Reassessment of 14C02 Compartmentation and of [14C]Formate Oxidation in Rat Liver*

Our previous report (Marsolais, C., Huot, S., David, F., Garneau, M., and Brunengraber, H. (1987) J. Biol. Chem. 262,2604-2607) had concluded that a fraction of [14C]formate oxidation in liver occurs in the mitochondrion. This conclusion was based on the labeling patterns of urea and acetoacetate labeled via 14C02 generated from [14C]formate and other ['4C]substrates. We reassessed our interpretation in experiments conducted in (i) perifused mitochondria and (ii) isolated livers perfused with buffer containing [14C]formate, ['4C]gluconolactone, 14C02, or NaH13C03, in the ab- sence and presence of acetazolamide, an inhibitor of carbonic anhydrase. Our data show that the cytosolic pools of bicarbonate and CO, are not in isotopic equi-librium when l4COZ is generated in the cytosol or is supplied as NaH14C03. We retract our earlier sugges- tion of a mitochondrial site of [14C]formate oxidation. the (SA)' the

' The abbreviations used are: SA, specific activity; tCO,, total CO,, Le. the sum of HCOi + CO, + H2C03; AcAc, acetoacetate; KIC, 2- When experiments were conducted in the presence of ["C] formate, we found that the SAs of urea and AcAc were increased in the presence of AZ. This was unexpected in the light of current concepts that ascribe the generation of 14C0, from [14C]formate to cytosolic methylene tetrahydrofolate dehydrogenase and peroxisomal catalase (5,6). In addition, in the presence of [14C]formate, the SA of C-1 of AcAc was double that of urea. This suggested channeling of I4CO2 derived from [14C]formate to mitochondrial methylcrotonyl-CoA carboxylase.
In the present study, we first investigated the oxidation of formate in isolated rat liver mitochondria. These were perifused with [14C]formate under conditions that stimulate citrulline synthesis from ornithine + NH: . In isolated mitochondria, the SA of the ureido carbon of citrulline reflects that of matrix bicarbonate.
Our data show that the cytosolic pools of CO, and bicarbonate are not in isotopic equilibrium, when labeled CO, is generated in the cytosol. We identified an error of interpretation of our previous data on [I4C]formate oxidation in liver. Therefore, we retract our earlier suggestion (1) of a mitochondrial site of formate oxidation in rat liver.
Before use, aliquots of [14C]formate stock were purified from traces of H"CO; as follows. The water ethanol solvent was evaporated with pure COZ. The tracer was dissolved in 0.5 ml of 10 mM Na2C03, and 0.02 ml of 1.2 M BaC1, was added to precipitate BaC03. After centrifugation, the supernatant was loaded onto a 0.5 X 3-cm AG50-H' column developed with 1 ml of water. The acidic effluent was neutralized with 1 N NaOH, counted, and used immediately. D-[l-"C]Glucose was converted to D-[l-'4C]g~uconolactone by reacting 0.5 mCi of tracer with 10 units of glucose oxidase and lo' units of catalase in 1 ml of 25 mM Tris buffer, p H 7.3. After 1-h incubation at 37 "C, pH was raised to 9. Fifteen min later, the reaction mixture was loaded onto a 0.5 X 1-cm column of AG-l-X8-C1 resin. The column was first developed with 6 X 0.5 ml of water to elute unreacted 19761 14C02 Compartmentation in Rat Liver glucose and then with 6 X 0.5 ml of HCI (0.04 N) to elute [1-"C] gluconolactone.
Liver Perfusions-Livers from fed or 2-day starved rats (Sprague-Dawley, Charles River) were perfused (10) with nonrecirculating Krebs-Ringer bicarbonate buffer (30 ml/min) equilibrated with 95% 0 2 + 5% cop and containing 1.2 mM NH;, 2.0 mM ornithine, 2.0 mM asparagine, 0.5 mM KIC and where indicated, (i) 0.25 mM formate + 1 mM methionine (11) or (ii) 10 mM glucose + 0.25 mM ghconolactone. In the latter case, gluconolactone (7.5 mM, pH = 3.0) was infused via a syringe pump just ahead of the liver, to minimize hydrolysis of the lactone. The pH of the perfusate was not affected by the infusion of the acidic stock solution of gluconolactone. In all perfusions, a 15-min equilibration period was allowed. Tracers were infused for 14 min, and AZ (0.2 mM) was added for the last 7-min period. Influent and effluent perfusates were sampled under NaOH during the 5th, 6th, 12th, and 13th min. All measured parameters were constant during the 4-to 7and 11to 13-min intervals of each tracer infusion. In some experiments, anterograde and retrograde perfusions were performed on the same liver.
Radioactive tracers were infused at rates calculated to achieve SAs in the range of 1 dpm/nmol for H"CO3 and I4CO2 and 1500 dpm/ nmol for ["Clformate. In experiments with ~-[l-~~C]gluconolactone, a mixed solution of tracer and substrate (7.5 mM, 700 dpm/nmol) replaced the unlabeled substrate infusion. "CO, was added to the inflowing perfusate by diffusion through a silastic tubing. In perfusions with [13C]bicarbonate, the concentration of unlabeled bicarbonate in the stock buffer was reduced to 22.5 mM, and 2.5 mM of pure NaH13C03 was infused into the inflowing perfusate to obtain a total bicarbonate concentration of 25 mM, and a molar percent enrichment (MPE) of 10%.

Analytical Techniques
Citrulline (12) and mitochondrial proteins (13) were assayed by standard colorimetric techniques. All assays of SAs were done as described previously (1). In perfusions with Hl3CO3, MPE of AcAc and urea were determined by gas chromatography-mass spectrometry analysis. Five ml of effluent samples were brought to pH 9-11 with NaOH and made 10 mM NaBH, to convert unstable AcAc to stable (R,S)-BHB. After 30 min at room temperature, the samples were saturated with NaCl and extracted four times with 15 ml of ethyl acetate. The pooled extracts containing urea were evaporated. The aqueous phase was acidified with HC1 to pH 1-2 and extracted three times with 10 ml of ethyl acetate. The pooled extracts containing BHB were evaporated. Both dry residues were incubated overnight at room temperature with 0.05 ml of N-methyl-N-(t-butyldimethyl- Analyses were performed on a Hewlett Packard 5890 mass spectrometer coupled with a gas chromatograph equipped with a HP-5 capillary column (30 m X 0.2 mm inner diameter, 0.33 pm film thickness; Ref. 14). Carrier gas was helium (0.8 ml/min), and the column head pressure was 20 kilopascals. The injection port was kept at 250 "C and the column temperature, initially set at 150 "C, was programmed to increase by 5 "C/min after a 3-min delay. tert-Butyldimethylsilyl derivatives of BHB and urea elute at about 8 and 9 min, respectively, after sample injection. The column was baked at 250 "C for 5 min between samples. The gas chromatography capillary column is directly connected to the mass detector which operates in the positive ion electron impact (70 eV) mode. For the determination of urea and BHB MPEs, the ions monitored are at m/z 231 and 232 corresponding to urea and [13C]urea and at m/z 275 and 276 corresponding to BHB and [1-13C]BHB.
MPE of tC02 in the influent perfusate was measured by a gas chromatography-mass spectrometry technique. Influent sample, 100 p1, was made 0.05 N NaOH in a glass vial (1 ml) capped with a gastight stopper. The air phase of the vial was purged with pure helium.
After acidification of the sample with 50 pl of 0.3 N H3P04, 10 p1 of the gas phase was injected manually into the gas chromatographymass spectrometry. The column temperature was kept at 200 "C and the gas injected eluted at 1 min. The ions monitored were a t m/z = 40, 44, and 45 corresponding to argon, C02, and %O2, respectively. Argon was monitored to identify samples contaminated with air. Areas under each fragmentogram were determined by computer integration.

RESULTS AND DISCUSSION
In perifused mitochondria from aminotriazole-treated rats, the rate of ["C]formate oxidation was 7 pmol/mg protein x min. This rate, equivalent to 1.7 nmol/g dry wt of liver X min, is only one-hundreth of the rate observed in the perfused livers (110-240 nmol/g dry weight x min). We ascribed this very low rate of [14C]formate oxidation to residual catalase activity in peroxisomes which contaminate most mitochondrial preparations (8). Since we could not demonstrate significant rates of [14C]formate oxidation in isolated mitochondria, we turned back to the perfused liver system. The first protocols were designed to test the hypothesis that part of formate oxidation occurs in mitochondria.
Data from liver perfusions with ['4C]formate f AZ (Table I, experiment 1) were similar to those (1) reported previously: (i) the SA of urea was equal to that of effluent tCOz, (ii) the SA of AcAc was 1.5 times that of urea, and (iii) in the presence of AZ, the SAs of AcAc and urea increased 6 to 8 times above that of bicarbonate. We tried to favor a putative mitochondrial site of ['4C]formate oxidation and thus to increase the differential labeling of urea and AcAc by inhibiting extramitochondrial formate oxidation with NzO (15) + 3-amino-1,2,4-triazole (9). Although the rate of formate oxidation was decreased by 70% as reported by others (16), there was no major change in the labeling patterns of urea, AcAc, and bicarbonate (Table   I, experiment 2). Note, however, that the SA of urea was 70% that of effluent tCO2.
In view of the above negative experiments, we suspected that we had made an error in the interpretation of some of our original data from perfused liver experiments. We had concluded that a fraction of [  Livers were perfused using conditions stimulating ureogenesis and AcAc production from 2-ketoisocaproate, with various labeled precursors in the absence and presence of 0.2 mM acetazolamide and under different conditions (see "Experimental Procedures"). Except where indicated, all perfusions were conducted in livers from 48 h starved rats in the anterograde mode at 37 "C. In experiment 4, 14C02 gas and NaH13C03 10% MPE were added to the inflowing perfusate just before it entered the liver. Sections A and C and B and D of experiment 4 refer to product labeling with "C and 13C, respectively. Data are presented as ratios of SA or MPE of urea/total CO, and of AcAc/ total COa in the effluent ( (1). Thus, we retract and apologize for our earlier suggestion (1) of a mitochondrial site of formate oxidation. We further investigated the different fates of cytosolic CO, and bicarbonate in the absence and presence of AZ. We perfused livers in anterograde and retrograde modes, with buffer that was enriched with 14C0, gas and NaH13C03, just before entering the organ. This protocol was designed to minimize the uncatalyzed isotopic equilibration of CO, and bicarbonate before these species enter the cytosol of liver cells. The SAs and MPEs of effluent urea and AcAc were compared to the average SA and MPE of tCO, measured in a sample of the influent perfusate (Table I, experiment 4). In the absence of AZ, the ratios (SA urea/SA tCOJ and (MPE urea/MPE tCOz) were different (1.24 uersus 0.89). The corresponding labeling ratios (AcAc/tCOz) were similarly different (1.24 uersus 0.78). This reflects an incomplete isotopic equilibration between the cellular pools of CO, and bicarbonate in spite of the presence of active carbonic anhydrase. In the presence of AZ, the ratios (SA urea/SA tCOz) and (SA AcAc/SA tCOJ increased 3-fold, whereas the ratios (MPE urea/MPE tC0,) and (MPE AcAc/MPE tCO,) decreased by We propose that the data from the above experiments and from our previous report can be explained by the compartmentation of CO, and bicarbonate shown in Fig. 1. One should keep in mind that at pH = 7.40, the concentration of bicarbonate is 20 times that of COz. In addition, the predominant bicarbonate does not diffuse through the mitochondrial membrane, unless it is converted to diffusible CO,. Finally, the equilibration between bicarbonate and CO, is maintained by two reactions: a rapid enzymatic reaction catalyzed by carbonic anhydrase and a slower spontaneous reaction. On these bases, let us discuss our data in the framework of Fig. 1.
The SA of 14C02 entering the cell or generated in the cytosol is greatly diluted by equilibration with the large pool of cytosolic bicarbonate before diffusing into the mitochondrion. There, the SA of I4CO, is further diluted by unlabeled C o n generated locally, as it equilibrates with mitochondrial bicarbonate. The latter is incorporated into urea and AcAc via 20-25%. carbamoyl phosphate synthetase and methylcrotonyl-CoA carboxylase. In the presence of AZ, which inactivates cytosolic carbonic anhydrase, dilution of the SA of cytosolic 14C02 is markedly decreased. Thus, the SA of 14C02 that diffuses into the mitochondrion as well as the SAs of urea and AcAc are increased. This occurs in spite of a decrease in the absolute rates of urea (1,19) and AcAc (1) synthesis caused by a limitation in the supply of mitochondrial bicarbonate.
The SA of 14C02 generated in the mitochondrion is diluted by unlabeled COz generated locally and diffusing from the cytosol. In the presence of AZ, diffusion of unlabeled CO, from the cytosol is decreased because of lower production from cytosolic bicarbonate. This leads again to an increase in the SAs of urea and AcAc.
The MPE of H13CO; entering the liver cell is slightly diluted by equilibration with the small pool of unlabeled CO,. Cytosolic COz becomes labeled and diffuses into the mitochondrion. There, it is further diluted by unlabeled C o n generated locally, as it equilibrates with mitochondrial bicarbonate. In the presence of AZ, labeling of cytosolic C02 from extracellular H13C03 is decreased, leading to a lowering in the MPE of urea and AcAc.
In the presence of AZ, the SAs of urea and AcAc labeled from l4CO2 increase 3-to 5-fold, whereas the MPE of urea and AcAc labeled from Hl3COi decrease by only 20-25%. In this and in the previous report (l), AZ induced large variations in the labeling of urea and AcAc when these are labeled from COz, the minor species of the C02/HCO; system. In contrast, when urea and AcAc are labeled from HCO;, the predominant species, much smaller variations in urea and AcAc labeling are observed. Thus, in the C02/HC0; system, the relative size of the pool to which label is added influences the extent to which label is diluted in the other species of the system. Note also that the scales of variation of a parameter (i) from 100 to 0% and (ii) from 100 to + infinity are not symmetrical.
Increases are thus much more striking than decreases. It is puzzling that in the presence of cytosolic sources of 14C02 ([14C]formate and [l-'4C]gluconolactone; Table I) and in the absence of AZ, the labeling ratio (AcAc /tC02) is greater than the labeling ratio (urea/tC02). In contrast, in the pres-  Table I of Ref. I), both labeling ratios were equal to 1. We cannot explain this discrepancy. It might involve some kind of channeling of l4COZ as has been described for other metabolic precursors such as urea cycle intermediates (20) and oxaloacetate in the tricarboxylic acid cycle (21). We recognize, however, that such an interpretation is speculative.
The above discussion of our data in relation to Fig. 1 requires the presence of an active carbonic anhydrase in the cytosol of liver cells. The liver is known to contain several carbonic anhydrase isoenzymes: carbonic anhydrase V in the mitochondrion (22), and carbonic anhydrase I1 and carbonic anhydrase I11 in the cytosol (23). It has been shown immunohistochemically that carbonic anhydrase 111, which is insensitive to AZ inhibition (24), occurs throughout the liver but in greater concentration around the perivenous hepatocytes (25). Carbonic anhydrase 11, also found throughout the liver, is present at much lower concentrations. Our results suggest the existence of a cytosolic AZ-sensitive carbonic anhydrase. This activity appears distributed throughout the liver lobule, since labeling patterns of urea and AcAc labeled from infused 14C02 + NaH13C03 were identical in livers perfused in anterograde and retrograde modes (Table I, experiment 4). A recent study has suggested the presence of carbonic anhydrase activity on the outside surface of the hepatocyte membrane (26). Our data (Table I, experiment 4) show that such extracellular carbonic anhydrase activity, if it indeed occurs, is insufficient to equilibrate the labeling of extracellular bicarbonate and CO,.
The present report illustrates errors of interpretation one can easily make when using labeled bicarbonate to trace the fate of labeled CO, generated inside the liver cell,