Plasmalogen loss caused by remodeling deficiency in mitochondria

³¹P NMR spectra in a SDS micellar solution show that the plasmenylcholine signal is not resolved as a minor component of the human lymphoblast and the mouse organs (brain, liver, and kidney) except the heart, which is a rare organ rich in plasmenylcholine (Diagne et al, 1984; May et al, 1988; Kikuchi et al, 1999; Kimura et al, 2018). The signal of plasmanylcholine as a minor component (Diagne et al, 1984; May et al, 1988; Kikuchi et al, 1999; Kimura et al, 2018) overlaps with that of diacyl PC (Kimura et al, 2018). The measured chemical shift value of a class or subclass of phospholipid extracted from those biological materials slightly depends on the organ and cell type as shown in Tables S1, S2, S3, and S4. This is due likely to some co-extracted materials that are present and depend on the organ/cell type, thus influencing differentially the molecular environment of the phosphate group of the same phospholipid class or subclass. The spectra are aligned with corrections for these minor chemical shift differences given in Tables S1, S2, S3, and S4.

Biosynthesis of plasmalogen initiates in peroxisomes, where dihydroxyacetone phosphate (DHAP) is acylated in its sn-1 position by glycerone phosphate O-acyltransferase (1: GNPAT), followed by formation of the characteristic ether linkage in the early precursor of plasmalogen via exchange of the sn-1 acyl group with an alkyl group by alkylglycerone phosphate synthase (2: AGPS). A fatty alcohol used for this introduction of the sn-1 alkyl ether group is derived from a fatty acid by the enzymatic action of fatty acyl-CoA reductase 1 (Far1) (Bishop & Hajra, 1981). Far1 is known as a rate-determining enzyme of the plasmalogen synthesis pathway, whose level is regulated by a feedback mechanism in response to the plasmalogen level (Honsho et al, 2010; Kimura et al, 2018). 1-O-alkyl-DHAP is reduced to 1-O-alkyl-2-hydroxy-glycero-3-phosphate (1-O-alkyl-2-hydroxy-G3P) by acyl/alkyl-DHAP reductase (3) located in both the peroxisomal and the ER membranes. Reaction proceeds in the ER to form 1-O-alkyl-2-acyl-G3P by alkyl/acyl-glycerophosphate acyltransferase (4) (Yamashita et al, 2014), 1-O-alkyl-2-acyl-G by PA phosphatase named lipin (5) (Zhang & Reue, 2017), 1-O-alkyl-2-acyl-GPE (GPC) by the ethanolaminephosphotransferase (EPT) or cholinephosphotransferase (CPT) activity (6), and 1-O-alk-1′-enyl-2-acyl-GPE, that is, plasmenylethanolamine by plasmanylethanolamine desaturase (7). Plasmenylcholine is formed from plasmenylethanolamine in pathways either with or without replacement of the sn-2 acyl chain. One pathway without the acyl chain replacement is via formation of 1-O-alk-1′-enyl-2-acyl-G, such as by phospholipase C among several possible routes (8), followed by the plasmenylcholine formation by the CPT activity (15). Another pathway including the acyl chain replacement is via formation of 1-O-alk-1′-enyl-2-hydroxy-GPE by phospholipase A₂ (9), and 1-O-alk-1′-enyl-2-hydroxy-GP either upon cleavage of the ethanolamine headgroup by lysophospholipase D (10) or upon cleavage of the phosphoethanolamine headgroup by lysophospholipase C (11) with subsequent phosphorylation by phosphotransferase (12). 1-O-alk-1′-enyl-2-hydroxy-GP is acylated by acyl-CoA acyltransferase (13) to form 1-O-alk-1′-enyl-2-acyl-GP, whose phosphate is released by phosphohydrolase (14) to form 1-O-alk-1′-enyl-2-acyl-G.

(A) High-resolution ³¹P NMR spectra of healthy control (blue trace) and BTHS (red trace) lymphoblast phospholipids in the region showing ethanolamine glycerophospholipids and CLs. The spectra are shown for the whole cell, the crude mitochondria fraction, and the ER fraction. Measurements were conducted in aqueous solution of SDS micelles dissolving the lipids. The opposing trends in (i) the relative levels of plasmenylethanolamine and diacyl PE in mitochondria and (ii) those in the ER accord with data in a previous report (Arthur et al, 1985). These trends contribute to their comparable levels in the healthy whole cell (blue trace). The plasmenylethanolamine loss (14.4 ± 1.9 → 10.8 ± 1.1 mol % of total phospholipid) and the diacyl PE gain (13.6 ± 1.1 → 21.1 ± 1.4 mol %) in tafazzin deficiency observed in the whole cell (Table S4; N = 3) reflect the plasmenylethanolamine loss (11.2 ± 1.0 → 9.2 ± 0.7 mol %) and the diacyl PE gain (22.6 ± 1.7 → 28.5 ± 1.0 mol %) observed in the crude mitochondria fraction. Three independent subcellular fractionation experiments on three independent culture of lymphoblast cells were conducted (N = 3). No significant changes in their levels are seen in the ER fraction. The observation of the changes in the state of CL is also limited to the crude mitochondria fraction because of the predominant location of CL in the inner mitochondrial membrane of the cell (Daum, 1985; Horvath & Daum, 2013). *A part of the CL loss is masked in the spectra by contributions from the resonance of a phosphate group in the diacyl half of MLCLs (Kimura et al, 2018). (B) MALDI-TOF MS experiments were conducted to measure changes in the levels of individual ethanolamine glycerophospholipid species in the crude mitochondria fraction of BTHS lymphoblast compared with healthy control. Plots of molar ratios of plasmenylethanolamine/diacyl PE are shown on the dominant species in lymphoblast (18:0/18:1 [36:1] and 18:0/20:4 [38:4]) (Xu et al, 2005) and on the species of particular interest in transacylation by tafazzin (18:0/18:2 [36:2]) (Vreken et al, 2000; Schlame et al, 2002, 2003, 2005; Valianpour et al, 2005; Kiebish et al, 2013). The error is shown as the SD (N = 5). P values are from the t test. The measured plasmenylethanolamine/diacyl PE ratios <1 and the decreases in the ratios in BTHS compared with control are in accord with the NMR data showing the same result from the total amounts.

(A) PA, which is abundant in the mammalian brain, undergoes the characteristic second deprotonation of its phosphate located at the headgroup terminus with a pK_a2 value of 7.9 (Kooijman et al, 2005). The molecular species shown is a representative species in the mammalian brain (Chan et al, 2012). (B) At pH 6.0, the PA signal appears in the spectral region where there is significant overlap with the 2-MLCL signal. The second deprotonation induced by increasing the pH to 8.5 shifts the PA signal largely downfield, so that both PA and 2-MLCL contents in the TAZ-KD mouse brain can be unambiguously quantified. (C) ³¹P NMR spectra of the WT (blue trace) and TAZ-KD (red trace) brain phospholipids measured at pH 8.5. *A part of the CL decrease is masked by contributions from the resonance of a phosphate group in the diacyl half of MLCLs (Kimura et al, 2018).