Metabolic analysis reveals evidence for branched chain amino acid catabolism crosstalk and the potential for improved treatment of organic acidurias

Abstract

Branched chain amino acid (BCAA) metabolism occurs within the mitochondrial matrix and is comprised of multiple enzymes, some shared, organized into three pathways for the catabolism of leucine, isoleucine, and valine (LEU, ILE, and VAL respectively). Three different acyl-CoA dehydrogenases (ACADs) are active in each catabolic pathway and genetic deficiencies in each have been identified. While characteristic metabolites related to the enzymatic block accumulate in each deficiency, for reasons that are not clear, clinical symptoms are only seen in the context of deficiency of isovaleryl-CoA dehydrogenase (IVDH) in the leucine pathway. Metabolism of fibroblasts derived from patients with mutations in each of the BCAA ACADs were characterized using metabolomics to better understand the flux of BCAA through their respective pathways. Stable isotope labeled LEU, ILE, and VAL in patient and control cell lines revealed that mutations in isobutyryl-CoA dehydrogenase (IBDH in the valine pathway) lead to a significant increase in isobutyrylcarnitine (a surrogate for the enzyme substrate isobutyryl-CoA) leading to metabolism by short-branched chain acyl-CoA dehydrogenase (SBCADH in the isoleucine pathway) and production of the pathway end product propionylcarnitine (a surrogate for propionyl-CoA). Similar cross activity was observed for SBCADH deficient patient cells, leading to a significant increase in propionylcarnitine, presumably by metabolism of 2 methylbutyryl-CoA via IBDH activity. Labeled BCAA studies identified that the majority of the intracellular propionyl-CoA pool in fibroblasts is generated from isoleucine, but heptanoic acid (a surrogate for odd-chain fatty acids) is also efficiently converted to propionate.

INTRODUCTION

Consumption and efficient catabolism of branched-chain amino acids (BCAA) are critical in maintaining adequate anabolism, protein turnover mediated via mTOR signaling, fatty acid oxidation, neurological energetic demands and proper immune function (1). Numerous monogenic rare genetic diseases characterized by improper BCAA breakdown have been identified, usually through increased concentrations of specific organic acids in urine or carnitine esters of CoA-activated carboxylic acids in blood (2). Elevation of either BCAA (in branched chain ketoacid dehydrogenase deficiency, also known as maple syrup urine disease) or the more distil carnitine esters can be identified through newborn screening via tandem mass spectrometry (3). Patients can present with life threatening neonatal symptoms, but the spectrum of disease ranges from severe, early symptoms to asymptomatic. Catabolism of leucine, isoleucine, and valine (LEU, ILE, and VAL, respectively) begins with their transport into mitochondria followed by the reversible deamination to 2-oxo-branched chain organic acids (Fig 1A). Oxidative decarboxylation, which forms branched-chain acyl-coenzyme A (CoA) products is considered the first irreversible step in the pathway. These two reactions are performed by enzymes common to all three pathways (aminotransferase and alpha-keto acid dehydrogenase), followed by acyl-CoA dehydrogenases (ACADs) specific to each amino aci (1). The ACADs are a family of related enzymes active in fatty acid and BCAA oxidation that catalyze the FAD-dependent α,β-dehydrogenation of acyl-CoA esters (4). While isovaleryl-CoA dehydrogenase (IVDH)is specific towards the leucine metabolite isovaleryl-CoA, short/branched chain acyl-CoA dehydrogenase (SBCADH), and isobutyryl-CoA dehydrogenase (IBDH; ACAD8) show promiscuity towards both 2-methylbutyryl-CoA (ILE degradation) and isobutyryl-CoA (VAL degradation), respectively (Table 1) (5). The balance of flux from BCAAs to the TCA cycle is complex and tissue specific, but the relative contribution of all propiogenic amino acids to TCA cycle intermediates has not been studied (6).

Figure 1:

A. Summary of the BCAA catabolic pathway with the ACADs identified. LEU and ILE can be used to produce acetyl-CoA, while ILE and VAL can be used to produce propionyl-CoA. B. Metabolomic analysis of the ACAD deficient cell lines. For each ACAD deficient cell line, the expected acyl-carnitine was observed to be significantly increased. The Y-axis shows the fold change of each metabolite measured. Each dot along the X-axis corresponds to a specific analyte, detailed in the Supplemental Table.

Table 1.

Comparative K_ms of human BCAA ACADs (μM).

	Isovaleryl-CoA	2-methylbutyryl-CoA	Isobutyryl-CoA
IVD	1.0	Not reported	N.D.
SBCAD	N.D.	2.7	130
IBD	N.D.	18	2.6

N.D., not detectable

Disorders of all three ACADs in the BCAA pathways have been described. Diagnosis of a specific ACAD deficiency is often first suspected on the basis of the presence of a characteristic acylcarnitine in blood, derived from the increased acyl-CoA species (Fig 1B). Interestingly, only patients with IVDH mutations have clear clinical symptoms, ranging from severe to mild, often with exacerbations with intercurrent illness. Treatment is by restriction of LEU along with supplementation with carnitine and glycine to promote excretion of potentially toxic metabolites. While symptomatic patients with SBCADH and IBDH mutations have been described, babies identified via newborn screening are asymptomatic regardless of the observed increases in the organic acids, methylbutyric and isobutyric acid, leading to the conclusion that they are not clinically relevant deficiencies. The reason for this discrepancy in the clinical impact of these three defects is unclear. We have previously proposed that it may be due to the substrate promiscuity of SBCADH and IBDH, allowing enough flux of accumulating metabolites to decrease toxicity and/or to produce enough propionyl-CoA to meet energetic demands. To further examine this phenomenon, we have performed studies of catabolism of labeled BCAA amino acids in cells deficient of IVDH, SBCADH, and IBDH.

MATERIALS AND METHODS

Cell lines

All cell lines were obtained in anonymized fashion from cultures previously established for clinical diagnostic purposes. Thus, these experiments were designated as exempt by the Insitutional Review Board of the University of Pittsburgh. Details of each cell line are indicated in Table 1. All cell lines were previously shown to have no immunoreactive protein for the mutated gene (not shown).

Untargeted metabolomics

Five replicates of control and IVDH, IBDH, and SBCADH deficient patient derived fibroblast cell lines were grown in culture in T175 flasks to 90% confluence in complete DMEM media containing 5% FBS. Cells were harvested and analyzed by Metabolon, Inc (Morrisville, NC) using their Meta GA platform as previously described (7,8). Data were transformed to multiples of the mean and the resultant Z-scores were analyzed using Metabolon proprietary software as described.

Stable isotope labeling

Control and BCAA pathway deficient patient derived fibroblast cell lines (Table 2) were grown in T175 culture flasks to 90% confluence. On day 1, cells were treated with trypsin, counted, and seeded into three 6-well plates at approximately 150,000 cells/well in complete DMEM media with 5% FBS. On day 2, media was switched to DMEM containing 5% dialyzed FBS. On day 3, wells were washed with ~2 mL 75 μM ammonium carbonate (pH 7.5) for <1 min, aspirated, then 2.5 ml of fresh DMEM + 5% dialyzed FBS containing 0.8 mM uniformly ¹³C_n,¹⁵N-labeled BCAA (Cambridge Isotopes, Cambridge, MA) was added to triplicate wells, calculated to bring labeled substrate to 50% enrichment. Cells were incubated at 37 °C for 10, 30, 60, and 180 mins. At each time point, wells were briefly washed with 75 μM ammonium carbonate (pH 7.5), cold extraction buffer (80/20 MeOH/H²O) was added in a 1 μL: 1,000 cell ratio, and plates were incubated on dry ice or −80 °C for 10 min. The extraction buffer was transferred to Eppendorf tubes, and samples were centrifuged at 12,000 rpm for 10 min at 4 °C. Supernatants were transferred to a 96-well plate, dried in a vacuum centrifuge (>1 hr), sealed and stored at −80 °C until analysis. Replicate wells undergoing identical growth and media change procedures were measured with a Vi-Cell instrument (Beckman Coulter, Indianapolis, IN) to provide cell count information.

Table 2.

Characteristic of fibroblasts used in this study

Gene defect	Gene mutation	Protein change	Passage number	Patient characteristics
Control	None	None	5-10	8 years old
IVDH	149G>C	R21RP	6-10	10 days old
ACADS	455T>C	M152T	2-10	Newborn
ACADSB	443C>T	T148I	2-10	Newborn
PCCA	Unknown*		2-10	Newborn

^*Diagnosed by enzyme assay and western blotting only

Labeled metabolite analysis

96-well plates were analyzed via mass spectrometer by Agios Pharmaceuticals as previously described (9). Briefly, qualitative LC/MS was conducted using a Thermo Vanquish Flex pump delivered a gradient of 0.025% heptafluorobutyric acid, 0.1% formic acid in water and acetonitrile at 400 μl min–1. The stationary phase was an Atlantis T3, 3 μm, 2.1 mm × 150 mm column. Data was acquired on a QExactive mass spectrometer operated at 70,000 resolving power in full-scan ESI positive mode. Data analysis was conducted in MAVEN and Spotfire (10). Peak areas derived from stable isotope labelling experiments were corrected for naturally occurring isotope abundance with in-house developed software. Data were mined for unlabeled and labeled amino acids and propionyl-carnitine.

RESULTS

Metabolomic analysis of ACAD deficient fibroblast recapitulate clinical measurements

Untargeted metabolomics was used to characterize metabolic differences in fibroblast cell lines derived from controls and ACAD deficient patients. The acylcarnitines related to each ACAD defect were the most significantly increased when compared to control fibroblasts (Figure 1B). Specifically, IVD, SBCAD, and IBD deficient cells accumulate isovalerylcarnitine, 2-methylbutyrylcarnitine, and isobutyrylcarnitine, respectively. In addition to being the diagnostic markers observed in patient blood, these acylcarnitines are reasonable surrogates to identify the associated levels of the acyl-CoAs which accumulate in cells but are analytically challenging to detect.

To probe how BCAAs are catabolized in control and ACAD deficient fibroblasts, we provided uniformly ¹³C, ¹⁵N-labeled amino acids individually and analyzed cell extracts for isotope incorporation into downstream metabolites. These catabolites included ¹⁵N-Glutamate (GLU), indicative of the combined branched chain amino acid transferase (BCAT1/2) reactions, as well as metabolites downstream of the ACADs, namely acylcarnitines as surrogates for acyl-CoAs (Figure 2A). In control cells, intracellular levels of isotopically labeled BCAAs rapidly equilibrated with media, reaching 50% enrichment after 10 min of labeling (Figure 2B). However, a greater percentage of GLU was ¹⁵N labeled from ILE indicating a preference ILE over VAL by BCAT1 and/or BCAT2 (Figure 2D). Similarly, it then follows that ILE was the main contributor to ¹³C₃-propionylcarnitine (Figure 2C).

Figure 2:

A. A representative labeling scheme from ILE to propionyl-CoA. Three carbons are used from both ILE and VAL to produce propionyl-CoA and propionyl-carnitine. The incorporation of ¹³C into propionyl-carnitine was used to assess ILE and VAL specific BCAA pathway activity. B. Accumulation of isotopically labeled amino acids in cells. Upon addition of isotopically labeled ILE and VAL, intracellular pool reached steady state in under 15 min. In control cells, ILE was shown to be the major BCAA source of carbon to the propionyl-CoA (C) pool which was regulated at the BCAT step (D).

Given the propensity of SBCADH to accept the alternate substrate isobuturyl-CoA when concentrations increase >130 μM, we next determined whether IBDH deficient fibroblasts, which show 150-fold increase in isobutyrylcarnitine, would shunt carbon from VAL through SBCADH to downstream metabolites. In IBDH deficient fibroblasts, we observed a small but consistent ¹³C₃ labeling of propionyl carnitine derived from VAL (Fig 3A). This suggests that with no IBDH activity, propionylcarnitine derived from VAL was produced through SBCADH. In control cells where isobutyryl-CoA does not accumulate, little to no labeling of propionylcarnitine was observed, reinforcing the idea that most propionylcarnitine is derived from ILE, with the additional insight that the promiscuity of the SBCADH enzyme is induced by accumulation of a non-preferred substrate in the context of IBDH deficiency.

Figure 3: ¹³C₃ propionyl-carnitine derived from VAL.

A. Labeled metabolite was only measured in IBD mutated cells as a result of the significant increase in isobutyryl-CoA, which is then dehydrogenated by SBCADH. B. ¹³C₃ propionyl-carnitine via ILE was observed in both control and SBCADH mutated cells. The SBCADH mutation leads to only 2-3% protein activity thus reducing carbon flux into ¹³C₃ propionyl-carnitine. Methylbutyryl-CoA also increases significantly, increasing the probability of dehydrogenation via IBDH. C. In cells with IBD deficiency, ILE does not appear to contribute more to the propionyl-carnitine pool.

Similarly, we probed the potential promiscuity of IBDH in the setting of SBCADH deficient cells. In this case, SBCADH deficient cells were grown in labeled ILE. We observed that 16% of the total propionylcarnitine pool was derived from the labeled ILE in SBCADH mutants, as compared to 22% in control cells (Fig 3B). Since this cell line had no residual SBCADH protein, these findings suggest that IBDH can metabolize the accumulated 2-methylbutyryl-CoA in the SBCADH deficient cell line. Of note, there was no compensatory increase in the flux from ILE to propionyl-CoA in IBDH deficient cells, the percentage of the propionylcarnitine labeled via ILE in IBDH deficient and control cells are similar (Fig 3C).

Propiogenic metabolites contribute differently to the propionyl-CoA pool

The results from control cells suggest that the popiogenic amino acids ILE and VAL contribute differentially to the intracellular propionate pool, an issue of importance when considering dietary management of patients with propionic and methylmalonic acidemias (11,12). To further examine this issue, control cells were separately cultured in media supplemented with universally labeled VAL, MET, ILE, or THR, and accumulation of propionylcarnitine was measured. Cells were also grown in the presence of an odd chain fatty acid, ¹³C₃-5,6,7-heptanoic acid, another propiogenic compound. Interestingly, the major contributors to the propionate pool were ILE and heptanoate (C7). Labeled VAL, MET, and THR produce little to no measurable propionylcarnitine (Fig. 4A). Similarly, cells deficient in propionyl-CoA carboxylase accumulated significant quantities of !3Cpropionylcarnitine only when supplemented with labeled ILE or heptanoate, though the relative realtionship of the two is reversed, suggesting that the contribution of MET, THR, and VAL are minimal (Figu. 4B).

Figure 4:

A. Isotopically labeled propiogenic substrates were added to control fibroblasts to determine contribution to the cellular propionyl-CoA pool. Accumulation of 13C labeled propionyl (C3) carnitine from labled ILE, VAL, LEU, MET, THR, and heptanoic acid is shown. B. Isotopically labeled propiogenic substrates were added to propionyl CoA carboxylase deficient patient derived fibroblasts to determine the contribution to the cellular propionyl-CoA pool.

DISCUSSION

Mutations that lead to deficiencies at the ACAD step in BCAA catabolism cause disparate clinical symptoms (1). Patients with IVDH mutations often present with severe symptoms requiring immediate dietary intervention (13,14). The pathogenesis of this disease could be either a result of increased isovaleryl-CoA derivatives or a lack of acetyl-CoA production from LEU. While flux studies show a decreased of acetyl-CoA production from LEU, undirected metabolomics showed an overall increase in acetyl-CoA in IVDH deficient cells, suggesting compensatory production from other pathways, likely fatty acid oxidation and/or other carbon sources via pyruvate dehydrogenase (15-18). Interestingly, patients with SBCADH or IBDH mutations are usually asymptomatic, despite significant acyl-CoA increases similarly to isovaleryl-CoA. The promiscuity of SBCADH and IBDH may help explain the disparate symptoms seen in deficiencies of these two enzymes compared to IVDH deficiency (5). In the case of IVD, the lack of alternative metabolism by a related ACAD could allow isovaleryl-CoA concentrations to accumulate to high enough levels to either be toxic, trigger creation of additional metabolites by alternative enzymatic reactions, or both (19,20). In contrast, the metabolic consequences of SBCADH and IBDH deficiencies, through utilization of the alternate ACAD, appear to be sufficient to reduce or eliminate associated pathophysiology. Our results raise an independent possibility. Efficient BCAA metabolism could also be facilitated if the branched chain enzymes existed in a multi-protein complex enabling rapid catabolism of BCAA, a possibility suggested by common proximal steps for the three pathways. We have previously presented data demonstrating association of fatty acid oxidation proteins in a common protein complex that optimizes catalytic efficiency of flux through the pathway (21). Additional experiments will be necessary to further investigate this hypothesis.

Propionyl-CoA, the end product of ILE and VAL metabolism, is a critical metabolic intermediate ultimately necessary as a precursor of succinyl-CoA for the TCA cycle. Our metabolic flux studies revealed that ILE contributed significantly more than VAL to the propionylcarnitine pool in cells. The differential flux appears to be regulated at the BCAT (Fig 2D) step of the pathway, possibly due to higher catalytic proficiency of BCAT on ILE over VAL as substrate (6,22). Another relatively surprising finding was that this ILE-derived metabolite does not effectively reach the TCA cycle as shown by the lack of incorporation of ¹³C from ILE into aspartate in cells. However, this may be a tissue specific limitation (6). Additionally, MET and THR, two other propiogenic amino acids contributed minimally to the cellular propionylcarnitine in control cells, while cells deficient in propionyl-CoA carboxylase that accumulate labeled propionylcarnitine only when supplemented with 1³C-ILE or heptanoate. These findings potentially have significant clinical implications, namely, that in patients with genetic defects of propionyl-CoA metabolism, only ILE may need restriction in the diet, making dietary management potentially simpler, a conclusion reinforced by experiments of BCAA metabolism in mice (6). Recent publications on treatment of propionyl-CoA carboxylase and methylmalonyl-CoA mutase deficiencies advocate relaxation of methionine and threonine restriction due to potential adverse central nervous system effects, and our results are also in keeping with these recommendations (11,12,23,24). Additional experiments are necessary to establish the fate of propionyl-CoA from odd chain carbons in vivo, but clinical studies on the use of a heptanoate to treat TCA cycle intermediate depletion resulting from long chain fatty acid oxidation defects supports the likelihood that it contributes to the TCA cycle (25,26). One might consider BCAAs as an alternate source of TCA cycle intermediates in these conditions; however, our experiments suggest that propionyl-CoA derived from catabolism from BCAAs is unlikely to be effective.

In conclusion, this work provides evidence for functional crosstalk between the enzymes of BCAAs and defines the contribution of the propiogenic amino acids into the cellular propionyl-CoA pool as unequal. In total, these results provide the opportunity for direct benefit of patients with defects in propionate metabolism through improved dietary manipulations and development of drugs that improve stability of the BCAA metabolites and/or abrogate the mitochondrial derangements induced by intracellular propionate deficiency.

Supplementary Material

1

Supplemental Table 1. The spreadsheet provides the name of the metabolite, a KEGG pathway reference, a Pubchem reference, an HMDB identifier, the cell line genotype, the fold change of WT, and the log2 fold change over wild type. These data were plotted in Figure 1, with each dot representing a data point from this file.