Plasmalogen biosynthesis is spatiotemporally regulated by sensing plasmalogens in the inner leaflet of plasma membranes

Alkenyl ether phospholipids are a major sub-class of ethanolamine- and choline-phospholipids in which a long chain fatty alcohol is attached at the sn-1 position through a vinyl ether bond. Biosynthesis of ethanolamine-containing alkenyl ether phospholipids, plasmalogens, is regulated by modulating the stability of fatty acyl-CoA reductase 1 (Far1) in a manner dependent on the level of cellular plasmalogens. However, precise molecular mechanisms underlying the regulation of plasmalogen synthesis remain poorly understood. Here we show that degradation of Far1 is accelerated by inhibiting dynamin-, Src kinase-, or flotillin-1-mediated endocytosis without increasing the cellular level of plasmalogens. By contrast, Far1 is stabilized by sequestering cholesterol with nystatin. Moreover, abrogation of the asymmetric distribution of plasmalogens in the plasma membrane by reducing the expression of CDC50A encoding a β-subunit of flippase elevates the expression level of Far1 and plasmalogen synthesis without reducing the total cellular level of plasmalogens. Together, these results support a model that plasmalogens localised in the inner leaflet of the plasma membranes are sensed for plasmalogen homeostasis in cells, thereby suggesting that plasmalogen synthesis is spatiotemporally regulated by monitoring cellular level of plasmalogens.

. In contrast, treatment with chlorpromazine (CPZ), an inhibitor of clathrin-dependent endocytosis and macropinocytosis 27 , lowered the synthesis of both plasmalogens and PE in a similar fashion, indicating that the plasmalogen synthesis was not specifically reduced by CPZ, rather incorporation of 14 C-Etn and/or the common pathway(s) for the synthesis of both plasmalogens and PE were suppressed (Fig. 2, lanes 2 and 4). However, expression and peroxisomal localisation of Dhapat and Agps were not altered in the presence of Dynasore ( Supplementary Fig. S1A,B). We next assessed the expression of Far1, a rate-limiting enzyme of plasmalogen synthesis 4,5 . Protein level of Far1 was reduced upon the treatment with Dynasore in CHO-K1 (Fig. 3A, lanes 2 and 3), which was further decreased in the presence of CHX (Fig. 3A, lane 4). In contrast, expression of Far1 was not altered by treatment with Dynasore in agps ZPEG251 (Fig. 3B), where accumulation of transferrin (Tf) in recycling endosomes was inhibited with Dynasore as reported 23 (Supplementary Fig. S1C, lower panel). Transcription of FAR1 in CHO-K1 cells was not reduced in the presence of Dynasore (Fig. 3C). Together, these results suggest that plasmalogen-induced degradation of Far1 is augmented by Dynasore. However, total cellular level of plasmalogens was not elevated in the presence of Dynasore (Fig. 3D).
Dynasore inhibits GTPase activity of dynamin-like protein 1 (Dlp1) both in vitro 23 and in vivo 28,29 . Therefore, we investigated whether Dynasore-mediated inactivation of Dlp1 induces degradation of Far1 using dlp1 CHO mutant ZP121 30 . Consistent with the partial defect of plasmalogen synthesis in ZP121 30 , expression level of Far1 in ZP121 was higher than that in CHO-K1 cells ( Supplementary Fig. S1D), implying that inactivation of Dlp1 is not the primary cause for the reduction of the expression level of Far1 upon the treatment with Dynasore in CHO-K1 cells. Dynasore also reduces cellular cholesterol and disrupts rafts 31 . However, we did not observe the reduction of cholesterol when CHO-K1 cells were incubated with Dynasore (Fig. 3E). Finally, we verified the expression level of Far1 with myristyl trimethyl ammonium bromide (MiTMAB), an inhibitor that inhibits GTPase activity of dynamin by disrupting the interaction between pleckstrin homologydomain of dynamin and phospholipids 32 , and found that the expression level of Far1 was reduced to about 70% of that in mock-treated CHO-K1 cells upon the treatment with MiTMAB (Fig. 3F). Taken together, these results suggest that inhibition of dynamin-mediated endocytic pathways most likely accelerates plasmalogen-induced degradation of Far1.
Knockdown of flotillin-1 enhances plasmalogen-induced degradation of Far1. We further investigated whether inhibition of caveolin-or flotillin-1 (Flot1)-mediated endocytosis enhances plasmalogen-induced degradation of Far1. To this end, plasmalogen-deficient agps ZPEG251 cells were cultured with purified plasmalogens in the presence of PP2, an inhibitor specific for Src-kinase essential for both types of caveolin-dependent and Flot1-mediated endocytosis 33,34 . Expression of Far1 was lowered in ZPEG251 cells upon culturing with purified plasmalogens as compared to that in untreated ZPEG251, which was further reduced in the presence of Dynasore (Fig. 4A, lanes 1-3), confirming that inhibition of dynamin-mediated endocytosis induces degradation of Far1 (Fig. 3A, lanes 2 and 3). Protein level of Far1 was similarly reduced in ZPEG251 in the presence of PP2 and purified plasmalogens, whereas PP2 alone did not affect the expression level of Far1 (Fig. 4A, lanes 4 and 5). Under these conditions, total amount of cellular plasmalogens was not altered by the treatment with Dynasore and PP2 as compared with those in ZPEG251 cells cultured with purified plasmalogens (Fig. 4B)    of Far1 was reduced by nearly 50% upon knocking down FLOT1 using two independent double-stranded RNAs (dsRNAs) (Fig. 5A, lower panel), where Flot1 protein level was lowered to about 20% of that in mock-treated cells (Fig. 5A, upper panel). Consistent with these results, the knockdown of FLOT1 reduced the plasmalogen synthesis to about 50% level of that in HeLa cells (Fig. 5B), whereas transcription of FAR1 and total cellular plasmalogen  level were not altered (Fig. 5C). In contrast, protein level of Far1 was not altered under the more than 60% reduction of caveolin1 (Cav1) expression in HeLa cells treated with CAV1 RNAi (Fig. 6A). Next, we examined whether elevation of plasmalogens stimulates degradation of Far1 in HepG2 cells, a cell line that shows no detectable caveolins and caveolae 35 but expresses Flot1 (Fig. 6B). When HeLa cells were cultured in the presence of Etn, cellular plasmalogen level was increased about 1.5-fold as compared to that in mock-treated cells 16 , thereby stimulating the degradation of Far1 by sensing the elevated level of plasmalogens ( Fig. 6B,C, lanes 1 and 2) 4,5 . Similar extent of the degradation of Far1 was observed in HepG2 cells as the plasmalogen level was increased by supplementation of Etn (Fig. 6B,C, lanes 3 and 4), implying that plasmalogen-induced degradation of Far1 is not abrogated under the suppressed caveolin-mediated endocytosis. Together, these results suggest that inhibition of Flot1-mediated endocytosis most likely enhances plasmalogen-induced degradation of Far1.

Sensing of plasmalogens in the inner leaflet of plasma membrane. Given the findings that
plasmalogen-induced degradation of Far1 is enhanced by the knockdown of Flot1 without elevating total cellular plasmalogen level (Fig. 5C), we suspected that the degradation of Far1 is accelerated by sensing plasmalogens  locally enriched in the plasma membrane upon inhibiting Flot1-mediated endocytosis. In this hypothesis, elevation of plasmalogens in lipid rafts might be a prerequisite for the sensing because plasmalogens are enriched in lipid rafts 2,36 and knocking down of FLOT1 inhibits the endocytosis of raft-associated proteins such as CD59, i.e. a glycosylphosphatidylinositol (GPI)-linked protein, and a cholera toxin B subunit in HeLa cells 37 . Therefore, we verified the amount of plasmalogens and sphingomyelin (SM) in the detergent-resistant membranes (DRMs) in FLOT1-knocked down HeLa cells (Fig. 7). When HeLa cells were solubilized with Triton X-100 on ice and separated into DRMs and Triton X-100 soluble fractions by floatation on a 30% iodixanol gradient, Cav1 was mostly recovered in DRMs, while Tf receptor (Tfr), a non-raft protein, was completely solubilized by Triton X-100 as reported 2 (Fig. 7B, lanes 1 and 2). In FLOT1-knocked down HeLa cells, Cav1 and SM, not Tfr, were recovered in DRMs similar to those in mock-treated HeLa cells where the transfection of one of the dsRNAs  n.s., not significant by two-way ANOVA. (G) Lipids were extracted from cells cultured as in (F). Cholesterol was detected as described in Fig. 3E. Relative amount of cholesterol is represented by taking as 100 that at (0 h, -Nys.). *p < 0.05 and **p < 0.01; two-way ANOVA with Tukey post hoc test. n.s., not significant. Note that cellular cholesterol was reduced by the treatment of CHX but not with nystatin. against FLOT1 induced protein level of Tfr (Fig. 7B,C). However, amount of plasmalogens in DRMs prepared from FLOT1-knocked down HeLa cells was about 1.5-fold higher than that in mock-treated cells (Fig. 7D), suggesting that plasmalogens are locally enriched in lipid rafts upon reducing the expression of Flot1. We further examined if plasmalogens in the plasma membrane are sensed by sequestering cholesterol in the plasma membrane. To this end, CHO-K1 cells were treated with nystatin, a cholesterol-chelating agent, inducing distortion of the structure and function of cholesterol-rich membrane domain 38 and assessed for the degradation of Far1 in the presence of CHX. Degradation of Far1 was significantly suppressed by the treatment with nystatin (Fig. 7E) without reducing total cellular level of plasmalogens and cholesterol (Fig. 7F,G). We interpreted these results to mean that the plasmalogen-induced degradation of Far1 is most likely retarded by interfering with the sensing steps of plasmalogens localised in the plasma membrane. Finally, we assessed if plasmalogens located in the inner leaflet of the plasma membrane are sensed. It is known that aminophospholipids such as phosphatidylserine (PS) and PE are concentrated in the inner leaflet of the plasma membrane, where type-IV P-type ATPases (P4-ATPases) function in the phospholipid transfer from the outer leaflet to the inner leaflet 39 . Since plasmalogens are also shown located in the inner leaflet of plasma membrane in red blood cells and myelin 40,41 , we verified the protein level of Far1 by reducing the expression of CDC50A encoding a β -subunit of P4-ATPases that is essential for exiting of most of P4-ATPases from the ER by forming a heterooligomer 42,43 . Knockdown of CDC50A using two independent dsRNAs elevated the protein level of Far1 (Fig. 8A,B). We also determined distribution of plasmalogens at the outer leaflet with 2,4,6-trinitrobenzene sulfonic acid (TNBS), a membrane impermeable amine-reactive reagent 44 . TNBS-modified PE was detected with a higher mobility on TLC than PE. Upon the trichloroacetic acid (TCA)-treatment, by which the vinyl ether linkage of plasmalogens is hydrolysed, thereby giving rise to two more slower-migrating bands each corresponding to TNBS-plasmalogens (TNBS-2-acyl-GPE) and unmodified plasmalogens (2-acyl-GPE) (Fig. 8C), where TNBS-plasmalogens and TNBS-PE represented 3.7 ± 0.2% and 6.8 ± 0.5%, respectively, of total plasmalogens and PE (Fig. 8C, lane 4). Knockdown of CDC50A elevated the levels of TNBS-plasmalogens and TNBS-PE (Fig. 8D) and augmented the synthesis of plasmalogen and PE (Fig. 8E) where total amount of PE and plasmalogens was not significantly increased (Fig. 8F). Collectively, these results suggest that plasmalogens located in the inner leaflet of plasma membrane are sensed for detection of the cellular level of plasmalogens.

Discussion
Plasmalogen synthesis is regulated by a feedback mechanism in a manner dependent on the cellular level of plasmalogens 4,5 . In the present study, we showed that plasmalogen-induced degradation of Far1 is enhanced by the inhibition of dynamin-, Src kinase-, and Flot1-mediated endocytosis (Figs 3,4 and 5), whereas perturbation of inner leaflet localisation of plasmalogens or sequestering of cholesterol in the plasma membrane suppressed the degradation of Far1 without altering total cellular level of plasmalogens (Figs 7 and 8). Together, these results suggest that plasmalogens localised in the inner leaflet of the plasma membrane is critical for sensing of the cellular level of plasmalogens.
Flot1 is shown to be enriched in lipid rafts of the plasma membrane 37,45 and its association with lipid rafts in DHAPAT-knockout mice is reduced 46 , suggesting that the association of Flot1 with lipid rafts is involved in sensing of plasmalogens. If so, knocking down of FLOT1 would abrogate the sensing system for plasmalogens, stabilize Far1, and elevate plasmalogen synthesis. However, our data showed the converse that knockdown of FLOT1 reduced the expression of Far1 and plasmalogen synthesis, suggesting that another unidentified protein, if any, monitors the cellular level of plasmalogens.
Flot1 is involved in signaling pathway, cell adhesion, and membrane trafficking 47 . Our findings that the inhibition of dynamin or Src-kinase reduce the expression of Far1 and knocking down of FLOT1 elevates the plasmalogen level in DRMs (Fig. 7D) enriched in cholesterol and SM, both most abundant in the plasma membrane 48 , suggest that inhibition of the Flot1-mediated endocytic pathway stimulates the degradation of Far1 by increasing the plasmalogen level in the plasma membrane. Based on these findings, we propose that newly synthesized plasmalogens are transported from the ER to the plasma membrane and dynamically endocytosed, rather than being statically localised in plasma membrane. Accordingly, Flot1-mediated endocytosis plays an important role in the plasmalogen homeostasis, while caveolae is less likely involved in the plasmalogen-dependent degradation of Far1 (Fig. 6). Noteworthily, caveolae is not found in neuron 49,50 and lymphocytes 51 . In contrast, Flot1 is expressed in several cell types including neuron 52,53 and lymphocytes 53,54 . Together, our findings suggest that Flot1-mediaed endocytosis plays a crucial role in the plasmalogen homeostasis in most tissues. At the next steps toward elucidation of the flotillin-mediated endocytic pathway in plasmalogen homeostasis, flotillin-2 (Flot2) may need to be verified since Flot2 that is 47% identical to Flot1 in the primary structure 55 functions together with Flot1 to generate flotillin microdomains by binding to plasma membrane inner leaflet via SPFH (stomatin, prohibitin, flotillin, Hflk/C) domain and acylation of the cysteine residues in its N-terminal hydrophobic stretch 56 .
The physiological consequence of the enrichment of plasmalogens in lipid rafts has not been defined. Plasmalogens do not seem to be a structurally essential constituent for the formation of lipid rafts 2 . The interference with cholesterol-mediated assembly and function of the membrane by nystatin significantly reduces degradation of Far1 (Fig. 7E). Nystatin does not cross the bilayer and enter the cells 57 , supporting a model for sensing of plasmalogens in the plasma membrane. Nystatin disrupts caveolae morphology and inhibits the function of caveolae without altering clathrin-mediated endocytosis 38 . However, the elevated level of plasmalogens was sensed, followed by stimulating degradation of Far1 in caveolae-lacking HepG2 cells (Fig. 6). Therefore, plasmalogens that reside within cholesterol-rich domain of the plasma membrane such as non-caveolae rafts seem to be important for the sensing of plasmalogens. Clearly, more precise manipulation of membrane property or lipid constituent in non-caveolae rafts and identification of the molecule sensing plasmalogens are required for full understanding of the sensing step(s) of plasmalogens. Autoradio.
-acyl-GPE  Plasmalogens are located in the inner leaflet of plasma membrane in red blood cells and myelin, which is shown by using spin-labelled plasmalogens in red blood cells 40 or by X-ray with mercuric chloride in myelin 41 , respectively. By a combination of TNBS modification and TCA treatment of the extracted lipids, one-step analysis was made feasible in the detection of TNBS-PE, TNBS-plasmalogens, unmodified PE, and plasmalogens (Fig. 8C). In the analysis using HeLa cells, 5.2 ± 0.2% of total Etn-containing phospholipids corresponding to plasmalogens plus PE were modified with TNBS, although the level of TNBS-modified Etn-phospholipids was slightly lower than that obtained from CHO-K1 in G1 phase 44 . Therefore, TNBS-modification assay is also useful and a simple procedure for the detection of outer leaflet-localised plasmalogens. In HeLa cells, about 4% of total plasmalogens were modified by TNBS, less than that of PE (Fig. 8C, lane 4), implying that asymmetric distribution of PE and plasmalogens, both types of phospholipids harboring Etn in their head group, is distinctly regulated. To date, fourteen P4-ATPases are identified in humans and the importance of ATP8A1-mediated PE translocation is assessed in cell migration 58 . We show here that knockdown of CDC50A significantly elevates localisation of plasmalogens at cell-surface, giving rise to the increase in plasmalogen synthesis by reducing plasmalogen-induced degradation of Far1, hence implying that P4-ATPase-mediated asymmetric distribution of plasmalogens is important for sensing the plasmalogens. We should await identification of P4-ATPase(s) responsible for the topogenesis of plasmalogens and delineation of the physiological roles of the P4-ATPase(s) in cellular functions.
In summary, plasmalogen biosynthesis is regulated by modulating the stability of Far1 by sensing the plasmalogens localised in the inner leaflet of plasma membrane, where several mechanisms involving Flot1-mediated endocytosis, P4-ATPase-dependent asymmetric distribution of plasmalogens, and recognition of plasmalogens by yet unidentified proteins, if any, coordinately contribute. Subsequently, a signal emerged from sensing the plasmalogen level is spatially transferred to peroxisomes, where the Far1 stability is modulated. Therefore, plasmalogen synthesis is spatiotemporally regulated via several machineries locating in the plasma membrane and peroxisomes.

Materials and Methods
Cell culture. HeLa 4 and HepG2 (a gift from Dr. H. Sumimoto. Kyushu Univ.) were cultured in DMEM supplemented with 10% FBS (Biowest) in 5% CO 2 and 95% air 4 . Chinese hamster ovary (CHO)-K1and adaps ZPEG251 2 were cultured in Ham's F-12 medium supplemented with 10% FBS in 5% CO 2 and 95% air 2 . Cells were cultured in the presence of several types of inhibitors as follows: CHO-K1 cells were cultured in the presence of 80 μ M Dynasore (Sigma) 23  Floatation. Cell homogenate (250 μ l) containing 200 μ g protein was treated with 1% Triton X-100 on ice for 30 min and was adjusted to 40% iodixanol with 500 μ l of OptiPrep (Invitrogen) containing 60% of iodixanol. The sample (700 μ l) in TLS55 centrifuge tubes (Beckman) was overlaid with 1.2 ml of 30% iodixanol/TNE (150 mM NaCl/2 mM EDTA in 50 mM Tris⋅ HCl, pH 7.4) and 0.1 ml of TNE, and centrifuged at 55,000 rpm (259,000 × g) for 2 h at 4 °C. Two 1-ml fractions were collected from the top. Two fractions each containing the detergent-resistant (R) and detergent-soluble (S) materials was collected. Total lipids were extracted from 700 μ l of each fraction and analysed for the amount of plasmalogens and SM by liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) 59 .
Lipid analysis. Cells were cultured in the presence of 0.1 μ Ci/ml of 14 C-Etn (Moravek). Equal aliquots (100 μ g protein) of cell lysates were subjected to lipid extraction by the Bligh and Dyer method 60 . For the detection of plasmalogens, cell lysates were treated with TCA to hydrolyse vinyl-ether bond of plasmalogens prior to lipid extraction, which generates 2-acyl-GPE 2 . Lipids were analysed on TLC plates (silica gel 60, Merck) with chloroform/methanol/acetic acid solution (v/v/v: 65/25/10) 2 or hexane/diethyl ether/acetic acid solution (v/v/v: 80/20/1.5) 16 . Total cellular plasmalogens were analysed by LC-ESI-MS/MS 59 . Distribution of plasmalogens in the outer leaflet of plasma membranes was assessed by TNBS (Wako) modification 44 . In brief, cells were metabolically labelled with 14 C-Etn for 18 h, washed with ice-cold SHT buffer (0.25 M sucrose, 10 mM Hepes-KOH pH 8.5) containing 1 μ g/ml taxol (Sigma), incubated on ice for 5 min, and further treated with 10 mM TNBS dissolved in SHT buffer on ice for 30 min in the dark. Excess TNBS was quenched for 15 min with 50 mM Tris-HCl pH 8.0. Lipids were extracted and analysed on TLC as described above. 14 C-labelled lipids were detected by autoradiography using a FLA-5000 imaging analyser or Typhoon TM FLA 9500 biomolecular imager and quantified using an image analyser software (Multi Gauge, Fuji Film).