Ltc1 is an ER-localized sterol transporter and a component of ER–mitochondria and ER–vacuole contacts

Organelle contact sites are critical for interorganellar lipid transfer, ion homeostasis, and organelle membrane dynamics. In yeast, a contact site is formed between the ER and mitochondria by the ER–mitochondria encounter structure (ERMES), composed of four subunits: outer membrane mitochondrial Mdm10 and Mdm34, cytosolic Mdm12, and integral ER Mmm1 (Kornmann et al., 2009; Stroud et al., 2011). Mdm34, Mdm12, and Mmm1 are members of the extended synaptotagmin domain family, which bind lipids (Toulmay and Prinz, 2012; Schauder et al., 2014), and consistently, ERMES has been implicated in lipid exchange between the ER and mitochondria (Kornmann et al., 2009). ERMES-marked ER–mitochondria contacts are also linked to mitochondrial division and mitochondrial DNA (mtDNA) segregation and mitophagy, suggesting they represent microdomains that coordinate cellular functions (Murley et al., 2013; Böckler and Westermann, 2014). Recently, additional mitochondrial contacts sites have been described whose functions are related to ERMES, including V-CLAMP, a contact between mitochondria and vacuoles (Elbaz-Alon et al., 2014; Hönscher et al., 2014) and the ER protein complex EMC, proposed to form an additional contact between ER and mitochondria (Lahiri et al., 2014). In the case of V-CLAMP and ERMES, these sites are coregulated. Thus, different contact sites between organelles are coordinated, but how this regulation is achieved remains unknown.

To understand the composition, mechanisms, functional scope, and modes of communication of organelle contact sites, we explored the environment of ERMES using proteomics and identified a conserved uncharacterized ER protein, Ylr072w, as an ERMES interactor, which belongs to a larger uncharacterized protein family. Based on cytological and biochemical analyses, we have renamed Ylr072w Lipid transfer at contact site 1 (Ltc1). Ltc1 localizes to both ER–mitochondria and ER–vacuole contact sites in partnership with the organelle-specific components Tom70/71 and Vac8, respectively. Our data suggest that Ltc1 functions to transport and/or sense sterols at contact sites to facilitate separate organelle-specific functions and support a model in which the dual organelle localization of Ltc1 is a means of coordinating the functions of mitochondria and vacuoles.

To explore the composition of ER–mitochondria contacts, we immunopurified GFP-tagged versions of ERMES subunits Mmm1, Mdm12, and Mdm34 from dithiobis succinimidyl propionate (DSP) cross-linked yeast extracts to identify interacting proteins by tandem mass spectrometry (IP-MS/MS). As shown in Table 1, proteomic analysis of ERMES subunit purifications identified multiple core ERMES complex components and the ERMES auxiliary subunit Gem1, consistent with previous observations (Kornmann et al., 2011; Stroud et al., 2011). In addition, we identified an uncharacterized protein, Ylr072w. Deletion of YLR072W from cells did not cause significant growth defects on media with fermentable (glucose) or nonfermentable (ethanol/glycerol) carbon sources (Fig. 1 A) nor defects in mitochondrial morphology (Fig. 1 B). However, deletion of YLR072W in combination with deletion of MDM34 caused a significant synthetic growth defect as compared with deletion of MDM34 alone, indicating that YLR072W has both a physical and functional relationship to ERMES (Fig. 1 C). We used this MDM34-dependent phenotype to determine that a YLR072W-yEGFP fluorescent protein fusion allele integrated at its endogenous locus was functional, as the growth of YLR072W-yEGFP Δmdm34 cells was similar to Δmdm34 cells (Fig. 1 C).

Ylr072w localized to punctate structures in cells, of which 35–40% colocalized with ERMES foci, marked by Mdm34-mCherry, suggesting that Ylr072w foci localize to ER–mitochondria contact sites (Fig. 1, D and G). However, Ylr072w foci formation was independent of ERMES, and we also observed ERMES foci formation in the absence of Ylr072w (Fig. S1, A and B). We directly examined whether Ylr072w marks ER–mitochondria contact sites by examining the localization of Ylr072w in cells relative to ER and mitochondria, using ER-targeted DsRed (DsRed-HDEL) and mitochondrial matrix–targeted BFP (mtBFP), respectively. We observed that a vast majority of Ylr072w foci (80–90%) were present at ER–mitochondria contact sites (Fig. 1, E [yellow arrowheads] and G). However, we also observed Ylr072w foci that were not localized near mitochondria (Fig. 1 E, open arrowhead). The localization of a proportion of nonmitochondrial Ylr072w foci to the nuclear envelope prompted us to examine their localization relative to nucleus–vacuole junctions (NVJs; Pan et al., 2000; Roberts et al., 2003; Millen et al., 2008). Relative to the ER and vacuoles, we observed that Ylr072w localized to both NVJs (Fig. 1 F, open arrowheads; and Fig. S1 C, open arrowheads) and other non-NVJ ER–vacuole contact sites (Fig. 1 F, closed arrowheads; and Fig. S1 C, closed arrowheads). Quantification of Ylr072w localization at ER–mitochondria contacts and ER–vacuole contacts yielded a proportion >100% (Fig. 1 G), which is likely the result of error introduced by the resolution limitation of fluorescence microscopy. In total, our data indicate that Ylr072w localizes to both ER–mitochondria and ER–vacuole contacts.

Phylogenetic analysis and structure prediction programs indicate that Ylr072w contains two predicted lipid-binding domains and is a member of a larger family with multiple homologues present in nearly all eukaryotes (Fig. 2 A). Both Phyre2 (Kelley and Sternberg, 2009) and iTasser (Yang et al., 2015) reveal a conserved N- to C-terminal domain architecture within the family: an unstructured N terminus, a GRAM domain structurally similar to pleckstrin homology (PH) domains (Begley et al., 2003), a previously unannotated StART-like domain also called a VASt domain (Khafif et al., 2014); and a C-terminal hydrophobic α helix. The VASt domain in this family shares no sequence similarity to StART domains, but is predicted to form a similar hydrophobic pocket that accommodates a lipid molecule (Khafif et al., 2014). We also include vertebrate GramD2 and GramD3 proteins in this family as phylogenetic analyses indicate they are derived from a common ancestor with GramD1a-c, suggesting that they arose through VASt domain loss. Based on these structural predictions and our observations, we rename YLR072W LTC1 for Lipid Transfer at Contact site 1 and the other members of the yeast family LTC2 (YFL042C), LTC3 (YHR080C), and LTC4 (YSP2/YDR326C). Although LTC1 is paralogous to LTC2, our physiological data suggest that it possesses nonredundant functions (Fig. 1 and see Fig. 4).

To characterize Ltc1 molecular features, we expressed and purified the predicted soluble domain of Ltc1 (amino acids 1–590, 6xHis-Ltc1(ΔTM)) from Escherichia coli. Size exclusion chromatography coupled with multiangle light scattering indicated that recombinant Ltc1 is a monomer (73,690 ± 0.283% D). To test whether Ltc1 facilitates lipid transfer, we used an established in vitro lipid transfer assay (Schulz et al., 2009), in which dense sucrose-loaded donor liposomes containing one of three different types of radiolabeled lipids, phospholipids, triolein (a triglyceride), or cholesterol, were incubated with lighter, unlabeled acceptor liposomes in the presence and absence of Ltc1(ΔTM) and subsequently separated by centrifugation (Fig. 2 B). Radioactive cholesterol, but not triolein or major classes of phospholipids, was transferred to acceptor liposomes in an Ltc1-dependent manner (Fig. 2 B), and transport efficiency was dependent on Ltc1(ΔTM) concentration and time (Fig. 2, C and D). Thus, Ltc1 selectively transports sterols between membranes in vitro, suggesting that it functions as a sterol transfer protein and/or sensor in vivo.

To examine the molecular basis for Ltc1’s dual localization, we immunopurified Ltc1-yEGFP from DSP cross-linked yeast whole cell lysates using anti-GFP antibodies. Consistent with our cytological and proteomic data (Fig. 1 and Table 1), we identified the ERMES subunit Mdm34 and the ERMES-associated protein Gem1 (Table 2, wild type [WT]). Based on the number of spectral counts, the most abundant proteins identified in Ltc1 purifications were the paralogous mitochondrial preprotein import receptors, Tom70/71 (Table 2, WT; Schlossmann et al., 1996; Schmidt et al., 2010), also identified in the ERMES complex subunits purifications (Table 1). Localization of a functional Tom71-yEGFP fusion relative to Ltc1 in cells revealed that, in addition to a uniform labeling of the mitochondrial outer membrane, Tom71 was present in focal structures that colocalized with Ltc1 (Fig. 3 A and Fig. S1 A). We tested whether Tom70/71 is required for Ltc1’s localization to mitochondria. In Δtom70 Δtom71 cells, Ltc1 was localized to foci, but in contrast to WT cells, a vast majority of these were not localized to mitochondria (Fig. 3 B) and instead were at ER–vacuole contacts (Fig. S1 C). Tom70/71 also functions redundantly to target the F-box protein Mfb1 to mitochondria (Kondo-Okamoto et al., 2008). Although not detected in our Ltc1 purifications, we tested whether localization of Ltc1 to mitochondria was Mfb1 dependent. In contrast to deletion of TOM70 and TOM71, deletion of MFB1 did not alter the localization pattern of Ltc1 relative to WT (Fig. S1 D). Thus, these data indicate a major and perhaps direct role of Tom70/71 in the localization of Ltc1 to ER–mitochondria contact sites.

Additional proteins were identified in Ltc1 purifications, including the vacuolar protein Vac8, which based on total spectral counts was the most abundant vacuolar protein present (Table 2, WT). Vac8 has diverse functions in autophagy, vacuolar transport, and in the formation of NVJs (Wang et al., 1998; Veit et al., 2001; Subramanian et al., 2006; Tang et al., 2006). Of note is that Tom70/71 and Vac8 have TPR repeats and Armadillo repeats, respectively, which form protein–protein interaction platforms (Wang et al., 1998; Wu and Sha, 2006; Grove et al., 2008; Li et al., 2009), suggesting that similar to Tom70/71 for mitochondria, Vac8 may function to localize Ltc1 to vacuoles. Consistently, in Δvac8 cells, Ltc1 localized to foci exclusively at ER–mitochondria contact sites (Fig. 3 B and Fig. S1 E). Vac8 is required for the formation of NVJs through a direct interaction with the outer nuclear envelope protein Nvj1 (Pan et al., 2000). Thus, we tested whether the Ltc1 localization to ER–vacuole contacts was dependent on NVJ formation. We observed that Ltc1-yEGFP localization was unaffected in Δnvj1 cells (Fig. S1 E), indicating that Ltc1-dependent ER–vacuolar contact is distinct from Nvj1-dependent NVJ contact sites. In cells harboring deletions in TOM70/71 and VAC8, Ltc1 foci were absent, and instead, Ltc1 was localized diffusely throughout the ER (Fig. 3 C), consistent with Ltc1 being an ER membrane–associated protein. In contrast, Ltc1 formed foci localized to vacuoles in Δnvj1 Δtom70 Δtom71 cells, similar to its localization in Δtom70 Δtom71 cells (Fig. S1 F). Together our data indicate that Tom70/71 and Vac8 are partners of Ltc1 in the formation of ER–mitochondria and ER–vacuolar contact sites/tethers, respectively, and suggest that they may regulate the relative distribution of Ltc1 to these distinct organelle contacts.

To test whether the dual localization of Ltc1 is a means for organelle communication, we first asked whether Ltc1 has distinct functions at mitochondria and vacuoles. Consistent with this, we found that deletion of TOM70 and TOM71, but not VAC8, produced a synthetic lethal/sick phenotype in combination with deletion of MDM34 (Fig. 4 A). However, given that Tom70 and Tom71 function in mitochondrial protein import, it is possible that the synthetic lethality/sickness observed in Δtom70 Δtom71 Δmdm34 cells is not a specific consequence of defects in Ltc1 localization to mitochondria. Thus, we searched for cis-regulators of Ltc1 localization to interrogate Ltc1 organelle-specific functions.

Ltc1 contains a GRAM domain that in other contexts controls the cellular localization of proteins via protein and/or phosphoinositide lipid interactions (Doerks et al., 2000; Begley et al., 2003). We asked whether the GRAM domain of Ltc1 determines its localization to ER–mitochondria and/or ER–vacuolar contacts. Ltc1(ΔGRAM) (deletion of amino acids 145–360) localized exclusively to ER–vacuole contact sites, a pattern similar to that observed in Δtom70 Δtom71 cells (Fig. 4 B). Thus, the GRAM domain is an essential determinant of Ltc1’s localization to ER–mitochondria contacts, although we do not presently understand the molecular basis for GRAM-dependent Ltc1 mitochondrial localization. Transformed plasmids containing full-length Ltc1-yEGFP, but not Ltc1(ΔGRAM)-yEGFP or an empty vector, rescued the synthetic lethal/sick interaction phenotype of Δltc1 Δmdm34 cells (Fig. 4 C). Together, these observations indicate that Ltc1-dependent functions at ER–mitochondria and ER–vacuole contact sites are nonredundant and separable.

In response to glucose starvation, cycloheximide treatment, or weak acid stress, the vacuole membrane rearranges into ergosterol-enriched (marked by Ivy1) and ergosterol-depleted (marked by Vph1) lipid domains (Toulmay and Prinz, 2013). Given that Ltc1 possesses sterol transport activity in vitro, we examined whether it is required for ergosterol-dependent vacuolar lipid domain formation in cells. Vacuolar domain formation was severely defective in Δltc1 cells after glucose starvation and cycloheximide treatment, but was unaffected in response to weak acid stress as compared with WT cells (Fig. 4 D). These observations are consistent with the in vitro ergosterol transport activity of Ltc1 and suggest that Ltc1 functions at ER–vacuole contacts in the stress-dependent reorganization of vacuolar membranes. These results also support the idea that different stresses mediate vacuolar domain formation through different mechanisms (Toulmay and Prinz, 2013).

We asked whether the relative localization of Ltc1 to ER–mitochondria and ER–vacuole contacts alters organelle-specific Ltc1 functions. Specifically, we increased the proportion of Ltc1 localized to the ER–vacuole contact sites and examined vacuolar domain formation in unstressed cells (Fig. 4, B and E; and Fig. S2). We accomplished this by deleting TOM70 and TOM71 or by replacing Ltc1 with Ltc1-ΔGRAM and assessed vacuolar domain formation based on patterned Vph1 localization in cells after exponential growth for over a dozen generations in nutrient-replete media. Under these conditions, vacuolar domains were readily observed in Δtom70 Δtom71 and Ltc1-ΔGRAM Δltc1 cells, but not in WT or Δltc1 Δtom70 Δtom71 cells (Fig. 4, B and E; and Fig. S2). These observations demonstrate that Ltc1 functions specifically at the ER–vacuole contact to control vacuole membrane domain formation and that shifting the localization of Ltc1 to ER–vacuole contact sites is sufficient to induce vacuolar membrane domain formation under normal growing conditions. In addition, we observed that Ltc1-ΔGRAM localization in Δltc1 cells was restricted to ergosterol-enriched domains (Fig. 4 B, open arrowheads; and Fig. S2). The exclusive localization of Ltc1 to ergosterol regions in the absence of its GRAM domain suggests that the VASt domain harbors the ergosterol binding/transport activity of Ltc1.

Our data show that ER membrane–associated Ltc1 is capable of transporting sterols and forming ER–mitochondria and ER–vacuole membrane contacts by partnering with the organelle-specific components Tom70/71 and Vac8. We envision that at these organelle contacts, Ltc1 acts as dynamic organizer of local membrane lipid composition rather than, or in addition to, mediating bulk sterol transfer in cells. Regulation of local lipid composition by Ltc1 may promote the formation of specialized domains, an extreme example of which is the stress-induced microscale vacuolar domains in yeast. Ltc1-induced lipid domains could act cooperatively to regulate the localization of proteins that, together with a selective lipid milieu, would act as regulators of organelle biogenesis and signaling.

Although the physiological roles of vacuole membrane domains is poorly understood (Toulmay and Prinz, 2013; Wang et al., 2014), our proteomic data suggest that the Target-Of-Rapamycin Complex 1 (TORC1) subunit Kog1 as well as the downstream TORC1 target Npr1 are in close proximity to Ltc1 at the vacuole (Table 2). Thus, one possible function of Ltc1-dependent domain formation may be in TORC1 regulation. Consistent with this possibility, Gtr2, a Rag GTPase component of the EGO (Exit from G₀) complex, a conserved upstream regulator of TORC1 (Dubouloz et al., 2005; Bonfils et al., 2012), is also localized with Ltc1 and Ivy1, a protein with a putative BAR-like domain, in the ergosterol-enriched vacuolar domain (Toulmay and Prinz, 2013). Domain formation at the vacuole may thus function similar to TORC2 regulation at the plasma membrane, where the BAR domain activators Slm1/2 are dynamically partitioned between the eisosome and TORC2 membrane domain compartments to regulate TORC2 activity in response to stress (Olivera-Couto et al., 2011; Berchtold et al., 2012; Niles et al., 2012). The exact molecular role of Vac8 in Ltc1-mediated vacuolar domain formation as well as whether Ltc1 affects and/or coordinates Vac8’s known functions in NVJ formation, vacuolar motility, and autophagy (Wang et al., 1998; Tang et al., 2006) remain to be determined.

Our data suggest Ltc1 possesses a separate lipid-linked function at mitochondria as the simultaneous loss of Ltc1 and ERMES functions produces a synthetic lethal/sick growth phenotype. Loss of ERMES subunits alters the levels of mitochondrial lipids relative to WT, including ergosterol (Tan et al., 2013). Thus, we speculate that Ltc1 function at mitochondria, similar to its function at vacuoles, is related to sterols. However, although sterols are important for the maintenance of mitochondrial morphology in yeast, their exact functions remain poorly described (Altmann and Westermann, 2005). In mammals, however, the nonvesicular transport of sterols from ER to mitochondria has been specialized for cell type–specific steroid hormone production. A mammalian homologue of Ltc1, GramD1b (Fig. 2 A), is highly expressed in steroidogenic Leydig cells (McDowell et al., 2012) and in the adrenal glands (Rosenbloom et al., 2015), suggesting that it may function in this capacity.

Our observations raise the possibility that Ltc1 via Tom70/71 might play an important regulatory role in mitochondrial biogenesis specific to Tom70/71 clients, which are the metabolite carrier proteins of the inner mitochondrial membrane (Hines et al., 1990; Söllner et al., 1990; Steger et al., 1990; Schlossmann et al., 1996). Cytosolic kinases, such as the nutrient-sensing kinase PKA, regulate Tom70/71 through phosphorylation (Schmidt et al., 2011), which may be a means of regulating Ltc1–Tom70/71 interactions in addition to regulating protein import. The exact function of Ltc1 at mitochondria and whether the canonical function of Tom70/71 in protein import is integrated with Ltc1 remain outstanding questions.

Although Ltc1 functions at mitochondria and vacuoles are separate, we show that increasing the proportion of Ltc1 at ER–vacuole contacts is sufficient to induce ergosterol-enriched vacuolar domains under normal growth conditions. Thus, we propose that the dual localization of Ltc1 to mitochondria and vacuolar contacts serves as a unique mode of interorganellar communication. Although the physiological pathways that regulate the relative distribution of Ltc1 at these distinct contacts are not known, it is likely that such coordination will be important for nutrient sensing and signaling (Hönscher et al., 2014) and cellular replicative aging (Hughes and Gottschling, 2012), which both involve extensive cross-talk between mitochondria and the vacuole.

All Saccharomyces cerevisiae strains constructed in this study are based on the W303 (ade2–1; leu2–3; his3–11, 15; trp1–1; ura3–1; can1–100 [Rothstein, 1983]) genetic background and are listed in Table S1. Strains were constructed by mating and/or by direct transformation using the lithium acetate method. Correct targeting of chromosomal integrations was confirmed by PCR. Fluorescent protein fusion alleles were derived from plasmids described in Sheff and Thorn (2004; yEGFP and mCherry) and derivatives described in Lackner et al. (2013; yEmCherry).

To generate pRS313::LTC1-yEGFP, genomic DNA from the W303 LTC1-yEGFP yeast strain was used to amplify a PCR product containing LTC1-yEGFP plus its native promoter (250 bp 5′ of start codon) and the entire ADH1 terminator used in the yEGFP::HIS cassette, which was then cloned into linearized pRS313 (Sikorski and Hieter, 1989) using a Gibson isothermal assembly reagent (New England Biolabs, Inc.) according to the manufacturer’s instructions. This was then used to generate pRS313::LTC1(ΔGRAM)-yEGFP by isothermal assembly of PCR products that when recombined would remove the nucleotides encoding amino acids 145–360 of Ltc1. To generate 6xHis-tagged Ltc1(ΔTM) for expression in E. coli, the DNA encoding amino acids 1–590 of the protein was amplified from yeast genomic DNA and cloned directionally into pET15b using 5′-XhoI/3′-BamHI sites. pVT100U-mtTagBFP was constructed by synthesizing yeast codon–optimized TagBFP de novo and replacing GFP in pVT100U-mtGFP (Westermann and Neupert, 2000) with 5′-KpnI/3′-XhoI sites.

pYES-mitoTagBFP, which drives galactose-inducible expression of mitoTagBFP from a GAL1 promoter (Murley et al., 2013); pYX142-mtDsRed, a CEN/ARS plasmid driving mitochondrial matrix–targeted DsRed (mtDsRed) expression from a TDI promoter (Westermann and Neupert, 2000); pKW1803 (DsRed-HDEL), which expresses a fusion protein containing the Kar2 signal sequence, DsRed, and an HDEL retention sequence (Madrid et al., 2006); and pRS305 3xBFP-Pho8, an integrating plasmid driving expression of 3xBFP-Pho8 from a PGK1 promoter (Graef et al., 2013), are all previously described in more detail.

Cells were grown to log phase (OD₆₀₀ ∼0.6–1.0) at 30°C in the appropriate synthetic medium to select for plasmids, concentrated by centrifugation, deposited directly on a glass slide, and then sealed under a #1.5 coverslip using nail polish. Cells in Fig. 1 E were grown in synthetic medium containing 2% galactose to drive expression of mtTagBFP from pYES-mtTagBFP; otherwise, cells were grown in synthetic medium with 2% glucose. Cells were imaged at 25°C.

Cells in Fig. 1 (B and D–F), Fig. S1 (A and C), Fig. 3 A, and Fig. 4 E were imaged on a DeltaVision Real-Time microscope (IX70 DeltaVision; Olympus) using a 60× 1.40 NA objective lens (Olympus) and a 100-W mercury lamp (Applied Precision). Light microscopy images were collected using an integrated, cooled charge-coupled device (CCD)–based camera (CoolSNAP HQ; Photometrics) equipped with an Interline Chip (Sony). Datasets were processed using softWoRx’s (Applied Precision) iterative, constrained three-dimensional deconvolution method to remove out-of-focus light.

Cells depicted or analyzed in Fig. 3 B, Fig. 4 B, Fig. S1 (B and D–F), and Fig. S2 were imaged using the spinning disc module of a Marianas SDC Real-Time 3D Confocal-TIRF microscope (Intelligent Imaging Innovations, 3i) fit with a Yokogawa spinning disk head, a 63× 1.40 NA (Olympus; Fig. S2) or 100× 1.46 NA objective (Olympus; Fig. 3 B, Fig. 4 B, Fig. S1 [B, D, and E], and Fig. S2), and EMCCD camera.

For stress treatments in Fig. 4 D, cells were grown for >12 generations to mid-log phase (to ensure no residual vacuole domains from cells in the colony) in synthetic complete (SC) media containing dextrose as a carbon source (SD) and harvested by centrifugation. They were then resuspended in an equal volume of SD media containing 60 mM sodium acetate, pH 5.2, 50 µg/ml cycloheximide, or SC media without glucose. For glucose starvation, cells were washed three times with water before resuspending them in an equal volume of SC-glucose. All treatments lasted 3 h. Cells were imaged on the DeltaVision wide-field deconvolution microscope described above. Experiments were performed in independent biological replicates.

Images were manipulated in Photoshop (Adobe). In some images found in this paper, linear adjustments were made to brightness and contrast.

Genes encoding ERMES subunits (MDM12, MDM34, and MMM1) were tagged at their C terminus with GFP at their native chromosomal locus. GFP fusion proteins were immunopurified from whole-cell lysates as previously described (Lackner et al., 2013). Cells were grown overnight to log phase (OD₆₀₀ = 1) in YPD medium (2% glucose, 2% peptone, and 1% yeast extract supplemented with adenine and tryptophan), harvested by centrifugation, washed once with distilled water, and resuspended 1:1 with IPLB buffer (20 mM Hepes, pH 7.4, 150 mM potassium acetate, 2 mM magnesium acetate, 1 mM EGTA, and 0.6 M sorbitol) containing protease inhibitor cocktail Set 1 (PIC; EMD Millipore). Cell suspensions were flash-frozen dropwise in liquid nitrogen. Frozen cells were lysed in a freezer mill (Spex 6970 EFM) and using three lysing periods of 2 min at speed “7” with 2 min of chilling between each. Lysates were stored at −80°C. Frozen cell lysates were thawed in a room temperature water bath with additional PIC, then clarified of unlysed cells and debris by centrifugation at 500 g for 5 min. Clarified lysates were treated with a reversible cross-linker DSP (Thermo Fisher Scientific) at a concentration of 1 mM for 30 min at 4°C, nutating at 50 rpm. The reaction was quenched by addition of Tris, pH 7.5, to 100 mM and incubation on ice for 10 min. Membranes were solubilized by addition of digitonin to a final concentration of 1% and nutating at 4°C at 50 rpm. Lysates were clarified by centrifugation at 12,000 g for 10 min at 4°C. Clarified lysates were incubated with 50 µl μMACs monoclonal mouse anti-GFP magnetic microbeads (Miltenyi Biotec) incubated at 4°C for 30 min. μMACS columns were equilibrated in IPLB + 1% digitonin + 1× PIC. Lysates and anti-GFP beads were flown over the column, and bound beads were washed three times with 800 µl IPLB + 0.1% digitonin + PIC and then twice more with 500 µl IPLB. On-bead digestion with trypsin was achieved by applying 25 µl of elution buffer 1 (EB1; 2 M Urea, 50 mM Tris-HCl, pH 7.5, 1 mM DTT, and 5 µg/ml Trypsin) to the column and incubating for 30 min. Digested proteins were eluted by adding 2× 50 µl EB2 (2 M Urea, 50 mM Tris-HCl, pH 7.5, and 5 mM chloroacetamide). Eluted proteins were incubated overnight at 25°C and then quenched by addition of 1 µl trifluoroacetic acid (TFA). Purified proteins were identified by tandem mass spectrometry in the University of California, Davis Proteomics Core. Mock purifications from cells that did not express GFP were used as a control for nonspecific interactions.

Ltc1-yEGFP was immunopurified using a modified protocol. Cells were grown in YPD and harvested routinely by centrifugation as above. Cells were then resuspended in a buffered solution (RIPA; 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 1 mM EDTA). Suspended cells were flash-frozen dropwise in liquid nitrogen and lysed in a freezer mill. Clarified lysates were then cross-linked with DSP and quenched as above. Lysates were then solubilized by adding detergents to a final concentration of 1% sodium deoxycholate, 1% NP-40, and 0.1% SDS. Lysates were clarified and incubated with anti-GFP μMACs beads as above. Bound anti-GFP beads were washed three times with RIPA + 0.1% sodium deoxycholate, 0.1% NP-40, 0.01% SDS + PIC, and then twice more with RIPA (no detergents or PIC). On-bead digest, protein elution, and MS/MS were performed as above.

Hexahistidine-tagged Ltc1 was purified from BL21 E. coli containing RIPL plasmid (encoding nonabundant tRNAs) and pET15b::YLR072W(aa1–590). Cultures were grown to mid-log phase in LB media at 37°C and then shifted to 18°C for overnight expression in 0.5 mM IPTG. Cells were harvested routinely by centrifugation, resuspended in a buffer (50 mM Hepes, pH 8, 500 mM NaCl + 350 µg/ml PMSF) and lysed in a microfluidizer, upon which Triton X-100 was added to 0.1%. Insoluble proteins and debris were removed by centrifugation at 35,000 rpm in a Beckman 45Ti rotor for 45 min. The soluble protein fraction was incubated with Ni-IDA resin and washed twice with buffer (50 mM Hepes, pH 8, 500 mM NaCl, 5 mM imidazole, and 0.1% Mega-8 [Dojindo Molecular Technologies Inc.]). Ni-IDA resin was loaded onto an FPLC column and proteins were eluted with 250 mM imidazole. Peak fractions were pooled, concentrated, and then run on a size-exclusion chromatography column (Superdex 200) in a buffer (50 mM Hepes, pH 8, 500 mM NaCl, 0.1% Mega-8, and 0.01% BME), and peak fractions were pooled and concentrated to ∼1 mg/ml protein.

Liposomes were prepared as described previously (Schulz et al., 2009) except that EDTA was omitted from the buffers. In brief, lipids dissolved in chloroform were mixed and dried under nitrogen, rehydrated in a buffer, subjected to five freeze–thaw cycles, and extruded through a 0.4-µm-pore-size track-etched Nuclepore membrane (Whatman) using a mini extruder (Avanti Polar Lipids, Inc.). The liposomes contained PC/PE/total yeast lipids (80:10:10 mol%). Egg PC and Egg PE were purchased from Avanti Polar Lipids, Inc., and total yeast lipids were obtained by extraction from WT cells in mid-logarithmic growth phase in synthetic complete medium and total lipid phosphate was determined (Bartlett, 1959). Some heavy (sucrose-filled) liposomes were made so that 1 ml liposomes contained either 3 µCi [³H]cholesterol (61 pmol) or [³H]triolein (50 pmol; American Radiolabeled Chemicals). To prepare heavy liposome with radiolabeled phospholipids, total yeast lipids were obtained from a strain grown for 4 h in a medium containing 5 µCi/ml [³H]palmitate (American Radiolabeled Chemicals).

Lipid transfer assay contained 40 µl of 1 mM heavy liposomes, 40 µl of 1 mM light liposomes, 10 µl of 1% Mega-8, and 10 µl protein. The reactions were incubated at 30°C and stopped by transfer to an ice bath, and 10 µl of 1 mg/ml Proteinase K (Sigma-Aldrich) was added. After 30 min on ice, the samples were centrifuged at 16,000 g for 10 min at 4°C, and the radioactivity in 80 µl of the supernatant was determined by scintillation counting. The amount of radioactivity in the supernatant from a reaction prepared without light liposomes and protein was subtracted from other reactions to calculate the amount of radioactivity transferred to light liposomes.

Fig. S1 depicts results from experiments further characterizing the localization of Ltc1 and the factors that determine it; for example, we show that Ltc1 localization to ER–mitochondria contact sites is ERMES independent and that localization to ER–vacuole contact sites is independent of NVJs, formed by Nvj1. Fig. S2 depicts a z-series through a yeast cell that is expressing Vph1-mCherry and Ltc1(ΔGRAM)-yEGFP, showing their anticorrelated localization pattern with respect to the vacuole. Table S1, included as a separate Excel file, lists strains used in this study.

We thank Dr. Michael Paddy in the University of California, Davis (UC Davis) MCB-CBS Imaging Facility for advice with fluorescence microscopy. We also thank Dr. Brett Finney and Mr. Darren Weber from the UC Davis Proteomics Core Facility for mass spectrometric analysis of protein samples. Finally, we thank members of the Nunnari, Prinz, and Powers laboratories for stimulating discussion.

A. Murley was supported by National Institutes of Health (NIH) training grant 5T32GM007377-34; J. Nunnari is supported by NIH grants R01GM062942, R01GM097432, and R01GM106019.

J. Nunnari is on the Scientific Advisory Board of Mitobridge, Inc. The authors declare no other competing financial interests.