Metabolic Implications of Non-Electrogenic ATP/ADP Exchange in Cancer Cells: A Mechanistic Basis for the Warburg Effect

Abstract

In post-mitotic cells, mitochondrial ATP/ADP exchange occurs by the adenine nucleotide translocator (ANT). Driven by membrane potential (ΔΨ), ANT catalyzes electrogenic exchange of ATP⁴⁻ for ADP³⁻, leading to higher ATP/ADP ratios in the cytosol than mitochondria. In cancer cells, ATP/ADP exchange occurs not by ANT but likely via the non-electrogenic ATP-Mg/phosphate carrier. Consequences of non-electrogenic exchange are: 1) Cytosolic ATP/ADP decreases to stimulate aerobic glycolysis. 2) Without proton utilization for exchange, ATP/O increases by 35% for complete glucose oxidation. 3) Decreased cytosolic ATP/ADP•Pi increases NAD(P)H/NAD(P)⁺. Increased NADH increases lactate/pyruvate, and increased NADPH promotes anabolic metabolism. Fourth, increased mitochondrial NADH/NAD⁺ magnifies the redox span across Complexes I and III, which increases ΔΨ, reactive oxygen species generation, and susceptibility to ferroptosis. 5) Increased mitochondrial NADPH/NADP⁺ favors a reverse isocitrate dehydrogenase-2 reaction with citrate accumulation and export for biomass formation. Consequently, 2-oxoglutarate formation occurs largely via oxidation of glutamine, the preferred respiratory substrate of cancer cells. Overall, non-electrogenic ATP/ADP exchange promotes aerobic glycolysis (Warburg effect) and confers growth advantages to cancer cells.

Introduction

As famously first described by Otto Warburg almost a century ago, proliferating tumor cells display a different pattern of bioenergetic metabolism compared to ‘normal’ postmitotic cells [1–3]. Specifically, tumor cells produce lactate from glucose under aerobic conditions, whereas normal tissues do not, a phenomenon named the ‘Warburg effect’ by Efraim Racker in 1972, two years after Warburg’s death [4]. Warburg proposed that damage to what we now know as mitochondrial oxidative phosphorylation accounted for this ‘aerobic glycolysis’, but more modern studies show that oxidative phosphorylation by isolated tumor mitochondria is well coupled and seemingly completely intact [5, 6]. Nonetheless, overall mitochondrial metabolism is suppressed within intact tumor cells, such that glycolysis seems to account for half or more of all ATP production versus less than 10% in aerobic postmitotic tissues [7, 8].

The advantages of the Warburg metabolic phenotype for tumor growth and proliferation have been widely discussed [7, 9, 10]. Glycolysis may simply generate ATP more rapidly than oxidative phosphorylation and do so with less protein machinery and with no need for complex membranous organelles like mitochondria. Glycolysis may also provide precursor molecules for diversion into various biosynthetic pathways for biomass formation. Growth of tumors typically outpaces that of their blood supply, and glycolysis becomes a useful adaptation to the resulting hypoxia. Other theories suggest that aerobic glycolysis promotes immune cell evasion, a protective microenvironment, changes of cell signaling, or other beneficial effect for cancer cell proliferation and survival. However, none of these explanations provides a compelling and unequivocal rationale for the Warburg metabolic phenotype (see [9, 10]). For example, ATP production does not seem rate-limiting for cell proliferation [10, 11]. Additionally, S. cerevisiae grow robustly in non-fermentable (non-glycolytic) glycerol-containing medium, which shows that intermediates from glycolytic flux are not obligatory for cell proliferation and growth [12]. Furthermore, non-cancer cells without the Warburg metabolic phenotype adapt rapidly to hypoxia/anoxia with a very vigorous upregulation of glycolytic flux that protects against cell death [13, 14]. Lastly, given the enormous plasticity of signaling pathways, immune responses, and adaptive mechanisms, the suggestion that cancer cells are obligatorily limited or enabled by a single factor like glycolysis seems implausible.

A second remarkable metabolic feature of cancer cells is the need for 10 to 100-times more glutamine than any other amino acid for growth, as first shown in cultured HeLa cells 65 years ago by Harry Eagle [15]. The transcription factor and proto-oncogene, c-myc, upregulates numerous enzymes involved in glutamine metabolism and transport, but less understood is why transformed cells become ‘addicted’ to glutamine in comparison to post-mitotic cells [16–18]. Such addicted cells undergo apoptosis or otherwise fail to survive after glutamine limitation or inhibition of glutamine metabolism. How this glutamine addiction is related to the Warburg metabolic phenotype of aerobic glycolysis remains poorly understood.

The Adenine Nucleotide Translocator Catalyzes Mitochondrial Exchange of ATP for ADP

As discovered and characterized by the laboratories of Martin Klingenberg and Pierre Vignais beginning in the 1960’s, the adenine nucleotide translocator (ANT), also known as the ATP/ADP carrier, is a 30 kDa protein in the mitochondrial inner membrane that catalyzes the exchange of ADP for ATP [19–23]. Essential for the overall process of oxidative phosphorylation in eukaryotes, ANT is the most abundant protein in the inner membrane on a molar basis. A member of the solute carrier family (SLC), ANT in humans comprises isoforms ANT1, ANT2, ANT3 and ANT4 that are encoded, respectively, by SLC25A4, SLC25A5, SLC25A6 and SLC25A31. ANT1 is expressed predominantly in post-mitotic differentiated tissues, especially heart and skeletal muscle. ANT2 is highly expressed in proliferating tissues, including cancer cells. ANT3 is ubiquitously expressed, and ANT4 is expressed in testis. Only 3 isoforms are found in mouse (ANT1, 2, and 4) [24–28]. Long considered to be a homodimeric membrane protein, more recent studies using a variety of approaches (e.g., crystallography, differential affinity purification, negative dominance) show that ANT exists and functions as a monomer in the inner membrane [29, 30].

The Adenine Nucleotide Translocator is Electrogenic and Drives a High Cytosolic ATP/ADP Ratio

Importantly for this discussion, ANT catalyzes the electrogenic exchange of ATP⁴⁻ for ADP³⁻. In this way ATP/ADP exchange is coupled to the mitochondrial membrane potential (ΔΨ) (see Fig. 1B). Thus, a negative inside ΔΨ drives mitochondrial release of ATP⁴⁻ in exchange for uptake of ADP³⁻. In cells from tissues like heart, brain, liver, and kidney, ATP⁴⁻/ADP³⁻ exchange can approach near equilibrium with ΔΨ. Since each 59 mV of ΔΨ drives a 10-fold concentration gradient, electrogenic ANT drives a cytosolic ATP/ADP ratio that is up to 100-times greater than that of the matrix at a mitochondrial ΔΨ of ~−120 mV, which is typical for these cells [31–34].

Figure 1. Schemes of mitochondrial adenine nucleotide uptake and exchange of ATP for ADP and Pi.

(A) In net uptake of ATP, Mg⁺²ATP⁻⁴ enters mitochondria through AMPC in exchange for HPO₄⁻². HPO₄⁻² protonates to H₂PO₄⁻ (Pi⁻), which returns to the mitochondrial matrix in exchange for OH⁻ via the phosphate carrier (formally equivalent to symport with H⁺, as depicted). H⁺ entering mitochondria can exchange for Na⁺ via the Na⁺/H⁺ exchanger (NHE) or for K⁺ via the K⁺/H⁺ exchanger. (B) In electrogenic exchange (left), extramitochondrial ADP³⁻, Pi⁻ and H⁺ exchange for intramitochondrial ATP⁴⁻ mediated by ANT and the phosphate carrier (PC) and driven by Δp. Electrogenic ATP/ADP exchange increases cytosolic ATP/ADP and amplifies cytosolic ΔG_P relative to mitochondrial ΔG_P, which inhibits glycolysis. In non-electrogenic exchange (right), AMPC and the phosphate carrier mediate exchange of cytosolic H⁺ADP⁻³, Pi⁻ and H⁺ for mitochondrial Mg⁺²ATP⁻⁴. Mg²⁺ and 2 H⁺ entering mitochondria return via the sodium-magnesium exchanger (NME) and NHE. Since non-electrogenic exchange does not amplify cytosolic ΔG_P relative to mitochondrial ΔG_P, glycolysis is stimulated.

The phosphorylation potential (ΔG_P) is the free energy required for ATP synthesis from ADP and Pi or, equivalently, the free energy released by ATP hydrolysis to ADP and Pi:

$$\Delta {\text{G}}_{\text{P}}=\Delta {\text{G}}_{\text{P}}°\prime +2.3\text{RT}\log ([\text{ATP}]/[\text{ADP}][\text{Pi}]$$

where ΔG_P°′ is the standard free energy change at prevailing conditions, primarily pH, Mg²⁺, and ionic strength. Because of electrogenic ATP/ADP exchange, ΔG_P is greater by ~12 kJ/mol in the cytosol compared to matrix [19, 33, 35, 36]. Driven by mitochondrial ΔpH, mitochondrial uptake of Pi⁻ (= H₂PO₄⁻) in exchange for OH⁻ via the phosphate carrier, a reaction that is formally equivalent to symport of Pi⁻ and H⁺ as depicted in Fig. 1, also contributes to ΔG_P amplification in the cytosol relative to mitochondria, but to a lesser extent, since ΔpH (~0.8 pH units) makes a smaller contribution to overall protonmotive force (Δp = ΔΨ – 59ΔpH) than ΔΨ. For processes like ion transport and muscle contraction that come to near equilibrium with ATP hydrolysis, the greater cytosolic ΔG_P produced by electrogenic ATP/ADP exchange translates to larger ion gradients, membrane potentials and contractile forces, among other benefits, which is a distinct advantage for eukaryotic organisms.

The Electroneutral ATP-Mg/Phosphate Carrier Mediates Net Uptake of Adenine Nucleotide into Mitochondria

ANT catalyzes an obligatory exchange of ADP for ATP. Consequently, ANT cannot catalyze the net uptake of adenine nucleotides, as required for mitochondrial biogenesis which must occur during cell proliferation, replacement of mitochondria turning over every 10 to 25 days even in post-mitotic differentiated tissues, and recovery after injury [37–39]. The pathway for net uptake of ATP by mitochondria is the ATP/Mg-Pi carrier (AMPC), which catalyzes the reversible electroneutral (non-electrogenic) exchange of Mg²⁺-ATP⁴⁻ for HPO₄²⁻ stimulated by cytosolic Ca²⁺ [40–42] (Fig. 1A). Since HPO₄²⁻ + H⁺ can reenter mitochondria via the phosphate carrier, AMPC facilitates net accumulation of ATP into mitochondria. AMPC also catalyzes electroneutral exchange of divalent anionic H⁺ADP³⁻ for Mg²⁺-ATP⁴⁻ or for HPO₄²⁻. Notably, mitochondrial adenine nucleotide loading by AMPC also catalyzes Mg²⁺ uptake. Mg²⁺ at near millimolar concentrations is required for the mitochondrial F₁F_O-ATP synthase and numerous other mitochondrial enzymes [43–45]. Linked to the phosphate carrier, the overall process of Mg-ATP uptake is driven by ΔpH. H⁺ entering mitochondria may then promote Na⁺ and K⁺ uptake by the Na⁺/H⁺ and K⁺/H⁺ exchangers to help provide matrix osmolytes as mitochondrial volume increases during biogenesis.

AMPC is encoded by five paralogues in mammals − SLC25A23, SLC25A24, SLC25A25, SLC25A54 and SLC25A41, the last a calcium-independent AMPC [41, 46–48]. SLC25A24, the dominant isoform, is overexpressed by as much as 20-fold in several cancer cell lines and in more than 60 tumor types compared to normal tissue counterparts [49]. SLC25A24 expression also increases in regenerating limbs of salamanders, suggesting that SLC25A24 upregulation is a common feature of proliferating cells, both cancerous and non-cancerous [50].

The function of AMPC has generally been considered in the context of mitochondrial biogenesis where net uptake of adenine nucleotide is needed. Additionally, AMPC together with the phosphate carrier can catalyze electroneutral exchange of ATP for ADP and Pi. Via AMPC, Mg²⁺-ATP⁴⁻ moves out in exchange for entry of H⁺ADP³⁻, whereas Pi⁻ moves in through the phosphate carrier together with H⁺, as depicted in Fig. 1B. The combined action of AMPC and the phosphate carrier also moves Mg²⁺ out of and the equivalent of 2 H⁺ into the mitochondrial matrix. In the overall exchange, these ions must return, as most likely achieved through concerted Mg²⁺/2Na⁺ and Na⁺/H⁺ exchange (Fig. 1) [51].

Mitochondrial ATP/ADP Exchange Occurs Independently of the Adenine Nucleotide Translocator in Cancer Cells

Oligomycin but not bongkrekic acid or carboxyatractyloside inhibits basal respiration in cancer cells.

In isolated mitochondria incubated with Pi and respiratory substrate (e.g., succinate, glutamate plus malate, 2-oxoglutarate), ADP markedly stimulates respiration due to ATP formation by oxidative phosphorylation. As ADP is exhausted due to conversion to ATP by oxidative phosphorylation, respiration slows to its original rate. As originally defined by Chance and Williams, ADP-stimulated respiration is State 3, whereas respiration in the absence of ADP (or other phosphate acceptor like AMP) is State 4 [52]. The ratio of State 3 to State 4 respiration is the respiratory control ratio, which is typically 5 to 8 for well coupled mitochondria. Oligomycin by inhibiting the F₁F_O-ATP synthase restores State 4 respiration even in the presence of ADP. In isolated mitochondria, the high affinity ANT inhibitors, carboxyatractyloside (CAT) and bongkrekic acid (BA), similarly inhibit State 3 respiration and restore State 4 respiration [19, 33, 53–55].

In intact cells, oligomycin also inhibits respiration. Likewise, in primary cultured hepatocytes, CAT and BA inhibit respiration to an identical degree as oligomycin [56]. Unexpectedly in several different cancer cell lines, however, CAT and BA do not inhibit respiration under conditions where oligomycin produces ~50% inhibition, as reported by independent laboratories, and which is illustrated for A549 lung adenocarcinoma cells in Fig. 2A [56, 57]. Remarkably, plasma membrane permeabilization with digitonin restores the ability of CAT and BA to inhibit ADP-stimulated respiration. Such findings might suggest that CAT and BA do not permeate the plasma membrane. BA, however, is highly membrane-permeant and must permeate mitochondrial inner membranes to bind to ANT from the matrix side to cause inhibition. By contrast, CAT binds ANT from the cytosolic side of the inner membrane and is more hydrophilic and less membrane-permeant than BA. Nonetheless, at high concentrations in comparison to their IC₅₀ for ANT inhibition in isolated mitochondria, both CAT and BA enter cells to induce and inhibit, respectively, apoptosis via up and down-regulation of the mitochondrial permeability transition (MPT), since ANT is a putative component of MPT pores whose activity is up and down regulated, respectively, by CAT and BA [58, 59]. Overall, the observation that oligomycin but not CAT and BA inhibits basal respiration was the first to suggest that mitochondrial adenine nucleotide movement occurs by a pathway other than ANT in cancer cells [56].

Figure 2. Mitochondrial ATP transport occurs independently of the adenine nucleotide translocator in cancer cells.

(A) Inhibition of respiration in A549 cancer cells by vehicle (VEH), oligomycin (OLIG), BA, and CAT. (B) Mitochondrial depolarization by BA and CAT in respiration-inhibited cultured rat hepatocytes. Mitochondrial ΔΨ was monitored by tetramethylrhodamine methylester (TMRM) fluorescence whose intensity is pseudocolored (see reference bar, upper right). Note that TMRM fluorescence did not decrease after the respiratory inhibitor myxothiazol (MYX) but did decrease markedly after subsequent addition of either OLIG or BA. (C) Absence of mitochondrial depolarization by BA and CAT in respiration-inhibited A549 cancer cells. Note that BA and CAT addition after MYX did not cause a loss of TMRM fluorescence. Subsequent OLIG, by contrast, markedly decreased TMRM fluorescence. (D) Preservation of mitochondrial ΔΨ after respiratory inhibition in ANT2,3 double knockdown A549 cells. Cells were transfected with non-target siRNA (upper panels) or siRNA against both ANT2 and ANT3 (lower panels). Note that TMRM fluorescence was unchanged after MYX in both targeted and non-targeted siRNA-treated cells. Subsequent addition of 2-deoxyglucose (2DG) markedly decreased fluorescence. Adapted from [56].

After respiratory inhibition, oligomycin and 2-deoxyglucose but not bongkrekic acid and carboxyatractyloside decrease mitochondrial membrane potential in cancer cells.

After respiratory inhibition, hydrolysis of ATP by the F₁F_O-ATP synthase working in reverse drives proton translocation and formation of a mitochondrial ΔΨ in both intact cells and isolated mitochondria. For intact cells, the principal source of ATP is glycolysis in the cytosol, which means that cytosolic ATP must move across the inner membrane to gain access to the ATP synthase. For example, hepatocytes maintain a robust mitochondrial ΔΨ after treatment with myxothiazol, a Complex III respiratory inhibitor, but depolarization then occurs rapidly after subsequent addition of oligomycin, as visualized by loss of the ΔΨ-indicating fluorophore, tetramethylrhodamine methylester (TMRM) (Fig. 2B, upper panels). Likewise, cancer cells maintain a high mitochondrial ΔΨ after myxothiazol, which declines rapidly after oligomycin (see Fig. 2C). In hepatocytes, both CAT and BA act like oligomycin in producing rapid depolarization after myxothiazol treatment (Fig. 2B, lower panels and not shown). In marked contrast, BA and CAT have no effect in four different cancer cell lines after an identical treatment, as shown for A549 cells in Fig. 2C. Nonetheless, subsequent exposure to oligomycin causes rapid depolarization of the myxothiazol-treated cancer cells. Treatment with the glycolytic inhibitor 2-deoxyglucose after myxothiazol also leads to mitochondrial depolarization and signifies that glycolysis is the source of the ATP that sustains ΔΨ generation after respiratory inhibition (see Fig. 2D) [56].

Knockdown of ANT2 and ANT3 does not prevent mitochondrial import of glycolytic ATP in cancer cells.

Genetic interventions also indicate that ANT does not mediate mitochondrial ATP/ADP exchange in cancer cells. A549 cells express only the ANT2 and ANT3 isoforms. After use of siRNA to knockdown mRNA and protein expression of both ANT2 and ANT3 by more than 90% (double knockdown), mitochondria of A549 cells polarize to virtually the same extent as cells treated with non-target siRNA and remain polarized after myxothiazol (Fig. 2D) [56]. Moreover, 2-deoxyglucose then causes depolarization, confirming that glycolytic ATP is entering mitochondria to sustain ΔΨ. These findings confirm the conclusion that glycolytic ATP enters mitochondria by a pathway that is not ANT2 or ANT3 [56].

The Electroneutral ATP-Mg/Phosphate Carrier May Mediate Mitochondrial ATP/ADP Exchange in Cancer Cells

As discussed above, mitochondria have two pathways for movement of ATP and ADP across the inner membrane: ANT and AMPC. Unlike ANT, AMPC can catalyze net uptake of adenine nucleotides as required for mitochondrial biogenesis. During cell proliferation, mitochondrial biogenesis must increase to match the rate of cell division. Accordingly, proliferating cancer cells and likely other proliferating cells overexpress AMPC [49, 50]. However, high expression of both ANT and AMPC leads to futile cycling of ATP and ADP with dissipation of Δp, effectively an uncoupling effect. In this context, suppression of ANT activity may be beneficial when AMPC is upregulated. How ANT is suppressed remains unknown, but a soluble cytosolic factor seems involved based on the recovery of sensitivity to CAT and BA after plasma membrane permeabilization [56].

Decreased Cytosolic ATP/ADP Due to Non-Electrogenic ATP/ADP Exchange Stimulates Glycolysis

The switch from electrogenic ANT to non-electrogenic AMPC in cancer cells has several important metabolic consequences. The first consequence is that cytosolic ATP/ADP ratios are lower by a factor of 10 or more in cancer cells using AMPC instead of ANT for mitochondrial adenine nucleotide exchange (Fig. 1). Such lower ATP/ADP ratios stimulate glycolysis directly and account for the aerobic glycolysis and net production of lactate that is characteristic of the Warburg metabolic phenotype [56]. With the switch to non-electrogenic ATP/ADP exchange, moreover, less stimulation of mitochondrial oxidative phosphorylation occurs as ADP increases, which also contributes to the greater relative contribution of glycolysis to overall ATP production. Arguably, glycolysis in cancer cells is activated the ‘old fashioned’ way, namely by increased ADP and decreased ATP, but this does not exclude other adaptive changes in expression of glycolytic enzymes and other proteins to accommodate an increased demand for glycolysis. Indeed, metabolic reprogramming is a fundamental characteristic of malignancy [60].

Decreased ATP/ADP ratios likely do not impair growth and proliferation of cancer cells. In eukaryotes, the pathways and energetics for protein and nucleic acid synthesis are largely unchanged from their prokaryotic ancestors. In bacteria like E. coli, ATP/ADP ratios are in the range of 3 to 10, substantially less than in the cytosol of post-mitotic eukaryotic cells, but like ATP/ADP ratios in the mitochondrial matrix [61–64]. Nonetheless, growth, proliferation and biomass formation by E. coli are robust.

Non-Electrogenic ATP/ADP Exchange Increases the ATP Yield of Oxidative Phosphorylation

During mitochondrial respiration, Complexes I, III and IV pump 10 protons from the mitochondrial matrix into the intermembrane space as NADH is oxidized to NAD⁺ and one atom of oxygen is reduced to water (see [33]). As protons re-enter the matrix through the mitochondrial F₁F_O-ATP synthase, hydrophilic F₁ rotates relative to membranous hydrophobic F_O in the enzyme complex. With each full rotation, 3 molecules of ATP are formed and released into the matrix, whereas rotation in the opposite direction leads to ATP hydrolysis. Rotation in one direction (ATP synthesis) versus the other (ATP hydrolysis) depends on the thermodynamic driving forces of Δp relative to ΔG_P. Since the c-ring of F₁ comprises 8 subunits in vertebrate mitochondria and F₁ is trimeric for its major α and β subunits, the prevailing view is that one full rotation of F₁ relative to F_O allows entry of 8 H⁺ through the synthase to drive synthesis of 3 ATP molecules [65, 66]. Thus, the H⁺/ATP stoichiometry of matrix ATP synthesis is 8/3 or 2⅔.

Intramitochondrial ATP so formed exchanges for extramitochondrial (cytosolic) ADP and Pi, which ordinarily occurs by ANT and the phosphate carrier (Fig. 1). Together, exchange of intramitochondrial ATP for cytosolic ADP and Pi is coupled to the inward movement of 1 H⁺, which increases the H⁺/ATP stoichiometry from 2⅔ to 3⅔ for the overall mitochondrial synthesis and release to the cytosol of ATP. Consequently, the ATP/O stoichiometry for oxidation of NADH is 2.73 (10 H⁺/NADH oxidized divided by 3⅔ H⁺/ATP). This ATP/O is an ideal number, since some proton leakage (uncoupling) always occurs, but tightly coupled mitochondria approach this efficiency when synthesizing ATP maximally [67, 68]. Taking into account H⁺ driving the malate/aspartate shuttle to bring in reducing equivalents from glycolytically generated NADH, as well as the cytosolic ATP equivalence of substrate level phosphorylation of GDP to GTP, the total yield of ATP released to the cytosol from complete oxidation of glucose to CO₂ and water is ideally 33.45 (Table 1) [33, 66].

Table 1. Cytosolic ATP/glucose stoichiometries for complete glucose oxidation using electrogenic versus non-electrogenic mitochondrial ATP/ADP exchange.

Shown are ATP yields for complete glucose oxidation to CO₂ and H₂O by glycolysis, the TCA cycle, and the respiratory chain. Cytosolic (cyto) glucose metabolism yields two molecules of ATP, NADH and pyruvate by the glycolytic pathway. Reducing equivalents of cytosolic NADH enter mitochondria by the malate/aspartate shuttle with the cost of 1 H⁺ moving into mitochondria for each NADH. Each pyruvate entering mitochondria yields 4 NADH, 1 succinate and 1 GTP via the TCA cycle. Respiratory oxidation of NADH and succinate translocates 10 and 6 H⁺, respectively, across the inner membrane to generate Δp. ATP synthesis and release to the cytosol requires 3⅔ H⁺ with electrogenic ATP/ADP exchange and 2⅔ H⁺ with non-electrogenic exchange. GTP equilibrates with the mitochondrial ATP pool by the nucleoside-diphosphate kinase reaction.

Cytosolic ATP Generated by Glucose Oxidation
		ATP Yield from Glucose
Intermediate Metabolite	Protons Translocated	Electrogenic (3⅔ H+/ATP)	Non-Electrogenic (2⅔ H+/ATP)
Glycolytic ATP	--	2	2
Cyto NADH × 2 × 9H+	18	4.91	6.75
Mito NADH × 2 × 4 × 10H+	80	21.82	30
Succinate × 2 × 6H+	12	3.27	4.5
GTP × 2	5⅓	1.45	2
TOTAL	115.33	33.45	45.25

When non-electrogenic AMPC replaces ANT, the ATP/O stoichiometry for NADH oxidation increases. Without a proton devoted to driving ATP, ADP, and Pi movement between mitochondria and cytosol, ATP/O rises to 3.75 for NADH-linked substrates, and the ATP yield from complete glucose oxidation increases from 33.45 to 45.25 (Table 1). Thus, the switch from electrogenic ANT to non-electrogenic AMPC results in an increase of ATP yield by 35%, a considerable advantage for proliferating cancer cells. This also means that cancer cells are correspondingly more reliant on mitochondrial oxidative phosphorylation that previous estimates would suggest.

Decreased Cytosolic ATP/ADP Leads to a More Reductive Intracellular Milieu Favoring Lactate Formation from Pyruvate

In the steady state for many metabolic pathways, individual reactions proceed to near equilibrium, whereas others are removed from equilibrium and often rate-controlling. For example, mitochondrial Δp is in near equilibrium with both intramitochondrial ΔG_P, representing the ATP synthase reaction, and the redox free energy change between NADH and cytochrome c (ΔG_RI-III), representing the oxidation-reduction reactions catalyzed by proton-pumping respiratory Complexes I and III [33, 69]. By contrast, the oxidation of cytochrome c by oxygen catalyzed by Complex IV is far from equilibrium and irreversible. Likewise in glycolysis, ATP formation from ADP and Pi via the glyceraldehyde phosphate dehydrogenase and 3-phosphoglycerate kinase reactions is in near equilibrium with NADH/NAD⁺ [19, 70]. Importantly, the proportionality of [ATP]/[ADP][Pi] to NADH/NAD⁺ is different in the cytosol versus the mitochondrial matrix. In mitochondria, [ATP]/[ADP][Pi] is linearly related to NADH/NAD⁺, whereas in the cytosol [ATP]/[ADP][Pi] is inversely proportional to NADH/NAD⁺. Thus, the decrease of cytosolic [ATP]/[ADP][Pi] after the switch from electrogenic to non-electrogenic ATP/ADP exchange leads to an increase of cytosolic NADH/NAD⁺ in proliferating cancer cells (Fig. 3). Moreover, the lactate dehydrogenase reaction is also at near-equilibrium. As a consequence, increased cytosolic NADH/NAD⁺ due to decreased [ATP]/[ADP][Pi] translates to an increased lactate/pyruvate ratio, which promotes net aerobic lactate generation and release, the hallmark of aerobic glycolysis and the Warburg metabolic phenotype (Fig. 3). In differentiated post-mitotic tissues that rely on electrogenic ANT, cytosolic [ATP]/[ADP][Pi] is higher, and NADH/NAD⁺ and lactate/pyruvate ratios are lower. Because lactate/pyruvate equilibrates with the blood, post-mitotic tissues act to consume the extra lactate generated by tumors, which in effect ‘pulls’ aerobic glycolysis by cancer cells. Similarly, fetal tissues release lactate that is disposed of metabolically by maternal tissues [71, 72]. Thus, electrogenic mitochondrial ATP/ADP exchange by non-proliferating tissues leads to consumption of excess lactate formed by aerobic glycolysis in proliferating cells using non-electrogenic ATP/ADP exchange.

Figure 3. Metabolic changes associated with non-electrogenic ATP/ADP exchange.

See text for details. BPG, 1,3-bisphosphoglycerate; ETC, electron transport chain; GAPDH, glyceraldehyde phosphate dehydrogenase; G3P, glyceraldehyde-3-phosphate; LDH, lactate dehydrogenase; Mal/Asp, malate/aspartate; 2-OG, 2-oxoglutarate; PGK, phosphoglycerate kinase.

Increased NADH/NAD⁺ Increases Mitochondrial Membrane Potential in Cancer Cells

The oxidation-reduction poise of various cellular redox systems, such as NADH/NAD⁺, NADPH/NADP⁺, and GSH/GSSH, is also in dynamic near equilibrium between different redox carriers and between compartments [73]. Reducing equivalents from cytosolic NADH transfer into mitochondria via the malate/aspartate shuttle, a process linked to mitochondrial H⁺ entry and consequently driven by Δp. As a result, NADH/NAD⁺ ratios in the mitochondrial matrix are much greater than in the cytosol. Also driven by Δp, mitochondrial NADH then reduces NADP⁺ to NADPH via the NAD⁺/NADP⁺ transhydrogenase (TH) reaction:

$$\text{NADH}+{\text{NADP}}^{+}{\text{H}}_{}^{+}{}_{\text{out}}\to {\text{NAD}}^{+}+\text{NADPH}+{\text{H}}^{+}{}_{\text{in}}$$

Thus, mitochondrial NADPH/NADP⁺ is greater than mitochondrial NADH/NAD⁺ to about the same extent as mitochondrial NADH/NAD⁺ is greater than cytosolic NADH/NAD⁺.

What happens to these equilibria when electrogenic ANT is replaced by non-electrogenic AMPC? First, increased cytosolic NADH/NAD⁺ drives an increase of mitochondrial NADH/NAD⁺, as communicated by the Δp-driven malate/aspartate shuttle (Fig. 3). Greater mitochondrial NADH/NAD⁺ increases the redox span across Complexes I and III of the respiratory chain. Since Δp is in near equilibrium with this redox span, Δp itself must increase in magnitude to maintain the near equilibrium. Taking into account that ΔΨ comprises ~70% of Δp in living cells, then this analysis predicts that ΔΨ becomes greater by 15–25 mV in proliferating cancer cells, which is consistent with old but otherwise unexplained observations that cancer cells and other proliferating cell types have a higher mitochondrial ΔΨ than non-transformed cells (Fig. 3) [74, 75].

Greater mitochondrial NADH/NAD⁺ ratios also produce correspondingly greater mitochondrial NADPH/NADP⁺ ratios via the TH reaction, which in turn leads to greater cytosolic NADPH/NADP⁺ ratios via exchanges of NADP⁺-linked metabolites like isocitrate and 2-oxoglutarate. Importantly for proliferating cancer cells, greater NADPH/NADP⁺ favors reductive anabolic metabolism as, for example, for fatty acid and cholesterol biosynthesis. Greater reduction of NADPH also promotes higher GSH to GSSG ratios via the NADPH-dependent glutathione reductase reaction. Lastly as discussed below, increased mitochondrial NADPH/NADP⁺ has implications for the direction of the mitochondrial isocitrate dehydrogenase (IDH) reaction (Fig. 3).

Increased Mitochondrial ΔΨ and a More Reduced Respiratory Chain Increase Mitochondrial Generation of Reactive Oxygen Species and Susceptibility to Ferroptosis

Mitochondria are an important source of reactive oxygen species (ROS) [76]. Mitochondrial generation of superoxide (O₂^•-) and hydrogen peroxide (H₂O₂) increases with increasing mitochondrial ΔΨ and a more reduced respiratory chain [77–81]. In the presence of iron, O₂^•- and H₂O₂ react to form highly reactive and toxic hydroxyl radical (•OH) by the iron-catalyzed Haber-Weiss or Fenton reaction (reviewed in [82, 83]). •OH reacts indiscriminately and very rapidly (t_1/2 ≈ 10⁻⁹ s) with every major type of macromolecule in living cells, including nucleic acids, proteins, carbohydrates and lipids [73, 84]. With lipids, •OH reacts to form alkyl radicals that initiate an oxygen-dependent chain reaction generating peroxyl radicals and lipid peroxides. Iron also itself catalyzes lipid peroxidation chain reactions sustained by lipid peroxyl, alkyl, and alkoxy radicals [85]. Subsequent beta scission of alkoxy radicals generates toxic aldehydes like malondialdehyde and 4-hydroxynonenal that contribute to various forms of disease, particularly in the liver (reviewed in [83, 86, 87]). Accordingly, chelation of iron with agents like desferal has long been shown to protect against cell death from oxidative stresses like ischemia-reperfusion, drug-induced hepatotoxicity and exposure to oxidant chemicals [88–96].

More recently, the term ferroptosis was introduced for non-apoptotic (necrotic) cell death driven by oxidative stress and subsequent iron-dependent lipid peroxidation [97, 98]. Ferroptosis occurs in cancer cell lines in response to inhibition of glutathione peroxidase-4 (GPX4), an enzyme that metabolizes and detoxifies lipid hydroperoxides, as well as by other interventions limiting glutathione availability [99, 100]. By increasing mitochondrial ΔΨ and respiratory chain reduction with consequent increased mitochondrial ROS generation, non-electrogenic ATP exchange may account for the greater susceptibility of cancer cells to ferroptosis compared to their normal counterparts (Fig. 3).

Partial Closure of Voltage Dependent Anions Channels Defends Against Increased Mitochondrial ΔΨ and Mitochondrial Generation of Reactive Oxygen Species in Cancer Cells

Voltage-dependent anion channels (VDAC) allow hydrophilic metabolites to diffuse freely across the mitochondrial outer membrane. Accumulating evidence indicates that modulation of VDAC conductance can act as a dynamic global regulator or ‘governator’ of mitochondrial metabolism [101, 102]. For example, during hepatic ethanol metabolism, relative VDAC closure inhibits mitochondrial uptake of hydrophilic respiratory substrates to promote selective detoxifying mitochondrial oxidation of membrane-permeant acetaldehyde formed by alcohol dehydrogenase [86, 103, 104]. Another proposed governator role is closure of VDAC to suppress mitochondrial metabolism globally in cancer cells [101].

Remarkably, low nanomolar free dimeric tubulin blocks conductance of the most abundant VDAC isoforms, VDAC1 and VDAC2, after reconstitution in lipid bilayers, whereas the least abundant VDAC3 is insensitive to tubulin [105, 106]. In cancer cells, microtubule stabilization with paclitaxel to decrease free tubulin increases mitochondrial ΔΨ, whereas microtubule depolymerization with colchicine or nocodazole to increase free tubulin leads to a decrease of mitochondrial ΔΨ. These increases and decreases of mitochondrial ΔΨ appear due to corresponding changes of respiratory substrate entry through VDAC (Fig. 3) [107]. Compared to non-proliferating cells, cancer and other proliferating cell types have high levels of free tubulin, because a reservoir of free tubulin is needed for rapid spindle formation at metaphase. Thus, in the steady state in cancer cells, VDAC is partially closed. Limiting access of substrate to the respiratory chain by limitation of VDAC conductance likely protects, at least in part, against mitochondrial hyperpolarization and ROS generation associated with non-electrogenic ATP/ADP exchange. In addition, by limiting mitochondrial release of ATP, VDAC closure causes ATP/ADP ratios to become still lower, which further stimulates aerobic glycolysis [108].

Protein kinase A (PKA) regulates these tubulin-dependent changes of VDAC conductance and mitochondrial ΔΨ. In reconstituted VDAC, phosphorylation by PKA increases the on-rate of tubulin binding and hence of tubulin dependent closure. In situ, PKA activation of cAMP analogues promotes a decrease of ΔΨ, whereas PKA inhibition does the opposite [109].

Erastin, the canonical inducer of ferroptosis, interacts physically with VDAC and blocks the inhibitory effect of free tubulin on VDAC conductance (Fig. 3) [106, 110]. Erastin also reverses mitochondrial depolarization induced by elevated free dimeric tubulin in intact cancer cells treated with nocodazole or colchicine. Furthermore, erastin by itself causes mitochondrial hyperpolarization that in turn promotes mitochondrial ROS formation and subsequent mitochondrial dysfunction (depolarization), which likely represents onset of the MPT. These events lead to a necrotic and presumptively ferroptotic cell death [102, 111]. A small molecule screen identified several structurally different compounds that like erastin antagonize mitochondrial depolarization after microtubule depolymerization without themselves altering microtubule dynamics [112]. These erastin-like compounds also increase mitochondrial ΔΨ and ROS generation, which is followed within about an hour by mitochondrial dysfunction and ultimately cell death [111]. Activation and translocation of c-Jun N-terminal kinase to mitochondria help mediate this mitochondrial ROS generation and dysfunction [113]. Antioxidants like the glutathione precursor N-acetylcysteine and the mitochondrially targeted antioxidant MitoQ block ROS generation, mitochondrial dysfunction and cell death after exposure to these agents.

VDAC is constitutively open in post-mitotic cells that have low levels of free tubulin. Consequently, erastin and erastin-like compounds do not induce mitochondrial hyperpolarization in primary cultured hepatocytes, unlike what happens in hepatocarcinoma cells. Despite an open VDAC, hepatocytes do not experience toxic mitochondrial ROS generation leading to cell death. The reason for this protection is that electrogenic ATP/ADP exchange by ANT in postmitotic cells gives rise to lower mitochondrial ΔΨ, greater oxidation of the respiratory chain, and hence decreased mitochondrial ROS generation compared to proliferating cancer cells that use non-electrogenic ATP/ADP exchange. In this way, the switch from electrogenic to non-electrogenic ATP/ADP exchange makes cancer cells susceptible to inducers of ferroptosis.

The increase of mitochondrial NADPH/NADP⁺ caused by the switch to non-electrogenic ATP/ADP exchange in cancer cells also enhances antioxidant defenses, since NADPH-dependent glutathione reductase will maintain the GSH/GSSG couple in a more highly reduced state (Fig. 3). NADPH and GSH together support the activity of several antioxidant enzymes, including glutathione S-transferases, thioredoxin, thioredoxin reductase, glutaredoxin, sulfiredoxin, peroxiredoxin, and glutathione peroxidases [73, 114–116]. Because of their greater propensity to generate ROS, cancer cells may be more vulnerable to agents impairing these antioxidant defenses. In particular, the GPX4 isoenzyme is important for eliminating •OH-generated lipid hydroperoxides and is implicated as being inhibited by inducers of ferroptosis, including erastin [99]. The relative importance of GPX4 inhibition and VDAC opening for erastin induction of ferroptosis remains to be determined.

High NADPH/NADP⁺ and Expression of NADP⁺-dependent Isocitrate Dehydrogenase Favor Mitochondrial Citrate Release for Biomass Formation and Addiction to Glutamine

IDH catalyzes oxidative decarboxylation of isocitrate to produce 2-oxoglutarate and CO₂ with the reduction of either NAD⁺ or NADP⁺ to NADH or NADPH. The NAD⁺-dependent IDH3 isoform resides in the mitochondrial matrix and catalyzes the third step of the tricarboxylic acid (TCA) cycle. Many cancers overexpress mitochondrial NADP⁺-dependent IDH2. This overexpression is associated with increased tumor aggressiveness and poorer clinical outcomes ([117–119]; reviewed in [120]).

Co-expression of NADP⁺-dependent IDH2 with NAD⁺-dependent IDH3 in the mitochondrial matrix can lead to futile cycling between intramitochondrial NADH and NADPH via the Δp-linked TH reaction, which is effectively an uncoupling reaction:

$$\text{Isocitrate}+{\text{NAD}}^{+}\stackrel{\text{IDH}3}{\to }2\text{-Oxoglutarate}+{\text{CO}}_2+\text{NADH}$$

$$\text{NADH}+{\text{NADP}}^{+}+{\text{H}}_{}^{+}{}_{\text{out}}\stackrel{\text{TH}}{\to }\text{NADPH}+{\text{NAD}}^{+}+{\text{H}}^{+}{}_{\text{in}}$$

$$\text{2-Oxoglutarate}+{\text{CO}}_2+\text{NADPH}\stackrel{\mathrm{IDH}2}{\to }\text{Isocitrate}+{\text{NADP}}^{+}$$

$${\text{Net:}\text{H}}_{}^{+}{}_{\text{out}}\to {\text{H}}^{+}{}_{\text{in}}$$

Such cycling may have a regulatory function to fine tune mitochondrial TCA cycle activity [121]. Other possibilities to minimize futile cycling are segregation of different IDH isoforms into different cells or even into different mitochondria within the same cell. More complex regulation of IDH isoforms by acetylation, Ca²⁺ and other factors may also modulate futile cycling [122, 123].

Because NADPH/NADP⁺ ratios are so much higher than NADH/NAD⁺ ratios and made even higher by non-electrogenic ATP/ADP exchange, expression of NADP⁺-dependent IDH2 favors accumulation of citrate in equilibrium with isocitrate via the iconitase reaction to promote mitochondrial citrate release for biosynthesis of fatty acids, an essential need for proliferating cells (Fig. 3). High NADPH/NADP⁺ together with citrate release depletes intermediates for forward TCA cycle reactions. To sustain activity, these intermediates must be replenished, a process of anaplerosis. In cancer cells, metabolism of glutamine by glutaminase and then glutamate dehydrogenase (glutaminolysis) becomes the source of 2-oxoglutarate to continue forward flux through the TCA cycle. Glutamine is also a precursor for amino acid synthesis, another important need of proliferating cells. Cancer cells frequently experience hypoxia as they outgrow their blood supply. In hypoxia, net reversal of the IDH reaction (reductive carboxylation) occurs to form citrate from 2-oxoglutarate, which in turn is supplied by glutaminolysis. This reverse flux may help cancer cells survive aglycemia during hypoxia [122, 124, 125]. Reductive decarboxylation of 2-oxoglutarate in cancer cells can also occur during normoxia and is driven by NADPH produced by the TH reaction [126]. Overall, higher NADPH/NADP⁺ ratios due to the switch to non-electrogenic ATP/ADP exchange together with mitochondrial expression of NADP⁺-dependent IDH2 promotes glutamine utilization and dependence in cancer cells.

Spontaneous mutations of IDH1 and IDH2 occur in some cancers, such as gliomas and certain leukemias (reviewed in [127, 128]). By contrast tumor-associated mutations of IDH3 have not been reported. Mutated forms of IDH1/2 catalyze a new reaction, namely the NADPH-dependent reduction of 2-oxoglutarate to the ‘oncometabolite’ D-2-hydroxyglutarate, which can accumulate intracellularly in greater than 100-fold excess compared to the absence of mutation. Mutated IDH 1 and 2 consume 2-oxoglutarate, which is replenished in large part by the glutaminase pathway [129]. 2-Hydroxyglutarate inhibits a variety of 2-oxoglutarate -dependent enzymes, including prolylhydroxylase whose inhibition stabilizes hypoxia inducible factor-1α and promotes aerobic glycolysis and cell proliferation [130]. 2-Hydroxyglutarate also inhibits histone demethylases and methyl cytosine dioxygenase, leading to increased DNA methylation and altered gene expression with activation of oncogenes and inactivation of tumor-suppressor genes [131–134].

Perspectives and Unanswered Questions

The switch from electrogenic to non-electrogenic ATP/ADP exchange provides advantages to proliferating cancer cells but also creates vulnerabilities. The foremost advantage is a 35% increase of ATP yield from complete oxidation of glucose, which drives correspondingly greater biomass formation. Higher ATP yield comes at the expense of a lower ΔG_P, which becomes similar to the ΔG_P in prokaryotes. Since the pathways and energetics of DNA, protein and lipid biosynthesis in eurkaryotes are largely of inherited from their prokaryotic ancestors, lower ΔG_P does not impair cell proliferation, which is, of course, quite robust in organisms like bacteria. By contrast, eukaryotic cells have evolved to differentiate and perform specialized tasks that require a greater ΔG_P. Such ‘heavy lifting’ cannot be performed by cells that have switched to non-electrogenic ATP/ADP exchange and have a smaller ΔG_P. This reasoning implies that proliferating cancer cells are undifferentiated, at least in part, because their energetics do not allow performance of differentiated functions. The switch to non-electrogenic ATP/ADP exchange enables more rapid cell growth but is not obligatory for cell proliferation. Consequently, the switch may not occur in all cancer cells, such as well differentiated tumors whose growth is typically slower.

In the immune system, certain cell types display polarization between different functional and metabolic phenotypes. For example, macrophages have different inflammatory (M1) versus anti-inflammatory (M2) phenotypes. M1 macrophages display Warburg-like metabolism with aerobic glycolysis and decreased respiration, whereas M2 macrophages mainly employ oxidative glucose metabolism (reviewed in [135]). Similarly, after allogenic activation, T cells undergo metabolic reprogramming and switch from fatty acid and pyruvate oxidation to aerobic glycolysis with increased reliance on glutaminolysis [136]. Future studies will be needed to determine whether reprogramming to aerobic glycolysis in macrophages and T-cells also involves a switch from electrogenic to non-electrogenic ATP/ADP exchange.

Despite their switch to non-electrogenic ATP/ADP exchange, cancer cells continue to express ANT isoforms, such as ANT2 and ANT 3. Another important question for future studies is the mechanism by which ANT activity becomes suppressed in cancer cells and how AMPC expression is activated. Regarding ANT inhibition, a soluble factor in the cytosol seems involved, since after plasma membrane permeabilization CAT and BA inhibit OCR similarly to oligomycin. Metabolites like polyamines that are present at high concentration in cancer cells, but not post-mitotic cells, might be responsible. Spermine, in particular, may alter ANT function [137]. Likewise, cancer cells may express proteins that inhibit ANT or activate AMPC directly or through post-translational modification (phosphorylation, acetylation, etc.).

The switch from electrogenic to non-electrogenic mitochondrial ATP/ADP exchange also creates vulnerabilities in cancer cells. Since cancer cells utilize non-electrogenic AMPC for mitochondrial ATP/ADP exchange, AMPC inhibition is a potential treatment target to induce selective bioenergetic stress in tumors. Increased intramitochondrial NADH/NAD⁺ causes an increase of ΔG_Ri-iii between NADH and cytochrome c and a corresponding increase of ΔΨ, changes that promote mitochondrial generation ROS. Although higher mitochondrial NADPH/NADP⁺ and GSH/GSSG ratios together with partial closure of VDAC act to protect against ROS-induced cell injury, cancers cells nonetheless become selectively more vulnerable than non-cancer cells to disruptors of antioxidant defenses (e.g., PDX4 inhibition, GSH depletion) and VDAC openers (e.g., erastin-like compounds) (Fig. 3). Moreover, increased utilization of glutamine associated with expression of IDH2 creates a selective vulnerability to interventions limiting glutamine supply or IDH2 activity. In this regard, TH inhibition shows promise as an anti-cancer treatment target [138], and suppression of TH activity would both suppress glutamine utilization and impair antioxidant defenses. These vulnerabilities create new opportunities for development of cancer chemotherapeutics.

Highlights.

• In cancer cells, mitochondrial ATP/ADP exchange occurs non-electrogenically

• Cytosolic ATP/ADP decreases to stimulate glycolysis (Warburg effect)

• ATP/O increases by 35% for complete glucose oxidation

• Increased NAD(P)H promotes lactate release, anabolic metabolism and glutamine utilization

• Increased mitochondrial NADH increases ΔΨ, reactive oxygen species, and ferroptosis

Abbreviations:

ΔΨ

membrane potential

ΔG_P

phosphorylation potential

ΔG_RI-III

redox free energy change across Complexes I and III

Δp

protonmotive force

AMPC

ATP/Mg-Pi carrier

ANT

adenine nucleotide translocator

BA

bongkrekic acid

CAT

carboxyatractyloside

FCCP

trifluoromethoxy carbonylcyanide phenylhydrazone

GPX4

glutathione peroxidase-4

IDH

isocitrate dehydrogenase

MPT

mitochondrial permeability transition

OCR

oxygen consumption rate

PKA

protein kinase A

ROS

reactive oxygen species

SLC

solute carrier family

TCA

tricarboxylic acid

TH

NAD⁺/NADP⁺ transhydrogenase

TMRM

tetramethylrhodamine methylester

VDAC

voltage-dependent anion channel