COQ9, a new gene required for the biosynthesis of coenzyme Q in Saccharomyces cerevisiae

Currently, eight genes are known to be involved in coenzyme Q6 biosynthesis in Saccharomyces cerevisiae. Here, we report a new gene designated COQ9 that is also required for the biosynthesis of this lipoid quinone. The respiratory-deficient pet mutant C92 was found to be deficient in coenzyme Q and to have low mitochondrial NADH-cytochrome c reductase activity, which could be restored by addition of coenzyme Q2. The mutant was used to clone COQ9, corresponding to reading frame YLR201c on chromosome XII. The respiratory defect of C92 is complemented by COQ9 and suppressed by COQ8/ABC1. The latter gene has been shown to be required for coenzyme Q biosynthesis in yeast and bacteria. Suppression by COQ8/ABC1 of C92, but not other coq9 mutants tested, has been related to an increase in the mitochondrial concentration of several enzymes of the pathway. Coq9p may either catalyze a reaction in the coenzyme Q biosynthetic pathway or have a regulatory role similar to that proposed for Coq8p.     

Genetic evidence for a multi-subunit complex in the O-methyltransferase steps of coenzyme Q biosynthesis

Coq3 O-methyltransferase carries out both O-methylation steps in coenzyme Q (ubiquinone) biosynthesis. The degree to which Coq3 O-methyltransferase activity and expression are dependent on the other seven COQ gene products has been investigated. A panel of yeast mutant strains harboring null mutations in each of the genes required for coenzyme Q biosynthesis (COQ1-COQ8) have been prepared. Mitochondria have been isolated from each member of the yeast coq mutant collection, from the wild-type parental strains and from respiratory deficient mutants harboring deletions in ATP2 or COR1 genes. These latter strains constitute Q-replete, respiratory deficient controls. Each of these mitochondrial preparations has been analyzed for COQ3-encoded O-methyltransferase activity and steady state levels of Coq3 polypeptide. The findings indicate that the presence of the other COQ gene products is required to observe normal levels of O-methyltransferase activity and the Coq3 polypeptide. However, COQ3 steady state RNA levels are not decreased in any of the coq mutants, relative to either wild-type or respiratory deficient control strains, suggesting either a decreased rate of translation or a decreased stability of the Coq3 polypeptide. These data are consistent with the involvement of the Coq polypeptides (or the Q-intermediates formed by the Coq polypeptides) in a multi-subunit complex. It is our hypothesis that a deficiency in any one of the COQ gene products results in a defective complex in which the Coq3 polypeptide is rendered unstable.     

The Saccharomyces cerevisiae COQ10 gene encodes a START domain protein required for function of coenzyme Q in respiration

Deletion of the Saccharomyces cerevisiae gene YOL008W, here referred to as COQ10, elicits a respiratory defect as a result of the inability of the mutant to oxidize NADH and succinate. Both activities are restored by exogenous coenzyme Q2. Respiration is also partially rescued by COQ2, COQ7, or COQ8/ABC1, when these genes are present in high copy. Unlike other coq mutants, all of which lack Q6, the coq10 mutant has near normal amounts of Q6 in mitochondria. Coq10p is widely distributed in bacteria and eukaryotes and is homologous to proteins of the "aromatic-rich protein family" Pfam03654 and to members of the START domain superfamily that have a hydrophobic tunnel implicated in binding lipophilic molecules such as cholesterol and polyketides. Analysis of coenzyme Q in polyhistidine-tagged Coq10p purified from mitochondria indicates the presence 0.032-0.034 mol of Q6/mol of protein. We propose that Coq10p is a Q6-binding protein and that in the coq10 mutant Q6 it is not able to act as an electron carrier, possibly because of improper localization.     

Coenzyme Q differentially modulates phospholipid hydroperoxide glutathione peroxidase gene expression and free radicals production in malignant and non-malignant prostate cells

The aim of this study was to investigate the role of coenzyme Q on the mRNA abundance of PHGPx and the reactive oxygen species (ROS) production in two different cell lines from human prostate, a line of non cancer cells (PNT2) and a line of cancer cells (PC3). Results showed that malignant cells markedly differ in their response to coenzyme Q compared to non-malignant cells, with no changes in PHGPx expression and greater ROS production. Furthermore coenzyme Q supplementation significantly lowered cell growth of the PC3 cancer line without affecting the PNT2. If these results are confirmed with additional experiments, it could represent a novel and interesting approach on the biomedical use of coenzyme Q10 in cancer therapy.     

Detection of suppressed maturation of the human COQ5 protein in the mitochondria following mitochondrial uncoupling by an antibody recognizing both precursor and mature forms of COQ5

Yeast Coq5p is required for the biosynthesis of coenzyme Q₆ (CoQ₆), but its human homolog has not been studied. We purified soluble recombinant human COQ5 protein under native conditions and generated an antibody recognizing both precursor and mature forms of COQ5. Mitochondrial localization of the mature form in 143B cells was demonstrated with this antibody. Moreover, a chemical uncoupler in a dose that suppressed CoQ₁₀ levels downregulated the mature form but augmented the precursor form of COQ5. The results that knockdown of the COQ5 gene reduced CoQ₁₀ levels further indicated the critical role of COQ5 in the biosynthesis of CoQ₁₀.     

A mutation in para-hydroxybenzoate-polyprenyl transferase (COQ2) causes primary coenzyme Q10 deficiency

Ubiquinone (coenzyme Q(10) or CoQ(10)) is a lipid-soluble component of virtually all cell membranes, where it functions as a mobile electron and proton carrier. CoQ(10) deficiency is inherited as an autosomal recessive trait and has been associated with three main clinical phenotypes: a predominantly myopathic form with central nervous system involvement, an infantile encephalomyopathy with renal dysfunction, and an ataxic form with cerebellar atrophy. In two siblings of consanguineous parents with the infantile form of CoQ(10) deficiency, we identified a homozygous missense mutation in the COQ2 gene, which encodes para-hydroxybenzoate-polyprenyl transferase. The A-->G transition at nucleotide 890 changes a highly conserved tyrosine to cysteine at amino acid 297 within a predicted transmembrane domain. Radioisotope assays confirmed a severe defect of CoQ(10) biosynthesis in the fibroblasts of one patient. This mutation in COQ2 is the first molecular cause of primary CoQ(10) deficiency.     

An analysis of the role of coenzyme Q in free radical generation and as an antioxidant.

The vital role of coenzyme Q in mitochondrial electron transfer and its regulation, and in energy conservation, is well established. However, the role of coenzyme Q in free oxyradical formation and as an antioxidant remains controversial. Demonstration of the existence of the semiquinone form of coenzyme Q during electron transport, coupled with recent evidence that hydrogen peroxide (but not molecular oxygen) may act as an oxidant of the semiquinone, suggests that the highly reactive OH. radical may be formed from the semiquinone. On the other hand, data exist implicating the Fe-S species as the source of electron transfer chain, free radical production. Additional data exist suggesting instead that the unpaired electron of the coenzyme Q semiquinone most likely dismutases superoxide radicals. These concepts and those arising from observations at several levels of organization including subcellular systems, intact animals, and human subjects in the clinical setting, supporting the concept of reduced coenzyme Q as an antioxidant, will be presented. The results of recent studies on the interaction between the two-electron quinone reductase--DT diaphorase and coenzyme Q10 will be presented. The possibility that superoxide dismutase may interact with reduced coenzyme Q, in conjunction with DT diaphorase inhibiting its autoxidation, will be described. The regulation of cellular coenzyme Q concentrations during oxidative stress accompanying aerobic exercise, resulting in increased protection from free radical damage, will also be presented.     

A tRNA(TRP) gene mediates the suppression of cbs2-223 previously attributed to ABC1/COQ8

The Saccharomyces cerevisiae gene ABC1 was originally isolated as a multicopy suppressor of a yeast strain harboring a mutation in a cytochrome b translational activator (cbs2-223). Based on this identification, Abc1p was postulated to activate the bc1 complex and function as a chaperone of cytochrome b. ABC1 was subsequently identified as COQ8 and found to be necessary for yeast coenzyme Q synthesis. In this work we show that a segment of yeast genomic DNA containing ABC1/COQ8 and neighboring genes suppresses the respiratory and Q-deficient phenotypes of the coq6 mutant, coq6-1. COQ6 is essential for yeast coenzyme Q biosynthesis. We show that a tRNA(TRP) gene located downstream of ABC1/COQ8 mediates suppression of the cbs2-223 and coq6-1 mutations, and each is identified here as containing UGA nonsense codons. The inability of ABC1/COQ8 to suppress the cbs2-223 allele in multicopy indicates it may not be a chaperone as previously reported.     

Molecular characterization of the human COQ5 C-methyltransferase in coenzyme Q10 biosynthesis

Coq5 catalyzes the only C-methylation involved in the biosynthesis of coenzyme Q (Q or ubiquinone) in humans and yeast Saccharomyces cerevisiae. As one of eleven polypeptides required for Q production in yeast, Coq5 has also been shown to assemble with the multi-subunit complex termed the CoQ-synthome. In humans, mutations in several COQ genes cause primary Q deficiency, and a decrease in Q biosynthesis is associated with mitochondrial, cardiovascular, kidney and neurodegenerative diseases. In this study, we characterize the human COQ5 polypeptide and examine its complementation of yeast coq5 point and null mutants. We show that human COQ5 RNA is expressed in all tissues and that the COQ5 polypeptide is associated with the mitochondrial inner membrane on the matrix side. Previous work in yeast has shown that point mutations within or adjacent to conserved COQ5 methyltransferase motifs result in a loss of Coq5 function but not Coq5 steady state levels. Here, we show that stabilization of the CoQ-synthome within coq5 point mutants or by over-expression of COQ8 in coq5 null mutants permits the human COQ5 homolog to partially restore coq5 mutant growth on respiratory media and Q₆ content. Immunoblotting against the human COQ5 polypeptide in isolated yeast mitochondria shows that the human Coq5 polypeptide migrates in two-dimensional blue-native/SDS-PAGE at the same high molecular mass as other yeast Coq proteins. The results presented suggest that human and Escherichia coli Coq5 homologs expressed in yeast retain C-methyltransferase activity but are capable of rescuing the coq5 yeast mutants only when the CoQ-synthome is assembled.     

The participation of coenzyme Q in free radical production and antioxidation

Published experimental data pertaining to the participation of coenzyme Q as a site of free radical formation in the mitochondrial electron transfer chain and the conditions required for free radical production have been reviewed critically. The evidence suggests that a component from each of the mitochondrial NADH-coenzyme Q, succinate-coenzyme Q, and coenzyme QH₂-cytochrome c reductases (complexes I, II, and III, most likely a nonheme iron-sulfur protein of each complex, is involved in free radical formation. Although the semiquinone form of coenzyme Q may be formed during electron transport, its unpaired electron most likely serves to aid in the dismutation of superoxide radicals instead of participating in free radical formation. Results of studies with electron transfer chain inhibitors make the conclusion dubious that coenzyme Q is a major free radical generator under normal physiological conditions but may be involved in superoxide radical formation during ischemia and subsequent reperfusion. Experiments at various levels of organization including subcellular systems, intact animals, and human subjects in theclinical setting, support the view that coenzyme Q, mainly in its reduced state, may act as an antioxidant protecting a number of cellular membranes from free radical damage.     

ADCK3, an ancestral kinase, is mutated in a form of recessive ataxia associated with coenzyme Q10 deficiency

Muscle coenzyme Q(10) (CoQ(10) or ubiquinone) deficiency has been identified in more than 20 patients with presumed autosomal-recessive ataxia. However, mutations in genes required for CoQ(10) biosynthetic pathway have been identified only in patients with infantile-onset multisystemic diseases or isolated nephropathy. Our SNP-based genome-wide scan in a large consanguineous family revealed a locus for autosomal-recessive ataxia at chromosome 1q41. The causative mutation is a homozygous splice-site mutation in the aarF-domain-containing kinase 3 gene (ADCK3). Five additional mutations in ADCK3 were found in three patients with sporadic ataxia, including one known to have CoQ(10) deficiency in muscle. All of the patients have childhood-onset cerebellar ataxia with slow progression, and three of six have mildly elevated lactate levels. ADCK3 is a mitochondrial protein homologous to the yeast COQ8 and the bacterial UbiB proteins, which are required for CoQ biosynthesis. Three out of four patients tested showed a low endogenous pool of CoQ(10) in their fibroblasts or lymphoblasts, and two out of three patients showed impaired ubiquinone synthesis, strongly suggesting that ADCK3 is also involved in CoQ(10) biosynthesis. The deleterious nature of the three identified missense changes was confirmed by the introduction of them at the corresponding positions of the yeast COQ8 gene. Finally, a phylogenetic analysis shows that ADCK3 belongs to the family of atypical kinases, which includes phosphoinositide and choline kinases, suggesting that ADCK3 plays an indirect regulatory role in ubiquinone biosynthesis possibly as part of a feedback loop that regulates ATP production.     

Yeast COQ4 encodes a mitochondrial protein required for coenzyme Q synthesis

The COQ4 gene coding for a component of the coenzyme Q biosynthetic pathway in the yeast Saccharomyces cerevisiae was cloned by a functional complementation of a Q-deficient mutant strain. Yeast coq4 mutant strains harboring the COQ4 gene on either single- or multicopy plasmids acquired the ability to grow on media containing a nonfermentable carbon source, synthesize Q(6), and respire. COQ4 encodes a polypeptide containing 335 amino acids with a calculated molecular mass of 38.6 kDa. By Western blot analysis with a specific antiserum, Coq4p was shown to peripherally associate with the matrix face of the mitochondrial inner membrane. The putative mitochondrial-targeting sequence present at the amino-terminus of the polypeptide efficiently imported it to mitochondria in a membrane-potential-dependent manner. Steady-state levels of COQ4 mRNA were increased during growth on glycerol-containing medium, in accordance with a function in Q biosynthesis. The function of Coq4p is unknown, although its presence is required to maintain a steady-state level of Coq7p, another component of the Q biosynthetic pathway. The results presented here, along with those available from literature, are discussed in light of the recently proposed existence of a multisubunit complex functioning in Q biosynthesis (A. Y. Hsu, T. Q. Do, P. T. Lee, and C. F. Clarke, 2000, Biochim. Biophys. Acta 1484, 287-297).     

Isolation and functional expression of human COQ3, a gene encoding a methyltransferase required for ubiquinone biosynthesis

The COQ3 gene in Saccharomyces cerevisiae encodes an O-methyltransferase required for two steps in the biosynthetic pathway of ubiquinone (coenzyme Q, or Q). This enzyme methylates an early Q intermediate, 3,4-dihydroxy-5-polyprenylbenzoic acid, as well as the final intermediate in the pathway, converting demethyl-Q to Q. This enzyme is also capable of methylating the distinct prokaryotic early intermediate 2-hydroxy-6-polyprenyl phenol. A full-length cDNA encoding the human homologue of COQ3 was isolated from a human heart cDNA library by sequence homology to rat Coq3. The clone contained a 933-base pair open reading frame that encoded a polypeptide with a great deal of sequence identity to a variety of eukaryotic and prokaryotic Coq3 homologues. In the region between amino acids 89 and 255 in the human sequence, the rat and human homologues are 87% identical, whereas human and yeast are 35% identical. When expressed in multicopy, the human construct rescued the growth of a yeast coq3 null mutant on a nonfermentable carbon source and restored coenzyme Q biosynthesis, although at lower levels than that of wild type yeast. In vitro methyltransferase assays using farnesylated analogues of intermediates in the coenzyme Q biosynthetic pathway as substrates showed that the human enzyme is active with all three substrates tested.     

Distribution of Coenzyme Q Homologues in Brain

Ubiquinone (coenzyme Q₁₀), in addition to its function as an electron and proton carrier in mitochondrial electron transport coupled to ATP synthesis, acts in its reduced form (ubiquinol) as an antioxidant, inhibiting lipid peroxidation in biological membranes and protecting mitochondrial inner-membrane proteins and DNA against oxidative damage accompanying lipid peroxidation. Tissue ubiquinone levels are subject to regulation by physiological factors that are related to the oxidative activity of the organism: they increase under the influence of oxidative stress, e.g. physical exercise, cold adaptation, thyroid hormone treatment, and decrease during aging. In the present study, coenzyme Q homologues were separated and quantified in the brains of mice, rats, rabbits, and chickens using high-performance liquid chromatography. In addition, the coenzyme Q homologues were measured in cells such as NG-108, PC-12, rat fetal brain cells and human SHSY-5Y and monocytes. In general, Q₁ content was the lowest among the coenzyme homologues quantified in the brain. Q₉ was not detectable in the brains of chickens and rabbits, but was present in the brains of rats and mice. Q₉ was also not detected in human cell lines SHSY-5Y and monocytes. Q₁₀ was detected in the brains of mice, rats, rabbits, and chickens and in cell lines. Since both coenzyme Q and vitamin E are antioxidants, and coenzyme Q recycles vitamins E and C, vitamin E was also quantified in mice brain using HPLC-electrochemical detector (ECD). The quantity of vitamin E was lowest in the substantia nigra compared with the other brain regions. This finding is crucial in elucidating ubiquinone function in bioenergetics; in preventing free radical generation, lipid peroxidation, and apoptosis in the brain; and as a potential compound in treating various neurodegenerative disorders.     

Interactions Between Ascorbyl Free Radical and Coenzyme Q at the Plasma Membrane

A role for coenzyme Q in the stabilization of extracellular ascorbate by intact cells has beenrecently recognized. The aim of this work was to study the interactions between reducedubiquinone in the plasma membrane and the ascorbyl free radical, as an approach to understandubiquinone-mediated ascorbate stabilization at the cell surface. K-562 cells stabilized ascorbateand decreased the steady-state levels of the semiascorbyl radical. The ability of cells to reduceascorbyl free radical was inhibited by the quinone analogs capsaicin and chloroquine andstimulated by supplementing cells with coenzyme Q₁₀. Purified plasma membranes also reducedascorbyl free radical in the presence of NADH. Free-radical reduction was notobserved inquinone-depleted plasma membranes, but restored after its reconstitution with coenzyme Q₁₀.Addition of reduced coenzyme Q₁₀ to depleted membranes allowed them toreduce the signalof the ascorbyl free radical without NADH incubation and the addition of an extra amount ofpurified plasma membrane quinone reductase further stimulated this activity. Reduction wasabolished by treatment with the reductase inhibitor p-hydroximercuribenzoate and by blockingsurface glycoconjugates with the lectin wheat germ agglutinin, which supports the participationof transmembrane electron flow. The activity showed saturation kinetics by NADH andcoenzyme Q, but not by the ascorbyl free radical in the range of concentrations used. Our resultssupport that reduction of ascorbyl free radicals at the cell surface involves coenzyme Qreduction by NADH and the membrane-mediated reduction of ascorbyl free radical.     

CK2-Mediated Hyperphosphorylation of Topoisomerase I Targets Serine 506, Enhances Topoisomerase I–DNA Binding,and Increases Cellular Camptothecin Sensitivity

Topoisomerase I is the target for a potent class of chemotherapeutic drugs derived from the plant alkaloid camptothecin that includes irinotecan and topotecan. In this study we have identified a novel site of CK2-mediated topoisomerase I (topo I) phosphorylation at serine 506 (PS506) that is relevant to topo I function and to cellular responses to these topo I-targeted drugs. CK2 treatment induced hyperphosphorylation of recombinant topo I and expression of the PS506 epitope, and resulted in increased binding of topo I to supercoiled plasmid DNA. Hyperphosphorylated topo I was approximately three times more effective than the basal phosphorylated enzyme at relaxing plasmid supercoils but had similar DNA cleavage activity once bound to DNA. The PS506 epitope was expressed in cancer cell lines with elevated CK2 activity, hyperphosphorylated topo I, and increased sensitivity to camptothecin. In contrast, PS506 was not detected in normal cells or cancer cell lines with lower levels of CK2 activity. By experimentally manipulating CK2 activity in cancer cell lines, we demonstrate a cause and effect relationship between CK2 activity, PS506 expression, camptothecin-induced cellular DNA damage, and cellular camptothecin sensitivity. Our results show that the PS506 epitope is an indicator of dysregulated, hyperphosphorylated topo I in cancer cells, and may thus serve as a diagnostic or prognostic biomarker and predict tumor responsiveness to widely used topo I-targeted therapies.     

Pathogenicity of two COQ7 mutations and responses to 2,4‐dihydroxybenzoate bypass treatment

Primary ubiquinone (co‐enzyme Q) deficiency results in a wide range of clinical features due to mitochondrial dysfunction. Here, we analyse and characterize two mutations in the ubiquinone biosynthetic gene COQ7. One mutation from the only previously identified patient (V141E), and one (L111P) from a 6‐year‐old girl who presents with spasticity and bilateral sensorineural hearing loss. We used patient fibroblast cell lines and a heterologous expression system to show that both mutations lead to loss of protein stability and decreased levels of ubiquinone that correlate with the severity of mitochondrial dysfunction. The severity of L111P is enhanced by the particular COQ7 polymorphism (T103M) that the patient carries, but not by a mitochondrial DNA mutation (A1555G) that is also present in the patient and that has been linked to aminoglycoside‐dependent hearing loss. We analysed treatment with the unnatural biosynthesis precursor 2,4‐dihydroxybenzoate (DHB), which can restore ubiquinone synthesis in cells completely lacking the enzymatic activity of COQ7. We find that the treatment is not beneficial for every COQ7 mutation and its outcome depends on the extent of enzyme activity loss.     

Mouse homologue of coq7/clk-1, longevity gene in Caenorhabditis elegans, is essential for coenzyme Q synthesis, maintenance of mitochondrial integrity, and neurogenesis.

coq7/clk-1 was isolated from a long-lived mutant of Caenorhabditis elegans, which showed sluggish behavior and an extended life span. Mouse coq7 is homologous to Saccharomyces cerevisiae coq7/cat5 that is required for biosynthesis of coenzyme Q (CoQ), an essential cofactor in mitochondrial respiration. Here we generated COQ7-deficient mice to investigate the biological role of COQ7 in mammals. COQ7-deficient mouse embryos failed to survive beyond embryonic day 10.5, exhibiting small-sized body and delayed embryogenesis. Morphological studies showed that COQ7-deficient neuroepithelial cells failed to show the radial arrangement in the developing cerebral wall, aborting neurogenesis at E10.5. Electron microscopic analysis further showed the enlarged mitochondria with vesicular cristae and enlarged lysosomes filled with disrupted membranes, which is consistent with mitochondriopathy. Biochemical analysis demonstrated that COQ7-deficient embryos failed to synthesize CoQ(9), but instead yielded demethoxyubiquinone 9 (DMQ(9)). Cultured embryonic cells from COQ7-deficient mice were rescued by adding bovine fetal serum in vitro, but exhibited slowed cell proliferation, which resembled to the phenotype of clk-1 with delayed cell divisions. The result implied the essential role of coq7 in CoQ synthesis, maintenance of mitochondrial integrity, and neurogenesis in mice.     

Genetic Evidence for Coenzyme Q Requirement in Plasma Membrane Electron Transport

Plasma membranes isolated from wild-type Saccharomyces cerevisiae crude membrane fractions catalyzed NADH oxidation using a variety of electron acceptors, such as ferricyanide, cytochrome c, and ascorbate free radical. Plasma membranes from the deletion mutant strain coq3, defective in coenzyme Q (ubiquinone) biosynthesis, were completely devoid of coenzyme Q₆ and contained greatly diminished levels of NADH–ascorbate free radical reductase activity (about 10% of wild-type yeasts). In contrast, the lack of coenzyme Q₆ in these membranes resulted in only a partial inhibition of either the ferricyanide or cytochrome-c reductase. Coenzyme Q dependence of ferricyanide and cytochrome-c reductases was based mainly on superoxide generation by one-electron reduction of quinones to semiquinones. Ascorbate free radical reductase was unique because it was highly dependent on coenzyme Q and did not involve superoxide since it was not affected by superoxide dismutase (SOD). Both coenzyme Q₆ and NADH–ascorbate free radical reductase were rescued in plasma membranes derived from a strain obtained by transformation of the coq3 strain with a single-copy plasmid bearing the wild type COQ3 gene and in plasma membranes isolated form the coq3 strain grown in the presence of coenzyme Q₆. The enzyme activity was inhibited by the quinone antagonists chloroquine and dicumarol, and after membrane solubilization with the nondenaturing detergent Zwittergent 3–14. The various inhibitors used did not affect residual ascorbate free radical reductase of the coq3 strain. Ascorbate free radical reductase was not altered significantly in mutants atp2 and cor1 which are also respiration-deficient but not defective in ubiquinone biosynthesis, demonstrating that the lack of ascorbate free radical reductase in coq3 mutants is related solely to the inability to synthesize ubiquinone and not to the respiratory-defective phenotype. For the first time, our results provide genetic evidence for the participation of ubiquinone in NADH–ascorbate free radical reductase, as a source of electrons for transmembrane ascorbate stabilization.