Compartmentation of Metabolism of the C12-, C9-, and C5-n-dicarboxylates in Rat Liver, Investigated by Mass Isotopomer Analysis

Background: We explored the compartmentation of n-dicarboxylate liver catabolism. Results: Dodecanedioate and azelate oxidation span peroxisomes and mitochondria; glutarate oxidation is mitochondrial; dodecanedioate is anaplerotic. Conclusion: Dodecanedioate is a potential substrate for anaplerotic therapy of reperfusion injury and some inborn disorders of metabolism. Significance: This metabolomic + mass isotopomer strategy can be used to better characterize substrate catabolism in disease models. We investigated the compartmentation of the catabolism of dodecanedioate (DODA), azelate, and glutarate in perfused rat livers, using a combination of metabolomics and mass isotopomer analyses. Livers were perfused with recirculating or nonrecirculating buffer containing one fully 13C-labeled dicarboxylate. Information on the peroxisomal versus mitochondrial catabolism was gathered from the labeling patterns of acetyl-CoA proxies, i.e. total acetyl-CoA, the acetyl moiety of citrate, C-1 + 2 of β-hydroxybutyrate, malonyl-CoA, and acetylcarnitine. Additional information was obtained from the labeling patterns of citric acid cycle intermediates and related compounds. The data characterize the partial oxidation of DODA and azelate in peroxisomes, with terminal oxidation in mitochondria. We did not find evidence of peroxisomal oxidation of glutarate. Unexpectedly, DODA contributes a substantial fraction to anaplerosis of the citric acid cycle. This opens the possibility to use water-soluble DODA in nutritional or pharmacological anaplerotic therapy when other anaplerotic substrates are impractical or contraindicated, e.g. in propionic acidemia and methylmalonic acidemia.

would complete its degradation in mitochondria. The catabolism of glutarate is mitochondrial (14).
The goal of the present study was to characterize the catabolism of azelate, contrasting it with the catabolism of DODA and glutarate. Our strategy was to use a combination of metabolomics and fully 13 C-labeled n-dicarboxylates so that the labeling patterns of CoA esters and of other compounds would reflect mitochondrial and/or peroxisomal oxidation of the substrates (15). We used proxies of the labeling of mitochondrial acetyl-CoA (acetyl moiety of citrate (16), C-1 ϩ 2 of ␤-hydroxybutyrate (17)) and peroxisomal acetyl-CoA (acetate released by the liver (18)) to interpret the labeling patterns of acetyl-CoA and malonyl-CoA in liver extracts. The labeling pattern of acetylcarnitine, made in mitochondria, microsomes, and peroxisomes (19) is influenced by the location(s) at which its acetyl-CoA precursor is formed. The data presented below support the compartmentation of the catabolism of DODA, azelate, and glutarate outlined in Scheme 1.
Liver Perfusions-Sprague-Dawley male rats were kept on standard rodent chow ad libitum for 8 -12 days. Isolated livers from overnight fasted rats were perfused (21) with recirculating bicarbonate buffer containing 4% bovine serum albumin (fatty acid-free and dialyzed), 4 mM glucose, and 2 mM unlabeled or fully 13 Clabeled dicarboxylate for 120 min. Other livers were perfused with non-recirculating albumin-free buffer containing 0 to 0.6 mM fully 13 C-labeled dicarboxylate for 30 min. One liver was perfused with 2 mM [ 13 C 2 ]acetate as a control where CAC intermediates are labeled only from M2 acetyl-CoA. At the end of the perfusions, livers were quick frozen and kept in liquid nitrogen until analysis.
Analytical Procedures-To assay the fatty acids and citric acid cycle intermediates, 200 mg of powdered frozen liver was homogenized in 2 ml of methanol. Protein was precipitated by centrifugation at 2000 ϫ g for 20 min. Supernatants were dried under nitrogen gas and derivatized with 100 l of BSTFA at 70°C for 1 h. The samples were analyzed on Agilent 6890 gas chromatograph coupled to Agilent 5973 mass spectrometer operated in electron impact ionization (EI) mode. The GC column was Agilent VF-5MS capillary column (60 m ϫ 0.25 mm ϫ 0.25 m). The inlet and transfer line temperatures were set at 290 and 300°C, respectively. The EI source and mass analyzer temperatures were 230 and 150°C. The temperature gradient was: start at 80°C, increase to 300°C at the rate of 5°C/min, and hold for 10 min.
The 13 C labeling of acetate (22) was a probe of the labeling of peroxisomal acetyl-CoA (18). The 13 C labeling of the C-1 ϩ 2 of ␤-hydroxybutyrate (BHB) was a probe of mitochondrial acetyl-CoA and was calculated from labeling of BHB and the C-3 ϩ 4 fragment according to our published method (14). Acyl-CoAs and acetylcarnitine were assayed by LC-MS/MS as previously described (17,23). The labeling of acetylcarnitine is a composite SCHEME 1. Compartmentation of the catabolism of M12 DODA (dodecanedioate), M9 azelate (nonanedioate), and M5 glutarate in liver. The mass isotopomer composition of most compounds is indicated in parentheses. Additional mass isotopomer distributions are shown in Table 1. Question marks on some arrows reflect uncertainties on the chain length of the catabolite being transferred from peroxisomes to mitochondria. The formation of M2 acetate in peroxisomes is an index of the labeling of peroxisomal acetyl-CoA.
of the labeling of acetyl-CoA in peroxisomes and mitochondria (24). The acetyl moiety of citrate, a proxy of mitochondrial acetyl-CoA, was assayed as acetyl-CoA after cleaving citrate with ATP citrate-lyase (16).
Calculations-Molar percentage enrichment of mass isotopomers 3 was calculated after accounting for the natural abundance mass isotopomer distributions. Correction of mass isotopomer distributions (MID) for natural enrichment was performed as in Ref. 22.

Results and Discussion
Labeling of Acetyl-CoA and Proxies-In livers perfused with recirculating buffer containing M12 DODA, M9 azelate, or M5 glutarate, we assayed the M2 labeling of total acetyl-CoA, acetylcarnitine, the acetyl moiety of citrate, and the C-1 ϩ 2 moiety of BHB (Fig. 1A) 4 . The acetyl moiety of citrate and the C-1 ϩ 2 moiety of BHB are proxies of mitochondrial acetyl-CoA (16,17). In all cases, the acetyl moiety of citrate and the C-1 ϩ 2 moiety of BHB were equally M2 labeled, as expected. However, in livers perfused with M12 DODA or M9 azelate, the M2 labeling of total acetyl-CoA was significantly higher than those of the acetyl moiety of citrate and the C-1 ϩ 2 moiety of BHB. This was not the case in livers perfused with M5 glutarate where the three labeling were about equal. The data reflect the partial peroxisomal ␤-oxidation of M12 DODA and M9 azelate (Scheme 1), based on the following rationale.
The enrichment of acetyl-CoA generated in mitochondria from a fully 13 C-labeled substrate becomes diluted by unlabeled acetyl-CoA derived from unlabeled endogenous and exogenous substrates (glucose, glycogen, and fatty acids). Thus the M2 enrichment of the acetyl moiety of citrate reflects the contribution of the labeled substrate to energy metabolism. In contrast, the M2 labeling of acetyl-CoA generated in peroxisomes (and transferred to the cytosol) can be much higher than that of mitochondrial acetyl-CoA because it is less diluted by unlabeled acetyl-CoA generated extra mitochondrially (18). Thus, although extra-mitochondrial acetyl-CoA accounts for a small fraction of total liver acetyl-CoA, it can substantially increase the average labeling of total liver acetyl-CoA. We observed a similar process in livers and hearts perfused with labeled fatty acids, which are partially degraded in peroxisomes ([1- 13 (20 -22). In the experiments with [1-13 C]octanoate and [3-13 C]octanoate (24 -26), we showed that labeling of substrates was acetate Ͼ acetylcarnitine Ͼ acetyl moiety of citrate. Thus, the labeling of acetylcarnitine reflects the labeling of peroxisomal acetyl-CoA, albeit to a lower degree than the labeling of acetate. Thus, our data indicate that DODA and azelate are partially degraded in peroxisomes, whereas glutarate is not degraded in peroxisomes to a detectable degree (Scheme 1).
In livers perfused with non-recirculating buffer containing increasing concentrations of M12 DODA or M9 azelate, the M2 labeling of total acetyl-CoA was greater than that of the acetyl moiety of citrate (Fig. 2). This was not the case in the presence of M5 glutarate. The M2 labeling ratio (acetyl-CoA)/(acetyl moiety of citrate) was 4 -6 with M12 DODA, 1.5 to 2 with M9 azelate, and about 1.0 with M5 glutarate. This is further support for the partial oxidation of DODA and azelate in peroxisomes.
A proxy of the labeling of liver peroxisomal acetyl-CoA is the labeling of free acetate released by livers perfused with nonrecirculating buffer (18). This is because there are very active acetyl-CoA hydrolases in peroxisomes (27). The acetate proxy is useful only in non-recirculating perfusions where the acetate is released from the liver (rather than metabolized in the liver). Fig. 2D shows that, in livers perfused with M12 DODA or M9 azelate, the enrichment of released acetate is greater than the enrichment of the acetyl moiety of citrate. In contrast, in livers perfused with M5 glutarate, the enrichment of released acetate is much lower than that of the acetyl moiety of citrate. The release of low-labeled acetate reflects probably the release of unlabeled acetate from (i) the peroxisomal ␤-oxidation of unlabeled peroxisomal substrates, plus (ii) a low rate of acetyl-CoA to acetate cycling in mitochondria and cytosol (28,29).
The following calculations support a substantial contribution of peroxisomes to the catabolism of DODA and azelate. If one were to assume that all acetyl-CoA derived from DODA, azelate, and glutarate are generated in mitochondria, one could estimate the total mitochondrial acetyl-CoA turnover in the livers, using (i) the uptake of the dicarboxylates (0.075 Ϯ 0.029, 0.097 Ϯ 0.02, and 0.081 Ϯ 0.047 mol min Ϫ1 g Ϫ1 , for DODA, azelate, and glutarate, respectively), (ii) the M2 enrichment of the acetyl moiety of citrate (Fig. 1A), and (iii) the total yields of acetyl groups/dicarboxylate. One would calculate turnover rates of mitochondrial acetyl-CoA at 0.94, 1.5, and 2.0 mol min Ϫ1 g Ϫ1 in livers perfused with DODA, azelate, and glutarate, respectively. It is unlikely that the turnover of mitochondrial acetyl-CoA would be much lower in livers perfused with DODA or azelate, compared with livers perfused with glutarate. With the latter substrate, the two derived acetyls are generated in mitochondria. The lower apparent turnover of mitochondrial acetyl-CoA in livers perfused with DODA or azelate, compared with perfusions with glutarate suggests that 2 of 4 acetyls derived from DODA were lost as free acetate and did not contribute to the turnover of mitochondrial acetyl-CoA. Similarly 1 or 2 of the 3 acetyls derived from azelate would be lost as free acetate. This is compatible with current concepts of dicarboxylate metabolism starting in peroxisomes, followed by a shortened dicarboxylyl being transferred to mitochondria (Scheme 1).
Labeling of Malonyl-CoA- Fig. 1B shows that, in livers perfused with recirculating buffer, malonyl-CoA is mostly M2 labeled from M12 DODA, M9 azelate, or M5 glutarate, with a small or very small M3 component in perfusions with M12 DODA and M9 azelate. The M2 malonyl-CoA was formed from M2 acetyl-CoA derived from the catabolism of the three fully 13 C-labeled dicarboxylates. In livers perfused with M12 DODA or M9 azelate, the M2 enrichment of malonyl-CoA is significantly lower than that of total acetyl-CoA (Fig. 1A), but not significantly higher than that of the acetyl moiety of citrate (Fig.  1A). Thus, in the presence of M12 DODA or M9 azelate, the pool of cytosolic acetyl-CoA used for malonyl-CoA synthesis consisted mostly of acetyl-CoA transferred from mitochondria, with a component of acetyl-CoA made in peroxisomes. In livers perfused with M5 glutarate, the M2 malonyl-CoA is not signif-icantly different from M2 of total acetyl-CoA, or the acetyl moiety of citrate. Thus, in the presence of M5 glutarate, the pool of cytosolic acetyl-CoA used for malonyl-CoA synthesis consisted mostly of acetyl-CoA transferred from mitochondria.
We had hypothesized that in livers perfused with M9 azelate or M5 glutarate, the presence of M3 malonyl-CoA would reflect the complete peroxisomal ␤-oxidation of part of the metabolized odd-chain dicarboxylate. In livers perfused with M9 azelate, we found a low M3 enrichment of malonyl-CoA, compatible with some degree of complete peroxisomal ␤-oxidation of azelate. No M3 malonyl-CoA was detected in livers perfused with M5 glutarate, as expected. The even-chain M12 DODA was not expected from M3 malonyl-CoA (but would form M4 succinyl-CoA if completely ␤-oxidized in peroxisomes, see below). Unexpectedly, in livers perfused with M12 DODA, malonyl-CoA had a small M3 component that was larger than in livers perfused with M9 azelate (Fig. 1B). We cannot explain this finding with current knowledge of the metabolism of peroxisomes. It is possible that a small amount of M3 malonyl-CoA was formed by the reaction of M2 acetyl-CoA with 13 CO 2 derived from the oxidation of M12 DODA. This could not be demonstrated with our experimental protocol.
Other Metabolites Labeled from Fully 13 C-Labeled Dicarboxylates- Table 1 compares the MIDs of liver metabolites labeled from the dicarboxylates (n ϭ 5 for each dicarboxylate). It also shows for comparison the MIDs of the same metabolites from one liver perfused with M2 acetate. In the latter case, all metabolites are labeled from M2 acetyl-CoA. This comparison allows detecting labeling patterns that do not derive only from M2 acetyl-CoA.
The MID of succinyl-CoA (Table 1) made from M12 DODA shows substantial proportions of M1 to M4 mass isotopomers. In contrast, the MIDs of succinyl-CoA made from M9 azelate, M5 glutarate, or M2 acetate show M1 and M2 isotopomers, which are relatively more abundant than the M3 and M4 isotopomers. The MIDs of free succinate, fumarate, malate (Table 1) show the same relative distributions as the MID of succinyl-CoA. Thus, in perfusions with M9 azelate or M5 glutarate, succinyl-CoA is labeled only from the metabolism of M2 acetyl-CoA in the CAC, and shows a low proportion of M4 isotopomer. In contrast, in perfusions with M12 DODA, some M4 succinyl-CoA is derived from four ␤-oxidation cycles of DODA-CoA. Thus DODA is an anaplerotic substrate. Its contribution to succinyl-CoA is 11% under the conditions of the experiments, based on the M4 enrichment of succinyl-CoA (Table 1).
Is there a possibility of complete ␤-oxidation of M12 DODA to M4 succinyl-CoA in peroxisomes? This question arises because of the identification of a peroxisomal thioesterase, ACOT4, specific for succinyl-CoA hydrolysis (with some activity toward glutaryl-CoA) (30). M4 succinate, released by peroxisomes, could be transferred to mitochondria, as suggested by Tserng and Jin (11) and Westin et al. (30). It is not clear that the ␤-oxidation capacity of peroxisomes could generate enough succinate to contribute substantially to the M4 enrichment of fast turning over CAC intermediates. This is because the ␤-oxidation capacity of peroxisomes decreases with the chain length of substrates and intermediates (10,32). Still, this possibility requires further investigations.

Metabolism of C 12 -, C 9 -, and C 5 -n-dicarboxylates in Rat Liver
M4 succinyl-CoA (Scheme 1). From the uptake of M12 DODA (0.075 mol min Ϫ1 g Ϫ1 ) and the M4 enrichment of succinyl-CoA (11%), the apparent turnover of succinyl-CoA would be 0.075/0.11 ϭ 0.68 mol min Ϫ1 (g wet wt) Ϫ1 . This rate is similar to calculated rates of acetyl-CoA oxidation in liver (21). This calculation assumes that (i) all M12 DODA taken up is converted to mitochondrial succinyl-CoA, and (ii) no M4 succinyl-CoA is recycled in the CAC. Similarly, the MID of citrate labeled from M12 DODA ( Table  1) shows larger relative contributions of heavy isotopomers (M3 to M5) than the MID of citrate labeled from M9 azelate or M5 glutarate. This is compatible with two parallel processes of citrate synthesis: (i) from M2 acetyl-CoA and mostly unlabeled oxaloacetate (as is the case in the presence of M9 azelate or M5 glutarate), and (ii) M2 acetyl-CoA and M2 and M3 oxaloacetate derived from M4 succinyl-CoA (in the presence of M12 DODA) with partial losses of label via exchange processes. These exchanges take place (i) in the oxaloacetate to phosphoenolpyruvate to pyruvate to oxaloacetate cycle, and (ii) by exchange of label from one carboxyl of oxaloacetate for unlabeled bicarbonate in the most rapidly reversible component of the PEP carboxykinase reaction (33). The MID of ␣-ketoglutarate is compatible with those of succinyl-CoA and citrate.
Labeling of the Glutaryl-CoA Pathway-M12 DODA does not generate labeled glutarate, hydroxyglutarate, or glutaryl-CoA (Table 1), but forms M2 ϩ M4 crotonyl-CoA, BHB-CoA, BHB, and butyryl-CoA. In these metabolites, the M2/M4 labeling ratio increases from crotonyl-CoA to BHB-CoA to BHB. The increasing contribution of the M2 isotopomers of these compounds reflects the reversibility of the reactions between crotonyl-CoA and mitochondrial acetyl-CoA (which is more M2 labeled than the C 4 CoA esters being discussed). Thus, these compounds are labeled from M2 acetyl-CoA and are not, as expected, primary metabolites of DODA.
In conclusion, our study has clarified some aspects of the compartmentation of dicarboxylate metabolism in rat liver (Scheme 1). First, the metabolism of glutarate is essentially all mitochondrial, based on the very low enrichment of acetate released by livers perfused with M5 glutarate. Any peroxisomal component of glutarate metabolism must be minuscule.
Second, the metabolism of M9 azelate has a clear peroxisomal component, based on the higher enrichment of total acetyl-CoA, acetylcarnitine, and of released acetate compared with the acetyl moiety of citrate.
Third, the metabolism of M12 DODA includes a very clear peroxisomal ␤-oxidation component as indicated by the higher M2 labeling of total acetyl-CoA, acetylcarnitine, malonyl-CoA, and acetate compared with the proxies of mitochondrial acetyl-CoA (acetyl moiety of citrate and C-1 ϩ 2 of BHB). That most of succinyl-CoA derived from M12 DODA is formed in mitochondria is indicated by the M4 component of the MIDs of succinyl-CoA and other CAC intermediates. Thus, in addition to being a precursor of acetyl-CoA, DODA is an anaplerotic substrate in the liver, and presumably in other organs. Mingrone had advocated the use of glycerol-DODA ester as a water-soluble, sodium-free, insulin-independent parenteral nutrient (12). Because one-third of the DODA carbon is anaplerotic, glycerol-DODA may also be a substrate for anaplerotic therapy (35) of reperfusion injury (myocardial infarction, stroke, and organ transplantation) and of conditions where the use of other anaplerotic substrates is not practical or is contraindicated. For example, heptanoate, a precursor of anaplerotic propionyl-CoA is used in the treatment of long-chain fatty acid oxidation disorders (31). However, heptanoate is contraindicated in disorders of propionyl-CoA metabolism, which are cataplerotic conditions because of the formation of methylcitrate and methylsuccinate, e.g. propionic acidemia, and methylmalonic acidemia. Patients with the latter conditions may benefit from dietary treatment with anaplerotic DODA.
Author Contributions-H. B. and G. F. Z. conceived and coordinated the study and wrote the paper. Z. J. and F. B. designed, performed, and analyzed all experiments. J. K. K. provided critical discussions of the data. K. T. provided technical assistance. All authors reviewed the results and approved the final version of the manuscript.