Syntaxin 17 Recruits ACSL3 to Lipid Microdomains in Lipid Droplet Biogenesis

Abstract

During lipid droplet (LD) formation, several key enzymes for neutral lipid biosynthesis, such as acyl-CoA synthetase 3 (ACSL3), translocate from the bilayer of the endoplasmic reticulum membrane or mitochondria-associated membrane to the monolayer surface of LDs. It has been recently shown that syntaxin 17 (Stx17) in cooperation with synaptosomal-associated protein of 23 kDa (SNAP23) facilitates the translocation of ACSL3 from the endoplasmic reticulum/mitochondria-associated membrane to LDs. In this study, we investigated whether lipid microdomains enriched in cholesterol and sphingolipids are important for the formation of LDs and the interaction of Stx17 with ACSL3 and SNAP23. Cholesterol depletion and blockage of ceramide synthesis by chemicals inhibited oleic acid (OA)-induced LD biogenesis and decreased the interaction of Stx17 with ACSL3 and SNAP23, whereas blockage of ganglioside GD3 synthesis by sialyltransferase knockdown interfered with LD biogenesis by affecting the interaction of Stx17 with SNAP23 but not ACSL3. Consistent with the requirement of GD3 in LD biogenesis, Stx17 was found to associate with GD3-containing membranes upon OA loading. SNAP23 and a minor fraction of Stx17 were found to reside in detergent-resistant membranes (DRMs), whereas OA treatment caused redistribution of ACSL3 and Stx17 to DRMs. Importantly, the redistribution of ACSL3 to DRMs was abrogated upon depletion of Stx17 or SNAP23. Taken together, our results highlight the importance of lipid microdomains enriched in cholesterol and sphingolipids as a platform for the interaction of Stx17 with ACSL3 and SNAP23 in LD biogenesis.

Introduction

Lipid droplets (LDs) are intracellular organelles that contain neutral lipids such as triacylglycerol and sterol esters, and their number and size continually change in response to energy status (Walther & Farese, 2012). When lipids, such as fatty acids or sterols, are excess over cell’s energy requirement, they are converted to neutral lipids and stored in nascent or expanding LDs. On the other hand, when cells need energy or membrane components, neutral lipids stored in LDs are degraded. LDs are unique among the cellular membrane compartments in that they are surrounded by a phospholipid monolayer. Depending on the maturation status and cell types, the protein composition of LDs varies and changes (Bersuker et al., 2018; Ohsaki, Suzuki, & Fujimoto, 2014; Pol, Gross, & Parton, 2014; Thiam & Beller, 2017). Recent studies have revealed that the redistribution of lipid-synthesizing enzymes such as acyl-CoA synthetase (ACSL) 3, an enzyme that converts fatty acids to the active species acyl-CoA, from the endoplasmic reticulum (ER) and others to LDs is critical for LD maturation (Kassan et al., 2013; Krahmer et al., 2011; Stone et al., 2009; Wilfling et al., 2013).

Autophagy-related SNARE protein, syntaxin 17 (Stx17), localizes to the ER, mitochondria-associated membrane (MAM), and mitochondria (Arasaki et al., 2015; Hamasaki et al., 2013; Hung et al., 2014; Itakura, Kishi-Itakura, & Mizushima, 2012; Steegmaier, Oorschot, Klumperman, & Scheller, 2000). Previously, Arasaki et al. (2015) reported that Stx17 is distributed in cholesterol-enriched and nonenriched microdomains of the MAM and that the distribution of Stx17 is closely related to its function: Binding of Stx17 to Drp1 for mitochondrial division and to ATG14L for autophagosome formation perhaps occurs on cholesterol-enriched and nonenriched microdomains, respectively. Recently, Kimura et al. (2018) revealed that Stx17 promotes LD maturation by facilitating the translocation of ACSL3 from the ER to the surface of LDs. For this role, both the SNARE motif and the C-terminal long hydrophobic and cytosolic regions of Stx17 are required. The SNARE motif is required for the binding to ACSL3 and the interaction with synaptosomal-associated protein of 23 kDa (SNAP23), a SNARE protein implicated in LD formation (Boström et al., 2007; Jägerström et al., 2009). Kimura et al. (2018) have put forward a model in which ACSL3 initially interacts with Stx17, and this interaction is later disrupted by SNAP23 for ACSL3 translocation to LDs.

Many enzymes responsible for the synthesis of neutral lipids as well as phospholipids localize to the MAM (Brasaemle & Wolins, 2012; Schon & Area-Gomez, 2013; Vance, 2014). As the MAM exhibits properties similar to those of lipid rafts (Area-Gomez et al., 2012; Hayashi & Fujimoto, 2010), we tested whether cholesterol/sphingolipid-enriched microdomains are important for LD biogenesis. Here, we show that such lipid microdomains are important for the interaction of Stx17 with ACSL3 and SNAP23 as well as LD expansion.

Results

Sequestering of Cholesterol Inhibits LD Formation

Lipid rafts are microdomains that are rich in cholesterol and sphingolipids including gangliosides and function in membrane signaling and trafficking (Lingwood & Simons, 2010). We examined whether the treatment of cells with drugs that disrupt cholesterol distribution in cells affects LD formation and the interaction of Stx17 with ACSL3 and SNAP23. First, we used methyl-β-cyclodextrin (MβCD) and nystatin, both of which are known as reagents capable of sequestering cholesterol and thereby disrupting lipid microdomains (Hayashi & Fujimoto, 2010; Ohtani, Irie, Uekama, Fukunaga, & Pitha, 1989; Rothberg et al., 1992). HeLa cells were incubated in the absence of fetal calf serum (FCS) for 24 hr to remove preformed LDs and then pretreated with 20 mM MβCD for 1 hr or 10 μg/ml nystatin for 20 min, followed by incubation with 150 μM oleic acid (OA) to start LD biogenesis. Upon treatment with these reagents, the cells were found to have smaller LDs with a slightly decreased number of LDs after OA loading (Figure 1(a) and (b)). In addition, proximity ligation assay (PLA; Figure 1(c)) and immunoprecipitation (Figure 1(d)) demonstrated that disruption of lipid microdomains with these reagents reduces the interaction of Stx17 with ACSL3 and SNAP23. Upon treatment with these reagents, Stx17 was redistributed from a mitochondria-like pattern to a slightly diffused pattern, as revealed by a decrease in Manders’ colocalization coefficient between Stx17 and Tom20 (Figure 1(e) and (f)). It should be noted that the uptake of BODIPY FL-C16, a fluorescent fatty acid that is efficiently incorporated into the triacylglycerol pool (Somwar, Roberts, & Varlamov, 2011), was not inhibited by MβCD or nystatin ( Supplementary Figure 1 ). Consistent with the finding that MβCD and nystatin blocked LD expansion (Figure 1(a) and (b)), BODIPY FL-C16 exhibited a diffuse pattern in cells treated with these reagents, in contrast to vehicle-treated cells showing a punctate pattern ( Supplementary Figure 1a ). These results suggest that the block of LD expansion by these reagents is not due to their inhibition on fatty acid uptake.

Figure 1.

Depletion of cholesterol inhibits LD formation. (a) HeLa cells were incubated without FCS for 24 hr (FCS−) to remove preformed LDs and then pretreated with 20 mM MβCD for 1 hr or 10 μg/ml nystatin for 20 min. After being washed, the cells were incubated with medium containing FCS and 150 µM OA for 8 hr, fixed, and then stained for LDs with LipidTox. Bar = 5 μm. (b) The number and relative size of LDs were determined by quantifying the data in panel (a). Values are means ± standard error of the mean (SEM; n (number of experiments) = 3). *p ≤ .05; ***p ≤ .001. (c) The fixed cells were prepared as described in panel (a) and subjected to PLA using antibodies against Stx17 and ACSL3 or SNAP23. PLA-positive dots were counted. Values are means ± SEM (n = 3). *p ≤ .05; **p ≤ .01. (d) HeLa cells stably expressing FLAG-Stx17 wild type (WT) were treated as described in panel (a), and cell lysates were immunoprecipitated using anti-FLAG M2 beads; 5% input and the precipitated proteins were analyzed using the indicated antibodies. (e) HeLa cells were treated with 20 mM MβCD for 1 hr or 10 μg/ml nystatin for 20 min, fixed, and double immunostained for Stx17 and Tom20. Bar = 5 μm. (f) The fraction of Stx17 overlapping with Tom20. Manders’ colocalization coefficients were determined by quantifying the data in panel (e). Values are means ± SEM (n = 3). ***p ≤ .001. LD = lipid droplet; PLA = proximity ligation assay; MβCD = methyl-β-cyclodextrin; FCS = fetal calf serum; ACSL3 = acyl-CoA synthetase 3; Stx17 = syntaxin 17; SNAP23 = synaptosomal-associated protein of 23 kDa.

GD3 Ganglioside Is Important for LD Formation

We next used fumonisin B₁ (FB1), an inhibitor of ceramide synthase and de novo sphingolipid biosynthesis (Wang, Norred, Bacon, Riley, & Merrill, 1991). FB1 treatment also inhibited the expansion of LDs (Figure 2(a)) and decreased the interaction of Stx17 with ACSL3 and SNAP23 (Figure 2(b)), as seen in the case of the treatment with MβCD and nystatin.

Figure 2.

Sphingolipids are important for LD formation. (a) HeLa cells were incubated without FCS for 24 hr (II and III) and then treated without or with 150 µM OA for 8 hr in the absence (II) or presence (III) of 10 µM FB1, fixed, and stained for LDs using LipidTox. Bar = 5 µm. For comparison, HeLa cells were cultured in FCS-containing medium and then incubated without or with OA (I). (b) The cells prepared as described in panel (a) were subjected to PLA using antibodies against Stx17 and ACSL3 or SNAP23. PLA-positive dots were counted. Values are means ± standard error of the mean (n = 3). *p ≤ .05; ***p ≤ .001. ACSL3 = acyl-CoA synthetase 3; Stx17 = syntaxin 17; PLA = proximity ligation assay; FCS = fetal calf serum; OA = oleic acid; FB1 = fumonisin B₁.

To further substantiate the requirement of cholesterol/sphingolipid-enriched microdomains for LD biogenesis, we knocked down α-N-acetylneuraminide α-2,8-sialyltransferase (ST8SIA1), the enzyme that catalyzes the transfer of sialic acid from CMP-sialic acid to GM3 to produce GD3. As two ST8SIA1 isoforms (transcript variants) are expressed in HeLa cells, we used relevant siRNAs for knocking down of ST8SIA1 ( Supplementary Figure 2a ). Knockdown of both isoforms (ST8SIA1 KD) using three siRNAs (one is common for the two isoforms and the other two are isoform-specific) caused a marked decrease in GD3 staining in cells ( Supplementary Figure 2b ). Concomitantly, LD expansion was blocked without a significant decrease in the number of LDs (Figure 3(a) and (b)). Under this condition, the interaction between Stx17 and SNAP23 was disrupted, whereas the interaction between Stx17 and ACSL3 was barely affected (Figure 3(c)).

Figure 3.

GD3 is important for LD formation. (a) HeLa cells were mock-transfected or transfected with siRNAs targeting ST8SIA1 isoforms 1 and 2 (ST8SIA1 KD). At 56 hr after transfection, OA was added at a final concentration of 150 µM, and the cells were incubated for 16 hr and fixed. The cells were stained with an anti-ACSL3 antibody and LipidTox. Bar = 5 µm. (b) The number and relative size of LDs were determined by quantifying the data in panel (a) Values are means ± standard error of the mean (SEM; n = 3). ***p ≤ .001; NS = not significant. (c) The cells treated and fixed as described in panel (a) were subjected to PLA using antibodies against Stx17 and ACSL3 or SNAP23. PLA-positive dots were counted. Values are means ± SEM (n = 3). ***p ≤ .001; NS = not significant. (d) FLAG-Stx17 WT or the K254C mutant stably expressing HeLa cells were incubated with or without 150 µM OA for 16 hr, and cell lysates were prepared without addition of detergent and immunoprecipitated using anti-FLAG M2 beads. The precipitated proteins were analyzed by Western and dot blotting using antibodies against FLAG and GD3, respectively. The bar graph shows the relative intensity of GD3 spots. Values are means ± standard deviation (n = 3). *p ≤ .05. (e) FLAG-Stx17 WT stably expressing HeLa cells were incubated as described in the legend to Figure 2a, and cell lysates were prepared, immunoprecipitated, and analyzed as described in panel (d). The bar graph shows the relative intensity of GD3 spots. Values are means ± standard deviation (n = 3). *p ≤ .05. (f) HeLa cells were incubated with or without 150 µM OA for 16 hr, fixed, and then double immunostained for Stx17 and GD3. Bar = 5 µm. (g) The fraction of Stx17 overlapping with GD3. Manders’ colocalization coefficients were determined by quantifying the data in panel (f). Values are means ± SEM (n = 3). ***p ≤ .001. (h) The fixed cells were prepared as described in the legend to Figure 1(a) and double immunostained for Stx17 and GD3. Bar = 5 μm. (i) The fraction of Stx17 overlapping with GD3. Manders’ colocalization coefficients were determined by quantifying the data in panel (h). Values are means ± SEM (n = 3). **p ≤ .005. LD = lipid droplet; PLA = proximity ligation assay; MβCD = methyl-β-cyclodextrinp; FCS = fetal calf serum; ACSL3 = acyl-CoA synthetase 3; Stx17 = syntaxin 17; SNAP23 = synaptosomal-associated protein of 23 kDa; OA = oleic acid; FB1 = fumonisin B₁; WT = wild type; KC = syntaxin 17 K254C mutant; KD = knockdown; ST8SIA1 = α-N-acetylneuraminide α-2,8-sialyltransferase.

Previous studies showed that membranes containing GD3 can be coprecipitated with MAM proteins such as calnexin (CNX; Garofalo et al., 2016; Matarrese et al., 2014). We therefore examined the association of Stx17 with GD3-containing membranes by immunoprecipitation using detergent free buffer and found that Stx17 association with GD3-containing membranes is enhanced upon OA treatment (Figure 3(d)). This association is specific because inhibition of ceramide formation and thereby GD3 synthesis by FB1 treatment abrogated this association (Figure 3(e), III). Moreover, the Stx17 K254C mutant, which cannot localize to the MAM (Arasaki et al., 2015) nor facilitate LD formation (Kimura et al., 2018), was found not to bind to GD3 (Figure 3(d)).

To gain insight into the mechanism underlying the OA-induced association of Stx17 with GD3-containing membranes, we next examined the localization of Stx17 and GD3. In the absence of OA, Stx17 was found to be only partially colocalized with GD3, whereas OA treatment significantly increased the area of GD3-containing membranes, resulting in the enhanced colocalization between Stx17 and GD3 (Figure 3(f) and (g)). Of note, this colocalization was prevented upon sequestering of cholesterol by treatment with MβCD or nystatin (Figure 3(h) and (i)), suggesting that Stx17 localizes to GD3-positive microdomains upon OA treatment.

Mitofusin 2-Dependent MAM–Mitochondria Connection Is Not Important for LD Formation

Under fed conditions, Stx17 promotes mitochondrial division by interacting with the mitochondrial fission factor Drp1 in a MAP1B-LC1-dependent manner, whereas starvation causes the dephosphorylation of MAP1B-LC1 at Thr217, allowing Stx17 to dissociate from Drp1 and associate with Atg14L for autophagosome formation ( Arasaki et al., 2015, 2018; Hamasaki et al., 2013). The interaction of Stx17 with Drp1 and Atg14L needs appropriate tethering between the MAM and mitochondria (Arasaki et al., 2015; Hamasaki et al., 2013). We therefore examined whether the appropriate MAM–mitochondria tethering is required for LD formation by knocking down mitofusin 2 (Mfn2). Mfn2 mediates a functional link between the MAM and mitochondria (Arasaki et al., 2015; Hailey et al., 2010; Hamasaki et al., 2013) by promoting (Naon et al., 2016) or inhibiting (Filadi et al., 2015) tethering between the MAM and mitochondria. Knockdown of Mfn2 neither affected LD formation ( Supplementary Figure 3a , bottom row) nor the proximity between Stx17 and ACSL3 ( Supplementary Figure 3b ). On the other hand, knockdown of PACS-2, a protein that maintains MAM integrity (Simmen et al., 2005), inhibited LD formation ( Supplementary Figure 3a , middle row) and abrogated the proximity between Stx17 and ACSL3 ( Supplementary Figure 3b ). These results suggest that the MAM, but not Mfn2-dependent connection of the MAM with mitochondria, is important for LD formation.

As digitonin has a high affinity for cholesterol, this reagent at low concentrations (e.g., 0.03 mg/ml) well below its critical micelle concentration (0.8–0.9 mg/ml) specifically permeabilizes the cholesterol-rich plasma membrane by extracting cholesterol (Oliferenko et al., 1999). As the MAM exhibits properties similar to those of lipid rafts (Area-Gomez et al., 2012; Hayashi & Fujimoto, 2010), it is also sensitive to digitonin. A previous study demonstrated that the localization of Stx17, in particular, moderately overexpressed Stx17, is sensitive to a low concentration of digitonin: Treatment with 0.03 mg/ml digitonin causes the redistribution of Stx17 from a mitochondria-like pattern to a moderately diffuse pattern (Arasaki et al., 2015). This diffusion can be quantified as a decrease in the overall staining intensity of Stx17 in cells ( Supplementary Figure 3c , top row and 3d). When Mfn2 was depleted, the staining intensity of FLAG-Stx17 was decreased ( Supplementary Figure 3c , bottom row, and 3d), as in the case of mock-treated cells. On the other hand, PACS-2 knockdown inhibited the reduction in the intensity of FLAG-Stx17 staining ( Supplementary Figure 3c, middle row, and 3d ), suggesting that some FLAG-Stx17 in PACS-2-depleted cells is localized in digitonin-insensitive structures.

Localization of Stx17 to a Digitonin-Sensitive Structure Is Regulated by SNAP23

Previous studies demonstrated that SNAP23 may be a critical factor for LD biogenesis (Boström et al., 2007; Jägerström et al., 2009; Kimura et al., 2018). Similar to the case of Stx17, SNAP23 was also found in the MAM fraction (Figure 4(a)) on Percoll-based fractionation (Wieckowski, Giorgi, Lebiedzinska, Duszynski, & Pinton, 2009). Of note is that ACSL3 was also fractionated into the MAM fraction as well as the microsome fraction (Figure 4(a)). Knockdown of SNAP23 by siRNA SNAP23 (837) or (534; Figure 4(b)) markedly inhibited triacylglycerol synthesis (Figure 4(c)) and reduced the size of LDs with a disrupted distribution of ACSL3 on LDs (Figure 4(d) and (e)), as seen in Stx17-depleted cells (Kimura et al., 2018).

Figure 4.

SNAP23 is required for LD biogenesis and regulates Stx17 localization. (a) HeLa cells were incubated with 150 µM OA for 16 hr and then subjected to MAM fractionation analysis. PNS = postnuclear supernatant; MAM = mitochondria-associated membrane; MS = microsomes; Mt = mitochondria. Equal amounts of proteins were analyzed by Western blotting using the indicated antibodies. The amounts of proteins recovered on fractionation were as follows: PNS (5.0 mg), cytosol (5.2 mg), MS (2.2 mg), MAM (0.53 mg), and Mt (0.33 mg). (b) HeLa cells were mock-transfected or transfected with siRNA SNAP23 (837) or (534), and protein levels were determined with antibodies against SNAP23 and CNX. (c) Alternatively, the amount of triacylglycerol was determined after incubation with OA for the indicated times. Values are means ± standard error of the mean (SEM; n = 3). *p ≤ .05; **p ≤ .01; ***p ≤ .001. (d) Mock-treated cells and cells depleted of SNAP23 using SNAP23 (837) or SNAP23 (534) were treated with 150 µM OA for 16 hr and then stained with an anti-ACSL3 antibody and LipidTox. Bar = 5 μm. (e) Quantification of the data in panel (d). The bar graph shows the relative ACSL3 staining intensity surrounding LDs. Values are means ± SEM (n = 3). *p ≤ 0.05. (f) FLAG-Stx17 WT stably expressing HeLa cells were mock-transfected or transfected with siRNA SNAP23 (837). After 72 hr, the cells were fixed and immunostained for FLAG and Tom20 or Sec61β. Bar = 5 µm. (g) The fraction of FLAG-Stx17 overlapping with Tom20 or Sec61β. Manders’ colocalization coefficients were determined by quantifying the data in panel (f). Values are means ± SEM (n = 3). **p ≤ .01; NS = not significant. (h) FLAG-Stx17 WT stably expressing HeLa cells were mock-treated or transfected with siRNA SNAP23 (837). After 72 hr, the cells were incubated with 0.03 mg/ml digitonin for 5 min at room temperature, fixed, and then immunostained for FLAG. Bar = 5 µm. (i) Quantification of the data in panel (h). The bar graph shows the relative staining intensity of FLAG-Stx17 WT per whole cell. Values are means ± SEM (n = 3). *p ≤ .05. ACSL3 = acyl-CoA synthetase 3; Stx17 = syntaxin 17; SNAP23 = synaptosomal-associated protein of 23 kDa; OA = oleic acid; TAG = triacylglycerol; CNX = calnexin; Stx3 = syntaxin 3; KD = knockdown; WT = wild type.

We examined whether SNAP23 regulates the localization of FLAG-Stx17. When SNAP23 was knocked down, the distribution of Stx17 was changed from a mitochondria-like pattern to a more diffuse pattern (Figure 4(f) and (g)). Mitochondrial localization may also have been altered. We then assessed the digitonin-sensitive localization of FLAG-Stx17. Digitonin treatment decreased the staining intensity of FLAG-Stx17 in mock-treated cells, whereas SNAP23 knockdown inhibited the reduction in the intensity of FLAG-Stx17 staining (Figure 4(h) and (i)), suggesting that FLAG-Stx17 was redistributed to digitonin-insensitive domains upon SNAP23 depletion.

Stx17 and SNAP23 Are Required for ACSL3 Association With Detergent-Resistant Membranes During LD Formation

Cholesterol/sphingolipid-enriched microdomains are resistant to nonionic detergents (Lingwood & Simons, 2010; Pike, 2003; Sezgin, Levental, Mayor, & Eggeling, 2017). To substantiate the presence of Stx17 in cholesterol/sphingolipid-enriched microdomains, we performed detergent-resistant membrane (DRM) isolation using Triton X-114 (Hayashi & Fujimoto, 2010). In cells without OA loading (Figure 5(a)), a large portion of SNAP23 was detected in the DRM fraction (fractions 2–5), whereas most ACSL3 was present in the non-DRM fraction (fractions 9–12). A minor fraction of Stx17 was observed in the DRM fraction, and most Stx17 was present in the non-DRM fraction.

Figure 5.

Redistribution of ACSL3 to DRMs upon OA treatment. (a and b) 293T cells were incubated with or without 10 µg/ml nystatin for 20 min and then without (a) or with 150 µM OA for 16 hr (b). The cells were solubilized in buffer containing 0.5% Triton X-114 and subjected to sucrose density centrifugation. Each fraction was analyzed by Western blotting using the indicated antibodies. (c) 293T cells transiently expressing FLAG-Stx17 WT or the K254C mutant were incubated with 150 µM OA for 16 hr and subjected to DRM isolation. Each fraction was analyzed by Western and dot blotting using the indicated antibodies. Of note, GD3 was detected in the DRM fraction. (d) 293T cells were mock-transfected or transfected with siRNA targeting Stx17 or SNAP23. At 56 hr after transfection, OA was added at a final concentration of 150 µM, and the cells were incubated for 16 hr and subjected to DRM isolation followed by Western blot analysis. ACSL3 = acyl-CoA synthetase 3; Stx17 = syntaxin 17; SNAP23 = synaptosomal-associated protein of 23 kDa; OA = oleic acid; KD = knockdown; WT = wild type; DRM = detergent-resistant membrane.

When OA was loaded, ACSL3 as well as Stx17 was redistributed to the DRM fraction (Figure 5(b), upper panel). Tip47/PLIN3, a protein that translocates from the cytosol to LDs, was not recovered in the DRM fraction regardless of whether OA was present (Figure 5(b), upper panel) or not (Figure 5(a)). Nystatin treatment markedly disrupted the association of all three proteins (ACSL3, Stx17, and SNAP23) with DRMs (Figure 5(b), lower panel). These results suggest that ACSL3 and Stx17 redistribute to cholesterol-enriched microdomains upon OA treatment. It should be noted that upon OA treatment the Stx17 K254C mutant, which was found not to be associated with GD3-containing membranes (Figure 3(d)), was neither recovered in the DRM fraction nor induced the redistribution of ACSL3 to the DRM fraction (Figure 5(c), middle). Nevertheless, GD3 association with the DRM was not affected by the expression of the Stx17 K254C mutant (Figure 5(c), bottom).

We next examined whether the depletion of Stx17 or SNAP23 affects the redistribution of ACSL3 to DRMs. When DRM isolation was conducted using cells with OA treatment, depletion of Stx17 or SNAP23 inhibited the association of ACSL3 with DRMs (Figure 5(d), middle and bottom), suggesting that these proteins are required for ACSL3 redistribution to cholesterol/sphingolipid-enriched microdomains.

Part of ACSL3 Surrounding LDs Is Sensitive to Digitonin

Finally, we tested whether ACSL3 around the surface of LDs is sensitive to digitonin. When cells treated with OA for 16 hr were incubated with 0.03 mg/ml for 5 min, ACSL3 staining was abolished in some area of the LD surface, yielding a crescent-like distribution (Figure 6(a) and (b)), reminiscent of that in Stx17-depleted cells (Arasaki et al., 2015). This sensitivity was not seen for a smooth ER marker reticulon 4 (RTN4) that is connected or adjacent to LDs (Figure 6(c)).

Figure 6.

The part of the LD surface and the ER adjacent to LDs are digitonin-sensitive. (a) HeLa cells were incubated with 150 µM OA for 16 hr, treated without (Dig−) or with 0.03 mg/ml digitonin (Dig+) for 5 min, fixed, and stained with an anti-ACSL3 antibody and LipidTox. Bar = 1 µm. (b) The ACSL3 distribution around LDs was classified into three patterns (left), and the data in panel (a) were quantified (right). Values are means ± standard error of the mean (n = 3). **p ≤ .01; NS = not significant. Of note is that the number and size of LDs were not affected upon digitonin treatment (data not shown). (c) HeLa cells were treated as described in panel (a) and stained with LipidTox and an antibody against ACSL3 or RTN4, and Manders’ colocalization coefficients were determined. Values are means ± standard error of the mean (n = 3). **p ≤ .01; NS = not significant. (d) OA-dependent redistribution of Stx17, SNAP23, and ACSL3. K and GATE denote Lys254 in Stx17 and the GATE domain of ACSL3 that interacts with the SNARE motif of Stx17, respectively. CS stands for cholesterol and sphingolipids. ACSL3 = acyl-CoA synthetase 3; Stx17 = syntaxin 17; SNAP23 = synaptosomal-associated protein of 23 kDa; OA = oleic acid; LD = lipid droplet; RTN4 = reticulon 4; SNARE = soluble N-ethylmaleimide-sensitive factor attachment protein receptor.

Discussion

At the onset of LD formation, several enzymes for neutral lipid and phospholipid synthesis translocate from the ER or the cytosol to the surface of LDs (Kassan et al., 2013; Krahmer et al., 2011; Stone et al., 2009; Wilfling et al., 2013). Proteins that translocate from the ER to LDs often have hydrophobic domains with hairpin-like structure (Class I), whereas proteins such as perlipins that move from the cytosol have amphipathic α helices or other hydrophobic domains (Class II; Kory, Farese, & Walther, 2016). ACSL3 contains an N-terminal hydrophobic domain responsible for the targeting to the LD surface (Poppelreuther et al., 2012) and is one of the proteins that translocate from the ER to emerging LDs at the earliest stage of LD formation (Kassan et al., 2013). Although several ACSL isoforms are expressed in cells, only ACSL3 mediates LD expansion, likely by supplying acyl-CoA at the on-site LD formation (Fujimoto et al., 2007; Kassan et al., 2013, Kimura et al., 2018). Kimura et al. (2018) recently showed that Stx17 promotes LD formation by facilitating the distribution of ACSL3 from the ER to nascent LDs.

In this study, we examined whether cholesterol/sphingolipid-enriched microdomains are important for LD expansion and the interaction between Stx17 and ACSL3. We used several approaches: cholesterol depletion by MβCD and nystatin, inhibition of sphingolipids synthesis by FB1, depletion of GD3 by ST8SIA1 knockdown, and association of Stx17 with DRMs including GD3-containing membranes. All data obtained emphasized the importance of cholesterol/sphingolipid-enriched microdomains for LD expansion and the interaction between Stx17 and ACSL3, although we could not completely exclude the possibility that some of the observed effects could result from perturbations in more cholesterol- and sphingolipid-enriched membranes.

The present observations combined with the results in a study by Kimura et al. (2018) may provide a clue of how Stx17 facilitates ACSL3 translocation from the ER membrane to the LD surface. SNAP23 and a minor fraction of Stx17 are localized in cholesterol/sphingolipid-enriched microdomains even in the absence of OA, whereas ACSL3 in the absence of OA is almost exclusively localized in the ER membrane deficient for cholesterol and sphingolipids (Figure 6(d), upper panel). Upon OA loading, Stx17 redistributes from a cholesterol/sphingolipid-deficient microdomain to a cholesterol/sphingolipid-enriched microdomain and assists the movement of ACSL3 from a cholesterol/sphingolipid-deficient microdomain to a cholesterol/sphingolipid-enriched microdomain (Figure 6(d), lower panel). Without Stx17 or SNAP23, ACSL3 may not translocate to or pass through cholesterol/sphingolipid-enriched microdomains on the ER and the surface of LDs. Of note is that a part of the LD surface and the ER region connected to LDs are cholesterol/sphingolipid-enriched microdomains and sensitive to digitonin (Figure 6(a)). This idea can explain the previous observation that ACSL3 exhibits crescent-like distribution around LDs in Stx17-depleted cells (Kimura et al., 2018). Cholesterol/sphingolipid-enriched microdomains may be a barrier for ACSL3 to translocate from the ER and surround the surface of LDs. This idea can also explain that, in Stx17-depleted cells, no or little recovery of green fluorescent protein (GFP)-ACSL3 on LDs was observed after photobleaching, whereas GFP-ACSL3 fluorescence was rapidly recovered when ER-localized GFP-ACSL3 was photobleached (Kimura et al., 2018).

Stx17 has a long C-terminal hydrophobic domain (44 amino acids) that is separated by Lys254. Replacement of Lys254 by other amino acids, including Cys but not Arg, causes the redistribution of the resultant mutants from mitochondria-like localization to the entire ER region. Moreover, the K254C mutant lost its ability to bind to Drp1, Rab32, Atg14L, (Arasaki et al., 2015), and MAP1B-LC1 (Arasaki et al., 2018) and failed to reverse LD formation in Stx17-depleted cells (Kimura et al., 2018). We showed that the K254C mutant neither binds to GD3 (Figure 3(d)) nor associates with DRMs (Figure 5(c)). It is likely that the C-terminal hydrophobic domain containing Lys254 forms a hairpin or w-shaped structure suitable for embedding in cholesterol/sphingolipid-enriched microdomains. Loss of the positive charge at residue 254 may affect the conformation of the long hydrophobic domain, leading to the exclusion of Stx17 from cholesterol/sphingolipid-enriched microdomains and the loss of the ability to bind to many proteins.

Based on the observations that the association between Stx17 and ACSL3 depends on the expression level of SNAP23 and that the amount of the Stx17-ACSL3 complex initially increased and then decreased during LD biogenesis, Kimura et al. (2018) proposed a model in which ACSL3 initially interacts with Stx17, and this interaction is later disrupted by SNAP23 for ACSL3 translocation to LDs. However, the present observations call for the revision of this model. The fact that SNAP23 depletion inhibited the Stx17-mediated recruitment of ACSL3 to cholesterol/sphingolipid-enriched microdomains (Figure 5(d), bottom) favors the idea that Stx17 and SNAP23 cooperate to recruit ACSL3 to these domains. Perhaps, the association between Stx17 and ACSL3 contributes to this recruitment. The mechanism for SNAP23 that recruits ACSL3 is obscure because it does not bind to ACSL3 (data not shown). This should be investigated in future studies.

In conclusion, the present results revealed that cholesterol/sphingolipid-enriched microdomains are important for LD biogenesis. Stx17, perhaps in coordination with SNAP23, supports LD formation by recruiting ACSL3 to cholesterol/sphingolipid-enriched microdomains.

Materials and Methods

Chemicals and Antibodies

OA was obtained from Sigma-Aldrich, dissolved in DMSO, mixed with bovine serum albumin, and used. Digitonin and Triton X-114 were obtained from Wako Chemicals. LipidTox and BODIPY FL-C16 were purchased from Thermo Fisher Scientific. Polyclonal antibodies against FLAG and SNAP23 were purchased from Sigma-Aldrich (No. F7425) and Proteintech (No. 10825-1-AP), respectively. A polyclonal goat anti-RTN4 antibody (sc-11027) and a monoclonal anti-SNAP23 antibody (sc-374215) were purchased from Santa Cruz Biotechnology. The following antibodies were obtained from BD Bioscience Pharmingen: CNX (No. 610523), Tom20 (No. 612278), and Tim23 (No. 611223). The following antibodies were purchased from Abcam: α-tubulin (No. ab15246) and ST8SIA1 (No. ab140344). A monoclonal anti-GD3 antibody (No. A2580) was purchased from Tokyo Chemical Industry. An antiserum against Sec61β (No. 07-205) was purchased from Millipore Corp. Polyclonal and monoclonal antibodies against ACSL3 were obtained from GeneTeX (No. GTX112431) and Abnova (No. H00002181-B01P), respectively. Alexa Fluor 488 and 594 goat anti-mouse and anti-rabbit IgG antibodies were obtained from Thermo Fisher Scientific (No. A-11001, A-11005, A-11008, and A-11012). A fluorescein isothiocyanate (FITC)-conjugated donkey anti-goat IgG antibody (No. AP180F) was from Millipore Corp. Rabbit antibodies against Sec22b and Stx3 were produced in this laboratory and affinity purified. An antibody against Stx17 was prepared as described by Arasaki et al. (2015) and used for immunofluorescence analysis. For Western blotting, an anti-Stx17 antibody (Sigma-Aldrich; HPA001204) was used.

Cell Culture

293T cells were grown in Dulbecco's modified Eagle's medium supplemented with 50 IU/ml penicillin, 50 µg/ml streptomycin, and 10% FCS. HeLa cells (RIKEN; RCB0007) were cultured in minimum essential medium Eagle – α modification supplemented with the same materials plus 2 mM glutamine. Stable transfectants were prepared as described by Arasaki et al. (2015). For the induction of LDs in HeLa cells, OA was added at a final concentration of 150 µM.

Plasmids and Transfection

Plasmids encoding human Stx17 full length and its derivatives were described previously (Arasaki et al., 2015). Transfection was carried out using LipofectAMINE2000 (Thermo Fisher Scientific) or polyethylenimine.

RNA Interference

The following siRNAs were used:

Stx17(440):5′-GGUAGUUCUCAGAGUUUGAUU-3′

SNAP23 (837): 5′-CAUUAAACGCAUAACUAAU-3′

SNAP23(534):5′-CUCAAAAUCCACAAAUAAA-3′

ST8SIA1 isoform 1: 5′-CCAUUGACAAUUCAACUUA-3′

ST8SIA1 isoform 2: 5′-CCAUCUUUGAGGGUUUAUU-3′

ST8SIA1 common to isoforms 1 and 2: 5′-GGAUGUUGGAUCCAAAAGU-3′

Mfn2: 5′-AGAGGGCCUUCAAGCGCCA-3′

PACS-2: 5′-AACACGCCCGUGCCCAUGAAC-3′

siRNAs were purchased from Japan Bio Services. The efficient knockdown of Stx17, SNAP23, Mfn2, and PACS-2 by the above siRNAs was verified previously (Arasaki et al., 2015; Kimura et al., 2018). HeLa and 293T cells were grown on 35 mm dishes, and siRNAs were transfected at a final concentration of 200 nM using Oligofectamine (Thermo Fisher Scientific) according to the manufacturer’s protocol.

Immunoprecipitation

HeLa cells expressing FLAG-tagged proteins were lysed in lysis buffer (20 mM HEPES-KOH [pH 7.2], 150 mM KCl, 2 mM EDTA, 1 mM dithiothreitol, 1 µg/ml leupeptin, 1 µM pepstatin A, 2 µg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride) containing 1 mg/ml digitonin. After centrifugation, the supernatants were collected and immunoprecipitated with anti-FLAG M2 affinity beads (Sigma-Aldrich). The precipitated proteins were eluted with SDS sample buffer and then analyzed by Western blotting. Experiments were repeated 2 or 3 times with similar results.

Immunofluorescence Microscopy and Image Analysis

For immunofluorescence microscopy, cells were fixed with 4% paraformaldehyde for 20 min at room temperature followed by permeabilization in 0.2% Triton X-100 in phosphate-buffered saline (PBS) for 10 min at room temperature and then blocked with 2% bovine serum albumin in PBS for 10 min. The cells were incubated with a primary antibody (ACSL3 [dilution, 1:50], Stx17 [1:50], Tom20 [1:300], Sec61β [1:200], RTN4 [1:500], SNAP23 [1:50], or GD3 [1:50]) for 1 hr at 37°C, followed by three washes in PBS and incubation with an Alexa Fluor 488- or 594-conjugated anti-mouse or anti-rabbit IgG antibody (1:200) or an FITC-conjugated anti-goat IgG antibody (1:100) for 1 hr at 37°C. When stained with LipidTox, the cells were incubated with LipidTox (dissolved in DMSO) for 30 min in PBS at room temperature. After washing in PBS, they were mounted with mounting medium (Dako) and observed using a 100× oil immersion objective lens (UPlan FI: NA = 1.3) under a laser scanning confocal microscope (OLYMPUS Fluoview FV1000-D) with a pinhole of 3 AU. All images were single confocal sections. ImageJ software (National Institutes of Health) was used to determine the size of LDs and the fluorescence intensity ratio between LipidTox and LD proteins. In each LD, the intensities of circular LipidTox fluorescence and surrounding FITC fluorescence were measured. This analysis was performed for randomly chosen 30 LDs in each cell, and 30 cells were analyzed in each experiment.

Proximity Ligation Assay

PLA was conducted using a PLA kit (Sigma-Aldrich) according to the manufacturer’s protocol. Thirty cells were analyzed in each assay. PLA dots were identified using the analyze particle program in the ImageJ software. Randomly, 30 cells were selected, and the number of PLA dots was measured in each sample. The experiments were repeated 3 times.

DRM Isolation

293T cells were harvested and pelleted at 3,800 g, and DRMs were isolated as previously described (Hayashi & Fujimoto, 2010). Cells were extracted for 30 min in 900 µl of buffer (10 mM Tris-HCl [pH 7.4], 150 mM NaCl, 5 mM EDTA, 1 µg/ml leupeptin, 1 µM pepstatin A, and 2 µg/ml aprotinin) containing 0.5% Triton X-114. Triton X-114 extracts (0.9 ml) were adjusted to contain 40% sucrose by mixing with an equal volume of 80% sucrose, placed in an ultracentrifuge tube, overlaid with sucrose gradients (35%, 2.25 ml; 15%, 0.45 ml; 5%, 0.45 ml; and 0%, 0.45 ml), and centrifuged at 100,000 g at 4°C for 20 hr in an SW55 rotor. After centrifugation, 12 fractions (0.4 ml each from top) were collected and analyzed by Western blotting. For GD3 detection, 1 µl of each fraction was spotted onto nitrocellulose membranes and subjected to immunoblotting.

Subcellular Fractionation

Subcellular fractionation to isolate the MAM was performed as described by Wieckowski et al. (2009).

Triacylglycerol Measurement

Triacylglycerol measurement was carried out using a triglyceride quantification colorimetric kit (BioVison: No-k662) according to the manufacturer’s protocol. The optical density at 570 nm was measured, and triacylglycerol amount was calculated using a standard curve.

Statistical Analyses

The results were averaged, expressed as the mean ± standard error of the mean or standard deviation, and analyzed using a Student’s test. The p values are indicated by asterisks in the figures with the following notations: *p ≤ .05; **p ≤ .01; ***p ≤ .001.

Supplemental Material

Supplemental material for Syntaxin 17 Recruits ACSL3 to Lipid Microdomains in Lipid Droplet Biogenesis

Supplemental Material for Syntaxin 17 Recruits ACSL3 to Lipid Microdomains in Lipid Droplet Biogenesis by Hana Kimura, Kohei Arasaki, Moe Iitsuka and Mitsuo Tagaya in Contact