Retromer mediates Ypt7-dependent lipid droplet internalization

Summary

Lipid droplets (LDs) are dynamic organelles central to cellular energy homeostasis and stress adaptation. Maintaining LD balance is critical for cell function, yet the mechanisms governing selective LD degradation remain unclear. Here, we reveal a pathway promoting efficient vacuolar sequestering and potential degradation of a Pdr16-marked LD subpopulation, thereby linking LD metabolism to organelle communication. During nutrient limitation, the Rab GTPase Ypt7 and its guanine nucleotide exchange factor (GEF), Mon1-Ccz1, specifically localize to Pdr16-LDs. This targeting relies on a conserved amphipathic helix within Ccz1. Importantly, Ypt7 activation recruits the SNX-BAR retromer complex to the vacuole-LD interface at sites of invagination. There, retromer potentially interacts with its cargo receptor, Vps10, to promote positive membrane curvature that drives the late stage of vacuolar invagination required for LD internalization and degradation. These findings suggest a link between LD turnover, organelle interactions, and cellular adaptation to metabolic stress, providing new insight into lipid homeostasis.

Graphical abstract

Highlights

• •Lipid droplet (LD) internalization depends on the Ypt7-retromer pathway
• •Under nutrient limitation, Mon1-Ccz1 and Ypt7 target Pdr16-marked LDs
• •Active Ypt7 recruits retromer to the vacuole-LD interface
• •Retromer may interact with Vps10 to promote LD internalization and degradation

Xu et al. show that nutrient stress triggers selective vacuolar internalization of lipid droplets through Mon1-Ccz1-mediated Ypt7 activation. Activated Ypt7 recruits retromer to facilitate lipid droplet engulfment and potential degradation, revealing a pathway linking organelle communication to cellular lipid homeostasis.

Introduction

Lipids are fundamental to life, serving as building blocks for cell membranes, a concentrated form of energy storage, and critical signaling molecules. Maintaining lipid homeostasis is, therefore, important for cellular function and adaptation to fluctuating environmental conditions.¹ Lipid droplets (LDs) are dynamic organelles present in nearly all eukaryotic cells and serve as the primary storage site for neutral lipids—including triacylglycerol (TAG) and sterol esters (SEs)—within a hydrophobic core surrounded by a phospholipid monolayer.^2,3,4,5 LDs are remarkably plastic, constantly undergoing formation, growth, and breakdown to meet cellular demands.^3,6 The consumption of LDs is mediated through two major pathways: lipolysis and lipophagy.⁷ Lipolysis involves the enzymatic degradation of stored triglycerides resulting in the release of fatty acids and glycerol.⁸ Alternatively, lipophagy is the lysosome-mediated degradation of LDs that proceeds through macrolipophagy, where LDs are sequestered within autophagosomes before lysosomal fusion, and microlipophagy, which involves direct lysosomal engulfment of LDs.^9,10,11

LDs are not a homogenous population and exhibit functional diversity, driven by distinct surface protein compositions. This heterogeneity allows cells to tailor LD metabolism to specific needs.^4,12,13,14 Recent findings have identified Pdr16 as a marker for a specific LD subpopulation localized in proximity to nuclear-vacuole junctions (NVJs), contact sites between the nuclear endoplasmic reticulum (ER) and the vacuole. Two additional proteins, Ldo16 and Ldo45, are residents of the Pdr16-LDs, and they cooperate with seipin to regulate LD composition.^15,16 Notably, the Ldo16-45 complex tethers Vac8, forming vacuole-LD contact sites (vCLIPs) that is essential for macrolipophagy.^17,18

Rab proteins act as molecular switches, cycling between inactive GDP-bound and active GTP-bound states, and are activated by guanine nucleotide exchange factors (GEFs).^{19,20,21,22,23,24,25,26,27} These activated Rabs then recruit effector proteins, which mediate diverse cellular processes including vesicle formation, transport, and fusion. Rab7, a key regulator of endosomal and autophagosomal trafficking, is activated by the Mon1-Ccz1 complex. This activation is crucial for the proper function of these organelles, particularly in the fusion of multivesicular bodies/late endosomes and autophagosomes with lysosomes.^{27,28,29,30,31,32} Interestingly, Mon1-Ccz1 targeting to autophagosomes requires an amphipathic helix (AH) in Ccz1 and Atg8, while targeting to endosomes relies on negative charges of phospholipids and Rab5-GTP.^27,33,34,35 Rab7 recruits multiple effectors, including the highly conserved retromer complex, which mediates the sorting of specific cargoes like Vps10 from endosomes to Golgi. These cargoes are concentrated into distinct domains of the endosomal membrane, where tubules form to facilitate protein transport to the Golgi.^{36,37,38,39,40} Importantly, Rab7 has also been shown to play a key role during LD breakdown.^41,42,43 However, the molecular mechanisms underlying this process have remained unclear.

Here, we found that Ypt7, a Rab7 homolog, and its GEF, the Mon1-Ccz1 complex, specifically target Pdr16-positive LDs during stationary phase. Deletion of MON1, CCZ1, or YPT7 results in a strong accumulation of these Pdr16-LDs. Importantly, targeting of the Mon1-Ccz1 complex to LDs depends on the same AH within Ccz1 that is also required for its binding to autophagosomes. A point mutation in this AH specifically abolishes Ccz1 localization to LDs, leading to Pdr16-LD accumulation in yeast cells, without disrupting its function in the endocytic and autophagic pathways. Upon activation by Mon1-Ccz1, Ypt7 recruits the retromer complex to the vacuole-LD interface, particularly at tips of vacuolar invaginations. Retromer mutants accumulate Pdr16-labeled LDs together with elevated cellular levels of triglycerides, ergosterol esters, and diacylglycerol (DAG). They also display a strong increase in the number of invaginated vesicles attached to the inner vacuolar membrane, suggesting a role for retromer at the late stage for vacuole invagination, potentially by promoting positive membrane curvature to facilitate LD internalization and subsequent degradation. Vps10, a retromer cargo receptor, targets Pdr16-LDs specifically during stationary phase and is also required for Pdr16-LD entry into the vacuole, indicating that it potentially functions in concert with retromer.

Results

Mon1-Ccz1 and Ypt7 localize to Pdr16-labeled LDs during stationary phase

LDs undergo substantial remodeling during the stationary phase of cell growth, which involves changes in their distribution, composition, and interactions with other cellular components. These alterations are critical for maintaining cellular lipid homeostasis and adapting to nutrient-limiting conditions.^3,44 Previous work has shown that Mon1-Ccz1 and Ypt7 localize to endosomes under nutrient-rich conditions and to autophagosomes during nitrogen starvation^28,33,35,45 (Figure 1A). Interestingly, during nitrogen starvation, we observed very few Ccz1 signals on round structures identified as LDs, indicating limited LD targeting under these conditions. We, therefore, asked which conditions promote robust recruitment of Mon1-Ccz1 and Ypt7 to LDs and compared their subcellular distributions in both log and stationary phases. Specifically, we assessed the localization of GFP-tagged Ccz1 and Ypt7 relative to Erg6, a functionally tagged LD marker carrying a mKate fluorophore. Under nutrient-rich conditions, Ccz1 displayed a punctate pattern consistent with endosomal localization,^28,33,47 with minimal co-localization with Erg6-marked LDs (Figures 1B and 1D). Similarly, Ypt7 localized to the vacuolar membrane with a single punctum that did not co-localize with LDs (Figure 1C). However, during stationary phase, Erg6-marked LDs turned into larger ring-like formations. Notably, Ccz1 also formed ring-like structures and significantly co-localized with Erg6 (Figure 1B), as did Ypt7. Interestingly, some LDs were observed inside the vacuole, surrounded by Ypt7 (Figures 1C and 1D). However, we observed that only approximately 60% of Erg6-marked LDs were positive for Ccz1. Considering the existence of distinct LD subpopulations within a single cell, each differing in protein and lipid composition and recruitment capacity, we hypothesized that Mon1-Ccz1 and Ypt7 might target a specific LD subpopulation. Pdr16 marks a specific LD subpopulation localized near NVJs during stationary phase; Ldo45 was previously identified as a modulator of Pdr16 localization in a yeast screen, as was Ccz1.¹⁵ We, therefore, investigated whether Mon1-Ccz1 and Ypt7 are specifically recruited to Pdr16-LDs during stationary phase. To address this, we first analyzed the co-localization of GFP-tagged Ccz1 or Ypt7 with mKate-tagged Pdr16 or with the Pdr16-LD-resident proteins Ldo16/45 (LDO) in the exponentially growing and stationary phase cells. Pdr16 puncta were found to be enriched on only a fraction of cellular LDs.^15,16 During stationary phase, Pdr16 and LDO exhibited strong co-localization with Ccz1 (over 80%) and remained significantly co-localized with Ypt7, though to a lesser extent than with Ccz1 (Figures 1E–1G; Figures S1A and S1B), whereas no co-localization was observed during log phase (Figures 1E–1G). To confirm whether this recruitment is specific for Pdr16-positive LDs, we compared GFP-Ccz1 or GFP-Ypt7 co-localization with Pdr16-mKate versus total LDs labeled using the LD-specific probe FAS.⁴⁶ As expected, GFP-Ccz1 and GFP-Ypt7 co-localized strongly with Pdr16-LDs but not with the FAS-positive LD pool that lacks Pdr16 (Figures 1H and 1I). These data strongly suggest that the Mon1-Ccz1 complex and Ypt7 are selectively targeted to Pdr16-LDs during stationary phase.

Figure 1.

Recruitment of Mon1-Ccz1 and Ypt7 to Pdr16-LDs during stationary phase

(A) A model of Mon1-Ccz1 targeting different organelles; details are in the text. Mon1-Ccz1 (MC1) binds to Rab5 (red) on late endosomes and interacts with Atg8 (green) on autophagosomes.

(B and C) Localization of Ccz1 or Ypt7 relative to Erg6. Cells expressing GFP-tagged Ccz1 or Ypt7 and mKate-tagged Erg6 were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy, and individual slices are shown. Scale bars, 5 μm. Bottom: schematic of the observed Ypt7 localization relative to vacuole.

(D) Determination of the percentage of Erg6 puncta co-localizing with Ccz1 or Ypt7. Erg6 dots (n ≥ 1,000), Ccz1 dots (n ≥ 500), and Ypt7 dots (n ≥ 250) were quantified using a custom graphical user interface based on ImageJ built-in routines (Arlt et al., 2015; Gao et al., 2018, 2025). Single cells were segmented from bright-field images, and protein dot signals were identified on sum projections of deconvolved image stacks; co-localization was assessed by counting particles with >50% overlapping area, with manual corrections to reduce background artifacts. Error bars represent standard deviation (SD). Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(E and F) Localization of Ccz1 or Ypt7 relative to Pdr16. Cells expressing GFP-tagged Ccz1 or Ypt7 and mKate-tagged Pdr16 were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy, and individual slices are shown. Scale bars, 5 μm.

(G) Determination of the percentage of Pdr16 puncta co-localizing with Ccz1 or Ypt7. Pdr16 dots (n ≥ 400), Ccz1 dots (n ≥ 500), and Ypt7 dots (n ≥ 200) were quantified by ImageJ, as described in Figure 1D. Error bars represent the SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(H) Localization of Ccz1 or Ypt7 relative to Pdr16 and FAS-stained LDs. Cells expressing GFP-Ccz1 or GFP-Ypt7 together with Pdr16-mKate were cultured in synthetic medium to stationary phase, stained with the neutral-lipid dye FAS,⁴⁶ and examined by fluorescence microscopy. Scale bars, 5 μm.

(I) The percentage of GFP-Ccz1 or GFP-Ypt7 puncta that (1) co-localized with both Pdr16-mKate and FAS-stained LDs, (2) co-localized with Pdr16 alone, or (3) co-localized with FAS-stained LDs alone. Pdr16 puncta (n ≥ 300), Ccz1 puncta (n ≥ 300), Ypt7 puncta (n ≥ 100), and FAS-positive LDs (n ≥ 600) were counted using ImageJ, as described for Figure 1D.

(J) Localization of Ccz1 relative to Pdr16 in wild-type and vac8Δ cells. Cells expressing GFP-tagged Ccz1 and mKate-tagged Pdr16 were grown in a synthetic medium to stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm.

(K) The percentage of Pdr16 puncta co-localizing with Ccz1 was determined in wild-type and vac8Δ cells. Pdr16 dots (n ≥ 500) and Ccz1 dots (n ≥ 500) were quantified by ImageJ, as described in Figure 1D. Error bars represent the SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(L) Localization of Pdr16 relative to BODIPY-stained LDs in wild-type and mutant cells. Cells expressing mKate-tagged Pdr16 were grown in a synthetic medium to stationary phase, with LDs stained with BODIPY 493/503 (green) and analyzed by fluorescence microscopy. Representative slides are shown. Scale bars, 5 μm.

(M) Pdr16 puncta and BODIPY-stained LD count per cell. Cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent the SD. ^∗, p < 0.05; ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(N) The percentage of Pdr16 puncta co-localizing with BODIPY-stained LDs was determined from (J). Pdr16 dots (n ≥ 600) and BODIPY dots (n ≥ 1000) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ns, p > 0.05 (two-tailed Student’s t test).

LDO form a tether with Vac8 (the vCLIP complex) that links Pdr16-positive LDs to the vacuole. Deletion of VAC8 or LDO eliminates the specialized LD subpopulation that normally attaches to the vacuole and accumulates at the NVJ during starvation.^17,18 To further demonstrate whether Mon1-Ccz1 specifically targets Pdr16-LDs, we examined GFP-Ccz1 or GFP-Ypt7 co-localization with Pdr16-mKate in the wild-type, vac8Δ, and ldoΔ cells. In wild-type cells, GFP-Ccz1 robustly co-localized with Pdr16-mKate, as previously shown (Figures 1E and 1G), whereas this co-localization was completely abolished in vac8Δ cells during stationary phase (Figures 1J and 1K). Similarly, in wild-type cells, GFP-Ypt7 co-localized with both Pdr16-mKate and FAS-stained LDs. However, in ldoΔ cells, Pdr16 became cytosolic and detached from LDs,^15,16,17 while Ypt7 completely lost its localization to FAS-stained LDs (Figures S1C and S1D). Taken together, these data demonstrate that the Mon1-Ccz1 complex and Ypt7 specifically localize to the Pdr16-LD subpopulation during stationary phase.

Given that Mon1-Ccz1 activates Ypt7/Rab7 and promotes tethering and fusion of late endosomes and autophagosomes with lysosomes/vacuoles for cargo degradation,^27,28,33 we examined whether Ypt7 influences LD abundance. To test this, we analyzed the number and distribution of Pdr16-LDs and total LDs (stained with a neutral lipid dye, BODIPY) in stationary phase wild-type, mon1Δ, ccz1Δ, and ypt7Δ cells. Interestingly, Pdr16-LDs accumulated significantly, while the number of BODIPY-stained LDs increased only slightly, though significantly, in the mutant cells (Figures 1L and 1M). Importantly, the co-localization of Pdr16-LDs with BODIPY-stained LDs remained unchanged in both wild-type and mutant cells (Figure 1N). Therefore, these data suggest that Ypt7 contributes to the regulation of LD abundance.

A conserved AH in Ccz1 facilitates the recruitment of Mon1-Ccz1 to LDs

To determine the function of Ypt7 on LDs, we next investigated how Mon1-Ccz1 targets LDs. Mon1-Ccz1 is recruited to endosomes by Rab5 proteins and to autophagosomes through interaction with Atg8 (LC3 in mammals).^33,34 We, therefore, first investigated whether the endocytic or autophagy pathway, specifically Rab5 or Atg8, is required for Mon1-Ccz1 targeting to LDs during stationary phase. To test this, we analyzed the co-localization of GFP-Ccz1 with Erg6-mKate in wild-type and various mutant cells during both log and stationary phases. Seipin is a protein involved in LD formation, and its absence disrupts normal LD growth and maturation, resulting in the formation of larger, aberrant LDs.^{48,49,50,51,52} Notably, in cells lacking Sei1, the yeast seipin homolog, Ccz1 formed elongated structures in stationary phase, rather than the ring-like structures overlapping and surrounding LDs in wild-type cells. However, these elongated Ccz1 structures still co-localized with Erg6-marked LDs (Figures S2A–S2C). Mon1-Ccz1 targets PI3P-positive endosomal membranes via interaction with Rab5 proteins.⁵³ Consistent with previous studies, loss of Mon1, Vps21/Rab5, or the CORVET subunit Vps8 (a Rab5 effector protein) strongly re-localized Ccz1 to the cytosol under growing conditions^28,45 (Figure S2A). In contrast, Ccz1 remained punctate, likely representing endosomes, in atg1Δ or atg8Δ cells with impaired autophagy. Interestingly, in these mutants, Ccz1 was still present on LDs during stationary phase, similar to wild-type cells (Figures S2A and S2B). These data suggest that Mon1-Ccz1 localization to LDs during stationary phase does not require the endocytic or autophagy pathways or its known interacting partners, Rab5 proteins and Atg8; however, normal LD biogenesis is essential for its distribution on LDs.

Proteins that target LDs from the cytoplasm often contain AHs. These AHs, characterized by opposing hydrophilic and hydrophobic faces, enable proteins to bind to the LD surface by inserting their hydrophobic residues into the monolayer—a mechanism important for various cellular processes involving lipid storage and metabolism.⁵⁴ Interestingly, a conserved C-terminal AH was identified in Ccz1, which is required for Mon1-Ccz1 function in autophagy but not endosomal maturation³⁵ (Figure 2A). We asked whether Mon1-Ccz1 targeting to LDs depends on this AH. To examine this, we first tested whether the AH alone is sufficient to target Mon1-Ccz1 to LDs. We compared the localization of a GFP-tagged Ccz1 fragment (amino acids 270–403) containing the AH³⁵ with that of full-length Ccz1, relative to Erg6-mKate, under both growing and stationary phase conditions. Intriguingly, the Ccz1 fragment strongly co-localized with LDs, even in log phase, whereas no co-localization was observed with full-length Ccz1 under this condition. Furthermore, this co-localization was significantly higher with the Ccz1 fragment than with full-length Ccz1 during stationary phase (Figures 2B and 2C). These data strongly indicate that the AH in Ccz1 is sufficient to target Mon1-Ccz1 to LDs. To further investigate the direct interaction between the Ccz1 AH and lipid surfaces, we generated a GFP-Ccz1 construct with an inverted AH sequence. This inversion was designed to preserve the overall physicochemical properties of the AH, including its amphipathic character (Figure S2D), and, therefore, minimize disruption of its membrane adsorption properties.⁵⁵ Notably, this inverted construct still localized to Erg6-marked LDs, although with reduced co-localization compared with wild-type Ccz1 (Figures S2E and S2F). This finding supports our model in which the Ccz1 AH contributes to the localization of Mon1-Ccz1 to LDs through direct interaction with the lipid surface. To further assess the function of this AH, we generated a point mutation in the AH (Ile341Ala), which introduced a less hydrophobic residue into the hydrophobic face to avoid complete disruption of the structure that would impair autophagy (Figure 2D). Strikingly, in the ccz1^Ile341Ala mutant, no co-localization of GFP-Ccz1 with Erg6-marked LDs or Pdr16-positive LDs was observed during stationary phase (Figures 2E–2G). These data demonstrate that the specific amphipathic character of the AH in Ccz1 facilitates its targeting to LDs. Furthermore, Pdr16-LDs accumulated significantly in the ccz1^Ile341Ala mutant, while the number of total Erg6-labeled LDs increased only slightly, though significantly (Figures 2E–2H). Importantly, mCherry-tagged Atg8, an autophagy marker, was still delivered to the vacuole lumen in the ccz1 mutant under nitrogen-starved conditions to induce autophagy, indicating that the autophagy pathway remained functional, even though it also requires the AH³⁵(Figures 2I and 2J). Taken together, these data support the model that specific features of the AH in Ccz1 contribute to the recruitment of Mon1-Ccz1 to both autophagosomes and Pdr16-LDs, where the Mon1-Ccz1 complex subsequently activates Ypt7 to potentially promote their degradation.

Figure 2.

The amphipathic helix in Ccz1 facilitates its targeting to LDs

(A) Schematic of the AH of the C-terminal part of Ccz1. Red and green indicate the AH and loop region, respectively.

(B) Localization of Ccz1 relative to Erg6 in wild-type and mutant cells. Cells expressing mKate-tagged Erg6 carrying the Ccz1 encoding plasmid pRS406-TEFpr-GFP-CCZ1 or pRS406-TEFpr-GFP-CCZ1 (aa270-403) were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm.

(C) The percentage of Erg6 puncta co-localizing with Ccz1 was determined. Erg6 dots (n ≥ 1000) and Ccz1 dots (n ≥ 500) were quantified by ImageJ, as described in Figure 1D. Error bars represent standard deviation (SD). Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(D) Helical wheel projection of the putative AH of Ccz1.

(E and F) Localization of Ccz1 relative to Erg6 or Pdr16 in wild-type and ccz1^Ile341Ala cells. Cells expressing GFP-tagged Ccz1 and mKate-tagged Erg6 or Pdr16 were grown in a synthetic medium to stationary phase and analyzed by fluorescence microscopy, and individual slices are shown. Scale bars, 5 μm.

(G) The percentage of Erg6 or Pdr16 puncta co-localizing with Ccz1 was determined in wild-type and ccz1^Ile341Ala cells. Erg6 dots (n ≥ 1,000), Pdr16 dots (n ≥ 400), and Ccz1 dots (n ≥ 500) were quantified by ImageJ, as described in Figure 1D. Error bars represent the SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(H) Quantification of the number of Erg6 and Pdr16 puncta per cell in wild-type and ccz1^Ile341Ala cells. Cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent the SD. ^∗, p < 0.05; ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(I) Localization of Ccz1 relative to Atg8 under nitrogen starvation conditions. Cells expressing GFP-tagged Ccz1 and mCherry-tagged Atg8 were grown in synthetic medium without nitrogen (labeled as N-starved) for 2 h and analyzed by fluorescence microscopy and are shown via individual slices. Scale bars, 5 μm.

(J) Quantification of mCherry-Atg8 localization from data in (I). Cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent the SD. ns, p > 0.05 (two-tailed Student’s t test).

Given the high degree of conservation of the Mon1-Ccz1 complex, we further investigated whether it maintains a similar function in mammalian cells. We first assessed the effects of CCZ1 knockdown in mammalian COS7 cells using small interfering RNAs (siRNAs), confirming reduction of CCZ1 mRNA levels by quantitative reverse-transcription PCR (RT-qPCR) (Figures S3A and S3B). To determine the potential role of MON1-CCZ1 in mammalian LD regulation, we quantified LD number and size in both wild-type and CCZ1-knockdown cells. As COS7 cells exhibit limited LD accumulation under basal conditions, we stimulated LD biogenesis with a 6-h treatment of 200 μM sodium oleate (OA), followed by immunofluorescence analysis (Figure S3A). In agreement with observations in CCZ1 deletion in yeast (Figures 1L and 1M), CCZ1 knockdown resulted in a significant increase in the number of LDs (Figures S3A and S3E). Interestingly, the induced LDs in CCZ1-knockdown cells were significantly smaller than those observed in wild-type cells (Figures S3A and S3F). Taken together, these data suggest a potentially conserved role of the Mon1-Ccz1 complex in mammalian LD abundance.

To investigate the role of the AH of CCZ1 in targeting the human MON1-CCZ1 complex to LDs, we re-expressed wild-type and CCZ1 mutant in CCZ1-knockdown COS7 cells. We utilized a construct encoding a CCZ1 protein with a Leu287Ala mutation within the AH domain, mirroring the Ile341Ala mutation in the yeast Ccz1 (Figures 2D and S3C). The wild-type CCZ1 was mostly found in the cytoplasm and additionally exhibited partial localization to LDs, which rescued the number and size of LDs induced by CCZ1 knockdown. In contrast, the CCZ1 (Leu287Ala) mutant localized less to LDs and did not rescue the altered LD phenotype (Figures S3D–S3F). Consistent with these findings, flotation experiments indicated an association of the wild-type MON1-CCZ1 complex with LDs labeled by PLIN3, while the mutant remained largely in the soluble fraction (Figure S3G). These results suggest that the function and targeting of MON1-CCZ1 to LDs is potentially conserved in mammalian cells.

Mon1-Ccz1 is required for the entry of Pdr16-LDs into the vacuole via invagination

To directly test the potential function of Mon1-Ccz1 and Ypt7 in LD internalization, we analyzed the localization of Pdr16-mNeon-labeled LDs relative to the vacuole in wild-type and ccz1^Ile341Ala cells. During log phase, one or two Pdr16-labeled LDs was observed adjacent to the vacuole. In contrast, during stationary phase, Pdr16-LDs partially localized within the vacuolar lumen in wild-type cells but strongly accumulated at the vacuolar periphery in ccz1^Ile341Ala cells (Figures 3A and 3B). These findings indicate that Pdr16-LD sorting into the vacuole lumen is impaired in ccz1^Ile341Ala cells, supporting our model that Ypt7 is potentially required for Pdr16-LD degradation. Additionally, the mutant cells exhibited reduced viability during stationary phase, whereas viability was comparable to that of wild-type during log phase (Figure 3C). These data suggest that the Ypt7-mediated pathway is important for cell survival under energy-demanding conditions.

Figure 3.

Mon1-Ccz1 is required for Pdr16-LD entry into the vacuole via invagination

(A) Localization of Pdr16 relative to the vacuole in wild-type and mutant cells. Cells expressing mNeon-tagged Pdr16 were grown in a synthetic medium to stationary phase for 24 h, vacuoles were stained with FM4-64, and analysis was performed using fluorescence microscopy. Scale bars, 5 μm and 1.2 μm (inset).

(B) Quantification of Pdr16-mKate localization from data in (A). Cells (n ≥ 200) were quantified by ImageJ, using the tool of region of interest (ROI). Bar graphs represent averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(C) Quantification of cell death rate in wild-type and mutant cells from data in (A). Cell death was assessed by quantifying increased autofluorescence in the green spectrum, a characteristic of dead yeast cells, using ImageJ software. ^∗∗∗, p < 0.001; ns, p > 0.05 (two-tailed Student’s t test).

(D–F) Analysis of LDs in wild-type and ccz1^Ile341Ala cells during stationary phase by cryo-EM (see STAR Methods). Scale bars, 1 μm and 0.25 μm (inset). Right: schematic representation of observed LD localization.

(G) Localization of Pdr16 relative to Ldo16/45 (LDO) in wild-type and mutant cells. Cells expressing mNeon-tagged Pdr16 and mKate-tagged LDO were grown in a synthetic medium to stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm.

(H) Determination of the percentage of Pdr16 co-localizing with LDO. Pdr16 dots (n ≥ 150) and LDO dots (n ≥ 150) were quantified by ImageJ, as described in Figure 1D. Error bars represent standard deviation (SD). Bar graphs represent the averages from three independent experiments. ns, p > 0.05 (two-tailed Student’s t test).

To further investigate LD morphology at higher resolution, we employed cryo-electron microscopy (cryo-EM) to compare LD distribution in wild-type and ccz1^Ile341Ala mutant cells during stationary phase. In wild-type cells, we observed some LDs, typically one or two, localized adjacent to NVJs, or isolated free LDs within the vacuole (Figures 3D and 3E). Notably, some LDs appeared entrapped within membrane structures inside the vacuole, potentially representing invaginated vacuolar membranes (Figures 3E and 1C). In contrast, ccz1^Ile341Ala mutant cells exhibited a marked accumulation of LDs along NVJs, with a significant proportion surrounded by invaginating vacuolar membrane, consistent with the localization of Pdr16-positive LDs (Figure 3F). These results suggest that Ypt7 may function at a late stage of vacuolar invagination toward the potential LD degradation. Next, we asked whether ccz1^Ile341Ala mutation disrupts the vCLIP tether and thereby blocks Pdr16-LD entry into the vacuole. To test this, we examined the localization of Pdr16-mNeon relative to LDO-mKate in wild-type and ccz1^Ile341Ala cells during stationary phase. Strikingly, the mutant cells retained the strong co-localization between Pdr16 and LDO, just as observed in wild-type cells (Figures 3G and 3H). These results indicate that the vCLIP tether localization is unchanged in the ccz1^Ile341Ala background, implying that Ypt7 may function downstream of vCLIP in this pathway.

Retromer, a Ypt7 effector, targets Pdr16-LDs at the vacuolar surface during stationary phase

Rab proteins activate downstream signaling by recruiting effector proteins to cellular membranes when bound to GTP.^24,56 We, thus, investigated the Ypt7 effector protein and its potential role in LD degradation. Ypt7 regulates multiple cellular processes through interactions with different effectors: it controls vacuole membrane homeostasis via Ivy1,^57,58,59 promotes fusion of late endosomes and autophagosomes with vacuoles through the HOPS complex (homotypic fusion and vacuole protein sorting),^60,61,62,63 and facilitates endosome maturation by interacting with the retromer complex to recycle membrane proteins^36,38,64 (Figure 4A). Among them, the highly conserved retromer complex retrieves specific proteins from endosomes or vacuoles to the Golgi, utilizing tubule formation for transport (Figure 4A). In yeast cells, a retromer consists of a core of Vps35, Vps26, and Vps29, also called cargo-selecting complex (CSC), that associates with the tubule/vesicle-forming complex (SNX; Vps5–Vps17)^39,40,65 or forms a subcomplex with the PX-only protein Snx3 to sort a subset of membrane proteins.^66,67,68,69 To identify Ypt7 effectors on LDs, we initially screened for the presence of known Ypt7 effectors on LDs during stationary phase. Surprisingly, only the retromer subunit Vps35, not Ivy1 or HOPS subunit Vps41, was significantly adjacent to LDs marked by Erg6 at the vacuolar surface under these conditions (Figures 4B–4E).

Figure 4.

Retromer is recruited to Pdr16-LDs at the LD-vacuole interface during stationary phase

(A) Model of Ypt7 effectors.

(B–D) Localization of Vps35 or Vps41 or Ivy1 relative to Erg6. Cells expressing mKate-tagged Erg6 and mNeon-tagged Vps35 or Vps41 or mGFP-tagged Ivy1 were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy, and individual slices are shown. Scale bars, 5 μm and 0.8 μm (inset).

(E) The percentage of Vps35 or Vps41 or Ivy1 puncta co-localizing with Erg6 was determined. Vps35 dots (n ≥ 200), Vps41 dots (n ≥ 200), Ivy1 dots (n ≥ 200), and Erg6 dots (n ≥ 1,000) were quantified by ImageJ, as described in Figure 1D. Error bars represent standard deviation (SD). Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(F) Localization of Vps29 relative to the periphery of Nvj1. Cells expressing mScarlet-tagged Nvj1 and mNeon-tagged Vps29 were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy, and individual slices are shown. Scale bars, 5 μm and 0.8 μm (inset).

(G) The percentage of Vps29 or Vps35 co-localizing with the periphery of Nvj1 was determined. Vps29 dots (n ≥ 150), Vps35 dots (n ≥ 250), and Nvj1 dots (n ≥ 150) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(H) Localization of Vps35 relative to Ldo16/45 (LDO). Cells expressing mNeon-tagged Vps35 and mKate-tagged LDO were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy. Scale bar, 5 μm and 0.3 μm (inset).

(I) The percentage of Vps35 co-localizing with LDO was determined. Vps35 dots (n ≥ 250) and LDO dots (n ≥ 150) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(J) Localization of Vps35 relative to Pdr16 in wild-type and vac8Δ cells. Cells expressing mNeon-tagged Vps35 and mKate-tagged Pdr16 were grown in a synthetic medium to stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm and 0.3 μm (inset).

(K) The percentage of Vps35 puncta co-localizing with Pdr16 was determined in wild-type and vac8Δ cells. Vps35 dots (n ≥ 250) and Pdr16 dots (n ≥ 500) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(L) Schematic of the observed retromer localization during stationary phase.

Given that Mon1-Ccz1 localizes to Pdr16-positive LDs and subsequently activates Ypt7 at those sites, we hypothesized that retromer would be recruited to the periphery of NVJs, where Pdr16-LDs localize. To test this, we analyzed the localization of mNeon-tagged retromer subunits Vps35 and Vps29 relative to NVJs labeled by Nvj1-mKate in both log and stationary phases. We found that both Vps35 and Vps29 strongly localized to the periphery of NVJs specifically under stationary phase conditions (Figures 4F and 4G; Figure S4A). As LDO are required for LD positioning near NVJs and vCLIP formation,^17,18 we further examined the co-localization of mNeon-tagged Vps35 with LDO-mKate. Consistent with the NVJ localization data, Vps35 significantly co-localized with LDO (Figures 4H and 4I). Furthermore, we assessed co-localization between Vps35 and Pdr16 in wild-type and vac8Δ cells. As observed with Ccz1 (Figures 1J and 1K), Vps35 and Pdr16 significantly co-localized in wild-type cells, but this co-localization was absent in vac8Δ cells (Figures 4J and 4K). Additionally, as with Ypt7, Vps35 co-localized with Pdr16-mKate and FAS-stained LDs in wild-type cells but completely lost localization to FAS-stained LDs in ldoΔ cells (Figures S4B and S4C). Collectively, these data support a model in which retromer functions as a Ypt7 effector, localizing to Pdr16-LDs and coordinating their interaction with the vacuole, thereby facilitating their potential subsequent degradation (Figure 4L).

SNX-BAR retromer functions at the late stage of vacuolar invagination and is required for LD internalization

The retromer complex, together with the SNX-BAR proteins Vps5–Vps17, form an SNX-BAR retromer that shuttles specific cargoes between endosomes (or vacuoles) and the Golgi. In addition, retromer interacts with Atg18 to facilitate membrane recycling from the vacuole.^70,71,72 Because Pdr16-labeled LDs accumulate in mon1Δ, ccz1Δ, ccz1^Ile341Ala, and ypt7Δ cells, we hypothesized that Mon1-Ccz1 activates Ypt7 on Pdr16-LDs at the vacuolar surface, and active Ypt7 then interacts with retromer at the LD-vacuole interface to promote LD internalization for potential subsequent consumption. To test this, we analyzed the localization and number of Pdr16-positive LDs relative to total LDs (stained with BODIPY) in wild-type, the SNX-BAR retromer mutants (vps35Δ and vps5Δ), and CROP mutant (atg18Δ) cells. We found that the number of Pdr16-positive LDs was significantly increased in both retromer mutants but not in the CROP mutant. Importantly, the co-localization of Pdr16 puncta with BODIPY-stained LDs remained unchanged in all strains examined (Figures 5A–5C). Furthermore, we assessed the localization of mNeon-tagged Vps5 or Atg18 with mScarlet-tagged Nvj1. Consistent with the Pdr16-LD accumulation data, only Vps5, but not Atg18, strongly localized to the periphery of NVJs specifically under stationary phase conditions (Figures 5D–5F), mirroring the localization of Vps35 and Vps29 (Figures 4F and 4G; Figure S4A). Taken together, these data support our hypothesis that SNX-BAR retromer functions as a Ypt7 effector involved in Pdr16-LD internalization.

Figure 5.

Disruption of SNX-BAR retromer function leads to the accumulation of Pdr16-LDs and NVJs

(A) Localization of Pdr16 relative to BODIPY-stained LDs in wild-type and mutant cells. Cells expressing mKate-tagged Pdr16 were grown in a synthetic medium to stationary phase, with LDs stained with BODIPY 493/503 (green) and analyzed by fluorescence microscopy. Representative slides are shown. Scale bars, 5 μm.

(B) The number of Pdr16 puncta and BODIPY-stained LDs per cell. Cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent standard deviation (SD). ns, p > 0.05, ^∗, p < 0.05; ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(C) The percentage of Pdr16 puncta co-localizing with BODIPY-stained LDs was determined from (J). Pdr16 dots (n ≥ 600) and BODIPY dots (n ≥ 1000) were quantified by ImageJ as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ns, p > 0.05 (two-tailed Student’s t test).

(D and E) Localization of Vps5 or Atg18 relative to the periphery of Nvj1. Cells expressing mScarlet-tagged Nvj1 and mNeon-tagged Vps5 or Atg18 were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm.

(F) The percentage of Vps5 or Atg18 co-localizing with the periphery of Nvj1 was determined. Vps5 dots (n ≥ 100), Atg18 dots (n ≥ 50), and Nvj1 dots (n ≥ 150) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001; ns, p > 0.05 (two-tailed Student’s t test).

(G) Localization of Pdr16 relative to the periphery of Nvj1 in wild-type and vps35Δ cells. Cells expressing mKate-tagged Pdr16 and mNeon-tagged Nvj1 were grown in a synthetic medium to stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm.

(H) Pdr16 and Nvj1 puncta counts per cell in wild-type and vps35Δ cells. Pdr16 dots (n ≥ 500) and Nvj1 dots (n ≥ 150) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

To investigate the role of retromer in the internalization of LDs and their potential subsequent degradation, we examined whether retromer mutants influence the Pdr16-positive LDs and NVJs. We analyzed the localization of Pdr16-mKate relative to Nvj1-mNeon in wild-type and vps35Δ cells during stationary phase. Consistent with previous observations, we found an increased number of Pdr16-LDs in the vps35 mutant. Surprisingly, NVJ expansion, as indicated by Nvj1 labeling, was impaired, and the number of NVJs was also increased in the mutant (Figures 5G and 5H). These data suggest that the SNX-BAR retromer complex may stabilize NVJs and potentially promote the association of Pdr16-LDs with both NVJs and the vacuole, ultimately facilitating LD internalization.

To investigate this further, we analyzed the localization of Vps35 relative to Ypt7 during both log and stationary phases. During log phase, Ypt7 exhibited a single punctum on the vacuolar surface that co-localized with Vps35 (Figure 6A). In stationary phase, Ypt7 accumulated significantly at the vacuolar surface, and Vps35 strongly co-localized with Ypt7 at the vacuolar membrane (Figures 6B–6D) and at the tips of invaginated vacuoles (Figure 6B). Furthermore, time-lapse imaging during stationary phase revealed that Ypt7 and Vps35 initially co-localized at the tips of invaginated vacuoles and were subsequently internalized within Ypt7-positive membrane structures (Figure 6E; Video S1). This dynamic process resembles the entrapment of LDs within membrane structures inside the vacuole (Figure 6E; Video S1), with these LDs likely surrounded by the Ypt7-positive membrane (Figure 1C). This specific localization and dynamic movement of SNX-BAR retromer at the vacuolar membrane prompted us to investigate whether retromer directly functions in vacuolar invagination. To test this, we analyzed vacuoles in wild-type and vps35Δ cells during log and stationary phases. Strikingly, vps35Δ cells exhibited a massive accumulation of invaginated vesicles that remained attached to the vacuole membrane specifically during stationary phase (Figures 6F, 6G, and 6I). Consistent with LD accumulation outside the vacuole in vps35Δ cells (Figure 6H) and the blocked vacuolar invagination at the late stage observed in the ccz1^Ile341Ala mutant (Figure 3), these data suggest that retromer promotes vacuolar invagination at the late stage. To assess whether the retromer complex is required for potential subsequent LD turnover, we measured neutral-lipid levels in wild-type cells and the cells lacking Vps35 during both log and stationary phases. We report three main observations. (1) During the stationary phase, TAG levels increased in both wild-type and vps35Δ cells; however, the increase was significantly more pronounced in mutant cells (Figure 6J). Given that steady-state TAG levels represent the balance between synthesis and degradation, the greater accumulation in the mutants may imply an impaired degradation process, suggesting that, in wild-type cells, a fraction of TAG is degraded during stationary phase. (2) DAG levels remained unchanged in wild-type cells across both growth phases. In contrast, vps35Δ cells showed a significant increase in DAG levels specifically during the stationary phase, indicating that DAG may act as a transient intermediate in the LD degradation pathway from TAG in the mutant cells (Figure 6J). (3) Ergosterol ester levels did not increase in wild-type cells during stationary phase. However, vps35Δ cells exhibited an accumulation of ergosterol esters, even under normal growth conditions, with a further marked increase during the stationary phase. The absence of ergosterol ester accumulation in wild-type cells, despite ongoing lipid synthesis, suggests active ergosterol turnover, a process that appears to be disrupted in the mutant cells (Figure 6J). Taken together, this lipid profile demonstrates that loss of retromer function may impair LD degradation, supporting the model in which the SNX-BAR retromer is essential for the processing of LDs at the vacuolar surface for subsequent degradation. Importantly, the vCLIP tether remains correctly positioned in retromer-deficient cells. As observed for both ccz1^Ile341Ala and wild-type strains (Figures 3G and 3H), LDO-mKate continued to co-localize strongly with Pdr16-mNeon in vps35Δ cells (Figures S4D and S4E). Thus, the localization of the vCLIP complex is unaffected by retromer loss, suggesting that Ypt7-retromer may function downstream of vCLIP in the LD-vacuole degradation pathway. Taken together, we propose that Mon1-Ccz1 activates Ypt7 on Pdr16-LDs at the vacuolar surface, facilitating SNX-BAR retromer recruitment to the LD-vacuole interface. This recruitment drives late-stage vacuolar invagination by inducing positive membrane curvature, forming tubules that “hug” LDs and ultimately mediate their internalization into the vacuole where they can be degraded.

Figure 6.

Retromer functions at the late stages of vacuole invagination

(A and B) Localization of Vps35 relative to Ypt7. Cells expressing mKate-tagged Vps35 and GFP-tagged Ypt7 were grown in a synthetic medium to log and stationary phases and analyzed by fluorescence microscopy; individual slices are shown. Scale bars, 0.5 μm and 0.1 μm (inset). Right: schematic of the observed retromer localization relative to Ypt7 during stationary phase.

(C) Ypt7 and Vps35 puncta count per cell. Cells (n ≥ 150) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent standard deviation (SD). ^∗∗∗, p < 0.001, ns, p > 0.05 (two-tailed Student’s t test).

(D) The percentage of Vps35 puncta co-localizing with Ypt7 was determined. Vps35 dots (n ≥ 300) and Ypt7 dots (n ≥ 150) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(E) Time-lapse analysis of the localization of Vps35 relative to Ypt7. Cells expressing mKate-tagged Vps35 and GFP-tagged Ypt7 were grown in a synthetic medium to stationary phase and analyzed by fluorescence microscopy. Scale bars, 0.5 μm. Right: schematic of observed retromer localization relative to Ypt7 during stationary phase.

(F and G) Vacuole morphology was analyzed in wild-type and vps35Δ cells. The cells were grown in a synthetic medium to log or stationary phase, and vacuoles were stained with FM4-64 and analyzed by fluorescence microscopy. Scale bars, 4 μm and 1.8 μm (inset). Right: schematic of the observed vacuole morphologies.

(H) Vacuole morphology and LDs were analyzed in wild-type and vps35Δ cells. The cells were grown in a synthetic medium to stationary phase for 48 h; vacuoles were stained with FM4-64, and LDs were stained with BODIPY 493/503 (green), and they were analyzed by fluorescence microscopy. Scale bars, 2.5 μm.

(I) Quantification on a particular vacuolar morphology from data in (G). The cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent SD. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(J) Quantification of neutral-lipid species in wild-type and vps35Δ cells during log and stationary phases. Neutral lipids—triacylglycerol (TAG), ergosterol esters (EG esters), and diacylglycerol (DAG)—were extracted from stationary phase cultures using a chloroform:methanol mixture (see STAR Methods). Bar graphs represent averages from three independent experiments. Error bars represent SD. ns, p > 0.05 ^∗, p < 0.05; ^∗∗, p < 0.01; ^∗∗∗∗, p < 0.0001 (two-tailed Student’s t test).

Vps10, a known retromer cargo, is required for the internalization of Pdr16-LDs into the vacuole

Given that retromer remodels membranes in the presence of its specific cargoes,⁷³ we hypothesized that retromer functions with its cargo to promote vacuolar membrane curvature surrounding LDs, thereby facilitating LD degradation via vacuolar invagination. To identify potential retromer cargoes, we employed label-free vacuolar proteomics,^74,75 allowing for comprehensive identification of vacuolar proteins in comparative analyses. We reasoned that the impaired retromer function in a mutant would lead to cargo accumulation on the vacuole. To ensure data reliability, we analyzed vacuolar proteomes from wild-type and vps35Δ cells during stationary phase, using both Orbitrap Eclipse and timsTOF Pro mass spectrometry (Figure 7A). We then plotted the accumulated cargoes, identified by comparing vps35Δ to wild-type ratios from Orbitrap Eclipse (x axis) and timsTOF Pro (y axis). Interestingly, both analyses revealed strong accumulation of the known retromer cargo Vps10 and its homolog Vth1, with less pronounced accumulation of cargo Pep12. Additionally, we observed increased abundance of glucose transporters (Hxt7, Hxt4, and Hxt3), the peptide transporter Pdr2, and plasma membrane proteins (Lem3 and Sso1) (Figure 7B). Given that Vps10 is a specific cargo of SNX-BAR retromer and facilitates retromer-mediated membrane remodeling,⁷³ we initially investigated whether Vps10 accumulates at the vacuole in retromer mutant cells. To examine this, we analyzed the localization of mGFP-tagged Vps10 relative to the vacuole in wild-type and vps35Δ cells during log and stationary phases. Consistent with previous reports,⁷⁶ Vps10 accumulated on the vacuolar membrane in vps35Δ cells during log phase, and this accumulation was also observed during stationary phase (Figures 7C–7E), corroborating our mass spectrometry data (Figure 7A). We then tested whether Vps10 is required for LD internalization into the vacuole, similar to Mon1-Ccz1 and retromer. To this end, we analyzed LD localization (stained with BODIPY) relative to the vacuole in wild-type and vps10Δ cells during stationary phase. Notably, LDs partially localized within the vacuolar lumen in wild-type cells but strongly accumulated at the vacuolar periphery in vps10Δ cells (Figure 7F), suggesting that Vps10 may be involved in LD internalization into the vacuole. Next, we examined whether Vps10 targets Pdr16-LDs at the vacuolar membrane and is required for their entry into the vacuole during stationary phase, similar to retromer. To examine this, we analyzed the localization of Vps10-GFP relative to Pdr16-mKate during both log and stationary phases. Intriguingly, Vps10 significantly co-localized with Pdr16 only during stationary phase (Figures 7G and 7H). We then tested whether Vps10 is required for Pdr16-LD internalization. Analyzing the localization of Pdr16-mNeon-labeled LDs relative to the vacuole in wild-type and vps10Δ cells during stationary phase revealed that Pdr16-LDs partially localized within the vacuolar lumen in wild-type cells but strongly accumulated at the vacuolar periphery in vps10Δ cells (Figures 7I–7K), mirroring the phenotype observed in ccz1^Ile341Ala cells (Figures 3A and 3B). These findings indicate that Pdr16-LD internalization into the vacuolar lumen is impaired in vps10Δ cells. Moreover, we investigated whether Vps10 stabilizes retromer on the vacuolar membrane and, thereby, promotes LD internalization by stimulating membrane curvature. We examined the co-localization of Vps35-mNeon with Nvj1-mKate in both wild-type and vps10Δ cells. As in the wild-type, Vps35-mNeon remained strongly co-localized with Nvj1-mKate in the vps10Δ background (Figures S4F and S4G). Thus, Vps10 does not appear to be essential for retromer recruitment or stability at the vacuolar membrane.

Figure 7.

Loss of Vps10, a retromer cargo protein, leads to Pdr16-LD accumulation

(A) Design of the experimental procedure to determine vacuolar proteomics.

(B) Vacuolar proteomic analysis. Wild-type (WT) and vps35Δ cells were grown in YPD for 22 h, as described in STAR Methods. The vacuoles were isolated and analyzed by mass spectrometry. Intensities of identified proteins are plotted in normalized ratios. Selected vacuolar proteins are marked, highlighting proteins with apparent changes in abundance.

(C and D) Localization of Vps10 relative to vacuoles. WT and vps35Δ cells expressing GFP-tagged Vps10 were grown in a synthetic medium to log phase or stationary phase; vacuoles were stained with FM4-64 and analyzed using fluorescence microscopy. Scale bars, 3 μm and 0.7 μm (inset).

(E) Quantification of Vps10-GFP localization from data in (C) and (D). Cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent standard deviation (SD).

(F) Vacuole morphology and LDs were analyzed in WT and vps10Δ cells. The cells were grown in a synthetic medium to stationary phase for 48 h; vacuoles were stained with FM4-64, LDs were stained with BODIPY 493/503 (green), and they were analyzed by fluorescence microscopy. Scale bars, 4.5 μm and 1 μm (inset).

(G) Localization of Vps10 relative to Pdr16. Cells expressing GFP-tagged Vps10 and mKate-tagged Pdr16 were grown in a synthetic medium to log phase or stationary phase and analyzed by fluorescence microscopy. Scale bars, 5 μm and 0.6 μm (inset).

(H) The percentage of Vps10 puncta co-localizing with Pdr16 was determined. Pdr16 dots (n ≥ 200) and Vps10 dots (n ≥ 400) were quantified by ImageJ, as described in Figure 1D. Error bars represent SD. Bar graphs represent the averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(I) Localization of Pdr16 relative to the vacuole in WT and mutant cells. Cells expressing mNeon-tagged Pdr16 were grown in a synthetic medium to stationary phase. Vacuoles were stained with FM4-64 and analyzed by fluorescence microscopy. Scale bars, 5 μm and 1.2 μm (inset).

(J) Quantification of Pdr16-mKate localization from data in (A). The cells (n ≥ 200) were quantified by ImageJ using the tool of region of interest (ROI). Bar graphs represent averages from three independent experiments. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

(K) Pdr16 puncta count per cell. The cells (n ≥ 200) were quantified by ImageJ. Bar graphs represent averages from three independent experiments. Error bars represent the SD. ^∗∗∗, p < 0.001 (two-tailed Student’s t test).

The ESCRT (endosomal sorting complex required for transport) machinery drives selective microautophagy in yeast cells by catalyzing vacuolar membrane invagination and fission.^77,78 We, therefore, asked whether ESCRT is also required for the Ypt7-retromer-dependent vacuolar invagination observed in stationary phase. To address this, we quantified the number and spatial distribution of Pdr16-positive LDs and of total LDs stained with BODIPY in wild-type and vps4Δ cells. Surprisingly, the vps4Δ mutant showed no significant change in (1) the total number of Pdr16-LDs, (2) the total LD pool, or (3) the proportion of Pdr16-LDs that co-localize with BODIPY-labelled LDs (Figures S4H and S4I). These results suggest that ESCRT activity is not essential for the Ypt7-retromer-mediated vacuolar invagination pathway during stationary phase.

Taken together, our data support a model in which Mon1-Ccz1 activates Ypt7 on Pdr16-marked LDs next to the NVJs. This activation recruits the SNX-BAR retromer complex to the LD-vacuole interface. Once there, retromer and its bound cargo generate positive curvature of the vacuolar membrane surrounding the LDs to promote vacuolar invagination. Within these retromer-derived tubules, Vps10 may function as a cargo receptor that extracts an as-yet-unidentified vacuolar cargo from retromer-generated tubules. This cargo-sorting process could also be important for driving the late-stage vacuolar invagination.

Discussion

LDs are central to cellular energy homeostasis, and their dynamic regulation—including formation, growth, and turnover—is essential for adaptation to fluctuating environmental conditions. While LD degradation by lipolysis and autophagy (called macrolipophagy) has been well studied, the direct degradation of LDs by vacuoles through invagination, termed microlipophagy, remains unclear. Here, we identify a pathway through which LDs are targeted for potential degradation via vacuole invagination during nutrient limitation. Specifically, we found that the Rab GTPase Ypt7, activated by its GEF complex Mon1-Ccz1, localizes to a subpopulation of LDs marked by the protein Pdr16 at the vacuolar surface. Deletion of MON1, CCZ1, or YPT7 resultes in the accumulation of these Pdr16-positive LDs. Further investigation revealed that an AH in Ccz1 facilitates its recruitment to LDs in both yeast and mammalian cells. Crucially, we discovered that the SNX-BAR retromer complex functions downstream of Ypt7 at the LD-vacuole interface, potentially promoting LD invagination. Retromer and its cargo Vps10 co-localize with Pdr16 specifically under nutrient-limited conditions, and retromer co-localizes with Ypt7 at the tips of invaginated vacuoles. Loss of either retromer or Vps10 resultes in Pdr16-LD accumulation. In addition, disruption of retromer function causes massive accumulation of invaginated vesicles that remain attached to the vacuole membrane, accompanied by a significant accumulation of TAG, ergosterol esters, and DAG during stationary phase. These findings support a model in which Mon1-Ccz1-mediated activation of Ypt7 on Pdr16-LDs at the vacuolar surface recruits SNX-BAR retromer to the LD-vacuole interface. There, retromer, likely together with its cargo Vps10, facilitates LD internalization by promoting positive membrane curvature during late stage of vacuole invagination for potential subsequent LD degradation.

Our data indicate that an AH within the C-terminal domain of Ccz1 contributes to its localization to LDs in both yeast and mammalian cells. In yeast cells, a GFP-tagged Ccz1 fragment (amino acids 270–403) containing the AH strongly co-localizes with LDs during both log and stationary phases, whereas full-length Ccz1 targets LDs only during stationary phase. This suggests a regulatory mechanism by which Ccz1 is excluded from LDs during log phase. The specificity of this AH for LD targeting is striking as the same AH is required for Mon1-Ccz1 targeting to autophagosomes,^27,47 yet a subtle reduction in AH hydrophobicity specifically abolishes LD targeting without affecting autophagic function. This differential targeting likely reflects differences in the lipid composition and membrane packing of LDs and autophagosomes. Alternatively, Mon1-Ccz1 may interact with a specific partner on LDs, a possibility we have not yet excluded.

Notably, the ccz1^Ile341Ala mutation causes the accumulation of Pdr16-positive LDs in yeast cells, and the corresponding human CCZ1^Leu287Ala mutation similarly accumulates LDs in COS7 cells. However, it remains unknown whether these accumulated LDs in COS7 cells are equivalent to yeast Pdr16-LDs.

Furthermore, we identify the SNX-BAR retromer complex as a crucial effector at the LD-vacuole interface. Retromer localizes to Pdr16-LDs at the periphery of NVJ and co-localizes with LDO during stationary phase. Consistent with a potential role in LD degradation, loss of retromer results in a significant increase in TAG, ergosterol esters, and DAG during stationary phase. It is important to note that the increase in TAG levels during the stationary phase occurs even in wild-type cells, indicating a simultaneous increase in lipid synthesis. Thus, steady-state TAG levels alone are not a straightforward indicator of degradation activity. The more telling comparison is between wild-type and retromer-mutant cells; the latter show a significant accumulation of TAG and a unique build-up of DAG and ergosterol esters during the stationary phase. This suggests that these neutral lipids, which might be degraded via lipophagy in wild-type cells, are not effectively processed in the mutant. The accumulation of ergosterol esters, in particular, underscores a disruption in normal lipid degradation pathways, as these esters are well-documented substrates of lipophagy. Their stable levels in wild-type cells, contrasted with their accumulation in mutant cells, indicate a blockade in LD degradation in the absence of retromer. While steady-state lipidomics provides valuable insights, it does not differentiate between lipid synthesis and degradation rates. Future studies using pulse-chase techniques could offer detailed insights into these dynamics. Additionally, the retromer-mutant phenotype is marked by an accumulation of LDs and invaginated vesicles connected to the vacuolar membrane. This observation is corroborated by our cryo-EM study in the ccz1^Ile341Ala mutant, where LDs accumulated and showed partial engulfment by the vacuolar membrane, implying an obstruction in the invagination phase. Taken together, these findings emphasize that the retromer complex operates at a late stage of vacuolar invagination to facilitate subsequent LD degradation. This underscores its broader role in maintaining lipid homeostasis. Interestingly, we observed that retromer co-localizes with Ypt7 at the tips of invaginated vacuoles, suggesting a specific role at these membrane regions. Retromer is well known for mediating cargo recycling and for generating membrane curvature through its SNX-BAR subunits Vps5 and Vps17.^36,72 We, therefore, propose that, together with its cargo, retromer creates positive curvature at the tips of the invaginated vacuole “hugging” LDs. This curvature drives the formation of tubular extensions that facilitate the final step of vacuolar invagination and enable LD internalization. Supporting this model, we detected the canonical retromer cargo Vps10 at the LD-vacuole interface. Loss of Vps10 caused accumulation of Pdr16-positive LDs adjacent to the vacuole. Vps10 is a sorting receptor that shuttles cargo between the Golgi and endosomes, and its retrieval from the vacuole depends on both retromer and Ypt7.⁷⁹ We, therefore, suggest that Vps10, together with retromer-generated SNX-BAR tubules, removes an as-yet-unknown vacuolar cargo and provides the membrane curvature necessary for the late stage of invagination, thereby permitting degradation of Pdr16-associated LDs. We note, however, that the buildup of Pdr16-LDs in vps10Δ cells might reflect secondary consequences of other Vps10 functions; so, indirect effects cannot be ruled out at this stage.

Our results extend recent reports that vCLIPs, composed of LDO and Vac8, are essential for microlipophagy.^17,18 Deletion of LDO or VAC8 causes mis-localization of Ccz1, Ypt7, and the retromer complex to Pdr16-positive LDs during stationary phase, whereas the ccz1^Ile341Ala mutant and loss of retromer do not disrupt vCLIP positioning. These observations suggest that the Ypt7-retromer complex may function downstream of vCLIP in the LD-vacuole degradation pathway. Although the precise interplay between vCLIPs and the Ypt7-retromer axis remains to be defined, it is likely that the two systems cooperate to coordinate LD targeting and degradation. In this model, the Ypt7-retromer complex could act at vCLIPs to sharpen LD positioning and promote membrane invagination required for LD clearance.

Future studies will focus on elucidating the precise molecular mechanisms by which Ypt7 and retromer promote LD invagination into the vacuole, identifying the specific retromer and Vps10 interaction partners involved, and determining how these proteins are regulated at membrane contact sites. Understanding these processes will provide valuable insights into the complex interplay between LD metabolism, organelle communication, and cellular adaptation to stress.

Limitations of the study

While this study identifies a pathway for Ypt7-retromer-mediated LD internalization, several limitations should be considered. Although our genetic, cell biological, and localization analyses support a role for retromer in LD internalization, the molecular mechanism by which retromer promotes LD engulfment remains incompletely defined. In particular, we lack direct biochemical evidence for the proposed interactions among retromer and downstream effectors, which limits our ability to distinguish between direct and indirect roles of retromer in this process. Additionally, our analysis relies predominantly on endpoint measurements, which preclude a detailed understanding of the temporal sequence of LD internalization and degradation. As a result, the order of these events and the kinetics of the process remain inferential rather than established.

Collectively, these caveats do not undermine our main conclusion that retromer mediates selective LD internalization, but they should be taken into account when interpreting the mechanistic details of this pathway.

Resource availability

Lead contact

Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Jieqiong Gao (jqgao@ust.hk).

Materials availability

Plasmids and strains generated in this study may be requested from the lead contact.

Data and code availability

Data supporting the findings of this study will be made available by the lead contact upon reasonable request. No original code was generated during this research. Additional information required for data re-analysis will also be provided by the lead contact upon request. The mass spectrometry proteomics data have been deposited in the Japan ProteOme STandard Repository (jPOSTrepo)⁸⁰ under the jPOST ID: JPST004765 and PXID: PXD080968. The mass spectrometry lipidomics data have been deposited to the MetaboLights,⁸¹ with identifier MTBLS15038. In addition, all quantitative results, mass spectrometry proteomics data, and lipidomics data are publicly available in the Mendeley Data: https://doi.org/10.17632/rj52r9kvhw.2.

Acknowledgments

We thank members of the Gao laboratory and Maria Bohnert (University of Münster, Germany) for insightful discussions and critical review of the manuscript. We also thank Yusong Guo (HKUST) for providing the COS7 cell line and assisting with the establishment of the mammalian system. We thank the Fröhlich lab (University of Osnabrück, Germany) for developing the vacuolar-proteomics workflow that enabled the identification of retromer cargo. We thank the Biosciences Central Research Facility at HKUST for providing access to mass spectrometry instruments and services. The cryo-EM dataset was collected at the Cryo-EM Center, made possible by the generous donation of the Lo Kwee Seong Foundation to HKUST. The graphical abstract was created in BioRender (https://BioRender.com/vhbp9lf). This work was supported by the grants of the Deutsche Forschungsgemeinschaft (DFG) to C.U. (UN111/13-2, grant number 448135520; SFB 1557, Project P14, grant number 516911046) and L.L. (grant number 534725166). R.W.K. acknowledges funding from the Wellcome Trust 311314/Z/24/Z and Diabetes UK grant 22/0006453. J.G. was funded by start-up funding (R9956) from HKUST and Hong Kong Research Grants Council Early Career Scheme (ECS) (26102725). Y.G. was funded by National Key R&D Program of China (2025YFA0923000) and Hong Kong Research Grants Council Grants (16104423 and 16103622).

Author contributions

Conceptualization, J.G.; investigation and formal analysis, Z.X., Y. Zhi, G.Q., Y.L., Y. Zhang, H.T.N., A.J.S.-P., Y.G., C.U., R.W.K., L.L., and J.G.; visualization, Z.X., Y. Zhi, G.Q., Y.L., Y. Zhang, H.T.N., A.J.S.-P., Y.G., C.U., R.W.K., L.L., and J.G.; writing – original draft, J.G.; writing – review & editing, Z.X., Y. Zhi, G.Q., Y.L., Y. Zhang, H.T.N., A.J.S.-P., Y.G., C.U., R.W.K., L.L., and J.G.; supervision and funding acquisition, J.G., L.L., R.W.K., C.U., and Y.G.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Myc-Tag (9B11) Mouse mAb	Cell Signaling Technology	Cat#2276T; RRID:AB_331783
Donkey anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488	Thermo Fisher Scientific	Cat# A21202;RRID:AB_141607
Rabbit polyclonal anti PLIN3-c-terminus	EMD Milipore	Cat#: ABS482
Mouse monoclonal anti TOM20	Abcam	Cat#ab56783;RRID:AB_945896
Mouse monoclonal anti C-Myc 9E10	Santa Cruz Biotech	Cat#: sc-40; RRID: AB_627268
Chemicals, peptides, and recombinant proteins
BODIPY 493/503	Thermo Fisher Scientific	Cat#D3922
LipidTOX DeepRed	Thermo Fisher Scientific	Cat#H34477
FM™ 4-64 Dye (N-(3-Triethylammoniumpropyl)-4-(6-(4-(Diethylamino) Phenyl) Hexatrienyl) Pyridinium Dibromide)	Thermo Fisher Scientific	Cat#T13320
Sodium hydroxide	Sigma-Aldrich	Cat#367176;CAS: 1310-73-2
Paraformaldehyde	Sigma-Aldrich	Cat#P6148;CAS:30525-89-4
TRIzol™	Thermo Fisher Scientific	Cat#15596018CN
Lipofectamine 2000	Thermo Fisher Scientific	Cat#11668019
ProLong Gold Antifade Mountant with DAPI	Thermo Fisher Scientific	Cat#P36935
Chloroform	Sinopharm Chemical Reagent Co., Ltd.	Cat#151902104; CAS: 67-66-3
Methanol (LC-MS grade)	Merck	Cat#1.06035.2500; CAS: 67-56-1
Citric acid	Sangon Biotech	Cat#A610055-0500; CAS: 77-92-9
FAS (AIEgen lipid droplet probe)	Tang Lab/Wang et al., 2016	N/A
DMEM	Gibco™	Cat#12100061
Fetal bovine serum	Thermo Fisher Scientific	Cat#A5256701
Optimized Minimum Essential Medium (Opti-MEM)	Gibco™	Cat#31985070
Pen Strep (Penicillin Streptomycin)	Gibco™	Cat# 15140122
0.25% Trypsin-EDTA (1×)	Gibco™	Cat# 25200072
Oleic acid	Sigma-Aldrich	Cat#O1383-5G; CAS: 112-80-1
Triton X-100	Sigma-Aldrich	Cat#:×100
Glycine	Thermo Fisher Scientific	Cat#15527013
Yeast extract	Gibco™	Cat#212730
Peptone	Gibco™	Cat #211830
D-(+)-Glucose	Sigma-Aldrich	Cat#G7021; CAS: 50-99-7
Trizma@basa	Sigma-Aldrich	Cat#T1503; CAS: 77-86-1
DTT	Thermo Fisher Scientific	Cat#R0862
D-Sorbitol	Sigma-Aldrich	Cat#S1876;CAS:50-70-4
Potassium hydroxide	Scharlab	Cat# PO02750500;CAS: 1310-58-3
PIPES	Sigma-Aldrich	Cat#P1851; CAS: 5625-37-6
Potassium phosphate dibasic	Sigma-Aldrich	Cat#P3786;CAS:7758-11-4
Ficoll® PM 400	Sigma-Aldrich	Cat#F4375;CAS:26873-85-8
DEAE Dextran Hydrochlorid	Sigma-Aldrich	Cat#D9885; CAS: 9064-91-9
Acetonitrile (LC-MS grade)	Merck	Cat#159014.2500; CAS: 75-05-8
2-Propanol (LC-MS grade)	Merck	Cat#1.02781.2500; CAS: 67-63-0
Water (LC-MS grade)	Merck	Cat#1.15333.2500; CAS: 7732-18-5
Ammonium acetate (LC-MS grade)	Sigma-Aldrich	Cat#73594-100G-F; CAS: 631-61-8
Dichloromethane (LC-MS grade)	Merck	Cat#1.06044.1000; CAS: 75-09-2
LIPIDOMIX Standards	Avanti	Cat#330707-1EA
Potassium dihydrogen phosphate	Scharlab	Cat#PO02600500;CAS: 7778-77-0
Critical commercial assays
PrimeScript RT reagent Kit with gDNA Eraser	Takara Bio Inc	Cat#RR047A
TB Green Premix Ex Taq II	TaKaRa Bio Inc	Cat#RR820A
Preomics iST kit (96×)	Thermo Fisher Scientific	Cat#NC1699272
Deposited data
Saccharomyces cerevisiae UniProt database	UniProt	https://www.uniprot.org/proteomes/UP000002311
Proteomic mass spectrometry data	This study	jPOST ID: JPST004765 and PXID: PXD080968.
Lipidomic mass spectrometry data	This study	MetaboLights with identifier MTBLS15038
Experimental models: Cell lines
Cell line: COS7	Gift from Yusong Guo Lab;originally from ATCC	RRID:CVCL_0224
Experimental models: Organisms/strains
Saccharomyces cerevisiae: BY4741: MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0	Gift of Christian Ungermann	N/A
Saccharomyces cerevisiae: SEY6210: MATalpha leu2-3 leu2-112 ura3-52 his3-Δ200 trp1-Δ101 lys2-801 suc2-Δ9 GAL	Gift of Christian Ungermann	N/A
Saccharomyces cerevisiae:JGY121:BY4741; vps10Δ::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY125:BY4741; vps35Δ::kanMX4	Gift of Christian Ungermann	N/A
Saccharomyces cerevisiae:JGY172:BY4741; VPS10-GFP::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY238:BY4741; GFP-CCZ1::URA3-PHO5pr-myc mCherry-ATG8::natNT2	Gift of Christian Ungermann (Gao et al., eLife, 2018)	N/A
Saccharomyces cerevisiae:JGY336:BY4741; PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY338:BY4741; mon1Δ::kanMX4 PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY340:BY4741; ccz1Δ::kanMX4 PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY342:BY4741; PDR16-mKate::hphNT1 NVJ1-mGFP::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY344:BY4741; GFP-YPT7::HIS5-PHO5pr-Myc PDR16-mKate:natNT2	This study	N/A
Saccharomyces cerevisiae:JGY356:BY4741; ypt7Δ::kanMX PDR16-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY386:SEY6210; ERG6-mKate::natNT2 ccz1Δ::hphNT1 pRS406-TEFpr-GFP-CCZ1::URA	This study	N/A
Saccharomyces cerevisiae:JGY387:SEY6210; ERG6-mKate::natNT2 ccz1Δ::hphNT1 pRS406-TEFpr-GFP-CCZ1(270aa-403aa)::URA	This study	N/A
Saccharomyces cerevisiae:JGY397:BY4741; ERG6-3xmCherry GFP-YPT7::URA	This study	N/A
Saccharomyces cerevisiae:JGY430:BY4741; GFP-CCZ1::URA3-PHO5pr-myc ERG6-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY431:BY4741; GFP-CCZ1::URA3-PHO5pr-myc mCherry-ATG8::natNT2 ccz1(I341A)::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY435:BY4741; GFP-CCZ1::URA3-PHO5pr-myc PDR16-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY437:BY4741; GFP-CCZ1::URA3-PHO5pr -myc ERG6-mKate::natNT2 ccz1(I341A)::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY438:BY4741; GFP-CCZ1::URA3-PHO5pr -myc PDR16-mKate::natNT2 ccz1(I341A)::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY452:BY4741; ERG6-mKate::natNT2 VPS41-mNeon::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY464:BY4741; ERG6-mKate::natNT2 VPS35-mNeon::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY475:BY4741; NVJ1-mScarlet::kanMX VPS29-mNeon::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY476:BY4741; NVJ1-mScarlet::kanMX VPS35-mNeon::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY478:BY4741; YPT7::HIS5-PHO5pr-Myc-GFP VPS35-mKate::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY596:BY4741; GFP-CCZ1::URA3-PHO5pr -myc PDR16-mKate::natNT2 vac8Δ::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY598:BY4741; PDR16-mKate::hphNT1 NVJ1-mKate::natNT2 vps35Δ::HIS	This study	N/A
Saccharomyces cerevisiae:JGY602:BY4741; VPS35-mNeon::kanMX PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY605:BY4741; VPS35-mNeon::kanMX PDR16-mKate::hphNT1 vac8Δ ::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY607:BY4741; GFP-CCZ1::URA3-PHO5pr-myc mon1Δ::kanMX ERG6-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY608:BY4741; GFP-CCZ1::URA3-PHO5pr-myc atg1Δ::hphNT1 ERG6-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY609:BY4741; GFP-CCZ1::URA3-PHO5pr-myc atg8Δ::hphNT1 ERG6-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY610:BY4741; GFP-CCZ1::URA3-PHO5pr-myc ERG6-mKate::natNT2 sei1Δ::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY611:BY4741; GFP-CCZ1::URA3-PHO5pr-myc ERG6-mKate::natNT2 vps8Δ::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY612:BY4741; GFP-CCZ1::URA3-PHO5pr-myc ERG6-mKate::natNT2 vps21Δ:: hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY613:BY4741; ccz1(I341A)::kanMX	This study	N/A
Saccharomyces cerevisiae:JGY614:BY4741; ccz1(I341A)::kanMX PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY615:BY4741; vps35Δ::kanMX PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY620:BY4741; vps5Δ::kanMX PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY621:BY4741; atg18Δ::kanMX PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY634:BY4741; PDR16-mNeon::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY638:BY4741; ccz1(I341A)::kanMX PDR16-mNeon::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY666:BY4741; PDR16-mNeon::natNT2 LDO16-45-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY667:BY4741; CCZ1(I341A)::kanMX PDR16-mNeon::natNT2 LDO16-45-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY669:BY4741; VPS35::mNeon-kanMX LDO16-45-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY670:BY4741; PDR16-mNeon::natNT2 vps35Δ::KanMX LDO16-45-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY671:BY4741; vps10Δ::kanMX PDR16-mNeon::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY676:BY4741; VPS10-GFP::kanMX PDR16-mKate::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY678:BY4741; NVJ1-mScarlet::kanMX VPS5-mNeon::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY679:BY4741; NVJ1-mScarlet::kanMX ATG18-mNeon::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY681:BY4741; YPT7::HIS5-PHO5pr-Myc-GFP LDO16-45-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY699:BY4741; VPS10-GFP::KanMX Vps35Δ::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY737:BY4741; vps4Δ::kanMX4 PDR16-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY770:BY4741; CCZ1::URA3-PHO5pr-GFP-myc LDO16-45-mKate::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY780:BY4741; NVJ1-mScarlet::kanMX VPS35-mNeon::hphNT1 vps10Δ::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY802:BY4741; YPT7::HIS5-PHO5pr-Myc-GFP Pdr16-mGFP::natNT2 LDOΔ::hphNT1	This study	N/A
Saccharomyces cerevisiae:JGY805:BY4741; VPS35::mNeon::kanMX PDR16-mKate::hphNT1 ldoΔ::natNT2	This study	N/A
Saccharomyces cerevisiae:JGY814:SEY6210; ERG6-mKate::natNT2 ccz1Δ::hphNT1 pRS406-TEF-GFP-CCZ1_AH_invert::URA	This study	N/A
Saccharomyces cerevisiae:JGY815:SEY6210; ERG6-mKate::natNT2 ccz1Δ::hphNT1 pRS406-TEF-GFP-CCZ1(270aa-403aa_AH_invert)::URA	This study	N/A
Oligonucleotides
siRNA-Ccz1:GGATGTCCGGGTCTGAGAA	This study	N/A
q-GADPH-1-F:ACAACTTTGGTATCGTGGAAGG	This study	N/A
q-GADPH-1-R:GCCATCACGCCACAGTTTC	This study	N/A
q-Ccz1-2-F:CCTGATTGAAGTAAACGAAGAGGT	This study	N/A
q-Ccz1-2-R:ATGACCAGTAACTTGCATATGACC	This study	N/A
Recombinant DNA
pRS406-TEFpr-GFP-CCZ1	This study	N/A
pRS406-TEFpr-GFP-CCZ1 (aa270-403)	This study	N/A
pcDNA5/FRT/TO/Myc CCZ1	Gift from Francis Barr	N/A
pcDNA/FRP/TO/GFP-MON1a	Gift from Francis Barr	N/A
pcDNA5/FRT/TO/Myc CCZ1 (Leu287Ala)	This study	N/A
pRS406-TEFpr-GFP-CCZ1 (aa270-403)_AH_invert	This study	N/A
pRS406-TEFpr-GFP-Ccz1_AH_invert	This study	N/A
Software and algorithms
PEAKS Studio X (v10.6)	Bioinformatics Solutions Inc.	https://www.bioinfor.com/peaks-studio/
Proteome Discoverer software (v2.4.1.15)	Thermo Fisher Scientific	https://www.thermofisher.com/
ImageJ	ImageJ	https://imagej.net/software/imagej/
Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
Rstudio v3.3.2	Rstudio	https://rstudio.com/
BioRender	BioRender	https://BioRender.com/vhbp9lf

Experimental model and study participant details

Yeast strains and molecular biology

Yeast strains and plasmids used in this study are listed in STAR Methods. Gene deletions and tagging were performed using PCR-based homologous recombination with appropriate primers and templates.^82,83 The basic yeast strains BY4741 and SEY6210, as well as a strain expressing both GFP-tagged Ccz1 and mCherry-tagged Atg8, were kindly provided by Christian Ungermann (University of Osnabrück, Germany). Plasmid encoding MON1 or CCZ1 was kindly provided by Francis Barr (University of Oxford, UK).

Cell line

COS7 cells (ATCC CRL-1651; RRID: CVCL_0224), a male African green monkey kidney fibroblast cell line, were kindly provided by Yusong Guo (Hong Kong University of Science and Technology, HK). All cells were routinely tested and confirmed to be free of Mycoplasma contamination. Cells were cultured at 37°C in DMEM (Thermo-Fisher) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich) and maintained in 5% CO2.

Method details

Plasmid transfection of COS7

For transfection, COS7 cells were seeded onto glass coverslips in 12-well plates and cultured in DMEM supplemented with 10% FBS until reaching 90% confluence. Cells were transfected using Lipofectamine 2000 (Thermo-Fisher) according to the manufacturer’s instructions, with modifications: 0.3 μg DNA and 1 μL transfection reagent were used per construct and well. Both DNA and Lipofectamine 2000 were prediluted in Optimem medium (Thermo-Fisher). The DNA-transfection reagent complex was incubated for 20 min at room temperature before being added dropwise to the culture medium. Cells were then incubated overnight with the transfection mix. The following day, cells were treated with sodium oleate (OA, see below) for 6 h and then processed for indirect immunofluorescence.

Sodium oleate preparation

Sodium oleate was prepared by dissolving 400 μL of 5 M NaOH (Sigma-Aldrich) and 400 μL of oleic acid (Sigma-Aldrich) in 25 mL of warmed, doubly deionized water (ddH₂O). The solution was vigorously stirred using a magnetic stirrer until a clear, micelle-free solution was obtained. If necessary, additional NaOH was added dropwise until complete transparency was achieved. The final volume was adjusted to 30 mL with ddH₂O.

CCZ1 knockdown

siRNA against CCZ1 was purchased from Ruibo Bio. The target sequence against CCZ1 is GGATGTCCGGGTCTGAGAA. To knock down the expression of Ccz1, COS7 cells were transfected with siRNA using Lipofectamine 2000 (Thermo-Fisher) according to the manufacturer’s instructions and cells were analyzed 48 h after siRNA transfection. Real-Time Quantitative RT-PCR was used to confirm mRNA level after knockdown.

Real-Time Quantitative RT-PCR

Total RNA was extracted from the cultured cells using TRIzol reagent (Thermo-Fisher), according to the manufacturer’s instructions. Total RNA (1 μg) was reverse transcribed into cDNA using a PrimeScript RT reagent Kit with gDNA Eraser (Takara Bio Inc, #RR047A). The cDNA determination was performed using TB Green Premix Ex Taq II (TaKaRa Bio Inc, #RR820A) on the Roche Lightcycler 480 Real-Time PCR system (Roche Diagnostics, Basel, Switzerland). After the reactions were completed, the mean CT was determined from triplicate PCRs. GAPDH were used as normalization. Quantification was performed using the ΔΔCT method.

Indirect immunofluorescence with COS7 cells

Cells were fixed with 4% paraformaldehyde in PBS for 20 min. Following fixation, cells were blocked with a buffer containing 2.5% fetal bovine serum, 0.1% Triton X-100, and 0.2M glycine in PBS for 30 min at room temperature. Cells were then incubated with the indicated primary antibody in blocking buffer at room temperature for 1 h, followed by three PBS washes. Secondary antibodies, fluorescently labeled, were then applied in blocking buffer for 1 h in the dark at room temperature, and cells were again washed three times with PBS. Cells were subsequently incubated with LipidTOX DeepRed (Invitrogen, 1:200 in PBS) for 30 min at room temperature, washed three times with PBS, and mounted onto slides using ProLong Gold Antifade Mountant with 4′,6-diamidino-2-phenylindole for microscopic analysis.

Fluorescence microscopy

Yeast cells were grown to log phase in synthetic complete medium (synthetic defined medium lacking amino acids with ammonium sulfate and 2% glucose; SDC+all) or to stationary phase in SDC+all for 20–24 h at 30°C. For visualization, cells were imaged at room temperature using two distinct microscopy systems. Firstly, images were acquired using a DeltaVision Elite imaging system based on an Olympus IX-71 inverted microscope. This system was equipped with 100× NA 1.49 objectives, a sCMOS camera (PCO, Kelheim, Germany), and an Insight SSI (TM) illumination system. Images were collected using 4′,6-diamidino-2-phenylindole (DAPI), GFP, mCherry, and Cy5 filters. Z-stacks consisting of 6–10 images were taken at intervals of 0.2–0.35 μm, and subsequent deconvolution was performed using SoftWoRx software (Applied Precision, Issaquah, WA). Secondly, cells were also imaged at room temperature on a Zeiss LSM 980 confocal microscope equipped with AiryScan 2 super-resolution technology (Zeiss, Jena, Germany). This system was equipped with a Plan-Apochromat 63× objective lens (oil) that has a 1.40 numerical aperture, and a 32 channel GaAsP-PMT spectral detector. To take images, 405nm/488 nm/568 nm/647nm lasers were used, respectively. The Zeiss ZEN blue software was used for image acquisition. Fiji software was used for image analysis.

Cryo-fixation of yeast cells

The wild-type and ccz1^I341A mutant yeast cells, with an estimated concentration about 108 cells/mL (∼3.5 μL), was applied onto a glow-discharged (PELCO easiGlowTM) R2/2 200 mesh holey carbon copper grid (Quantifoil Micro Tools GmbH). A plunge-vitrification device (Vitrobot VI, Thermo Fisher Scientific) operating at 22°C and 100% humidity was applied to plunge-freeze the grid using a blotting force of 25, blotting time of 15 s into liquid ethane. The cryo-grids after frozen were clipped into CryoFIB AutoGrids (Thermo Fisher Scientific), then stored for Cryo lamella preparation.

Cryo lamella preparation and Cryo-EM imaging

The CryoFIB AutoGrids were mounted onto a 35-degree Autogrid shuttle and transferred into a Cryo Focused ion beam scanning electron microscope (Cryo-FIB SEM) dual beam system (Aquilos 2, Thermo Fisher Scientific) operated at 2 keV (Electron beam) and 30 keV (Ion beam) with an Everhart-Thornley detector (ETD) for imaging at liquid nitrogen temperature. After coating, a map of each AutoGrid was collected with Maps 3.27 (Thermo Fisher Scientific) and several lamella sites were added and further milled using AutoTEM Cryo 2.4 (Thermo Fisher Scientific) with a milling degree around 12, milling current from 0.5 nA to 30 pA. The prepared cryo lamellas have their size about 10–14 μm in length, 15-18 μm in depth and 180-240 nm in thickness.

The Cryo-FIB lamellas after preparation were restored and transferred into a Titan Krios G3i microscope (Thermo Fisher Scientific Inc.) operated at 300 keV. The lamellas were screened and imaged with a Gatan K3 direct electron detector positioned at the end of an energy filter with a slit width of 20 eV. The images were taken with a total dose about 30 eV/Å2 at a couple of magnifications with calibrated pixel size of 1.37 nm/pixel (6500×), 3.4Å/pixel (26000×) and 2.7Å/pixel (33000×) without dosefractionation.

Vacuole isolation and proteomics

Wild-type and mutant cells were grown in YPD (1% yeast extract, 1% peptone, and 2% glucose) at 30°C for 22 h. The vacuole isolation assay was performed as described before (Gao et al., 2018b). These isolated vacuoles were digested and purified using the Preomics iST kit (96×) with a PreON automated system. Purified peptides were dried via SpeedVac vacuum centrifugal concentrator at 37°C and resuspended to 200 ng/μL in 0.1% formic acid (FA) in water.

Bruker timsTOF pro data acquisition and processing

Samples (1 μL injection volume) were subjected to nanoflow liquid chromatography on a Bruker nanoElute system using an IonOpticks Aurora Ultimate column (25 cm × 75 μm ID, 1.6 μm C18) at 0.3 μL/min. For liquid chromatography, solvent A was 98% water and 2% acetonitrile with 0.1% formic acid; solvent B was 100% acetonitrile with 0.1% formic acid. The gradient transitioned from 2% to 5% solvent B within 0.5 min, from 5% to 40% solvent B over 26.5 min, and finally from 40% to 95% solvent B within 0.5 min. Following a 0.5 min retention at 95% solvent B, the mobile phase was reduced to 2% solvent B in 0.1 min. A 2% solvent B equilibration phase occurred for 2 min before the next injection. The eluted peptide was detected by Bruker timsTOF Pro mass spectrometer operating in data-dependent acquisition with parallel accumulation-serial fragmentation (DDA-PASEF) mode. Instrument control was managed by timsControl 3.0 with a capillary voltage of 1400 V and a dry gas flow of 3.0 L/min at 180°C using Captive Spray (Bruker Daltonics).

For protein identification, the acquired spectra were analyzed with PEAKS Studio X (v10.6) by searching against a Saccharomyces cerevisiae UniProt database. Search parameters included carbamidomethylation as a fixed modification, oxidation and N-terminal acetylation as variable modifications, with maximum of 3 variable PTM allowed per peptide, and mass tolerances of 15 ppm (precursor) and 0.06 Da (fragment). Estimate FDR with decoy-fusion is activated. All identifications were filtered to achieve a 1% FDR at the peptide and protein levels.

Thermo eclipse data acquisition and processing

Chromatographic separation was performed using an Easy-nLC 1200 system coupled to an Orbitrap Eclipse™ Tribrid™ mass spectrometer (Thermo Scientific) equipped with an EasySpray ion source and a FAIMS Pro interface. Peptides were separated on an IonOpticks Aurora Ultimate series column (25 cm × 75 μm ID, 1.7 μm C18) maintained at 50°C with a flow rate of 300 nL/min. Solvent A is 0.1% FA in water and solvent B is 80% ACN in water with 0.1% FA. The gradient was ramped from 1% B to 4% B in 1 min, from 4% B to 23% B in 51 min, from 23% B to 35% B in 7 min, from 35% B to 100% B in 1 min and retain at 100% B for 5 min. Mass spectrometry analysis was conducted in a data-dependent acquisition mode. Data acquisition used a 1-s cycle time with two experiments performed at distinct FAIMS CVs: −45V and −65 V.

Data were processed using Proteome Discoverer software (version 2.4.1.15, Thermo Scientific). Spectra were searched against an UniProt Saccharomyces cerevisiae database. Search parameters included a precursor mass tolerance of 10ppm and a fragment mass tolerance of 0.02 Da. A maximum of two missed cleavages were set. Carbamidomethylation of cysteine was set as a fixed modification, while oxidation of methionine and protein N-terminal acetylation were set as variable modifications. Peptide identifications were filtered to a 1% FDR.

Final data analysis and plotting were performed using the R software package.

Lipid extraction and quantitative LC-MS/MS analysis

Yeast lipids were extracted with chloroform:methanol as described previously (2:1, v/v).⁸⁴ Briefly, two OD600 units of yeast cells were harvested and resuspended in 0.5 mL MS-grade methanol. Cell lysis was achieved via bead-beating, and the resulting lysates were transferred to glass tubes. Chloroform and citric acid were used to achieve phase separation. The bottom organic phase containing lipids was collected and dried using a vacuum concentrator system (Beijing JM Tech). The dried lipid extracts were reconstituted in a 1:1 (v/v) mixture of methanol and dichloromethane supplemented with 5 mM ammonium acetate. Quantitative analysis was performed using Lipidomix standards (Avanti Polar Lipids) as internal spike-in controls. Chromatographic separation was carried out on an ACQUITY UPLC BEH C18 column (1.7 μm, 130 Å, 2.1 mm × 50 mm; Waters) coupled to a QTRAP 6500+ triple quadrupole mass spectrometer (AB SCIEX). The mobile phases consisted of Buffer A (methanol/acetonitrile/water, 1:1:1, v/v/v, with 5 mM ammonium acetate) and Buffer B (100% isopropanol with 5 mM ammonium acetate). The flow rate was set at 0.15 mL/min with the following gradient program: 0–1 min, 20% B; 1–3 min, 20%–60% B; 3–13 min, 60%–98% B; 13.0–13.1 min, 98%–20% B; and 13.1–16 min, equilibration at 20% B. Data acquisition was conducted in positive ionization mode using multiple reaction monitoring (MRM) transitions specific for diacylglycerols (DAG), triacylglycerols (TAG), and ergosterol esters. Lipid species were identified by comparing retention times with authentic standards, and quantification was performed based on peak areas using the software OS v3.4 (AB SCIEX).

Quantification and statistical analysis

Fluorescence microscopy images were processed and quantified using ImageJ. All quantifications involved at least three independent experiments. Details of the quantification methods, along with the number of cells and protein puncta analyzed, are provided in each figure legend. In the bar graphs, data are presented as the mean ± standard deviation (SD). Statistical analyses of all fluorescent microscopy images were conducted using Microsoft Excel. We assumed a normal distribution of the data, although this assumption was not formally tested. To compare two groups, we utilized a two-sided Student’s t test. For comparisons involving multiple groups, a one-way ANOVA was performed, followed by Tukey’s post-hoc test. Statistical significance is indicated as follows: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001 and ^∗∗∗∗p < 0.0001. All statistical analyses and corresponding p values are detailed in the figure legends.

Published: August 12, 2026