Can Oxaloacetate cross the outer mitochondrial membrane?

I am aware of the Malate-Aspartate Shuttle, but something is not clear to me and different sources seem to contradict each other. Some show oxaloacetate (OAA) being reduced to malate in the mitochondrial inter-membrane space (IMS), whereas others show the reduction happening in the cytosol.

Where does the OAA → malate reduction happen? i.e., can OAA cross the outer mitochondrial membrane (from the cytosol into the IMS) so that it can be reduced in the IMS, or must it be reduced in the cytosol before crossing either of the two mitochondrial membranes?

In general the outer mitochondrial membrane is thought to be basically permeable (through porins) to small molecules such as OAA. As is typical in biology, the situation may actually be more complex -- see for example this paper. But I think the default assumption is that metabolites freely cross the mitochondrial outer membrane.

You might also ask whether either the mitochondrial MDH or the cytoplasmic MDH enzyme is likely to be found or be active inside the intermembrane space.

Oxaloacetate (OAA) cannot cross the inner mitochondrial membrane.

The process of oxidative phosphorylation and the electron transport system (ETS) occur in the mitochondrion, whereas $ce{NADH}$ generated by the reduction of $ce{NAD+}$ in glycolysis is in the cytoplasm. The problem is that the inner mitochondrial membrane is not permeable to $ce{NADH}$ , so a shuttle system is required for the transport of the reducing equivalents through the mitochondrial membrane. There are two of these, one of which is the 'Malate-Aspertate' shuttle.

In the process (see above) oxaloacetate (OAA) takes up the reducing equivalents from $ce{NADH}$ to form malate in a reaction catalysed by malate dehydrogenase. The inner mitochondrial membrane is permeable to malate, which passes through carrier proteins (Malate-$alpha$-ketogluterate transporter) into the mitochondrial matrix where it is converted back to OAA. As this happens the concentration of OAA decreases in the inter-membrane space, and as OAA cannot pass directly through the inner mitochondrial membrane it is converted into aspartate in the mitochondrial matrix by reacting with glutamate to produce $alpha$-ketoglutarate and aspartate. The aspartate then travels to the inter-membrane space through specific carriers(Glutamate-aspartate Transporter). In the inter membrane space the aspartate combines with $alpha$-ketoglutarate to form glutamate and OAA (the reverse of what happened in the mitochondrial matrix). Thus the concentration of OAA is maintained in the inter-membrane space, and the reaction continues.


OAA is converted into malate in the inter-membrane space. In the mitochondrial matrix malate is converted back to OAA, as illustrated in the illustration below.

Image source : Malate-Aspartate Shuttle, Wikipedia

Mitochondria—hubs for regulating cellular biochemistry: emerging concepts and networks

Mitochondria are iconic structures in biochemistry and cell biology, traditionally referred to as the powerhouse of the cell due to a central role in energy production. However, modern-day mitochondria are recognized as key players in eukaryotic cell biology and are known to regulate crucial cellular processes, including calcium signalling, cell metabolism and cell death, to name a few. In this review, we will discuss foundational knowledge in mitochondrial biology and provide snapshots of recent advances that showcase how mitochondrial function regulates other cellular responses.

1. Introduction

All modern-day eukaryotes are believed to have arisen from a primordial ancestor that engulfed an α-protobacterium with the capacity for respiration [1]. This event gave rise to modern-day mitochondria, an event that is now deeply integrated in eukaryotic cell homeostasis and survival. Mitochondria are dynamic networks capable of remodelling their morphology and activity. They provide energy and biomolecules for the cell, in addition to contributing to pathways of cell stress, immune responses, intra- and intercellular signalling, cell-cycle control and cell death. The unique biology of mitochondria underpins their influence on the cell and the ability to calibrate their structure and proteome is an efficacious means of adapting their function. As such, we will begin with a brief outline of three fundamental concepts in mitochondrial biology: (i) mitochondrial ultrastructure (ii) mitochondrial protein import and (iii) mitochondrial dynamics. This will inform subsequent discussion of mitochondria as key players in broad and diverse roles, including metabolism, signal transduction, immunity, cell cycle, cell differentiation, cell death and stress.

2. Mitochondrial ultrastructure, dynamics and protein import

2.1. Mitochondrial ultrastructure

Mitochondria have a double membrane that defines four compartments: the outer membrane, the intermembrane space, the inner membrane and the matrix. The architecture of the inner membrane is malleable and typically convoluted into folded invaginations, called cristae, that dictate the spatial arrangement of proteins [2]. Remodelling cristae structure of cristae can also alter enzymatic flux between the compartments, consistent with the diverse cristae structures observed across cell types with different metabolic demands [2]. The recently described MICOS complex (mitochondrial contact site and cristae organizing system) is required to maintain cristae morphology [3] (figure 1). Loss of MICOS assembly ablates cristae junctions and manifests severe defects in energy metabolism, calcium handling and lipid trafficking [4]. However, it remains unclear how MICOS is regulated by cellular conditions to produce diverse cristae morphologies. Interestingly, disruption of MICOS alters the activity and/or abundance of mitochondrial morphology proteins [5,6]. Perturbations to organelle function have long been associated with gross morphological changes in the mitochondrial network, therefore cristae reorganization by MICOS assembly/disassembly may be an intermediary between function and dynamics. Recently identified associations between MICOS and protein import complexes point to the broad influence of MICOS on mitochondrial function [7,8].

Figure 1. Nuclear-encoded mitochondrial proteins are imported by multi-subunit translocases. Mitochondrial proteins synthesized in the cytosol are imported into mitochondria post-translationally. The TOM complex at the outer membrane serves as a general protein entry gate. hTom40 forms the pore of the translocase, while hTom20, hTom22 and hTom70 function as receptors. hTom22 plays an additional role in the assembly of the complex. hTom5, hTom6 and hTom7, collectively called the small TOMs, regulate the dynamics and assembly of the complex. The TIM22 complex at the inner membrane mediates the import of multi-pass transmembrane proteins into the inner membrane. hTim22 forms the pore through which proteins are inserted, while AGK and hTim29 function as receptors and in complex assembly. The TIM23 complex can translocate precursor proteins into the matrix or the inner membrane. hTim23 and hTim17 form the channel pore, and hTim50 functions as a receptor for precursors. The core complex associates with an import motor that helps to translocate proteins into the matrix in an ATP-dependent manner. The MIA complex mediates the import of soluble intermembrane space proteins by catalysing the formation of disulfide bonds. hMia40 carries out the disulfide bond formation and is anchored to the inner membrane through an interaction with AIF. ALR removes electrons from hMia40 so that it can undergo further rounds of catalysis. The SAM complex of the outer membrane mediates insertion of β-barrel proteins into the outer membrane. hSam50 associates with MTX1 and MTX2. Cristae, the large invaginations of the inner mitochondrial membrane, are stabilized by a multi-subunit complex called MICOS. Mic60 is the core subunit of MICOS, which additionally contains Mic10, Mic13, Mic14, Mic19, Mic25, Mic26 and Mic27. MICOS also associates with the SAM complex at the outer membrane to form a structure known as the mitochondrial intermembrane space bridging complex (MIB).

(Recommended further reading on cristae, MICOS and ultrastructure: [2,9,10].)

2.2. Mitochondrial protein import

From their endosymbiont origins, human mitochondria have retained only 37 genes in a small circular genome known as the mitochondrial DNA (mtDNA), which encodes 13 polypeptides, 22 tRNAs and 2 rRNAs. The remaining 1000–1500 mitochondrial proteins are nuclear encoded and must be imported and sorted to the relevant mitochondrial compartment following synthesis in the cytosol. Fundamentally, mitochondrial protein import is mediated by multimeric protein complexes known as translocases, which are located at mitochondria (figure 1). Briefly, two major translocases reside in the outer membrane of mitochondria: the Translocase of the Outer Membrane (TOM) complex and the Sorting and Assembly Machinery (SAM). The TOM complex is the initial point of contact for almost all mitochondrial precursors and provides a means of entry into the organelle. Following translocation through TOM, precursor import pathways diverge based on their targeting information and ultimate location within the organelle. β-barrel proteins of the outer membrane are sorted to the SAM complex for integration into the membrane. There are two translocases embedded in the inner membrane of mitochondria: the Translocase of the Inner Membrane (TIM) 22 and 23 (TIM22 and TIM23) complexes. TIM22 mediates the insertion of non-cleavable polytopic membrane proteins into the inner membrane, while the TIM23 complex is responsible for importing precursors across the inner membrane into the matrix or in some instances can laterally release transmembrane precursors into the inner membrane. Finally, the Mitochondrial Intermembrane space Assembly (MIA) machinery mediates the import of small cysteine-rich intermembrane space proteins and couples their import to their oxidation [11]. These import pathways and machines have been predominately characterized in fungal organisms however, in more recent years, analysis in higher eukaryotes has uncovered important physiological consequences due to perturbations in protein import. Specifically, mutations in genes encoding protein import subunits cause distinct mitochondrial diseases with phenotypes ranging from severe muscular defects to neurodegeneration and congenital growth defects [12].

(Recommended further reading on mitochondrial protein import: [13–15].)

2.3. Mitochondrial dynamics: fission, fusion and organelle contact sites

As an organellar network, mitochondria undergo fission and fusion to replicate, be recycled, and alter their bioenergetics. Fusion of the outer membrane is mediated by homotypic interactions between the GTPases Mfn1 and Mfn2 on adjacent mitochondria (figure 2) [16], but the domains involved and stepwise mechanism of fusion are still debated. Fusion of the inner membrane is controlled by Opa1, which exists as five isoforms generated by mRNA splicing and proteolytic cleavage (figure 2) [17]. It is believed that the stoichiometry of these isoforms governs Opa1 interactions with the mitochondrial-specific lipid cardiolipin and, subsequently, fusion events [18]. Mitochondrial fusion is associated with increased ATP production by oxidative phosphorylation and protects against oxidative and proteostatic stress [19]. Conversely, mitochondrial fission is concomitant with a reliance on glycolysis and precedes mitochondrial turnover. Fission is largely dependent on the dynamin-related and cytosolic protein Drp1, which oligomerizes around and constricts mitochondrial tubules (figure 2). The recruitment of Drp1 from the cytosol requires adaptor proteins on the mitochondrial outer membrane, including Mff, Mid49 and Mid51 [20,21], although human Fis1 can promote Drp1-independent mitochondrial fragmentation through inhibition of fusion proteins [22] (figure 2). While conflicting models of Drp1 recruitment have been proposed, its localization and activity are known to be regulated by numerous post-translational modifications [23]. The scission ability of Drp1 oligomers is sterically limited to tubules up to 250 nm diameter, indicating pre-constriction is required for larger mitochondria [24]. This is achieved by the endoplasmic reticulum (ER), which wraps around and constrict tubules to mark future fission sites and aid correct partitioning of mitochondrial contents [25,26].

Figure 2. Cellular machineries mediating mitochondrial fission, fusion and formation of contact sites with the endoplasmic reticulum. Mitochondria continuously undergo fission and fusion. Fission is mediated by the GTPase Drp1, which can be recruited to the outer mitochondrial membrane by a variety of receptors, including Mff, Fis1, Mid49 and Mid51. Drp1 at the outer membrane can oligomerize into fibrils that constrict mitochondria to initiate fission. Mitochondrial fusion is initiated by tethering of mitochondria through homotypic interactions between Mfn1 and Mfn2 on opposing mitochondria. Inner membrane fusion is mediated by OPA1, which exists as long and short forms generated through proteolysis. Contact sites between the mitochondria and the endoplasmic reticulum (ER) are established and maintained through protein–protein interactions. Interactions occur between Mfn2 molecules on the ER membrane and the outer mitochondrial membrane, and between VAPB on the ER membrane and RMDN3 on the mitochondrial outer membrane. Interactions also occur between IP3R3, a calcium channel on the ER membrane, and VDAC1 and hTom70 on the mitochondrial outer membrane.

Mitochondria also engage in extensive dynamic inter-organelle contacts that coordinate functional exchanges between mitochondria and other cellular components [27]. In particular, ER–mitochondria contact sites (ERMCs) facilitate a multitude of functions including mitochondrial fission, coenzyme Q biosynthesis, lipid transfer, Ca 2+ transfer, mtDNA replication and autophagy [25,26,28–31]. The ER–mitochondria encounter structure (ERMES) has been well characterized in Saccharomyces cerevisiae [32], however no human equivalent has been identified [33]. Preliminary work in humans suggests that metazoan ERMCs are tethered by interactions between hTom70 and IP3R3, VDAC1 and IP3R3, RMDN3 and VAPB, Mfn2 homodimers, Vps13a and Pdzd8 with an unknown partner (figure 2) [34–39]. Furthermore, acetylated microtubule ‘tracks’ have been proposed to maintain these contacts despite the movements and remodelling of the two organellar networks [40]. Other inter-organelle contacts have been described between mitochondria and Golgi [27,41], peroxisomes [42], lysosomes [43], lipid droplets [44] and the plasma membrane [45]. The interconnectivity of mitochondria with cellular components enables significant interplay across various pathways, examples of which we will highlight throughout this review (table 1).

Table 1. Full names and identifiers of proteins discussed in this review.

(Recommended further reading on mitochondrial dynamics: [46–48] on organelle contacts: [49–51].)

3. Mitochondria and metabolism

Mitochondria are well known for providing energy to the cell, predominantly by coupling the tricarboxylic acid (TCA) cycle with oxidative phosphorylation. The TCA cycle is a series of eight enzymatic reactions that occur in the matrix to harvest electrons from citrate and its catabolic intermediates (figure 3a). The typical input to the cycle is acetyl-CoA, which can be produced from glucose (via glycolysis), fatty acids (via β-oxidation) and amino acids (via deamination) (figure 3a). The electrons scavenged throughout the cycle are transferred by NADH and FADH2 to the complexes of the electron transport chain. Complexes I–IV of the electron transport chain shuttle electrons, using their energy to pump protons into the intermembrane space and establish an electrochemical gradient across the inner membrane. Complex V (ATP synthase) releases the protons back into the matrix, using the energy of the electrochemical gradient to produce ATP, the cell's energy currency (figure 3a) [52]. Although normally efficient, oxidative phosphorylation is negatively regulated by the accumulation of its toxic by-product, reactive oxygen species (ROS). If unchecked, ROS can cause damage to mitochondria, induce protein aggregation and introduce mutations in DNA [53–55]. Recent advances in cryoelectron microscopy have revealed Complexes I, III and IV can assemble to form supercomplexes thought to reduce the amount of ROS produced during electron transport, as well as enhance respiration rates [56].

Figure 3. Mitochondria coordinate essential metabolic processes. (a) Mitochondria are best known for housing the protein machinery required for generating ATP. When oxygen is available, most cells will generate ATP through oxidative phosphorylation, where electrons harvested through catabolic reactions are used to power ATP synthase. Electrons are obtained through the TCA cycle, which occurs in the matrix and consists of eight enzymatic reactions. Acetyl-CoA is the primary input for the TCA cycle, and can be obtained through metabolism of glucose, fatty acids and amino acids. Electrons extracted during the TCA cycle are loaded onto NAD + and FAD 2+ . Electrons are subsequently transferred from NADH and FADH2 onto Complexes I and II of the electron transport chain. Electrons are passed through Complexes III and IV, which transport protons into the intermembrane space. Protons are allowed to flow back into the matrix through ATP synthase (Complex V), which uses the energy of the proton gradient to convert ADP to ATP. (b) Mitochondrial one-carbon (1C) metabolism comprises a series of parallel and reversible reactions which occur in the cytosol and mitochondrial matrix. In proliferating cells, the reaction normally proceeds in a specific direction such that formate produced within mitochondria can be used for biosynthetic processes in the cytosol. Within the mitochondria, THF and serine imported from the cytosol are acted upon sequentially by SHMT2, MTHFD2 and MTHFD1 L to produce formate, which is exported back into the cytosol. Cytosolic MTHFD1 loads formate onto THF to form charged folate intermediates that can be used to synthesize purine and pyrimidine nucleotides. Mitochondrial 1C metabolism is also an important source of glycine. (c) The mitochondrial matrix functions as an important storage site for calcium ions. Mitochondrial calcium uptake often occurs at ER contact sites, where large volumes of Ca 2+ can be released through IP3R3. Calcium can pass freely through the outer membrane via VDAC channels and is transported across the intermembrane space and inner membrane through the coordinated function of a MICU1/MICU2 dimer docking to MCU in the inner membrane. Calcium can exit the mitochondrial matrix through LETM1 or SLC8B1 (in exchange for H + or Na + , respectively) and can cross the outer membrane through VDACs or NCX3.

Mitochondria also produce fatty acids, amino acids, nucleotides and haem groups for the cell through biosynthetic pathways [57–59]. One such process, one-carbon (1C) metabolism, produces glycine, methionine, nucleotides, phosphatidylcholine and 1C units (methyl-like groups) from serine catabolism through the redox chemistry of folate and its derivatives (figure 3b) [60]. These 1C units charge the universal methyl donor S-adenosylmethionine required for the methylation of proteins and chromatin [61]. There is now significant evidence of metabolic enzymes and metabolites altering gene expression as reporters of environmental conditions (nutrient availability, hypoxia, oxidative stress) or mitochondrial dysfunction. This has been shown for acetyl-CoA, TCA intermediates, ketones, lactate, fatty acids and amino acids [62–68]. Emerging studies also indicate cellular nutrient and energy sensing by mTOR kinase regulates mitochondrial biogenesis and protein synthesis [69]. Through downstream effectors of transcription and translation, mTORC1 stimulates mitochondrial biogenesis and oxidative metabolism to meet the energy demand of anabolism [70–72]. Interestingly, the tumour suppressor p53 inhibits mTOR-mediated growth and proliferation to prevent oncogenesis [73,74]. p53 activity increases electron transport chain efficacy [75], mtDNA stability [76,77] and reduced glutathione (GSH) levels [78] to limit ROS production as well as inhibiting glycolysis [79,80], which contributes to the replicative potential of tumour cells [79,81,82]. Thus, metabolism is intimately integrated with other cellular pathways, but is not the sole contribution of mitochondria to signalling mechanisms.

(Recommended further reading on metabolism: [60,83] on metabolite signalling: [68,84] on mTOR/p53: [85,86].)

4. Signalling

4.1. Mitochondria control calcium homeostasis

Calcium ions are common to diverse signalling pathways. The outer mitochondrial membrane is permeable to Ca 2+ , in part due to channel-forming VDAC proteins [87] and export via SLC8A3 [88]. The mitochondrial inner membrane calcium uniporter (MCU) complex regulates transport into the matrix (figure 3c). Permeability of the MCU complex is calibrated by two regulatory subunits, MICU1 and MICU2, that are linked by an intermolecular disulfide bond introduced by hMia40 [89,90]. The ability of mitochondria to accumulate Ca 2+ up to 20-fold higher concentrations than the cytosol allows them to function as buffering systems and re-establish homeostasis following Ca 2+ bursts [91,92]. Bursts of Ca 2+ into the cytosol, from across the plasma membrane or intracellular stores, can initiate neurotransmitter release, muscle fibre contraction and transcriptional regulation. In neurons, mitochondrial Ca 2+ buffering modulates both the propensity and duration of neurotransmitter release [93,94]. In cardiac muscle, contraction is coupled to enhanced mitochondrial ATP production via Ca 2+ -increased activities of TCA cycle enzymes, Complex V and the ADP/ATP transporter [95–98] an effect maximized by local Ca 2+ concentrations at ERMCs [29,99] (figure 3c). Additionally, mitochondrial Ca 2+ regulation influences hormone secretion [100], tissue regeneration [101] and interferon-β signalling via the mitochondrial antiviral signalling protein, MAVS [102].

(Recommended further reading on mitochondrial Ca 2+ signalling: [92,103,104].)

4.2. Roles of mitochondria in immune responses

The contribution of mitochondria to immune responses is a growing area of research. Cell-autonomous immune signalling is driven by MAVS at the outer membrane, which acts as a relay point for immune signal transduction. Rig-like receptors in the cytosol undergo conformational changes upon detecting viral RNA or DNA and are recruited to MAVS, particularly at ERMCs [105]. MAVS then dimerizes to enable the binding of multiple downstream signalling adaptors including TRADD, TRAF3 and STING to activate NF-κB and IRF-3/7 transcription of interleukins and pro-inflammatory cytokines [106–108] (figure 4a). Interestingly, MAVS dimers and many of its adaptors co-immunoprecipitate with hTom70 of the TOM complex, the overexpression of which increases the signalling response [109]. MAVS signalling is also affected by ROS and negatively regulated by Nlrx1, a binding partner of Complex III and MAVS [110,111] (figure 4a). As mitochondrial protein import and oxidative metabolism can be hijacked by virulence factors [112], these interactions may make MAVS sensitive to consequences of infection. Finally, if mitochondria are compromised by infection, the increased ROS and release of mtDNA into the cytosol can activate the NLRP3 inflammasome to evoke an inflammatory response [113,114] (figure 4a).

Figure 4. Mitochondria make crucial contributions to diverse cellular processes. (a) The mitochondrial outer membrane is the site of important signalling events during the innate immune response. Detection of viral nucleic acids by Rig-like receptors (RLRs) induces dimerization of MAVS, a protein of the mitochondrial outer membrane. Dimerized MAVS recruits signalling adaptors that initiate downstream activation of IRF3/7 and NF-κB, transcription factors that induce expression of type I interferons and pro-inflammatory cytokines. MAVS is regulated by NLRX1, a protein which downregulates MAVS when localized to the outer membrane, but activates MAVS when at the inner membrane by interacting with Complex III to induce ROS production. Release of mtDNA during infection can also activate the NLRP3 inflammasome. (b) Mitophagy is a process that allows damaged mitochondria to be identified and destroyed. Under normal conditions, PINK1 is imported into mitochondria and degraded by PARL. When mitochondria are damaged, import is impaired and PINK1 accumulates in the TOM complex at the outer membrane. Autophosphorylated and active PINK1 at the outer membrane phosphorylates monoubiquitin molecules on outer membrane proteins, recruiting and activating the E3 ubiquitin ligase Parkin. Activated Parkin synthesizes polyubiquitin chains that recruit autophagy receptors to initiate mitophagy. (c) Mitochondrial proteostatic stress is sensed through the partitioning of the transcription factor ATF5 between the mitochondria and the nucleus. Under normal conditions, ATF5 is imported into and sequestered within mitochondria. If mitochondrial protein import becomes compromised, ATF5 is trafficked into the nucleus, where it upregulates expression of genes that enhance proteostasis. (d) Mitochondria play crucial roles in the initiation of apoptosis. In response to pro-apoptotic stimuli, Bax and Bak oligomerize in the outer membrane to form pores that allow for efflux of apoptogenic proteins (Cytochrome c, Diablo, AIF and Endonuclease G) from the intermembrane space into the cytosol. Cytochrome c binds to Apaf-1 to induce formation of the apoptosome and activation of caspases. Diablo blocks inhibitors of apoptosis (IAPs) which would otherwise mitigate the effect of caspases. AIF and Endonuclease G translocate into the nucleus where they contribute to destruction of the genome.

Mitochondrial metabolism also directs rapid changes to specialized immune cells during infection. Changes in membrane potential can activate or supress M2 macrophages [115,116] and M1 macrophages shunt intermediates from the TCA cycle to generate nitrous oxide, IL-1β and the antibacterial itaconic acid [117,118]. Furthermore, the phagocytic abilities of macrophages depend on mitochondrial ROS production to destroy internalized pathogens [119]. Naive T-cells display increases in mitochondrial mass, mtDNA copy number, glycolysis, and glutamine metabolism during differentiation for rapid proliferation and to escape quiescence [120,121]. Metabolic remodelling then also decides the T-cells' mature fate [122,123], by altering cristae architecture [124] or by direct effect of metabolites on epigenetic transcription regulation [125].

(Recommended further reading on mitochondrial immune signalling: [106,118,126].)

5. Cell cycle, differentiation and death

Mitochondria are implicitly tied to cell-cycle control as providers of energy and nucleotides however, they also coordinate checkpoints and respond to signals of proliferation. To meet the metabolic demand of mitosis, mitochondrial mass and membrane potential increase from G1/S until late mitotic stage [127]. Indeed, hyperpolarization and increased ATP production inhibit AMP kinase to allow cyclinE-mediated entry to S-phase [128]. In the late G2 stage of dividing S. cerevisiae, the cyclinB1/Cdk1 complex traffics to mitochondria to phosphorylate Complex I subunits and Tom6, stimulating oxidative metabolism both directly and indirectly via increased protein import [129,130]. During mitosis, a highly fused and reticular mitochondrial network progressively fragments to small tubular organelles that segregate in anticipation of cytokinesis [127,131]. Mitochondria can also delay cell-cycle progression to increase their biogenesis [132], because of insufficient nucleotide production [133], or because of ROS accumulation [134]. Moreover, the fusion mediator Mfn2 can sequester both Ras and Raf to inhibit proliferative signalling [135].

Stem cell differentiation also relies on mitochondria as a ‘metabolic switch’. Human embryonic stem cells are glycolytic however, they develop mature cristae, rapidly replicate mtDNA and increase ATP production upon differentiation [136]. In haematopoietic stem cell differentiation, the downregulation of Pdk2, an inhibitor of pyruvate dehydrogenase, releases suppression of acetyl-coA production and enables oxidative phosphorylation [137]. The subsequent increase in ROS production and oxidative phosphorylation during differentiation drives upregulation of mitochondrial antioxidant proteins by the transcription factors Oct4, Sox2 and Nanog [138]. Mitochondrial fusion is believed to facilitate these metabolic changes, although the importance of specific proteins and fission/fusion balance may be cell-type specific [139–141]. This is supported by somatic cell reprogramming studies showing deletion of Mfn2 permits pluripotency as glycolysis becomes predominant over oxidative phosphorylation [142] the same effect being achieved by the pluripotency factor ZFP42 activation of Drp1 [143].

If cellular conditions or external insults are too harsh, mitochondria can trigger multiple forms of cell death. Apoptosis, or programmed cell death, can be elicited from extrinsic signalling via the Fas, TRAIL and TNFα receptors or intrinsic insults such as DNA damage, Ca 2+ overload, ROS and ER stress [144]. Mitochondria contribute to the extrinsic pathway but are the nexus of the intrinsic apoptotic pathway. In the latter pathway, cytosolic pro-apoptotic Bax oligomerizes with Bak at the outer membrane to permeabilize mitochondria and release pro-apoptotic proteins, including cytochrome c, Diablo, Htra2, Endonuclease G and AIF (figure 4d) [145]. In the cytosol, cytochrome c nucleates the formation of the apoptosome and activation of the caspases that dismantle the cell in an immunologically silent manner. Cytosolic Diablo and Htra2 block inhibitors of caspase activation, which would otherwise protect the cell from basal cytochrome c leakage [146,147]. Endonuclease G and AIF translocate to the nucleus to fragment DNA (figure 4d), AIF first requiring proteolytic cleavage of its transmembrane domain [148–150]. AIF is normally part of the intermembrane space import machinery, or MIA complex, anchoring the oxidoreductase hMia40 to the inner membrane. The outer membrane protein VDAC2 protects against apoptosis by sequestering Bak [151,152], yet new evidence suggests it may be required for Bax-mediated apoptosis [153]. Emerging research also implicates mitochondria in alternate and less-studied cell-death pathways such as ROS-induced necrosis [154], immune-activated necroptosis [155], ferroptosis [156,157] and parthanotosis [158].

(Recommended further reading on mitochondria in the cell cycle: [159,160] on differentiation [161–163] on cell death: [164,165].)

6. Mitochondrial quality control

The loss of mitochondrial function has profound negative effects on cellular health therefore, multiple quality control and stress response mechanisms have evolved. The mitochondrial unfolded protein response (mtUPR) detects proteostatic stress within mitochondria [166]. Central to the mtUPR is the transcription factor ATF5. When stress causes protein import and/or electron transport chain dysfunction ATF5 accumulates in the nucleus to transcribe mitochondrial chaperones and protease genes (figure 4c) [167,168]. The Caenorhabditis elegans homologue ATFS-1 has also been shown to repress translation of the electron transport chain subunit and assembly proteins from both mitochondrial and nuclear genomes [169]. Translation of ATF5 is partly controlled by its homologue ATF4, both of which are upregulated in the integrated stress response (ISR) [170,171]. The ISR can be triggered by ER stress, amino acid starvation or degradation of hTim17A, a TIM23 complex subunit [172,173]. The ISR is characterized by phosphorylation of eIF2α, leading to global reduction of translation and selective induction of cytoprotective genes including pro-survival MCL1 and autophagy proteins. This illustrates the preference for clearance of defective organelles over controlled cell death although the response may alter with cell type or insult [174].

The selective autophagic clearance of mitochondria is termed mitophagy and is controlled by the mitochondrial serine/threonine protein kinase PINK1 and the E3 ubiquitin ligase Parkin. PINK1 is constitutively imported into healthy mitochondria through the TOM complex and laterally released into the inner membrane by TIM23 [175] before cleavage by the PARL protease (figure 4b) [176]. Depolarization of the inner membrane in defective mitochondria prevents import of PINK1, causing it to oligomerize at the outer membrane TOM complex [177], where it becomes auto-phosphorylated [178]. This triggers phospho-PINK1 phosphorylation of basal outer membrane monoubiquitin and recruits Parkin to rapidly poly-ubiquitinate outer membrane proteins for the recruitment of autophagosome factors (figure 4b) [179,180]. Recent data suggest that mitochondria can identify and initiate mitophagy of specific tubules [181], while mitophagy induced by CSNK2/CK2 phosphorylation of hTom22, FUNDC1 and BCL2L13 suggests a potential cytoplasmic influence or pathway [182–185]. Additionally, observations of transcellular mitophagy in astrocytes illustrate much is still unknown in these processes [186].

New stress responses are emerging that demonstrate the reciprocal communication between mitochondria and cytoplasm. Ablation of MIA import pathways in S. cerevisiae activates the proteasome to mitigate mitochondrial precursor accumulation in the cytosol [187]. This correlates with the mammalian, intermembrane space-specific mtUPR (mtUPRIMS) where ERRα transcriptional activity upregulates intermembrane space proteases and activates the proteasome [188,189] the proteasome being previously shown to degrade unfolded intermembrane space proteins that retrotranslocate to the cytosol [190]. In S. cerevisiae, the proteasome is also engaged by Ubx2 to clear mitochondrial protein precursors arrested during translocation, blocking the TOM complex [191]. Reciprocally, mitochondria can degrade defective proteins to aid cytosolic proteostasis. In S. cerevisiae, cytosolic Vms1 can remove mistranslated mitochondrial precursors from stalled ribosomes and direct their import for intra-mitochondrial degradation [192] and aggregation-prone cytosolic proteins may be imported for intra-mitochondrial degradation if cytosolic Hsp70s fail [193]. Intriguing for further research are reports of lysosomal fusion of mitochondria-derived vesicles enriched for non-natively oxidized proteins [194,195] and the extracellular jettison of aggregates by neurons of C. elegans [196].

(Recommended further reading on mitochondrial quality control: [197–199] on mitophagy: [200,201].)

7. Concluding remarks

This review illustrates the importance of mitochondria to eukaryotic cellular functions. As mitochondrial biologists we are frequently surprised by novel pathways or protein networks that involve mitochondria and/or mitochondrial proteins. Mitochondrial protein import and structural dynamics provide the means for rapid alterations in activity to facilitate biological responses to signalling molecules, nutrient availability and pathogenic insult. The temporal coordination of mitochondrial energetics and their biosynthetic capacity drives cell proliferation and differentiation. However, the highly reactive biochemistry compartmentalized in the organelle makes it capable of inducing cell death and necessitates quality control mechanisms. An understanding of this interplay between mitochondrial functions and their diverse cellular implications is therefore critical to a comprehensive holistic model of cellular homeostasis and biochemistry. The importance of this is evident in the escalating occurrence of mitochondria in post-genomic medical research [202]. Although mitochondria are undeniably hubs of cellular biochemistry, further fundamental research is required. In particular, elucidating how the mitochondrion regulates and integrates the various pathways it is associated with, in specialized cells/tissue types and in the context of health and in disease, will help uncover the true depth of influence this amazing organelle has on eukaryotic cells.

Oxidative Phosphorylation

Movement of Electrons from Cytoplasmic NADH to the Mitochondrial ETC

Intact mitochondrial membranes are impermeable to NADH and NAD + , and in order for glycolysis to continue, NAD + must be continually regenerated in the cytoplasm (see Chapter 26 ). Therefore, reducing equivalents (i.e., electrons) from NADH, rather than NADH itself, are carried across mitochondrial membranes by either malate (Mal) or glycerol 3-phosphate ( Fig. 36-1 ), thus allowing for cytoplasmic NAD + reformation, and for NADH and/or FADH2 utilization in the mitochondrial ETC. In the Mal shuttle, reducing equivalents from NADH are accepted by oxaloacetate (OAA), thus forming Mal which crosses mitochondrial membranes via an α-ketoglutarate (α-KG = )-Mal antiporter. Inside mitochondria, NADH is regenerated from Mal, and OAA, which also cannot cross mitochondrial membranes, is returned to the cytoplasm via reversible conversion to aspartate (Asp see Chapters 9 and 35 ). The amine group from Asp is transferred to α-KG = in the cytoplasm, thus forming glutamate (Glu), which is returned to mitochondria via an Asp-Glu antiporter. Inside mitochondria, Glu transfers its amine group to OAA, thus reforming Asp and completing the shuttle.

Another carrier of reducing equivalents is glycerol 3-P, which, like Mal and Asp, readily traverses mitochondrial membranes. This shuttle transfers electrons from NADH to dihydroxyacetone phosphate (DHAP), thus forming glycerol 3-P (and NAD + ) in the cytoplasm. Glycerol 3-P then crosses the outer mitochondrial membrane, and is reoxidized to DHAP by the FAD prosthetic group of glycerol 3-P dehydrogenase. FADH2 is thus formed on the inner mitochondrial membrane, and DHAP diffuses back into the cytosol to complete the shuttle. In some species, activity of this shuttle decreases after thyroidectomy. Although it is present in insect flight muscle, the brain, brown adipose tissue, white muscle tissue and the liver of mammals, in other tissues (e.g., heart muscle), mitochondrial glycerol 3-P dehydrogenase is deficient. It is therefore believed that the malate shuttle is of more universal utility than the glycerol 3-P shuttle, particularly since 3 rather than 2 ATP can be generated per atom of O2 consumed (see below).

2. Compartmentalized Amino Acid Metabolism

2.1. Glutamine

Under physiological conditions, glutamine is one of the most abundant amino acids in circulation [4,5]. Glutamine supply is derived from both dietary sources and de novo synthesis, the latter of which requires glutamate and ammonia and is catalyzed by glutamine synthetase (GS) in the cytosol. Activity of GS is well described in the brain as a means of removing excess ammonia by astrocytes, and dysfunctional ammonia metabolism can lead to hepatic encephalopathy and cerebral edema [6,7]. In cells with high proliferative rates (e.g., cancer cells, activated T lymphocytes), glutamine demand outweighs supply and environmental access becomes 𠇌onditionally essential” [8,9,10,11,12]. When nutrients become locally limited, several cancers, including pancreatic, glioblastoma, and ovarian, hijack stromal glutamine synthesis as an alternative supply line to fulfill their increased demands [13,14,15]. Furthermore, glutamine deprivation suppresses expansion of activated T lymphocytes, and competition for nutrients within tissues may affect immune responses to pathological states that exhibit hallmark increases in nutrient consumption (e.g., viral infection, cancer) [10,16,17,18].

The abundance of glutamine in circulation reflects its robust versatility to satisfy metabolic requirements beyond protein translation ( Figure 1 ). Glutaminolysis represents one of the major catabolic pathways important for the generation of TCA intermediaries, fatty acids, reducing equivalents necessary for oxidative phosphorylation, and non-essential amino acids. The first step of glutaminolysis occurs in the mitochondria and is catalyzed by the enzyme glutaminase (GLS), which converts glutamine to glutamate. There are two isozymes of GLS, the kidney-type (GLS1) and the liver-type (GLS2) [19]. Mitochondrially-produced glutamate serves a number of direct and indirect metabolic roles. For example, glutamate can be oxidized by the NAD(P) + -dependent enzyme glutamate dehydrogenase (GLUD1) or contribute its amino-nitrogen for non-essential amino acid synthesis by cytosolic and/or mitochondrial transaminases (e.g., GOT1/2 for aspartate, PSAT1 for serine, GPT1/2 for alanine). Glutamate is also required for glutathione (GSH) synthesis and is used as a backbone for proline and arginine biosynthesis. On the other hand, catabolism of other amino acids (e.g., proline) has been shown in several contexts to be an important source of glutamate. To provide metabolic flexibility in nutrient-limited conditions, pancreatic cancer cells scavenge collagen peptides from the extracellular matrix and utilize proline as an anaplerotic source when glutamine levels are low [20]. Furthermore, proline catabolism by proline dehydrogenase (PRODH) has also been shown to be an important source of glutamate in metastasizing breast cancer cells [21]. In other contexts, mitochondrial pyrroline 5-carboxylate reductase 1 (PYCR1) redirects excess mitochondrial NADH and/or glutamate towards proline synthesis in isocitrate dehydrogenase 1 (IDH1) mutant glioma cells, leading to a partially uncoupled TCA cycle that allows cells to regulate mitochondrial NAD + /NADH [22].

Biochemical pathways and transporters involving glutamine (Gln) and related intermediates. Glutamine is transported by plasma membrane transporters (e.g., SLC1A5/ASCT2, SLC6A14/ATB 0,+ , SLC38A1/SNAT1, SLC38A2/SNAT2) and fuel nucleotide, amino acid, and glycosyl synthesis via asparagine synthetase (ASNS), carbamoyl phosphate synthetase I (CPS1), phosphoribosyl pyrophosphate amidotransferase (PPAT), and glutamine-fructose 6-phosphate aminotransferase (GFPT1). Sodium (Na + ) and chloride (Cl − ) gradients across the plasma membrane determine the intracellular concentration of glutamine. Glutamine directly contributes nitrogen for purine and pyrimidine biosynthesis (marked in red). Cytosolic glutamine can also transport into mitochondria via a mitochondrial-targeted (MTS) SLC1A5 variant (MTS-SLC1A5 also referred to as SLC1A5_var) where it acts as a major anaplerotic source for tricarboxylic acid (TCA) cycle metabolism (‘glutaminolysis’). Glutaminolysis is inhibited by BPTES or CB-839, which specifically target glutaminase (GLS). Glutamine-derived glutamate is a significant source of carbon and nitrogen for non-essential amino acid synthesis. αKG, α-ketoglutarate Ala, alanine Asp, aspartate CP, carbamoyl phosphate F6P, fructose 6-phosphate GlcN6P, glucosamine 6-phosphate Gln, glutamine Glu, glutamate Oac, oxaloacetate P5C, pyrroline 5-carboxylate PRA, 5-phospho-β- d -ribosylamine Pro, proline Pyr, pyruvate.

While the capability to utilize glutaminolysis pathways is transcriptionally inherent in the genome of all cells, the tendency to shunt glutamine towards them is likely tissue-specific, cell-type specific, and depends on conditional metabolic needs [23]. For example, neonatal mammals utilize proline, not glutamine, as the major anabolic input for arginine synthesis [24]. On the other hand, a shift to glutaminolysis as the major glutamate source is intrinsic across many cancer contexts. Upregulation of GLS by oncogenic pathways is a common mechanism for this metabolic shift. The oncogenic transcription factor c-Myc has been implicated as one driver of GLS1 upregulation by suppressing the inhibitory effects on GLS1 translation by miR-23a/b [25]. This upregulation of GLS1 through post-transcriptional mechanisms has also been attributed to other pro-neoplastic factors. NF-㮫 member p65 downregulates miR-23a transcription in leukemic cells, while HSF1, which is ubiquitously expressed in several cancer types, suppressed transcription of the GLS1-inhibitor miR-137 [26,27]. Additionally, the transcription factor c-Jun, a downstream effector of oncogenic-Dbl and the JNK-MAP kinase pathway, was shown to directly bind the GLS1 promoter and promote its upregulation [28]. In contrast to GLS1, GLS2 has been identified as a target for the tumor suppressor p53 and thus may support anti-cancer properties [29,30]. This apparent contradiction may be a result of differences in properties between the two GLS isozymes and requires further investigation.

Other oncogenes have been further shown to mediate pro-cancer effects through upregulation of glutaminolysis. In pancreatic ductal adenocarcinoma cells (PDAC), oncogenic KRAS was shown to shift glutamine metabolism by upregulating cytoplasmic GOT1 and downregulating GLUD1, stimulating a pathway in which glutamine-derived aspartate from the mitochondria is used as a metabolite to generate cytosolic OAA [11]. Subsequent activity of malate dehydrogenase (MDH) and cytosolic malic enzyme (ME1) supply PDAC cells with reduced pyridine nucleotides necessary for redox homeostasis [11]. In a separate study, PDAC cells cultured in acidic conditions also exhibited an increased dependence on glutamine for redox homeostasis and anaplerosis [31]. In colorectal cancer (CRC), mutations in the PIK3CA gene, which encodes for the p100α subunit of PI3K, lead to upregulation of GPT2 and reliance on glutamine-derived TCA intermediates to sustain growth [32]. Additionally, the liver receptor homolog 1 (LRH1) has been implicated as a transcription factor, which drives tumor formation via effects on glutamine metabolism in hepatocellular carcinoma (HCC) [33]. Overall, the upregulation of glutaminolysis in cancer cells is near ubiquitous and achieved through many different oncogenic effectors.

Inhibiting aberrant glutaminolysis has largely focused on targeting mitochondrial glutaminase activity. BPTES and the more soluble and bio-available CB-839 selectively inhibit GLS1 and have been investigated as anti-neoplastic agents in several contexts [34,35,36,37,38]. However, certain cancer types (e.g., pancreatic, lung) demonstrate contradicting sensitivity to GLS inhibition in vitro and in vivo, suggesting that tumors may be more glutaminolysis independent in vivo than modeled in culture [39,40]. Further, these studies highlight the plasticity of glutamine and glutamate metabolism and suggest that cells may autonomously re-route metabolic flux to supply glutamate through other means ( Figure 1 ). Glutamate is produced during the synthesis of purine and pyrimidine nucleobases and the glycosylation subunit N-acetyl-glucosamine (GlcNAc). Purine and pyrimidine synthesis utilize the γ-nitrogen of glutamine to generate 5-phospho-β- d -ribosylamine (PRA) and carbamoyl phosphate (CP) by phosphoribosyl pyrophosphate amidotransferase (PPAT) and the carbamoyl phosphate synthetase (CPS) domain of the CAD complex, respectively [41,42,43]. Cytosolic glutamine is also a substrate for glutamine-fructose 6-phosphate aminotransferase (GFPT1), which is used to produce glutamate and glucosamine 6-phosphate (GlcN6P) a precursor for O-linked N-acetylglucosaminylation [44]. Furthermore, cytosolic asparagine synthesis by asparagine synthetase (ASNS) yields glutamate as well. Given the numerous glutamate supply routes, efforts to target the glutamine demands in cancer have broadened to identify antagonists targeting more than one glutamine-dependent enzyme simultaneously. 6-diazo-5-oxo-L-norleucine (DON) was developed decades ago as a potential anti-neoplastic agent for its inhibitory activity against many glutamine-dependent enzymes, including glutaminase and glutamine amidotransferases [45]. However, gastrointestinal (GI) toxicity in the majority of patients receiving DON limited its clinical use [46]. More recently, pro-drug forms of DON have been developed with enhanced delivery properties to either the brain or tumors and reduced GI toxicity, which has reinvigorated interest in using glutamine antagonists as antitumor agents [47,48,49,50,51]. An alternative strategy is limiting cancer cell access to glutamine by inhibiting transporter-dependent uptake. Two of the most well documented glutamine transporters are from the SLC1A and SLC38A solute carrier (SLC) families, and the more promiscuous Na + /Cl − -dependent SLC6A14/ATB 0,+ transporter can also play a role in importing glutamine [52,53,54]. SLC1A5/ASCT2 inhibitors have been identified and exhibit promising anti-tumor properties in preclinical models [55,56,57,58,59]. Furthermore, targeting secondary glutamine transporters (e.g., SLC38A2, SLC6A14) genetically or pharmacologically (e.g., α-methyltryptophan) significantly suppresses amino acid homeostasis and tumor growth in pancreatic cancer [60,61].

2.2. Aspartate

Aspartate is an acidic non-essential amino acid that can be acquired by either de novo synthesis and/or import from external sources. However, circulating levels of aspartate in physiological conditions are low (

10 µM) and maintained by liver aspartate transaminases thus, synthesis likely provides the majority of cellular aspartate in most contexts [4]. Biosynthesis of aspartate is carried out via aspartate aminotransferases (glutamic-oxaloacetic transaminases) in the cytosol (GOT1) and in the mitochondrial matrix (GOT2), which as discussed above utilize glutamate as the amino-nitrogen source. Aspartate has many biosynthetic fates within the cell (e.g., proteins, nucleotides, and amino acids) and also serves as an exchange factor for the aspartate-glutamate carrier (AGC1/AGC2), an essential component of the malate-aspartate-shuttle (MAS) ( Figure 2 ). MAS is responsible for transferring electrons from cytosolic NADH to mitochondrial NADH, as reducing equivalents (e.g., NAD(P)H) cannot directly cross the inner mitochondrial membrane. However, recent studies identified that SLC25A51 and SLC25A52 facilitate mitochondrial NAD + transport [62,63,64]. Subsequent activity of the MAS and/or UCP2 is required to export aspartate into the cytosol where it can be used as a proteinogenic source and/or a precursor for arginine and asparagine synthesis [65,66].

Biochemical pathways and transporters involving aspartate (Asp) and related intermediates. Aspartate is transported by the plasma membrane transporter SLC1A3, which also transports glutamate. Aspartate is synthesized by glutamic-oxaloacetic transaminases (GOT) present in the cytosol (GOT1) or mitochondria (GOT2). Mitochondrial efflux of aspartate mainly occurs through SLC25A12 or SLC25A13, which counter-exchange glutamate and are critical components of the malate-aspartate-shuttle (MAS), and UCP2. Cytosolic aspartate is used as a substrate for asparagine and arginine synthesis via asparagine synthetase (ASNS) and argininosuccinate synthase (ASS1) and as a substrate for nucleotide biosynthesis, contributing carbon and nitrogen to purine and pyrimidines (marked in red). Cytosolic asparagine is used as an exchange factor for several amino acids through an unknown plasma membrane transporter. AA, amino acid AcCoA, acetyl-coenzyme A αKG, α-ketoglutarate Asn, asparagine Asp, aspartate FH, fumarate hydratase Gln, glutamine Glu, glutamate GSH, reduced glutathione GSSG, oxidized glutathione Mal, malate Oac, oxaloacetate Pyr, pyruvate SDH, succinate dehydrogenase UCP2, uncoupling protein 2.

In many contexts, aspartate is predominantly synthesized by mitochondrial GOT2 and is suggested to be one output of mitochondrial electron transport chain (ETC) activity in proliferating cells [11,67,68,69,70,71,72]. Although ATP is another major output of ETC activity, proliferating cells with sufficient access to glucose can switch to aerobic glycolysis to largely satisfy these requirements [67]. Aspartate serves a biosynthetic role, acting as a nitrogen donor for adenine synthesis and a carbon backbone via orotate for pyrimidine synthesis. The availability of aspartate has been suggested to be limiting for the proliferation of certain cancers. Sullivan et al. utilized a guinea pig asparaginase (gpASNase1) to supply tumors with asparagine-derived aspartate and observed enhanced tumor growth in HCT116 and AL1376 colorectal and murine PDAC cell lines, respectively [73]. Interestingly, gpASNas1 had little to no effect on the human AsPC1 tumor growth. Similarly, some cancer cells utilize a plasma membrane glutamate and aspartate transporter, SLC1A3, to provide aspartate in conditions where de novo synthesis is restricted by ETC inhibition [74]. Environmental acquisition of aspartate by SLC1A3 has also been implicated in hypoxic microenvironments or in response to glutamine restriction [72,74,75]. Hypoxia reportedly suppresses mitochondrial aspartate biosynthesis via HIF1α-dependent down-regulation of GOT1 and GOT2 in Von Hippel-Lindau (VHL)-deficient renal carcinoma cells [76]. However, pancreatic cancer cells have been shown to sustain aspartate biosynthetic fluxes in oxygen tensions as low as 0.1% O2 through activity of Complex III+IV containing respiratory supercomplexes, which are suggested to promote efficient respiration in limiting oxygen environments [77]. Notably, maximal HIF stabilization occurs in

1% O2, well above tensions where oxygen becomes limiting for mitochondrial respiration [78]. Although glutaminolysis provides cells with the majority of carbon necessary to synthesize aspartate, in cancer subtypes driven by TCA cycle deficiencies (e.g., SDH- or FH- deficiency), pyruvate carboxylase activity can divert glucose-derived pyruvate to supply oxaloacetate necessary for this anabolic function [79,80,81,82,83,84]. This shift to PC-dependent aspartate synthesis was also observed in PDAC tumors in vivo and in breast and lung cancer cell lines exposed to hypoxic oxygen tensions [77,85]. Taken together, aspartate is a critical anabolic metabolite necessary to supply nucleotides for proliferating cancer cells however, its synthesis from glutamine- and/or glucose-derived pathways are complex and highly dependent on the environmental context and nutrient availability.

Cytosolic aspartate is utilized by ASNS and ASS1 for asparagine and arginine biosynthesis, respectively ( Figure 2 ). The production of these amino acids supports protein translation, but also play indirect roles for proliferation. For example, asparagine acts as an amino acid exchange factor to facilitate the influx of other amino acids (e.g., serine, threonine) necessary to regulate mammalian target of rapamycin complex 1 (mTORC1) activity and proliferation [86]. Arginine is a major source of cellular nitric oxide (NO), through activity of iNOS, or is catabolized by arginase as the final step of the urea cycle. Arginine, and other basic amino acids (e.g., lysine, ornithine), can also transport into the mitochondria by SLC25A29 [87]. Expression of SLC25A29 was shown to be elevated in several cancer cell lines and important for NO production by a mitochondrial NOS [88]. Activity of extrahepatic arginase 2 (ARG2) also regulates mitochondrial NO production [87,89,90]. Notably, several cancers down-regulate the activity of these pathways via silencing of ASS1 and/or ASNS expression, creating a dependence (auxotrophy) for environmental and/or stromal acquisition of these amino acids [91,92,93,94]. Silencing of ASS1 and/or ASNS may provide a selective advantage for cancer cells, allowing for diversion of aspartate towards other anabolic pathways such as nucleotide biosynthesis. Importantly, activity of the MAS and/or other mitochondrial aspartate transporters (e.g., UCP2) represent a key step for regulating compartmental availability of this critical amino acid [65,66].

2.3. Serine, Glycine and Alanine

The metabolism of small neutral amino acids serine, glycine, and alanine occurs in both the cytosol and mitochondria and has implications for physiology and human diseases ( Figure 3 ). Serine consists of a simple hydroxymethyl side chain and is either taken up or synthesized de novo from the glycolytic intermediate 3-phosphoglycerate by three cytosolic enzymes, phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase (PSAT1), and phosphoserine phosphatase (PSPH). Activity of serine synthesis is important for normal development, as deletion of Phgdh causes embryonic lethality in part due to neurological defects [95]. Serine, specifically D-serine, is thought to be a critical excitatory neurotransmitter acting as a co-agonist of the N-methyl D-aspartate (NMDA) receptor on neurons, and depletion due to deficient synthesis or racemase activity likely leads to catastrophic neurotoxicity [96]. Abnormal D-serine levels in the brain are thought to contribute to neurodegenerative disorders such as Alzheimer’s disease and schizophrenia [97,98,99,100]. Expression of serine biosynthesis enzymes are highly regulated by several factors, including NRF2-ATF4 [101,102], c-Myc [103], and hypoxia inducible factors [104]. Many of these transcriptional regulators are altered in cancer and contribute to increased serine synthesis flux, but PHGDH expression was also found to be amplified through copy number gain of a genomic region on chromosome 1p12 in a subset of breast and melanoma [105,106]. PHGDH expression has been demonstrated to support tumor growth specifically in low serine environments, such as cerebrospinal fluid where concentrations are significantly lower than plasma, and dietary restriction of serine and glycine reduces tumor growth in preclinical cancer models and enhances activity of mitochondrial inhibitors [107,108,109,110,111]. Furthermore, a subset of PDAC cells downregulate serine synthesis enzymes, and neuronal supply has been shown to supply cancer cells with serine specifically in null environments [112]. As PHGDH is thought to be the rate-limiting step of serine synthesis and important for the proliferation of PHGDH-amplified cancer cell lines, several inhibitors have been developed [113,114,115]. Notably, serine synthesis and uptake can occur in parallel with catabolism, depending on the context and cell intrinsic demand for serine and/or its many catabolic outputs.

Metabolism of small, neutral amino acids including serine (Ser), glycine (Gly), and alanine (Ala). Serine, glycine, and alanine are mainly transported by the plasma membrane transporters SLC38A1, SLC38A2, SLC1A4, and SLC1A5 or synthesized de novo by cytosolic and/or mitochondrial pathways. Sodium (Na + ) gradients across the plasma membrane drive intracellular concentration of serine, glycine, and alanine. Serine is synthesized from 3-phosphoglycerate (3pg) through a three-step process involving 3-phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase (PSAT1), and phosphoserine phosphatase (PSPH). Cytosolic serine is used for several metabolic pathways, including the transsulfuration pathway for de novo cysteine synthesis and folate-mediated one carbon metabolism (FOCM) involving serine hydroxymethyltransferase (SHMT) and methylenetetrahydrofolate dehydrogenases (MTHFD). FOCM occurs in both the cytosol and mitochondria and can produce glycine. In the cytosol, glycine is utilized for glutathione synthesis and purine synthesis (marked in red) and acts as a substrate for the methionine cycle important for methylation reactions. In the mitochondria, glycine can be cleaved by the glycine cleavage system (GCS) to provide one carbon unites for FOCM and is a substrate for δ-aminolevulinic acid (δ-ALA) synthesis. When GCS activity is low, mitochondrial glycine can lead to the accumulation of methylglyoxal. Components of FOCM, including 5,10-methylene-tetrahydrofolate (5,10-meTHF) and 10-formyl-tetrahydrofolate (10-formylTHF) can contribute to nucleotide biosynthesis (marked in dark blue and dark red, respectively). Mitochondrial serine import is facilitated by sideroflexin 1 and 3 (SFXN1/3), formate is transported by an unknown transport(er) mechanism, and glycine is imported by SLC25A38 and other transporters may be involved. Alanine is synthesized from pyruvate by cytosolic or mitochondrial glutamic-pyruvic transaminases (GPT) localized to the cytosol (GPT1) or mitochondria (GPT2). Pyruvate is imported into the mitochondria by the mitochondrial pyruvate carrier (MPC1/2) an obligate heterodimer. Alanine is exchanged between the cytosol and mitochondria by an unknown transporter. 10-formylTHF, 10-formyl-tetrahydrofolate 3pg, 3-phosphoglycerate 5,10-meeTHF, 5,10-methenyl-tetrahydrofolate 5,10-meTHF, 5,10-methylene-tetrahydrofolate 5-mTHF, 5-methyl-tetrahydrofolate accoa, acetyl-coenzyme A αKG, α-ketoglutarate Ala, alanine Cys, cysteine δ-ALA, δ-aminolevulinic acid DMG, dimethylglycine Gap, glyceraldehyde 3-phosphate Gluc, glucose Gly, glycine GSH, reduced glutathione Homocys, homocysteine Lac, lactate Met, methionine Oac, oxaloacetate p-Pyr, 3-phosphopyruvate p-Ser, phosphoserine Pyr, pyruvate SAH, S-adenosylhomocysteine SAM, S-adenosylmethionine Ser, serine Succ-coa, succinyl-coenzyme A THF, tetrahydrofolate.

Serine is utilized for glycine synthesis as well as ceramide and sphingolipid synthesis, nucleotide synthesis, folate-mediated one carbon metabolism (FOCM), S-adenosyl methionine regeneration for methylation reactions, and transsulfuration for cysteine biosynthesis ( Figure 3 ). Glycine synthesis requires the cytosolic and mitochondrial enzymes serine hydroxymethyltransferase (SHMT1/2) producing one-carbon units in the form of 5,10-methylene-tetrahydrofolate (5,10-meTHF). Subsequent activity of methylenetetrahydrofolate dehydrogenases (MTHFD1/1L/2) releases formate, which can be transported across mitochondrial membranes. The resulting metabolic cycle acts as a shuttle for NAD(P)H reducing equivalents and one-carbon units required for thymidine and purine synthesis. The directionality of the FOCM metabolic cycle operates predominantly oxidatively in the mitochondria but can reverse to maintain one-carbon supply for nucleotide synthesis in cases where mitochondrial isoforms are deleted [116,117,118]. High activity and dependence on SHMT1/2 for proliferation has been demonstrated in multiple cancer contexts, and inhibitors targeting these enzymes have been developed [105,106,119,120]. In replete environments, serine catabolism by SHMT2 can occur in excess and release of glycine and formate, termed 𠇏ormate overflow”, has been reported for several cancer and non-transformed cell lines and in mice [121]. Through this mechanism, serine catabolism acts as a significant source of both ATP and NAD(P)H, and flux through this pathway was demonstrated to support oxidative mitochondrial metabolism [122]. In response to pharmacological inhibition of respiration expected to increase mitochondrial NADH/NAD + , mitochondrial serine catabolism is sustained, whereas other enzymatic NADH sources (e.g., pyruvate dehydrogenase) were feedback inhibited [123]. Thus, serine metabolism is highly complex and can provide cells with glycine, one-carbon units, NAD(P)H, and ATP depending on the context and cellular demand [124].

Glycine can also contribute one-carbon units through mitochondrial activity of the glycine cleavage system (GCS), which releases CO2, NH3, and 5,10-meTHF ( Figure 3 ). Notably, activity of GCS in many cancer cell lines was found to be low relative to serine catabolism [125]. However, in the context of cancer cell lines with high mitochondrial SHMT2 flux, activity of GCS is necessary to clear excess glycine and prevent a build-up of the toxic byproducts aminoacetone and methylglyoxal derived from the interconversion of glycine and threonine [126]. Methylglyoxal was found to accumulate in non-small cell lung cancers relative to normal tissue, and sequestration of toxic methylglyoxal requires glutathione (GSH) and activity of glyoxalase (GLO1) to prevent cellular damage [127]. Activity of the GCS has been shown to be important for the maintenance of stem cell pluripotency through epigenetic regulation [128]. Extracellular glycine can also be used for SHMT-dependent serine synthesis but requires an exogenous source of formate [125]. In addition to FOCM, glycine is also a precursor required for the synthesis of GSH and δ-aminolevulinic acid (δ-ALA mitochondrial) necessary for heme biosynthesis. Synthesis of glutathione occurs in the cytosol, requiring mitochondrial export of glycine and GSH import into intracellular organelles including the mitochondria, which contains

10�% of cellular GSH at a similar concentration to the cytosol [129]. Maintenance of GSH pools is important for regulating the activity of proteins sensitive to post-translational oxidation of cysteine residues (e.g., PTP1B) [130,131,132]. Insight into the activity and downstream role of serine and glycine metabolism can be gained from examination of extracellular uptake and secretion however, cytosolic-mitochondrial exchange is equally important and requires a number of plasma membrane and mitochondrial transporters [121,133].

Alanine synthesis requires cytosolic and/or mitochondrial glutamic-pyruvic transaminases (GPT1/2). Physiological synthesis occurs in skeletal muscle from pyruvate and glutamate derived from glycolysis and BCAA catabolism, respectively. Alanine secreted by muscles provides the carbon necessary for gluconeogenesis in the liver, which in turn provides glucose back to muscles and sequesters the nitrogen produced from alanine catabolism as urea [134,135,136,137]. The resulting glucose/alanine cycle, referred to as the �hill cycle”, is an important organ crosstalk relevant during normal physiology, exercise, fasting, and disease [138]. Dysregulation of this cycle has been proposed to occur in cancer patients, whereby increased protein turnover and/or muscle breakdown (�hexia”) releases alanine, BCAAs, and other amino acids for hepatic gluconeogenesis [139,140,141]. Elevated hepatic alanine-to-glucose conversion was measured in lung and other cancer patients, but plasma alanine levels remained mostly stable [142,143,144,145]. Hepatocytes express both cytosolic (GPT1) and mitochondrial (GPT2) isoforms required for de novo alanine synthesis and catabolism. However, biochemical parameters and studies suggest that GPT1 (Km, ala = 34 mM) and GPT2 (Km, ala = 2 mM) exhibit preference towards alanine anabolism and catabolism, respectively, although this was highly dependent on the method used to ascertain directionality [146,147,148,149,150]. Recent evidence suggests that pancreatic cancer cells have a high demand for alanine, and scavenge alanine from stromal sources (e.g., activated stellate cells) [60,151]. The majority of human pancreatic cancer cells selectively express GPT2, at both the transcript and protein level, suggesting that alanine metabolism occurs mainly in the mitochondria [60]. Notably, mitochondrial alanine catabolism by GPT2 requires activity of a mitochondrial alanine transporter, which was functionally identified decades ago but remains unknown [147]. Low expression of cytosolic GPT1, which catalyzed alanine synthesis from pyruvate in hepatocytes, and alanine uptake was suggested to provide pancreatic cancer cells with the capacity to retain pyruvate in the cytosol and support aerobic glycolysis [60,152]. In contrast, naïve T lymphocytes require alanine for activation as neither GPT1 nor GPT2 are expressed at sufficient levels [153]. Alanine production from pyruvate was found to be important for the metastasis of breast cancer cells, providing a source of α-ketoglutarate used for collagen hydroxylation and extracellular matrix (ECM) remodeling [154]. Importantly, it has been suggested that transport across the plasma membrane may be the main rate-limiting step of alanine metabolism at extracellular concentrations ρ mM [147,155,156]. Normal plasma levels of alanine are

0.2𠄰.4 mM, but elevated levels (

1 mM) have been measured intratumorally, suggesting altered alanine metabolism and availability in cancer and tumor-associated stromal cells [157,158]. Notably, SLC38A2/SNAT2 was identified to be the main concentrative alanine transporter utilized by pancreatic cancer cells and targeting SLC38A2 was sufficient to suppress alanine uptake by pancreatic cancer cells and cause significant re-wiring of compartmentalized pyruvate metabolism [60]. Taken together, these studies suggest that perturbing alanine metabolism in cancer is possible by altering plasma membrane transport, and mitochondrial alanine transport may be a key player in glucose-pyruvate-alanine metabolism by skeletal muscle, hepatocytes, and cancer cells.

2.4. Branched-Chain Amino Acids

Branched chain amino acids (BCAAs) include leucine, isoleucine, and valine and are derived from dietary sources. Because of their essentiality in mammals, BCAA transport and sensing in addition to catabolic mechanisms of acquisition (e.g., autophagy, macropinocytosis) has attracted much interest. Cellular uptake of BCAAs is mainly facilitated by the L-type amino acid transporter (SLC7A5/LAT1), which requires dimerization with SLC3A2/CD98 to function. It also transports aromatic amino acids (e.g., tyrosine, phenylalanine) ( Figure 4 ) [159,160]. Notably, LAT1 is sodium-independent and relies on other amino acids to serve as exchange factors to facilitate net BCAA import [159]. Leucine is well characterized to influence mTORC1 signaling, which is aberrantly activated across many cancer types [161]. Leucine can activate mTORC1 signaling through direct sensing by Sestrin2 and disruption of the Sestrin2-Gator2 interaction, triggering a signaling cascade through downstream effectors (e.g., eukaryotic translation initiation factor 4E binding protein 1, p70-S6 kinase, ULK1) [161,162,163,164]. These signals coordinate proliferation through activity of autophagy and protein, lipid, and nucleotide synthesis.

Biochemical pathways and sensing mechanisms involving the branched-chain amino acids (BCAAs) leucine (Leu), isoleucine (Ile), and valine (Val). BCAAs are mainly imported by the large amino acid transporter (LAT1), a dimer consisting of SLC7A5 and SLC3A2/CD98, which functions as an amino acid exchanger and the Na + -dependent SLC16A9/B0AT1. Glutamine—mainly transported by SLC38A1, SLC38A2, and SLC1A5—is thought to provide the chemical driving force necessary to influx BCAAs through LAT1. Cytosolic leucine is directly sensed by Sestrin2 and regulates mTORC1-dependent signals that control autophagy, protein synthesis, and proliferation. BCAAs can be metabolized by the BCAA transaminase (BCAT) present in the cytosol (BCAT1) or mitochondrial (BCAT2), which produce branched-chain ketoacids (BCKAs). Cytosolic BCKAs can be transported through SLC16A monocarboxylate transporters present on the plasma membrane or mitochondria. BCAAs are imported into the mitochondria by SLC25A44 and can contribute to acyl-CoA production through activity of BCAT2 and BCKA dehydrogenase (BCKDH), which is regulated by BCKDH kinase (BCKDK) and Mg 2+ /Mn 2+ -dependent 1K protein phosphatase (PPM1K). Acyl-CoA produced by BCAA catabolism can fuel TCA cycle metabolism and de novo lipogenesis. Acetyl-CoA levels are sensed through EPS300-dependent acetylation of RAPTOR, which in-turn regulates mTORC1 activity. Mitochondrial propionyl-CoA, produced either from valine or isoleucine catabolism, is metabolized to produce succinyl-CoA, but can produce the byproduct methylmalonate (MMA) when vitamin B12 levels are low. AcCoA, acetyl-coenzyme A αKG, α-ketoglutarate Gln, glutamine Glu, glutamate Ile, isoleucine KIC, α-ketoisocaproic acid KIV, α-ketoisovaleric KMV, α-keto-β-methylvaleric Leu, leucine Oac, oxaloacetate PropCoA, propionyl-coenzyme A SucCoA, succinyl-coenzyme A Val, valine.

Because of the abundant expression of SLC1A5/ASCT2 and LAT1, it has been suggested that ASCT2-dependent glutamine uptake may serve as the exchange factor for BCAA influx by LAT1. However, ASCT2 is dispensable for the proliferation and mTORC1 signaling in many cancer lines, and ASCT2 functions primarily as an exchanger unable to concentrate glutamine sufficiently to drive LAT1 activity [55,165,166]. Thus, secondary active glutamine transporters (e.g., SNAT1/SLC38A1, SNAT2/SLC38A2, SLC6A14/ATB 0,+ ) are more likely to contribute to glutamine concentration for LAT1-mediated exchange. However, deletion of SLC38A2 in pancreatic cancer failed to impact either BCAA or glutamine uptake flux despite significantly decreasing intracellular glutamine levels [60]. Rather, transporter cooperativity between glutamine and BCAA transporters may be more important for level maintenance. Indeed, LAT1 knockout results in a

90% decrease in leucine transport in hepatocellular carcinoma cells but fails to illicit proliferative defects, and knockdown or inhibition of LAT1 did not negatively impact mTORC1 re-activation following EAA stimulation [165,167]. Furthermore, knockout of SLC3A2/CD98 abolished

90% of leucine uptake by LAT1 in colon adenocarcinoma cells, but proliferative defects and activation of the GCN2-linked amino acid stress response were not observed [168]. Thus, plasma membrane transport of BCAA and/or glutamine may not be limiting or is highly dependent on the cellular context, and minimal transport capacity may be sufficient to satisfy the biosynthetic and catabolic demands for these amino acids. In contrast, LAT1 was significantly upregulated in an Apc fl /fl LSL-Kras G12D /+ Villin CreER mouse model of colorectal cancer, and targeted deletion of Slc7a5 resulted in delayed tumorigenesis and improved survival [169]. Furthermore, JPH203, a small molecule inhibitor of LAT1, has shown significant pre-clinical efficacy in colorectal cancer and T-cell lymphoblastic lymphoma/leukemia and was well-tolerated in a Phase I study in patients with advanced solid tumors [170,171,172]. Other transport systems can facilitate BCAA uptake, including the Na+-dependent SLC6A19/B0AT1, which may contribute to differing sensitivity in response to LAT1-deletion [173,174]. Inhibitors targeting SLC6A19/B0AT1 have been developed using in silico and high-throughput screening approaches [175,176].

Aside from being used for protein synthesis, BCAAs can contribute to anabolic and bioenergetic outputs important for human physiology and dysregulated activity is attributed to multiple diseases (reviewed in [161,177,178]) ( Figure 4 ). Through catalytic activity of highly reversible branched chain aminotransferases (BCAT1/2) localized within the cytosol (BCAT1) or mitochondrial matrix (BCAT2), BCAA catabolism provides cells with amino-nitrogen for glutamate synthesis as well as branched chain ketoacids (BCKAs) (e.g., α-ketoisocaproic, KIC α-ketoisovaleric, KIV α-keto-β-methylvaleric, KMV) that contribute to acyl-CoA synthesis, lipogenesis, and TCA cycle metabolism. While BCAT2 is ubiquitously expressed, BCAT1 is selectively expressed in the brain, ovary, and placenta [179]. BCAT1 is commonly up-regulated in many different cancer lines, such as human glioblastoma, breast cancer, and non-small cell lung carcinoma (NSCLC), while BCAT2 seems more important for pancreatic cancer [180]. Furthermore, elevated plasma BCAA levels are associated with several diseases, including cardiovascular disease, pancreatic cancer, and breast cancer [139,181,182]. In the mitochondria, BCKAs can undergo irreversible decarboxylation by the branched chain α-ketoacid dehydrogenase (BCKDH) complex, which consists of three subunits (E1, E2, and E3). Activity of BCKDH is negatively regulated by the phosphorylation status of the E1 subunit. BCKDH kinase (BCKDK) and the Mg 2+ /Mn 2+ -dependent 1 K protein phosphatase (PPM1K) coordinate the activity of BCKA oxidation. Activity of PPM1K was shown to positively regulate BCAA catabolism important for leukemogenesis [177,183]. Furthermore, defective BCKA oxidation drives the inborn error of metabolism maple syrup urine disease (MSUD), and dysregulated BCKDH activity is also attributed to several human diseases (e.g., diabetes, cancer) [184].

Acyl-CoA products of BCAA oxidation (e.g., acetyl-CoA, propionyl-CoA, succinyl-CoA) have the potential to contribute carbon for oxidative TCA cycle activity and/or lipogenesis, suggesting that BCAA may serve as an important fuel source for proliferative cells. In addition, acetyl-CoA derived from leucine can provide direct proliferative signals through acetylation of Raptor via EP300, which in turn negatively regulates autophagosome formation and activates mTORC1 signaling [185,186]. Whether this represents a major metabolic contribution, particularly to the TCA cycle, depends highly on the context. The metabolic contribution of BCAA-derived acyl-CoA has been extensively characterized in mutant Kras-driven tumors (e.g., pancreatic, lung) given the correlation between elevated plasma levels and disease progression [139]. In acute myeloid leukemia (AML), human pancreatic cancer, and colorectal cancer cells, as well as in LSL-Kras G12D /+ Trp53 flox /flox -driven lung and pancreatic tumors, 13 C-labeled BCAAs contributed minimally to mitochondrial TCA cycle intermediates irrespective of which BCAT1/2 isoform is expressed in each context [180,187,188,189,190]. In contrast, cancer-associated fibroblasts derived from human pancreatic tumors showed higher BCAA oxidation flux than pancreatic cancer cells, and BCKAs secreted from CAFs were incorporated into the TCA cycle in human pancreatic cancer cells through subsequent oxidation [191]. Similarly, 13 C-KIC, derived from leucine catabolism, was shown to be oxidized by tumors in a rat glioma model using hyperpolarized nuclear magnetic resonance (NMR) spectroscopy [192]. Transport of BCKAs across the plasma membrane is mainly facilitated by monocarboxylate transporters, MCT1/SLC16A1 and MCT4/SLC16A4, allowing cells to share pools of circulating BCKAs to convert to BCAAs if needed [193,194,195,196]. In adipocytes, BCAAs represent a major anaplerotic and lipogenic source. Acetyl-CoA or propionyl-CoA is utilized for even- or odd-chain fatty acid synthesis and, in addition to succinyl-CoA, contributes significantly to TCA cycle intermediates (e.g., citrate) [197,198,199]. Adipose tissue can also utilize BCAA catabolism to generate mono-methylated branched-chain fatty acids through promiscuous activity of carnitine acetyltransferase (CRAT) and fatty acid synthase (FASN) [200]. Notably, the methylmalonyl-CoA mutase required to convert propionyl-CoA to succinyl-CoA is B12-dependent, and odd-chain fatty acids and methylmalonic acid (MMA) accumulate in adipocytes only when cultured in media deficient in cobalamin (e.g., DMEM) [199]. In a recent study, increased MMA levels in circulation correlate with increasing age, and MMA was found to promote an epithelial-mesenchymal transition (EMT)-like phenotype and contribute to increased tumorigenesis [201].

Biogenesis of porin of the outer mitochondrial membrane involves an import pathway via receptors and the general import pore of the TOM complex

Porin, also termed the voltage-dependent anion channel, is the most abundant protein of the mitochondrial outer membrane. The process of import and assembly of the protein is known to be dependent on the surface receptor Tom20, but the requirement for other mitochondrial proteins remains controversial. We have used mitochondria from Neurospora crassa and Saccharomyces cerevisiae to analyze the import pathway of porin. Import of porin into isolated mitochondria in which the outer membrane has been opened is inhibited despite similar levels of Tom20 as in intact mitochondria. A matrix-destined precursor and the porin precursor compete for the same translocation sites in both normal mitochondria and mitochondria whose surface receptors have been removed, suggesting that both precursors utilize the general import pore. Using an assay established to monitor the assembly of in vitro-imported porin into preexisting porin complexes we have shown that besides Tom20, the biogenesis of porin depends on the central receptor Tom22, as well as Tom5 and Tom7 of the general import pore complex (translocase of the outer mitochondrial membrane [TOM] core complex). The characterization of two new mutant alleles of the essential pore protein Tom40 demonstrates that the import of porin also requires a functional Tom40. Moreover, the porin precursor can be cross-linked to Tom20, Tom22, and Tom40 on its import pathway. We conclude that import of porin does not proceed through the action of Tom20 alone, but requires an intact outer membrane and involves at least four more subunits of the TOM machinery, including the general import pore.


Import of F 1 β and porin into N . crassa mitochondria with…

Assembly of in vitro–imported porin…

Assembly of in vitro–imported porin into preexisting complexes. (A) Radiolabeled precursor of yeast…

Insertion of porin is inhibited…

Insertion of porin is inhibited by blocking the translocation pore with import intermediates.…

Porin forms high molecular weight…

Porin forms high molecular weight complexes in yeast. (A) Purified S . cerevisiae…

Import of porin requires components…

Import of porin requires components of the GIP. (A) Levels of Tom proteins…

Import of porin is not affected in tom40-3 mutant mitochondria. (A) The experiment…

Two new mutant alleles of…

Two new mutant alleles of TOM40 . (A) Predicted amino acid sequences (single…

Porin is in the vicinity of Tom20, Tom22, and Tom40 on its insertion…

Import of porin is inhibited…

Import of porin is inhibited into tom40 mutant mitochondria. (A) Radiolabeled porin precursor…


Aspartate transaminase catalyzes the interconversion of aspartate and α-ketoglutarate to oxaloacetate and glutamate.

L-Aspartate (Asp) + α-ketoglutarate ↔ oxaloacetate + L-glutamate (Glu)

As a prototypical transaminase, AST relies on PLP (Vitamin B6) as a cofactor to transfer the amino group from aspartate or glutamate to the corresponding ketoacid. In the process, the cofactor shuttles between PLP and the pyridoxamine phosphate (PMP) form. [6] The amino group transfer catalyzed by this enzyme is crucial in both amino acid degradation and biosynthesis. In amino acid degradation, following the conversion of α-ketoglutarate to glutamate, glutamate subsequently undergoes oxidative deamination to form ammonium ions, which are excreted as urea. In the reverse reaction, aspartate may be synthesized from oxaloacetate, which is a key intermediate in the citric acid cycle. [7]

Two isoenzymes are present in a wide variety of eukaryotes. In humans:

These isoenzymes are thought to have evolved from a common ancestral AST via gene duplication, and they share a sequence homology of approximately 45%. [8]

AST has also been found in a number of microorganisms, including E. coli, H. mediterranei, [9] and T. thermophilus. [10] In E. coli, the enzyme is encoded by the aspCgene and has also been shown to exhibit the activity of an aromatic-amino-acid transaminase (EC [11]

X-ray crystallography studies have been performed to determine the structure of aspartate transaminase from various sources, including chicken mitochondria, [12] pig heart cytosol, [13] and E. coli. [14] [15] Overall, the three-dimensional polypeptide structure for all species is quite similar. AST is dimeric, consisting of two identical subunits, each with approximately 400 amino acid residues and a molecular weight of approximately 45 kD. [8] Each subunit is composed of a large and a small domain, as well as a third domain consisting of the N-terminal residues 3-14 these few residues form a strand, which links and stabilizes the two subunits of the dimer. The large domain, which includes residues 48-325, binds the PLP cofactor via an aldimine linkage to the ε-amino group of Lys258. Other residues in this domain – Asp 222 and Tyr 225 – also interact with PLP via hydrogen bonding. The small domain consists of residues 15-47 and 326-410 and represents a flexible region that shifts the enzyme from an "open" to a "closed" conformation upon substrate binding. [12] [15] [16]

The two independent active sites are positioned near the interface between the two domains. Within each active site, a couple arginine residues are responsible for the enzyme's specificity for dicarboxylic acid substrates: Arg386 interacts with the substrate's proximal (α-)carboxylate group, while Arg292 complexes with the distal (side-chain) carboxylate. [12] [15]

In terms of secondary structure, AST contains both α and β elements. Each domain has a central sheet of β-strands with α-helices packed on either side.

Aspartate transaminase, as with all transaminases, operates via dual substrate recognition that is, it is able to recognize and selectively bind two amino acids (Asp and Glu) with different side-chains. [17] In either case, the transaminase reaction consists of two similar half-reactions that constitute what is referred to as a ping-pong mechanism. In the first half-reaction, amino acid 1 (e.g., L-Asp) reacts with the enzyme-PLP complex to generate ketoacid 1 (oxaloacetate) and the modified enzyme-PMP. In the second half-reaction, ketoacid 2 (α-ketoglutarate) reacts with enzyme-PMP to produce amino acid 2 (L-Glu), regenerating the original enzyme-PLP in the process. Formation of a racemic product (D-Glu) is very rare. [18]

The specific steps for the half-reaction of Enzyme-PLP + aspartate ⇌ Enzyme-PMP + oxaloacetate are as follows (see figure) the other half-reaction (not shown) proceeds in the reverse manner, with α-ketoglutarate as the substrate. [6] [7]

  1. Internal aldimine formation: First, the ε-amino group of Lys258 forms a Schiff base linkage with the aldehyde carbon to generate an internal aldimine.
  2. Transaldimination: The internal aldimine then becomes an external aldimine when the ε-amino group of Lys258 is displaced by the amino group of aspartate. This transaldimination reaction occurs via a nucleophilic attack by the deprotonated amino group of Asp and proceeds through a tetrahedral intermediate. As this point, the carboxylate groups of Asp are stabilized by the guanidinium groups of the enzyme's Arg386 and Arg 292 residues. formation: The hydrogen attached to the a-carbon of Asp is then abstracted (Lys258 is thought to be the proton acceptor) to form a quinonoid intermediate. formation: The quinonoid is reprotonated, but now at the aldehyde carbon, to form the ketimine intermediate.
  3. Ketimine hydrolysis: Finally, the ketimine is hydrolyzed to form PMP and oxaloacetate.

This mechanism is thought to have multiple partially rate-determining steps. [19] However, it has been shown that the substrate binding step (transaldimination) drives the catalytic reaction forward. [20]

AST is similar to alanine transaminase (ALT) in that both enzymes are associated with liver parenchymal cells. The difference is that ALT is found predominantly in the liver, with clinically negligible quantities found in the kidneys, heart, and skeletal muscle, while AST is found in the liver, heart (cardiac muscle), skeletal muscle, kidneys, brain, and red blood cells. [ citation needed ] As a result, ALT is a more specific indicator of liver inflammation than AST, as AST may be elevated also in diseases affecting other organs, such as myocardial infarction, acute pancreatitis, acute hemolytic anemia, severe burns, acute renal disease, musculoskeletal diseases, and trauma. [21]

AST was defined as a biochemical marker for the diagnosis of acute myocardial infarction in 1954. However, the use of AST for such a diagnosis is now redundant and has been superseded by the cardiac troponins. [22]

AST is commonly measured clinically as a part of diagnostic liver function tests, to determine liver health. However, it is important to keep in mind that the source of AST (and, to a lesser extent, ALT) in blood tests may reflect pathology in organs other than the liver. In fact, when the AST is higher than ALT, a muscle source of these enzymes should be considered. For example, muscle inflammation due to dermatomyositis may cause AST>ALT. This is a good reminder that AST and ALT are not good measures of liver function because they do not reliably reflect the synthetic ability of the liver and they may come from tissues other than liver (such as muscle).

Laboratory tests should always be interpreted using the reference range from the laboratory that performed the test. Example reference ranges are shown below:


The translocation of mitochondrial precursor proteins into and across the OM is an actively studied process (Pfanner et al., 2004 Neupert and Herrmann, 2007 Chacinska et al., 2009 Schmidt et al., 2010 Endo et al., 2011 Dimmer and Rapaport, 2012). However, in contrast to presequence-containing precursors, carriers, or β-barrel proteins, very little is known about the OM translocation of proteins that are targeted to the IMS via the dedicated oxidative folding pathway, MIA. Our in organello competition import assay between radiolabeled precursors and precursors in saturating amounts raised the possibility that MIA-dependent precursor proteins use an import route that is different from the classic one taken by presequence-containing precursor proteins. However, this discrete import pathway involves the Tom40 channel, because we were able to inhibit the entrance of MIA-dependent proteins with an alkylating agent that clogged the Tom40 channel.

Of importance, we demonstrated an in vivo interaction between a model MIA substrate, Mix17FLAG, and Tom40. The latter observation is interesting because no physical interaction between MIA-dependent precursor proteins and TOM or any other OM components has been reported. We were able to observe an intermediate stage in the transient and dynamic process of transiting across the OM, which may have two explanations. First, we applied an in vivo approach to express a precursor protein in the cell and follow its partners in the biogenesis using affinity purifications. Second, the choice of Mix17 as a model MIA-dependent protein may have advantages in monitoring rapid interactions during translocation through the OM. Mia40 serves as a specific and efficient receptor for its substrates on a trans side of the OM (Milenkovic et al., 2009 Sideris et al., 2009 von der Malsburg et al., 2011). This is likely preceded by a rapid interaction with the machinery that is responsible for the transfer of these precursor proteins to the trans side of the OM. Our model substrate, Mix17, belongs to the largest MIA-dependent proteins (Gabriel et al., 2007). Of interest, the twin CX9C motif is localized to the C-terminal end of Mix17 (Böttinger et al., 2012). These features may affect the speed of its translocation across the OM. Supporting this possibility, the formation of the Mia40-Mix17 intermediate that follows OM translocation is less effective compared with other MIA-dependent substrates (Böttinger et al., 2012). This, in turn, can favor the accumulation of earlier OM transport intermediates. A similar translocation intermediate is formed between Tom40 and Pet191, albeit with lower efficiency, which can be explained by its faster mitochondrial import, followed by more efficient recognition by Mia40. Thus we identified a transport intermediate of Mix17FLAG formed with Tom40 and subsequently showed that other MIA-dependent precursor proteins, such as Pet191, also formed this intermediate.

Of interest, we did not identify any other TOM or OM components in the translocation intermediate formed by Tom40 and the IMS-destined proteins. This was surprising because TIM23-dependent and presequence-containing proteins in transit interact with the entire TOM complex, including its core subunits, Tom22 and Tom5 (Dekker et al., 1997 Chacinska et al., 2003, 2010 Frazier et al., 2003 Tamura et al., 2009). Furthermore, various imported precursor proteins were purified using Tom22HIS (Chacinska et al., 2003, 2010 Wrobel et al., 2013). In agreement with the absence of Tom22 in the Tom40 translocation intermediate, the import of radiolabeled MIA-dependent precursor proteins into mitochondria that lack Tom22 but also Tom70 and Tom20 was unaffected. The minimized transport requirements were also reported previously for cytochrome c (Wiedemann et al., 2003).

The situation with Tom5 is different. The Tim9 requirement for Tom5 reported earlier (Kurz et al., 1999 Vögtle et al., 2012) was confirmed in the present experiments. However, the functional dependence on Tom5 seems not to be a universal feature of IMS-destined proteins. The import of a subfraction of MIA-dependent proteins, including Mix17, was unaffected in the absence of Tom5. Of importance, Tom5 was not present in the Tom40 translocation intermediates. On the basis of our data, we propose a scenario in which the function of Tom5 is indirectly needed on the stage of OM translocation. The absence of Tom5 may alter other, yet-unknown proteins involved in the recognition and transport of specific IMS proteins across the OM. Alternatively, mitochondria lacking Tom5 may also be impaired in oxidative folding reactions. This impairment would result in unproductive trapping of proteins in the IMS. Our results raise a possibility of existence of the altered TOM complex that does not contain all typical TOM subunits. Further, dynamics of the TOM machinery, that is, transient dissociation upon precursor binding, cannot be excluded. Finally, a putative different or more dynamic form of the Tom40 translocase may be preferentially formed in vivo. In summary, IMS-destined proteins cross the OM via a Tom40 translocase that is architecturally distinct from the Tom22-containing TOM complex.


The acquisition of a bacterial endosymbiont by a primitive host cell and its subsequent conversion into the mitochondrion marks one of the most important transitions in biology—the advent of the eukaryotic cell (Embley and Martin, 2006). A defining aspect of mitochondria is their capability to import proteins from the cytosol, a process that is driven by a set of characteristic protein translocases (Neupert and Herrmann, 2007 Lithgow and Schneider, 2010 Schmidt et al., 2010 Endo et al., 2011).

In the outer membrane we find two such translocases. The most important one—the general entry gate for essentially all imported proteins—is the translocase of the outer mitochondrial membrane (TOM). It consists of the pore-forming core subunit Tom40 (Hill et al., 1998 Ahting et al., 2001), a β-barrel–structured protein, protein import receptors (such as Tom70, Tom20, and Tom22 in yeast), and of a number of smaller subunits. The other translocase in the outer membrane is the sorting and assembly machinery (SAM), whose function is the insertion of β-barrel proteins. Its core subunit is Sam50, which is itself a β-barrel protein (Chacinska et al., 2009).

Extensive studies have revealed that most components and much of the architecture of the TOM and the SAM are conserved between yeast and mammals (Hoogenraad et al., 2002 Dolezal et al., 2006 Schneider et al., 2008 Hewitt et al., 2011). However a more global analysis shows that only the core components of the mitochondrial protein translocases are conserved in all eukaryotes. Among these, Sam50 is the most highly conserved: it is found in all groups of eukaryotes and even has a bacterial orthologue, the Omp85-like protein BamA, that functions in inserting β-barrel proteins into the outer bacterial membrane (Bos et al., 2007). Tom40 is also conserved and found in virtually all eukaryotes however, although the β-barrel structure of Tom40 would indicate a bacterial ancestry, no bacterial orthologue has been identified (Dolezal et al., 2006 Zeth, 2010).

The receptor subunits of the TOM complex are much less conserved. Thus, in plants Tom22 is severely truncated, and a conventional Tom20 is absent (Mac´asev et al., 2004). Instead we find a structural analogue of Tom20 that is coded in reverse, suggesting that the structural similarity between the mammalian and fungal Tom20 on one side and the plant Tom20 on the other side is due to convergent evolution and not to common descent (Perry et al., 2006). Tom70 appears to be missing in plants (Chan et al., 2006) but has recently been detected in some representatives of the Chromalveolata, which are unrelated to fungi, mammals, and plants, indicating that Tom70 might be more widespread than previously thought (Tsaousis et al., 2011).

Recent studies in a number of different protozoa revealed interesting but for the most part minor deviations to the canonical protein import systems described in yeast and mammals (Dolezal et al., 2005). It was therefore a surprise that trypanosomatids have a fundamentally different mitochondrial outer membrane translocation machinery. Trypanosomatids lack Tom40 (Pusnik et al., 2009) and instead employ an outer membrane protein translocation channel termed ATOM, for archaic translocase of the outer mitochondrial membrane (Pusnik et al., 2011). Unlike Tom40, ATOM has a bacterial orthologue. It shows similarities to a subgroup of the bacterial Omp85-like protein family that is distinct from the Sam50 orthologue BamA. This suggests that trypanosomatids have retained an archaic outer membrane protein import system that might have preceded the Tom40-based protein import channel.

The discovery of ATOM provided unexpected new insights into the evolution of the mitochondrial outer membrane protein import system and the evolution of eukaryotes in general. Moreover, it raised the question of what other factors might be required for protein translocation across the outer membrane of trypanosomal mitochondria. Using Trypanosoma brucei as an experimental model system, we discovered such a novel factor. This protein is an essential mitochondrial outer membrane protein that is required for import of a large subset but not all matrix proteins and therefore might have a receptor-like function.

How do shuttle mechanisms differ from one another?

One carrier system that has been extensively studied in insect flight muscle is the glycerol–phosphate shuttle. This mechanism uses the presence on the outer face of the inner mitochondrial membrane of an FAD-dependent enzyme that oxidizes glycerol phosphate.

The glycerol phosphate is produced by the reduction of dihydroxyacetone phosphate in the course of the reaction, NADH is oxidized to NAD + . In this reaction, the oxidizing agent (which is itself reduced) is FAD, and the product is FADH2 (Figure 20.23). The FADH2 then passes electrons through the electron transport chain, leading to the production of 1.5 moles of ATP for each mole of cytosolic NADH. This mechanism has also been observed in mammalian muscle and brain.

A more complex and more efficient shuttle mechanism is the malate–aspartateshuttle, which has been found in mammalian kidney, liver, and heart. Thisshuttle uses the fact that malate can cross the mitochondrial membrane, while oxaloacetate cannot. The noteworthy point about this shuttle mechanism is that the transfer of electrons from NADH in the cytosol produces NADH in the mitochondrion. In the cytosol, oxaloacetate is reduced to malate by the cyto-solic malate dehydrogenase, accompanied by the oxidation of cytosolic NADH to NAD + (Figure 20.24). The malate then crosses the mitochondrial mem-brane. In the mitochondrion, the conversion of malate back to oxaloacetate is catalyzed by the mitochondrial malate dehydrogenase (one of the enzymes of the citric acid cycle). Oxaloacetate is converted to aspartate, which can also cross the mitochondrial membrane. Aspartate is converted to oxaloacetate in the cytosol, completing the cycle of reactions.

The NADH that is produced in the mitochondrion thus passes electrons to the electron transport chain. With the malate–aspartate shuttle, 2.5 moles of ATP are produced for each mole of cytosolic NADH rather than 1.5 moles of ATP in the glycerol–phosphate shuttle, which uses FADH2 as a carrier.


Integral membrane β-barrel proteins are found in the OMs of Gram-negative bacteria and mitochondria. Their assembly requires an evolutionarily conserved protein of the Omp85 family, but significant adaptations of the assembly machineries have occurred during evolution. The accessory proteins, that is, four lipoproteins in the case of bacteria and two cytosol-exposed membrane-associated proteins in mitochondria, do not show any sequence similarity. Also, the Omp85-like component adapted during evolution besides a membrane-embedded β-barrel domain, the bacterial proteins contain five periplasm-exposed POTRA domains, whereas the mitochondrial variant contains only one such domain. Of note, a single POTRA domain was recently demonstrated to be sufficient for Omp85 function in N. meningitidis ( Bos et al. 2007b), but in E. coli BamA, three of these domains were essential although also the other two were important for optimal functioning ( Kim et al. 2007).

Not only the assembly machineries but also the signals they recognize underwent adaptations during evolution. In addition, these signals are recognized by different components of the machinery, that is, by Omp85 in the bacterial system ( Robert et al. 2006) and by Tob38 in the mitochondrial system ( Kutik et al. 2008). We recently demonstrated that, despite these variations, the primordial bacterial OMP signature sequence is still properly recognized and processed by the mitochondrial machinery ( Walther et al. 2009), suggesting that the basic mechanism of β-barrel protein assembly might be evolutionarily conserved. To test this hypothesis, we investigated here whether a mitochondrial OMP, VDAC, can be assembled into the OM of bacteria, which was indeed the case, and it was shown to be dependent on a functional Bam complex. Of note, the structure of VDAC, that is, a 19-stranded β-barrel ( Bayrhuber et al. 2008 Hiller et al. 2008 Ujwal et al. 2008), deviates from those of bacterial OMPs, which all contain an even number of β-strands ( Koebnik et al. 2000). However, this difference apparently does not prevent the incorporation of VDAC into the bacterial OM. Consistent with this observation, a mutant form of porin PhoE lacking the N-terminal transmembrane β-strand was previously reported to be functionally incorporated into the E. coli OM ( Bosch et al. 1988), demonstrating that, indeed, the bacterial Bam machinery can deal with β-barrels with an odd number of strands.

Efficient assembly of VDAC into the bacterial OM was dependent on its β-signal, which is also required for its assembly in the mitochondrial OM ( Kutik et al. 2008). Thus, the bacterial system is able to decode the evolved signals in the mitochondrial OMPs even though they are recognized in the authentic mitochondrial system by a component that has no equivalent in the bacterial system. At very high-expression levels, a mutant form of VDAC that lacks five C-terminal amino acid residues and is thereby affected in its β-signal was undetectable at the bacterial cell surface in immunofluorescence microscopy experiments ( fig. 3B). However, at lower expression levels, at least some of the total quantity of the mutant protein produced appeared to be incorporated into the OM ( figs. 4A and 5A), demonstrating that an intact β-signal is not absolutely essential for OM assembly of VDAC in E. coli. Similarly, an intact C-terminal signature sequence is not absolutely essential for the correct assembly of a bacterial OMP into the OM. At high-expression conditions, a mutant form of porin PhoE lacking the C-terminal Phe residue was not incorporated at all in the E. coli OM, but formed dense periplasmic aggregates that pellet with the membrane fraction during centrifugation and could be visualized in the cell by electron microscopy ( Struyvé et al. 1991 de Cock et al. 1997). However, at lower expression conditions, this mutant protein was assembled into the OM ( de Cock et al. 1997). Presumably, the lower aggregation kinetics at low-expression conditions increases the time span for the Bam machinery to deal with substrates with an imperfect recognition signal. Similarly, when the bacterial OMP PhoE was highly expressed in yeast, it accumulated as unfolded aggregates in the mitochondria, whereas it was efficiently assembled into the mitochondrial OM at lower expression conditions ( Walther et al. 2009).

Also, the lipid composition could potentially affect the assembly of integral membrane proteins. Indeed, LPS, an abundant lipid component of the bacterial OM, has been implicated in the biogenesis of OMPs in E. coli ( Bos et al. 2007a), although viable LPS-deficient mutants of N. meningitidis have been described in which OMP assembly appears unaffected ( Steeghs et al. 2001). The lipid composition of the bacterial OM is, by the presence of LPS and the absence of sterols, entirely different from that of the mitochondrial OM, but this difference is apparently not an insurmountable impediment to the assembly of VDAC into the bacterial OM.

In conclusion, it appears that signal recognition is flexible enough to allow for assembly of a mitochondrial OMP into the bacterial OM. Together with the previous observation that bacterial OMPs expressed in yeast are assembled into the mitochondrial OM ( Walther et al. 2009), these data allow the conclusion that the basic mechanism of β-barrel protein assembly is evolutionarily conserved.