Physiological cyanide concentrations do not stimulate mitochondrial cytochrome c oxidase activity

Randi et al. (1) claim that vanishingly small concentrations of cyanide added to human cell lines 1) increases cell proliferation, 2) reduces the basal glutathionylation of bovine cytochrome c oxidase (CcO), 3) increases the steady-state activity of CcO, and 4) produces spectral changes of purified CcO.

Our criticism arises from the following.

First, we have reproduced the effect of 1 nM cyanide on the respiration of mouse liver mitochondria by using the tetramethyl-p-phenylene diamine (TMPD)–ascorbate reducing system and found no evidence for increased CcO activity (Fig. 1A).

Fig. 1.

(A) Effect of 1 nM and 100 μM cyanide (CN) on the respiration of rotenone- and antimycin-treated mouse liver mitochondria by using the TMPD–ascorbate system. Briefly, 20 mg of purified mitochondria were incubated with 1 nM or 100 μM KCN for 15 min in resuspension buffer (75 mM sucrose, Hepes 20 mM, 50 mM KCl, 1 mM NaH₂PO₄, 1 mM MgCl₂, and 1 mM EGTA, pH 7.4). The respiratory activity of mouse liver mitochondria was measured with a Clark-type electrode, in an all-glass reaction chamber magnetically stirred, at 25 °C. Where indicated, 5 μM rotenone/1 μM antimycin A and 0.4 mM TMPD/5 mM ascorbate were added. The rate of O₂ consumption (in units of nanomoles of O₂ per minute per mg of protein) was calculated 10 to 14 s after the TMPD/ascorbate injection. Values are mean ± SEM; statistics of control vs. 1 nM is are nonsignificant while control vs. 100 μM KCN is P ≤ 0.01, one-way ANOVA. (B) Time course of scission of the SS bond in 20 μM DTNB (Ellman’s reagent) by 2 mM total cyanide at 25 °C. The buffer used was 0.1 M Na-phosphate, pH 7. The data are represented as difference spectra relative to DTNB alone. The data were filtered by singular value decomposition (6) and reconstructed with 2 s values (s₁ = 4.3483, s₂ = 0.4910, s₂ = 0.0438). Analogous time courses were determined at 5, 8, and 11 mM total HCN. Time courses at 412 nm (slice through difference spectra) were fitted to an exponential process [$y=\alpha \left(1-e^{-k_{obs}t}\right)$]. A plot of k_obs, measured at different cyanide concentrations, was linear, with a slope equal to the second-order rate constant reported herein. All calculations were carried with MathWorks MATLAB.

The reduced basal glutathionylation of CcO is based on Western blot and Ellman’s reagent. As for the former, the claimed differences are barely visible and detailed methods on densitometric analysis are missing. In the latter, purified CcO is reduced with Tris (2-carboxyethyl) phosphine and, after removal, treated with 0.1 nM cyanide. Following cyanide removal, free cysteines are shown to increase. Thus, unless one or two disulfide bridges are not reduced and can react with cyanide the reported result is flawed. However, the putative target cysteine residues (conserved subunit I M-side C498 and nonconserved subunit III C-side C115) are solvent-exposed. From Protein Data Bank ID code 3ASO the distances between the Sγ atom of these residues from the closest prosthetic group is about 18 Å (heme a) and 36 Å (Cu_B), respectively. In view of these large distances, glutathione-induced spectral perturbation appears unlikely. Cyanide binding to the a₃-Cu_B site is readily probed by ultraviolet-visible spectroscopy. The spectra shown in Fig. 1J of ref. 1 appear to be consistent with cyanide binding to a_a3-CuB, at least at high cyanide concentration; also, the low-cyanide (green) spectrum appears identical to the control and no isosbestic point is seen throughout the titration owing to an increase in light scattering.

Second, it is well known that cyanide reacts with disulfides, yielding a free thiol and a thiocyanate (2). The reaction is bimolecular with second-order rate constants of 0.03 (cyanide + cystine) and 0.002 (cyanide + penicillamine-cysteine disulfide) M⁻¹⋅s⁻¹ in aqueous buffered solutions (2), and sluggish. We confirm this result by studying the scission kinetics of the SS bond in Ellman’s reagent by cyanide. The results are shown in Fig. 1B, in which time-resolved spectra are reported. The time course was exponential and the calculated second-order rate constant determined to be 10.0 ± 0.2 M⁻¹⋅s⁻¹. We are not aware of the reaction of cyanide with S-glutathionylated proteins including CcO. However, 1) the reactive species is the cyanide anion and 2) the free cyanide concentration is about 1.5% of total HCN at pH 7.4 (the pK_A of HCN being 9.2). Therefore, when using 0.1 nM HCN (or its salt) the concentration of the cyanide ion would be about 2 pM. Cyanide is known to bind to a₃-Cu_B with second-order rate constants of ca. 10³ M⁻¹⋅s⁻¹ in the resting and pulsed states and of ca. 10⁶ M⁻¹⋅s⁻¹ during turnover (3–5). These rate constants are two to five orders of magnitude higher than the corresponding rate constant measured here for 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB). Under the experimental conditions of ref. 1 the cyanide reaction would be complete in weeks and the partitioning of cyanide on a₃-Cu_B would exceed 99.9%.