Coupling of cargo to the autophagy receptor is a critical step in ER-phagy Authors: Shuliang Chen, Subhrajit Banerjee, Dongmei Liu, Kamal Kumar, Christopher J Obara, Peter Novick, William A. Prinz, Susan Ferro-Novick Published: 2026 ABSTRACT During cell stress, endoplasmic reticulum autophagy (ER-phagy) receptors remodel the ER by sequestering membrane proteins (cargo) into autophagosomes for degradation. The conserved ER-phagy receptor, Atg40, contains a motif that binds to Atg8 and a reticulon homology domain that is needed for vacuolar/lysosomal delivery. Cargo capture, however, requires the Atg40 binding partner Lst1/SEC24C. To address whether lipids regulate cargo capture during ER-phagy, we analyzed autophagy in neutral lipid–deficient cells. Unexpectedly, we found that Atg40 was delivered to the vacuole in autophagosomes without Lst1/SEC24C or cargo in mutant cells. Lipidomic analysis revealed changes in the ratio of phosphatidylethanolamine to phosphatidylcholine in the neutral lipid–deficient cells that are predicted to alter ER membrane bendability. Our findings imply that phospholipids control cargo sequestration by regulating receptor-cargo coupling at autophagic sites. INTRODUCTION Endoplasmic reticulum autophagy (ER-phagy) is a selective autophagy pathway that remodels the ER to maintain cellular homeostasis. Damaged ER domains and misfolded proteins that fail to be degraded by ER-associated degradation (ERAD) are removed from the ER by autophagy (1, 2). During autophagosome-mediated ER-phagy, ER domains are sequestered into sealed double-membrane structures called autophagosomes that are delivered to the vacuole (yeast) or lysosome (mammals) for degradation (1–3). Selective autophagy pathways use autophagy receptors that bind to lipidated Atg8 to link cargo, targeted for degradation, to the autophagy machinery (4). Although several ER-phagy receptors have been reported in mammalian cells, only one conserved ER-phagy receptor (Atg40) and one nucleophagy receptor (Atg39) have been identified in the yeast Saccharomyces cerevisiae (5). Atg40 resides in tubules and the curved edges of sheets that line the cell cortex (5) and inserts into membranes via a reticulon homology domain (RHD) that facilitates membrane bending (5, 6). When autophagy is induced with the Tor kinase inhibitor rapamycin, Atg40 expression is up-regulated, and Atg40-containing puncta accumulate in the ER (5). These puncta associate with Atg8 to traffic ER membrane proteins (cargo) to the vacuole for degradation (5). Studies in yeast identified the Coat Protein Complex II (COPII)coat subunit, Lst1, as a binding partner of Atg40 (7). Lst1, which forms a stable complex with Sec23, is a dual functioning protein; it acts in secretion and in ER-phagy with Atg40 (7). On the secretory pathway, Lst1-Sec23 sorts proteins into COPII-coated transport carriers that traffic to the Golgi (8). In rapamycin-treated cells, Atg40 recruits the Lst1-Sec23 complex from the cytosol to the Atg8-marked perivacuolar autophagosome assembly site (called ER-phagy site or ERPHS). The role of Lst1 at these sites is currently unclear. In mammalian cells, the homolog of Lst1, SEC24C, works in an analogous manner with the ER-phagy receptor Reticulon-3 Long isoform (RTN3L) (9). The RTN3L-SEC24C receptor complex targets disease-causing, ERAD-resistant, misfolded proteins for degradation at defined ER domains, peripheral tubule junctions, sites in the ER where two tubules have fused (9, 10). Most cargoes that are degraded by selective autophagosome-mediated autophagy (herein called autophagy) are small enough to be captured within autophagosomes (4). However, because the ER is too large to be sequestered inside an autophagosome, resident ER membrane proteins that undergo ER-phagy must partition into an ER domain that then severs from the network (3). Lipids likely play a critical role in the ER remodeling events needed for cargo capture during ER-phagy. However, it is difficult to assess how ER-phagy–specific functions are affected by changes in ER lipid composition since many of the approaches used to modify lipids also block autophagosome formation (11, 12). As recent studies have revealed that phospholipids can be altered in rapamycin-treated neutral lipid–deficient mutant cells without disrupting autophagosome formation (13), we used this mutant to address whether and how lipids regulate cargo capture during ER-phagy. The use of this mutant has provided us with the opportunity to disrupt ER-phagy without affecting general autophagy. RESULTS ER-phagy is blocked in neutral lipid–deficient cells To ask whether the loss of neutral lipids blocks ER-phagy, we deleted the four genes required for their synthesis in the same strain: DGA1, LRO1, ARE1, and ARE2 (table S1) (14, 15). Dga1 and Lro1 synthesize triacylglycerol (TAG), whereas Are1 and Are2 produce steryl esters (SEs). Neutral lipid–deficient cells are depleted of lipid droplets, storage organelles that maintain ER homeostasis by releasing neutral lipids in response to cell stress (14, 15). We initially assessed the effects of the loss of the neutral lipids TAG and SE on ER-phagy using the ER translocon protein Sec63 in a ΔTAGΔSE mutant that we will herein refer to as ΔLD. Sec63–green fluorescent protein (GFP) delivery to the vacuole was assayed by fluorescence, 16 hours after rapamycin treatment (fig. S1A), and the vacuolar cleavage of Sec63-GFP to GFP was monitored by Western blot analysis (Fig. 1A). Whereas no defect was observed in ΔTAG or ΔSE cells, delivery was impaired in ΔLD cells that lacked all four genes (Fig. 1A and fig. S1A). This defect was not an indirect consequence of altering ER structure, as high-resolution fluorescence microscopy revealed that the morphology of the ER was unaffected in rapamycin-treated mutant cells (fig. S1B). Consistent with this conclusion, increasing ER tubulation by overexpressing the RHD-containing ER protein, Pex30 (16), failed to restore ER-phagy in ΔLD cells (fig. S1C). The vacuolar delivery of three other cargoes, the reticulon protein Rtn1 (Fig. 1B and fig. S2A), Sec66 (Fig. 1C, left), and Per33 (Fig. 1C, right), was also severely disrupted in neutral lipid–deficient cells. In addition to the cortical ER-phagy defect, vacuolar delivery of the nuclear ER marker, Hmg1, was disrupted (fig. S3A). ER-phagy, but not bulk autophagy, is blocked in ΔLD mutant cells. (A) Left: Cells expressing Sec63-GFP were cultured in SD medium with rapamycin (Rapa) for 0 and 16 hours at 30°C. Lysates were prepared as described in Materials and Methods, and the cleaved GFP product was analyzed by Western blot analysis. Right: Quantification of the percentage of cleaved GFP divided by total GFP (free GFP plus Sec63-GFP) for the 16-hour time point. The data were normalized to the WT control (100%). (B) Left: Cells expressing Rtn1-GFP were cultured, processed, and quantitated (right) as in (A). (C) Left: Cleavage and quantitation of Sec66-GFP and cleavage and quantitation of Per33-GFP (right) 16 hours after rapamycin treatment. (D) Left: The cleavage of GFP-Atg8 to GFP was assessed by Western blot analysis following rapamycin treatment. Right: Quantification of the percentage of free GFP divided by total GFP was assessed. The data were normalized to the WT control (100%). Error bars represent the SEM, N = 3. Asterisks denote statistical significance. NS, not significant (P ≥ 0.05); ****P < 0.0001, Student’s unpaired t test. To confirm that bulk autophagic flux is not disrupted in the ΔLD mutant, we compared the cleavage of GFP-Atg8 in wild-type (WT), atg1Δ, and ΔLD cells that were treated with rapamycin for 8 to 24 hours. While the loss of the Atg1 kinase, which is required for autophagosome formation (17), blocked the delivery of GFP-Atg8 to the vacuole (Fig. 1D), GFP-Atg8 cleavage was unaltered at all time points examined in ΔLD cells (Fig. 1D). Next, we measured autophagosome number to confirm that autophagosome formation was not affected in our experiments. Autophagosome number can be measured by transmission electron microscopy (TEM) or by quantitating GFP-Atg8 puncta in ypt7Δ cells. Ypt7 is required for the fusion of autophagosomes with the vacuole, and in its absence, mature autophagosomes accumulate in the cytoplasm (18). For our studies, autophagosome number was assessed using the latter approach as organelle integrity was not preserved after rapamycin-treated ΔLD mutant cells were fixed for TEM. Consistent with autophagic flux measurements, we did not see a significant change in GFP-Atg8 puncta number in ΔLD cells when compared to WT (fig. S2B). When autophagosomal structures were examined in more detail by expanding GFP-Atg8 marked preautophagosomal membranes on a giant complex formed by aminopeptidase I (Ape1) overexpression, the membranes appeared similar in WT and the ΔLD mutant (fig. S2C). These findings corroborate previous studies showing that autophagosome formation is not disrupted during bulk autophagy in rapamycin-treated neutral lipid–deficient cells (13). Atg40 is sequestered in autophagosomes without cargo in ΔLD cells When we examined the trafficking of the RHD-containing ER-phagy receptor Atg40 in rapamycin-treated neutral lipid–deficient cells, we did not expect to find that Atg40-GFP was delivered to the vacuole as efficiently as in WT (fig. S2D). In contrast, the vacuolar delivery of the single-pass transmembrane nucleophagy receptor, Atg39, was blocked (fig. S3B) (5). Atg40 delivery to the vacuole was dependent on Atg1 and Atg8, indicating that it is autophagosome-mediated (Fig. 2A and fig. S4A). Trafficking of Atg40 also required interaction with Atg8, as mutations in its AIM (Atg8 interacting motif), which are known to block binding (Fig. 2B, top) (5), abrogated vacuolar delivery (Fig. 2B, bottom, and fig. S4B). Together, these findings indicate that Atg40, but not cargo, is sequestered into autophagosomes in ΔLD cells. Atg40 traffics to the vacuole without cargo in rapamycin-treated ΔLD cells. (A) Left: The cleavage of Atg40-GFP was examined by Western blot analysis 16 hours after rapamycin treatment. Quantitation is on the right. The data were normalized to the WT control. (B) Top left: Schematic diagram showing the topology of Atg40-GFP. The HR (hydrophobic region), AIM, and AIM mutations (Atg40AIMmut) used to disrupt the Atg40-Atg8 interaction are shown. Top right: Membrane topology of Atg40. Bottom left: Atg40-GFP cleavage was examined by Western blot analysis as in (A) and quantitated on the bottom right. The data were normalized to the WT control. (C) Top right: Quantification of GFP-Atg8 puncta associated with Rtn1-mCherry puncta per cell at 0 and 3 hours of rapamycin treatment; representative images are shown on the left. At 0 hours, 1536 WT and 1668 ΔLD cells were quantitated. At 3 hours, 2250 WT and 2285 ΔLD cells were quantitated. Atg8 (arrowheads) and Rtn1 (arrows) puncta associated with each other are marked. Bottom right shows the median peak-to-peak distance between Atg8 and Rtn1 fluorescence emissions from 202 Atg8-Rtn1 associated puncta in ΔLD cells versus the control, an estimation of the comparison of Atg8 to itself (see Materials and Methods). Error bars represent SEM, N = 3. NS, not significant (P ≥ 0.05); **P < 0.01; ****P < 0.0001, Student’s unpaired t test. Scale bar, 2 μm (C). To ask where cargo delivery is blocked in the mutant, we imaged Rtn1-mCherry. Rtn1 was found in small highly concentrated structures (puncta) that were spatially associated with perivacuolar Atg8 puncta (Fig. 2C, left). We noticed a nearly twofold increase in the accumulation of Rtn1 puncta proximal to Atg8 in neutral lipid–deficient cells (Fig. 2C, top right), suggesting that cargo accumulated at or near autophagosome assembly sites. To address whether cargo capture into the ERPHS is disrupted in the neutral lipid–deficient mutant, we measured the peak-peak distance between Atg8 and Rtn1 fluorescence emissions. Notably, the distance between Atg8 and Rtn1 was found to be significantly greater than the error measured when Atg8 was compared to itself (see Materials and Methods for details), suggesting physical displacement of the Rtn1 puncta from the Atg8 signal (Fig. 2C, bottom right). These findings imply that cargo does not enter the ERPHS in the ΔLD mutant. To start to understand why Atg40, but not Atg39, traffics to the vacuole in ΔLD cells, we asked whether the AIM and RHD are both needed for vacuolar delivery. The C-terminal region of Atg40 that contains the AIM (Atg40Ct, amino acids 194 to 256) was fused to the C terminus of the RHD-containing protein Rtn1 to generate Rtn1-Atg40Ct-GFP (fig. S5A, top; see location of AIM). We found that the presence of the AIM increased the vacuolar trafficking of Rtn1-Atg40Ct-GFP in WT and neutral lipid–deficient cells, largely bypassing the need for Atg40 (fig. S5, B and C; fig. S5C, lanes 4 to 6, compare with lanes 1 to 3 and lanes 7 to 9). These observations are consistent with previous studies showing that Rtn1-Atg40Ct is delivered to the vacuole in rapamycin-treated WT cells that lack Atg40 (19). In contrast to Rtn1, fusing the Atg40 AIM to the single-pass transmembrane protein Sec66 (Sec66-Atg40Ct; fig. S5D, top) did not enable Sec66-Atg40Ct to traffic to the vacuole in ΔLD cells (fig. S5D, bottom), indicating that an AIM and RHD are both needed for vacuolar delivery. Lst1/SEC24C is required for receptor-cargo coupling When we monitored the trafficking of Lst1 to the vacuole in neutral lipid–deficient cells, we found a block in delivery that was comparable to atg40Δ cells (Fig. 3, A and B, and fig. S4C). Like Rtn1, an increase in Lst1 puncta proximal to Atg8 was observed in ΔLD cells (Fig. 3C, top right), suggesting that Lst1 accumulated at or near autophagosome assembly sites. We noted that the distance between the peaks of Atg8 and Lst1 fluorescence emissions in the mutant was significantly larger than when Atg8 was compared to itself (Fig. 3C, bottom right), implying that Lst1 accumulates at a site proximal to the ERPHS. A modest increase in Atg8-associated Atg40 (fig. S4D) puncta was also observed in the ΔLD mutant. However, the Atg40 fluorescence signal was too weak to reliably measure the distance between Atg8 and Atg40. The Atg40 binding partner, Lst1, fails to traffic to the vacuole in ΔLD cells. (A) Left: The cleavage of Lst1-GFP was examined in ΔLD cells. Right: Quantification of the percentage of cleaved GFP. The data were normalized to the WT control. (B) Quantification of Lst1-GFP in the vacuole 16 hours after rapamycin treatment for the images shown in fig. S4C. A total of 1142 WT, 951 atg40Δ, and 984 ΔLD cells were examined. (C) Top right: Quantification of GFP-Atg8 puncta associated with Lst1–3× mCherry puncta per cell at 0 and 3 hours of rapamycin treatment; representative images are shown on the left. At 0 hours, 2594 WT and 2091 ΔLD cells were examined. At 3 hours, 2083 WT and 2061 ΔLD cells were examined. Atg8 (arrowheads) and Lst1 (arrows) puncta associated with each other are marked. Bottom right shows the median peak-to-peak distance between Atg8 and Lst1 fluorescence emissions from 211 Atg8-Lst1 associated puncta in ΔLD cells versus the control, an estimation of the comparison of Atg8 to itself (see Materials and Methods). (D) Quantification of Rtn1-GFP in the vacuole 16 hours after rapamycin treatment for the images shown in fig. S7B. A total of 688 WT and 558 lst1Δ cells were examined. (E) Top: Cells expressing Atg40-GFP, under the Rtn1 promoter, were cultured in SD media supplemented with rapamycin for 6 hours at 30°C. Atg40-GFP recovery was assessed by FRAP analysis as described in Materials and Methods. Arrowheads mark the regions that were photobleached. Bottom left: Fluorescence intensity, normalized to the prebleach value, was plotted over time. Bottom right: The half time of fluorescence recovery (t½) after photobleaching for the data shown on the left. A total of 52 WT and 57 ΔLD cells were examined. Error bars represent SEM, N = 3. **P < 0.01; ***P < 0.001; ****P < 0.0001, Student’s unpaired t test. Scale bars, 2 μm [(C) and (E)]. Although Rtn1-Atg40Ct trafficked to the vacuole in the absence of Atg40, it could not restore ER-phagy (see Sec63-GFP cleavage in fig. S6A, left) or the vacuolar delivery of Lst1 (fig. S6A, right) in the atg40Δ mutant (see lanes 2 to 5 and graphs below). Unlike Atg40, Rtn1-Atg40Ct did not bind to Lst1 (fig. S6B, right lane 3, compare to input and Atg40 binding in fig. S6B, left), suggesting that it failed to restore ER-phagy because it could not interact with Lst1. These findings indicate that an AIM and RHD are not sufficient for ER-phagy, and other machinery, such as Lst1, may be needed. Although Lst1 delivery to the vacuole was dependent on Atg40 (Fig. 3, A and B, and fig. S4C), a significant fraction of Atg40-GFP (~73% of WT) trafficked to the vacuole in lst1Δ cells (fig. S7A), indicating that Lst1 facilitates, but is not essential for the sequestration of Atg40 into autophagosomes. Consistent with previous studies showing that Lst1 is needed for ER-phagy (7), disengaging Atg40 from the Lst1-Sec23 complex in the lst1Δ mutant disrupted the vacuolar delivery of Rtn1-GFP (Fig. 3D and fig. S7B). In total, these findings demonstrate that the coupling of cargo to Atg40 requires the Lst1-Sec23 complex, whereas Atg40 delivery to the vacuole occurs largely in the absence of Lst1. As the oligomerization of RHD-containing ER-phagy receptors has been reported to drive cargo capture and membrane fission during ER-phagy (19), we asked whether Atg40 still oligomerizes in the ΔLD mutant. RHD-containing oligomers were visualized by irreversibly cross-linking intact membranes with the cross-linker ethylene glycol bis(succinimidyl succinate) (EGS). Atg40 oligomers were found to form equally as efficiently in rapamycin-treated mutant membranes when compared to control membranes (fig. S7C), and only a small decrease in the conversion of monomer to oligomers was observed at high concentrations of cross-linker with mutant membranes (fig. S7C, right). These findings show that RHD-mediated oligomerization and binding to Atg8 are sufficient for Atg40’s vacuolar delivery, but not for the capture of ER membrane proteins (cargo) into autophagosomes. Lipids regulate the coupling of cargo to the autophagy receptor during ER-phagy The finding that Atg40 is delivered to the vacuole in ΔLD mutant cells without cargo or Lst1 suggested that bulk ER membrane properties may regulate Atg40’s ability to function as a cargo receptor. To begin to understand how Atg40 behaves in the mutant, fluorescence recovery after photobleaching (FRAP) analysis was performed to assess its diffusional mobility in ΔLD cells (Fig. 3E). As a control for these studies, we analyzed the multipass transmembrane protein, Sec61 (fig. S8A), as it is an Atg40 cargo (fig. S8B). When we photobleached a small region of Atg40-GFP on the cortical ER, fluorescence recovery was significantly slower in the mutant when compared to WT (Fig. 3E). This observation suggests that most Atg40 molecules move slower in ΔLD ER membranes. In contrast to Atg40, the diffusion of Sec61-GFP did not significantly decrease (fig. S8A). To better understand how bulk membrane properties are altered in the ΔLD mutant, we analyzed the lipidome of WT and the mutant in the absence and presence of rapamycin. A complete analysis of all glycerolipids, the major lipids in the ER, is reported in tables S2 and S3. The phospholipid composition of Sec63–3× hemagglutinin (HA)–tagged immunopurified ER membranes (fig. S9, A and B) was analyzed at 0, 3, and 6 hours of rapamycin treatment (table S2), and whole cells were examined 16 hours after treatment (table S3). Similar results were obtained with both preparations. As shown in Fig. 4A, phosphatidylcholine (PC) and phosphatidylethanolamine (PE) were the most abundant phospholipids in our ER membrane preparations, whereas phosphatidylinositol (PI) and phosphatidic acid (PA) were barely detectable, and phosphatidylserine (PS) was undetectable. The induction of ER-phagy with rapamycin led to a significant increase in PC at 6 hours (Fig. 4A) and a decrease in the ratio of PE to PC (Fig. 4B) in ΔLD, but not WT, ER membranes. A change in the unsaturation index of phospholipid acyl chains in ΔLD ER membranes was also observed (fig. S9, C to E). However, this change is unlikely to be relevant to ER-phagy as it was not dependent on rapamycin (fig. S9C) and not observed 16 hours after rapamycin treatment (compare fig. S9C with fig. S10D). Lysophosphatidylcholine (Lyso-PC) and lysophosphatidylinositol (Lyso-PI), which are enriched in ER-derived COPII vesicles (20), were not significantly different in the mutant and WT (figs. S9F and S10A). Therefore, the only observed significant change that was rapamycin-dependent in the mutant was the increase in PC and the decrease in the ratio of PE:PC. Changes in the ratio of PE:PC affect biophysical properties of membranes such as bendability, which is influenced by membrane fluidity (21–23). ER-phagy and Atg40 mobility are restored in the ΔLD mutant at 20°C. (A) LC-MS analysis of ER membrane lipids was performed as described in Materials and Methods. Cells were incubated in SD media at 30°C with or without rapamycin for this analysis. Each bar shows the indicated level of phospholipid as a relative percent of total glycerophospholipids that were obtained from the LC-MS analysis. (B) The PE:PC ratio was calculated for the indicated time points from the data shown in (A). (C and D) Cells were grown overnight at 20°C, and the vacuolar delivery of Rtn1-GFP (C) or Lst1-GFP (D) was assessed by GFP cleavage 16 hours after rapamycin treatment at 20°C. (E) FRAP analysis was performed as in Fig. 3E, except that the cells were grown overnight at 20°C and treated for 6 hours with rapamycin at 20°C. A total of 52 WT and 57 ΔLD cells were examined. Error bars represent SEM, N = 3. NS, not significant (P ≥ 0.05); *P < 0.05; **P < 0.01; ****P < 0.0001, Student’s unpaired t test. As the ratio of PE:PC in membranes has been reported to increase at low temperatures (21), we asked whether the ER-phagy defect in ΔLD cells could be suppressed at 20°C. We did not observe a significant ER-phagy defect in the LD mutant at 20°C when either Rtn1-GFP (Fig. 4C) or Sec63-GFP (fig. S10B) degradation was examined. Furthermore, the delivery of Lst1-GFP to the vacuole was also restored at 20°C (Fig. 4D). Consistent with the suppression of the ER-phagy defect, FRAP analysis revealed that the diffusional mobility of Atg40-GFP was restored in the ΔLD mutant at 20°C (Fig. 4E and fig. S10C). To ask whether the suppression of the ER-phagy defect in the ΔLD mutant was the consequence of reestablishing the ratio of PE:PC in mutant membranes, we compared the lipidome of WT, atg40Δ, and neutral lipid–deficient cells at 30° to 20°C 16 hours after rapamycin treatment. atg40Δ was examined as a control in this analysis to rule out the possibility that the lipid changes we observed are an indirect consequence of blocking ER-phagy. At 30°C, 16 hours of rapamycin treatment yielded similar results to 6 hours. PC levels dramatically increased (Fig. 5A, left, compare to Fig. 4A), whereas PE levels only modestly increased in ΔLD, but not atg40Δ cells (Fig. 5B, left). Overall, the dramatic increase in PC resulted in a significant decrease in the PE:PC ratio in the ΔLD mutant at 30°C (Fig. 5C, left). In sharp contrast, PC (Fig. 5A, right) and PE (Fig. 5B, right) levels were unaltered at 20°C, and the PE:PC ratio was the same in ΔLD cells and WT (Fig. 5C, right). In total, these findings indicate that the suppression of the ER-phagy defect in the ΔLD mutant at 20°C is the consequence of balancing the ratio of PE:PC in the ER. The balance of PE:PC in the ER regulates Atg40’s ability to sequester cargo. (A) PC levels were quantitated from WT, atg40Δ, and ΔLD cells that were treated with rapamycin for 16 hours at either 30° or 20°C as described in Materials and Methods. (B) Same as (A), only that PE levels were measured. (C) The PE:PC ratio was calculated from the data shown in (A) and (B). Error bars represent SEM, N = 4. NS, not significant (P ≥ 0.05); *P < 0.05; **P < 0.01; ***P < 0.001, one-way ANOVA. (D) Receptor-cargo coupling is defective in rapamycin-treated ΔLD cells. When ER-phagy is induced, Atg40 binds to Atg8 and recruits the Lst1-Sec23 complex to negatively curved membranes (left) to facilitate the recognition and concentration of cargo (ER membrane proteins) into autophagic sites (ERPHS) (left) (7). In the ΔLD mutant (right), at 30°C, a decrease in the ratio of PE:PC in the ER alters the bendability and fluidity of the membrane (right). As a consequence, Atg40 still enters the ERPHS via its ability to bind to Atg8; however, cargo concentration into ERPHS is impeded, and the receptor is delivered to the vacuole without cargo or Lst1 (right). DISCUSSION Autophagy receptors are defined by their ability to recognize cargo and link it to the autophagy machinery via their AIM (4). Although the conserved ER-phagy receptor Atg40 contains an AIM that binds to autophagosomal membranes and an RHD that mediates membrane fission, these domains are not sufficient for cargo capture into autophagosomes. Here, we show that when ER-phagy is induced, receptor-cargo coupling is disrupted in the ΔLD and lst1Δ mutants. Our analysis of the phospholipid composition of rapamycin-treated ΔLD cells implies that the defect in ER-phagy is caused by increased levels of the cylindrical lipid PC that lead to a decrease in the ratio of PE:PC in the ER. Based on theoretical modeling (Martini modeling) studies, the decreased level of PE:PC in the ΔLD mutant is predicted to decrease ER membrane bendability (23). Membrane bending is believed to create a physical force that facilitates cargo concentration and the partitioning of proteins into highly curved membranes (24). As cargo capture and membrane bending are closely linked to each other, we propose that the defect in cargo capture in ΔLD cells is the consequence of the decreased ratio of PE:PC in the ER. Coat proteins, on secretory and endocytic pathways, recognize sorting motifs in transmembrane cargoes and concentrate them into transport carriers (25–27). On the secretory pathway, COPII vesicle cargoes are thought to enter transport vesicles from negatively curved membranes that form at the rim of ER exit sites (20). By analogy, ER transmembrane cargoes that undergo ER-phagy may enter autophagic sites (ERPHS) from negatively curved membranes (Fig. 5D, left). We propose that the decrease in ER bendability in the ΔLD mutant leads to a less curved membrane that cannot recruit Lst1 and cargo into an ERPHS at an ER-autophagosomal membrane contact site (Fig. 5D, compare the curvature of the membrane on the right and left). Atg40 enters the ERPHS in the mutant via its ability to bind to Atg8, while Lst1 and cargo accumulate next to the ERPHS and fail to be sequestered within autophagosomes (Fig. 5D, right). Membrane curvature, which is likely determined in part by membrane deformability, may also play an important role in mammalian ER-phagy. Recent studies have revealed that SEC24C associates with RTN3L at ER junctions (10), domains in ER tubules with increased negative curvature (28, 29). The formation of the RTN3L-SEC24C receptor complex at junctions appears to restrict ER-phagy to this site (10). Misfolded ERAD-resistant proteins, which are degraded by RTN3L-SEC24C-mediated ER-phagy, also accumulate at junctions (10). In total, our studies imply that lipids regulate receptor-cargo coupling to restrict ER-phagy to defined sites in ER tubules. These findings could have important implications for the therapeutic treatment of neurological and metabolic disorders that are associated with neutral lipid defects (15). In the future, it will be important to address whether other ER-phagy receptors function with SEC24C. Although the ER-phagy receptor FAM134B does not interact with SEC24C, it binds to SEC24C-containing COPII vesicles that harbor misfolded ERAD-resistant alpha-1-antitrypsin, and targets these vesicles to lysosomes for degradation (30, 31). Thus, SEC24C may play multiple roles in cargo capture during ER-phagy. MATERIALS AND METHODS Experimental design Strains and growth conditions Yeast strains used in this study are described in table S1. Unless otherwise stated, cell growth in liquid culture was monitored by measuring optical density at 600 nm (OD600). Yeast deletion mutants were generated using the Longtine toolbox (32) in combination with tetrad dissection. Strains expressing tagged fusion proteins were constructed using integrating, centromere (CEN), or 2-μm vectors. ER-phagy was induced with 200 to 400 ng/ml of rapamycin using a protocol described before (33). Briefly, yeast cells expressing GFP-tagged ER markers were cultured to exponential phase (OD600 = 0.2) in synthetic complete medium that was supplemented with 2% glucose (SD) and amino acids. Subsequently, rapamycin was added to the culture media, and the cells were incubated at 30°C for the times indicated in the legends and figures. To assess bulk autophagy, cells expressing GFP-Atg8 were grown to OD600 = 0.5 and then shifted to synthetic media containing 2% glucose without amino acids and ammonium sulfate (SD-N media). Giant prApe1 complexes and phagophore formation was induced as described before (34), using pYEXBX[prApeI], a 2-μm plasmid that expresses prApe1 from the CUP1 promoter. Briefly, cells harboring GFP-Atg8, red fluorescent protein–prApe1, and pYEXBX[prApeI] were cultured overnight in SDCA media (0.17% yeast nitrogen base without amino acids and ammonium sulfate, 0.5% casamino acids, 0.5% ammonium sulfate, and 2% glucose). Cells were then reinoculated into fresh SDCA media containing 250 μM CuSO4 and cultured for 24 hours to exponential phase. Subsequently, autophagy was induced with rapamycin during a 3-hour incubation. Fluorescence microscopy Yeast cells expressing tagged proteins, equivalent to ~0.2 to 0.5 OD600 units, were collected and imaged using an Axio Imager Z1 fluorescence microscope. The microscope was equipped with an A-Plan 10× 0.25 Ph1 objective lens (Zeiss), a Compact Light Source HXP 120V, and an Axiocam 506 mono digital camera. Image acquisition was performed with ZEN software and subsequently deconvolved using Open Lab (Improvision) following the manufacturer’s recommended parameters. Further analysis of the images was conducted using either Fiji ImageJ or Photoshop CS4 (Adobe). To stain the vacuolar membrane, 0.5 OD600 units of cells were pelleted at 3000g for 5 min. The cell pellet was resuspended in 100 μl of yeast extract, peptone, and dextrose (YPD) media containing FM4-64 (1.6 μM) and incubated at 30°C for 30 to 60 min. Afterward, the cells were centrifuged, and the pellet was washed twice with YPD media to remove excess FM4-64 before the cells were imaged. For high-resolution imaging, cells were visualized using a Nikon AXR confocal microscope equipped with a four-line LUA-S4 laser engine (405, 488, 561, and 640 nm), a 100×/1.45 numerical aperture (NA) Apo objective and an NSPARC spectral detector. Images were captured in Galvano scanning mode at 1024 × 1024 pixels with a zoom setting that produced 0.064-μm pixels using NIS Elements software. Measurement of cargo (Rtn1) and Lst1 that accumulate adjacent to autophagosome assembly sites To quantify the displacement of the Rtn1 and Lst1 puncta from the Atg8 structures, we took advantage of the diffraction-limited nature of these structures and used traditional fluorescence maxima localization to measure the effective distance between the puncta. Briefly, Atg8 puncta with neighboring structures (Rtn1 or Lst1 puncta) were identified, and the spot displacement axis was manually annotated using a custom javascript code in Fiji. The center of each puncta was estimated as the maximum of the point spread function (fit as a Gaussian function) along the linear displacement axis, and the distance between the two centers was recorded. To test whether these distances were significantly larger than zero, we compared the observed distances to a simulated sample of equal size where the measured distances were the error in localization of the Atg8 puncta (estimated as a normal distribution at the size of the microscope’s point spread function, full width at half maximum = 250 nm). Statistics were performed using a nonparametric Kruskal-Wallis test with Welch’s correction. Cleavage assays to measure autophagy Approximately 5.0 OD600 units of cells, expressing a GFP-tagged fusion protein (Sec61, Sec63, Sec66, Rtn1, Hmg1, Atg40, Lst1, Per33, or Atg8), were collected by centrifugation at 3000g for 5 min. After two washes with distilled water, the cells were incubated with 1.0 ml of ice-cold 10% trichloroacetic acid (TCA) for a minimum of 30 min. The cells were then pelleted at 13,000g for 5 min and washed two times with 1 ml of ice-cold acetone to remove residual TCA. The resulting cell pellet was air-dried at room temperature and resuspended in 50 μl of boiling buffer [50 mM tris-HCl, 1 mM EDTA, 1% SDS, and 6 M urea (pH 7.5)]. Silica beads, equal to the sample volume, were added to the tube, and the samples were vortexed for 5 min. This was followed by a 5-min incubation at 55°C. Subsequently, 50 μl of 2× sample buffer [150 mM tris-HCl, 6% SDS, 6 M urea, 10% 2-mercaptoethanol, and 0.002% bromophenol blue (pH 6.8)] was added, and the mixture was vortexed for 5 min before it was incubated for 5 min at 55°C. Cell debris and silica beads were eliminated during a 10-min centrifugation at 13,000g before the samples were subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE). Immunoblotting, used to detect the GFP fusion protein and free GFP, was performed with an anti-GFP mouse monoclonal antibody (1:3000 dilution; Clone 7.1 and 13.1, Roche). Adh1, which was used as a loading control, was probed with an anti-Adh1 rabbit polyclonal antibody (1:10,000 dilution; AB1202, EMD Millipore). In vitro binding experiments Yeast lysates, prepared from cells expressing tagged Atg40 or Rtn1, were treated with rapamycin (400 ng/ml) at an OD600 of 0.5 to 0.8 and then incubated at 30°C for 3 hours with shaking. Subsequently, the cells were pelleted at 4230g for 10 min at 4°C and washed with ice-cold spheroplasting buffer [1.4 M sorbitol, 10 mM sodium azide, 50 mM potassium phosphate, and β-mercaptoethanol (final concentration of 0.35%)]. The washed cells were resuspended in 20 ml of ice-cold spheroplasting buffer containing 500 μl of zymolase (2 mg/ml) and incubated in a water bath at 37°C with intermittent mixing. Afterward, the spheroplasts were centrifuged at 500g for 10 min at 4°C and washed two times with ice-cold spheroplasting buffer (without zymolase). The spheroplasts were resuspended in 6 ml of lysis buffer [25 mM Hepes (pH 7.4), 150 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol, and 5× protease inhibitor] and lysed in a dounce homogenizer. After lysis, 1 M tris-HCl (pH 7.5) (100 μl/1 ml lysate) and Triton X-100 (final concentration of 0.5%) were added to the lysate. The lysate was incubated at 4°C for 30 min with intermittent mixing and centrifuged for 15 min at 4°C at 10,000g. The supernatant was collected, and protein was measured by the bicinchoninic acid method. For the in vitro binding reactions, the lysate was diluted with lysis buffer (without Triton X-100) to a final concentration of 0.3% Triton X-100. In vitro binding reactions were performed with 30 μl of glutathione beads containing 0.1 μM of purified glutathione S-transferase (GST), GST-Sec13, or GST-Lst1 and incubated with 2 mg of lysate in a 1-ml reaction. Samples were then incubated at room temperature for 1.5 hours on an end-to-end rotor. The beads were centrifuged at 900g for 2 min at 4°C and washed four times with 1 ml of lysis buffer (containing 0.3% of Triton X-100). The washed beads were resuspended in 50 μl of SDS-PAGE sample loading buffer, heated to 100°C for 10 min, and electrophoresed on a 10% gel. Oligomerization studies Oligomers of Atg40 were visualized by irreversibly cross-linking intact membranes with the cross-linker EGS. The method we used was a modification of a protocol used before to monitor the formation of oligomers in other RHD-containing proteins (35). Briefly, cells expressing Atg40–3× FLAG were grown to an OD600 of 0.2 in SD medium that was supplemented with 2% glucose. ER-phagy was induced with rapamycin (200 ng/ml) for 6 hours at 30°C. Subsequently, the cells were pelleted, washed once with water, flash-frozen in liquid nitrogen, and stored at −80°C. The cell pellet was then ground to a fine powder by submerging it in liquid nitrogen. The grinding was performed with a mortar and pestle that was prechilled with liquid nitrogen. The powder was resuspended in 1 ml of TKMG lysis buffer [50 mM tris (pH 7.0), 150 mM KCl, 2 mM MgCl2, 10% glycerol, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 mM 4-(2-aminoethyl)benzenesulfonylfluoride hydrochloride], and debris was removed during centrifugation at 2000g at 4°C for 5 min. Samples were pelleted at 100,000g at 4°C for 30 min, and the pellet was resuspended in 300 μl of HKP buffer [50 mM Hepes (pH 7.0), 150 mM KCl, and 1 mM PMSF]. The resuspension was incubated at a protein concentration of 3 mg/ml with different amounts (0, 0.1, 0.25, 0.5, and 1 mM) of EGS (Pierce) for 30 min at room temperature. The reactions were quenched with the addition of 1 M tris (pH 7.5) (1:10 dilution) for 15 min. Samples were analyzed by SDS-PAGE (4 to 15% gel) and immunoblotted with mouse anti-FLAG antibody (Sigma-Aldrich) to detect Atg40–3× FLAG. FRAP analysis Yeast cells expressing Sec61-GFP or Atg40-GFP, under the Rtn1 promoter, were precultured at 25°C to late log phase, then reinoculated into SD-Leu medium and grown overnight at either 20° or 30°C to exponential phase (OD600 = 0.2 to 0.5). Rapamycin (400 ng/ml) was then added, and the cultures were maintained at the respective temperatures for an additional 6 hours prior to imaging. Cells were imaged on a Nikon AR confocal microscope attached to a Ti2-E system featuring a Plan Apo 100× 1.45 NA objective. The AXR was equipped with a LUA-S4 laser engine using four lines (405, 488, 561, and 640 nm), and the newest DUX-VB tunable emission GaAsP detectors. To achieve a faster imaging rate of two frames per second, images were captured at 1024 × 1024 pixels with a zoom setting that produced 0.15-μm pixels. The pinhole was fully opened to optimize for the detection of emitted photons. During scanning, the 488-nm laser was set to 10%, and for bleaching, the 488-nm laser intensity was adjusted to 100% to obtain selected regions of interest (ROIs) that were drawn manually. Three prebleached images were acquired without delay, and postbleaching images were taken continuously for 5 min with 50-s intervals. The Time Measurement tool within NIS Elements was used to quantify the intensity of the bleached ROIs after the background intensity was subtracted from the images. FRAP quantification followed the method outlined by Phair and Misteli (36). Briefly, following image acquisition, fluorescence intensities within the ROIs were quantified using the Time Measurement tool within NIS Elements. Raw intensity values were corrected for background by subtracting fluorescence from noncellular regions and subsequently normalized to compensate for photofading during image acquisition. The normalized recovery curves were then fitted to mathematical models implemented in NIS Elements to determine the half-time of recovery (t½). FRAP experiments were independently performed in three replicates, with 20 to 30 cells analyzed per replicate. Data are presented as the means ± standard error of the mean (SEM), as indicated in the figure legends and the “Statistical analysis” section. Purification of ER membranes ER membranes were isolated from 30 OD600 units of WT and ΔLD mutant cells containing 3× HA–tagged Sec63 using the rapid immunopurification of ER membranes method described by Banerjee et al. (37). Briefly, cells were treated with 10 mM sodium fluoride and 10 mM sodium azide before they were centrifuged in a swinging bucket rotor at 1900g for 5 min at 4°C. After the spin, the cells were washed in 0.5 ml of KPBS buffer [136 mM KCl and 10 mM potassium phosphate (pH 7.25)] and centrifuged at 3381g for 3 min at 4°C. The cell pellet was then resuspended in 300 μl of lysis buffer (KPBS containing a protease inhibitor cocktail, Roche catalog no. 11836170001, supplemented with 1 mM PMSF) and 300 μl of glass beads (Sigma-Aldrich, G8772). Cells were lysed using a Precellys 24 Homogenizer (Bertin Technologies) at 30-s intervals for 3 min with 1-min gaps in between. To remove cell debris, the mixture was centrifuged at 400g for 5 min at 4°C. The supernatant was collected and then transferred to a fresh tube. The remaining beads were incubated with 600 μl of lysis buffer, vortexed, and centrifuged, and 550 μl of the supernatant was combined with the first supernatant. Supernatant was saved (100 μl) to measure cellular lipids. A total of 100 μl of anti-HA antibody magnetic beads (Pierce catalog no. 88837) that were prewashed in lysis buffer were mixed with the remaining supernatant and incubated by rocking at 4°C for 15 min. The beads were then washed three times in cold lysis buffer at 4°C, and membranes were eluted during a 30-min incubation at 30°C with 100 μl of HA-peptide (1 mg/ml; Thermo Fisher Scientific, 26184) in KPBS buffer. The eluted membranes were resuspended in 500 μl of KPBS on ice. Samples analyzed by Western blot analysis were prepared by resuspending the elute in 4× LDS buffer (Invitrogen, 2399420) with 2% 2-mercaptoethanol. The remaining membranes were stored on ice. Lipidomics For whole-cell lipidomics (see table S3), yeast cells were precultured at 25°C to late log phase, then reinoculated into SD complete medium and grown overnight at either 20° or 30°C to exponential phase (OD600 = 0.2 to 0.5). Rapamycin (400 ng/ml) was then added, and the cultures were incubated for an additional 16 hours. Subsequently, 30 to 50 OD600 units of cells were harvested for lipidomic analysis. To perform lipidomics analysis on purified ER membrane fractions (see table S2), Sec63–3× HA–tagged ER membranes were purified from WT and ΔLD mutant cells. Cells were grown to an OD600 of 0.6 and treated with rapamycin (400 ng/ml) for 0, 3, and 6 hours, and ER membranes were isolated from 30 OD600 units of cells using the rapid immunopurification method described above and by Banerjee et al. (37). Lipids were extracted from ER membranes as well as whole cell lysates and analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) following the method described in Banerjee et al. (37). Briefly, lipid extracts equivalent to total fatty acid content of 25 nmol were dissolved in 100 μl of chloroform/methanol (2:1, v/v) spiked with internal standards. The internal standards used were 125 pmol of PE 18:0 to 18:0 and DAG 18:0 to 22:6 from Avanti Polar Lipid and hydrogenated spinach thylakoid sulfoquinovosyldiacylglycerol 16:0 to 18:0 (38, 39). The lipids were then separated and quantified by high-performance liquid chromatography and MS/MS as described before (38, 39). Statistical analysis All experiments were independently replicated at least three times. Results are presented as the means ± SEM or the median and 95% confidence interval of the median, if the population was not confidently shown to be normal. The specifics are detailed in the figure legends. Statistical significance was determined using Student’s t test or a one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test for normally distributed samples, or a Kruskal-Wallis nonparametric test with Dunn’s correction for samples that could not be confidently shown to be normal. Statistical P values < 0.05 were considered to be significant in all comparisons. The number of cells analyzed in the FRAP analysis and immunofluorescence studies are stated in the legends.