Many biological systems including the oxidative catabolic pathway for branched-chain amino acids (BCAAs) are affected in vivo by valproate therapy. In this study, we investigated the potential effect of valproic acid (VPA) and some of its metabolites on the metabolism of BCAAs. In vitro studies were performed using isovaleryl-CoA dehydrogenase (IVD), isobutyryl-CoA dehydrogenase (IBD), and short branched-chain acyl-CoA dehydrogenase (SBCAD), enzymes involved in the degradation pathway of leucine, valine, and isoleucine. The enzymatic activities of the three purified human enzymes were measured using optimized high-performance liquid chromatography procedures, and the respective kinetic parameters were determined in the absence and presence of VPA and the corresponding CoA and dephosphoCoA conjugates. Valproyl-CoA and valproyl-dephosphoCoA inhibited IVD activity significantly by a purely competitive mechanism with Ki values of 74 ± 4 and 170 ± 12 μM, respectively. IBD activity was not affected by any of the tested VPA esters. However, valproyl-CoA did inhibit SBCAD activity by a purely competitive mechanism with a Ki of 249 ± 29 μM. In addition, valproyl-dephosphoCoA inhibited SBCAD activity via a distinct mechanism (Ki = 511 ± 96 μM) that appeared to be of the mixed type. Furthermore, we show that both SBCAD and IVD are active, using valproyl-CoA as a substrate. The catalytic efficiency of SBCAD turned out to be much higher than that of IVD, demonstrating that SBCAD is the most probable candidate for the first dehydrogenation step of VPA β-oxidation. Our data explain some of the effects of valproate on the branched-chain amino acid metabolism and shed new light on the biotransformation pathway of valproate.
Valproic acid (VPA; 2-n-propylpentanoic acid) is a simple branched-chain fatty acid that is known worldwide for its anticonvulsant properties. It is used mostly for the treatment of several types of seizures, bipolar disorders, various psychiatric syndromes, and migraines (Perucca, 2002; Peterson and Naunton, 2005; Bialer and Yagen, 2007). In addition, VPA has emerged recently as a drug that shows potential in cancer treatment (Duenas-Gonzalez et al., 2008). Although VPA has a broad range of clinical use, it is associated with several adverse effects. Hepatotoxicity is a well recognized complication of VPA therapy (Gerber et al., 1979; Bissell et al., 2001; Sztajnkrycer, 2002). Knowledge of the pathophysiological mechanisms underlying this hepatotoxicity is incomplete. Inhibition of mitochondrial catabolic pathways, for instance, fatty acid β-oxidation, has been implicated strongly in the hepatotoxicity of VPA (Silva et al., 2001a, 2008; Bjornsson, 2008). In patients receiving valproate, an increase in serum levels of the branched-chain amino acids (BCAAs; leucine, isoleucine, and valine) and their intermediates as well as an increase in urinary excretion of these amino acids have been reported (Anderson et al., 1994; Silva et al., 2001a). These findings suggest that valproate adversely affects BCAA metabolism.
VPA and the intermediates in the degradation of the two BCAAs isoleucine and valine undergo β-oxidation (Silva et al., 2002) that generates structurally similar metabolites, which suggests that VPA might use key enzymes of the BCAA catabolic pathways for its own oxidation (Fig. 1). Isovaleryl-CoA, 2-methylbutyryl-CoA, and isobutyryl-CoA, the intermediates in BCAA catabolism, are converted to 3-methylcrotonyl-CoA, tiglyl-CoA, and methacrylyl-CoA by isovaleryl-CoA dehydrogenase (IVD; EC 188.8.131.52), short branched-chain acyl-CoA dehydrogenase (SBCAD; EC 184.108.40.206), and isobutyryl-CoA dehydrogenase (IBD; EC 220.127.116.11), respectively. These enzymes are members of the acyl-CoA dehydrogenase (ACD) family, which consists of homologous mitochondrial flavoproteins that catalyze the α,β-dehydrogenation of acyl-CoA thioester substrates to the corresponding trans-2-enoyl-CoAs (Battaile et al., 2004). These enzymes share many similar molecular and catalytic properties but differ with respect to their substrate specificities regarding length and configuration of the carbon backbone of their substrates.
Previous work from Li et al. (1991), Ito et al. (1990), and our group (Silva et al., 2001b, 2002, 2004) has led to the partial resolution of the mitochondrial β-oxidation pathway of VPA. We concluded that the enzymes involved in the oxidation of straight-chain fatty acids, including very long-, long-, medium-, and short-chain acyl-CoA dehydrogenases, are not involved in the first dehydrogenation reaction of VPA. Ito et al. (1990) originally showed that the rat 2-methyl-branched-chain acyl-CoA dehydrogenase was active with valproyl-CoA (VP-CoA) as substrate, but Willard et al. (1996) subsequently demonstrated that the human homolog was much less active with VP-CoA than its rat counterpart. In the current study, we have investigated the involvement of the three BCAA acyl-CoA dehydrogenases with respect to the first step of the β-oxidation of VPA. We report the kinetic characterization of the SBCAD reaction with VP-CoA as substrate using the human purified enzyme. We also demonstrate that IVD is able to produce Δ2(E)-valproyl-CoA (Δ2(E)-VP-CoA), although with a much lower catalytic efficiency. Furthermore, we have investigated the potential inhibitory effect of VPA and some of its mitochondrial metabolites, especially in the form of the corresponding acyl-CoA intermediates VP-CoA and valproyl-dephosphoCoA (VP-DephCoA) (Silva et al., 2001b, 2004), on the activity of IVD, IBD, and SBCAD. The implications of these findings with respect to the treatment of patients by VPA are discussed.
Materials and Methods
Valproic acid, human and bovine serum albumin (BSA), bicinchoninic acid, ferrocenium hexafluorophosphate, FAD, isovaleryl-CoA, and isobutyryl-CoA were obtained from Sigma-Aldrich (St. Louis, MO). Tris was obtained from Merck (Darmstadt, Germany).
Heterologously expressed IVD, IBD, and SBCAD were obtained as described previously (Mohsen and Vockley, 1995; Gibson et al., 2000; Nguyen et al., 2002). VP-CoA (Silva et al., 2001b), VP-DephCoA (Silva et al., 2004), and 2-methylbutyryl-CoA (Rasmussen et al., 1990) were synthesized as described previously.
Enzymatic Activity Measurement of the Branched-Chain Acyl-CoA Dehydrogenases.
Experimental conditions were optimized for each enzyme in terms of protein, time, and pH. The reaction mixture contained 200 mM Tris-HCl, pH 8.0, 50 μM FAD, 400 μM ferrocenium hexafluorophosphate, 0.1 mg/ml BSA, and substrate (isovaleryl-CoA, isobutyryl-CoA, 2-methylbutyryl-CoA) or VP-CoA. This mixture was added to an enzyme solution containing 0.1 mg/ml BSA. Incubations were performed at 37°C for variable time periods depending on the enzyme and the substrate used. The reactions were terminated by adding 10 μl of HCl (2 M), and, afterward, the samples were placed on ice. After neutralization with (2 M KOH)/(1 M MES, pH 6.0), 10 μl of l-cysteine (10 mM) and 30 μl of methanol were added to the mixture. l-Cysteine is added to reduce oxidized ferrocenium hexafluorophosphate, which may interfere with the chromatographic separation. The samples were centrifuged at 20,000g for 5 min, and the metabolites in the supernatants were analyzed by high-performance liquid chromatography (HPLC).
Sample Analysis by HPLC.
Acyl-CoA esters were quantified by HPLC. Analysis was performed at room temperature with a pump (PE series 200; PerkinElmer Life and Analytical Sciences, Waltham, MA) and a Gilson 234 auto-sampling injector (Gilson, Inc., Middleton, WI). A frit C-402X (Upchurch Scientific, Oak Harbor, WA), a 4.6 × 250-mm Supelcosil LC-18-DB (5 μM) column (Supelco, Bellefonte, PA), and a guard column (4.6 × 20 mm) filled with the same packing material were used. For gradient elution of branched-chain acyl-CoAs, a binary system of methanol and 50 mM potassium phosphate, pH 5.3, was used, whereas for the analysis of VPA intermediary metabolites, a system of acetonitrile and 17 mM sodium phosphate, pH 6.9, was used. Acyl-CoA esters were detected with a UV detector (SPD-10A VP UV-VIS; Shimadzu, Kyoto, Japan) at 260 nm.
Identification of Δ2-VP-CoA by HPLC-Electrospray Ionization-Tandem Mass Spectrometry.
HPLC-electrospray ionization (ESI)-tandem mass spectrometry analysis was performed using a triple-quadrupole TSQ Quantum HPLC tandem mass spectrometer (Thermo Fisher Scientific, Waltham, MA) in the negative ESI mode. The samples were injected onto a YMC-Pack Pro C4 column (2.1 × 100 mm; YMC Europe GMBH, Dinslaken, Germany) using a HPLC system consisting of a Surveyor MS-pump with degasser, a Surveyor autosampler, and a column oven. The flow rate was set at 250 μl/min. Elution of acyl-CoAs was achieved with a tertiary system using solvent A (50 mM ammonium acetate, pH 7.0), solvent B (100% acetonitrile), and solvent C (20 mM ammonium bicarbonate).
Separation was performed at 40°C, and nitrogen was used as nebulizing gas, whereas argon was used as collision gas at a pressure of 1.5 mTorr. The capillary voltage was 2.5 kV, and the respective temperature was 350°C, with an optimal collision energy of 30 eV. Acyl-CoAs were measured using multiple reaction monitoring in the negative ionization mode, using the transitions m/z 445.5 → 79.0 for C8-CoA (VP-CoA) and m/z 444.5 → 79.0 for C8:1-CoA (Δ2-VP-CoA). The system was controlled by Xcalibur software (version 2.0; Thermo Fisher Scientific).
The characterization of IVD and SBCAD activities in the absence and presence of different inhibitors was performed by plotting the measured reaction rates as a function of the substrate concentration (at a fixed concentration of the remaining components of the reaction). The steady-state kinetic data of IVD, IBD, and SBCAD activities were determined by nonlinear regression analysis using the SigmaPlot 10.0 Technical Graphing software (Systat Software, Inc., San Jose, CA) and the Enzyme Kinetics Module (version 1.3; Systat Software, Inc.). The Michaelis-Menten equation was used to calculate kinetic parameters (Km and Vmax) of the enzyme reaction using the respective substrates. The inhibition constant (Ki) was calculated by nonlinear regression of the respective inhibition curves, using the mentioned software.
Effects of VP-CoA and VP-DephCoA on the Activity of IVD, IBD, and SBCAD.
Kinetic studies were performed with the three heterologously expressed and purified acyl-CoA dehydrogenases IVD, IBD, and SBCAD. The corresponding activity was determined with their natural substrates, isovaleryl-CoA, isobutyryl-CoA, and 2-methylbutyryl-CoA, respectively. The kinetic constants Km and Vmax were calculated and are summarized in Table 1.
The effect of VP-CoA and VP-DephCoA was tested on the three enzymes. The activities of IVD and SBCAD, as measured at 0.2 mM isovaleryl-CoA and 0.2 mM 2-methylbutyryl-CoA, were decreased by approximately 75 and 45%, respectively, at 1 mM VP-CoA. After incubation with 1 mM VP-DephCoA, both enzyme activities were decreased approximately 45%. The activity of IBD was not affected in the presence of VP-CoA and VP-DephCoA. To further characterize the inhibition mechanism, the activity of IVD and SBCAD was determined as a function of the inhibitor concentration.
The obtained results were analyzed by three linearization methods of the Michaelis-Menten equation, namely Lineweaver-Burk, Eadie-Hofstee, and Hanes-Woolf plots. Analysis of the data pointed to a purely competitive mechanism of inhibition of both VP-CoA and VP-DephCoA with respect to IVD, as depicted in Fig. 2. However, with respect to SBCAD, these CoA esters appeared to have different inhibitory mechanisms. Indeed, whereas VP-CoA was found to inhibit SBCAD by a purely competitive mechanism, VP-DephCoA appeared to be a mixed-type inhibitor, as shown in Fig. 3. The Lineweaver-Burk linearization plots and corresponding Dixon plots (Dixon, 1953) are exhibited in parallel, in either Fig. 2 or Fig. 3.
Assuming a purely competitive or mixed mechanism of inhibition, the Ki was calculated. With respect to IVD activity, a Ki value of 74 ± 4 μM was estimated for VP-CoA and a Ki value of 170 ± 12 μM was estimated for VP-DephCoA. The Ki values for VP-CoA and VP-DephCoA were 249 ± 29 and 511 ± 96 μM, respectively, for SBCAD activity.
Involvement of Branched-Chain Acyl-CoA Dehydrogenases in the Dehydrogenation of VP-CoA.
To identify the metabolizing enzyme of the first dehydrogenation step of the oxidative metabolism of VPA, the purified enzymes (IVD, IBD, and SBCAD) were incubated with 1 mM VP-CoA. As shown in Fig. 4, Δ2(E)-VP-CoA was not observed using IBD as the enzyme (Fig. 4B), suggesting that IBD is not involved in the metabolism of VPA. However, both IVD (Fig. 4A) and SBCAD (Fig. 4C) showed activity with VP-CoA as a substrate, amounting to 0.13 and 1.20 nmol/mg protein · min−1, respectively. Although SBCAD dehydrogenates VP-CoA at a much lower rate than its natural substrate, i.e., 2-methylbutyryl-CoA, the kinetic parameters were determined (Km = 304 ± 51 μM and Vmax = 0.27 ± 0.02 nmol/mg protein · min−1).
An extra incubation was performed with SBCAD to identify the product of VP-CoA. Figure 5B shows the HPLC chromatograms related to the synthesis of Δ2(E)-VP-CoA, where SBCAD was incubated for 1 h with VP-CoA. In Fig. 5A, the same incubation was performed for 0 h, and, consequently, there is no product formation. The same samples were analyzed by HPLC-ESI-tandem mass spectrometry using multiple-reaction monitoring in the negative ionization mode. Figure 5, C and D, shows the mass spectra of singly charged ions that are formed more abundantly. The mass of a C8:1-CoA (889.8) was detected in samples in which SBCAD was incubated for 1 h with VP-CoA (Fig. 5D). This mass corresponds to Δ2(E)-VP-CoA, the dehydrogenation product of VP-CoA.
This study shows that valproate interferes with enzymes involved in the oxidative metabolism of leucine and isoleucine. In particular, two branched-chain acyl-CoA dehydrogenases (BCADs), IVD and SBCAD, were found to participate in the oxidation of VPA. We have shown that human IVD is able to convert VP-CoA into Δ2(E)-VP-CoA, although at a lower rate than with its natural substrate, isovaleryl-CoA. VP-CoA has been previously shown to be oxidized at a significant rate by SBCAD from rat liver mitochondria (Ito et al., 1990) and by bacterially expressed rat and human SBCAD (Willard et al., 1996). We have used human SBCAD expressed in Escherichia coli (Willard et al., 1996) to confirm and characterize the active role of SBCAD in the β-oxidation of VPA. These results are in agreement with the inhibition studies performed with VP-CoA and VP-DephCoA and the BCADs. Both VPA metabolites inhibit IVD by a purely competitive mechanism. IBD activity was not affected by these VPA metabolites. However, VP-CoA did inhibit SBCAD activity by a purely competitive mechanism, whereas VP-DephCoA inhibited SBCAD activity through a distinct mechanism.
The observed inhibitory effects of the VPA metabolites, as tested in this study, on the activity of the BCADs are most likely due to structural similarities between the substrates of these enzymes and the valproate metabolites VP-CoA and VP-DephCoA. IVD oxidizes β-branched acyl-CoAs such as isovaleryl-CoA, but both SBCAD and IBD oxidize short-chain acyl-CoAs with a branched α-position (Fig. 6). As a result of its α-branched configuration, VP-CoA would have appeared to be a good substrate analog for both SBCAD and IBD. However, VP-CoA is a substrate analog for SBCAD but not for IBD. The natural substrate of IBD is isobutyryl-CoA, which has a smaller acyl moiety than the VP-CoA acyl moiety; hence, the apparent lack of interaction between VP-CoA and IBD is probably because of the limited size of the isobutyryl moiety binding pocket (Battaile et al., 2004). SBCAD is able to oxidize α-branched acyl-CoAs such as 2-methyl-butyryl-CoA and straight-chain substrates such as butyryl-CoA and hexanoyl-CoA (He et al., 2003), and therefore its active site seems to be the best accommodating the dehydrogenation of VP-CoA.
Unexpectedly, VP-CoA was found to be oxidized by IVD. Although the reaction occurs at a very low rate, it is still rather surprising because this enzyme handles β- but not α-branched acyl-CoAs. IVD has a larger binding pocket than IBD (Battaile et al., 2004), which apparently allows the enzyme to act upon VP-CoA with its five-carbon backbone. This result is in agreement with the competitive mechanism of inhibition seen for IVD when VP-CoA was used as an inhibitor.
Because SBCAD dehydrogenates VP-CoA to Δ2(E)-VP-CoA and is significantly inhibited by both VP-CoA and VP-DephCoA, it would be expected that patients under VPA therapy would accumulate the endogenous substrate of SBCAD, 2-methylbutyryl-CoA. However, no studies have reported the increase of metabolites derived from 2-methylbutyryl-CoA perhaps because IBD also has activity toward this substrate (Nguyen et al., 2002) substituting, at least in part, SBCAD activity in case of its malfunction or inhibition.
It has been shown that a dysregulated BCAA metabolism makes an independent contribution to development of insulin resistance and glucose intolerance in obese humans (Newgard et al., 2009). Insulin resistance associated with weight gain has also been reported during VPA treatment (Masuccio et al., 2010; Verrotti et al., 2010). However, it is unknown to what extent the interference of valproate on the BCAA oxidation shown in this study is related to the significant weight gain potentially associated with VPA. It remains to be established whether the inhibitory effects, as observed in this study, are enough to account for the well recognized potential liver toxicity of VPA. In this respect, it is important to mention that, although there are a significant number of patients suffering from side effects, VPA-induced hepatotoxicity seems to be idiosyncratic and only affects a small group of patients.
In summary, we have shown that VPA can interfere with the activity of some enzymes from the ACD family, namely IVD and SBCAD. Taking into account that ACDs have overlapping activity with different substrates (Tiffany et al., 1997; Battaile et al., 2004), the drug might exacerbate the impairment of genetically affected routes and elicit a toxic condition (Silva et al., 2008). In fact, it has been suggested already that the intake of VPA should be avoided in patients with inborn errors affecting mitochondrial metabolism (Silva et al., 2008). For that reason, we conclude that VPA administration should be avoided in cases of inborn deficiencies affecting certain ACDs or affecting the leucine and isoleucine oxidative pathways (Korman, 2006; Vockley and Ensenauer, 2006).
Participated in research design: Luís, Ruiter, IJlst, Duran, Wanders, and Silva.
Conducted experiments: Luís.
Contributed new reagents or analytic tools: Mohsen and Vockley.
Performed data analysis: Luís, Ruiter, IJlst, and Silva.
Wrote or contributed to the writing of the manuscript: Luís, IJlst, Almeida, Duran, Vockley, Wanders, and Silva.
We thank Arno van Cruchten for expert technical assistance.
This work was supported in part by the National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases [Grant R01-DK54936] (to J.V.); and Fundação para a Ciência e a Tecnologia, Lisboa, Portugal [Grant SFRH/BD/25913/2005].
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
- 2-n-propylpentanoic acid or valproic acid
- 2-n-propyl-2-pentenoic acid
- branched-chain amino acid
- isovaleryl-CoA dehydrogenase
- bovine serum albumin
- isobutyryl-CoA dehydrogenase
- short branched-chain acyl-CoA dehydrogenase
- high-performance liquid chromatography
- electrospray ionization
- branched-chain acyl-CoA dehydrogenase
- acyl-CoA dehydrogenase.
- Received December 7, 2010.
- Accepted March 14, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics