GPAT4-Generated Saturated LPAs Induce Lipotoxicity through Inhibition of Autophagy by Abnormal Formation of Omegasomes.

Excessive levels of saturated fatty acids are toxic to vascular smooth muscle cells (VSMCs). We previously reported that mice lacking VSMC-stearoyl-CoA desaturase (SCD), a major enzyme catalyzing the detoxification of saturated fatty acids, develop severe vascular calcification from the massive accumulation of lipid metabolites containing saturated fatty acids. However, the mechanism by which SCD deficiency causes vascular calcification is not completely understood. Here, we demonstrate that saturated fatty acids significantly inhibit autophagic flux in VSMCs, contributing to vascular calcification and apoptosis. Mechanistically, saturated fatty acids are accumulated as saturated lysophosphatidic acids (LPAs) (i.e. 1-stearoyl-LPA) possibly synthesized through the reaction of GPAT4 at the contact site between omegasomes and the MAM. The accumulation of saturated LPAs at the contact site causes abnormal formation of omegasomes, resulting in accumulation of autophagosomal precursor isolation membranes, leading to inhibition of autophagic flux. Thus, saturated LPAs are major metabolites mediating autophagy inhibition and vascular calcification.

Introduction

Vascular calcification is a major complication in aging populations and pan class="Species">patients with chronic kidney disease (CKD), who display characteristics of premature aging (Kovacic et al., 2011a, Kovacic et al., 2011b, Liu et al., 2013, Shanahan et al., 2011). Vascular calcification is highly associated with cardiovascular mortality (Blacher et al., 2001, Block et al., 2007). Recent studies revealed that saturated fatty acids (SFAs) contribute to the development of vascular calcification by promoting vascular apoptosis and transdifferentiation of vascular smooth muscle cells (VSMCs) to osteoblastic-like cells (Durham et al., 2018, Proudfoot et al., 2000). “Lipotoxicity” of excess SFAs is manifested as apoptosis, endoplasmic reticulum (ER) stress, or production of oxidative stress that accelerates vascular calcification (Anderson et al., 2012, Brodeur et al., 2013, Masuda et al., 2012). Stearoyl-CoA desaturase (SCD) catalyzes the conversion of SFAs to unsaturated fatty acids (UFAs) by introducing a double bond at the Δ9 position (Miyazaki and Ntambi, 2003). In our previous in vivo and in vitro studies, SCD inhibition induced vascular calcification and cell apoptosis associated with excess accumulation of metabolites of SFAs in VSMCs (Masuda et al., 2015). Using lipidomic and shRNA screening approaches, we identified three acyltransferases, glycerol-3-phosphate acyltransferase-4 (GPAT4), and acylglycerol-3-phosphate acyltransferase-3 and -5 (AGPAT3 and AGPAT5), that mediate SFA-induced ER stress and vascular calcification by generating fully saturated phosphatidic acids (PAs) such as distearoyl-phosphatidic acid in VSMCs. Importantly, two recent bias studies have confirmed our observations that GPAT4 is a central enzyme mediating SFA-induced global cellular lipotoxicity (Piccolis et al., 2019, Zhu et al., 2019). However, whether other lipotoxic pathways and other SFA metabolites generated through GPAT4 contribute to vascular calcification is still obscure.

The autophagy-related gene (ATG) family generates the autophagosome membrane (Dikic and Elazar, 2018, Mizushima and Komatsu, 2011). During the activation state of autophagy, the phosphatidylinositol 3-kinase pan class="Gene">VPS34 and Beclin-1 complex is recruited to the ER-mitochondria contact site, the mitochondria-associated membrane (MAM) (Hamasaki et al., 2013), and generates phosphatidylinositol 3-phosphate (PI3P) as an early event of autophagy (Itakura et al., 2008, Matsunaga et al., 2009, Sun et al., 2008, Zhong et al., 2009). PI3P-enriched ER membrane, known as the origin for autophagosome membrane formation, can be visualized by specific PI3P-binding proteins such as double FYVE domain-containing protein-1 (DFCP1) and WD repeat domain phosphoinositide-interacting protein-2 (WIPI2), which are positive membranes called “omegasomes” whose structure is shaped as the letter “Ω” (Axe et al., 2008, Polson et al., 2010). Omegasomes are a platform to form isolation membranes (an autophagosomal precursor) that are unclosed autophagosome membrane structures visualized by either LC3 or ATG16L1 in combination with omegasome markers such as WIPI2 and DFCP1 (Axe et al., 2008, Itakura and Mizushima, 2010, Polson et al., 2010). When the omegasome structure is deformed by inhibition of the capping protein (CapZ)-dependent actin filament assembly, the abnormal isolation membrane structure accumulates DFCP1-positive elongated membrane structures (Mi et al., 2015). However, metabolites that can affect the omegasome structure are still not identified. LC3 is conjugated to the PE molecule on the isolation membrane as lipidated LC3 (LC3-II) by the ATG12-ATG5-ATG16 complex, similar to ubiquitin conjugation after cleavage of glycine 120 residue of LC3 by ATG4 (Fujita et al., 2008, Ichimura et al., 2000, Kabeya et al., 2004, Matsushita et al., 2007, Mizushima et al., 2003, Tanida et al., 2004). The isolation membrane matures to an autophagosome with a double membrane structure while loading autophagic cargo such as Sequestosome-1 (p62) or damaged mitochondria (Dikic and Elazar, 2018, Yoshii and Mizushima, 2017). The mature autophagosome then fuses with a lysosome to generate autolysosomes for degradation inside of Lc3-II and cargo proteins (Gatica et al., 2018, Lamark et al., 2017). Inhibition of autophagy machinery such as an autophagosome-lysosome fusion or protein degradation in autolysosomes induces accumulation of Lc3-II and autophagic substrates (Yoshii and Mizushima, 2017). Because a number of membrane formations and modulations are involved in autophagy machinery, phospholipid modulations such as PI3P and PE are known to affect autophagic flux (Mitroi et al., 2017, Petiot et al., 2000, Rockenfeller et al., 2015). Besides their role in phospholipids, fatty acyl chains are involved in autophagic machinery. For instance, SCD contributes to autophagosome formation at an early stage of autophagy initiation through an unknown mechanism (Ogasawara et al., 2014). In addition, there are a number of studies showing that saturated and unsaturated fatty acids (SFAs and UFAs) modulate autophagy (Choi et al., 2009, Las et al., 2011, Mei et al., 2011, Tan et al., 2012), although a precise conclusion has not been made due to the lack of a detailed mechanistic study. In addition, emerging evidence suggests that the modulation of autophagy contributes to lipotoxicity-mediated diseases such as vascular calcification (Leidal et al., 2018, Levine and Kroemer, 2008, Nussenzweig et al., 2015, Rogers and Aikawa, 2019, Ueno and Komatsu, 2017). Inhibition of autophagy by phosphatidylinositol 3-kinase (PI3K) inhibitor worsens inorganic phosphate (Pi)-induced vascular calcification through secretion of matrix vesicles with alkali phosphatase (Dai et al., 2013), whereas an activator of autophagy, rapamycin, inhibits vascular calcification (Zhao et al., 2015). CKD induces the accumulation of aortic Lc3-II (Dai et al., 2013). However, there is no direct evidence that autophagy plays a causative or protective role in regulating vascular calcification in vivo.

In this study, we present evidence that SFA-mediated autophagy inhibition contributes to lipotoxicities including pan class="Disease">vascular calcification and apoptosis and identify the mechanisms by which SFAs mediate autophagic flux.

Results

Saturated Fatty Acids Induced Accumulation of Lipidated Lc3 and p62 in Mouse Aortas and Cultured Vascular Smooth Muscle Cells

Smooth muscle cell (SMC)-specific SCD1/2 deficiency induced severe medial pan class="Disease">calcification, consistent with our previous report (Masuda et al., 2015) (Figure S1A and S1B). Because recent studies suggest that autophagy contributes to vascular calcification, we examined levels of Lc3b-II and p62 proteins in the aortas from SMC-Scd1/2 knockout (KO) mice by immunoblot analysis (Figure 1A). Scd1/2 KO induced accumulation of lipidated Lc3b (Lc3b-II) and p62 proteins. Immunofluorescence analysis revealed a significant increase in Lc3 punctation in the aortas of SMC-Scd1/2 KO mice (Figure 1B). In addition, another mouse model with vascular calcification, mice with 5/6 nephrectomy (Figure S1C), also had significantly higher protein levels of the aortic autophagy substrates Lc3b-II and p62 than sham-operated control mice (Figure S1D). These results suggest that SMC-specific Scd1/2 KO and CKD modulate autophagic flux in vivo. To examine the effect of fatty acids on autophagy in VSMCs, we treated VSMCs with 15 major fatty acids in mammals and analyzed levels of Lc3b and p62 proteins under nutrient-rich (complete media) and starvation (EBSS) conditions (Figures 1C and S1E). SFAs such as palmitic (C16:0), stearic (C18:0), and arachidic (C20:0) acids but no unsaturated fatty acids (UFAs) induced accumulation of Lc3b-II and p62. Levels of Lc3b-II and p62 were dose-dependently increased by treatment with C18:0 under both nutrient-rich and starvation (EBSS) conditions (Figure 1D and S1F). Treatment with CAY10566 (an SCD-specific inhibitor) potentiated the effect of C18:0 on autophagy in VSMCs (Figure 1E). Consistent with the immunoblot analysis, treatment of VSMCs with C18:0 significantly increased the number of Lc3 puncta (Figures 1F and 1G). In addition, GFP-p62-stable VSMCs treated with SFAs and SCD inhibitor but not UFAs significantly induced accumulation of GFP-p62 puncta (Figures S1G and S1H). Similar to VSMCs, treatment with C18:0 in various cell lines such as NIH3T3 and HeLa also induced accumulation of Lc3b-II and p62 (Figures S1I and S1J). To further analyze the effect of SFA-induced autophagy inhibition, we assessed ER-phagy. C18:0 blocked starvation-induced degradation of ER-phagy receptor proteins such as Testis-expressed 264 (Tex264) (An et al., 2019, Chino et al., 2019) and Fam134b (Khaminets et al., 2015, Kohno et al., 2019), accompanied by accumulation of Lc3b-II (Figure S1K).

Figure 1

Saturated Fatty Acids (SFAs) Inhibited Autophagic Flux in Mouse Models and Cultured Vascular Smooth Muscle Cells (VSMCs)

(A) Immunoblot analysis of aortic media of control and smooth muscle cell (SMC)-specific Scd1/2 knockout (KO) mice.

(B) Confocal microscopy merged images of alpha-smooth muscle actin (aSMA, green), Lc3 (red), and DAPI (blue) in aortas of control and SMC-Scd1/2 KO mice with CKD. Scale represents 50 μm. Arrow indicates Lc3 punctation.

(C) Immunoblot analysis of human VSMCs treated with 250 μM of 15 major saturated or unsaturated fatty acids for 6 h.

(D) Dose-dependent effect of C18:0 (SFA) on Lc3b and p62 proteins.Immunoblot analysis of VSMCs treated with C18:0 for 6 h.

(E) Immunoblot analysis of VSMCs treated with C18:0 and 3 μM CAY10566 for 6 h.

(F) Immunofluorescence of Lc3 (red) and DAPI (blue) in VSMCs treated with vehicle or 200 μM C18:0 for 2 h.

(G) Number of Lc3 puncta in Figure 1F (n = 25, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(H) Scheme of measurement of autophagic flux with GFP-LC3-RFP-LC3ΔG probe.

(I) Measurement of autophagic flux using stable VSMCs expressing GFP-LC3-RFP-LC3ΔG co-treated with Veh, 300 μM C16:0, C16:1, C18:0, C18:1, or 200 nM bafilomycin A1 under EBSS conditions for 6 h. GFP/RFP ratio data were expressed as fold value against Veh (n = 4, One-way ANOVA). ∗p < 0.05.

(J) Measurement of autophagic flux in stable VSMCs expressing GFP-LC3-RFP-LC3ΔG treated with C18:0 or bafilomycin A1 under EBSS conditions for 6 h (n = 4).

Saturated Fatty Acids (pan class="Chemical">SFAs) Inhibited Autophagic Flux in Mouse Models and Cultured Vascular Smooth Muscle Cells (VSMCs)

(A) Immunoblot analysis of aortic media of control and smooth muscle cell (SMC)-specific Scd1/2 knockout (KO) pan class="Species">mice.

(B) Confocal microscopy merged images of pan class="Gene">alpha-smooth muscle actin (aSMA, green), Lc3 (red), and DAPI (blue) in aortas of control and SMC-Scd1/2 KO mice with CKD. Scale represents 50 μm. Arrow indicates Lc3 punctation.

(C) Immunoblot analysis of human VSMCs treated with 250 μM of 15 major saturated or pan class="Chemical">unsaturated fatty acids for 6 h.

(D) Dose-dependent effect of C18:0 (pan class="Chemical">SFA) on Lc3b and p62 proteins.Immunoblot analysis of VSMCs treated with C18:0 for 6 h.

(E) Immunoblot analysis of VSMCs treated with C18:0 and 3 μM pan class="Chemical">CAY10566 for 6 h.

(F) Immunofluorescence of Lc3 (red) and pan class="Chemical">DAPI (blue) in VSMCs treated with vehicle or 200 μM C18:0 for 2 h.

(H) Scheme of measurement of autophagic flux with GFP-LC3-pan class="Gene">RFP-LC3ΔG probe.

(I) Measurement of autophagic flux using stable VSMCs expressing GFP-LC3-pan class="Gene">RFP-LC3ΔG co-treated with Veh, 300 μM C16:0, C16:1, C18:0, C18:1, or 200 nM bafilomycin A1 under EBSS conditions for 6 h. GFP/RFP ratio data were expressed as fold value against Veh (n = 4, One-way ANOVA). ∗p < 0.05.

(J) Measurement of autophagic flux in stable VSMCs expressing GFP-LC3-pan class="Gene">RFP-LC3ΔG treated with C18:0 or bafilomycin A1 under EBSS conditions for 6 h (n = 4).

SFAs Block Autophagic Flux

SFAs induced accumulation of pan class="Chemical">Lc3b-II and p62. Accumulation of Lc3b-II possibly occurs by both induction and inhibition of autophagy. We next examined whether SFAs either induce or inhibit autophagic flux using an autophagy probe, the GFP-LC3-RFP-LC3ΔG probe (Kaizuka et al., 2016). The probe is cleaved by Atg4 under autophagy induction and divided into GFP-LC3 and RFP-LC3ΔG. GFP-LC3 conjugates with PE on an isolation membrane through glycine 120 residue. After autophagosome-lysosome fusion, interior GFP-LC3 is degraded and GFP fluorescence is diminished by the acidic environment of the autolysosome. Because RFP-LC3ΔG cannot be degraded by incapability of conjugation to PE due to deletion of glycine 120, RFP fluorescence can be used as internal control due to its stability. A decline of the GFP/RFP ratio indicates autophagic flux (Figure 1H). To examine the effect of fatty acids on autophagic flux, VSMCs stably expressing GFP-LC3-RFP-LC3ΔG were treated with SFAs and UFAs. Treatment with SFAs such as C16:0 and C18:0 but not UFAs such as C16:1 and C18:1 dose-dependently increased the GFP/RFP ratio compared with vehicle (Figures 1I and 1J). These results conclude that SFAs block autophagic flux, resulting in significant accumulation of autophagic substrates such as Lc3b-II and p62 in VSMCs.

Inhibition of Autophagy Induces Mortality and Vascular Calcification in CKD

Because SFAs inhibit autophagic flux, we next examined whether inhibition of autophagy induces pan class="Disease">vascular calcification. In VSMC cultures, Atg5 KO significantly induced mineralization of VSMCs (Figures 2A, 2B, and S2A). We next generated SMC-specific Atg5 KO mice to study the effect of Atg5 on the development of autophagy in CKD in vivo. The deletion of the gene of interest was observed in smooth muscle cells but no other cells, as previously reported (Figure 2C and data not shown). As expected, SMC-Atg5 deficiency blocked Lc3b lipidation, blocking autophagy and inducing p62 accumulation (Figure 2C) in the aortic media. To study CKD-dependent vascular calcification, CKD was induced by 5/6 nephrectomy. Interestingly, CKD drastically shortened the life span of SMC-specific Atg5 KO mice (Figure 2D) without affecting levels of serum creatinine, phosphorus, or calcium (Figures S2B–S2D). More importantly, SMC-Atg5 deficiency significantly aggravated aortic medial calcification (Figures 2E–2G) and apoptosis (Figures 2H and 2I) and damaged mitochondria accumulation (Figures 2J) 6 weeks after CKD induction.

Figure 2

Deficiency of Atg5-Dependent Autophagy in Smooth Muscle Cells (SMCs) Induced Death at a High Mortality Rate, Severe Calcification, and Apoptosis in 5/6 Nephrectomized Mice

(A) Mineralization of scrambled sg (control) or Atg5 knockout (KO) VSMCs treated with 2.6 mM Pi every 2 days for 6 days. (n = 6, 2-tailed Student's t test) ∗p < 0.05. Data are represented as mean ± SEM.

(B) Alizarin red staining of control and Atg5 knockout VSMCs treated with Pi for 6 days.

(C) Immunoblot analysis of aortic medias from control and SMC-Atg5 KO mice.

(D) Survival rate of control and SMC-Atg5 KO mice after 5/6 nephrectomy (n = 20, log rank and Wilcoxon signed rank method) ∗p < 0.05.

(E) Photograph (x10 magnification) of the lesions of aortic sinuses with von Kossa staining in control and SMC-Atg5 KO mice with CKD. Scale represents 50 μm.

(F) Quantitative analysis of aortic calcified regions (n = 8, 2-tailed Student's t test). ∗∗∗p < 0.001. Data are represented as mean ± SEM.

(G) Aortic calcium content (n = 8, one-way ANOVA). ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SEM.

(H) Photograph of TUNEL staining (magenta) and DAPI (blue) from aortas of control and SMC-Atg5 KO mice with CKD. Scale represents 50 μm.

(I) Quantitative analysis of TUNEL-positive nuclei in aortic lesions (n = 8, 2-tailed Student's t test). ∗∗∗p < 0.001. Data are represented as mean ± SEM.

(J) Electron micrograph of aortic lesions in control and SMC-Atg5 KO mice with CKD. Scale represents 1 μm. SMC, smooth muscle cells; EC, endothelial cells; AP, autophagosome; Mt, mitochondria.

Deficiency of Atg5-Dependent Autophagy in Smooth Muscle Cells (SMCs) Induced pan class="Disease">Death at a High Mortality Rate, Severe Calcification, and Apoptosis in 5/6 Nephrectomized Mice

(B) Alizarin red staining of control and pan class="Gene">Atg5 knockout VSMCs treated with Pi for 6 days.

(C) Immunoblot analysis of aortic medias from control and SMC-Atg5 KO pan class="Species">mice.

(D) Survival rate of control and SMC-Atg5 KO pan class="Species">mice after 5/6 nephrectomy (n = 20, log rank and Wilcoxon signed rank method) ∗p < 0.05.

(E) Photograph (x10 magnification) of the lesions of aortic sinuses with von Kossa staining in control and SMC-pan class="Gene">Atg5 KO mice with CKD. Scale represents 50 μm.

(H) Photograph of TUNEL staining (magenta) and DAPI (blue) from aortas of control and SMC-pan class="Gene">Atg5 KO mice with CKD. Scale represents 50 μm.

(I) Quantitative analysis of TUNEL-positive nuclei in pan class="Disease">aortic lesions (n = 8, 2-tailed Student's t test). ∗∗∗p < 0.001. Data are represented as mean ± SEM.

(J) Electron micrograph of aortic lesions in control and SMC-pan class="Gene">Atg5 KO mice with CKD. Scale represents 1 μm. SMC, smooth muscle cells; EC, endothelial cells; AP, autophagosome; Mt, mitochondria.

SFAs Inhibit Starvation-Induced Autophagy at Steps Prior to Degradation in Autolysosomes

Because SFAs inhibit autophagic flux, we put in extensive efforts to elucidate the precise mechanism by which pan class="Chemical">SFAs block autophagy. To enhance the effect of SFAs on autophagy, we performed all of the mechanistic studies under starvation conditions with EBSS treatment. We first performed confocal microscopic analysis with co-immunostaining of Lc3 and Lamp1 to assess whether SFAs affect lysosomal function (Figure S3A). Treatment of VSMCs with SFAs such as C18:0 exhibited significant accumulation of Lc3 compared with vehicle (Figure S3B), but co-localization of Lc3 and Lamp1 was not increased (Figure S3C). Treatment with SCD inhibitor (CAY) also increased total Lc3 but not co-localization of Lc3 and Lamp1, similar to C18:0 treatment. In contrast, treatment with various lysosomal inhibitors such as the protease inhibitors leupeptin, pepstatin A, and bafilomycin A1 increased co-localization of Lc3 and Lamp1 in VSMCs (Figure S3A–S3C). To further evaluate the effect of SFAs on lysosome function and other lysosomal degradation pathways, we examined degradation of EGFR, which can be degraded through the endocytosis-lysosomal pathway in response to EGF (Alwan et al., 2003, Tsuboyama et al., 2016) (Figure S3D). C18:0 did not influence EGF-induced lysosomal EGFR degradation, whereas bafilomycin A1 blocked degradation of EGFR after EGF treatment in VSMCs. These results revealed that SFAs do not affect lysosome function but inhibit autophagy prior to autolysosome formation. We next generated VSMCs stably expressing GFP-Syntaxin 17 (GFP-STX17), a matured autophagosome marker (Itakura et al., 2012, Tsuboyama et al., 2016, Uematsu et al., 2017, Yoshii and Mizushima, 2017), to examine which step of autophagy is affected by SFAs. Based on the immunofluorescence analysis, unlike bafilomycin A1, SFA treatment significantly reduced co-localizations between Lc3 and STX17 (Figures S3E and S3F) and between Lc3 and Lamp1 (Figures S3E and S3G), whereas STX17 and Lamp1 double-negative Lc3 was significantly increased in VSMCs treated with C18:0 (Figures S3E and S3H). These results suggest that SFAs block a step that occurs prior to mature autophagosome formation.

SFAs Induce Abnormal Formation of DFCP1/WIPI2-Positive Omegasomes and Accumulation of Isolation Membranes

We expect that SFAs inhibit autophagy machinery at a step prior to formation of the mature autophagosome. Formation of omegasomes is an early event of autophagosome generation. To determine whether pan class="Chemical">SFAs influence omegasome formation and structures, we generated VSMCs stably expressing GFP-DFCP1 and analyzed them with confocal microscopy (Figure 3A). Because DFCP1 localizes on omegasome structures by binding to PI3P, DFCP1 forms punctate-like structures (Axe et al., 2008). In normal VSMCs (vehicle treatment), DFCP1 formed punctate-like structures in a large proportion (97.1%) and punctates with long tube-like structures in a small proportion (2.9%) (Figures 3A and 3B). Surprisingly, on the other hand, C18:0 treatment induced abnormally enlarged DFCP1 structures that formed large ring shapes similar to the letter Ω and long tube shapes with branches in over 95% of VSMCs, whereas less than 5% showed normal punctate-like structures (Figures 3A and 3B). To confirm that SFAs induce abnormally enlarged omegasome formation, we examined the structures of omegasomes using endogenous WIPI2, another omegasome marker, and analyzed co-localization with Lc3 in VSMCs treated with C18:0 (Figure 3C). WIPI2 formed normal punctate structures in VSMCs treated with vehicle. Similar to DFCP1-postive membranes, C18:0 treatment drastically induced abnormal enlargement of WIPI2-positive omegasomes (Figures 3C and 3D). In addition, Lc3 was more significantly accumulated on WIPI2-positive membranes in VSMCs treated with C18:0 than VSMCs treated with vehicle (Figures 3C and 3E), suggesting accumulation of unclosed immature autophagosomes (isolation membranes) in the omegasomes. To confirm the accumulation of isolation membranes, we used ATG16L1 as an isolation membrane-specific marker. VSMCs treated with C18:0 had significantly higher DFCP1-postive ATG16L1 punctates (Figures 3F and 3G). Consistent with detailed confocal immunofluorescence analyses, electron microscopy analysis revealed that accumulation of abnormal enlarged membrane structures and isolation membranes was specifically observed in VSMCs treated with C18:0 (Figure 3H). We next examined whether the modulations of PI3 kinase- (Vps34) and PI3P phosphatase- (myotubularin-related phosphatase, Mtmr) mediated PI3P homeostasis contribute to SFA-induced omegasome formation. The total amount of PI3P was not changed in C18:0-treated VSMCs (Figure S4A). Abnormal omegasome formation by C18:0 treatment was significantly attenuated by the inhibition of a PI3 kinase, VPS34 (Figures S4B and S4C). In addition, unlike C18:0 treatment, the modulation of PI3P by the overexpression of VPS34-FLAG or the knockdown of Mtmr did not induce abnormal DFCP1-positive omegasome formation in VSMCs (Figures S4D, S4E, and Table S1). Two studies consistently reported that autophagy-related Mtmr-6, -8, and -9 did not induce abnormal omegasome formation (Mochizuki et al., 2013, Zou et al., 2012). These data suggest that modulations of Vps34 and Mtmr-regulated PI3P did not contribute to SFA-induced abnormal omegasome formation. SFA-induced abnormal omegasome formation and accumulation of isolation membranes occurred not only in VSMCs but also in HeLa cells (Figures S4F–S4H). These data suggest that SFAs lead to abnormal formation of omegasomes, resulting in accumulation of Lc3-II-anchored isolation membranes, which is a common effect of SFAs.

Figure 3

SFAs Lead to Abnormal Formation of DFCP1/WIPI2-Positive Omegasomes and Accumulation of Lc3 in VSMCs

(A) Confocal microscopy images of GFP-DFCP1 (green) in fixed stable VSMCs expressing GFP-DFCP1 treated with EBSS and BSA (Veh) or 200 μM C18:0 (SFA) for 2 h. Scale represents 10 μm. Arrow indicates enlarged tube and ring structure of GFP-DFCP1 with a Ω shape.

(B) Percentage of VSMCs with one or more enlarged GFP-DFCP1 structures in analysis of Figure 3A. (n = 3, from 35 cells in each experiment; 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(C) Confocal microscopy images of endogenous WIPI2 (green) and Lc3 (red) in wild-type VSMCs treated with EBSS and Veh or SFA for 2 h. Scale represents 10 μm. Arrow indicates enlarged WIPI2 and arrowhead indicates WIPI2-positive Lc3.

(D) Percentage of VSMCs with enlarged WIPI2 structures (n = 3, from 33 cells in each experiment; 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(E) Number of Lc3 punctation with WIPI2 (n = 22, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(F) Confocal microscopy images of GFP-DFCP1 (green) and endogenous ATG16L1 (red) in VSMCs treated with EBSS and Veh or SFA for 2 hr. Scale represents 10 μm. Arrowhead indicates DFCP1-positive ATG16L1.

(G) Number of ATG16L1 punctation with GFP-DFCP1 (n = 26, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(H) Electron micrography in wild-type VSMCs treated with EBSS and Veh or SFA for 2 h.

Scale represents 1 μm. Asterisk indicates enlarged membrane structure, arrowhead indicates unclosed membrane as isolation membrane, and arrow indicates Ω-like structure. ER, endoplasmic reticulum Mt, mitochondria; AP, autophagosome; AL, autolysosome; Nuc, nucleus.

SFAs Lead to Abnormal Formation of pan class="Gene">DFCP1/WIPI2-Positive Omegasomes and Accumulation of Lc3 in VSMCs

(A) Confocal microscopy images of GFP-pan class="Gene">DFCP1 (green) in fixed stable VSMCs expressing GFP-DFCP1 treated with EBSS and BSA (Veh) or 200 μM C18:0 (SFA) for 2 h. Scale represents 10 μm. Arrow indicates enlarged tube and ring structure of GFP-DFCP1 with a Ω shape.

(C) Confocal microscopy images of endogenous pan class="Gene">WIPI2 (green) and Lc3 (red) in wild-type VSMCs treated with EBSS and Veh or SFA for 2 h. Scale represents 10 μm. Arrow indicates enlarged WIPI2 and arrowhead indicates WIPI2-positive Lc3.

(E) Number of Lc3 punctation with pan class="Gene">WIPI2 (n = 22, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(F) Confocal microscopy images of GFP-pan class="Gene">DFCP1 (green) and endogenous ATG16L1 (red) in VSMCs treated with EBSS and Veh or SFA for 2 hr. Scale represents 10 μm. Arrowhead indicates DFCP1-positive ATG16L1.

(G) Number of ATG16L1 punctation with GFP-pan class="Gene">DFCP1 (n = 26, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

Scale represents 1 μm. Asterisk indicates enlarged membrane structure, arrowhead indicates unclosed membrane as isolation membrane, and arrow indicates Ω-like structure. ER, endoplasmic reticulum Mt, mitochondria; AP, autophagosome; AL, autolysosome; Nuc, pan class="Gene">nucleus.

Characterization of SFA-Induced Abnormal Formation of Omegasomes

To examine how SFAs induce the abnormal formation of omegasomes, we captured a live image of VSMCs expressing GFP-DFCP after pan class="Chemical">SFA treatment. As shown in Video S1, the peak of SFA-induced abnormal formation of omegasomes occurred between 0.5 and 1 h after C18:0 treatment. Five major characteristics of abnormal omegasome formation were detected (Video S1 and Figure S4): (1) enlargement: a single omegasome is simply enlarged (Figure S4I); (2) multiplication: an original omegasome produces another omegasome, which produces another omegasome (Figure S4J); (3) elongation: the omegasome membrane is simply elongated (Figure S4K); (4) clusterization: multiple omegasomes approach each other and form a cluster (Figure S4L); (5) fusion: some of the omegasome clusters fuse and generate a single large omegasome (Figure S4M). These abnormal omegasome formations were not observed in VSMCs treated with vehicle (Video S2).

Video S1. Time Lapse Imaging of SFA-induced Abnormal Omegasome Formation in Stable VSMCs Expressing GFP-DFCP1, Related to Figure 3

Images of stable VSMCs expressing GFP-DFCP1 treated with 200 μM pan class="Chemical">C18:0 (SFA) were acquired at 1 frame per 5.58 s for 27.5 min under EBSS conditions. Scale represents 5 μm and time indicates time elapsed after SFA treatment. SFA-induced abnormal characteristics of omegasomes (enlargement, multiplication, elongation, clusterization, and fusion) are indicated by white dotted squares, related to Figure S4I–S4M

Video S2 Time Lapse Imaging of Omegasome Formation in Stable VSMCs Expressing GFP-DFCP1 under Vehicle Treatment, Related to Figure 3

Images of stable VSMCs expressing GFP-DFCP1 were acquired at 1 frame per 5.58 s for 18.5 min under EBSS conditions or BSA (vehicle) conditions. Scale represents 5 μm and time indicates time elapsed after EBSS treatment. Normal omegasome formation, in which punctate-like structure appears and then quickly disappears under starvation (white dotted squares). Most GFP-pan class="Gene">DFCP1s are localized in the Golgi body on the right side of the video image, consistent with Figure S5A (GFP-DFCP1 co-localized with the Golgi marker RCAS1)

Gpat4 Mediates Accumulation of Lc3 and Enlargement of WIPI2-Positive Omegasomes through SFA Treatment

To examine whether free SFAs or pan class="Chemical">SFA-derived metabolites affect autophagy, we first modified acyl-CoA synthetase (ACS) with Triacsin C, an inhibitor of ACS. Triacsin C treatment completely blocked the incorporation of fatty acids into acyl lipids (Figure 4A). In the presence of Triacsin C, C18:0 treatment no longer increased levels of autophagic substrates such as Lc3b-II and p62 (Figure 4B). These data suggest that SFAs must be incorporated into an acyl-lipid to inhibit autophagic flux. We therefore next examined which acyltransferase contributes to SFA-mediated autophagy inhibition using an shRNA-mediated acyltransferase knockdown (KD) VSMC library that we previously generated (Masuda et al., 2015) (Figure 4C). Each acyltransferase KD VSMC was treated with C18:0 for 2 h. Similar to Figures 1F and 1G, C18:0 treatment significantly induced accumulation of Lc3 puncta compared with vehicle. Interestingly, Gpat4 and Agpat3 KD reciprocally regulated SFA-induced accumulation of Lc3 in VSMCs (Figures 4C and 4D). Gpat4 KD reduced SFA-induced Lc3 puncta accumulation by 79%, whereas Agpat3 knockdown significantly increased the number of Lc3 puncta in the presence of SFAs. In addition, immunoblot analysis confirmed that Gpat4 KD reduced levels of Lc3b-II increased with C18:0 treatment, whereas Agpat3 KD increased levels (Figure 4E). We next examined whether Gpat4 contributes to SFA-induced abnormal enlargement of WIPI2-positive omegasomes and accumulation of isolation membranes. Gpat4 KD reduced levels of abnormally enlarged WIPI2 membranes and accumulation of Lc3 on WIPI2-positive omegasomes under SFA treatment, suggesting that Gpat4 mediates SFA-induced abnormal enlargement of omegasomes, leading to accumulation of Lc3-positive isolation membranes (Figures 4F–4H). To analyze the effect of Gpat4 knockdown on SFA-induced autophagic flux inhibition, stable VSMCs expressing GFP-LC3-RFP-LC3ΔG probe were transfected with Gpat4 siRNA and then treated with SFAs under starvation conditions. Gpat4 siRNA treatment significantly reduced levels of Gpat4 in VSMCs by 88% (Figure 4I). Gpat4 knockdown significantly attenuated autophagic flux inhibition by C18:0 compared with scrambled siRNA (Figure 4J).

Figure 4

Gpat4 Mediates Accumulation of Lipidated Lc3 and Abnormal Formation of WIPI2-Positive Omegasomes by SFAs

(A) Scheme of fatty acid metabolism with acyl-CoA synthetase (ACS), Gpat4, and Agpat3. TriacsinC: an acyl-CoA synthetase inhibitor.

(B) Immunoblot analysis of VSMCs treated with 250 μM C18:0 with or without 10 μM Triacsin C for 6 h.

(C) Number of Lc3 punctations in shRNA-based acyltransferase knockdown (KD) VSMCs treated with BSA (Veh) or 200 μM C18:0 for 6 h under complete media conditions (n = 4, one-way ANOVA). ∗∗∗p < 0.001 versus Empty shRNAVeh. ###p < 0.001 versus Empty shRNA SFA. Data are represented as mean ± SEM.

(D) Immunofluorescence of Lc3 (red) and DAPI (blue) in empty, Gpat4 and Agpat3 KD VSMCs treated with 200 μM C18:0 for 6 h under complete media conditions. Scale represents 10 μm.

(E) Immunoblot analysis of Gpat4 or Agpat3 KD VSMCs treated with Veh or 250 μM C18:0 for 6 h under complete media conditions.

(F) Immunofluorescence of endogenous WIPI2 (green) and Lc3 (red) in empty and Gpat4 KD VSMCs treated with Veh or 200 μM C18:0 for 2 h under complete media conditions. Arrow indicates enlarged WIPI2 and arrowhead indicates WIPI2-positive Lc3 punctation. Scale represents 10 μm.

(G) Percentage of VSMCs with enlarged WIPI2 structures (n = 3, from 33 cells in each experiment; 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(H) Number of Lc3 punctation with WIPI2 from Figure 4F analysis (n = 15, one-way ANOVA). ∗p < 0.05; N.S., not significant. Data are represented as mean ± SEM.

(I) Levels of Gpat4 mRNA by qRT-PCR in 50 nM negative control or Gpat4 siRNA-transfected VSMCs. (n = 3, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(J) Measurement of autophagic flux with Gpat4 knockdown. Stable VSMCs expressing GFP-LC3-RFP-LC3ΔG were transfected with 50 nM negative control or Gpat4 siRNA. After 24 h, VSMCs were co-treated with Veh or 500 μM C18:0 under starvation conditions for 6 h. GFP/RFP ratio data was expressed as fold value against Veh (n = 6, one-way ANOVA). ∗p < 0.05.

Gpat4 Mediates Accumulation of pan class="Chemical">Lipidated Lc3 and Abnormal Formation of WIPI2-Positive Omegasomes by SFAs

(A) Scheme of fatty acid metabolism with pan class="Gene">acyl-CoA synthetase (ACS), Gpat4, and Agpat3. TriacsinC: an acyl-CoA synthetase inhibitor.

(B) Immunoblot analysis of VSMCs treated with 250 μM C18:0 with or without 10 μM pan class="Chemical">Triacsin C for 6 h.

(C) Number of Lc3 punctations in shRNA-based acyltransferase knockdown (KD) VSMCs treated with BSA (Veh) or 200 μM pan class="Chemical">C18:0 for 6 h under complete media conditions (n = 4, one-way ANOVA). ∗∗∗p < 0.001 versus Empty shRNAVeh. ###p < 0.001 versus Empty shRNA SFA. Data are represented as mean ± SEM.

(D) Immunofluorescence of Lc3 (red) and pan class="Chemical">DAPI (blue) in empty, Gpat4 and Agpat3 KD VSMCs treated with 200 μM C18:0 for 6 h under complete media conditions. Scale represents 10 μm.

(E) Immunoblot analysis of Gpat4 or pan class="Gene">Agpat3 KD VSMCs treated with Veh or 250 μM C18:0 for 6 h under complete media conditions.

(F) Immunofluorescence of endogenous WIPI2 (green) and pan class="Gene">Lc3 (red) in empty and Gpat4 KD VSMCs treated with Veh or 200 μM C18:0 for 2 h under complete media conditions. Arrow indicates enlarged WIPI2 and arrowhead indicates WIPI2-positive Lc3 punctation. Scale represents 10 μm.

(H) Number of Lc3 punctation with pan class="Gene">WIPI2 from Figure 4F analysis (n = 15, one-way ANOVA). ∗p < 0.05; N.S., not significant. Data are represented as mean ± SEM.

(I) Levels of Gpat4 mRNA by qRT-PCR in 50 nM negative control or pan class="Gene">Gpat4 siRNA-transfected VSMCs. (n = 3, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(J) Measurement of autophagic flux with Gpat4 knockdown. Stable VSMCs expressing GFP-pan class="Gene">LC3-RFP-LC3ΔG were transfected with 50 nM negative control or Gpat4 siRNA. After 24 h, VSMCs were co-treated with Veh or 500 μM C18:0 under starvation conditions for 6 h. GFP/RFP ratio data was expressed as fold value against Veh (n = 6, one-way ANOVA). ∗p < 0.05.

Saturated LPAs Induce Autophagic Flux by Inducing Abnormal Formation of DFCP1-Positive Omegasomes

Gpat4 and pan class="Gene">Agpat3 catalyze the initial two steps of glycerolipid synthesis (Fagone and Jackowski, 2009). Gpat4 is the first enzyme to catalyze the conversion of glycerol-3-phosphate to 1-acyl-lysophosphatidic acid (LPA), which is further acylated by Agpat3 to generate phosphatidic acid (PA) (Figure 4A). Our previous lipidomic study revealed that abnormally excessive amounts of fully saturated LPAs (i.e 18:0-LPA) in addition to fully saturated PAs (i.e. 18:0/18:0-PA) were accumulated in the aortic medias of SMC-Scd1/2 KO mice and cultured VSMCs treated with SFAs (Masuda et al., 2015). We therefore hypothesized that accumulation of saturated LPAs, LPAs derived from SFAs, mediate inhibition of autophagic flux. To confirm our hypothesis, we treated VSMCs with an SFA (C18:0), saturated and unsaturated LPAs (18:0-LPA and 18:1-LPA), and fully saturated and fully unsaturated PAs (18:0/18:0-PA and 18:1/18:1-PA). The immunoblot analysis with Lc3b-II shows that 18:0-LPA but not 18:1-LPA dose-dependently induced accumulation of Lc3b-II (Figure 5A). However, treatment with 18:0-LPA was unable to activate ER stress signals such as ATF4, CHOP, and BiP (Figure 5B). Unlike C18:0 and 18:0-LPA, both 18:0/18:0-PA and 18:1/18:1-PA did not induce accumulation of Lc3b-II (Figure 5C). In addition, treatment with 18:0/18:0-PA did not affect total amounts of 18:0-LPA in VSMCs (Figure 5D). To confirm that 18:0-LPA inhibits autophagic flux similar to SFAs, VSMCs expressing GFP-LC3-RFP-LC3ΔG probe were treated with 18:0-LPA. 18:0-LPA treatment significantly increased the ratio of GFP/RFP fluorescence, indicating that 18:0-LPA inhibits autophagic flux (Figure 5E). To evaluate whether 18:0-LPA modulates DFCP1-positive membrane structures, we perfomed confocal microscopy analysis with stable GFP-DFCP1 VSMCs treated with 18:0-LPA and 18:0/18:0-PA (Figures 5F and 5G). DFCP1 formed enlarged structures under 18:0-LPA treatment, similar to C18:0 treatment, whereas 18:0/18:0-PA did not induce enlargement of DFCP1-postive membrane structures. Electron microscopy analysis confirmed abnormally enlarged membrane structures in VSMCs treated with 18:0-LPA but not in VSMCs treated with vehicle (Figure 5H).

Figure 5

Saturated LPAs, but Not Fully Saturated PAs, Induce Accumulation of Lipidated Lc3b-II and Abnormal Enlargement of DFCP1-Positive Omegasomes in VSMCs

(A) Immunoblot analysis of VSMCs treated with BSA (−), C18:0, 18:0-LPA, 18:1-LPA, or bafilomycin A1 under EBSS conditions for 2 h.

(B) Immunoblot analysis of ER stress proteins in VSMCs treated with BSA (−), C18:0, or 18:0-LPA under complete media conditions for 6 h.

(C) Immunoblot analysis of VSMCs treated with BSA (−), C18:0, 18:0/18:0-PA, 18:1/18:1-PA, or bafilomycin A1 under EBSS conditions for 2 h.

(D) LC-MS-based absolute levels of 18:0-LPA species of VSMCs after 6 h of treatment with 300 μM C18:0, 100 μM 18:0-LPA, or 100 μM 18:0/18:0-PA (n = 3, one-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

(E) Measurement of autophagic flux using stable VSMCs expressing GFP-LC3-RFP-LC3ΔG treated with BSA (Veh), 300 μM C18:0, 100 μM 18:0-LPA, or 200 nM bafilomycin A1 under EBSS conditions for 6 h. GFP/RFP ratio data were expressed as fold value against vehicle (n = 6, one-way ANOVA). ∗p < 0.05.

(F) Confocal microscopy images in stable VSMCs expressing GFP-DFCP1 treated with EBSS and Veh, 200 μM C18:0, 100 μM 18:0-LPA, or 100 μM 18:0/18:0-PA for 2 h. Scale represents 20 μm and arrow indicates VSMCs with enlarged GFP-SFCP1 structures.

(G) Percentage of VSMCs with enlarged GFP-DFCP1 structures in Figure 5F analysis (n = 3, from 33–34 cells in each experiment, one-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

(H) Electron micrograph of wild-type VSMCs co-treated with 100 μM 18:0-LPA and EBSS for 2 h. Scale represents 1 μm. Asterisk indicates enlarged membrane structure. ER, endoplasmic reticulum Mt, mitochondria; Nuc, nucleus.

Saturated LPAs, but Not Fully Saturated pan class="Chemical">PAs, Induce Accumulation of Lipidated Lc3b-II and Abnormal Enlargement of DFCP1-Positive Omegasomes in VSMCs

(A) Immunoblot analysis of VSMCs treated with BSA (−), C18:0, 18:0-pan class="Chemical">LPA, 18:1-LPA, or bafilomycin A1 under EBSS conditions for 2 h.

(B) Immunoblot analysis of ER stress proteins in VSMCs treated with BSA (−), C18:0, or 18:0-pan class="Chemical">LPA under complete media conditions for 6 h.

(C) Immunoblot analysis of VSMCs treated with BSA (−), C18:0, 18:0/18:pan class="Chemical">0-PA, 18:1/18:1-PA, or bafilomycin A1 under EBSS conditions for 2 h.

(D) LC-MS-based absolute levels of 18:0-LPA species of VSMCs after 6 h of treatment with 300 μM pan class="Chemical">C18:0, 100 μM 18:0-LPA, or 100 μM 18:0/18:0-PA (n = 3, one-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

(E) Measurement of autophagic flux using stable VSMCs expressing GFP-LC3-pan class="Gene">RFP-LC3ΔG treated with BSA (Veh), 300 μM C18:0, 100 μM 18:0-LPA, or 200 nM bafilomycin A1 under EBSS conditions for 6 h. GFP/RFP ratio data were expressed as fold value against vehicle (n = 6, one-way ANOVA). ∗p < 0.05.

(F) Confocal microscopy images in stable VSMCs expressing GFP-pan class="Gene">DFCP1 treated with EBSS and Veh, 200 μM C18:0, 100 μM 18:0-LPA, or 100 μM 18:0/18:0-PA for 2 h. Scale represents 20 μm and arrow indicates VSMCs with enlarged GFP-SFCP1 structures.

(H) Electron micrograph of wild-type VSMCs co-treated with 100 μM 18:0-LPA and EBSS for 2 h. Scale represents 1 μm. Asterisk indicates enlarged membrane structure. ER, endoplasmic reticulum Mt, mitochondria; pan class="Gene">Nuc, nucleus.

SFAs Accumulate as Saturated LPAs in the Contact Sites between DFCP1-Positive Omegasomes and Mitochondria-Associated Membranes

To determine which organelle is responsible for abnormal formation of omegasomes by SFAs, we performed immunostaining of various organelle markers in GFP-pan class="Gene">DFCP1 cells (Figures S5A and S5B). Normally, DFCP1 co-localizes with a golgi marker, RCAS1, at the close region of the nucleus (Figure S5A). Although enlarged DFCP1-postive omegasomes in VSMCs treated with C18:0 do not localize with either RCAS1, a lysosomal marker Lamp1, or lipid droplets stained with LipidTox Red (Figures S5A and S5B), they are located in the close vicinity of an ER marker, Calnexin, and a mitochondria marker, Tomm20 (Figure S5A, magnified images). In addition, abnormally enlarged DFCP1 was localized next to other ER markers, KDEL and Tomm20, suggesting that abnormally enlarged omegasomes are possibly associated with contact sites between the ER and mitochondria (Figure S5C). The contact sites between ER-mitochondria and DFCP1 were significantly increased in VSMCs treated with C18:0 (Figures S5C and S5D). Moreover, electron micrographs of SFA-treated VSMCs showed that most of the enlarged membrane structures were connected to the contact sites of the ER and mitochondria (Figure 6A). ER membranes and mitochondria membranes form mitochondria-associated membranes (MAMs), which play many physiological roles such as autophagosome formation with the Beclin-1-Vps34 complex (Hamasaki et al., 2013). In addition, the MAM is a major site of glycerolipid synthesis (Man et al., 2006, Vance, 1990). DFCP1 proteins are translocated to the MAM under starvation (Hamasaki et al., 2013). To determine tight associations between abnormally enlarged DFCP1-positive omegasomes and the MAM, we prepared MAM fractions from vehicle- and SFA-treated VSMCs (Figure 6B). The immunoblot analysis shows that MAM fractions from both vehicle- and SFA-treated VSMCs expectedly contain MAM markers such as Calnexin and Facl4 proteins. Furthermore, PI3K complexes such as Beclin and Vps34 were present in MAM fractions. In the MAM of SFA-treated VSMCs, levels of DFCP1 protein were increased compared with those from vehicle-treated VSMCs (Figure 6B), consistent with the confocal analysis shown in Figures S5C and S5D. Next, we profiled the amount of lysophosphatidic acid species in MAM fractions using an LC/MS-based measurement (Figure 6C). Levels of 18:0-LPA were significantly increased by 14-fold in the MAM from VSMCs treated with C18:0 compared with vehicle. Because MAM fractions from VSMCs treated with C18:0 contain significant amounts of DFCP1, the elevation of 18:0-LPA could be associated with abnormally enlarged DFCP1-membranes. Therefore, we next performed OptiPrep-fractionation to isolate DFCP1-positive membranes (Figures S6A and S6B). Immunoblot analysis showed that GFP-DFCP1 protein was mainly detected on fractions 11, 12, 13, and 14 in both vehicle- and SFA-treated VSMCs (Figure S6A). Because only fraction 11 of SFA-treated VSMCs contains Lc3b-II, Tomm20, and Calnexin proteins, we used fraction 11 as a DFCP1-enriched membrane fraction. LC/MS analysis showed that levels of 18:0-LPA increased by 6-fold in DFCP1-enriched membranes from VSMCs treated with C18:0 (Figure S6B). Interestingly, endogenous Gpat4 protein was detected in MAM fractions from both control and C18:0-treated VSMCs (Figure 6B). To examine whether Gpat4 localizes in the contact site of the MAM and DFCP1 membranes to generate 18:0-LPA as a source of membrane lipids, we analyzed VSMCs overexpressing GFP-DFCP1 and GPAT4-tdTomato by confocal microscopy. As shown in Figures 7A and 7B, SFA treatment showed an increased number of GPAT4 close to the double-positive sites of Tomm20 and DFCP1. In addition, over 85% of Lc3 puncta are observed in proximity with GPAT4 in VSMCs (Figures S6C–S6E). These data suggest that Gpat4 might localize at the MAM and autophagosome formation sites and maybe one of the enzymes indirectly providing lipids for omegasome formation and enlargement. To confirm that DFCP1-omegasomes associated with the MAM are the site of isolation membrane formation and accumulation in VSMCs treated with C18:0 and 18:0-LPA, we performed immunofluorescence analysis to analyze the co-localization of GFP-DFCP1 with Lc3 and Tomm20 (Figures 7C and 7D). Lc3 puncta were more accumulated on the DFCP1 membrane next to Tomm20-positive signals in C18:0- and 18:0-LPA-treated VSMCs than in vehicle-treated VSMCs.

Figure 6

SFAs Lead to Accumulation of Saturated LPAs on Mitochondria-Associated Membrane (MAM) Contact Sites with DFCP1-Positive Omegasomes in VSMCs

(A) Electron micrograph of wild-type VSMCs co-treated with 200 μM C18:0 and EBSS for 2 h. Scale represents 1 μm or 500 nm in the magnified image. Asterisk indicates enlarged membrane structure and arrow indicates connections between enlarged membrane and ER, mitochondria, or ER-mitochondria contact site. ER, endoplasmic reticulum Mt, mitochondria; Nuc, nucleus.

(B) Immunoblot analysis of post nuclear supernatant (PNS), microsome (Micro), cytosol (Cyto), mitochondria (Mito), and MAM fractions of stable VSMCs expressing GFP-DFCP1 co-treated with EBSS and BSA (Veh) or C18:0 (SFA) for 2 h.

(C) LC-MS-based absolute levels of LPA species of purified MAM fractions (n = 3, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

Figure 7

SFAs Increase GPAT4 Localized in Close Proximity to the Double-Positive Site of Tomm20 and DFCP1 Omegasome in VSMCs

(A) Confocal microscopy images of Tomm20 (magenta), GFP-DFCP1 (green), or GPAT4-tdTomato (GPAT4-T) (red) in VSMCs stably expressing GFP-DFCP1 and GPAT4-tdTomato treated with Veh or SFA under EBSS conditions for 2 h. Scale represents 10 μm. Arrow indicates Tomm20, GFP-DFCP1, and GPAT4-tdTomato contact sites.

(B) Number of GFP-DFCP1-GPAT4-Tomm20 contact sites (n = 16, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(C) Confocal microscopy images of Tomm20 (magenta), GFP-DFCP1 (green), and Lc3 (red) in stable VSMCs expressing GFP-DFCP1 co-treated with Veh, C18:0, or 18:0-LPA under EBSS conditions for 2 h. Scale represents 10 μm. Arrow indicates Tomm20, DFCP1, and Lc3-contact site.

(D) Number of GFP-DFCP1-Lc3-Tomm20 contact sites in Figure 7C (n = 20, One-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

SFAs Lead to Accumulation of Saturated pan class="Chemical">LPAs on Mitochondria-Associated Membrane (MAM) Contact Sites with DFCP1-Positive Omegasomes in VSMCs

(A) Electron micrograph of wild-type VSMCs co-treated with 200 μM C18:0 and EBSS for 2 h. Scale represents 1 μm or 500 nm in the magnified image. Asterisk indicates enlarged membrane structure and arrow indicates connections between enlarged membrane and ER, mitochondria, or ER-mitochondria contact site. ER, endoplasmic reticulum Mt, mitochondria; pan class="Gene">Nuc, nucleus.

(B) Immunoblot analysis of post nuclear supernatant (PNS), micpan class="Chemical">rosome (Micro), cytosol (Cyto), mitochondria (Mito), and MAM fractions of stable VSMCs expressing GFP-DFCP1 co-treated with EBSS and BSA (Veh) or C18:0 (SFA) for 2 h.

(C) LC-MS-based absolute levels of LPA species of purified pan class="Gene">MAM fractions (n = 3, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

SFAs Increase pan class="Gene">GPAT4 Localized in Close Proximity to the Double-Positive Site of Tomm20 and DFCP1 Omegasome in VSMCs

(A) Confocal microscopy images of pan class="Gene">Tomm20 (magenta), GFP-DFCP1 (green), or GPAT4-tdTomato (GPAT4-T) (red) in VSMCs stably expressing GFP-DFCP1 and GPAT4-tdTomato treated with Veh or SFA under EBSS conditions for 2 h. Scale represents 10 μm. Arrow indicates Tomm20, GFP-DFCP1, and GPAT4-tdTomato contact sites.

(B) Number of GFP-DFCP1-pan class="Gene">GPAT4-Tomm20 contact sites (n = 16, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(C) Confocal microscopy images of pan class="Gene">Tomm20 (magenta), GFP-DFCP1 (green), and Lc3 (red) in stable VSMCs expressing GFP-DFCP1 co-treated with Veh, C18:0, or 18:0-LPA under EBSS conditions for 2 h. Scale represents 10 μm. Arrow indicates Tomm20, DFCP1, and Lc3-contact site.

(D) Number of GFP-DFCP1-pan class="Gene">Lc3-Tomm20 contact sites in Figure 7C (n = 20, One-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

Unsaturated LPAs Block SFA-Induced Abnormal Enlargement of DFCP1-Positive Omegasomes, Accumulation of Lipidated Lc3, and Calcification in VSMCs

UFAs attenuate pan class="Chemical">SFA-induced lipotoxicity (Listenberger et al., 2003, Masuda et al., 2012, Masuda et al., 2015). We examined whether co-treatment of unsaturated LPA (18:1-LPA) blocks SFA-induced abnormal formation of GFP-DFCP1 structures using confocal microscopy (Figures S7A and S7B). UFAs (C18:1) significantly blocked SFA-induced enlargement of DFCP1 membranes. Co-treatment with 18:1-LPA also reduced SFA-induced abnormally enlarged GFP-DFCP1 structures (Figures S7A and S7B). In contrast, fully unsaturated PAs (18:1/18:1-PA) did not influence SFA-induced enlargement of GFP-DFCP1 omegasomes. In the immunoblot analysis, C18:1 inhibited SFA-induced accumulation of Lc3b-II (Figure S7C). Similarly, 18:1-LPA also dose-dependently inhibited C18:0-induced Lc3b-II accumulation (Figure S7D). LPA is a signaling metabolite that works through lysophosphatidic acid receptors (LPARs) in plasma membranes. To study whether the effect of 18:1-LPA occurs via LPARs, VSMCs were treated with LPAR inhibitors such as Brp-LPA, AM095, and H2L5186303 in the presence of 18:1-LPA and C18:0. However, none of the LPAR inhibitors affected the effects of C18:0 and LPA on autophagy (Figure S7E). These results suggest that effects of SFAs and LPAs on the modulation of Lc3b-II are not dependent on LPA receptor signaling.

Gpat4-Generated Saturated LPAs Elicit Lipotoxic Effects In Vitro and In Vivo

We next examined whether 18:0-LPA mediates pan class="Chemical">SFA-induced lipotoxicities such as VSMC mineralization and apoptosis. Similar to C18:0, 18:0-LPA significantly increased levels of matrix calcium in VSMCs, whereas supplementation with 18:1-LPA, similar to C18:1, reduced mineralization (Figures 8A–8D) and levels of the cleavage (active) form of caspase 3 induced by C18:0 treatment (Figure 8E). To study SFA-mediated lipotoxicity through Gpat4-mediated saturated LPA production in vivo, we crossed SMC-Scd1/2 KO mice with Gpat4 KO mice to generate SMC-Scd1/2 KO; Gpat4 triple KO mice. Consistent with our previous report (Masuda et al., 2015), SMC-Scd1/2KO mice developed severe aortic medial calcification, vascular apoptosis, and accumulation of autophagic substrates such as p62 and Fam134b2, accompanied by significantly increased levels of saturated LPAs such as 16:0-LPA and 18:0-LPA (Figures 8F–8J). More importantly, Gpat4 deficiency completely blocked SCD deficiency-induced vascular calcification, apoptosis, and inhibition of autophagy (Figures 8F–8J). In addition, Gpat4 deficiency completely blocked accumulation of aortic saturated LPAs by SMC-SCD deficiency (Figure 8J).

Figure 8

Gpat4 Knockout Blocked Vascular Calcification in Scd1/2 Knockout Mice After 5/6 Nephrectomy

(A and B) (A) Mineralization and (B) ALP activity of VSMCs co-treated with 2.6 mM Pi and Veh or 18:0-LPA every 2 days for 6 days (n = 6, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(C) Alizarin red staining of VSMCs co-treated with Pi and Veh or 18:0-LPA for 6 days.

(D) Mineralization of VSMCs co-treated with 2.6 mM Pi and 200 μM C18:0 and BSA, 200 μM C18:1, or 20 μM 18:1-LPA every 2 days for 6 days (n = 6, one-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

(E) Immunoblot analysis of caspase 3 and Gapdh in VSMCs treated with 200 μM C18:0 and BSA, 200 μM C18:1, or 20 μM 18:1-LPA for 16 h. Arrow indicates cleaved (active) form of Caspase-3.

(F) Photograph (x10 magnification) of the lesions in aortic arches with von Kossa staining in SMC-Scd1/2 KO mice and SMC-Scd1/2 and Gpat4 triple KO mice. Scale represents 200 μm.

(G) Quantitative analysis of calcified lesions in Figure 8F analysis (n = 8, one-way ANOVA). p< 0.001. Data are represented as mean ± SEM.

(H) Levels of autophagy-specific substrates in the aortic medias of SMC-Scd1/2 KO; Gpat4 KO mice.

(I and J) (I) Percentage of aortic media apoptosis and (J) levels of LPA in the aortic medias of SMC-Scd1/2 KO; Gpat4 KO mice (n = 8, One-way ANOVA). ∗p < 0.05 versus control, #p < 0.05 versus SMC-Scd1/2 KO, and ∗p < 0.001. Data are represented as mean ± SEM.

(K) A model of SFA-induced vascular calcification through impairment of autophagosome maturation. Excess SFAs created through SCD inhibition and SFA supplementation are converted to saturated LPAs by GPAT4 at the MAM. Accumulation of saturated LPAs in MAM-associated omegasomes disrupts the proper balance of SFA-LPAs and UFA-LPAs, which causes abnormal formation of omegasomes, and excess production and accumulation of Lc3-II-positive isolation membranes, resulting in the inhibition of autophagic flux. Saturated LPA-mediated autophagy inhibition induces vascular apoptosis and calcification. G-3-P; grycerol-3-phosphate, LPA; lysophosphatidic acid.

Gpat4 Knockout Blocked pan class="Disease">Vascular Calcification in Scd1/2 Knockout Mice After 5/6 Nephrectomy

(A and B) (A) Mineralization and (B) ALP activity of VSMCs co-treated with 2.6 mM Pi and Veh or 18:0-pan class="Chemical">LPA every 2 days for 6 days (n = 6, 2-tailed Student's t test). ∗p < 0.05. Data are represented as mean ± SEM.

(C) Alizarin red staining of VSMCs co-treated with Pi and Veh or 18:0-pan class="Chemical">LPA for 6 days.

(D) Mineralization of VSMCs co-treated with 2.6 mM Pi and 200 μM C18:0 and BSA, 200 μM pan class="Chemical">C18:1, or 20 μM 18:1-LPA every 2 days for 6 days (n = 6, one-way ANOVA). ∗p < 0.05. Data are represented as mean ± SEM.

(E) Immunoblot analysis of caspase 3 and pan class="Gene">Gapdh in VSMCs treated with 200 μM C18:0 and BSA, 200 μM C18:1, or 20 μM 18:1-LPA for 16 h. Arrow indicates cleaved (active) form of Caspase-3.

(F) Photograph (x10 magnification) of the lesions in aortic arches with von Kossa staining in SMC-pan class="Gene">Scd1/2 KO mice and SMC-Scd1/2 and Gpat4 triple KO mice. Scale represents 200 μm.

(H) Levels of autophagy-specific substrates in the aortic medias of SMC-Scd1/2 KO; pan class="Gene">Gpat4 KO mice.

(I and J) (I) Percentage of aortic media apoptosis and (J) levels of LPA in the aortic medias of SMC-pan class="Gene">Scd1/2 KO; Gpat4 KO mice (n = 8, One-way ANOVA). ∗p < 0.05 versus control, #p < 0.05 versus SMC-Scd1/2 KO, and ∗p < 0.001. Data are represented as mean ± SEM.

(K) A model of SFA-induced pan class="Disease">vascular calcification through impairment of autophagosome maturation. Excess SFAs created through SCD inhibition and SFA supplementation are converted to saturated LPAs by GPAT4 at the MAM. Accumulation of saturated LPAs in MAM-associated omegasomes disrupts the proper balance of SFA-LPAs and UFA-LPAs, which causes abnormal formation of omegasomes, and excess production and accumulation of Lc3-II-positive isolation membranes, resulting in the inhibition of autophagic flux. Saturated LPA-mediated autophagy inhibition induces vascular apoptosis and calcification. G-3-P; grycerol-3-phosphate, LPA; lysophosphatidic acid.

Discussion

SFA-induced lipotoxicity is mediated by modifications of various pathways including apoptosis, ER stress, autophagy, oxidative stress, cell cycle pan class="Disease">inflammation, and membrane dynamics (Anderson et al., 2012, Brodeur et al., 2013, Holthuis and Menon, 2014, Hsiao et al., 2014, Listenberger et al., 2003, Masuda et al., 2012, Masuda et al., 2015, Mei et al., 2011, Park et al., 2019, Wen et al., 2011). We previously showed that inhibition of SCD induces aortic medial calcification by accumulating SFA-derived metabolites in VSMCs (Masuda et al., 2015). In addition, CKD increases circulating SFAs in humans (Masuda et al., 2015). These results suggest that SFA-induced lipotoxicity is a major cause of vascular calcification in CKD. However, which lipotoxic pathway is responsible for the development of vascular calcification is not completely understood. The autophagy-lysosomal degradation pathway is a protective mechanism in various organs against diseases such as neurodegenerative diseases, liver diseases, muscular disorders, and pathogen infections (Dikic and Elazar, 2018, Leidal et al., 2018, Maiuri and Kroemer, 2019, Ren and Zhang, 2018). In this study we showed direct evidence that autophagy in VSMCs plays a protective role in the pathogenesis of vascular calcification in CKD. CKD and SCD deficiency induced significant accumulation of aortic proteins (Lc3b-II, p62 and Fam134b2) specifically degraded through autophagy. More strikingly, SMC-specific Atg5 deficiency, and therefore inhibition of VSMC autophagy, strongly induces vascular calcification and apoptosis, resulting in significant increases in mortality of CKD mice. These results demonstrate that SFA-mediated autophagy inhibition plays a causative role in vascular calcification.

Due to the lack of detailed autophagic flux and mechanistic studies, the effect of SFAs and pan class="Gene">SCD on autophagy regulation is controversial (Khan et al., 2012, Liu et al., 2015, Tan et al., 2012, Tan et al., 2014). Most of the conclusions in the previous studies were made based on the results of immunoblot analysis and/or confocal microscopic analysis with LC3B. We therefore made extensive efforts to elucidate the effect of SFAs on autophagic flux and the precise mechanism by which SFAs modulate autophagy machinery. Although VSMCs were mainly used in this study, the mechanisms underlying the modulation of autophagy by SFAs are very similar in any cell type as SFAs also elicit the inhibitory effect of autophagy in two other common cell lines, HeLa and NIH-3T3. In this study, we conclude that SFAs such as C18:0 and C16:0 and the inhibition of SCD both inhibit autophagic flux based on the study using the GFP-LC3-RFP-LC3ΔG probe that enabled us to most accurately analyze it (Kaizuka et al., 2016, Yoshii and Mizushima, 2017). We therefore conclude that the accumulation of Lc3b-II and p62 in cells treated with SFAs and in the aortic media of SMC-Scd1/2 KO mice is due to SFA-induced autophagy inhibition. More importantly, in this study we have elucidated the precise mechanism by which SFAs mediate the inhibition of autophagy and demonstrate that SFA accumulation induces abnormal enlargement of MAM-originated omegasomes, which leads to accumulation of Lc3b-II positive isolation membranes on omegasomes.

We previously reported that GPAT4 is one of the enzymes in pan class="Chemical">SFA-induced lipotoxicity and vascular calcification (Masuda et al., 2015), which has also been shown by two recent studies (Piccolis et al., 2019, Zhu et al., 2019). In addition, we identified that fully saturated PAs produced by the GPAT4-AGPAT3/5 reaction mediate SFA-induced ER stress However, which SFA metabolite mediates the inhibitory effect of SFAs in autophagy has not been studied. Although SFA-induced ER stress was blocked by the inhibition of three enzymes, Gpat4, Agpat3, and Agpat5 (Masuda et al., 2015), SFA-induced autophagy inhibition was inhibited only by Gpat4 inhibition. Interestingly, Agpat3 blockage augments SFA-induced Lc3b accumulation. SFA treatment drastically induces accumulation of saturated LPAs (Masuda et al., 2015). Treatment with exogenous 18:0-LPA, a saturated LPA, but not fully saturated PAs such as 18:0/18:0-PA, replicates SFA-induced autophagy inhibition including the inhibition of autophagic flux and the generation of abnormally enlarged omegasomes, suggesting that saturated LPAs are specific SFA metabolites that affect omegasome formation and autophagic flux under excess SFAs. Furthermore, co-treatment with unsaturated LPAs such as 18:1-LPA, but not unsaturated PAs such as 18:1/18:1-PA, attenuates SFA-induced autophagy inhibition. These data demonstrate that Gpat4 converts SFAs to saturated LPAs, disrupting the proper balance of SFA-LPAs and UFA-LPAs (ie, 18:0-LPA/18:1-LPA), causing abnormal enlargement of omegasomes. LPAs containing UFAs are a termination factor for the enlargement of omegasomes. Another interesting finding in this study is that Gpat4 is located in the contact sites between abnormally enlarged omegasomes and the MAM, which are increased by SFAs. The sources of isolation membranes have not been fully defined yet, but they possibly come from the ER, mitochondria, and Golgi, recycling endosomes and the plasma membrane through Atg9-mediated vesicle transport (Hailey et al., 2010, Lamb et al., 2013, Mari et al., 2010, Ohashi and Munro, 2010, Puri et al., 2013, Ravikumar et al., 2010, Yamamoto et al., 2012). In addition, the ER-mitochondria contact site (the MAM) is a major site of autophagosome production (Hamasaki et al., 2013). Furthermore over 85% of autophagosome generation sites (Lc3 positive) are closely co-localized with GPAT4 in the cells treated with C18:0 (Figures S6C–S6E). The ER, mitochondria, and contact sites could directly supply lipids to omegasomes and isolation membranes. In fact, SFA treatment induces accumulation of saturated LPAs in the MAM and DFCP1-enriched membranes. The source of omegasome membrane lipids is also not clear, but our data suggest that glycerophospholipids such as LPAs synthesized on the MAM are also used as a major omegasome lipid source under excess SFAs. Gpat4 on the MAM supplies saturated LPAs, which causes the abnormal enlargement of omegasomes. In addition, exogenous 18:0-LPA induces Lc3 accumulation at the contact site of omegasomes and the MAM (Figure 7). Although several factors such as PI3P, EPG-6, CapZ, and DFCP1 that influence omegasome formation have been identified (Axe et al., 2008, Lu et al., 2011, Mi et al., 2015, Mochizuki et al., 2013), LPAs such as 18:0-LPA are critical lipid metabolites that affect omegasome formation.

In conclusion, our findings indicate that saturated LPAs could perturb the autophagy machinery. As illustrated in our model (Figure 8K), we speculate that pan class="Gene">Gpat4 localizes on the MAM, utilizing excess SFAs by SCD inhibition and increases SFA intake and converts them into saturated LPAs, which are used as a lipid source for omegasome formation. However, excess production of saturated LPAs such as 18:0-LPA unbalances SFA-LPAs and UFA-LPAs on omegasomes, causing abnormal enlargement. Abnormally enlarged omegasomes containing excess saturated LPAs produce and accumulate isolation membranes, block autophagic flux, and cause vascular apoptosis and calcification. We previously reported that fully saturated PAs (i.e. 18:0/18:0-PA) generated through the Gpat4-Agpat3/5 reaction are responsible for SFA-mediated ER stress (Masuda et al., 2015). These results suggest that each SFA metabolite mediates a distinct pathway of SFA lipotoxicity. Taken together, Gpat4 could be a major therapeutic target for the treatment of lipotoxicity-induced chronic diseases such as vascular calcification.

Limitations of the Study

Although artifacts might arise due to the EM sample preparation, abnormal enlarged omegasome structures were only observed in cells treated with SFAs or pan class="Chemical">SFA-LPAs but not vehicles. To ensure the results, the experiments have been at least triplicated. In addition, this study has revealed that GPAT4-generated SFA-LPAs indirectly affect autophagy through the generation of abnormal omegasomes. Follow-up studies will be required to determine whether GPAT4 and SFA-LPAs directly affect autophagosome formation.

Resource Availability

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Makoto Miyazaki (makoto.miyazaki@cuanschutz.edu).

Materials Availability

All materilas used and generated for this study are included in the published article and its supplementary information files or are available from the lead contact upon request.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.