Main

Five different enzymes with 3-ketoacyl-CoA thiolase activity have been characterized in different subcellular compartments of mammalian tissues. Cytosolic acetoacetyl-CoA thiolase catalyzes the acetoacetyl-CoA synthesis required for cholesterol biosynthesis(1). Peroxisomal 3-ketoacyl-CoA thiolase catalyzes the last step of peroxisomal β-oxidation, which has a role in the degradation of long chain fatty acids(2). In mitochondria, three thiolases have been identified: mitochondrial acetoacetyl-CoA thiolase, a key enzyme of hepatic ketogenesis(3), and two enzymes of the β-oxidation pathway: long chain 3-ketoacyl-CoA thiolase (the β-subunit of the TFP) and MCKAT, which catalyzes the cleavage of long to medium chain and medium to short chain 3-ketoacyl-CoAs(2, 4), respectively. With the exception of the mitochondrial medium chain enzyme, inborn errors of all other known thiolases have been reported(510). Overall, 18 single gene defects related to the fatty acid oxidation pathway have been characterized(11), and new disorders as well as new enzymes are continuously being discovered.

Common clinical features of fatty acid oxidation disorders are metabolic decompensation during fasting, hypoketotic hypoglycemia, and acute or chronic dysfunction of one or more fatty acid-dependent tissues (in particular skeletal muscle, heart, and liver)(1214). Although our understanding of the biochemical and molecular basis of these disorders has improved substantially in recent years, a growing number of candidate patients remain undiagnosed despite extensive laboratory investigations.

We describe here a patient who presented in 1989 with clinical manifestations and biochemical findings that were strongly suggestive of a possible fatty acid β-oxidation disorder. Eight years after his death, we have gathered multiple lines of evidence in vitro suggesting that this patient was affected with mitochondrial MCKAT deficiency, a newly identified mitochondrial fatty acid β-oxidation disorder.

METHODS

Case report. The patient was a full-term male infant born to unrelated parents after an uncomplicated pregnancy. He was delivered by cesarean section because of fetal bradycardia. The weight at birth was 2690 g, Apgar scores were 9 and 9 at 1 and 5 min, respectively. On the 2nd d, he was reffered to a neonatal care center because of excessive weight loss (230 g), poor feeding, and vomiting. Physical examination revealed no abnormality except mild dehydration. Nasogastric feeding with 5% dextrose was well tolerated, but reintroduction of formula feedings caused vomiting and diarrhea which were treated with parenteral fluids and glucose. With the exception of plasma lactate dehydrogenase (1153 IU/L; controls: 88-196), laboratory data on the 3rd d of life were essentially normal, although blood gases and glucose were not measured. A sepsis work-up was negative. Based on these results, formula feedings were gradually reintroduced and the patient remained asymptomatic for 3 d. On the 7th d, he developed respiratory distress, muscle hypotonia, and opisthotonos. Significant laboratory results were as follows: blood glucose 43 mg/dL (2.4 mmol/L), lactate dehydrogenase 714 IU/L, creatinine phosphokinase 257 IU/L (controls, 10-50), and ammonia >400 μg/dL (30-80). Blood gas analysis showed arterial pH 7.43, Po2 85 mm Hg, Pco2 16 mm Hg, HCO3 10 mEq/L, and base excess -10 mEq/L. The calculated anion gap was 30.1 mEq/L.

On the 8th d, the diagnostic possibility of a metabolic disorder was raised, and as a precautionary measure enteral nutrition was discontinued. Intravenous glucose (8.7%) and sodium bicarbonate (52 mEq/L) were given at a rate of 7 mL/h. The metabolic acidosis and hyperammonemia persisted after the i.v. administration of glucose and sodium bicarbonate. On the 9th d, in view of the progressive worsening of his condition and the biochemical parameters, peritoneal dialysis and insulin infusion (2 U/g of glucose) were attempted to correct a persistent hyperglycemia (>400 mg/dL). On the 10th d, fixed, dilated pupils and decerebrate rigidity were noticed, and he required assisted ventilation. A needle muscle biopsy and skin biopsy were taken on the 11th d. At that time, the serum creatinine phosphokinase value was further elevated to 4586 IU/L. Urinary myoglobin and β2-microglobulin were increased to levels exceeding 500 mg/mL (10-100) and 8000 μg/L (50-200), respectively. The patient became anuric and died at 13 d of age after withdrawal of assisted ventilation. Permission for postmortem examination was not granted.

Materials. Fatty acids and dicarboxylic acids were purchased from Sigma Chemical Co. (St. Louis, MO), and Tokyo Kasei (Tokyo, Japan). Fatty acyl-CoA esters were prepared as described previously(10). [1,12-13C]Dodecanedioic acid (99 atom% 13C), [1,14-13C]tetradecanedioic acid, and [1,16-13C]hexadecanedioic acid were custom synthesized by Merck, Sharp and Dohme Isotopes (Montreal, Canada). Proteins used as molecular mass markers for SDS-PAGE were obtained from Bio-Rad (Richmond, VA). All the other reagents were of the highest analytical grade commercially available and used without further purification.

Organic acid analysis. Urine organic acids were analyzed by GC/MS after ethylacetate extraction of a urine volume equivalent to 0.25 mg of creatinine and derivatization as trimethylsilyl derivatives(15). Quantitative determinations of lactic acid, 3-hydroxy butyric acid, and C6-C10 dicarboxylic acids were performed by capillary gas chromatography in reference to the internal standard pentadecanoic acid (100 μg), after positive identification of each peak's mass spectrum by GC/MS. Concentrations were determined by use of calibration curves prepared with commercially available standards added in six increasing concentrations (12.5 to 200 μg/mL; in a ratio to the internal standard concentration 0.25 to 4)(15).

Quantitative determinations of C12-C16 dicarboxylic acid methyl esters were performed by GC/MS-selected ion monitoring using a stable isotope dilution method and ammonia chemical ionization(15, 16). A mixture of the three labeled internal standards (100 μL, 25 μM each in methanol) was added and allowed to equilibrate overnight at +4 °C. For calibration purposes, unlabeled C12, C14, and C16 dicarboxylic acids were added in six increasing concentrations in a ratio to the internal standard concentration 0.1 to 10(16).

Culture of fibroblasts. Fibroblasts from skin biopsy specimens of the patient and normal individuals used as simultaneous controls were grown in modified Eagle's minimum essential medium containing 10% (vol/vol) fetal bovine serum. Monolayers were harvested by trypsinization and washed twice with PBS. Oxidation rate assays were performed on freshly harvested cells; individual enzyme assays were performed on pelleted cells frozen at -80 °C and thawed immediately before experiments.

Fatty acid β -oxidation enzyme assays. The acyl-CoA dehydrogenase activities were assayed in fibroblast homogenates by a dye-reduction method using phenazine ethosulfate and 2,6-dichlorophenolindophenol as described by Verity and Turnbull(17). The enoyl-CoA hydratase activity was assayed by measuring the decrease in absorbance of 2-enoyl-CoAs at 280 nm(18). The 3-hydroxyacyl-CoA dehydrogenase activity was measured by following the decrease in absorbance of NADH at 340 nm using 3-ketoacyl-CoA substrates(19). The activity of 3-ketoacyl-CoA thiolase was assayed by following the decrease in absorbance at 303 nm(20). In the 3-ketoacyl-CoA thiolase assay with the C4, C6, or C8 substrates, the buffer was as follows: 0.1 M Tris-Cl, pH 8.3, 25 mM MgCl2, 50 mM KCl, 0.1% Triton X-100, 0.2 mg/mL BSA, and 20 μM substrate. The thiolase activity with C4 substrate was also measured in the absence of KCl. In the 3-ketoacyl-CoA thiolase assay with the C12 or C16 substrates, the substrate concentration was reduced to 10 μM.

Extraction of enzyme activities from skin fibroblasts. For extraction of acyl-CoA dehydrogenase activities, the fibroblasts were sonicated in 0.1 M potassium phosphate (pH 7.5), 0.5 M sodium chloride, 0.2% (vol/vol) Tween 20, 0.1% (vol/vol) hexamethylphosphoric triamide, 2 mM 2-mercaptoethanol, and 0.5 mM EDTA. For extraction of enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase, and 3-ketoacyl-CoA thiolase activities, fibroblasts were sonicated in 50 mM potassium phosphate (pH 7.5), containing 0.2 M sodium chloride, 0.1% (vol/vol) hexamethylphosphoric triamide, 2 mM 2-mercaptoethanol, and 0.5 mM EDTA. Tween 20, 20% (vol/vol), was added to give a final concentration of 0.5% (vol/vol).

Immunoprecipitation. For 3-ketoacyl-CoA thiolase assays in immunoprecipitation experiments, the fibroblasts were sonicated in 50 mM sodium phosphate, pH 7.5, containing 0.1% (vol/vol) hexamethylphosphoric triamide, 2 mM 2-mercaptoethanol, and 0.5 mM EDTA. After incubation on ice for 30 min, the homogenate was centrifuged at 18 000 × g for 5 min. The supernatant was used for enzyme assays or immunoprecipitation experiments. To completely extract the membranebound fatty acid β-oxidation enzymes, 50 mM potassium phosphate, pH 7.5, 0.2 M sodium chloride, 0.1% (vol/vol) hexamethyl phosphoric triamide, 2 mM 2-mercaptoethanol, 0.5 mM EDTA, and 1.0% (vol/vol) Tween 20 were included in the extracting buffer.

Immunotitration. Fibroblasts' extracts, prepared for the 3-ketoacyl-CoA thiolase assays, were mixed with 5-fold the equivalent amounts of the anti-(rat mitochondrial MCKAT) antibody(21). The mixtures were incubated for 1 h at 4 °C, then centrifuged at 18 000 × g for 5 min, and the supernatants were used for assay of the enzyme. The amount of antibody required to completely precipitate the enzyme protein from the fibroblasts' extracts was determined by antibody titration monitoring residual enzyme activities in the supernatant fractions obtained after centrifugation.

Assay and analysis of protein. Protein concentration was determined by a modification(22) of the method of Lowry et al.(23). SDS-PAGE on a 10% gel was performed according to Laemmli(24). Immunoblot analysis was made according to the manufacturer's protocol for the immunoblot system for color development and ECL® Western blotting detection kits (Amersham, Japan). The bands were quantified using a densitometer, Densitograph® (ATTO, Japan).

Oxidation studies. The oxidation rate of [1-14C]palmitic acid and [1-14C]octanoic acid to acid-soluble products was determined as described earlier(25, 26).

RESULTS

Biochemical investigations. Analyses of amino acids in blood and urine revealed no significant abnormality. Blood lactate and pyruvate levels were 6.8 mmol/L (controls, 1-2) and 0.34 mmol/L (0.08-0.15), respectively. Urine organic acid screening by GC/MS showed lactic aciduria and a very unusual pattern of relatively hypoketotic C6-C16 dicarboxylic aciduria, where saturated and unsaturated C12-C16 dicarboxylic acids were unusually elevated (Fig. 1). Table 1 shows the quantitative results obtained in three urine specimens collected at d 7, 9, and 10, respectively. In the same specimens, the excretions of several short chain (C4-C8) acylglycine species, particularly isobutyrylglycine and butyrylglycine, were variably elevated, with no apparent correlation to the magnitude and relative distribution of the dicarboxylic aciduria.

Figure 1
figure 1

Capillary gas chromatographic profiles of urine organic acid trimethylsilyl derivatives in a patient with MCKAT deficiency. Urine volume is equivalent to 0.25 mg of creatinine. Peak legends: 1, lactic acid; 2, 3-hydroxybutyric acid; 3, adipic acid; 4, octenedioic acid (two isomers); 5, suberic acid; 6, hippuric acid; 7, 3-hydroxy suberic acid; 8, decenedioic acid (two isomers); 9, sebacic acid; 10, 4-hydroxy phenyllactic acid; 11, pentadecanoic acid (internal standard); 12, 4-hydroxyphenylpyruvic acid; 13, 3-hydroxy sebacic acid plus dodecadiendioic acid; 14, dodecenedioic acid; 15, dodecanedioic acid; 16, unknown, probable contaminant; 17, tetradecadiendioic acid; and 18, tetradecenedioic acid. All peak identifications were confirmed by gas chromatography/mass spectrometry. The analysis is completed in 48.0 min.

Table 1 Quantitative analysis of urine organic acids

On d 8, plasma total carnitine concentration was 42 μmol/L (age-matched controls, 17-41), with a prominent esterified fraction (34; controls, 3-24) and an acyl/free carnitine ratio of 4.3 (controls, 0.2-1.4). Unfortunately, no specimens were stored and retrospectively available to perform a plasma acylcarnitine profile.

Histologic analysis of skeletal muscle. Serial frozen sections were stained with hematoxylin and eosin and modified Gomori trichrome. Stains for ATPase, oil-red O, neuronspecific enolase, acid phosphatase, cytochrome c oxidase, AMP deaminase, acetylcholinesterase, phosphorylase, phosphofructokinase, and succinate dehydrogenase were also performed. All histochemistry investigations were uninformative. No fat accumulation was found in the skeletal muscle, and enzyme activities of the electron transport system were normal (data not shown).

Oxidation of [1-14C]palmitic and [1-14C]octanoic acid. [1-14C]Palmitic acid oxidation in the cells from the patient was 1.11 nmol/h/mg of protein, a slight decrease in comparison with control values (1.63 ± 0.41, n = 6). [1-14C]Octanoic acid oxidation in the patient's fibroblasts was 0.59 nmol/h/mg of protein, a value that was below 2 SDs of the mean control value (1.93 ± 0.65, n = 3).

Enzyme activities in cultured skin fibroblasts. Table 2 summarizes the results obtained by assay of nine mitochondrial enzyme activities in fibroblast extracts from controls and from the patient using the corresponding optimal substrate. In fibroblasts from the patient, the levels of activity of three acyl-CoA dehydrogenases, two enoyl-CoA hydratases, two 3-hydroxyacyl-CoA dehydrogenases and long chain 3-ketoacyl-CoA thiolase were comparable to those measured in control cell lines.

Table 2 Activities of fatty acid β-oxidation enzymes in cultured fibroblasts

The only abnormality detected in this set of experiments was a deficiency of thiolase activity. In the absence of K+, the assay of 3-ketoacyl-CoA thiolase activity using 20 μM acetoacetyl-CoA (C4) as the substrate revealed a decreased activity, 38% of controls (Table 2). In the presence of K+, the activity with the same substrate was up to 19.5 nmol/min/mg of protein. These results are consistent with a normal mitochondrial acetoacetyl-CoA thiolase activity.

To better characterize the significant deficiency of thiolase activity with the C4 substrate, we assayed the thiolase activity with substrates of increasing chain length (C4 to C16, ± K+ for the C4 substrate) in total cell extracts and in the immunoprecipitates obtained after incubation with anti-(MCKAT) antiserum (Table 3). In the total cell extract of the patient, all values were below 2 SDs of the respective controls (Table 3 and Fig. 2A). The activity levels in the corresponding immunoprecipitates were 0, 0, 10, 51, and 27% of the control mean, respectively, with the exception of the C16 experiments in which there was no detectable activities in both the patient and controls (Table 3 and Fig. 2C). Figure 2B summarizes the differences of activity observed between the total cell extract and the supernatant after immunoprecipitation. These values were also below 2 SDs of the respective control mean, especially those measured with the C4 and C6 substrates (<5%).

Table 3 Substrate specificity of 3-ketoacyl-CoA thiolase activities in extracts from cultured fibroblasts
Figure 2
figure 2

Carbon chain length specificities of 3-ketoacyl-CoA thiolase activities in fibroblasts extracts. Control (n = 5) () and patient (). The activities were measured before (A), after immunoprecipitation of mitochondria MCKAT, and with the immunoprecipitants. (C). The activities were expressed as nanomoles/min/mg protein. Differences between value before and after immunoprecipitation are plotted (B).

Immunoblot analysis. Figure 3 shows the results of immunoblot analyses of the three mitochondrial thiolases. The signal for mitochondrial MCKAT in the patient's cells (lane 4) was about 60% of two controls (Fig. 3A,lanes 2 and 3). There were no appreciable differences for the signals of mitochondrial acetoacetyl-CoA thiolase and trifunctional protein (Figs. 3,B and C, respectively). Immunoblot analyses of three acyl-CoA dehydrogenases (short, medium, and very long chain), long chain enoyl-CoA hydratase, 3-ketoacyl-CoA dehydrogenase, mitochondrial TFP, peroxisomal bifunctional protein, and peroxisomal 3-ketoacyl-CoA thiolase also showed no appreciable differences between the patient and control cell lines (data not shown).

Figure 3
figure 3

Immunoblot analysis of mitochondrial MCKAT. MCKAT (A), acetoacetyl-CoA thiolase (B), and trifunctional protein (C). Lane 1, purified enzyme (10 ng); lanes 2 and 3, control fibroblasts (10 μg of proteins); lane 4, fibroblasts of patient (10 μg of protein). Arrowheads indicate positions of subunits of the mature enzymes (C, β-subunit of trifunctional protein).

We also examined the amount of mitochondrial MCKAT in immunoprecipitation experiments. The immunoprecipitants obtained by incubation with anti-(mitochondrial MCKAT) antiserum were subjected to SDS-PAGE and immunoblot analysis. In the patient's cell line (Fig. 4,lane 3), the band of mitochondrial MCKAT was about 60% of two controls (Fig. 4A,lanes 2 and 4). Immunoblot signals were not detected using anti-(peroxisomal 3-ketoacyl-CoA thiolase) antibody as control (Fig. 4B).

Figure 4
figure 4

Immunoblot analysis of immunoprecipitants with anti-(mitochondrial MCKAT) antibody. The immunoprecipitants were suspended in the extracting buffer (see “Methods”) and subjected to SDS-PAGE analysis. The applied samples were immunoprecipitated from 30 μg of protein extracts. The transferred membranes were immunoblotted with anti-(mitochondrial MCKAT) antibody (A) or anti-(peroxisomal 3-ketoacyl-CoA thiolase) antibody (B). Lane 1, purified enzyme (30 ng); lane 2, control 1; lane 3, patient; lane 4, control 2.

Because the amount of mitochondrial MCKAT was slightly reduced in the patient's cells, we also studied the degradation rate of the protein in pulse-chase experiments. In cells from the patient, the incorporation of [35S]methionine into mitochondrial MCKAT after 1-h pulse labeling was 140% of the control as determined by densitometric measurement of the fluorographic bands (Fig. 5). However, the catalytically deficient protein was visibly degraded at 6 and 24 h, whereas the control protein remained virtually unchanged.

Figure 5
figure 5

Pulse-chase analysis of mitochondrial MCKAT. Lane 1, 1-h pulse; lane 2, 6-h chase; lane 3, 24-h chase; lane 4, competition experiment of 1-h pulse. Arrowhead indicates positions of the subunits of mature enzymes.

DISCUSSION

Mitochondrial fatty acid β-oxidation is mediated by a sequence of enzymes, all of which exhibit specificity for substrates of different chain length. The enzyme-step sequence is: acyl-CoA dehydrogenase (VLCAD, LCAD, MCAD, and SCAD); 2-enoyl-CoA hydratase (α-TFP and crotonase); 3-hydroxyacyl-CoA dehydrogenase (α-TFP and SCHAD); and 3-ketoacyl-CoA thiolase (β-TFP and MCKAT). The degree of biochemical and molecular characterization of these enzymes varies considerably, and our understanding of the tissue expression, substrate specificity, and regulation of several of them is still limited.

Long chain fatty acid oxidation disorders frequently present with fasting-induced hypoketotic hypoglycemia and dicarboxylic aciduria, acute liver dysfunction, myopathy with potentially severe cardiac involvement, and fatty infiltration of parenchymal organs(1214). In medium to short chain fatty acid oxidation disorders, fasting intolerance and hypoketotic hypoglycemia are the main clinical findings at presentation. Cardiac involvement is less common in these disorders.

In this case, the biochemical findings were not characteristic of a particular disorder, and for several years after his death this patient remained undiagnosed. In our experience, this is not an uncommon outcome, as a growing number of cases worldwide do not receive a specific diagnosis despite compelling clinical and biochemical evidence consistent with an underlying metabolic disorder. After the initial presentation with vomiting and failure to thrive, the patient's course was complicated by persistent metabolic acidosis, hyperammonemia, and hypoglycemia poorly responsive to glucose infusion. Terminally, myoglobinuria and renal failure supervened with marked elevation of serum creatine phosphokinase. Myoglobinuria may occur in a number of disorders of fatty acid β-oxidation, such as carnitine palmitoyltransferase II deficiency(27), VLCAD deficiency(28), and TFP deficiency(29). Because the plasma total and acyl/free carnitine ratio were essentially normal in association to marked dicarboxylic aciduria, a diagnosis of severe infantile form of the carnitine palmitoyltransferase II deficiency was considered unlikely. Myoglobinuria is also typical of muscle glycogen storage diseases, such as phosphofructokinase deficiency(30, 31) and muscle phosphorylase deficiency(32). Rare clinical variants of muscle phosphorylase deficiency have been diagnosed in the infantile period. However, excessive glycogen accumulation was not observed in the muscle of our patient by histologic analysis. This finding and the absence of hemolytic anemia were not consistent with a possible diagnosis of phosphofructokinase deficiency.

Although the initial biochemical evidence pointed to a possible mitochondrial fatty acid oxidation disorder, the markedly elevated excretion of saturated and unsaturated C12-C16 dicarboxylic acids and the moderate but unusual excretion pattern of acylglycine species were not indicative of a specific disorder(33). Although long chain C12 to C16 dicarboxylic aciduria has been reported in VLCAD deficiency(28), a possible diagnosis of this disorder was excluded by in vitro studies.

The enzyme assays of the fibroblast extracts and immunotitration studies pointed consistently to a deficiency of MCKAT as the underlying primary defect in this patient. In view of the evidence gathered in vitro, a possible explanation of the unusual biochemical results may be provided by the toxic accumulation and inhibitory effects of accumulated 3-keto fatty acid intermediates. Davidson and Schulz(34) showed that SCAD is strongly inhibited by acetoacetyl-CoA, whereas 3-ketodecanoyl-CoA has a similar effect on MCAD and LCAD. Because we documented a significant decrease of MCKAT activity in the patient's fibroblasts, especially with C4 and C6 substrates, we postulate that the accumulation of 3-ketoacyl-CoAs could cause a secondary inhibition of different acyl-CoA dehydrogenases, leading to the unusual biochemical profile seen in this patient.

The residual activity found with C6 to C12 substrates in the total fibroblast extract (Table 3 and Fig. 2A) was unexpected. We suspect this activity could be the contribution of the peroxisomal 3-ketoacyl-CoA thiolase. By Western blot analysis, this enzyme was indeed identified in the cell extract used for the immunoprecipitation experiments. Unfortunately, our attempt to completely eliminate this activity from the total cell extracts using either anti-(rat peroxisomal 3-ketoacyl-CoA thiolase) polyclonal antibody or anti-(human peroxisomal 3-ketoacyl-CoA thiolase) polyclonal antibody were not successful (data not shown).

Using anti-(mitochondrial MCKAT) antibody, we have shown that the decrease of the 3-ketoacyl-CoA thiolase activities was caused by the quantitative and qualitative abnormality of the thiolase. Previous reports have shown that mutant thiolases degraded more rapidly than the normal enzyme(10, 35, 36). Although the rate of synthesis of the subunit was increased in fibroblasts from the patient, we observed a rapid degradation of the protein in pulse-chase experiments, suggesting that protein instability plays a role in determining the defect in catalytic activity. However, immunoprecipitation experiments showed that the level of medium chain 3-ketoacyl-CoA thiolase was approximately 60% of control cells, with no evidence of differences in molecular weight. Immunotitration studies indicated that 3-ketoacyl-CoA thiolase activities with short chain substrates (C4 and C6) were especially impaired. Therefore, the qualitative abnormality of the subunit was likely to have changed the substrate specificity of mitochondrial MCKAT. Each one of the mitochondrial thiolases has different substrate specificity. The affinity of mitochondrial acetoacetyl-CoA thiolase is limited to the C4 substrate(3). MCKAT exhibits higher activities with short to medium chain substrates, but TFP is more active with medium to long chain substrates and inactive with the C4 substrate(37, 38). Although the cDNAs of two thiolases have been sequenced and the amino acid sequences are known(39, 40), the domain that determines the substrate specificity is unknown. Mutation analysis of our patient could be instrumental in the recognition of the catalytic domain of this protein.

As in the case of other inborn errors of metabolism of recent characterization(89, 41), we hope this report could lead to both retrospective and prospective recognition of additional patients with the same disorder, and eventually to a deeper insight of the mechanisms underlying the unique biochemical phenotype observed in our patient. Our results also underscore the importance of complementing standard catalytic assays with immunotitration experiments using specific antibodies as an effective strategy to reach a diagnosis in atypical and unexplained defects of fatty acid β-oxidation.