GHB (γ-hydroxybutyrate) is both a neurotransmitter and a drug of abuse (date-rape drug). We investigated the catabolism of this compound in perfused rat livers. Using a combination of metabolomics and mass isotopomer analysis, we showed that GHB is metabolized by multiple processes, in addition to its previously reported metabolism in the citric acid cycle via oxidation to succinate. A substrate cycle operates between GHB and γ-aminobutyrate via succinic semialdehyde. Also, GHB undergoes (i) β-oxidation to glycolyl-CoA+acetyl-CoA, (ii) two parallel processes which remove C-1 or C-4 of GHB and form 3-hydroxypropionate from C-2+C-3+C-4 or from C-1+C-2+C-3 of GHB, and (iii) degradation to acetyl-CoA via 4-phosphobutyryl-CoA. The present study illustrates the potential of the combination of metabolomics and mass isotopomer analysis for pathway discovery.
- γ-hydroxybutyrate (GHB)
- mass isotopomer analysis
GHB (γ-hydroxybutyrate) is a neurotransmitter derived from the transamination of GABA (γ-aminobutyrate). The known catabolism of GHB involves (i) conversion into succinic semialdehyde by a cytosolic NADP+ dehydrogenase  and by a mitochondrial pyridine nucleotide-independent enzyme system , and (ii) the oxidation of succinic semialdehyde to succinate, an intermediate of the CAC (citric acid cycle). Humans with an inborn disorder of succinic semialdehyde dehydrogenase have high GHB concentrations in their bodily fluids, mental retardation and suffer from seizures . Mice deficient in succinic semialdehyde dehydrogenase fail to thrive and suffer from chronic epileptic seizures . The identification of glycolate in the urine of humans with succinic semialdehyde dehydrogenase deficiency led Vamecq et al.  and Brown et al.  to hypothesize that GHB can undergo β-oxidation to acetyl-CoA. Furthermore, the identification of 2,4-dihydroxybutyrate and 3-hydroxypropionate in the urine of humans with succinic semialdehyde dehydrogenase deficiency led Brown et al.  to hypothesize that GHB can undergo α-oxidation. GHB is also a drug of abuse which, at millimolar concentrations in body fluids, causes amnesia and impairs the ability of the subject to exercise judgement (date-rape drug) . The disposal of exogenous GHB in vivo is inhibited by ethanol . GHB is used for the treatment of narcolepsy with associated cataplexy .
We reported previously that 4-hydroxyacids, including GHB, are converted in liver into 4-phosphoacyl-CoAs, a new class of acyl-CoA esters [10,11]. When generated from 4-hydroxyacids with more than four carbons, 4-phosphoacyl-CoAs undergo a series of reactions leading to 3-hydroxyacyl-CoAs, which are intermediates in the β-oxidation of fatty acids. The catabolism of 4-hydroxyacids with more than four carbons leads to acetyl-CoA, formate and, for odd-chain 4-hydroxyacids, propionyl-CoA. In mice deficient in succinic semialdehyde dehydrogenase, the brain and liver concentrations of 4-hydroxy-4-phosphobutyryl-CoA is 40-fold greater than in wild-type and heterozygote mice .
The aim of the present study was to characterize the metabolism of GHB via the CAC and other processes, using a combination of metabolomics [12–14] and mass isotopomer analysis [15,16]. (Mass isotopomers are designated M, M1, M2,…Mn where n is the number of heavy atoms in the molecule.) We report that, in addition to metabolism through the CAC, GHB is degraded by β-oxidation and by two parallel processes, which remove C-1 of GHB via α-oxidation, and C-4 of GHB via unknown mechanisms.
Chemicals and reagents
General chemicals, [13C4]GHB lactone, [2H6]GHB lactone and 2H2O were from Sigma–Aldrich Isotec®. trans-4-Hydroxycrotonic acid was obtained from Tocris Bioscience. [1,2-13C2]GHB and [3,4-13C2]GHB lactones were synthesized using methods available on request from H.B. Briefly, the synthesis of [1,2-13C2]GHB was started from the glycoaldehyde dimer (an in situ source of α-hydroxyacetaldehyde), which is converted into a hydroxy-α,β-unsaturated ester of butyric acid after Wittig olefination using [1,2-13C2](ethoxycarbonylmethylene)triphenylphosphorane. Hydrogenation of this molecule followed by acid-catalysed lactonization afforded [1,2-13C2]GHB lactone. [3,4-13C2]GHB lactone was synthesized starting with 13C2-labelled ethyl bromoacetate, which was reacted with the sodium salt of benzyl alcohol followed by DIBAL-H (di-isobutyl aluminium hydride) reduction to give the α-(benzyloxy)acetaldehyde. This was subjected to Wittig olefination using (ethoxycarbonylmethylene)triphenylphosphorane followed by hydrogenation and acid-catalysed lactonization as above to produce [3,4-13C2]GHB lactone. The lactones were purified with silica flash chromatography using a 90:10 (v/v) hexane/ethyl acetate solvent system.
The purity of the labelled GHB lactones was assessed using 1H-NMR and judged to be greater than 98% pure. For γ-[1,2-13C2]GHB lactone: 1H-NMR (400 MHz, [2H]chloroform) δ 2.25 (m, 2 H), 2.49 (dm, 2 H, J 35.1 Hz), 4.32 (m, 2 H); 13C-NMR (100 MHz, [2H]chloroform) δ 22.3 (d, J 33 Hz), 28.0 (d, J 49.4 Hz, enhanced), 68.5 (d, J 8.4 Hz), 177.9 (d, J 49.4 Hz, enhanced). Electrospray ionization high-resolution MS (positive mode): for C213C2H6O2·Na+ [M+Na+] m/z calculated 111.03271, found 111.03269. For γ-[3,4-13C2]GHB lactone: 1H-NMR (400 MHz, [2H]chloroform) δ 2.25 (dm, 2 H, J 135.2 Hz), 2.50–2.58 (m, 2 H), 4.36 (dtd, 2 H, J 151.6 Hz, 6.8 Hz, 2.0 Hz); 13C-NMR (100 MHz, [2H]chloroform) δ 22.3 (d, J 31.3 Hz, enhanced), 28.2, 68.7 (d, J 31.3 Hz, enhanced), 177.7. Electrospray ionization high-resolution MS (positive mode): for C213C2H6O2·Na+ [M+Na+] m/z calculated 111.03271, found 111.03271. Also, GC (gas chromatography)–MS of the di-trimethylsilyl derivatives of unlabelled GHB, [1,2-13C2]GHB and [3,4-13C2]GHB confirmed the position of 13C atoms in the molecules: (i) the mass of the parent ion at m/z 233 increased to 235 for both compounds, and (ii) the mass of ion at m/z 204 (C-1+C-2 of GHB) increased from 204 to 206 with [1,2-13C2]GHB, but remained at 204 for [3,4-13C2]GHB. The unlabelled and labelled GHB lactones were hydrolysed with 10% excess NaOH at 60°C for 1 h before use.
All experiments involving animals were performed in accordance with IACUC (Institutional Animal Care and Use Committee) guidelines at Case Western Reserve University.
Livers from overnight-fasted Sprague–Dawley rats were perfused  for 2 h with 150 ml of recirculating bicarbonate buffer containing 4% (w/v) dialysed BSA (fraction V, fatty-acid-free; InterGen), 4 mM glucose with or without 2 mM GHB (unlabelled or 13C4-, 1,2-13C2-, 3,4-13C2- or 2H6-labelled). One liver was perfused with 2 mM unlabelled GHB in 100% 2H2O buffer. In some experiments, 10 mM ethanol or 5 mM glucuronolactone was added at 60 min. Rat hearts were perfused for 60 min in the Langendorff mode with 100 ml of recirculating buffer containing 3% (w/v) BSA and 4 mM glucose with or without 1 mM [13C4]GHB. Livers and hearts were quick-frozen at the end of the perfusions.
The concentrations and 13C-mass isotopomer distributions of CAC intermediates, glutamate  and CoA esters  were assayed as described previously. For the assay of GHB and its carboxylic acid metabolites, perfusate samples (200 μl) spiked with 0.2 μmol of [2H6]GHB internal standard were deproteinized with 2 ml of acetonitrile. After centrifugation and evaporation under N2, the residue was allowed to react with 100 μl of trimethylsilyl reagent and heated for 1 h at 60°C. Then, 2 μl was injected into an Agilent 6890 gas chromatograph linked to a 5973 MSD mass spectrometer. The chromatograph was equipped with a 60 m Varian CP 9017 VF-5 capillary column. The carrier gas was helium (26.8 ml/min), and the injection mode was splitless. The injector temperature was set at 290°C, and the transfer line was held at 290°C. The column temperature was increased by 3°C/min from 50°C to 300°C, where it was held for 10 min. The mass spectrometer was operated under electron impact ionization. The quadrupole and ion source temperatures were 150 and 230°C respectively. The mass scan ranged from 50 to 700. The retention times of 3-hydroxypropionate, GHB, 2,4-dihydroxybutyrate and 3,4-dihydroxybutyrate were 25.2, 29.7, 37.4 and 38.3 min respectively.
The concentrations and labelling of formate and acetate were assayed as the pentafluorobenzyl derivatives by ammonia-negative chemical ionization . Because of the ubiquitous contamination of solutions with traces of unlabelled formate , the labelling of formate is not expressed as molar percent enrichment. We present concentrations of M1 formate, after proper correction of the fraction of the M1 signal, which is derived from the natural enrichment of contaminant formate. In experiments with [2H6]GHB or with unlabelled GHB in 100% 2H2O buffer, the 2H-mass isotopomer distribution of GHB was assayed by negative chemical ionization of the pentafluorobenzylacetyl derivative .
Calculations and statistics
Correction of raw mass isotopomer profiles for natural enrichment at each mass was conducted using CORMAT software . In perfusions with [13C4]GHB, relative anaplerosis was calculated as the enrichment ratio (M4 succinate)/(M4 GHB), because M4 succinate cannot be formed from recycling of label in the CAC. Statistical differences were tested using a paired Student's t test (GraphPad Prism version 3).
Uptake of GHB by the liver
The uptake of GHB by the perfused liver, starting with a 2 mM concentration, was linear with final concentrations above 1 mM at 120 min. Addition of 10 mM ethanol at 60 min significantly decreased the rate of GHB uptake from 205±29 to 92±34 nmol/min per g of dry weight (P<0.05). Kaufman and Nelson  reported a similar effect of ethanol on GHB metabolism in vivo. In two perfusions, addition of 10 mM t-butanol, a non-metabolizable alcohol, did not affect the rate of GHB uptake from 60 to 120 min (results not shown). This demonstrates that the inhibition of GHB uptake by ethanol requires ethanol oxidation. Kaufman and Nelson  reported that after injection of tracer [U-14C]GHB to two rabbits, the uptake of the tracer from the plasma was accelerated by an intravenous injection of glucuronate to one of the rabbits. This suggested the possibility that infusing glucuronate or glucuronolactone could treat humans intoxicated by GHB. In our experiments, the uptake of GHB (at millimolar concentration) by the perfused rat liver was not affected by 5 mM glucuronolactone. A similar negative effect of glucuronate on the clearance of a GHB load in rats was reported previously .
Metabolism of GHB via the CAC
For livers perfused with [13C4]GHB, Table 1 shows the mass isotopomer distribution of CAC intermediates, related compounds (glutamate and GABA) and glucose. The metabolism of M4 GHB yields initially M4 succinate. Cycling of M4 succinate in the CAC yields M1 and M2 isotopomers of succinate. Dividing the M4 enrichment of succinate (7.6%) by the M4 enrichment of GHB (95%) yields the relative anaplerosis from GHB of 8%. This represents the contribution of GHB to the catalytic intermediates of the CAC, which carry acetyl-CoA as it is oxidized. The M4 enrichment of succinyl-CoA was similar to that of succinate. This reflects the isotopic equilibration of succinate and succinyl-CoA through the activities of succinyl-CoA synthetase and succinyl-CoA hydrolase. The substantial abundances of M1 and M2 succinate reflect recycling of label in the CAC. Lastly, GABA was M4 labelled (59%, Table 1). This high M4 enrichment reflects the reversibility of the GABA transaminase and GHB dehydrogenase at high GHB concentration. Traces of M1, M2 and M3 labelling of GABA probably result from the decarboxylation of glutamate labelled by transamination of 2-oxoglutarate.
Labelling of acetyl-CoA was not detectable by LC (liquid chromatography)–MS/MS (tandem MS). At the end of the experiments with [13C4]GHB, perfusate glucose was enriched (M3=4.4±1.8%, M2=2.3±0.7%, M1=2.6±2.6%; means±S.E.M., n=6). This shows that carbon from anaplerotic GHB left the CAC via PEPCK (phosphoenolpyruvate carboxykinase) and followed the reactions of gluconeogenesis. Propionyl-CoA and methylmalonyl-CoA were substantially labelled (Table 1), probably via reversal of the methylmalonyl-CoA mutase and propionyl-CoA carboxylase reactions . Addition of ethanol did not significantly affect the mass isotopomer distribution of CAC intermediates and the anaplerosis from GHB (results not shown).
Cycling between GHB and succinic semialdehyde
To test for cycling between GHB and succinic semialdehyde, an intermediate in the formation of succinate and GABA, we ran one perfusion with unlabelled GHB in 100% 2H2O buffer, and a series of perfusions with an initial 2 mM [2H6]GHB in regular buffer. Our strategy was to test for gain or loss of 2H in GHB. To optimize the measurement of the mass isotopomer distribution of GHB, we derivatized it with pentafluorobenzyl bromide and assayed the derivative by ammonia-negative chemical ionization . The pentafluorobenzyl group of the derivative is lost as a neutral fragment by dissociative electron capture in the mass spectrometer source. Thus the isotopomer distribution is measured on the GHB ion, the natural M1 enrichment of which is low (4.5%). In the perfusion with unlabelled GHB in 100% 2H2O buffer, the mass isotopomer distribution of perfusate GHB, corrected for natural enrichment, showed a progressive accumulation of M1 to M4 mass isotopomers (Figure 1A). The accumulation of the dominant M1 isotopomer follows a sigmoidal curve, which reflects two processes. Early in the experiment, because of the chiral specificity of alcohol dehydrogenase , the oxidation of unlabelled GHB transfers an unlabelled hydrogen from the pro-R position on C-4 of GHB to the pro-4R position of NADH. Reduction of succinic semialdehyde to GHB would not introduce 2H into GHB because NADH was initially unlabelled. However, as the experiment proceeds, glucose and CAC intermediates become extensively labelled from the deuterated perfusate. As a result, NADH becomes labelled on its 4R hydrogen via dehydrogenases of the same chirality as alcohol dehydrogenase (e.g. malate dehydrogenase, NAD–isocitrate dehydrogenase, phosphoglycerate dehydrogenase  and succinic semialdehyde dehydrogenase). This explains the delay in M1 labelling of GHB, which is most likely to be [4-R-2H]GHB.
The stereospecificity of the alcohol dehydrogenase requires an alternative explanation for the formation of M2 to M4 isotopomers of GHB in 100% 2H2O buffer. If [4-2H]succinic semialdehyde undergoes non-enzymatic keto–enol tautomerization before being re-converted into GHB, two additional 2H atoms from the buffer could be incorporated on C-3 of GHB (after reduction of succinic semialdehyde to GHB). Incorporation of 2H into the pro-4S hydrogen of GHB could be effected during the transamination reaction by transfer from the enzyme general acid/base group. Alternatively, multiple 2H atoms could be incorporated by partial β-oxidation and reversal (see below). The predominance of the increase in the M1 isotopomer of GHB (Figure 1A) indicates that the cycling between GHB and succinic semialdehyde is more rapid than the keto–enol tautomerization at physiological pH.
In the perfusions with [2H6]GHB in normal buffer, we observed the release of M5 and M4 mass isotopomers of GHB in the perfusate (Figure 1B). The baseline proportions of M5 and M4 mass isotopomers result from the fact that commercial deuterated compounds are 98–99% 2H-labelled on each hydrogen. On the basis of the above rationale, the R hydrogen on C-4 of M5 GHB ([2H5]GHB) is most likely to be unlabelled. The small amount of M4 GHB ([2H4]GHB) is likely to result from keto–enol tautomerism undergone by [2H5]succinic semialdehyde. The above results clearly demonstrate the cycling between GHB and GABA via succinic semialdehyde.
Metabolomics of GHB catabolism
We conducted metabolomic analyses of the final perfusates from livers perfused with buffer containing 0 or 2 mM GHB (unlabelled or [13C4]GHB). Figure 2 shows the analytical steps leading to the identification of GHB metabolites in the final perfusate of a liver that had been perfused for 120 min with recirculating buffer containing unlabelled GHB (initially 2 mM). A liver perfused without GHB served as control. The top panel of Figure 2 shows mirror images of low-resolution time profiles of mass spectrometric data (total ion current under electron ionization) of trimethylsilyl derivatives. High-resolution plotting of the data in areas of the scan where the signal was very low identified peaks in the GHB perfusion that appeared to be absent from, or present in very low amounts in, the control perfusion (Figure 2, middle panels). Processing of the data using NIST software identified 3-hydroxypropionate, 2,4-dihydroxybutyrate and 3,4-dihydroxybutyrate (Figure 2, bottom panels), as well as 4-hydroxycrotonate (4-hydroxybutene-2-oate) (results not shown). The latter had been identified in the brain of rats injected with GHB [27,28].
Similar analyses from liver perfusions with [13C4]GHB led to the identification of labelled 3-hydroxypropionate (M3), 2,4-dihydroxybutyrate (M4), 3,4-dihydroxybutyrate (M4 as expected, but also M2), 4-hydroxycrotonate (M4), oxalate (M2) and glycolate (M2). Note that small amounts of unlabelled 3,4-dihydroxybutyrate, 3-hydroxypropionate, glycolate and oxalate were present in control experiments without GHB. Additional analyses demonstrated the formation of low concentrations of [13C]formate and [13C]acetate. These were assayed as pentafluorobenzyl derivatives under negative chemical ionization to minimize natural enrichment of the analytes. The very low labelling of free acetate (<0.6%) reflects a very low labelling of acetyl-CoA which could not be detected in the LC–MS/MS assay of acetyl-CoA because of the high M1 and M2 natural enrichments of acetyl-CoA (30% and 12% respectively). Some of the results gathered at this point were compatible with a β-oxidation of GHB to glycolate+acetyl-CoA as hypothesized by Vamecq et al. .
To gain further insight into the catabolism of GHB, we conducted a series of liver perfusions with 1 mM unlabelled 4-hydroxycrotonate, a compound identified in the brain of rats injected with GHB [27,28]. 4-Hydroxycrotonyl-CoA is a probable intermediate of the catabolism of GHB via β-oxidation. We reasoned that 4-hydroxycrotonate would be activated in the liver to 4-hydroxycrotonyl-CoA. This strategy allowed us to inject carbon in a metabolite pool, which is part of the putative β-oxidation of GHB. In livers perfused with 4-hydroxycrotonate, we observed the formation of 3-hydroxypropionate and of GHB (Figure 3). Also, we observed the formation of 3,4-dihydroxybutyrate, but not of 2,4-dihydroxybutyrate. Note that in perfusions with GHB, we had observed the accumulation of both 3,4- and 2,4-dihydroxybutyrate. This suggested that the catabolism of GHB bifurcates before 4-hydroxycrotonyl-CoA. To test this hypothesis, we synthesized [3,4-13C2]GHB and [1,2-13C2]GHB and conducted additional liver perfusion experiments with these substrates.
The six panels of Figure 4 show the release of labelled formate, 4-hydroxycrotonate, 2,4-dihydroxybutyrate, 3,4-dihydroxybutyrate, 3-hydroxypropionate and glycolate respectively from livers perfused with 2 mM [1,2-13C2]GHB or [3,4-13C2]GHB. Figure 4(A) shows a striking difference between the production of [13C]formate from [3,4-13C2]GHB compared with [1,2-13C2]GHB. Also, labelled glycolate (M2) was formed only from [3,4-13C2]GHB, and not from [1,2-13C2]GHB, as shown in Figure 4(F). No M1 glycolate was detected in perfusions with [13C4]GHB, [1,2-13C2]GHB or [3,4-13C2]GHB.
Figure 5 shows the mass isotopomer distributions of 3,4-dihydroxybutyrate, 2,4-dihydroxybutyrate and 3-hydroxypropionate released by livers perfused with [1,2-13C2]GHB or [3,4-13C2]GHB. The mass isotopomer distributions of 3,4-dihydroxybutyrate derived from [1,2-13C2]GHB and [3,4-13C2]GHB are very different. Note that traces of 3,4-dihydroxybutyrate are released by control livers perfused without GHB. 3,4-Dihydroxybutyrate formed from [3,4-13C2]GHB is mostly M2, with a trace of endogenous unlabelled M. In contrast, 3,4-dihydroxybutyrate formed from [1,2-13C2]GHB is 55% M2 and 45% unlabelled. The interpretation of these data is presented in the Discussion.
Fates of 4-hydroxy-4-phosphobutyryl-CoA
In our previous study , we showed that 4-phosphoacyl-CoAs with more that four carbons (in the acyl group) are intermediates in the isomerization of 4-hydroxyacyl-CoAs to 3-ketoacyl-CoAs which are normal products of fatty acid β-oxidation. In the case of 4-hydroxy-4-phospho-[13C4]butyryl-CoA, such isomerization would generate [13C]acetoacetyl-CoA which would (i) equilibrate with [13C2]acetyl-CoA via thiolase, and (ii) label ketone bodies, i.e. acetoacetate and β-hydroxybutyrate via the HMG-CoA (3-hydroxy-3-methylglutaryl-CoA) cycle. We could only detect traces of 13C in ketone bodies. These could also be explained by incorporation of label from acetyl-CoA derived from the β-oxidation of GHB to glycolyl-CoA+acetyl-CoA (via the reversal of thiolase and the HMG-CoA cycle). Thus the experiments with [13C]GHB do not provide unequivocal information on the fate of 4-phosphobutyryl-CoA. However, in perfusions with [2H6]GHB, the M6 enrichments of γ-hydroxybutyryl-CoA and 4-phosphobutyryl-CoA were 86.9±0.7% and 79.0±1.7% respectively (P<0.01), whereas the M5 enrichments of γ-hydroxybutyryl-CoA and 4-phosphobutyryl-CoA were 13.1±0.7% and 21.0±1.7% respectively (P<0.01). The lower M6 and higher M5 enrichments of 4-phosphobutyryl-CoA compared with γ-hydroxybutyryl-CoA reflect a partial loss of 2H in reversible reactions downstream of 4-hydroxy-4-phosphobutyryl-CoA, as we had shown in perfusions with unlabelled 4-hydroxynonanoate or 4-hydroxypentanoate in 100% 2H2O compared with plain H2O buffer . In parallel perfusions with 4-hydroxy-[3,4-13C2]nonanoate or [13C5]laevulinate (reduced in the liver to 4-hydroxy-[13C5]pentanoate), we had shown that the corresponding 4-phosphoacyl-CoAs are degraded via β-oxidation after conversion into 3-ketoacyl-CoAs. Thus it is likely that GHB is also β-oxidized to acetyl-CoA via 4-hydroxy-4-phosphobutyryl-CoA.
Metabolism of GHB in perfused rat hearts
In hearts perfused with recirculating perfusate containing 1 mM [13C4]GHB, the very low uptake of the substrate was not measurable with precision. The mass isotopomer distribution of succinate (M1=2.0±0.5%, M2=0.43±0.20%, M3=0.25±0.07%, M4=1.0±0.2%; means±S.E.M., n=6) shows that relative anaplerosis from GHB was only approximately 1% compared with 8% in liver. Thus very little GHB is metabolized in the heart via the CAC. The only detected substantial labelling of a metabolite was the M2 enrichment of glycolate (27±9%). This shows that a small amount of GHB undergoes normal β-oxidation in the heart. Also, unlabelled glycolate was formed from endogenous substrates. The enrichment of acetyl-CoA was only 0.5±0.4 (n=6).
New pathways of GHB metabolism in liver
The present study identified a number of processes through which GHB is metabolized in liver, in addition to entry into the CAC via succinate. The identification of processes was based on metabolomics and mass isotopomer distribution of metabolites labelled from commercial and custom-synthesized labelled substrates. Although new processes were identified, the only measurable flux was the uptake of GHB by the perfused livers. This is because no end metabolite accumulated. Also, the labelling of 13CO2 was too low to be measured because of (i) the large pool of CO2+bicarbonate in the perfusate, and (ii) the gassing of the perfusate with 1 litre of 95% O2+5% CO2 per min. The 13C label from GHB was found in a number of standard components of intermediary metabolism: CAC intermediates, glutamate, GABA, formate, propionyl-CoA and acetyl-CoA (reflected by the low labelling of acetate). However, we showed that the label of GHB passes through non-standard CoA esters, some of which were identified as such (γ-hydroxybutyryl-CoA, 4-hydroxy-4-phosphobutyryl-CoA, compounds 2 and 3 respectively in Figure 6), whereas others were identified from the corresponding free acids (4-hydroxycrotonyl-CoA, 2,4-dihydroxybutyryl-CoA, 3,4-dihydroxybutyryl-CoA, 3-hydroxypropionyl-CoA and glycolyl-CoA, compounds 7, 4, 8, 5/5′ and 10 respectively in Figure 6).
From all of the labelling data, especially those obtained with [1,2-13C2]GHB and [3,4-13C2]GHB, one can construct a scheme of GHB catabolism via β-oxidation and via two processes that remove C-1 or C-4 of GHB. This scheme is presented in two ways. First, Figure 6, in which C-1+C-2 of GHB are shown as tall characters, whereas C-3+C-4 are shown as short bold characters, presents the complete scheme in detail. Secondly, Figure 7 is a simplified version of Figure 6. In the following description, the labelling pattern of CoA esters discussed was inferred from the labelling pattern of the corresponding free acids assayed in the perfusate.
The α-oxidation process (Figure 6) goes from γ-hydroxybutyryl-CoA to 2,4-dihydroxybutyryl-CoA to 3-hydroxypropionyl-CoA+formate (compounds 2→4→5+6). Through this sequence, M2 [1,2-13C2]GHB forms M1 3-hydropxypropionyl-CoA + M1 formate, whereas [3,4-13C2]GHB forms M2 3-hydroxypropionyl-CoA+unlabelled formate.
The β-oxidation process (Figure 6) goes from γ-hydroxybutyryl-CoA to 4-hydroxycrotonyl-CoA to 3,4-dihydroxybutyryl-CoA to 4-hydroxy-3-ketobutyryl-CoA (putative) to glycolyl-CoA + acetyl-CoA (compounds 2→7→8→9→10+11). Through this sequence, M2 [1,2-13C2]GHB forms M2 3,4-dihydroxybutyryl-CoA, and M2 acetyl-CoA+unlabelled glycolyl-CoA. Through the same sequence, M2 [3,4-13C2]GHB forms M2 3,4-dihydropxybutyryl-CoA and unlabelled acetyl-CoA+M2 glycolyl-CoA. M2 [1,2-13C2]GHB also forms M1 formate, presumably via the catabolism of M2 glycolyl-CoA (Figure 6, compounds 10→6) .
Lastly, a process that removes C-4 of GHB branches off from the β-oxidation pathway: 3,4-dihydroxybutyryl-CoA to 3-hydroxypropionyl-CoA (Figure 6, compounds 8→5′). Through this sequence, M2 [1,2-13C2]GHB forms M2 3-hydroxypropionyl-CoA. Through the same sequence, M2 [3,4-13C2]GHB forms M1 3-hydroxypropionyl-CoA.
This scheme explains why the production of M1 formate (Figure 4A) is much greater from [3,4-13C2]GHB than from [1,2-13C2]GHB. This is because formate is formed from the catabolism of glycolyl-CoA (Figure 6, compounds 10 and 6′) . The latter is M2 labelled from [3,4-13C2]GHB, but unlabelled from [1,2-13C2]GHB. The scheme also explains why, in a liver perfused with 4-hydroxycrotonate (compound 12), we detected 3,4-dihydroxybutyrate (compound 8), but no 2,4-dihydroxybutyrate (compound 4).
Because of the complexity of Figure 6, we present a simplified description of its conclusions in Figure 7. This Figure follows the fates of the four carbons of GHB through the degradation processes that we have identified. The Figure emphasizes that 3-hydroxypropionyl-CoA is formed from C-1+C-2+C-3, as well as from C-2+C-3+C-4 of GHB.
Processes that remove C-1 or C-4 of GHB
First, let us consider the removal of C-1 of GHB in the sequence: γ-hydroxybutyryl-CoA to 2,4-dihydroxybutyryl-CoA to 3-hydroxypropionyl-CoA (Figure 6, compounds 2, 4 and 5). This appears as an α-oxidation sequence because it involves (i) the addition of a hydroxy group on C-2 of γ-hydroxybutyryl-CoA, and (ii) the removal of C-1 of 2,4-dihydroxybutyryl-CoA, presumably as formate. Fatty acid α-oxidation reactions are usually detected by the formation of labelled formate from a fatty acid labelled on C-1 [30–34]. However, labelled formate is also formed from processes unrelated to α-oxidation. For example, the high production of [13C]formate from [3,4-13C2]GHB (Figure 4A) probably results mostly from the catabolism of glycolyl-CoA  (see Figure 6). Also, we have shown previously that part of the label from anaplerotic compounds (such as [13C3]propionyl-CoA derived from [13C5]laevulinate ) is found in glycine and formate. Because GHB is anaplerotic via succinate  (Figure 6, top left), labelled formate should arise from 13C-labelled GHB. Therefore we cannot infer α-oxidation processes in the degradation of [1,2-13C2]GHB or [3,4-13C2]GHB solely from the production of [13C]formate. However, we can infer an α-oxidation process from specific precursor-to-product relationships between the mass isotopomer distributions of sequential metabolites, besides formate. For example, in perfusions with [1,2-13C2]GHB, M2 γ-hydroxybutyryl-CoA, M2 2,4-dihydroxybutyryl-CoA and M1 3-hydroxypropionyl-CoA are in a clear precursor–product relationship (Figure 6, compounds 2, 4 and 5, tall carbons). As in other α-oxidation processes, the sequence involves the addition of a hydroxy group on C-2 of γ-hydroxybutyryl-CoA (compound 2) to form 2,4-dihydroxybutyryl-CoA (compound 4). Thus the removal of C-1 of GHB in the sequence of compounds 2 to 4 to 5+6 is clearly an α-oxidation process.
Secondly, let us consider the removal of C-4 of GHB in the sequence 3,4-dihydroxybutyryl-CoA to 3-hydroxypropionyl-CoA (Figure 6, compounds 8 and 5′). In perfusions with [3,4-13C2]GHB, M2 3,4-dihydroxybutyryl-CoA and M1 3-hydroxypropionyl-CoA are in a clear precursor–product relationship (Figure 6, compounds 8 and 5′, short bold carbons). The production of compound 5′ from compound 8 is also confirmed by data from perfusions with unlabelled 4-hydroxycrotonate (Figure 3). This compound forms 3,4-dihydroxybutyrate and 3-hydroxypropionate, but not 2,4-dihydroxybutyrate (Figures 3 and 6). Therefore there must be a sequence of unknown reactions that remove C-4 of dihydroxybutyryl-CoA, forming 3-hydroxypropionyl-CoA and probably either formate or CO2. This is why, in Figure 6, we label the conversion of 8 into 5′ as ‘removal of C-4’. The removal of C-4 probably involves the conversion of the primary alcohol function into a carboxy group or a carboxyl-CoA, followed by the removal of C-4 as CO2 or formate.
Reversibility of some reactions of GHB β-oxidation
The mass isotopomer distribution of 3,4-dihydroxybutyrate formed from [3,4-13C2]GHB was mostly M2, with a trace of endogenous unlabelled M (Figure 5). In contrast, 3,4-dihydroxybutyrate formed from [1,2-13C2]GHB was 55% M2 and 45% unlabelled. This led us to hypothesize that C-1+C-2 of 3,4-dihydroxybutyryl-CoA (Figure 6, compound 8) exchange with unlabelled acetyl-CoA (compound 11) via the putative 4-hydroxy-3-ketobutyryl-CoA (compound 9) and the reactions catalysed by 3-hydroxyacyl-CoA dehydrogenase and 3-ketoacyl-CoA thiolase. To check this hypothesis, we perfused one liver with unlabelled GHB+unlabelled glucose+5 mM [13C2]acetate (a precursor of M2 acetyl-CoA). We found that 3,4-dihydroxybutyrate was 12% M2 labelled, whereas 2,4-dihydroxybutyrate was unlabelled. This confirmed the isotopic exchange of C-1+C-2 of 3,4-dihydroxybutyryl-CoA with acetyl-CoA via 3-hydroxyacyl-CoA dehydrogenase+thiolase.
The present study shows that the metabolism of GHB differs from that of longer-chain 4-hydroxyacids that we described previously . First, as shown by others, one fate of GHB is its oxidation to a dicarboxylic acid, succinate, which is an intermediate of the CAC. There is no equivalent process in the fates of longer-chain 4-hydroxyacids. We had shown that the 4-hydroxyacids with five of more carbons are catabolized by two pathways: (i) an isomerization pathway to 3-hydroxyacyl-CoAs via 4-phosphoacyl-CoAs followed by β-oxidation degradation, and (ii) sequential β-oxidation, α-oxidation and β-oxidation steps. Both pathways lead to acetyl-CoA, formate and propionyl-CoA. We were able to assess the metabolism of the longer 4-hydroxyacids from (i) the uptake of the substrate, and (ii) the ratio of pathway fluxes (A/B) by the labelling ratio (M2 acetyl-CoA)/(M1 acetyl-CoA) measured in the presence of 4-hydroxy-[3,4-13C2]nonanoic acid. This type of calculation is not possible for GHB catabolism because it is catabolized by five pathways, none of which yields a product that accumulates.
4-Phosphobutyryl-CoA accumulates in much lower concentrations in livers perfused with GHB than in livers perfused with longer-chain 4-hydroxyacids (C5 to C11). Thus one cannot identify intermediates of its catabolism. The 2H-labelling of 4-phosphobutyryl-CoA in perfusions conducted in 100% 2H2O buffer (as was observed with longer-chain 4-phosphoacyl-CoAs) suggests that it is sequentially metabolized to 3-hydroxybutyryl-CoA, acetoacetyl-CoA and acetyl-CoA. It is not possible to separate 3-hydroxybutyryl-CoA from the γ-hydroxybutyryl-CoA present at fairly high concentration. The 6-fold higher concentration of M1 formate in perfusions with [3,4-13C2]GHB compared with [1,2-13C2]GHB (Figure 4A) suggests that the β-oxidation of GHB (Figure 6) is its most abundant catabolic pathway. This is because M1 formate derives mostly from the oxidation of glycolate formed from C-3+C-4 of GHB.
The groups of Pollitt and Vamecq had hypothesized that GHB undergoes β-oxidation [5,6] and α-oxidation  because of the excretion of glycolate, 2,4-dihydroxybutyrate and 3-hydroxypropionate by patients affected by an inborn disorder of succinic semialdehyde dehydrogenase . Our data confirm and extend these hypotheses by the identification and characterization of two parallel β-oxidation processes and two parallel processes which remove C-1 or C-4 of GHB and yield 3-hydroxypropionate. These newly identified pathways may play a role in the physiopathology of succinic semialdehyde dehydrogenase deficiency. Lastly, the present study illustrates the potential of the combination of metabolomics and mass isotopomer analysis for pathway discovery .
Guo-Fang Zhang, Stephanie Lauden, Chia-Ying Chuang, Sophia Sushailo and Priya Chatterjee performed the liver perfusion experiments and analysed the samples. Sushabhan Sadhukhan synthesized 13C-labelled GHB. Rafael Ibarra performed the heart perfusion experiments. Vernon Anderson and Gregory Tochtrop designed the experiments and interpreted the data. Henri Brunengraber designed the experiments, interpreted data and wrote the paper.
This work was supported by the National Institutes of Health [Roadmap grant number R33DK070291 (to H.B.), and grant numbers R01ES013925 (to H.B.) and R01HL053315 (to G.P.T)] and the Cleveland Mt. Sinai Health Care Foundation].
Abbreviations: CAC, citric acid cycle; GABA, γ-aminobutyric acid; GHB, γ-hydroxybutyrate; HMG-CoA, 3-hydroxy-3-methylglutaryl-CoA; LC, liquid chromatography; MS/MS, mass spectrometry
- © The Authors Journal compilation © 2012 Biochemical Society