mPTP: mitochondrial permeability, cyclophilin D and cyclosporine

The mitochondrial inner membrane must retain differences in electrical potential and proton concentration while admitting selected metabolites through dedicated carriers. The mitochondrial permeability transition disrupts this selectivity: a conducting pathway allows small solutes and ions to cross the membrane, dissipating stored electrochemical energy. mPTP, the mitochondrial permeability transition pore, names the pore associated with this phenomenon. Its regulation is understood much better than its definitive molecular construction.

Calcium overload, oxidative stress, inorganic phosphate and energetic state change the probability of the transition. Matrix cyclophilin D increases sensitivity to opening; cyclosporine A can suppress this regulatory component in experiments. Genetic protection in a mouse, reduced swelling of isolated mitochondria and improved myocardial infarction outcomes in humans are nevertheless distinct claims. Combining them creates a misleading impression of an established pore structure and an available universal treatment.

The membrane barrier and the first observations

The outer membrane is relatively permeable to small metabolites through VDAC. The inner membrane, rich in respiratory proteins and carriers and with a distinctive lipid composition including cardiolipin, maintains the proton gradient. Respiratory complexes move protons from the matrix into the intermembrane space. Their return through ATP synthase couples electrochemical energy to ATP production. Even a small additional leak therefore affects coupling efficiency, while a large nonselective conductance can rapidly destroy the energetic barrier.

Define $\Delta\Psi_m=\Psi_{matrix}-\Psi_{IMS}$ and $\Delta pH=pH_{matrix}-pH_{IMS}$. Under this convention, the signed proton-motive force is:

$$\Delta p=\Delta\Psi_m-\frac{2.303RT}{F}\Delta pH$$

The matrix is normally electrically negative and more alkaline, so both contributions make $\Delta p$ negative under this convention. Stored energy is often discussed through its magnitude $|\Delta p|$. A negative membrane potential cannot be combined with a positive “total force” without explaining the signs. Values depend on tissue, substrates, workload and method; a number measured in isolated mitochondria is not a universal opening threshold. $R$ is the gas constant, $T$ absolute temperature and $F$ the Faraday constant.

Hunter and Haworth's 1979 series examined a calcium-induced transition to increased permeability and mitochondrial swelling, separating phosphate involvement, the calcium trigger and transitional states. Chelating available calcium and supplying adenine nucleotides could alter reversibility under particular conditions. Magnesium, polyamines including spermine, acidity and oxidation state also influenced the transition. These were observations of a regulated membrane phenomenon, rather than a visual identification of a single protein hole.[1–3]

In 1988, Crompton and colleagues showed inhibition by cyclosporine A of a calcium-dependent pore in heart mitochondria activated by phosphate and oxidative stress. The connection with cyclophilin D later gained genetic support. The experimental conditions matter: CsA sensitivity describes a particular regime and does not mean the drug completely prevents every form of inner-membrane damage.[4,9–11]

High-conductance states are often described as having conductance on the order of a nanosiemens and permeability to solutes of roughly up to 1.5 kDa. These are useful characteristics of some measurements, not a precise pore diameter shared by all cells. Passage depends on shape, charge, time and channel state as well as molecular mass. Ions, some nucleotides, NAD and glutathione are small enough to belong to the potentially permeant pool; matrix proteins and mitochondrial DNA do not become freely permeable through this pathway.[6–8]

Cytochrome c lies primarily in the intermembrane space and has a mass of about 12 kDa. It does not leave the matrix through mPTP. Its entry into the cytosol requires an outer-membrane change, such as mechanical damage following swelling or the separate BAX/BAK-regulated process of MOMP. This topology determines which damage events are sequential and which can occur independently.

What is established about pore composition?

The historical model combined outer-membrane VDAC, inner-membrane ANT and matrix-side CypD, with TSPO and hexokinase II also discussed nearby. It connected pharmacological associations and membrane contacts, but assigned an inner-membrane phenomenon to a proposed complex spanning both envelopes. Experiments with VDAC deficiency showed that the transition persists, arguing against VDAC as an obligatory universal pore-forming component. Different studies used different genetic interventions; they should not be collapsed into an imaginary single triple knockout of all VDACs.[13,14]

ANT, the adenine nucleotide translocator, normally exchanges one matrix ATP for one incoming ADP. The human family includes carriers encoded by SLC25A4, SLC25A5, SLC25A6 and SLC25A31; expression and functional compensation depend on tissue and species. Six transmembrane helices form three repeated structural units. Salt-bridge networks alternately restrict access from the matrix and intermembrane sides. During normal transport, both entrances should not simultaneously form an unrestricted through-channel.[17,18]

The $c$-state is accessible from the intermembrane, conventionally cytosolic, side; the $m$-state is accessible from the matrix. Carboxyatractyloside stabilizes the $c$-state and facilitates the transition in several models. Bongkrekic acid stabilizes the $m$-state and can restrain it. Both ligands also block ADP/ATP exchange and are toxic. Their effects support a connection between ANT and permeability regulation, but do not establish that calcium universally breaks both gating networks to produce one defined pore type.

In Kokoszka's study, deleting ANT1 and ANT2 did not eliminate the transition, although sensitivity and ligand responses changed. Karch's 2019 study found strong restriction of the transition with broader ANT-family deficiency in the studied system, with particularly marked protection after additional CypD deletion. Remaining isoforms, genetic background and the magnitude of the calcium challenge must be considered when comparing these results. They support a substantial role for ANT without proving that it is the sole pore protein in every tissue.[15,16]

Another group of models implicates $F_1F_o$-ATP synthase. Its catalytic $F_1$ head faces the matrix; the membrane $F_o$ sector provides the proton pathway, while the stator including OSCP maintains mechanical coupling. Dimers and their rows contribute to crista curvature. Proposed mechanisms include calcium replacing magnesium at catalytic sites, altered mechanical coupling, a dimer-interface pathway and a channel within the $c$-subunit ring. The dimer interface and the $c$-ring are distinct structural hypotheses and should not silently become one model.[19,20]

Reconstituting purified components and recording a channel in an artificial membrane demonstrates conduction in those conditions. Genetic removal tests whether a candidate is necessary in a cellular system. In He 2017 and Carroll 2019, transition features persisted without the $c$-subunit and without assembled ATP synthase, respectively. Channel-formation data therefore cannot be presented as the final resolution of the debate. Method differences, residual activity, metabolic remodeling and definitions of the measured event remain subjects of discussion.[21,22]

Associations of ATP synthase, ANT and the phosphate carrier are sometimes described as an ATP synthasome. Such organization may matter for substrate exchange and coupling. Co-localization or interaction alone does not identify the wall of a water-conducting channel. The claim that ANT is the primary pore, ATP synthase a backup and their complex the modern consensus exceeds the evidence. A working account should distinguish competing models and allow the mechanism to depend on the experimental system.

Cyclophilin D regulates sensitivity

CypD is encoded by PPIF. A mitochondrial targeting signal directs its precursor into the matrix, where the mature protein functions. It belongs to the cyclophilins, peptidyl-prolyl cis–trans isomerases. Their fold creates a pocket that interacts with substrate and cyclosporine, accelerating interconversion of the peptide bond preceding proline:

$$\mathrm{peptide-Pro}_{cis}\;\underset{CypD}{\rightleftharpoons}\;\mathrm{peptide-Pro}_{trans}$$

This is a catalytic function, not proof that CypD itself lines a large transmembrane channel. ANT and OSCP interactions, protein conformational changes and regulatory contacts are discussed. Without a specifically supported structural experiment, one loop or one proline cannot be assigned as the obligatory universal pore lock. The isomerization barrier also depends on the substrate and conditions.

Mitochondria without CypD generally require a greater calcium challenge to undergo the transition; typical CsA sensitivity decreases or disappears in the corresponding regime. The pore need not vanish: a sufficiently strong challenge can still cause permeability without CypD. PPIF is therefore a persuasive regulatory target, but knockout resistance is not evidence that all routes of damage have disappeared.[9–11]

Post-translational modifications connect CypD to metabolic and redox state. Hafner and colleagues demonstrated deacetylation by NAD-dependent SIRT3, involving Lys166 in the mouse sequence they used. SIRT3-deficient mice showed altered mitochondrial swelling and age-related cardiac phenotypes. Residue numbering must be interpreted in the context of species and precursor versus mature protein. These findings do not establish a clinical method of closing everyone's pore by raising NAD.[25]

Thiol state also matters. Nguyen's study found that the NO donor GSNO reduced peroxide-induced opening in cells containing CypD, while replacing Cys203 with serine reduced sensitivity in the studied cellular and mouse models. Mutation and S-nitrosylation are not identical, but these results rule out portraying CypD S-nitrosylation as an unconditional pore activator. NO effects depend on dose, redox environment, target and timing; excessive reactive nitrogen species can also cause injury.[33]

Akt/ERK signaling, grouped under RISK in cardioprotective models, and GSK3β inhibition, including through its Ser9, connect extracellular signals to mitochondrial resistance. Phosphorylation of a signaling kinase, altered CypD contacts and demonstrated direct modification of CypD are different events. Reducing all protection to an invariant reaction in which GSK3β phosphorylates CypD and opens the pore is inappropriate.[6,8,32]

Calcium, phosphate, pH and redox state

Calcium enters the matrix primarily through the MCU complex, driven by electrical potential. NCLX and other efflux mechanisms limit accumulation. Modest calcium elevation can support metabolism, whereas excess combined with an unfavorable membrane state increases transition risk. The result depends on entry, exit and binding rather than solely on calcium concentration outside the mitochondrion.

Free matrix $Ca^{2+}$ differs from total calcium. Phosphate helps bind calcium and form calcium–phosphate complexes, allowing considerable total accumulation without a proportional rise in free ion. One idealized balance can be written:

$$3Ca^{2+}+2HPO_4^{2-}\rightleftharpoons Ca_3(PO_4)_2+2H^+$$

This conserves atoms and charge; it does not claim that every living mitochondrion forms a pure crystal of this exact composition. Buffered, amorphous and protein-associated forms are more complex. Phosphate can change both calcium buffering and transition sensitivity, with direction depending on the medium and CypD-dependent regime.[1–3,6,11]

A Hill function sometimes approximates the probability of a response:

$$\theta(C)=\frac{C^n}{K_{0.5}^{\,n}+C^n}$$

$C$ must be an explicitly chosen measured concentration, $K_{0.5}$ its value at half the modeled response, and $n$ the empirical steepness. The Hill number does not automatically count ions bound to a pore molecule. Fitted parameters cannot be transferred between tissues, species, buffers or methods. Binding, accumulation, potential-dependent feedback and population heterogeneity may produce similar nonlinear behavior.

Calcium retention capacity, CRC, is often measured using successive calcium pulses until accumulated calcium is suddenly released. It is expressed as the amount of calcium challenge per mitochondrial protein mass, rather than a universal free matrix concentration. Pulse size and interval, temperature, substrates, phosphate, respiration, calcium efflux and preparation quality all affect the result. Two preparations with equal CRC may accumulate calcium at different rates and suffer different forms of damage.

Thiol oxidation, changes in NAD(H) state and cardiolipin injury connect the transition to ROS. This is a feedback loop: oxidative stress may facilitate opening, while energetic collapse and metabolite loss alter ROS generation and removal. A fall in potential does not by itself prove increased ROS under every condition, because respiratory reduction state and electron transport change as well. The oxidant source and the consequences of opening must be separated in each experiment.

Acidosis often lowers transition probability, particularly in ischemia models, and pH restoration may release that restraint. There is no single pH at which all mitochondria are guaranteed closed. Magnesium and adenine nucleotides generally oppose the transition in specified regimes through calcium regulation, protein states and Mg–ATP/ADP complexes. Potential influences both calcium accumulation and permeability state; the statement that a more negative matrix always keeps the pore closed ignores these opposing effects.[6–8]

From brief opening to energetic injury

The transition can be transient and reversible. In some systems, such events are considered a means of calcium release and redox adjustment. A flickering potential-dye signal, however, does not identify mPTP: other ion fluxes, respiratory oscillations, photodamage and mitochondrial movement can produce similar dynamics. Physiological function should be linked to a confirmed mechanism and conditions rather than assigned automatically to every brief drop in $\Delta\Psi_m$.[6,8]

Sustained high permeability dissipates the gradient through proton and other ion movement, reducing oxidative ATP synthesis. Initial uncoupling can accelerate respiration by removing proton-gradient backpressure. Prolonged opening can instead lose small metabolites and cofactors, disturb matrix volume and damage respiratory complexes. Respiration therefore need not always become maximal after pore opening; at a later stage it may decline.

Under suitable conditions, reversible ATP synthase can hydrolyze ATP and use the energy to pump protons out of the matrix:

$$ATP+H_2O\rightleftharpoons ADP+P_i$$

The chemical contribution for hydrolysis is:

$$\Delta G_{hyd}=\Delta G^{\circ\prime}_{hyd}+RT\ln\frac{a_{ADP}a_{P_i}}{a_{ATP}}$$

The $a$ terms represent activities; the transformed standard state assumes a specified pH. Ionic forms, magnesium binding and experimental conditions affect numerical values. Coupled direction also depends on proton-transfer work and enzyme stoichiometry. A fall in $\Delta p$ favors reversal but does not establish continuous hydrolysis at a universal rate: ATP and functioning enzyme must remain available. IF1 can restrain reverse ATPase activity depending on system state and pH. With a damaged membrane, attempts to sustain the gradient by ATP hydrolysis can worsen energetic depletion.[6–8,23]

The osmotic component arises from movement of solutes and water. Impermeant matrix macromolecules, ion redistribution and Donnan effects favor swelling. The van't Hoff relation for an ideal dilute solution provides an orientation:

$$\Pi\approx RT\sum_i i_iC_i$$

It illustrates the role of osmotically active particle concentration, rather than providing a ready quantitative model of a protein-dense matrix. Electroneutrality, binding, nonideality and membrane mechanics limit direct application. Unfolding cristae provides some inner-surface reserve, while the outer membrane has different geometry and may be damaged by a large volume increase.[6,8]

If the outer barrier changes, intermembrane proteins may enter the cytosol. Cytochrome c helps assemble the apoptosome with Apaf-1 and ATP/dATP, followed by caspase-9 and effector caspase-3/7 activation. Smac/DIABLO helps relieve IAP-mediated caspase inhibition, including XIAP; Omi/HtrA2 and other proteins may contribute to specific injury routes. AIF and EndoG require consideration of release, processing and relocation, rather than simply the presence of an inner pore. This list does not imply that every protein is always released simultaneously by the same process.

Cell-death claims require particular precision. Nakagawa's CypD deficiency protected against some necrotic insults without eliminating the studied apoptotic forms. BAX/BAK-dependent outer-membrane permeabilization can occur separately from CypD-dependent inner-membrane transition. Energetic reserves influence execution of apoptosis and progression to necrotic injury, but no universal percentage of residual ATP separates all cells into these two modes.[9,10]

Measuring the transition: why one marker is insufficient

In isolated mitochondrial suspensions, swelling is commonly assessed by reduced light scattering, calcium release by an external indicator and potential by a potential-sensitive dye. Cellular approaches include matrix calcein retention with appropriate quenching of extramitochondrial signals; electrophysiology records high-conductance inner-membrane states. Each measurement addresses a different question. Swelling reports volume change, dye behavior reports signal distribution, and current reports a conductive mode in a particular preparation.

Potential can fall because of uncoupling, respiratory inhibition, substrate shortage or membrane injury without established mPTP. Calcein loss requires localization and cell-quality controls; calcium release also depends on the ability to retain calcium. CsA sensitivity and PPIF manipulation strengthen interpretation, but different pore-forming or damaging regimes can respond differently. A time sequence, multiple independent signs and appropriate controls are more persuasive than a single attractive graph.

Preventing opening must also be distinguished from recovering after established injury. Higher CRC before a challenge does not demonstrate pore closure after cofactor loss and outer-membrane rupture. Brief recovery of potential does not guarantee cell survival. Conversely, reversible opening is not necessarily an immediate point of irreversibility: duration, the number of affected mitochondria, metabolic reserves and other injuries determine the outcome.[6–8,11]

Ischemia and reperfusion combine several stresses

During ischemia, oxygen shortage reduces oxidative ATP production. ATP-dependent pumps and calcium-control systems fail; acidosis, metabolite accumulation and ionic changes alter mitochondrial state. In several models low pH temporarily restrains the transition despite an increasing damage burden. This does not mean all pores remain closed during ischemia or that mitochondrial injury is absent.

Restored blood flow simultaneously supplies oxygen, changes pH, delivers substrates and alters ROS formation. Removal of acid inhibition in the presence of calcium overload and oxidative stress favors the transition, one version of the reperfusion pH paradox. MCU entry, NCLX exit, cellular Na/Ca exchange and ATP-dependent SERCA influence the result. Tissue and stage alter the sequence; every infarction cannot be assigned one identical opening window or one fraction of necrosis attributable solely to mPTP.[6,8,32]

Calcium overload also injures cells through other routes. Calpain-1/2 cleave cytoskeletal and membrane-associated proteins in particular models, including spectrin, talin, dystrophin-complex components and other structural elements. Changes in titin, ankyrin, connexin-43 and respiratory proteins including NDUFS1 have been discussed, but targets and causal contributions vary between experiments. Hypercontracture and mechanical sarcolemmal damage aggravate cardiomyocyte death. Troponin release indicates myocardial injury, not a specific mPTP-opening test.

Preconditioning and postconditioning connect brief protective interventions with Akt/ERK, GSK3β and altered mitochondrial resistance. This is important experimental cardioprotection, but a signaling pathway cannot substitute for a clinical outcome. Protection from a model challenge, reduction in injured area and lower mortality require separate assessment.

Nervous-system injury and collagen VI myopathies

In brain ischemia, excitation through NMDA receptors and calcium-permeable AMPA variants can raise intracellular calcium. Not all AMPA receptors have the same calcium permeability: subunit composition and processing matter. Matrix calcium accumulation depends on remaining driving force, while potential loss alters both transport and ATP-dependent restoration of ionic gradients. These interactions are particularly relevant to the ischemic penumbra. Protection in mice with altered CypD after focal ischemia supports transition involvement without proving inhibitor efficacy in stroke patients.[12]

Du and colleagues examined the CypD–amyloid-β connection and reported improvement in mitochondrial, neuronal and behavioral measures with CypD deficiency in an Alzheimer's disease model. This supports a mechanism in that system rather than a treatment for human dementia. Parkinson's disease models discuss α-synuclein and complex I impairment; Huntington's models discuss mutant huntingtin, and some ALS models mutant SOD1 and mitochondrial changes. Shared calcium and redox links do not establish an identical primary cause or a fixed fold increase in pore sensitivity across these disorders.[24,29]

Collagen VI deficiency connects the extracellular matrix to muscle-cell injury and mitochondrial regulation. COL6A1, COL6A2 and COL6A3 variants can underlie a myopathy spectrum including Bethlem and Ullrich phenotypes. Irwin and colleagues described mitochondrial dysfunction and apoptosis in collagen VI-deficiency models. Tiepolo and colleagues reported improved mitochondrial characteristics and mouse phenotype with Debio 025. The latter study explicitly concerns mice: it cannot establish alisporivir efficacy in phase II/III human muscular-dystrophy treatment.[30,31]

Cyclosporine A: mechanism and clinical testing

CsA is a cyclic peptide binding several cyclophilins. Its CypD-mediated effects may raise transition resistance in mitochondrial experiments, but it is not an exclusively selective PPIF inhibitor. The complex with cytosolic cyclophilin A inhibits calcineurin and contributes to immunosuppression. Clinical risks, including nephrotoxicity and drug interactions, do not disappear by describing the drug as acting only in mitochondria.

Piot's 2008 pilot reported a signal of reduced injury when CsA was given before reperfusion in myocardial infarction. It assessed particular biomarker and imaging measures in a small sample; it cannot be retold as a guaranteed reduction of necrosis in each patient or proof of improved survival. The signal justified a larger test.[26]

CIRCUS randomized 970 patients with anterior STEMI and an occluded culprit artery to intravenous CsA 2.5 mg/kg or placebo before coronary recanalization. Its primary composite included death, worsening heart failure during the initial admission, heart-failure rehospitalization and adverse left-ventricular remodeling at one year. In the population evaluable for this endpoint, rates were 59.0% with CsA and 58.1% with control, without a statistically convincing advantage ($P=0.77$). This is a clinically neutral result for the tested strategy.[27]

Delivery into injured tissue, formulation, administration timing, no-reflow and background interventions including antiplatelet therapy have been discussed as possible explanations for the difference between model protection and trial results. CIRCUS did not establish delivery failure as the cause. Nor does its neutral outcome disprove permeability transition itself. The drug, dose, timing and population were tested; another strategy requires its own clinical evidence.

MITOCARE tested TRO40303 alongside primary coronary intervention and did not confirm the expected reduction in infarct injury using its chosen measures. Associations with TSPO and the outer membrane do not make this negative clinical result a genetic demonstration that TSPO cannot participate in any mechanism. A drug trial and testing a protein's necessity are different tasks.[28]

Metabolic support and the limits of supplement claims

Magnesium is important for Mg–ATP-dependent reactions, and deficiency has independent clinical consequences. Ionic competition in a mitochondrial experiment nevertheless cannot establish a universal oral dose that closes heart or brain mPTP. Matrix concentration, tissue exposure and initial state differ from buffer conditions. Correcting an established deficiency and experimentally inhibiting a pore are different goals.

CoQ participates in electron transfer and lipid redox reactions. Maintaining its reduced pool may alter oxidative load, but supplement effects depend on delivery, tissue and disease. The article on CoQ10 and polyphenols separates molecular function from evidence for taking a product. A general antioxidant role does not establish guaranteed direct CypD or pore blockade.

Lipoic acid and dihydrolipoate participate in redox reactions; enzyme-bound lipoyl groups have a different role from freely administered material. Experimental thiol reduction does not prove specific correction of Cys203 by a drug or supplement in living tissue. Acetyl-L-carnitine contributes to acyl-group metabolism and substrate availability, discussed further in the acetyl-L-carnitine article. NADH should not be called a SIRT3 activator: its deacetylase reaction consumes NAD⁺, and changes in the NAD⁺/NADH ratio require separate analysis.

Combining magnesium, CoQ, lipoic acid and acetyl-L-carnitine does not become a proven antipore regimen merely because a biochemical arrow can be assigned to each. Actual exposure, target tissue, measured mechanism, clinical outcome and combination safety need evidence. This boundary preserves the metabolic biology while limiting claims about a specific therapeutic effect.

Interpreting the evidence and preparation method

mPTP can be considered at three levels: functional loss of inner-membrane selectivity, molecular models of its conducting pathway and clinical ability to improve disease outcomes. Electrophysiological, biochemical and genetic evidence supports the first; competing ANT and ATP-synthase interpretations remain at the second; the third requires trials of specific strategies. CypD is a well-supported sensitivity regulator in numerous models without automatically resolving the other two questions.

Sustained permeability can unite calcium overload, gradient loss, ATP depletion and osmotic injury. Each step has conditions: respiration depends on substrates, hydrolysis on ATP and enzyme constraints, protein release on the outer membrane, and cell death on combined energetic and signaling disturbances. These conditions determine reversibility and explain why one pore diagram is insufficient for diagnosis or prescribing.

This narrative scientific review was prepared from the supplied article while retaining its principal mechanistic and clinical branches. On 8 October 2026, targeted checks of primary publications addressed identified conflicts in ANT/ATP-synthase genetics, CypD-dependent death, SIRT3/CypD, Cys203 and the clinical CIRCUS result. Mismatched bibliographic identifiers were corrected, and an unsupported original CypD-structure entry was not used as evidence. This was not a systematic search of all literature; neither formal GRADE assessment nor external peer review is claimed. Numerical parameters are tied to experimental context, and preclinical findings are separated from human results.