnover of newly synthesized Cox3p is increased (Ogunbona et al., 2018). Together, these results establish a novel link between nucleotide transport and mitochondrial translation of complex IV subunits. A big question moving forward is whether this neoteric mechanism is evolutionarily conserved in mammals. Ongoing and future efforts focused on dissecting the functional significance of distinct aspects of the AAC/ANT interactome are likely to shed further insight into how it supports complex IV biogenesis.

PERSPECTIVES
The MCF is an extensive cluster of transport proteins with essential biochemical and physiological functions. Individually and collectively, they have a broad role in intermediary metabolism and are pathophysiologically significant. In spite of this, there remain numerous fundamental gaps in our understanding of their cellular functions and if and how their activities are regulated. At the molecular level, the mechanistic basis for their role(s) in the regulation of many biologic processes, including OXPHOS, is still missing. Even for the best characterized MCF protein, Aac2p, new biology was recently uncovered suggesting that there is likely a ton of new biology to uncover in the biggest subfamily of SLCs. Future research should primarily focus on establishing the biological functions of each and every protein in the family. For proteins with assumed roles in mitochondrial energy production, it will be important to carefully dissect their molecular contributions to OXPHOS. Using the recently uncovered translational regulation of complex IV activity by Aac2p as a paradigm, upcoming research work should remain open-minded to the identification of unanticipated functional relationships that may exist between MCF members and other biological processes. Discovery of new members and novel functions of existing members will give us a better understanding of their pathophysiological roles. Long term, such information may guide development of effective therapeutic strategies to correct or better manage diseases resulting from their dysfunction. With recent advances in biomedical research such as the advent of gene editing technologies (e.g., CRISPR/Cas9), substantial and rapid progress in the field of MCFs is anticipated.

AUTHOR CONTRIBUTIONS
OO and SC reviewed the literature. OO wrote the initial draft of the review which also served as the Introduction of his Ph.D. thesis (Ogunbona, 2018), submitted to Johns Hopkins University School of Medicine, and generated all of the figures. SC provided guidance and edited the manuscript.

FUNDING
This work was supported by the National Institutes of Health grant (R01HL108882) to SC and a pre-doctoral fellowship from the American Heart Association (15PRE24480066) to OO.