Status of the Mitochondrial Pool of Glutathione in the Isolated Hepatocyte*

Using the selective membrane-solubilizing properties of digitonin and a rapid centrifugation method to separate cytoplasmic and mitochondrial components, the metabolic state of mitochondrial glutathione was in-vestigated in isolated rat hepatocytes. T w o pools of GSH were released from hepatocytes incubated with increasing concentrations of digitonin. The largest pool (about 85% of cellular total) was released simultane-ously with lactate dehydrogenase, the other pool with citrate synthase, indicating cytoplasmic and mitochon- drial locations, respectively. The tlIz of the mitochondrial pool was estimated by linear regression analysis to be 30 f 3 h, while the cytoplasmic pool turned over with a tllz of about 2 f 0.1 h. The rate of incorporation of [36S]methionine or cysteine into the cytoplasmic pool of GSH, when corrected for turnover, was 15 times greater than into the mitochondrial pool. Mitochondrial GSH was not depleted after 60 min with 185 PM diethyl maleate with or without 75 FM bis-l,3-(2-~hloroethyl)- 1-nitrosourea, a specific inhibitor of glutathione reductase, whereas cytoplasmic levels were reduced to 40% and 10% of control values, respectively. In vivo experiments, using ~-(aS,5S)-a-amino-3-chIoro-4,5-dihydro-5- isoxazoleacetic acid to inactive y-glutamyl transpeptidase to limit cysteine formation from plasma GSH, demonstrated that in the absence of label reincorporation, NaHC03, prepared for pressure liquid chromatography analysis as described earlier (17). Specific radioactivity glutathione was determined by counting I-ml fractions of the high pressure liquid chromatography eluate in 3 ml of Formula 963 scintillation fluid (New England Nuclear). In experiments measuring 35S-labeled glutathione, hepatocytes were supplied with 0.67 mM unlabeled methionine or 0.16 mM cysteine. Amino acid concentration contribution made by labeled material was negligible. Radiochemical purity of sulfur amino acids was determined prior each by thin layer chromatography on Avicel plates 2-propano1:formic acidH20 Carrier amino acid and %labeled compound were co-chromatographed and the ninhydrin-positive scraped the plate and specific radioactivity

Using the selective membrane-solubilizing properties of digitonin and a rapid centrifugation method to separate cytoplasmic and mitochondrial components, the metabolic state of mitochondrial glutathione was investigated in isolated rat hepatocytes. T w o pools of GSH were released from hepatocytes incubated with increasing concentrations of digitonin. The largest pool (about 85% of cellular total) was released simultaneously with lactate dehydrogenase, the other pool with citrate synthase, indicating cytoplasmic and mitochondrial locations, respectively. The tlIz of the mitochondrial pool was estimated by linear regression analysis to be 30 f 3 h, while the cytoplasmic pool turned over with a tllz of about 2 f 0.1 h. The rate of incorporation of [36S]methionine or cysteine into the cytoplasmic pool of GSH, when corrected for turnover, was 15 times greater than into the mitochondrial pool. Mitochondrial GSH was not depleted after 60 min with 185 PM diethyl maleate with or without 75 FM bis-l,3-(2-~hloroethyl)-1-nitrosourea, a specific inhibitor of glutathione reductase, whereas cytoplasmic levels were reduced to 40% and 10% of control values, respectively. In vivo experiments, using ~-(aS,5S)-a-amino-3-chIoro-4,5-dihydro-5isoxazoleacetic acid to inactive y-glutamyl transpeptidase to limit cysteine formation from plasma GSH, demonstrated that in the absence of label reincorporation, liver glutathione exhibits a biphasic turnover. The rates of decay (half-lives) and percentages of total GSH under these conditions correlate well with the half-lives and pool distribution seen in the mitochondrial and cytoplasmic populations of GSH found in the isolated hepatocytes.
The presence of a discrete mitochondrial pool of glutathione was proposed by Vignais and Vignais (1) and later confirmed (2, 3) by Jocelyn and co-workers. Working with isolated rat mitochondria, they demonstrated that an amount equivalent to about 10% of the total hepatic glutathione was found in the mitochondria with about 90% as reduced glutathione (2).
Wahllander et al. (4), using a nonaqueous extraction procedure, found slightly higher glutathione levels, 13% of total liver content. These data suggested that the mitochondrion maintains a higher glutathione concentration than the cytoplasm, 10 mM uersus 7 mM, respectively, calculated on basis of compartment water space (4, 5). Other investigators (6), however, have found much lower concentrations of glutathione in the mitochondria, 4 to 5 mM. *This work was supported by Grant ES-10978 awarded by the National Institute of Environmental Health Services and by National Research Service Award AM 05876. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. It has been noted by several workers (3,4,6 ) that the glutathione content of isolated mitochondria is not significantly affected by brief incubations, in vitro. This presumed impermeability of at least the inner mitochondrial membrane has led to speculation that mitochondria maintain intramitochondrial glutathione by in situ synthesis (4). The absence of any documented mitochondrial glutathione transport system adds tacit support to this supposition. Jocelyn (3) observed the apparent efflux of GSSG from isolated mitochondria in response to the respiratory-dependent swelling effects of phosphate. However, the loss of GSSG was attributed to abnormal alteration of the membrane properties.
The selective membrane solubilization properties of digitonin in combination with sedimentation through an inert oil have been employed to measure cytoplasmic and mitochondrial metabolites and enzyme levels (7-10). The demonstration by Brock et al. (11) that this methodology does not significantly alter the distribution or concentration of mitochondrial metabolites further supports the applicability of the techniques to the measurement of mitochondrial glutathione. We have applied these methods to effect a rapid separation of the mitochondrial and cytoplasmic glutathione, thus allowing assessment of the glutathione status of the mitochondria only briefly removed from the intact hepatocyte. In this report, data are presented showing the mitochondrial glutathione pool in the isolated hepatocyte to be metabolically separate from that in the cytoplasm with regard to both synthetic rate, turnover, and sensitivity to chemical depletion. In addition, experiments evaluating the role of y-glutamyltranspeptidase in the reclamation of the cysteine moiety of glutathione and subsequent re-utilization for glutathione synthesis are presented. These experiments suggest that the "stable pool" of liver glutathione observed by Higashi et al. (12) may have been the result of kidney reabsorption of cyst(e)ine derived from the long lived mitochondrial pool of glutathione.

MATERIALS AND METHODS
Rat hepatocytes were prepared from 150-to 200-g male Sprague-Dawley rats as previously described (13) and incubated in Fisher's medium deficient in sulfur amino acids unless stated otherwise. Incubation was carried out at 37 "C in a gyratory shaker with each 25ml flask maintained under a constant flow of water-saturated 95% 0 2 , 5% COZ. Cell viability assessed by trypan blue exclusion and glutathione reductase activity were measured as described by Babson et al. (14). Lactate dehydrogenase leakage was measured with a Beckman TR analyzer according to the method of Lindstrom et al. (15). Citrate synthase was assayed by the method of Srere (16). Glutathione was measured as described by Reed et al. (17). Total lactate dehydrogenase and citrate synthase were determined in cell extracts prepared by sonication of hepatocytes for 10 s in the presence of 0.1% (v/v) Triton X-100.
Isolation of Mitochondrial and Cytoplasmic Components-Digitonin disruption was done by a modification of the method of Zuurandank and Tager (7). Hepatocyte suspensions were adjusted to contain: digitonin, 0.1 to 1 mg/n& EDTA, 2.5 mM; mannitol, 250 mM; Disrupted hepatocytes, 1 to 3 X lo6 cells in 0.7 ml , were layered onto 0.36 ml of dibutyl phthalate over 0.5 ml of 20% (v/v) perchloric acid in a 1.8-ml microcentrifuge tube. After incubation at room temperature for 30 s, centrifugation for 3 min at 13,000 X g at room temperature yielded rapid, quantitative sedimentation of the intact mitochondria into the perchloric acid, while cytoplasmic components remained above the oil layer. To allow recovery of mitochondrial enzymes, 40% (v/v) glycerol was substituted for perchloric acid.
Measurement of Liver Glutathione Turnouer-Male Sprague-Dawley rats, 200 to 215 g, were fasted 12 h before intraperitoneal injection of 100 pCi of [35S]methionine. After 4 h, a l l rats were given 100 mmol/kg of unlabeled methionine and half were treated with 0.5 mmol/kg of AT-125. Both chemicals were dissolved in sterile 0.9% NaCl and injected intraperitoneally. At various times, rats were killed by decapitation, livers rapidly excised and rinsed in ice-cold saline, then frozen with liquid nitrogen and stored at -80 "C until analyzed. For glutathione analysis, livers were weighed and homogenized, while still frozen, in 2 volumes of 5% (v/v) perchloric acid, centrifuged 10 min at 15,000 X g at 4 "C and the supernatant neutralized with excess NaHC03, then prepared for high pressure liquid chromatography analysis as described earlier (17). Specific radioactivity of glutathione was determined by counting I-ml fractions of the high pressure liquid chromatography eluate in 3 ml of Formula 963 scintillation fluid (New England Nuclear).
In experiments measuring 35S-labeled glutathione, hepatocytes were supplied with 0.67 mM unlabeled methionine or 0.16 mM cysteine. Amino acid concentration contribution made by labeled material was negligible. Radiochemical purity of sulfur amino acids was determined prior to each experiment by thin layer chromatography on Avicel plates developed in 2-propano1:formic acidH20 (7010:20). Carrier amino acid and %labeled compound were co-chromatographed and the ninhydrin-positive spots scraped from the plate and the specific radioactivity determined.

Zelease of Glutathione and Marker Enzyme by Digitonin
Disruption-Selective disruption of plasma and mitochondrial membranes on the basis of cholesterol content was possible by varying the digitonin concentration in the absence of the osmolarity-preserving agents, sucrose or mannitol. By careful attention to incubation time and conditions, a reproducible release profiie ( Fig. 1) was obtained for glutathione, the cytoplasmic marker lactate dehydrogenase and the mitochondrial matrix enzyme citrate synthase. About 85 to 90% of the total hepatocyte glutathione was released at low digitonin levels, tracing closely the release profiie of lactate dehydrogenase. In addition, all hepatocyte catalase was released coincident with the rapidly appearing cytoplasmic components (data not shown). A second pool of glutathione appeared with the mitochondrial marker at a higher digitonin concentration, confirming the feasibility of resolving the mitochondrial and cytoplasmic pools of glutathione by selective membrane disruption.
As had been previously accomplished with chick liver cells and 3T3-Ll and C-2 fibroblasts (8, lo), digitonin disruption in the presence of mannitol to preserve mitochondrial integrity was combined with sedimentation through a layer of dibutyl phthalate to produce an essentially cytoplasm-free mitochondrial preparation. As seen in Table I contamination of the mitochondrial pellet was observed. Additionally, the data in Table I  (1 X lo6 cells/&) were resuspended after centrifugation at 100 X g for 2 min in buffer containing varying amounts of digitonin in the absence of mannitol. After incubation for 2 min, cell debris was removed and marker enzymes and glutathione assayed as described under "Materials and Methods." TABLE I Separation of mitochondrial and cytoplasmic compartments of digitonin disrupted hepatocytes with dibutyl phthalate Isolated hepatocytes (1 X 106/ml) were pelleted by centrifugation at 80 x g and resuspended in 1 ml of buffer containing: digitonin as noted, 2.5 m~ EDTA, 250 m~ mannitol, 17 m~ MOPS, pH 7.4.
Cytoplasmic and mitochondrial compartments were separated as described under "Materials and Methods." Each entry represents the mean of two assays from duplicate samples and standard deviation was less than 1% throueh three exoeriments. activity found in samples sonciated in 0.1% Triton X-100 (total "Enzymatic activity remaining above oil layer as a percent of enzyme activity.
Enzyme activity found in glycerol layer as a per cent of total enzyme activity.
Cells reconstituted in 0.9% NaCl. Cells reconstituted in above buffer without digitonin. After a 2-h labeling period, cells were collected by centrifugation at 100 X g for 2 min, washed twice in Fisher's medium containing 0.67 m~ methionine, and resuspended (2 X IO6 cells/ml) in the same medium. Samples were taken and prepared for glutathione analysis as described under "Materials and Methods." Samples were taken at the times after treatment shown and rapidly prepared for compartment separation and analysis of glutathione as described under "Materials and Methods." Open symbols represent mitochondrial glutathione values.

TABLE I1
Inactivation of hepatocyte glutathione reductase by diethyl maleate and BCNU Hepatocytes (2 X 10" cells/ml) were suspended in Fischer's medium deficient in sulfur amino acid and treated with 185 PM diethyl maleate and, where designated, 75 p~ BCNU. Samples were taken and, following digitonin disruption, mitochondrial and cytoplasmic glutathione reductase were separated as described under "Materials and Methods." Data represent the summation of three experiments with duplicate assays of replicate flasks for each entry.  as reduced glutathione, at a rate equivalent to the loss of label from the cytoplasm (data not shown). The observed rate of glutathione efflux was approximately 7 nmol/h/106 cells. The synthetic rates for the compartmentalized glutathione pools were also found to be markedly different. As shown in Fig. 3, [35S] from methionine was incorporated into cytoplasmic glutathione much more rapidly than into the mitochondrial pool. When cysteine was used as the sulfur source, cytoplasmic and mitochondrial glutathione specific activities were 52 f 3 and 11 -+ 2% of medium cysteine specific activity, respectively, after 2 h of incubation. These values are not significantly different than those observed for methionine incorporation.
Differential Stability of Glutathione Pools during Chemical Intoxication-To measure the stability of the glutathione pools, diethyl maleate was used as a GSH-depleting agent. DEM rapidly (less than 10 min) reduced cytoplasmic glutathione levels while mitochondrial concentration did not vary significantly from control values (Fig. 4). Addition of bis-1,3-(2-chloroethyl)-l-nitrosourea, an antineoplastic-carbamylating agent which is known to inactivate glutathione reductase (14.18) and to enhance glutathione depletion by DEM, caused a more extensive loss of cytoplasmic glutathione than DEM alone. However, mitochondrial GSH levels were unchanged after a combined treatment with DEM and BCNU. The presence of BCNU in the mitochondrial compartment was c o n f i e d by measuring glutathione reductase levels. As seen in Table 11, mitochondrial and cytoplasmic glutathione reductase activity is dramatically reduced by BCNU treatment. Mitochondrial glutathione was found to be depleted in the intact hepatocyte by addition of ethacrynic acid, a penetrating thiol reagent found to decrease GSH levels in isolated mitochondria (6). If was found that addition of increasing concentrations of ethacrynic acid led to a proportionate loss of hepatocyte glutathione (Fig. 5, A and B ). These depressed levels were maintained in the absence of the glutathione precursors, cysteine or methionine. Depletion of glutathione by ethacrynic acid was found to affect significantly lactate dehydrogenase leakage only at the concentration causing total depletion of the mitochondrial glutathione pool (Fig. 6).
Effect of Inhibition of y-Glutamyl Transpeptidase on Liver GSH Turnover-After intraperitoneal administration of 0.5 mmol/kg body weight of AT-125, liver glutathione was measured in perchloric acid-deproteinated homogenates. Fig. 7 reveals the effect of blocking transpeptidase activity and, therefore, cysteine recovery on hepatic glutathione content. Over an 8-h period, glutathione decreased to 40% of the zero time level, before GSH content began to recover. Control values remained constant at about 4 mmol/g wet weight of liver. Determination of the specific radioactivity of the liver glutathione which had been prelabeled with [35S]methionine, showed in the control animals a rapid decrease, tlIz = 2 h, of GSH specific radioactivity to about 50% of the zero time value. The remaining one-half of GSH decayed with an apparent half-life of 36 h (Fig. 8). These decay times and apparent pool proportions agreed well with those seen by Higashi et al. (12) in similar experiments using [35S]cysteine to label liver GSH. The majority of the radioactive GSH (85%) in AT-125-treated livers was lost by 6 h, disappearing with a t1,2 of 2 h. The tII2 for the stable glutathione pool was 38.5 h.

DISCUSSION
As demonstrated for the isolated chick liver cell (9), selective solubilization of plasma and organelle membranes by digitonin complexation of cholesterol can be used to localize enzymes and metabolites within the rat hepatocyte. The clear displacement of appearance of the mitochondrial marker, citrate synthase, from the cytoplasmic components (see Fig.  1) makes isolation of the mitochondrial fraction very quick and efficient. When combined with a centrifugal separation through dibutyl phthalate, a rapid, quantitative examination of mitochondrial metabolites was possible. Brock et al. (ll), Zuurendank and Tager (7) and others (8-10) have shown, with numerous isolated cell systems, that the speed of isolation allowed no significant redistribution of organelle enzymes or metabolites. The advantages of this method are maximally exploited when mitochondrial components are centrifuged into perchloric acid, thus quenching metabolism within 30 to 45 s of the initial sampling.
The suggestion of multiple, organelle-bound pools of glutathione is not new. Using data from whole animal experiments, Edwards and Westerfeld (19) in 1952 proposed the existence of "stable" and "metabolized" pools of GSH, based on observations of nutritional effects on liver content. Barford and Eden (20) proposed that the majority of cellular GSH is contained in the cytoplasm.
Compelling evidence supporting the contention that the mitochondrial glutathione pool is not only spacially separate, but metabolically distinct from the cytoplasm was provided by the data in Figs. 2 and 3. In both synthesis and degradation, the rates of mitochondrial glutathione metabolism are significantly lower than those in the cytoplasm of isolated hepatocytes. The rate of loss of labeled cytoplasmic glutathione into the medium, measured to be about 7 nmol/106 hepatocytes/ h, agrees closeiy with rates of glutathione efflux observed previously (21). The low mitochondrial rate of glutathione turnover, estimated at about 0.05 nmol/106 hepatocytes/h, indicated that the two pools are not free to equilibrate. The impermeability of the mitochondrial membrane to GSH is suggested in experiments by Jocelyn (3) with isolated rat liver mitochondria. The independence of the mitochondrial pool of GSH was further demonstrated by the label incorporation studies shown in Fig. 3. As shown here, the rate of synthesis of mitochondrial glutathione was much lower than of glutathione in the cytoplasm. The finding that glutathione pools of similar compartmental concentration, but very different specific activity were maintained during these experiments suggests that the mitochondrial and cytoplasmic supplies of glutathione are not free to equilibrate. In addition, these data also suggest that the mitochondrial supply of glutathione is synthesized in situ.
The independent nature of the mitochondrial pool of glutathione was further demonstrated by the experiments involving glutathione depletion with diethyl maleate and BCNU. Several reports of GSH depletion by these agents are present in the literature (14,(21)(22)(23), and in each study, depletion proceeded only to about 10 to 15% of total glutathione, even at cytotoxic concentrations of depleting reagent. As seen in Fig. 4, neither DEM nor DEM plus BCNU induced a decline in mitochondrial GSH content. The failure of BCNU to decrease the level of mitochondrial GSH cannot be ascribed to insufficient intraorganelle concentration of either BCNU or its isocyanate degradation product. Mitochondrial glutathione reductase activity was inactivated by greater than 90% within 60 min of BCNU addition (Table 11). Whether this represents biotransformation of the nitrosourea allowing mitochondrial entry of the carbamylating moiety while retaining alkylating activity, but not the ability to stimulate depletion of GSH, or some other mechanism, is currently under investigation. It is also signifcant to note that under conditions of total cytoplasmic glutathione depletion, no efflux of GSH from the mitochondria (as registered by a drop in compartment concentration) was observed.
Depletion of mitochondrial glutathione by ethacrynic acid was found to be time-and concentration-dependent (Fig. 5B), although somewhat slower than the observed loss of cytoplasmic glutathione (Fig. 5A). In experiments where glutathione resynthesis was permitted by inclusion of cysteine or methionine, no effect was noted on lactate dehydrogenase leakage, although mitochondrial glutathione was transiently reduced below detectable levels (data not shown). However, when no sulfur amino acids were supplied, 0.2 m~ ethacrynic acid caused a marked increase in lactate dehydrogenase appearance in the medium (Fig. 6). Examination of Figs. 5 and 6 suggests that reduction of cytoplasmic glutathione to less than 5% of normal had no significant effect on lactate dehydrogenase leakage. These data c o n f m earlier observations made in this laboratory (14,24) that depletion of glutathione by DEM and BCNU does not induce cellular leakage of lactate dehydrogenase. Only after the mitochondrial pool of glutathione had been eliminated (after 60 min in Figs. 5 and 6) did lactate dehydrogenase leakage increase significantly.
The profound effect exerted by the small mitochondrial pool of glutathione on cell viability may well be related to the control of hydrogen peroxide formation, and ultimately lipid peroxidation, through the mitochondrial glutathione peroxidase/reductase system. Perfused rat liver mitochondria produce, as shown by Oshino et al. (25), about 50 nmol of H202/ g/min, representing 1.7% of the liver respiration rate. Interference with the mitochondrial ability to deal with the high localized peroxide levels would give rise to higher cytoplasmic HzOz levels, clearly presenting a peroxidative threat to membrane lipids.
Experiments with isolated hepatocytes demonstrated the independence of the two pools of glutathione measured as well as differences in half-life. The studies using AT-125 to block kidney reabsorption of the cysteine moiety of glutathione further delineate the defined nature of these stores of GSH. Previous studies (26) have shown AT-125 to inhibit effectively glutathione cleavage by y-glutamyl transpeptidase. Such inhibition would prevent reincorporation of labeled cysteine into glutathione, allowing an accurate description of the kinetics of glutathione turnover. The importance of cysteine reabsorption is emphasized by Fig. 8. When the loss of labeled liver glutathione was measured without AT-125 treatment, a decay curve essentially identical with that observed by Higashi et al. (12) was produced. Although the half-lives for the two discrete decay segments, the labile and stable pools, are comparable to those measured for the cytoplasmic and mitochondrial pools, the reported quantity of the stable pool, 50%, was much higher than could be accounted for as mitochondrial glutathione, based on hepatocyte experiments. After AT-125 addition, the rapid loss of label continued in a linear manner to a point equivalent to the level of the hepatocyte mitochondrial pool of glutathione. This evidence suggests that the larger, apparently stable pool observed here in control experiments and by others (12) is the result of kidney reabsorption of radioactive cysteine, derived from degraded glutathione to provide the liver and possibly other organs with recovered cysteine for glutathione synthesis.
The data of Reed and Ellis (26,27) predict excretion of glutathione in the urine of about 60 pmol/6 h by a 175-g rat, thus removing approximately 10 pmol of cysteine/h. Assuming a liver weight of 12 g, and 1.2 X 10' hepatocytes/g of liver (28), a cysteine supply of 7.1 nmol/106 hepatocytes/h is available from glutathione degradation. Given a glutathione content of 40 nmol/106 hepatocytes, and a ti,2 of 2 h for the cytoplasmic pool, 8.5 nmol of glutathione/h/106 hepatocytes must be replaced by new synthesis to maintain a constant liver glutathione concentration. The importance of reabsorbed cysteine, derived from degradation of extrahepatic glutathione is then proportional to the plasma level of the amino acid. Reliable values for plasma cysteine are difficult to obtain, due to oxidation during sample preparation and analysis. Frimpter et al. (29) have published plasma clearance studies with cysteine and cystine concentrations of 4 and 22 p~, respectively. Clearly, recirculation of 7 nmol/h of recovered cysteine represents a significant quantity, about 15% of the plasma cysteine supply. In the case of a radioactive cysteine-labeled population of liver glutathione, and neglecting radioactive cysteine released by protein turnover, kidney reabsorbtion provides cysteine at approximately 15% of the original glutathione specific activity, thus artificially lengthening the observed half-life of liver glutathione by allowing reincorporation of radioactive cysteine.
The observation by Crawhall and Segal (30) that cystine is the predominant form of cyst(e)ine in the plasma suggests cysteine uptake must be very efficient. Previous studies with hepatocytes (21) and certain other cell types' have shown that both uptake and glutathione synthesis is greater than 10-fold more efficient for cysteine than cystine. Although a small amount of plasma cystine is reduced (30), these studies support the contention that once oxidized, cysteine equivalents may be largely lost to glutathione synthesis occurring primarily in the liver. Considering the sporadic nature of the dietary cysteine supply, a constant supply of cysteine equivalents from glutathione may be of great importance in maintaining liver glutathione levels. The recent finding that human plasma GSH is rapidly removed from circulation, exhibiting a halflife of 1.6 min (31), is in agreement with these observations.
Recently, Lauterburg and Mitchell (31) concluded that no kinetically distinct pools of glutathione were found in the liver. The specific activities of glutathione trapped as the acetaminophen conjugate and excreted in the bile as well as hepatic glutathione were measured and found to be identical, within a standard deviation of 11%. It should be pointed out that their experiments allowed insufficient time to label significantly the mitochondrial glutathione pool. Additionally, considering that the mitochondrial pool ranges, depending on the method of measurement, between 5 and 15% of total hepatocyte glutathione, it is possible their experimental procedures overlooked the small but highly important contribution of mitochondrially located glutathione.