Hexacosenoyl-CoA is the most abundant very long-chain acyl-CoA in ATP binding cassette transporter D1-deficient cells[S]

X-linked adrenoleukodystrophy (X-ALD) is an inherited disorder caused by deleterious mutations in the ABCD1 gene. The ABCD1 protein transports very long-chain FAs (VLCFAs) from the cytosol into the peroxisome where the VLCFAs are degraded through β-oxidation. ABCD1 dysfunction leads to VLCFA accumulation in individuals with X-ALD. FAs are activated by esterification to CoA before metabolic utilization. However, the intracellular pools and metabolic profiles of individual acyl-CoA esters have not been fully analyzed. In this study, we profiled the acyl-CoA species in fibroblasts from X-ALD patients and in ABCD1-deficient HeLa cells. We found that hexacosenoyl (26:1)-CoA, but not hexacosanoyl (26:0)-CoA, was the most abundantly concentrated among the VLCFA-CoA species in these cells. We also show that 26:1-CoA is mainly synthesized from oleoyl-CoA, and the metabolic turnover rate of 26:1-CoA was almost identical to that of oleoyl-CoA in both WT and ABCD1-deficient HeLa cells. The findings of our study provide precise quantitative and metabolic information of each acyl-CoA species in living cells. Our results suggest that VLCFA is endogenously synthesized as VLCFA-CoA through a FA elongation pathway and is then efficiently converted to other metabolites, such as phospholipids, in the absence of ABCD1.

severity or onset of clinical phenotypes (6). Thus, it remains unknown how VLCFA accumulation contributes to the progression of X-ALD, although recent observations propose that aberrant VLCFA metabolism may be associated with a proinflammatory state leading to the progression of X-ALD (7). Most VLCFAs are not present in the free form but are incorporated into complex lipids such as phospholipids (PLs) in humans (3) and mice (8). Total VLCFA contents have been profiled by gas chromatography-MS (9,10), and recent progress in LC-MS analysis with a comprehensive lipidomic approach has revealed the endogenous form of VLCFAs in X-ALD patients to comprise part of PLs, glycolipids, and neutral lipids (11)(12)(13).
Fatty acyl-CoA (acyl-CoA) is an active form and serves as a metabolic intermediate of FAs (14). Acyl-CoA is composed of both a hydrophobic fatty acyl moiety and a hydrophobic CoA joined with a thioester linkage. This amphiphilic property hampers the efficient separation of individual acyl-CoA species when using conventional hydrophobic or hydrophilic interaction chromatography, and thus, the intracellular pool of each acyl-CoA ester has not yet been fully analyzed. In this study, we developed a LC-MS-based method to comprehensively profile the quantities and rates of metabolism of each acyl-CoA species in ABCD1-deficient cells.

Ethics
This study abides by the Declaration of Helsinki Principles, and the research protocol was approved by the Ethics Committee of Teikyo University (#12-078-6). Primary human fibroblasts were established from skin biopsy samples of four CCALD, three AMN, and five control patients. Informed consent was obtained from the human subjects or their representatives.

Synthesis of D 31 -palmitoyl-CoA
Deuterium-labeled palmitoyl-CoA (D 31 -16:0-CoA) was synthesized as described previously with slight modifications (15). Briefly, 2.5 mg of FA D 31 -16:0 and 2 ml of oxalyl chloride were mixed by stirring overnight at room temperature in a 10 ml ground-glass test tube to yield acyl chlorides. The excess oxalyl chloride was removed with nitrogen gas, and 2 ml of oxalyl chloride were added into the same test tube. After mixing for 1 h at room temperature, the excess oxalyl chloride was removed completely with nitrogen gas. The oil residue of FA chloride was dissolved in 0.5 ml of freshly distilled tetrahydrofuran (THF) and added into a CoA solution [10 mg of CoA dissolved in 2.3 ml of THF containing 0.1 M of Tris-HCl (pH 7.4) in a 2:1 ratio by volume] in a 10 ml screw-capped test tube. The acid chloride solution was added slowly to maintain the pH at 8 with the use of 1 M NaOH. Once all of the acid chloride solution had been added, the reaction mixture was adjusted to pH 4 with 10% HClO 4 , and the THF was removed with nitrogen gas. Then, 1 ml of double-distilled water and 0.3 ml of 10% HClO 4 were added into the reaction mixture, and the resulting white precipitate was collected by centrifugation (2,150 g for 10 min) at 4°C. The precipitate was washed with 4 ml of diethylether-petroleum ether (1:1 by volume), and the residue was dissolved in 100 l of isopropanol followed by purification with HPLC. An Inertsil SIL-100A column (7.6 mm i.d. × 250 mm, particle size 5.0 m; GL Science, Tokyo, Japan) was used at room temperature. The mobile phases were methanol/water (1:1 by volume) supplemented with 5 mM ammonium formate and 0.032% NH 4 OH. The flow rate was 1.5 ml/min. Each eluent fraction was diluted with methanol and directly infused into the mass spectrometer (Thermo LXQ; Thermo Fisher Scientific, Waltham, MA) to obtain MS/MS spectra. Fractions containing pure D 31 -16:0-CoA were combined and evaporated to dryness with the EZ-2 centrifugal evaporator (Genevac, Ipswich, UK). The obtained D 31 -16:0-CoA was reconstituted with methanol and stored at 20°C.

Sample preparation
Sample preparation for acyl-CoA analysis was conducted as reported previously (16). Cell layers in the 100 mm culture dish were washed twice with PBS, scraped from the dishes, and collected by centrifugation (1,000 g for 5 min) at 4°C. Cell pellets were homogenized in 0.9 ml of acetonitrile/isopropanol (3:1 by volume). Homogenate protein concentrations were determined with a BCA protein assay kit (Thermo Fisher Scientific). Cell homogenates (approximately 0.5-1 mg protein) were spiked with 100 pmol of D 31 -16:0-CoA as the internal standard (IS) and 300 l of 0.1 M KH 2 PO 4 (pH 6.7) were added. After centrifugation (19,120 g for 5 min) at 4°C, each supernatant was mixed with 300 l of acetic acid and loaded onto a 2-(2-pyridyl)ethyl silica gel column (Sigma-Aldrich) preconditioned with 1 ml of wash buffer (acetonitrile/isopropanol/water/acetic acid at 9:3:4:4 by volume). The loaded column was washed once with the wash buffer followed by the application of 1 ml of elution buffer (5 mM ammonium formate in a water/isopropanol mixture of 1:4 by volume). The eluent was evaporated completely with the EZ-2, and the resulting precipitate was reconstituted with 0.1 ml of methanol followed by filtration with a YMC Duo-Filter (4 mm i.d., pore size 0.2 m; YMC Co., Ltd., Kyoto, Japan). Samples were stored at 20°C until analysis. For PL analysis, cell pellets were homogenized with 1.0 ml of methanol and the total lipid fraction was extracted as previously described (17).

LC-MS/MS analysis
Quantitation of each acyl-CoA species was conducted with a QTRAP 4500 (Sciex, Framingham, MA) linked to a Nexera XR HPLC system (Shimadzu Corp., Kyoto, Japan). A Capcell Pak C 8 UG120 column (1.5 mm i.d. × 35 mm, particle size 5.0 m; Shiseido Co., Ltd., Tokyo, Japan) was used at 40°C. The mobile phases were 5 mM ammonium formate in water (pH 9.0) (mobile phase A) and 5 mM ammonium formate in a water/isopropanol solution (5:95 by volume; pH 9.0) (mobile phase B). The programmed solvent gradient consisted of solvents A/B at a 60/40 ratio for 2 min, programmed linear increments to 0/100 over 13 min, after which it was held at 0/100 for 2 min, and then linear increments to 60/40 over 1 min, after which it was held at 60/40 for 2 min. The flow rate of the mobile phase was 200 l/min, and the volume of the injected samples was 5 l. Multiple reaction monitoring (MRM) transitions were constructed to cover acyl-CoA species with 12-32 carbons and zero to six double bonds present in the acyl moieties. Each MRM transition was constructed by selecting protonated molecules ([M+H] + ) and fragmented molecules ([MC 10 H 15 N 5 O 13 P 3 ] + ) corresponding to the loss of a neutral fragment of a 3′-phosphoadenosine diphosphate as the precursor and product ions, respectively. The time period for data collection was 10 ms per cycle for each MRM transition. The following conditions were used for positive ion MRM: ion spray voltage, 5,500 V; temperature (TEM), 300°C; curtain gas (CUR), 40 arbitrary units (A.U.); collision gas (CAD), 9 A.U.; ion nebulizer gas (GS1), 40 A.U.; auxiliary gas (GS2), 80 A.U.; quadrupole mass filter (Q1) and Q3 linear ion trap (Q3/LIT) resolution, "unit"; declustering potential (DP), 1 V; entrance potential (EP), 10 V; collision energy (CE), 52.5 V; and collision cell exit potential, 12 V. Nitrogen was used as the nebulizer, curtain, and collision gas. Analyst software and MultiQuant software (Sciex) were used for data acquisition and processing. For structural analysis of VLCFA-CoA species, each acyl-CoA species was separated by HPLC, and the product ion spectra of VLCFA-CoA were obtained in the negative ion mode employing the MS 3 scan to increase the signal-to-noise ratio of the product ion. The structure of each acyl-CoA species was assigned by the detection of the product ions corresponding to [M-phosphate-adenosine monophosphate]  (m/z = M-348) and [adenosine diphosphate-H 2 O]  (m/z = 408). Quantitative and structural analyses of the PL species were conducted with the LC-MS/MS and LC-MS 3 method, respectively (8,18).

Method validation
Sample solutions for a spiked calibration curve were prepared as follows. A 100 M stock solution of 17:0-CoA (Sigma-Aldrich) in methanol was prepared as the standard and this was diluted further with methanol to prepare standard solutions of 0.05, 0.1, 0.5, 1, 5, 10, 50, and 100 M. A 10 M stock solu- ) were observed. Statistical analysis was performed with one-way ANOVA followed by the Tukey post hoc test in B and the Student's t-test in C. *P < 0.05 for comparisons between pH 4.0, pH 6.5, pH 9.0, and pH 10.5 (B) or between NH 4 OH and +NH 4 OH (C). tion of D 31 -16:0-CoA in methanol was prepared as the IS. Then, 50 l of IS solution and each diluted standard solution were placed into a 2.0 ml siliconized plastic tube and mixed with 0.6 ml of HeLa cell homogenate containing 0.6 mg protein in acetonitrile/isopropanol (3:1 by volume), followed by the addition of 200 l of 0.1 M KH 2 PO 4 (pH 6.7). Next, the supernatant was collected by centrifugation (19,120 g for 5 min) at 4°C and mixed with 200 l of acetic acid. Each sample was loaded onto a 2-(2-pyridyl)ethyl silica gel column (Sigma-Aldrich) preconditioned with 1 ml of wash buffer (acetonitrile/isopropanol/water/acetic acid at a ratio of 9:3:4:4 by volume), and processed using a procedure identical with that previously described above (Sample preparation section). For validation of the method, three samples containing 0.05, 1, and 10 pmol of standard per injection were analyzed for quality control (QC) purposes. To generate a linear regression curve, 1/x 2 was used as a weighting factor (19). Accuracy was calculated as: [(observed concentration  endogenous concentration)/nominal concentration  1] × 100 (%) and the coefficient of variation was evaluated to determine measurement precision.

Cell lines and cell culture
HeLa cells were obtained from the cell bank of the Riken Bioresource Center (Ibaraki, Japan). Three clonal WT HeLa cell lines were generated by the limiting dilution method, and one of these clones was used to generate ABCD1-KO HeLa cell lines. Human skin fibroblasts and HeLa cells were cultured in minimum essential medium (Sigma-Aldrich) supplemented with 10% FBS (Biowest, Nuaillé, France), 2 mM l-glutamine (Thermo Fisher Scientific), 100 U ml

Generation of ABCD1-KO HeLa cells with the CRISPR/Cas9 system
To establish the ABCD1-KO HeLa cell lines, we designed two guide RNAs for the human ABCD1 gene using the CRISPR Three samples with 0.050, 1.0, and 10 pmol of 17:0-CoA were mixed with 10 pmol of (D 31 )-16:0-CoA per injection and were analyzed as QC compounds QC-L (low), QC-M (middle), and QC-H (high), respectively.

Fig. 2.
Quantity of each acyl-CoA species in the X-ALD fibroblasts. A: Each acyl-CoA species in fibroblasts from four CCALD, three AMN, and five control patients was quantified in the positive ion mode, and was classified according to the number of carbons and double bonds in the acyl moiety. The ratio of peak area for each acyl-CoA/D 31 -16:0-CoA was used to calculate the amount of each acyl-CoA species, and the mean quantity of each acyl-CoA species in the X-ALD fibroblasts is represented with a color key. The data are also summarized in Table 2     The total amount of each VLCFA-CoA species that contain very long-chain fatty acyl moieties with the number of double bonds (DBs) as indicated.

Statistical methods
Statistical analysis was performed with either one-way ANOVA followed by the Dunnett T3 or Tukey post hoc test or with the Student's t-test or Mann-Whitney U test. Differences were considered to be significant if the P-value was <0.05. All statistical analyses were conducted with IBM SPSS Statistics version 23 (IBM, Armonk, NY). Spectral data were plotted with MjoGraph software (Ochiai Laboratory, Yokohama National University, Japan).

Development of an LC-MS method for acyl-CoA analysis
Free FAs are linked to CoA by acyl-CoA synthetase and pooled as intermediates of metabolic processes. To explore how ABCD1 deficiency influences the metabolism of each FA, we developed an LC-MS method for acyl-CoA analysis and used it to profile intracellular acyl-CoA species. Due to the amphiphilic property of acyl-CoA, ion-pairing reagents have been used to separate each acyl-CoA species by the reversed-phase HPLC methods developed thus far (22). However, it is desirable to avoid the usage of ion-pairing reagents because it is difficult to remove these reagents completely from the HPLC column after use. Among the LC-MS methods developed to date for acyl-CoA analysis, we referred to the method reported by Magnes et al. (23) for further development in which ion-pairing reagents were not used to separate each acyl-CoA species. In their study, C 18 reversed-phase columns were used at high pH (pH 10.5) with an ammonium hydroxide and acetonitrile gradient. Here, we utilized C 8 reversed-phase columns with an ammonium hydroxide and isopropanol gradient to efficiently elute the VLCFA-CoA species. We found that each acyl-CoA species was separated with high sensitivity when both mobile phases were adjusted to pH 9.0 by ammonium hydroxide (Fig. 1A). We also tested acidic (pH 4.0), weakly acidic (pH 6.5), and high pH (pH 10.5) conditions and found that the maximum number of theoretical plates for each acyl-CoA species was obtained when the mobile phases were adjusted to pH 9.0 (Fig. 1B). Because the thioester bond of acyl-CoA is sensitive to hydrolysis at high pH, we next examined whether the acyl-CoA species were degraded in the mobile phases at pH 9.0. We incubated 17:0-CoA with the mobile phases for 1 h before analysis with various ratios of water to isopropanol and either with or without ammonium hydroxide. We did not observe any significant effect of the pH of the mobile phases (Fig. 1C) and concluded that the acyl-CoA species were not degraded under the HPLC gradient conditions used in this study.
IS compounds are critical for LC-MS-based quantitation (24). Stable isotopically labeled analogs are desirable as IS compounds because these nonnatural compounds can be quantified without any signal overlap with natural compounds, and such standards can properly compensate for variability in sample extraction and LC-MS analysis. Therefore, we chemically synthesized D 31 -16:0-CoA and confirmed its structure by MS/MS analysis (Fig. 1D). We then validated our method for the quantitation of each acyl-CoA species. Two acyl-CoA species were used to construct a spiked calibration curve that was applicable to acyl-CoA species with cell homogenates as a biological matrix, namely 17:0-CoA and D 31 -16:0-CoA, which served as the standard compound and the IS, respectively. Spiked standard solutions were analyzed using a scheduled MRM mode and the linearity was examined over the range of 0.05-100 pmol per injection ( Table 1). Accuracy and precision values for the 1 and 10 pmol per injection were within 15%, and these values for the 0.05 pmol per injection were identical with the lower limit of the quantitation range and were almost within 20% (Table 1). These results showed that 0.05-100 pmol of acyl-CoA species could be quantified with the present quantitative method.

Quantitative analysis of acyl-CoA species in X-ALD fibroblasts
Using the LC-MS method we developed for acyl-CoA analysis, we first profiled each acyl-CoA species in fibroblasts from four CCALD, three AMN, and five control patients. Among the 112 acyl-CoA species analyzed in the fibroblasts cultured with medium containing 10% FBS, we found 43, 42, and 44 species to be within the quantitative range in X-ALD fibroblasts from CCALD, AMN, and the control patients, respectively ( Fig. 2A, Table 2). VLCFA-CoA species with no less than 24 carbons in their fatty acyl moiety were significantly accumulated in both CCALD and AMN fibroblast samples compared with the control, and hexacosenoyl-CoA (26:1-CoA) was the most and the second most abundant acyl-CoA species in both AMN and CCALD fibroblast samples, respectively (Fig. 2B, Table 2). It is also notable that the proportion of VLCFA-CoA species with one and two double bonds was significantly higher in both CCALD and AMN fibroblasts ( Fig. 2A, Table 2). Accumulated acyl-CoA pools in cells are partially used for the synthesis of a variety of PL species (25). To examine possible correlations between the profiles of acyl-CoA species and PL species, we quantified each PL species in both CCALD and AMN fibroblasts. Among the PL species observed within the quantitative range, eleven PC species ( Table 4. Acyl-CoA species observed to be present below the quantitation range are indicated in gray. C: Each PC, PE, and SM species present in significantly higher quantities (magenta) in the ABCD1-KO HeLa cell lines (#1 to #3) compared with three independent clones of WT HeLa cells were classified according to the number of carbons and double bonds in the two acyl moieties. No PL species were present in significantly lower quantities than in the control. PL species observed to be present below the quantitation range are indicated in gray. The species for which there were no MRM channels designed are indicated as N.A. The quantity of each PL species in the ABCD1-KO HeLa cells is listed in supplemental Table S2 at significantly higher levels in CCALD fibroblasts as compared with control fibroblasts (Fig. 2C, supplemental Table  S1). Notably, structural analysis by LC-MS 3 revealed that five (24:0, 24:1, 24:2, 26:0 and 26:1) and two (26:0 and 26:1) fatty acyl moieties were mainly present as VLCFAs in six of the PC species (PC 40:1, 42:1, 42:2, 42:3, 44:1, and 44:2) and two of the SM species (SM 44:1 and 44:2), respectively ( Table 3, supplemental Figure S1). Only one SM species (SM 38:1), which was identified as SM d18:1/20:0, was present at significantly lower levels in CCALD fibroblasts (Fig. 2C, supplemental Table S1, supplemental Fig. S1). In contrast, two PC species (PC 32:0 and 32:2), two PE species (PE 34:4 and 40:3), and five SM species (SM 32:1, 34:2, 44:1, 44:2, and 44:3) were present at a significantly higher level in AMN fibroblasts as compared with control fibro-blasts (Fig. 2C, supplemental Table S1, supplemental Fig.  S1). Significant difference was not observed in the amount of PC species with VLCFAs, such as PC 44:1 and 44:2, mainly due to the wide penetrance differences in the quantity of these PC species in three AMN fibroblasts (supplemental Table S1). These results show that acyl-CoA species with 26 carbons and zero to two double bonds in their acyl moieties are mainly accumulated as VLCFA-CoA pools and are possibly transferred into complex lipids, such as PC and SM. Interestingly, significant differences were observed in the amount of three acyl-CoA species (20:2-, 22:5-, and 24:5-CoA) and two PC species (PC 32:0 and PC 38:0) between CCALD and AMN fibroblasts ( Fig. 2A, C; Tables 2, 3; supplemental Table S1). These results may reflect the difference in the metabolism of saturated and polyunsaturated long-chain FAs between CCALD and AMN fibroblasts.

Quantitative analysis of acyl-CoA species in ABCD1deficient HeLa cells
Skin fibroblast cells from X-ALD patients senesce and show limited cell proliferation. To further examine the effect of ABCD1 deficiency on cellular acyl-CoA metabolism in other cell types, we generated three independent ABCD1-KO HeLa cell lines with the CRISPR/Cas9 system (Fig. 3A) (21) and compared them with three independent clones of WT HeLa cells. Penetrance differences were observed in the quantity of each acyl-CoA species, although mono-and di-unsaturated VLCFA-CoA species were significantly accumulated in all three ABCD1-KO HeLa cells (Fig. 3B, Table 4), as was likely observed in the X-ALD fibroblasts ( Fig. 2A, Table 2). The amount of 26:1-CoA was highest among the VLCFA-CoA species in ABCD1-KO (#1 to #3) and control HeLa cells (Fig. 3B, Table 4). Through quantitative and structural PL analysis, only two PC species (PC 44:0 and 46:4), which contain 26:0-and 28:0-fatty acyl moieties, respectively, were found to be present in significantly higher quantities in the ABCD1-KO HeLa cells (#1 to #3) (Fig. 3C, supplemental Fig. S2, Table 5,  supplemental Table S2). Taken together, these results show that the metabolism of mono-and di-unsaturated VLCFA-CoAs is mainly affected through ABCD1 dysfunction in both fibroblasts and HeLa cells, and that the accumulated VLCFA-CoAs are preferably incorporated into PC and SM species.

Hexacosenoyl-CoA (26:1-CoA) is not efficiently synthesized from 26:0-CoA in ABCD1-deficient HeLa cells
To  (Fig. 4A). In contrast, D 4 -26:1-CoA was present in much smaller quantities in the ABCD1-KO cells compared with the unlabeled 26:1-CoA, and it was not accumulated in a time-dependent manner (Fig. 4A, B). These results suggest the presence of acyl-CoA synthases, which utilize FA 26:0 as a substrate, although 26:1-CoA was not efficiently synthesized from 26:0-CoA through a FA desaturation process. The quantity of each PL species in the three WT and three ABCD1-KO (KO) HeLa cell lines (picomoles per milligram of protein). n.q., below the quantitation range.  It was previously shown that FA 26:0 is intracellularly synthesized from FA 18:0 through FA elongation, and the protein, elongation of very long-chain FAs 1 (ELOVL1), plays a critical role in the synthesis of VLCFAs (1,26). Thus, we next examined whether 26:1-CoA can be synthesized through the FA elongation machinery from the 18:1-CoA that is abundantly present as a long-chain fatty acyl-CoA species in cells (Figs. 2A, 3B; Tables 2, 4). To this end, we labeled the ABCD1-KO HeLa cells with deuterium-labeled oleic acid (FA D 2 -18:1) and analyzed the endogenous rate of synthesis of D 2 -24:1- and D 2 -26:1-CoA as well as D 2 -18:1-CoA (Fig. 5). The D 2 -18:1-, D 2 -24:1-, and D 2 -26:1-CoA levels immediately increased and reached a plateau within an hour after treatment with FA D 2 -18:1 in ABCD1-KO HeLa cells (Fig. 5A), showing that 26:1-CoA can be efficiently synthesized from 18:1-CoA through the endogenous FA elongation process after the activation of FA 18:1 by acyl-CoA synthetase. Note that the levels of D 2 -18:1-CoA and D 2 -26:1-CoA did not significantly change, while D 2 -24:1-CoA markedly decreased in the ABCD1-KO cells during the first 1-8 h after treatment (supplemental Fig. S3A). These results are consistent with the results obtained for the PL species with VLCFAs in their acyl moieties in ABCD1-KO HeLa cells (Table 5). Thus, C 24 fatty acyl-CoAs are transferred to PC species more efficiently than C 26 fatty acyl-CoAs. Interestingly, the quantity of nonlabeled 18:1-CoA did not significantly alter during the observation period following treatment (Fig. 5B); in contrast, the nonlabeled VLCFA-CoAs (24:1-, 24:2-, 26:1-, 26:2-, 28:1-, and 28:2-CoA) were significantly reduced within the first hour after treatment and remained at low levels (150 pmol/mg protein) after treatment with 30 M of FA D 2 -18:1 (Fig. 5B, supplemental  Fig. S3B). These results suggest that FA 18:1 itself or the metabolites of FA 18:1 could negatively regulate the endogenous synthesis of VLCFA-CoAs. Given that the pool of each fatty acyl-CoA species was maintained by some homeostatic machinery, it may be also possible that the amount of D 2 -26:1-CoA increased and then nonlabeled 26:1-CoA decreased inversely.

The pool of monounsaturated VLCFA-CoAs is turned over as efficiently as 18:1-CoA in both WT and ABCD1deficient HeLa cells
It is believed that acyl-CoA species are stored intracellularly and function as essential intermediates for FA metabolism. However, little is known regarding the metabolic turnover of each acyl-CoA species. This is at least partially due to the lack of a highly sensitive and precise quantitative method for acyl-CoA detection. To clarify the difference in metabolic turnover between each acyl-CoA species under the ABCD1-deficient condition, a pulse-chase study was conducted using deuterium-labeled oleic acid (FA  (Fig. 6A). Most of the deuterium-labeled acyl-CoA species decreased immediately after medium replacement (Fig.  6A). In contrast, D 2 -28:1-CoA levels did not significantly change within at least the first 6 h after medium replacement and decreased gradually over the following 18 h (Fig.  6A). This suggests that the elongation process from 18:1-CoA to 26:1-CoA is highly efficient, while the conversion from 26:1-CoA to 28:1-CoA is not. To further analyze the VLCFA-CoA metabolism quantitatively, we conducted regression analysis by applying the exponential model, thereby obtaining data on the half-life of each acyl-CoA species. The half-life of most deuterium-labeled acyl-CoA species ranged from 0.5 to 2 h, and the differences in this value between the WT and ABCD1-KO HeLa cells were no more than 10% (18:1-CoA, 20:2-CoA, 22:1-CoA, 24:1-CoA,   Table 6). Notably, the half-life of 26:1-CoA was almost identical with that of 18:1-CoA in both the WT and ABCD1-deficient HeLa cells, suggesting that these VL-CFA-CoAs are metabolized with high efficiency even in the absence of ABCD1. Interestingly, the half-life of 24:1-CoA was almost half that of the other acyl-CoA species (Table  6), supporting the hypothesis that C 24 fatty acyl-CoAs are transferred to PC species more efficiently than to C 26 fatty acyl-CoAs in ABCD1-KO HeLa cells, resulting in the accumulation of PC species with C 24 fatty acyl moieties as described in the previous section. Taken together, these results suggest that VLCFA-CoA species with one or two double bonds are efficiently metabolized, possibly through being incorporated into PC and SM. It is notable that while the quantities of nonlabeled VLCFA-CoAs (22:1-CoA, 24:1-CoA, 24:2-CoA, 26:1-CoA, 26:2-CoA, 28:1-CoA, and 28:2-CoA) initially increased during the first 4 h and then gradually decreased over the remainder of the chase period, those of nonlabeled 18:1-CoA and 20:1-CoA were not significantly altered (Fig. 6B). These results may be attrib-utable to certain unknown and unstable factors in FBS, which stimulate the synthesis and the metabolic flux of VLCFAs.  (Figs. 4, 5). Thus, the quantity of FA 18:1 present as the initial substrate will affect the VLCFA-CoA profile in ABCD1deficient cells. Furthermore, the activity of acyl-CoA synthase, which synthesizes 18:1-CoA from FA 18:1, will also affect the quantity of VLCFA-CoAs in ABCD1-deficient cells.

DISCUSSION
It is clinically important to develop biochemical markers to estimate the progression of X-ALD. In the present study, The half-life (t 1/2 ) was based on t 1/2 = ln (2)/k.  Tables 2, 3; supplemental Table S1). Considering that -linolenic acid (FA 18:3) is converted to FA 22:5 and FA 24:5 resulting in docosahexaenoic acid (FA 22:6) (2), these results may indicate the difference in the efficiency of the metabolism of polyunsaturated FAs involving several elongases and desaturases between CCALD and AMN fibroblasts. In this study, we analyzed human fibroblasts from skin biopsy samples of two sibling X-ALD patients with an identical mutation on the ABCD1 gene (c1825G>A); one X-ALD fibroblast was from a patient with CCALD phenotype, and the other X-ALD fibroblast was from a patient with AMN phenotype. We could not find significant differences in the amount of each acyl-CoA and PL species between these two X-ALD fibroblasts (data not shown). Further analysis including the dynamics of VLCFA metabolism is necessary to discover the novel features underlying differences in the severity of clinical symptoms between X-ALD subtypes. Several experiments using stable isotope labeling have been conducted to analyze the fate of acyl-CoA species. For example, U-13 C-palmitic acid was administered to rabbits and 13 C isotope enrichment based on the ratio of 16:0-CoA to U-13 C-16:0-CoA was determined in the muscle tissues by LC-MS (27). Similarly, deuterium-labeled 4-hydroxynonenal was used to analyze the catabolism of 4-hydroxy acids (C 4 to C 11 ) (28). However, the intracellular kinetics of each long-chain fatty acyl-CoA species has not yet been analyzed. In this study, we precisely analyzed the kinetics of each acyl-CoA species with pulse-chase experiments using FA D 2 -18:1. Interestingly, the kinetics of 18:1-CoA and 26:1-CoA were almost identical, and 24:1-CoA was turned over twice as rapidly as 18:1-CoA in both the WT and ABCD1deficient cells (Table 6). Possible explanations for this finding could be that VLCFA-CoAs are rapidly released from the cells or that VLCFA-CoAs are efficiently introduced into PC and SM species in the absence of ABCD1. To address this issue, we conducted the pulse-chase experiments using D 2 -FA 18:1 and harvested cells at 0, 1, 2, 4, 6, and 24 h after the replacement of medium, and found that the amount of D 2 -PC 42:2, D 2 -PC 44:2, and D 2 -SM 42:2 were increased in a time-dependent manner. Especially, the amount of D 2 -SM 42:2 was much higher than that of D 2 -PC 42:2 and D 2 -PC 44:2 (data not shown). These results indicate that VLCFA-CoA is preferably introduced into sphinganine to form ceramide and SM in HeLa cells. We observed significant accumulation of VLCFA-containing PC and SM in both fibroblasts and ABCD1-deficient HeLa cells. VLCFA moieties, such as 24:1- or 26:1-fatty acyl, are located at the sn-1 position of the glycerol backbone of PC or bind to the amine of the long-chain sphingoid base of SM. Therefore, these results strongly suggest that 2-acyl lysophosphatidylcholine (LPC) acyltransferases and ceramide synthase are likely involved in the synthesis of VLCFA-containing PC and SM, respectively (29,30).
Total FAs have been conventionally quantified so far by analyzing the methyl-esterified acids from both free FA and esterified FAs (9). Upon comparison of these acyl-CoA profiles in the X-ALD fibroblasts with the FA profiles previously reported (10,31), it was clear that the relative abundance of each acyl-CoA was significantly different to that of the corresponding FA. Among the VLCFA-CoAs, for example, 26:1-CoA was the most abundantly concentrated in X-ALD fibroblasts and ABCD1-KO HeLa cells and the level of 26:1-CoA was 5- to 6-fold higher than that of 26:0-CoA, while the amount of total FA 26:1 was similar to that of FA 26:0 (10, 31). As another example, the quantities of 26:0-CoA and 24:0-CoA were almost identical, although the amount of total FA 26:0 was much lower than that of FA 24:0 in the X-ALD fibroblasts (10,31). These discrepancies may be explained by the endogenous forms of the VLCFAs present; 26:0-CoA may be more efficiently transferred into complex lipids such as PL and then stored compared with 26:1-CoA. It is likely that 24:0-CoA may be a preferred substrate for PL synthesis compared with 26:0-CoA.
In previous studies, odd-numbered long-chain fatty acyl-CoA species, such as 17:0-CoA, were used as the IS (22,23). However, significant levels of PLs with odd-numbered FAs are observed in mammalian tissues as we have reported previously (8,32), raising the possibility that the total amount of IS in each sample is influenced by the endogenous odd-numbered fatty acyl-CoA species. To circumvent this issue, we chemically synthesized D 31 -16:0-CoA as the IS compound. As expected, significant amounts of odd-numbered acyl-CoA species were observed in both the fibroblasts and HeLa cells. For instance, the proportion of 17:0-CoA to total acyl-CoA quantified was 2.6-2.9% (11.4-42.1 pmol/mg protein) in the fibroblasts and 1.1-1.4% (1.8-2.5 pmol/mg protein) in the ABCD1-KO HeLa cells, showing the advantages of stable isotopically labeled fatty acyl-CoA in the quantitative analysis of acyl-CoA species by LC-MS. In this study, D 31 -16:0-CoA was used as the IS compound and the peak area of each acyl-CoA species was normalized by the peak area of D 31 -16:0-CoA. Because each acyl-CoA did not co-elute with the IS, our present method corresponds to the level 3 type of quantitation as defined by the Lipidomics Standards Initiative (https://lipidomicsstandards-initiative.org). To estimate the difference of ionization efficiency between each long chain acyl-CoA and VLCFA-CoA species, we generated the two dilution series of 17:0-CoA and 26:0-CoA and analyzed the peak area of each analyte. We found that the electrospray ionization efficiency of 26:0-CoA was almost one-ninth of 17:0-CoA (data not shown). Because both acyl chain length and retention time of D 31 -16:0-CoA are close to those of 17:0-CoA rather than 26:0-CoA, these results indicate that the amounts of VLCFA-CoA species with fewer number of double bonds in their acyl chain moiety were underestimated in this study, and the amount of 26:1-CoA may be much higher than that of 18:1-CoA in both X-ALD fibroblasts and ABCD1-deficient HeLa cells. The ionization efficiency can be influenced by ion suppression or ion enhancement depending on the mobile phase conditions and the presence of coeluting compounds (24), and it is desirable to prepare an IS compound for each acyl-CoA species for absolute quantitation. If stable isotopically labeled CoA is abundantly available, the preparation of each IS compound may be facilitated by a condensation reaction between each nonlabeled FA and the labeled CoA as conducted in this study.
In conclusion, we profiled the pool of intracellular acyl-CoA in X-ALD fibroblasts and ABCKD-KO HeLa cells and revealed that 26:1-CoA, but not 26:0-CoA, was the most abundantly accumulated VLCFA-CoA in both cell types. In addition, we conducted pulse-chase experiments using stable isotope-labeled FA and clarified that 26:1-CoA was efficiently synthesized from FA 18:1. Furthermore, we showed that 26:1-CoA is efficiently turned over in the absence of ABCD1, illustrating the efficiency of the ABCD1-independent machinery that metabolizes VLCFA-CoA.
The PX458 vector was a gift from Feng Zhang through Addgene (Watertown, MA). The authors thank Drs. A. Inoue and J. Aoki (Tohoku University) for their assistance with the CRISPR/Cas9 system. The authors also thank T. Yamanobe, K. Kurosaki, and colleagues for their technical assistance with MS.