To the editor

The 'chemiosmotic hypothesis', formulated in the 1960s, revolutionized the study of the mitochondrion and made it accessible to thermodynamic analysis. The central aspect of the hypothesis is that the energy from electron transport is transduced into a proton electrochemical gradient (Δp) across the inner membrane, and that protons re-entering the matrix through the F1Fo ATPase force this proton pump to run in reverse to generate ATP. The magnitude of Δp is given by Δp = ΔΨ − 61.5 ΔpH, where ΔΨ is the membrane potential and ΔpH is the transmembrane pH gradient. Although the magnitudes of these components are difficult to estimate with precision, it is firmly established that electron transport is primarily governed by the slight disequilibrium between Δp and the redox spans at complexes I and III and that the net rate (and direction) of the ATP synthase is determined by the disequilibrium between Δp and the Gibbs energy of the matrix ATP/ADP plus phosphate pool (reviewed in ref. 1).

The second mitochondrial 'revolution', detailing the involvement of the organelle in apoptotic and necrotic cell death, is now at its height, as evidenced by the article by Matsuyama et al. in Nature Cell Biology (2, 318–325; 2000). It is, however, unfortunate that the authors have not drawn attention to the fact that key aspects of their paper demand re-evaluation either of the chemiosmotic theory or of the data presented. The central proposal is that staurosporine- and Bax-induced apoptosis increases ΔpH by about −0.9 units. At the same time the 'relative [mitochondrial] membrane potential' increased by some 20%. The consensus in the literature is that intracellular mitochondria maintain a Δp of 160–210 mV, a ΔΨ of 120–180 mV, and a ΔpH of −0.5 units. Staurosporine is therefore proposed to increase Δp by 80 mV, by a process of ATP-synthase reversal.

There are three fundamental problems with this interpretation. First, the respiratory chain is thermodynamically incapable of supporting such an increased Δp without a hypothetical 'change of gear', that is, a reduction in the stoichiometry of proton translocation.

Second, the distribution of Δp between ΔΨ and ΔpH is determined by the balance between the uptake of ions such as Ca2+ and the availability of permeant weak acids such as phosphate, bicarbonate and carboxylic acids, which equilibrate across the inner membrane as the first, second or third power of ΔpH depending on the number of acidic groups and buffer against large changes in ΔpH. It is difficult to visualize how a ΔpH of −1.3 units could be generated in the presence of physiological concentrations of weak acids, as the only situation in which mitochondria can maintain a large ΔpH is when permeant weak acids are absent and Ca2+ is in excess, in which case ΔΨ is decreased proportionately.

The third, and most fundamental, problem comes from the mechanism advanced to account for these findings. The authors adopt a proposal by Vander Heiden et al.2,3,4, reviewed in this journal (Nature Cell Biol. 1, E209–E216; 1999), that Bcl-2 related proteins may be required for exchange of adenine nucleotides between the matrix and the cytoplasm. In this controversial hypothesis, pro-apoptotic stimuli would inhibit this exchange, leading to hyperpolarization, accumulation of metabolites, osmotic swelling and release of cytochrome c. However, the assay of adenine-nucleotide exchange upon which the Vander Heiden hypothesis is based3 is invalid as it involves a 10-min exposure of mitochondria to 14C-ADP, rather than the sub-second rapid quenching required to determine initial rates of this extremely active process5. Leaving aside further discussion of this hypothesis, Matsuyama et al. propose that such inhibition would lead to an increase in the matrix ATP/ADP ratio, 'favouring reverse operation of the F1F0-ATPase' and that this proton extrusion would contribute to hyperpolarization and increased ΔpH. However, as the ATPase must function in the direction of ATP synthesis to generate an increased ATP/ADP ratio, it cannot drive the ATP/ADP ratio past thermodynamic equilibrium and then start to hydrolyze it again to increase Δp. The only situation in which ATP-synthase reversal is possible is when Δp is decreased below that required for equilibrium with the matrix ATP/ADP pool, and certainly not in the presence of a supposed 80-mV hyperpolarization.

How do we resolve these discrepancies? One possibility is that the interpretation of the fluorescence changes is inaccurate. Changes in matrix volume could affect the ΔpH signal, and use of fluorescence-activated cell sorting to determine ΔΨ is dependent on the use of low concentrations of probes such as DiOC6(3), as 40 nM DiOC6(3) renders the matrix signal largely ΔΨ-independent and causes severe inhibition of mitochondrial complex I6.

An understanding of mechanisms that underlie programmed cell death is one of the main goals in mitochondrial physiology. The results presented in this paper are intriguing and are evidently reporting an important event. However, interpretations that contravene the laws of thermodynamics, or at least require re-evaluation of fundamental tenets of bioenergetics, may serve to increase, rather than resolve, the current confusion within the field.