Overexpression of autophagy genes activates lipophagy in Arabidopsis Authors: Jilian Fan, Saroj Kumar Sah, Dongling Xie, Changcheng Xu Published: 2026 ABSTRACT ABSTRACTCells possess an intrinsic recycling system called macroautophagy/autophagy, which delivers obsolete or damaged cytoplasmic components into the vacuole for degradation and reuse. Autophagy-mediated breakdown of organellar membrane lipids supplies fatty acids for the synthesis of triacylglycerols (TAGs), which are then packaged into subcellular organelles called lipid droplets (LDs). Conversely, autophagy contributes to TAG turnover by delivering LDs into vacuoles for breakdown by resident acid lipases. Additionally, LDs can undergo degradation through cytosolic lipolysis mediated by SDP1 (Sugar-Dependent1) lipase in Arabidopsis. Autophagy-mediated LD, a process referred to as lipophagy, has been described in plants. Nevertheless, the precise mechanisms of lipophagy and the specific types of vacuoles involved remain unclear. Here, we show that overexpression of autophagy genes ATG5 or ATG8 promotes autophagic activity and significantly reduces LD accumulation. We demonstrate that the decreased LD abundance is not due to increased LD degradation mediated by SDP1 but is dependent on autophagy. The abundance of vacuoles specifically labeled by δ-TIP3 (delta-tonoplast intrinsic protein 3) was markedly decreased in ATG5-overexpressing lines. In addition, disruption of autophagic genes prevented the formation of both δ-TIP3-positive vacuoles and the mysterious vacuolar lumen structures previously termed bulbs, indicating their autophagic origin. Furthermore, confocal imaging analysis revealed close associations between LDs and δ-TIP3-labeled vacuoles, as well as the presence of LDs within vacuoles delimited by δ-TIP3. Together, our findings indicate that overexpression of autophagy genes triggers lipophagy and identify a distinct type of vacuole involved in this process.Abbreviation: ATG: autophagy related; AV: autophagic vacuole; ConA: concanamycin A; DsRed: Discosoma sp red fluorescent protein; ER: endoplasmic reticulum; GFP: green fluorescent protein; LD: lipid droplet; OLE: oleosin; SDP1: Sugar-Dependent1; TAG: triacylglycerol; TIP: tonoplast intrinsic protein; TLC: thin-layer chromatography Introduction Autophagy acts as a crucial cellular recycling system in eukaryotes, including fungi, plants, and animals. It targets unnecessary or damaged cellular components, like organelles and large molecules, for breakdown and reuse. There are two main types of autophagy distinguished by their cargo delivery method [1–4]. Macroautophagy involves the formation of a special membrane sac called a phagophore. The phagophore engulfs cytoplasmic material and then seals itself around the cargo, forming a double-membraned autophagosome. Finally, the autophagosome fuses with a vacuole (in fungi and plants) or lysosome (in animals), delivering its cellular debris for degradation by digestive enzymes within. More than 30 ATG (autophagy related) proteins play various roles in macroautophagy in yeast and many of them are conserved in plants. Among them, ATG2, ATG3, ATG4, ATG5, ATG7, ATG8, ATG10, ATG12, and ATG16 are essential components of two ubiquitin-like conjugation systems that mediate the lipidation of ATG8 with phosphatidylethanolamine (PE) and the formation of autophagy-related membranes [4,5]. ATG5 plays a central role in the formation of autophagosomes, specifically in the conjugation of ATG12 to ATG5, which is essential for the expansion of phagophore membranes [3]. ATG8 is associated with phagophores, autophagosomes and autophagic bodies, making this protein an excellent marker to monitor the process of autophagy [6]. In addition, because ATG8–PE is involved in multiple steps of autophagosome formation, including phagophore membrane expansion and closure, monitoring the relative abundance of lipidated versus nonlipidated ATG8 is widely used as a measure of macroautophagic activity [7]. An alternative form of autophagy that occurs ubiquitously in all eukaryotes is microautophagy. This pathway involves direct engulfment of cytoplasmic content via invagination or protrusion of vacuolar or lysosomal membranes for degradation in vacuoles or lysosomes [3,8]. Several forms of microautophagy with remarkably diverse molecular machinery exist in yeast and mammals [8]. In plants, microautophagy is involved in the degradation of various cytoplasmic components [9,10] including lipid droplets (LDs) [11,12]. LDs are key subcellular organelles responsible for the storage and metabolism of neutral lipids, predominantly triacylglycerols (TAGs) [13–18]. They are formed in the endoplasmic reticulum (ER), where enzymes catalyzing the assembly of TAGs from diacylglycerols and fatty acids reside. In plants, LDs are mainly found in seeds. Seed LDs are composed of TAG cores coated with a monolayer of phospholipids and LD-specific proteins such as OLEs (oleosins). Genetic studies in Arabidopsis have suggested roles of OLEs in preventing LD coalescence and protecting LDs from adventitious hydrolysis during seed maturation and germination [19,20]. Plant vegetative tissues such as leaves do not accumulate TAG to a significant extent, even though they contain high enzymatic activities for TAG synthesis and turnover [21]. One possible reason for limited accumulation of LDs in leaves is the absence of OLE coating and hence protection against TAG hydrolysis, since OLE genes are not expressed in leaves. In line with this possibility, ectopic expression of seed specific OLE1 (oleosin1) has been shown to result in an increased accumulation of TAG and LDs in leaves [22]. In addition, disruption of SDP1 (Sugar-Dependent1), a TAG lipase catalyzing the critical step in TAG hydrolysis in the cytosol in Arabidopsis, results in significant increases in TAG accumulation in leaves, stems and roots [23,24], indicating a role of cytosolic hydrolysis in TAG degradation in vegetative tissues. Autophagy-mediated degradation of LDs is often termed lipophagy. This LD catalytic pathway was first discovered in mammalian cells [25] and later in yeast [26] and plants [11,12,27]. In mammalian cells, early studies have shown that lipophagy can occur via the sequestration of LDs within autophagosomes prior to their delivery to lysosomes for degradation [25,28]. This LD macroautophagic process is therefore named macrolipophagy. More recent studies, however, suggest a direct transfer of neutral lipids and proteins from LDs to lysosomes for degradation in the absence of the core ATG proteins of macroautophagy [29]. In yeast and plants, the autophagy-mediated LD degradation is featured by the direct engulfment of LDs by vacuolar membranes in a process resembling microautophagy and hence termed microlipophagy. In plants, lipophagy has been observed in leaves under starvation conditions [11] and during pollen maturation [12] and seed germination [27]. This process requires at least some components of macroautophagic machinery [11,12], but the exact role of core ATG proteins in plant microlipophagy remains largely unknown. Plant cells contain multiple vacuoles, each serving distinct functions believed to be associated with the presence of specific isoforms of TIPs (tonoplast intrinsic proteins) [30–32]. In Arabidopsis, three main isoforms of TIPs have been identified: α-, δ-, and γ-TIP [33,34]. Their transcripts exhibit tissue-specific expression patterns, with α-TIP expression being confined to seeds, while δ-TIP and γ-TIP are predominantly present in vegetative tissues [35]. Vacuoles marked by the presence of α-TIP are dedicated to protein storage, whereas those marked with δ-TIP are primarily involved in digestive processes due to their lytic nature [36,37]. Interestingly, α-TIP-labeled vacuoles have been linked to autophagic processes [38], though their involvement in lipophagy remains unexplored. Prior research has indicated that knockout of core autophagic genes impairs TAG synthesis and LD accumulation in leaves [11]. Additionally, it affects lipid homeostasis during senescence [39], lipophagy in pollen [12] and leaves under nutrient starvation [11]. Overexpression of ATG5, ATG7 or ATG8 promotes autophagic flux and affects plant growth, development, nutrient remobilization and stress response [40,41], but whether and how overexpression of autophagic genes impacts lipid metabolism and storage remains unexplored. To examine the possible role of ATG proteins in TAG storage, we generated stable Arabidopsis transgenic lines coexpressing OLE1 with ATG5 or ATG8. We found that enhanced autophagic activity due to ATG5 overexpression resulted in decreases in TAG and LD accumulation in leaves. Our results demonstrated that the decreased LD accumulation in ATG5 overexpressors is due to increased TAG degradation mediated by lipophagy, rather than cytosolic hydrolysis. We additionally found a marked decrease in the abundance of autophagic vacuoles (AVs) labeled specifically by an isoform of δ-TIPs, namely δ-TIP3 [42] in ATG5 overexpressing lines. Moreover, confocal fluorescence imaging unveiled associations between LDs and vacuoles, along with the presence of LDs within lytic vacuoles. Results Overexpression of ATG5 activates autophagy We recently reported that constitutive expression of green fluorescent protein (GFP)-tagged OLE1 resulted in an accumulation of OLE1-GFP-coated LDs in leaves [11]. To further explore the role of autophagy in lipid metabolism, we generated stable transgenic lines overexpressing ATG5 under the control of 35S promoter in the OLE1 (OLE1-GFP overexpressor) background. Multiple transgenic lines were obtained and three independent lines overexpressing ATG5 were propagated to homogeneity and used for detailed analysis. To test whether overexpression of ATG5 enhances autophagic activity, we analyzed the expression of ATG5 and ATG8 using quantitative polymerase chain reaction (qPCR). We detected 9-, 27- and 7-fold increases in levels of ATG5 transgenes in line 2 (ATG5-2 OLE1), 6 (ATG5-6 OLE1) and 7 (ATG5-7 OLE1), respectively, compared with OLE1 (Figure 1(A)). Overexpression of a single core autophagic gene can often lead to increases in transcript and protein abundance of other autophagic genes, thus increasing overall autophagic flux [40,41,43]. Consistent with this, the transcript levels of ATG8 were approximately 4-, 3- and 9-fold higher in ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1, respectively, compared with the control (Figure 1(B)). Figure 1.Overexpression of ATG5 enhances autophagic flux. (A) and (B) qRT-PCR analysis of transcript levels of ATG5 (A) and ATG8e (B) in leaves of wild-type (WT) and transgenic lines overexpressing ATG5. (C) Confocal images of epidermal root cells from seedlings overexpressing DsRed-ATG8e, treated with or without ConA. Bars: 10 µm. (D) the average number of DsRed-positive puncta in the images shown in (C). Three to five roots were imaged per genotype. Asterisks indicate statistically significant differences from OLE1 based on Student’s t test (*p < 0.05, **p < 0.01, ***p, 0.001). (E to G). Western blot analysis of levels of free DsRed (E), ATG8–PE (F) and NBR1 (G) in leaves of WT, OLE1 and ATG5 OLE1 lines under carbon-starvation conditions. NBR1 levels were analyzed in seedlings, treated with or without ConA. Equal protein amounts were subjected to SDS-PAGE followed by immunoblotting with antibodies against DsRed (E), ATG8 (F) or NBR1 (G). An anti-actin antibody was used as a loading control. Ratios of free DsRed to DsRed-ATG8, of ATG8–PE to ATG8 and of NBR1 to actin were calculated based on the relative band intensities quantified using ImageJ and are shown in (E), (F) and (G), respectively. [Use a colon for ratios as in “Free DsRed:DsRed-ATG8.” Use a standard dash for “DsRed-ATG8”].The image A showing a bar chart titled ATG5. The x-axis lists WT, OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. The y-axis is Relative expression (no unit), from 0 to 30 with ticks at 10, 20, 30. Bars: WT about 1; OLE1 about 1; ATG5-2 OLE1 about 9; ATG5-6 OLE1 about 28 to 29; ATG5-7 OLE1 about 8. Asterisks appear above ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. The image B showing a bar chart titled ATG8. The x-axis lists WT, OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. The y-axis is Relative expression (no unit), from 0 to 10 with ticks at 2, 4, 6, 8, 10. Bars: WT about 1; OLE1 about 1; ATG5-2 OLE1 about 4; ATG5-6 OLE1 about 2.5; ATG5-7 OLE1 about 9. Asterisks appear above ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. The image C showing confocal micrographs arranged as 2 rows by 4 columns. Column headers: OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. Row labels: minus ConA (top) and plus ConA (bottom). Each micrograph includes a scale bar. The image D showing a grouped bar chart with legend minus CoA and plus CoA. The x-axis lists OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. The y-axis is Normalized puncta per cell (no unit), from 0 to 6 with ticks at 1 through 6. Minus CoA bars are about 1 for all genotypes. Plus CoA bars: OLE1 about 1; ATG5-2 OLE1 about 2.3; ATG5-6 OLE1 about 3.3; ATG5-7 OLE1 about 4.6 to 4.8. Asterisks appear above the plus CoA bars for ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. The image E showing a western blot labeled kDa with markers at 40 and 25. Lanes: OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1, atg5 OLE1. Bands are labeled DsRed-ATG8 and Free DsRed. A second blot below is labeled Actin. Text below lists Free DsRed:DsRed-ATG8 ratios: 1.2, 6.5, 7.1, 5.4. The image F showing a western blot labeled kDa with marker at 15 and 40. Lanes: OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1, atg5 OLE1. Bands are labeled ATG8 and ATG8–PE, with an Actin blot below. Text below lists ATG8–PE:ATG8 ratios: 0.3, 0.8, 0.9, 0.7. The image G showing two stacked western blot sets labeled minus ConA (top set) and plus ConA (bottom set). Lanes: OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. Each set shows NBR1 with an Actin blot below. Text lists NBR1:Actin ratios for minus ConA: 0.9, 0.2, 0.3, 0.5; and for plus ConA: 1.2, 8.9, 10.6, 7.2.Multi-panel plots and blots of ATG5 and ATG8 expression and DsRed puncta across genotypes and treatments. Overexpression of ATG5 enhances autophagic flux. (A) and (B) qRT-PCR analysis of transcript levels of ATG5 (A) and ATG8e (B) in leaves of wild-type (WT) and transgenic lines overexpressing ATG5. (C) Confocal images of epidermal root cells from seedlings overexpressing DsRed-ATG8e, treated with or without ConA. Bars: 10 µm. (D) the average number of DsRed-positive puncta in the images shown in (C). Three to five roots were imaged per genotype. Asterisks indicate statistically significant differences from OLE1 based on Student’s t test (*p < 0.05, **p < 0.01, ***p, 0.001). (E to G). Western blot analysis of levels of free DsRed (E), ATG8–PE (F) and NBR1 (G) in leaves of WT, OLE1 and ATG5 OLE1 lines under carbon-starvation conditions. NBR1 levels were analyzed in seedlings, treated with or without ConA. Equal protein amounts were subjected to SDS-PAGE followed by immunoblotting with antibodies against DsRed (E), ATG8 (F) or NBR1 (G). An anti-actin antibody was used as a loading control. Ratios of free DsRed to DsRed-ATG8, of ATG8–PE to ATG8 and of NBR1 to actin were calculated based on the relative band intensities quantified using ImageJ and are shown in (E), (F) and (G), respectively. [Use a colon for ratios as in “Free DsRed:DsRed-ATG8.” Use a standard dash for “DsRed-ATG8”]. To investigate the impact of ATG5 overexpression on autophagic flux, we crossed ATG5-2 OLE1 with a transgenic line overexpressing the autophagic marker ATG8 fused to a DsRed (Discosoma sp red fluorescent gene) (DsRed-ATG8) [11] and obtained a transgenic line overexpressing both ATG5 and DsRed-ATG8 in the OLE1 background. To visualize autophagic structures, 7-day-old seedlings were treated with concanamycin A (ConA) to inhibit vacuolar degradation, thus leading to the accumulation of undegraded autophagic bodies labeled with DsRed-ATG8 in the vacuolar lumen. From confocal microscopic imaging, we observed that the number of DsRed-ATG8-labeled puncta was significantly increased in the ATG5 OLE1 line compared with OLE1 (Figure 1(C,D)), indicating an increase in autophagic activity due to ATG5 overexpression. Autophagic activity was further assessed by monitoring the release of free DsRed from the DsRed-ATG8 fusion protein. Upon autophagy induction, DsRed-ATG8 is delivered to the vacuole, where ATG8 is degraded and the relatively stable DsRed moiety accumulates; therefore, the ratio of free DsRed to DsRed-ATG8 serves as an indicator of autophagic activity. Consistent with the microscopy results, substantial accumulation of free DsRed was observed in ATG5 overexpressing lines under carbon starvation, whereas a significantly lower DsRed:DsRed-ATG8 ratio was detected in the OLE1 under the same conditions (Figure 1(E)). Autophagic flux can also be assessed by monitoring the abundance of ATG8–PE [7], which exhibits a faster migration on protein gels in the presence of urea compared to its unmodified forms [44,45]. Western blot analysis, employing antibodies against ATG8, demonstrated notable increases in levels of ATG8–PE conjugates in ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1 under carbon starvation conditions (Figure 1(F)). NBR1 is a selective autophagy adaptor that recognizes cargo proteins and interacts directly with ATG8 to promote their sequestration into autophagosomes, and its turnover is enhanced when autophagic flux is increased [7]. We observed reduced NBR1 protein levels in all ATG5 overexpressing lines under carbon starvation, indicating stimulated NBR1 degradation in these plants (Figure 1(G)). To further assess autophagic flux, we treated seedlings with ConA, which inhibits vacuolar H+ -ATPases and blocks the degradation of autophagic bodies and NBR1 in the vacuole. ConA treatment led to a marked accumulation of NBR1, with greater increases in ATG5 overexpressing lines than in OLE1, consistent with elevated autophagic flux in the ATG5 overexpressors. Together, these results suggest that overexpression of ATG5 enhances autophagy activity. LD abundance is drastically decreased in leaves of ATG5 overexpressors To test whether increased autophagic activity affects TAG and LD accumulation, total leaf lipids were separated by TLC (thin-layer chromatography) and TAG content was quantified. TAG content was negligible in leaves of wild-type plants but increased in OLE1 (Figure 2(A,B)), consistent with our previous reports [11,22]. Overexpression of ATG genes resulted in drastic decreases in TAG levels (Figure 2(A,B)). When the OLE1 was constitutively expressed in leaves, it targeted specifically to LDs, leading to the formation of LD clusters. Such clusters were abundantly present in leaves of OLE1 but were rarely observed in ATG5 OLE1 lines (Figure 2(D)). Quantitative analysis showed that the number of LD clusters decreased by 79%, 67% and 89% in ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1, respectively, compared with OLE1 (Figure 2(C)). It should be noted that among the three ATG5 overexpressors, the most significant negative impact of ATG5 overexpression on TAG content and LD abundance was observed in the ATG5-7 OLE1 line. This transgenic line also exhibited the highest autophagic flux, as determined by changes in ATG8–PE abundance and ATG8 expression (Figure 1). These findings support the hypothesis that the decreased LD accumulation in ATG5 lines is attributed to increased autophagic activity. Figure 2.Overexpression of ATG5 results in decreases in TAG content and LD cluster number. (A) Thin layer chromatogram of neutral lipids in leaves of wild type (WT), transgenic lines overexpressing OLE1 (OLE1) or ATG5 in the OLE1 background (ATG5 OLE1). Lipids were visualized with 5% sulfuric acid by charring. (B) and (C) TAG content (B) and the number of LD clusters (C) in leaves of WT, OLE1 and ATG5 OLE1 lines. Data are means of three independent replicates with sd. Asterisks indicate statistically significant differences from OLE1 based on Student’s t test (*p < 0.01, ***p, 0.001). (D) Images of LD clusters (OLE1-GFP, green) in leaves of OLE1 and ATG5 OLE1 lines. Bars: 10 µm.The image A shows a thin layer chromatogram of neutral lipids in leaves of wild type (WT), OLE1, ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. TAG is marked. The image B shows a bar graph of TAG content (mg/g DW) in leaves of WT, OLE1, ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. Asterisks indicate statistically significant differences from OLE1 based on Student’s test (< 0.01, p < 0.001). The image C shows a bar graph of LD clusters per view in leaves of OLE1, ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. Asterisks indicate statistically significant differences from OLE1 based on Student’s test (< 0.01, p < 0.001). The image D shows images of LD clusters (OLE1-GFP) in leaves of OLE1, ATG5-2 OLE1, ATG5-6 OLE1 and ATG5-7 OLE1. Bars represent 10 µm.Four panels show lipid analysis in plant leaves: chromatogram, TAG content graph, LD cluster graph and LD images. Overexpression of ATG5 results in decreases in TAG content and LD cluster number. (A) Thin layer chromatogram of neutral lipids in leaves of wild type (WT), transgenic lines overexpressing OLE1 (OLE1) or ATG5 in the OLE1 background (ATG5 OLE1). Lipids were visualized with 5% sulfuric acid by charring. (B) and (C) TAG content (B) and the number of LD clusters (C) in leaves of WT, OLE1 and ATG5 OLE1 lines. Data are means of three independent replicates with sd. Asterisks indicate statistically significant differences from OLE1 based on Student’s t test (*p < 0.01, ***p, 0.001). (D) Images of LD clusters (OLE1-GFP, green) in leaves of OLE1 and ATG5 OLE1 lines. Bars: 10 µm. Both rates of TAG synthesis and breakdown are enhanced in ATG5 overexpressors In plants, fatty acids, the building blocks of TAG, are exclusively synthesized in the chloroplast, while the assembly of fatty acids into TAG occurs in the ER following their export from the chloroplast and TAG is packaged into LDs in the cytosol. In leaf tissues, TAG and LD undergo rapid turnover mediating by SDP1 [24]. To test whether the decreased TAG accumulation in ATG5 overexpressors is due to decreased fatty acid and TAG synthesis or increased TAG turnover, we carried out pulse-chase radiotracer labeling experiments with 14C-acetate, a precursor for fatty acid synthesis [46]. To achieve this, detached expanding leaves were initially incubated with 14C-acetate for a 1-h (pulse). Subsequently, the label was removed by repeatedly washing with water, and the leaves were further incubated for 3 days in the presence of cold acetate (chase). Samples were collected at 1-day intervals for lipid analysis by TLC, and the radioactivity associated with TAG and total lipids was measured by scintillation counting. We observed no differences in the initial rate of radiolabel incorporation into total lipids, indicating that overexpression of OLE1 or ATG5 had no impact on rates of fatty acid synthesis (Figure S1). However, the radioactivity in TAG was significantly higher in ATG5 OLE1 lines compared to OLE1 immediately following a 1-hour pulse, with the highest level of radiolabeled TAG found in the ATG5-7 OLE1 line (Figure 3(A)). This suggests an increase in membrane lipid conversion into TAG via autophagy due to ATG5 overexpression. No difference in radiolabeled TAG was found between OLE1 and WT. The radiolabel in TAG declined during the chase (Figure 3(B)). Compared with WT, the level of radiolabeled TAG was significantly higher in OLE1 at the end of the 3 d of chase, supporting our previous finding that the increased TAG accumulation in the OLE1 line is due to decreased TAG turnover [22]. Importantly, rates of radiolabeled TAG loss during the 3 d of chase in ATG5 OLE1 lines were significantly higher compared with WT and OLE1 (Figure 3(B)). Again, the largest loss in radiolabeled TAG during the chase occurred in the ATG5-7 OLE1 line. Together, these results suggest increases in membrane lipid and TAG turnover due to ATG5 overexpression and that the reduced LD accumulation in leaves of ATG5 OLE1 lines is due to increased TAG turnover. Figure 3.Overexpression of ATG5 results in increased rates of both TAG synthesis and SDP1-independent TAG degradation. (A) and (B) Rates of TAG synthesis (A) and degradation (B) in leaves of wild type (WT), transgenic lines overexpressing OLE1 (OLE1) or ATG5 in the OLE1 background (ATG5 OLE1). (C) and (D) Rates of TAG synthesis (C) and degradation (D) in leaves of wild type (WT), sdp1 and transgenic lines overexpressing OLE1 (OLE1 sdp1), ATG5 (ATG5 sdp1) or OLE1 and ATG5 in the sdp1 background (ATG5 OLE1 sdp1). Detached leaves were incubated with 14C-acetate for 1 h, and the radiolabeled tag was then chased for 3 d. Radiolabeled tag loss was expressed as the percentage of the label remaining after a 3-d chase period, relative to the initial amount measured at 1 h. Data are means of three independent replicates with sd. Asterisks indicate statistically significant differences from OLE1 (A and B) or sdp1 (C and D) based on Student’s t test (*p < 0.05, **p < 0.01).Image A: Bar chart with categories WT, OLE1, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. Y-axis: Label incorporation into TAG (DPM/mg FW/h), 0-150. Heights: WT 55, OLE1 60, ATG5-2 OLE1 110, ATG5-6 OLE1 95, ATG5-7 OLE1 100. Asterisks on ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. Image B: Same categories, y-axis: TAG label loss during chase (% of initial label), 0-21. Heights: WT 15, OLE1 7, ATG5-2 OLE1 14, ATG5-6 OLE1 12, ATG5-7 OLE1 18. Asterisks on WT, ATG5-2 OLE1, ATG5-6 OLE1, ATG5-7 OLE1. Image C: Bar chart with categories WT, sdp1, OLE1 sdp1, ATG5-2 sdp1, ATG5-2 OLE1 sdp1. Y-axis: Label incorporation into TAG (DPM/mg FW/h), 0-125. Heights: WT 62, sdp1 60, OLE1 sdp1 63, ATG5-2 sdp1 95, ATG5-2 OLE1 sdp1 88. Asterisks on ATG5-2 sdp1, ATG5-2 OLE1 sdp1. Image D: Same categories, y-axis: TAG label loss during chase (% of initial label), 0-21. Heights: WT 14, sdp1 8, OLE1 sdp1 7, ATG5-2 sdp1 11, ATG5-2 OLE1 sdp1 14. Asterisks on WT, ATG5-2 sdp1, ATG5-2 OLE1 sdp1.A grouped bar chart showing tag label incorporation and tag label loss across five genotypes in four plots. Overexpression of ATG5 results in increased rates of both TAG synthesis and SDP1-independent TAG degradation. (A) and (B) Rates of TAG synthesis (A) and degradation (B) in leaves of wild type (WT), transgenic lines overexpressing OLE1 (OLE1) or ATG5 in the OLE1 background (ATG5 OLE1). (C) and (D) Rates of TAG synthesis (C) and degradation (D) in leaves of wild type (WT), sdp1 and transgenic lines overexpressing OLE1 (OLE1 sdp1), ATG5 (ATG5 sdp1) or OLE1 and ATG5 in the sdp1 background (ATG5 OLE1 sdp1). Detached leaves were incubated with 14C-acetate for 1 h, and the radiolabeled tag was then chased for 3 d. Radiolabeled tag loss was expressed as the percentage of the label remaining after a 3-d chase period, relative to the initial amount measured at 1 h. Data are means of three independent replicates with sd. Asterisks indicate statistically significant differences from OLE1 (A and B) or sdp1 (C and D) based on Student’s t test (*p < 0.05, **p < 0.01). Accelerated TAG degradation is not due to an increase in cytosolic lipolysis mediated by SDP1 Both cytosolic lipolysis and lipophagy contribute TAG hydrolysis in plants. The former is initiated by SDP1 and represents the major pathway for the metabolic breakdown of TAG in LDs in seeds [23] and leaves [24] under normal growth conditions. We crossed ATG5-2 OLE1 with sdp1 and recovered OLE1, ATG5 and ATG5 OLE1 in the sdp1 background (OLE1 sdp1, ATG5 sdp1 and ATG5 OLE1 sdp1, respectively). Quantitative PCR analysis confirmed that ATG5 transcript levels were strongly increased in the ATG5 overexpressors (Figure S2). To assess the impact of SDP1 on TAG turnover, we again carried out pulse-chase labeling experiments with 14C-acetate. We found that genetic manipulation of ATG5 expression resulted in a significant increase in the rate of radiolabel incorporation into TAG during 1 h of pulse (Figure 3(C)). During the chase, radiolabeled TAG decreased in WT but to a significantly lesser extent in sdp1 (Figure 3(D)), confirming that the major pathway for TAG hydrolysis is mediated by SDP1 in leaves. Additionally, the relative amount of radiolabeled TAG loss was significantly higher in ATG5 OLE1 sdp1 compared to OLE1 sdp1, suggesting that the increased TAG degradation caused by ATG5 overexpression is not primarily due to cytosolic lipolysis mediated by SDP1. δTIP3-DsRed labels AVs Our data thus far suggest that autophagy is involved in the degradation of LDs in ATG5 overexpressing lines. To provide additional evidence supporting this possibility and to investigate the mechanistic basis for lipophagy, we crossed ATG5-2 OLE1 with the transgenic line overexpressing δ-TIP3-DsRed (DsRed-tagged tonoplast marker δ-TIP3) [11] and recovered plants carrying δ-TIP3-DsRed in the ATG5-2 (ATG5 OLE1 δ-TIP3) or OLE1 (OLE1 δ-TIP3) background. Interestingly, whereas abundant δ-TIP3-DsRed-labeled vacuoles were found in leaves of OLE1 δ-TIP3, only a few of them were observed in the ATG5 OLE1 δ-TIP3 (Figure 4). To test whether δ-TIP3-DsRed-labeled vacuoles are autophagic structures, we generated T-DNA knockout mutants of ATG2, ATG5 and ATG7 overexpressing δ-TIP3 (atg2 δ-TIP3, atg5 δ-TIP3 and atg7 δ-TIP3). Strikingly, confocal microscopy analysis revealed that there were no or only a few intact δ-TIP3-DsRed-labeled vacuoles in δ-TIP3 lacking ATG2, ATG5 or ATG7 (Figure 5), indicating that δ-TIP3-DsRed-labeled vacuoles are likely of autophagic origin. Consistent with this, δ-TIP3-DsRed did not label the central vacuole (Figures 4–6 and Figure S3). These results indicate that δ-TIP3 is specifically present in membranes of AVs and that the drastic reduction in δ-TIP3-DsRed-labeled structures in ATG5 overexpressing lines is likely due to increased autophagic flux. Figure 4.Overexpression of ATG5 results in drastic decreases in numbers of LDs and δ-TIP3-marked vacuoles. Fluorescent images showing LD clusters (OLE1-GFP, green dots) and δ-TIP3-marked vacuoles (δ-TIP3-DsReD, red dots) in transgenic lines overexpressing δ-TIP3-DsRed (OLE1 δ-TIP3) and ATG5 (ATG5 OLE1 /δ-TIP3) in the OLE1 background. Bars: 20 µm.Image A: Fluorescence micrograph labeled OLE1-GFP, black background with small green puncta and larger green spots, white scale bar in lower right. Image B: Fluorescence micrograph labeled delta-TIP3-DsRed, black background with small red puncta and brighter red spots, white scale bar in lower right. Image C: Overlay micrograph labeled Overlay, grayscale background with scattered green and red puncta, some close together, white scale bar in lower right. Image D: Fluorescence micrograph labeled OLE1-GFP, black background with few green puncta spaced apart, white scale bar in lower right. Image E: Fluorescence micrograph labeled delta-TIP3-DsRed, black background with few red puncta and brighter spots, white scale bar in lower right. Image F: Overlay micrograph labeled Overlay, grayscale background with sparse green and red puncta, white scale bar in lower right.Fluorescence grid: green/red puncta on black, labeled OLE1-GFP, delta-TIP3-DsRed and Overlay. Figure 5.Autophagy-dependent accumulation of δ-TIP3-labeled vacuoles. Overlay of confocal images showing drastic decreases in the number of δ-TIP-labeled autophagic vacuoles in transgenic lines overexpressing δ-TIP3-DsRed in the atg2, atg5 or atg7 background. Bars: 25 µm.Microscopy images showing δ-TIP3-DsRed vacuoles in WT, atg2, atg5 and atg7 backgrounds with bright field and overlay views.Image A is a microscopy figure with three rows and four columns, separated by white borders. Columns are labeled WT, atg2, atg5 and atg7; rows are delta-TIP3-DsRed, Bright field and Overlay. The first row shows red fluorescence micrographs: WT has many small red puncta, atg2 has few, atg5 has a small number and atg7 has a small number with a brighter region. Each sub-image includes a white scale bar. The second row displays grayscale micrographs: WT shows darker patches, atg2 has a uniform texture, atg5 features a darker structure and atg7 has darker regions. Each sub-image has a white scale bar. The third row, Overlay, combines grayscale backgrounds with red puncta: WT has numerous puncta in the lower left, atg2 has sparse puncta, atg5 has puncta over the central structure and atg7 has puncta over the background, including one near the right side. Each sub-image contains a white scale bar. Figure 6.The formation of γ-TIP-labeled vacuolar lumen structures are partially dependent on autophagy. (A) Confocal images showing lack of vacuolar lumen structures (bulbs) in transgenic lines overexpressing γ-TIP-GFP in the atg2 or atg5 background. Bars: 20 µm). (B) Confocal images showing partial colocalization of bulbs (γ-TIP-GFP, green) with δ-TIP3-labeled autophagic vacuoles (δ-TIP3-DsRed, red) in transgenic lines coexpressing δ-TIP3-DsRed and γ-TIP-GFP. Bars: 20 µm.Confocal images showing γ-TIP and δ-TIP structures in transgenic lines. Two panels with three images each. Overexpression of ATG5 results in drastic decreases in numbers of LDs and δ-TIP3-marked vacuoles. Fluorescent images showing LD clusters (OLE1-GFP, green dots) and δ-TIP3-marked vacuoles (δ-TIP3-DsReD, red dots) in transgenic lines overexpressing δ-TIP3-DsRed (OLE1 δ-TIP3) and ATG5 (ATG5 OLE1 /δ-TIP3) in the OLE1 background. Bars: 20 µm. Autophagy-dependent accumulation of δ-TIP3-labeled vacuoles. Overlay of confocal images showing drastic decreases in the number of δ-TIP-labeled autophagic vacuoles in transgenic lines overexpressing δ-TIP3-DsRed in the atg2, atg5 or atg7 background. Bars: 25 µm. The formation of γ-TIP-labeled vacuolar lumen structures are partially dependent on autophagy. (A) Confocal images showing lack of vacuolar lumen structures (bulbs) in transgenic lines overexpressing γ-TIP-GFP in the atg2 or atg5 background. Bars: 20 µm). (B) Confocal images showing partial colocalization of bulbs (γ-TIP-GFP, green) with δ-TIP3-labeled autophagic vacuoles (δ-TIP3-DsRed, red) in transgenic lines coexpressing δ-TIP3-DsRed and γ-TIP-GFP. Bars: 20 µm. In contrast to δ-TIP3, γ-TIP, another dominant TIP form predominantly present in vegetative tissues, exclusively labeled the tonoplast of central vacuole (Figure 6, Figure S3), along with its vacuolar lumen structures previously referred to as bulbs [35,47], whose identity and function remain unknown. These bulbs were of various size, with diameter ranging from 1 to 10 µm (Figure 6(A)). To investigate the autophagic nature of bulbs, we crossed a transgenic line expressing γ-TIP-GFP with atg2 or atg5 to generate γ-TIP atg2 and γ-TIP atg5 lines. Confocal microscopy revealed the complete disappearance of bulbs in the absence of ATG2 or ATG5 (Figure 6(A)), suggesting their autophagic origin. In support of this, confocal imaging analysis of a transgenic line co-expressing γ-TIP-GFP and δ-TIP-DsRed showed partial colocalization of smaller and brighter vacuolar lumen structures and AVs (Figure 6(B), Figure S3). δ-TIP3-DsRed labeled autophagic vacuoles are involved in LD degradation To provide evidence that δ-TIP3-DsRed-labeled vacuoles are involved in LD degradation, we generated a stable transgenic line overexpressing δ-TIP3-DsRed in the OLE1 background. Confocal imaging analysis revealed the presence of LDs inside the δ-TIP3-labeled autophagic structures (Figure 7(A-D)) and a tight association of δ-TIP3-labeled AVs with LDs (Figure 7(C-F)). These results suggest the involvement of δ-TIP3-labeled AVs in the degradation of LDs. Figure 7.δ-TIP3-labeled autophagic vacuoles are involved in LD degradation. Confocal images of leaf cells of transgenic plants coexpressing OLE1-GFP (LD, green) and δ-TIP3-DsRed (autophagic vacuole, red). Seedlings were treated overnight with 0.5 µm ConA prior to imaging. Overlay of red and green fluorescence showing the presence of LDs in autophagic vacuoles (A to D) and the close association of LDs with autophagic vacuoles (C to F). Bars: 10 µm.Image A: Fluorescence micrograph with two red structures, each with a green signal, labeled 'OLE1-GFP' and 'δ-TIP3-DsRed'. Image B: Bright field micrograph showing an elongated structure with two rounded features, labeled 'Bright field'. Image C: Fluorescence micrograph with an elongated structure, green signals and red signals at ends. Image D: Bright field micrograph with a vertically aligned elongated structure and thicker central region. Image E: Fluorescence micrograph with a vertical structure, dense central green signal and multiple red signals, including a prominent one at the top. Image F: Bright field micrograph with a vertical structure and granular darker region near the upper half.Composite micrograph: 3 fluorescence (green/red) and 3 bright field images on dark/gray backgrounds. δ-TIP3-labeled autophagic vacuoles are involved in LD degradation. Confocal images of leaf cells of transgenic plants coexpressing OLE1-GFP (LD, green) and δ-TIP3-DsRed (autophagic vacuole, red). Seedlings were treated overnight with 0.5 µm ConA prior to imaging. Overlay of red and green fluorescence showing the presence of LDs in autophagic vacuoles (A to D) and the close association of LDs with autophagic vacuoles (C to F). Bars: 10 µm. To provide additional evidence that autophagy is involved in LD degradation, we generated multiple independent ATG8 OLE1 (transgenic lines overexpressing ATG8 under the control of the 35S promoter in the OLE1 background) and two of these lines with high ATG8 expression (Figure S4B) were propagated to homogeneity and subjected to further analysis. Overexpression of ATG8 is known to upregulate other autophagy genes such as ATG5, leading to increased autophagic flux [40,41]. Consistent with this, both ATG8 OLE1 lines displayed reduced TAG content (Figure S4A) and LD cluster size (Figure S4C), and increased expression of ATG5 (Figure S4B). We then crossed ATG8-1 OLE1 with atg5 sdp1 [48] and obtained sdp1 ATG8 and sdp1 atg5 ATG8 (transgenic lines carrying ATG8-1 in the sdp1 or atg5 sdp1 background, respectively, in the F2 generation. To determine whether autophagy is involved in TAG breakdown, we again conducted a pulse-chase experiment using 14C-acetate to track changes in radioactivity in TAG over a 3-d period. Rate of TAG synthesis was significantly higher in sdp1 overexpressing ATG8 compared with sdp1 (Figure 8(A)). On the other hand, TAG breakdown was slower in sdp1 compared to WT but was unaffected in atg5 (Figure 8(B)), indicating that the primary mechanism for breaking down TAG under our experimental conditions involves SDP1, which operates outside of the cell’s recycling system mediated by autophagy. This conclusion is supported by the observation that TAG breakdown rates were similar in plants lacking either SDP1 or both SDP1 and ATG5. However, when we increased the activity of autophagy by overexpressing ATG8, TAG breakdown accelerated in plants lacking SDP1 (Figure 8(B)). This effect was reversed when autophagy was blocked, confirming that the increased TAG breakdown was indeed due to autophagy. Figure 8.TAG degradation induced by ATG8 overexpression is dependent on autophagy. (A) and (B) Rates of tag synthesis (A) and degradation (B) in leaves of wild type (wt), atg5, sdp1, atg5 sdp1 and transgenic lines overexpressing ATG8 in the sdp1 (ATG8-1 sdp1) or atg5 sdp1 (ATG8-1 atg5 sdp1) background. Detached leaves were incubated with 14C-acetate for 1 h, and the radiolabeled tag was then chased for 3 d. Radiolabeled tag loss was expressed as the percentage of the label remaining after a 3-d chase period, relative to the initial amount measured at 1 h. Data are means of three independent replicates with sd. Asterisks indicate statistically significant differences from sdp1 based on Student’s t test (*p < 0.05).Two bar graphs showing label incorporation and loss in various plant lines.The image A showing a bar graph with the y-axis label “Label incorporation into TAG (DPM/mg FW/h)” and a scale from 0 to 100. The x-axis categories are WT, atg5, sdp1, ATG8-1 sdp1, atg5 sdp1 and ATG8-1 atg5 sdp1. Approximate bar heights are: WT about 66, atg5 about 59, sdp1 about 63, ATG8-1 sdp1 about 81 (marked with asterisk), atg5 sdp1 about 61, ATG8-1 atg5 sdp1 about 58. Each bar includes an error bar. The image B showing a bar graph with the y-axis label “Label loss during chase (% of the initial label)” and a scale from 0 to 21. The x-axis categories are WT, atg5, sdp1, ATG8-1 sdp1, atg5 sdp1 and ATG8-1 atg5 sdp1. Approximate bar heights are: WT about 12, atg5 about 13, sdp1 about 8 (marked with asterisk), ATG8-1 sdp1 about 15 (marked with asterisk), atg5 sdp1 about 9 (marked with asterisk), ATG8-1 atg5 sdp1 about 10 (marked with asterisk). Each bar includes an error bar. TAG degradation induced by ATG8 overexpression is dependent on autophagy. (A) and (B) Rates of tag synthesis (A) and degradation (B) in leaves of wild type (wt), atg5, sdp1, atg5 sdp1 and transgenic lines overexpressing ATG8 in the sdp1 (ATG8-1 sdp1) or atg5 sdp1 (ATG8-1 atg5 sdp1) background. Detached leaves were incubated with 14C-acetate for 1 h, and the radiolabeled tag was then chased for 3 d. Radiolabeled tag loss was expressed as the percentage of the label remaining after a 3-d chase period, relative to the initial amount measured at 1 h. Data are means of three independent replicates with sd. Asterisks indicate statistically significant differences from sdp1 based on Student’s t test (*p < 0.05). Discussion Autophagy, a cellular recycling process, functions at a basal level under normal conditions but is significantly upregulated during starvation. Our previous research indicated that autophagy plays dual roles in lipid metabolism: promoting TAG synthesis under normal conditions and breaking down LDs through a process called microlipophagy during stress [11]. Overexpression of ATG genes has been shown to activate autophagy in plants [40,41] and reduce lipid accumulation in mammalian cells [43], but whether and how ATG overexpression affects lipid metabolism and storage remains largely unexplored in plants. In this study, we found that enhanced autophagic activity by overexpressing core autophagy genes, ATG5 and ATG8 led to a marked reduction in LDs and TAG content in leaves. We additionally found that the decreased TAG and LD accumulation in transgenic lines overexpressing ATG genes is not due to the decreased TAG synthesis or impaired LD formation. Importantly, while disruption of SDP1 had no obvious effect on TAG and LD accumulation in ATG5 overexpressing lines, blocking autophagy reversed the LD-reducing effect of ATG8 overexpression, confirming autophagy’s essential role in LD degradation. Microscopic analysis revealed LDs within δ-TIP3-positive vacuoles, providing direct evidence for the involvement of these structures in LD breakdown. Several lines of evidence support that these δ-TIP3-labeled vacuoles are AVs. First, specific labeling by TIPs indicates that these structures are vacuoles; Second, a large increase in their abundance under carbon starvation [11] is consistent with the established role of autophagy under these conditions; Third, the loss of δ-TIP3-labeled vacuoles (Figures 5 and 6) and starvation-induced vacuoles [11] in ATG knockout mutants demonstrates that their formation depends on autophagy. The reduced δ-TIP3-DsRed signal in ATG knockout mutants is unlikely to result from marker mislocalization or transcript silencing. Instead, it may reflect increased turnover of fluorescent proteins in non-autophagic pathways. These observations collectively support our conclusion that these δ-TIP3-labeled vacuoles are AVs. Depending on cell types and physiological status, both macrolipophagy and microlipophagy have been observed in mammals [25,29]. In contrast, autophagy-mediated LD degradation has only been observed to occur via direct engulfment of LDs by vacuoles in a process termed microlipophagy in yeast, microalgae and plants [49]. Plant cells harbor multiple types of vacuoles, each distinguished by specific TIP proteins [31,33,34]. While these TIP markers appear to differentiate between vacuole subtypes [37,50,51], the precise roles of these organelles, particularly in autophagy and lipid droplet breakdown, remain largely elusive. Furthermore, the Arabidopsis genome encodes ten TIP isoforms, and functional roles of most of them remain unknown. Our findings demonstrate that vacuoles marked by one of the δTIP isoforms, specifically δ-TIP3, as defined by Johanson et al. [42], are of autophagic origin, therefore are bona fide AVs. It is important to note that another isoform of δ-TIPs, δ-TIP1, has been observed to localize in central vacuoles [35]. The variation in localization patterns between δ-TIP1 and δ-TIP3 is mostly likely due to their isoform-specific vacuolar functions. Overexpression of ATG5 significantly reduced the number of δ-TIP3-labeled AVs. While enhanced autophagy would typically be expected to increase autophagic vesicle formation and potentially lead to more lytic vacuoles, our results suggest that overexpression of ATG5 enhances autophagic flux, which accelerates the degradation process, including the turnover of AVs. As a result, rather than accumulating, AVs are degraded more rapidly, leading to a decrease in their number. AVs are the mature forms of autophagosomes essential for efficient cargo degradation [32,52]. They arise from the fusion of autophagosomes with vacuoles, although the precise molecular mechanisms underlying this process remain largely elusive [53–55]. Disruption of core autophagy machinery prevents AV formation, but whether this is due to impaired autophagosome generation, defective vacuole function, or compromised fusion requires further investigation. Plant cells contain distinctive, spherical structures within the central vacuole called “bulbs,” which are marked by specific TIP proteins. While previous studies have identified these bulbs [35,47], their origin and function remain unclear. Our findings suggest a clear link between bulbs and autophagy, as their formation is disrupted in atg2 and atg5 mutants. Furthermore, their morphology, their tendency to disappear during starvation [47], and their partial colocalization with AVs, support this association. However, it should be mentioned that autophagy-deficient mutants may have fewer membrane resources, which could impact tonoplast folding. Further studies are needed to clarify the identity, origin, and functional significance of these tonoplastic folds. Under normal conditions, lipophagy likely operates at a basal level to support membrane lipid recycling and general cellular maintenance, rather than serving as a major pathway for storage lipid mobilization. In the absence of SDP1 activity, this basal autophagic lipid turnover is insufficient to compensate for impaired TAG hydrolysis. Only when lipophagy is artificially upregulated does it provide an alternative route for lipid catabolism. This suggests that the regulatory mechanisms controlling lipophagy activation are tightly constrained under physiological conditions, possibly to avoid unintended degradation of essential membrane systems. Lipophagy is a selective form of autophagy targeting LDs for degradation within vacuoles. While the precise mechanisms underlying this process remain to be fully elucidated, studies in yeast have identified several proteins involved in LD degradation via selective autophagy, including Niemann-Pick type C sterol transporters [56–58] and components of the endosomal sorting complexes required for transport [59]. Several ATG proteins including Arabidopsis ATG8 and yeast Atg14 have been shown to localize on LDs [27,60,61]. Mammalian ATG14 is involved in LD degradation via interactions with adipose triglyceride lipase and its activator comparative gene identification-58 in mammalian cells [61]. In another study, ATG14 has been shown to function as an autophagic receptor mediating LD degradation [62]. In plants, ATG8 colocalizes with LDs during seed germination via interactions with the LD proteins CLO (caleosin) [27]. Overexpression of a single ATG gene often triggers the upregulation of other autophagy-related genes [40,41]. It is plausible that some of these upregulated ATG proteins contribute to the selectivity of autophagy-mediated degradation of LDs in plants. Materials and methods Plant materials and growth conditions The experiments were conducted using Arabidopsis thaliana ecotype Col-2. The atg2 and atg5 mutant lines were previously described by Fan et al. [11]. The atg7 (CS870111, SAIL_11_H07) mutant was obtained from Arabidopsis Biological Resource Center (ABRC). The sdp1 mutant line lacking functional SDP1 lipase was characterized by Eastmond et al. [23]. The atg5 sdp1 double mutant was previously described by Fan et al. [11]. ϒTIP-GFP seeds were obtained from ABRC (CS16257). For growth on plates, sterilized Arabidopsis seeds were germinated on a half-strength Murashige-Skoog (MS; Sigma-Aldrich, M9274) medium with 1% sucrose (Sigma-Aldrich, S0389) under controlled light conditions (16 h light at 22°C, 8 h dark at 18°C) with a light intensity of 50 μmol m−2 s−1. For growth on soil, seedlings grown on the nutrient medium for 10 days were then transferred to soil and grown under similar light conditions (16 h light at 22°C, 8 h dark at 18°C) but with a higher light intensity (100 μmol m−2 s−1). Fertilizer solution was applied once within the first four weeks following the transplanting and then every two weeks thereafter. For carbon starvation, 7-d-old seedlings grown on ½ x MS containing 1% sucrose plates were transferred to ½ x MS liquid medium lacking sucrose and maintained in darkness for 3 to 4 d. Plasmid construction Two binary constructs carrying the δ-TIP3 (at5g47450) or ATG8e fused to a DsRed gene under the control of 35S promoter were previously described [11]. We used DsRed-Express2 (Clontech, 632,538), a rapidly maturing, tetrameric fluorescent protein that is soluble and nontoxic [63]. These constructs were subsequently transformed into OLE1. To construct ATG5 overexpression vector, the coding region of ATG5 was amplified with primers 5’-agagagctcATGGCGAAGGAAGCGGTCAA-3’ and 5’-ctaggtaccCCTTTGAGGAGCTTTCACAAG-3,’ then cloned into modified binary vector pPZP212 [64]. In vivo 14C-acetate pulse-chase labeling In vivo 14C-acetate pulse-chase assay was carried out according to Fan et al. [11]. Briefly, rapidly growing leaves of 5-week-old plants were cut on a round disk and labeled for 1 h with 14C-acetate. After washing three times with water, the leaves were incubated further with unlabeled solution (20 mM MES, pH 5.5, one-tenth strength of MS salts (Sigma-Aldrich, M9274), and 0.01% [v:v] Tween 20 (Sigma-Aldrich, S6635) containing 1 mM cold acetate under a 16-h-light/8-h-dark cycle for 3 d. Total lipids were extracted and separated by TLC as previously described [22]. Label incorporation into total lipids was determined by liquid scintillation counting. The 14C-labeled TAG was separated by TLC and the radiolabeled TAG bands were visualized by brief exposure to iodine vapor and the radioactivity in TAG was quantified by liquid scintillation counting. Lipid and fatty acid analyses Plant samples were collected for lipid analysis and other tests approximately four hours after the onset of light. Lipids were extracted from the plant material using a solvent mixture and separated into different classes based on their chemical properties by TLC, following a previously described method [65]. Fatty acid methyl esters were prepared as described [66] and subsequently separated and identified using gas chromatography-mass spectrometry according to the established protocol [65]. The quantity of each fatty acid was accurately determined using heptadecanoic acid as an internal standard. RNA extraction and qRT-PCR Total RNAs were extracted from mature leaves of 4-week-old plants using the Direct-zol RNA kits (Zymo research, R2052) according to manufacturer’s instructions. First-strand cDNA was synthesized using M-MLV reverse transcriptase (New England Biolabs, M0253). The qRT-PCR analysis was performed with the Bio-Rad CFX96 Real time system using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, 1,725,270) according to manufacturer’s instructions. The results were normalized to the expression level of ACTIN2 (At3g18780), which was amplified with a forward primer 5’-CTTGCACCAAGCAGCATGAA-3’ and a reverse primer 5’-CCGATCCAGACACTGTACTTCCTT-3.’ The 2−ΔΔCT method [67] was used for quantitative comparisons of the gene transcripts levels. Immunoblot analysis Approximately 100–200 mg of seedlings was ground in liquid nitrogen and homogenized in 2× Laemmli sample buffer at a ratio of 1:2 (w:v). Homogenates were boiled at 70°C for 5 min and centrifuged at 16,000 g for 5 min. For ATG8 lipidation assays, 7-day-old seedlings were subjected to carbon starvation for 4 days. Proteins were separated on 15% SDS-PAGE gels containing 6 M urea in the resolving gel and transferred to PVDF membranes. ATG8 and ATG8–PE were detected using a rabbit anti-ATG8A antibody (1:4,000; Agrisera, AS142811). For NBR1 analysis, 7-day-old seedlings grown on sucrose-supplemented MS medium were subjected to carbon starvation by transfer to ½ MS medium without sucrose and incubated in the dark for 3 days. Seedlings were then treated with 0.5 µM ConA (Cayman, 11,050) or, for control samples, an equal volume of DMSO, and incubated overnight with gentle shaking on an orbital shaker in the dark. Protein samples were separated by 10% SDS – PAGE and analyzed by immunoblotting using an anti-NBR1 antibody (Agrisera, AS142805A). For analysis of DsRed-ATG8 cleavage, seedlings were subjected to 4 days of carbon starvation, and immunoblotting was performed using a monoclonal anti-RFP antibody (RF5R; Invitrogen, 3,248,684). An anti-actin antibody (MyBioSource, MBS8500610) was used as a loading control. Secondary antibodies were HRP-conjugated goat anti-mouse IgG (Invitrogen, 31,430) and HRP-conjugated goat anti-rabbit IgG (Invitrogen, 31,460). HRP activity was detected using SuperSignal West Femto maximum sensitivity substrate detection reagents (Thermo Scientific, 34,095) and visualized with Cytiva Amersham ImageQuant 800 (GE Healthcare Life Sciences). Microscopy For epifluorescence microscopy imaging, leaf tissues were observed under a Zeiss microscope (Carl Zeiss, Axiovert 200 M, Germany) with a green fluorescent protein filter (zeiss filter set 38) or a DsRED filter (Zeiss filter set 43). GFP was excited with a BP 470/40 nm and the emission was captured at BP 525/50 nm. For colocalization studies, leaf samples were mounted in water on slides and were directly examined using a Leica TCS SP5 laser scan confocal microscope with sequential scanning. GFP was excited with a wavelength of 488 nm and detected at 500–530 nm. DsRed was excited at 543 nm and detected at 560–630 nm. Co-visualization was achieved through sequential scanning mode. Images, overlays, and z-stacks were acquired and processed using the Leica analysis software LAS AF2.2.1. To examine the autophagosomes, six-d-old seedlings were transferred to liquid half-strength MS medium containing 0.5 µM ConcA and incubated around 20 h in the dark. Root tips were observed under confocal microscopy as described above. Statistical analysis Statistical significance between means was assessed using Student’s t-test, and differences were considered significant at p ≤ 0.05. Supplementary Material Supplemental materials_R5.docx