Coupled sterol synthesis and transport machineries at ER–endocytic contact sites

A pronounced heterogeneity in the sterol content of cellular membranes is maintained (Menon, 2018; Mesmin et al., 2013). How such heterogeneity is generated, and how it contributes to particular functions, is fairly unknown. Despite sterols are synthesized at the ER, their level in this compartment is kept low (Mesmin et al., 2013). This implies that sterol synthesis and extraction must be coupled. Nonvesicular sterol transport is thought to convey ER sterols to the plasma membrane (PM) in yeast (Baumann et al., 2005), but detailed information on the sterol routes is missing. Proteins with capacity to transfer sterols include OSBP (oxysterol-binding protein)–related proteins (ORPs; Wong et al., 2019). In yeast, OSH1 to OSH7 encode ORP-like proteins. Osh1, Osh2, Osh4, and Osh5 extract sterols in vitro (Raychaudhuri et al., 2006; Schulz et al., 2009), whereas Osh6 and Osh7 transport phosphatidylserine (Maeda et al., 2013). Oshes are thought to counter-transport phosphatidylinositol-4 phosphate (PI4P) to fuel lipid transport against gradient (de Saint-Jean et al., 2011; Moser von Filseck et al., 2015b), but direct in vivo evidence for the hypothesis is missing for many Oshes. Oshes localize at specialized membrane contact sites (MCSs; Prinz et al., 2020; Wong et al., 2019), where they sustain different processes (Beh et al., 2001). In this context, we found that Osh2 and Osh3 are components of ER–endocytic MCSs (Encinar del Dedo et al., 2017). Contact of the ER with endocytic sites promotes actin polymerization and membrane invagination (Encinar del Dedo et al., 2017). Endocytosis in yeast requires sterols, and the endocytic function of Osh2 depends on its sterol transfer activity (Encinar del Dedo et al., 2017; Maeda et al., 2013; Moser von Filseck et al., 2015a). Thus, Osh2 could locally transfer sterols from the ER to endocytic sites, but the functional need for this transport is difficult to reconcile with the high levels of sterols at the PM (Menon, 2018; Schneiter et al., 1999). Nonetheless, one must take into account that the pool of active cytosolic-accessible sterol is tuned not only by its concentration but also by the membrane composition and curvature (Menon, 2018; Mesmin et al., 2013; Moser von Filseck et al., 2015b).

Here, we followed the sterol dynamics in yeast with the D4H domain of perfringolysin (PFO), which only binds sterols when its C3 hydroxyl group is exposed (Maekawa, 2017; Savinov and Heuck, 2017). We found that accessible sterols were highly polarized, accumulating at the PM of daughter cells. In mothers, the probe was instead recruited to ER subdomains labeled with a subset of sterol biosynthetic enzymes (Ergs), which also interacted with Myo5 and Osh2. Further, we showed that Osh2 extracts sterols from these subdomains, which we name ERSESs (ER sterol exit sites). Finally, we found that although sterols are critical for endocytosis, on-site coupling of the sterol synthesis and transport machineries was needed in mother cells, but not in daughters, where endocytosis and PM loading with sterols were linked to secretion.

To investigate the dynamics of cytosolic-accessible sterols in yeast, we expressed GFP-D4H (Johnson et al., 2012; Koselny et al., 2018; Maekawa, 2017; Marek et al., 2020; Savinov and Heuck, 2017; Fig. S1 A) at levels that did not alter the cortical dynamics of the endocytic coat component Sla1-GFP (Fig. S1 B). Inspection of yeast expressing GFP-D4H unveiled a strong PM staining in small to medium-sized daughter cells (Fig. 1 A; Video 1 and Video 2). The strong staining persisted from bud emergence to initial phases of cytokinesis, where labeling accumulated at the bud neck (Video 1). At bud emergence, small vesicles moved toward the bud (Video 1). In addition, cortical, less-dynamic GFP-D4H patches were visible in mother cells at longer exposures (Fig. 1 A and Video 2). The probe recognized sterols, since the vesicular structures and the PM staining disappeared in erg11 and erg28 mutants (Fig. S1, C and D) or in yeast incubated with the Erg1 inhibitor terbinafine (TBF; Bhattacharya et al., 2018; Fig. S1, C and E). These results suggested that secretory vesicles enriched in sterols (Klemm et al., 2009) might load the PM of daughter cells. Consistently, the polarized GFP-D4H staining was lost in secretory mutants (Spang, 2015; sec1-1, sec8-6, sec6-4, and sec4-8) at restrictive temperature (Figs. 1 A and S1 F). The Golgi-localized Osh4, which contributes to polarized secretion (Alfaro et al., 2011; Antonny et al., 2018; Fang et al., 1996; Kozminski et al., 2006; Li et al., 2002; Ling et al., 2014; Smindak et al., 2017), also played a prominent role in the process, as compared with other Oshes (Fig. S1 G).

In mother cells, more than 85% of the cortical GFP-D4H (cGFP-D4H) patches localized at rims of the cortical ER (cER; Fig. 1 B and Video 3). The GFP-D4H patches followed the ER dynamics (Video 4) and sometimes spread toward the ER (Fig. 1 B), indicating that they were linked. More than 65% of the cGFP-D4H also associated with transient foci of Abp1-mCherry, a marker of endocytic actin (Fig. 1 C). Triple labeling further demonstrated mCherry-D4H patches at cER rims, some associated with endocytic sites (Fig. 1 D). Quantitative immunoelectron microscopy (QEM) showed specific HA-GFP-D4H labeling at cER rims and the invagination neck and base (Figs. 1 E and S2 A). Analysis of the gold relative position (GRP; Fig. S2 A) showed coincidence of the HA-GFP-D4H labeling with that of Myo5 and Osh2/3 and, to a lesser extent, with the yeast amphiphysin Rvs167 and N-WASP, Las17 (Fig. 1 F).

The data suggested that sterols were accessible at certain cER subdomains, from where they could be extracted by Osh2. Consistently, the number and brightness of cGFP-D4H patches increased in the absence of Osh2 (Fig. 2 A). QEM further demonstrated the presence of engrossed HA-GFP-D4H–labeled cER rims in osh2Δ cells (Fig. 2 B). Deletion of Osh4 also caused accumulation of GFP-D4H foci, but mostly noncortical (Fig. S2 B). In this strain, the ER was grossly altered, with very little cER (Fig. S2 B). Depletion of the endocytic Osh3, which does not extract sterols (Encinar del Dedo et al., 2017; Schulz et al., 2009; Tong et al., 2013), or depletion of the sterol transporter Osh1, which sits on the nucleus/vacuole junction (Kvam and Goldfarb, 2004; Levine and Munro, 2001) and the Golgi (Levine and Munro, 2001), had no effect on the cGFP-D4H patches (Fig. 2 A).

To directly investigate if Osh2 extracted sterols from cER rims, we followed individual cGFP-D4H patches upon addition of TBF in either the presence or absence of Osh2. The GFP intensity immediately dropped upon drug imposition in WT cells but remained unaltered in the absence of Osh2 (Figs. 2 C and S2 C). Fenpropimorph (FPM), a downstream inhibitor, had a similar effect (Fig. S1 C; and Fig. S2, C and D; Marcireau et al., 1990). Upon 1-h TBF incubation, the GFP-D4H patch intensity dropped in WT, osh1Δ, and osh3Δ cells but was unaltered in osh2Δ yeast (Fig. S2 E). Thus, the data indicated that the cGFP-D4H patches decorated Osh2-dependent ERSESs. In agreement, Osh2-YFP transiently colocalized with mCherry-D4H at cER rims (Fig. 2 D).

Interestingly, treatment with OSW-1, an inhibitor of OSBP (Mesmin et al., 2017), had the opposite effect of TBF on the cGFP-D4H patch intensity in WT cells (Fig. 2 C; and Fig. S2, C and F). Similar to TBF, the OSW-1 effect required Osh2, but not Osh1 or Osh3 (Fig. 2 C; and Fig. S2, C and F). 1-h OSW-1 treatment did not grossly alter the polarized sterol pattern (Fig. S2 G).

As expected, Osh2 function at ERSESs required its sterol transfer activity. Deletion of its ORD (OSBP-related lipid-binding domain; osh2-ORDΔ) increased the cGFP-D4H patch intensity, which was restored to WT levels by expression of a chimera of the Osh2 N terminus with the Osh4 ORD (osh2-ORD4), but not with that of Osh6 (osh2-ORD6; Figs. 2 E and S2 H). In addition, mutation of the “double F in an acidic track” (FFAT) motif (osh2-FFAT*) impaired sterol extraction, indicating that Osh2 function required binding to VAPs (Encinar del Dedo et al., 2017; Figs. 2 E and S2 H). Finally, sterol extraction relied on PI4P counter-transport. Mutation of the Osh2 H1000, H1001, and R1230, predicted to impair PI4P and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂)binding (osh2-HHR*; de Saint-Jean et al., 2011; Wang et al., 2019), increased the cGFP-D4H patch intensity, while mutation of the surface patch III K1114, K1116, and K1118, required to sustain ORP2 PI(4,5)P₂ sterol counter-transport (Wang et al., 2019), did not. Further, depletion of Osh2 and Osh3 significantly increased the PM PI4P levels in mother cells (Stefan et al., 2011; Fig. S2 I), and expression of Osh2, but not Osh2-HHR*, partially restored the PI4P levels in the mutant (Fig. S2 I). Instead, no significant elevation of the PM PI(4,5)P₂ was detected in the mutants (Fig. S2 I).

Most Ergs localize throughout the ER (Dubreuil et al., 2019). However, the observation that GFP-D4H accumulated at discrete ER zones suggested that some sterol modifications recognized by the probe might occur at ERSESs. Consistently, we found that Erg7, Erg27, Erg6, and Erg2 (Fig. S1 C) showed a punctuated pattern, with bright spots near the nuclear envelope and fainter cortical patches associated to cER rims (Fig. 3 A and Video 5; Dubreuil et al., 2019). Erg7 closes the aromatic rings of the squalene epoxide, and Erg27 reduces the C3 oxygen, two molecular features recognized by PFO (Fig. S1 C; Maekawa, 2017). Consistent with a dense network of Erg contacts (Mo and Bard, 2005), double labeling demonstrated that these enzymes colocalized in all structures (Fig. S3 A and Video 6). The bright perinuclear Erg6-GFP spots, but not the fainter cortical patches, were lipid droplets (LDs) stainable with BODIPY (Fig. S3 B; Joshi et al., 2018). Reciprocally, a significant fraction of cortical Erg patches, but not LDs, associated with GFP-D4H (Fig. 3 B; and Video 7 and Video 8), labeling cortical ERSESs. In agreement, the cortical Erg patches marked endocytic hot spots (Figs. 3 C and S3 C; Video 9 and Video 10).

The data indicated that ERSESs accumulated a subset of sterol biosynthetic enzymes, which could physically contact the components of the ER–endocytic MCSs. In fact, coimmunoprecipitation (Fig. 3 D) and two-hybrid assays (Fig. 3 E) detected interactions between Myo5 and Erg6 or Erg27. The Erg-binding region in Myo5 included the TH2 (tail homology 2) domain and did not overlap with that for Osh2 (Encinar del Dedo et al., 2017), indicating a role for myosin-I as a scaffold, holding the sterol biosynthetic and transport machineries (Fig. 3 F). In addition, we demonstrated binding of Erg6 with the ORD of Osh2 in immunoprecipitation (Fig. 3 D; and Fig. S3, D and E), two-hybrid (Fig. 3 G), and pull-down assays (Fig. S3 F). Again, the Osh2-binding sites for Erg6 and Myo5 did not overlap (Fig. 3 G; Encinar del Dedo et al., 2017). Fluorescence microscopy also showed contact between the sterol synthesis and transport machineries (Fig. 3 H). No interaction between Erg6 and other Oshes was detected in two-hybrid assays, nor could we find binding between Osh4 and Erg2, Erg6, Erg7, or Erg27 in immunoprecipitations (Figs. 3 G and S3 E), suggesting that Osh2 plays a prominent role at cortical ERSESs. This did not discard specialized ER subdomains serving lipids to other transporters. This is likely for Osh4, which interacts with Erg11 (Tarassov et al., 2008) and whose depletion results in the accumulation of GFP-D4H foci (Fig. S2 B).

To test the functional significance of the Erg-Osh2 interaction, we mutated the PPPVP motif that mediates binding of Osh2 to Myo5 (Encinar del Dedo et al., 2017) within the osh2-ORD4 chimera, to generate a mutant osh2-ORD4-PPPVP* unable to directly or indirectly interact with the Ergs at ERSES (Fig. 3, E–G). Osh2-ORD4-PPPVP* failed to restore sterol extraction in an osh2Δ background (Fig. 3 I). Further sustaining the functional relevance of the Erg/Osh interaction, a chimera of Erg6 and Osh4 reduced the brightness of the cGFP-D4H patches to WT levels in osh2Δ cells, whereas neither of the two proteins did when separately expressed (Fig. 3 I). Also consistent with a key nonenzymatic scaffolding role of Erg6 sustaining sterol extraction was the observation that a catalytically inactive Erg6 (erg6-D152L) restored growth and sterol extraction in an erg6Δ strain, similar to the WT (Nes et al., 2004; Fig. 3, J and K).

Even though sterols are essential for endocytosis in yeast (Encinar del Dedo et al., 2017; Munn et al., 1999), our work showed that the sterol transport routes markedly differ in mother and daughter cells. Thus, we hypothesized that the mechanisms supporting endocytosis might diverge. PM accessible sterols in mothers are kept low, perhaps to prevent ectopic activation of the PI4P5 kinase (Nishimura et al., 2019) or assembly of the Exocyst (Inoue et al., 2006). This might impose the need for a localized sterol transport to endocytic sites. In daughters, loaded with sterols in a secretory pathway–dependent manner (Fig. 1 A), on-site sterol transfer should be dispensable. Consistent with this view, treatment of yeast with FPM and OSW-1 immediately delayed and weakened actin polymerization at endocytic sites in mother cells and consequently extended the life span of Sla1-GFP (Fig. 4, A and B). However, endocytosis in daughters smaller than 2 µm in diameter, essentially devoid of ERSESs, was unaltered under these experimental conditions (Fig. 4 A). This was despite sterols being equally required for endocytosis, as demonstrated by acute sterol sequestration with filipin or prolonged FPM and OSW-1 treatments (Fig. 4 A), which completely altered the sterol homeostasis (Fig. S3 G).

Also in agreement with the diverse sterol routes, depletion of Osh2 significantly expanded the life span of Sla1 in mothers (Fig. 5 A), but not in daughters (Fig. 5 A). The capacity of osh2 alleles to sustain endocytosis correlated with their capacity to extract sterols (Fig. S3 H; Encinar del Dedo et al., 2017). Reciprocally, preventing secretion, strongly influenced endocytic uptake in daughter cells (Johansen et al., 2016; Riezman, 1985), but not in mothers (Fig. 5 B).

Further sustaining a universal requirement for sterols in endocytosis, depletion of Erg28, cofactor of the essential Erg27, affected endocytic uptake in mother and daughter cells, similar to filipin (Fig. 5 C). In contrast, depletion of Erg6, which played a catalytically independent role at ERSESs (Fig. 3, J and K), affected endocytic uptake only in mothers (Fig. 5 C). Finally, deletion of the Myo5 TH2 domain, which mediated binding to Erg6 and Erg27 (Fig. 3 E), sensitized mother cells, but not daughter cells, to very low concentrations of filipin (Fig. 5 D). Altogether, the data demonstrated that on-site coupling of the sterol synthesis and transport machineries was required to sustain endocytosis in mother cells, but not in daughters, where PM loading with accessible sterols and endocytosis were linked to secretion.

The yeast strains used in this study are listed in Table 1. GFP, mCherry, YFP, and HA tags were fused at the C terminus of each protein by homologous recombination in the genome as described previously (Wach et al., 1997) using 50-nt oligonucleotides upstream and downstream of the stop codon. Genome-edited strains behaved as WTs. Strains without plasmids were grown in complete yeast peptone dextrose and strains with plasmids were selected on synthetic dextrose complete (SDC) lacking the appropriate nutrient (Dulic et al., 1991) at 25°C. Transformation of yeast was accomplished by the lithium acetate method (Ito et al., 1983). Strains used in this work are listed below.

DNA manipulations were performed as described previously (Sambrook et al., 1989). Enzymes for molecular biology were obtained from New England Biolabs. PCRs were performed with a Vent polymerase (New England Biolabs) and a TRIO thermoblock (Biometra). All plasmids used in this study bear ampicillin resistance for selection in Escherichia coli and are listed in Table 2, where the yeast features and inserts are described. The osh2-HHR* ORF was synthesized by Genescript. Information about the construction strategies are available upon request. The oligonucleotides used in this work for plasmid construction are listed below (Table 3).

SDS-PAGE was performed as described previously (Laemmli, 1970) using precasted Mini-PROTEAN TGX 4–20% acrylamide gels (Bio-Rad). For immunoblot, nitrocellulose membranes (Protan BA85; GE Healthcare) were probed with anti-HA (Anti-HA-Peroxidase High Affinity 3F10; Roche), peroxidase anti-peroxidase (Sigma-Aldrich) or anti-GFP (monoclonal antibody JL-8; Living Colors), followed by a peroxidase-conjugated secondary goat anti-rabbit IgG antibody (Sigma-Aldrich). Protein transfer, blotting, and chemiluminescence detection were performed using standard procedures. Detection of proteins was performed using the ECL kit (GE Healthcare).

For the anti-GFP-agarose immunoprecipitations from P13000 extracts, cells were glass bead-lysed in LB (25 mM Tris and 5 mM EDTA, pH 8.5) in the presence of protease inhibitors (1 mM PMSF, 5 µg/ml Leupeptine, 2.5 µg/ml Antipain, 1 µg/ml Pepstatin, and 1 µg/ml Aprotinin). Unbroken cells were eliminated at 700 g for 15 min at 4°C. The supernatant was diluted with the same volume of BB (10 mM Tris, 0.2 mM EDTA, and 0.2 mM DTT, pH 7.5) containing protease inhibitors. The post–700 g supernatants were spun at 13,000 g for 20 min, and the pellet was recovered and resuspended in 1 ml IP buffer (50 mM Tris, 150 mM NaCl, 5 mM EDTA, and 1% Triton X-100, pH 7.5). The extract was adjusted to 1 mg protein per ml, incubated 30 min at 4°C, and spin at 700 g for 15 min. The supernatant was incubated with 20 µl of 50% anti-GFP-Agarose (GFP-trap; Chromotek) for 1 h at 4°C. Beads were washed with IP buffer and subsequently with IP buffer without Triton X-100 and boiled in 25 µl SDS-PAGE sample buffer.

The Interaction Trap two-hybrid system was used (Gyuris et al., 1993). Plasmids pEG202, pJG4-5, and pSH18-34 and the strain EGY48 were kindly provided by Dr. R. Brent (Fred Hutchinson Cancer Research Center, Seattle, WA). To measure β-galactosidase activity, EGY48 cells bearing the lexAop-lacZ reporter plasmid pSH18-34 were cotransformed with the appropriate pEG202- and pJG4-5–derived plasmids and streaked out on 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside containing Synthetic Dextrose/Gal/Raf-His-Trp-Ura plates (0.67% yeast nitrogen base [Difco], 7 g/liter Na₂HPO₄, 3 g/liter NaH₂PO₄, 2% galactose, 1% raffinose, 40 mg/ml leucine, 80 mg/liter 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside [Sigma Chemicals], and 2% agar, pH 7). Pictures were taken after 2 d of growth at 24°C.

Unless otherwise stated, samples were observed on a Leica DMI6000 wide-field microscope equipped with a 63× APO (Apochromatic; NA 1.4) oil-immersion objective, a Hamamatsu Orca-R2 camera and a SOLA-SMII white LED illumination source. A fast external filter wheel for GFP (BP470/40) and mCherry (BP572/35) excitation was used to minimize channel acquisition mismatch. The system was controlled under the LAS X (Leica Application Suite X) software. Cells were grown to mid-log phase, harvested, diluted in a small volume of SDC medium, and imaged as wet mounts on a 2% agarose patch on the same medium and at the same temperature used for culture growth. For time lapses longer than 5 min, cells were mounted on slides with agarose pads prepared by dissolving 2% agarose in SDC medium containing appropriate nutrients.

Confocal live imaging was performed in an LSM Zeiss780 system equipped with a 63× APO (NA = 1.4) oil immersion objective. Two-channel time-lapse movies were acquired at 2-s intervals by line sequential scanning. Pixel size was set to 60 nm, and pinhole aperture was kept close to 1 a.u. to avoid resolution loss. Photon counting mode detection allowed to drastically reducing the amount excitation laser power, thus minimizing photobleaching and phototoxicity. GFP and mCherry conjugates were excited at 488-nm and 594-nm laser lines and registered at detection windows of 489–549 nm and 597–696 nm, respectively. For three-channel live acquisitions, the YFP was excited at 514 nm and detected at 517–553 nm, and the CFP conjugate was excited at 458 nm and detected at 463–509 nm. GFP-D4H dynamics assays were performed in a Leica Thunder 3D live cell, equipped with a 63× water immersion lens (NA 1.2), a fast and sensitive sCMOS camera DFC9000 and a Spectra-X Light Engine. GFP was excited/detected using BP475/28 and BP510/40 filters. Algorithm-driven high-speed autofocus was used to keep the cells in focus before and after addition of the drug. Built-in instant computational clearing algorithms were applied to remove out-of-focus light and allow for accurate fluorescence intensity quantification. Instant computational clearing algorithm parameters were always adjusted according to the objective used to avoid unwanted changes in object intensity or shape. Cells expressing GFP-D4H were grown to mid-log phase in SDC lacking the appropriate nutrient, harvested, diluted in a small volume of SDC medium, and placed in a 35-mm glass-bottom dish with 20-mm micro-well (ref. D35-20-1.5-N; Cellvis) previously coated with 15 µl of 0.2 mg/ml concanavalin A (ref. C7275; Sigma-Aldrich). For the 12-min dynamics, cells were imaged in the Leica Thunder 3D, at 5 s temporal resolution for two minutes, then they were treated with the properly drug (25 µM FPM [36772; Sigma-Aldrich], 5 µg/ml TBF [T8826; Sigma-Aldrich], or 8 µM OSW-1 [B0005-092456; BOC Sciences]), and the time lapse continued for 10 min. Kymographs of the cortical dots were performed and analyzed with ImageJ. For steady analysis of GFP-D4H cortical patches, only patches in touch with the PM in mother cells were considered. For analysis of endocytosis, SLA1 was genome edited to express C-terminal GFP in the corresponding WT or mutant strains, except for erg mutants, where we failed to integrate the GFP tag in the genome. In this case, Sla2-GFP expressed from a centromeric plasmid was used instead of Sla1-GFP, because it generates less cytosolic background. Cells were imaged every 0.5 s, and kymographs were generated and analyzed by ImageJ. To inspect the kinetics of endocytic proteins in daughter cells, we only considered daughter cells smaller than 2 µm of diameter, where Erg6-labeled cortical ERSESs are essentially not detectable.

Immunoelectron microscopy was performed as described previously (Idrissi et al., 2008). Briefly, cells were grown in yeast extract peptone dextrose medium (to 4–5 × 10⁶ cells/ml) and harvested over a disposable Stericup 0.22-µm-filter unit, leaving 5 ml of media. 25 ml of 1.2× fixative solution was immediately added to obtain final concentrations of 0.04 M KPO₄, pH 6.6, 0.6 M sorbitol, 4% formaldehyde (Polysciences), 0.4% glutaraldehyde (Fluka), 1 mM MgCl₂, 0.5 mM EGTA, 10 mM NaF, and 10 mM NaN₃. Cells were then transferred to a 50-ml Falcon, and fixation was continued overnight at 4°C with rolling. Subsequent steps (metaperiodate and ammonium chloride treatments, dehydration, infiltration, embedding, and sectioning) were performed as described previously (Mulholland et al., 1994) without further modifications. For immunolabeling, ultrathin sections were incubated for 15 min in blocking buffer (10 mM KPO₄, pH 7.5, 150 mM NaCl, 2% BSA, and 0.05% Tween-20), transferred to a 25-µl drop of primary antibody in blocking buffer for 3 h, and washed over 30 min in washing buffer (10 mM KPO₄ buffer, pH 7.5, 150 mM NaCl, and 0.05% Tween-20). After blocking for 15 min, grids were incubated with the corresponding gold-conjugated secondary antibody for 60 min and washed first for 30 min in washing buffer and 30 min in washing buffer without Tween-20. All steps were performed at room temperature. Grids were then washed in double distilled water, fixed for 20 min in 8% glutaraldehyde, and poststained with uranyl acetate (2% in water) over 30 min and lead citrate for 30 s. Anti-HA rat monoclonal antibody (3F10; Roche) was used as primary antibodies. 12 nm gold-conjugated goat anti-rat IgG (Jackson ImmunoResearch Laboratories) was used as a secondary antibody (diluted 1:25). Ultrathin sections were examined using a Jeol 1010 transmission electron microscope at 50 kV accelerating voltage. Micrographs of the yeast PM invaginations were acquired at 100,000× magnification with a MegaView III CCD camera and the image acquisition software analySIS (Soft Imaging System). Adjustments of image size, brightness, and contrast were performed on ImageJ. Labeling was considered specific when the labeling for a specific cellular structure (ER rims or invaginations) was at least three times higher for the HA-tagged strain that for the nontagged strain. Labeling was performed with the lowest possible antibody concentration to obtain one gold per image to statistically study independent events. The GRP of the GFP-D4H was compared with the GRP for immunogolds decorating the different proteins reported previously (Encinar del Dedo et al., 2014; Fernández-Golbano et al., 2014; Idrissi et al., 2012; Idrissi et al., 2008).

Quantifications and kymographs were performed with ImageJ (1997–2016, W.S. Rasband, National Institutes of Health; http://imagej.nih.gov/ij). Average, standard deviation, and P values for the two-sided Student’s t test of statistically significant differences were calculated with Microsoft Excel. Data distribution was assumed to be normal, but this was not formally tested.

Fig. S1 shows a scheme of the GFP-D4H sterol probe, the effect on the Sla1-GFP cortical dynamics of the GFP-D4H probe expression, a scheme of the late ergosterol biosynthetic pathway, the controls showing that the GFP-D4H probe recognizes sterols in yeast, and the effect of different sec and osh mutants on the polarization of the GFP-D4H probe. Fig. S2 shows the QEM data on the localization of the GFP-D4H probe, the effect of FPM on the GFP-D4H cortical patch intensity; the effects of Osh4 depletion on the GFP-D4H cortical patch intensity and numbers; kymographs of representative GFP-D4H cortical patches upon treatment with TBF, FPM, or OSW-1; effects of TBF and OSW-1 on control strains lacking Osh1 or Osh3; effect of 1-h treatment with OSW-1 on the polarity of GFP-D4H; representative GFP-D4H cortical patches of yeast expressing different osh2 alleles; and the effects of osh2 mutation on the PI4P and PI(4,5)P₂ PM levels. Fig. S3 shows colocalization of Erg proteins, BODIPY staining of Erg6 patches, time lapse of Abp1-mCherry in cells expressing Erg27-GFP, control immunoprecipitations or Erg-GFP–expressing cells, pull-down of the ORD of Osh2 and Erg6-GFP, the effect of 2-h treatments of OSW-1 and FPM on the GFP-D4H staining, and complementation of the endocytic defect of an osh2Δ strain with different OSH2 alleles and the ERG6-OSH4 chimera. Video 1 and Video 2 show WT yeast expressing GFP-D4H. Video 3 shows a 3D view of a section of a WT cell where GFP-D4H patches can be seen at the rims of cER. Video 4 shows the dynamics of a GFP-D4H patch at the rim of the cER. Video 5 shows a 3D view of a section of a WT cell where Erg6-mCherry can be observed in big perinuclear structures, which correspond to LDs, and small cortical patches, which correspond to ERSESs. Video 6 shows colocalization of Erg6-mCherry and Erg27-GFP. Video 7 and Video 8 show colocalization of Erg6-mCherry and GFP-D4H at the cortex. Video 9 shows a 3D view of a section of a WT cell demonstrating colocalization of Abp1-mCherry and Erg6-GFP. Video 10 shows that Erg27-GFP cortical patches label endocytic hot spots.

We acknowledge M. Muñiz, H. Riezman, and M. Molina for critical reading; Leica Microsystems for support of the Molecular Imaging Platform; and C. Roncero (Instituto de Biología Funcional y Genómica, Salamnca, Spain), M. Munson (University of Massachusetts Medical School, Worcester, MA), K. Kozminski (University of Virginia, Charlottesville, VA), W. Prinz (NIH. Bethesda, MD), O. Gallego (Universitat Pompeu Fabra, Barcelona, Spain), M. Molina (Universidad Autonoma de Madrid, Madrid, Spain), A. Pol (Universitat de Barcelona, Barcelona, Spain), R. Piper (University of Iowa, Iowa City, IA), and A. Minard (University of Iowa, Iowa City, IA) for sharing plasmids and reagents.

This study was financed by the Spanish "Ministerio de Economia Industria y Competitividad" (grant BFU2017-82959-P).

The authors declare no competing financial interests.

Author contributions: J. Encinar del Dedo contributed to conceptualization, execution, and analysis of most experiments. I.M. Fernández-Golbano performed and analyzed QEM and some fluorescence micrographs experiments and generated plasmids and strains. P. Meler performed two-hybrid and pull-downs assays. L. Pastor generated plasmids and strains and purified proteins. C. Ferrer-Orta helped with protein purification. E. Rebollo designed the fluorescence imaging methodology. M.I. Geli contributed to conceptualization and analysis, funding acquisition, and supervision; performed some fluorescence micrograph and biochemical experiments; and wrote the original draft.