Increased Phospholipid Flux Bypasses Overlapping Essential Requirements for the Yeast Sac1p Phosphoinositide Phosphatase and ER-PM Membrane Contact Sites

In budding yeast cells, much of the inner surface of the plasma membrane (PM) is covered with the endoplasmic reticulum (ER). This association is mediated by seven ER membrane proteins that confer cortical ER-PM association at membrane contact sites (MCSs). Several of these membrane “tether” proteins are known to physically interact with the phosphoinositide phosphatase Sac1p. However, it is unclear how or if these interactions are necessary for their interdependent functions. We find that SAC1 inactivation in cells lacking the homologous synaptojanin-like genes INP52 and INP53 results in a significant increase in cortical ER-PM MCSs. We show in sac1Δ, sac1tsinp52Δ inp53Δ, or Δ-super-tether (Δ-s-tether) cells lacking all seven ER-PM tethering genes that phospholipid biosynthesis is disrupted and phosphoinositide distribution is altered. Furthermore, SAC1 deletion in Δ-s-tether cells results in lethality, indicating a functional overlap between SAC1 and ER-PM tethering genes. Transcriptomic profiling indicates that SAC1 inactivation in either Δ-s-tether or inp52Δ inp53Δ cells induces an ER membrane stress response and elicits phosphoinositide-dependent changes in expression of autophagy genes. In addition, by isolating high-copy suppressors that rescue sac1Δ Δ–s-tether lethality, we find that key phospholipid biosynthesis genes bypass the overlapping function of SAC1 and ER-PM tethers and that overexpression of the phosphatidylserine/phosphatidylinositol-4-phosphate transfer protein Osh6 also provides limited suppression. Combined with lipidomic analysis and determinations of intracellular phospholipid distributions, these results suggest that Sac1p and ER phospholipid flux controls lipid distribution to drive Osh6p-dependent phosphatidylserine/phosphatidylinositol-4-phosphate counter-exchange at ER-PM MCSs.

In budding yeast cells, much of the inner surface of the plasma membrane (PM) is covered with the endoplasmic reticulum (ER).This association is mediated by seven ER membrane proteins that confer cortical ER-PM association at membrane contact sites (MCSs).Several of these membrane "tether" proteins are known to physically interact with the phosphoinositide phosphatase Sac1p.However, it is unclear how or if these interactions are necessary for their interdependent functions.We find that SAC1 inactivation in cells lacking the homologous synaptojanin-like genes INP52 and INP53 results in a significant increase in cortical ER-PM MCSs.We show in sac1Δ, sac1 ts inp52Δ inp53Δ, or Δ-super-tether (Δs-tether) cells lacking all seven ER-PM tethering genes that phospholipid biosynthesis is disrupted and phosphoinositide distribution is altered.Furthermore, SAC1 deletion in Δ-stether cells results in lethality, indicating a functional overlap between SAC1 and ER-PM tethering genes.Transcriptomic profiling indicates that SAC1 inactivation in either Δ-s-tether or inp52Δ inp53Δ cells induces an ER membrane stress response and elicits phosphoinositide-dependent changes in expression of autophagy genes.In addition, by isolating highcopy suppressors that rescue sac1Δ Δ-s-tether lethality, we find that key phospholipid biosynthesis genes bypass the overlapping function of SAC1 and ER-PM tethers and that overexpression of the phosphatidylserine/phosphatidylinositol-4-phosphate transfer protein Osh6 also provides limited suppression.Combined with lipidomic analysis and determinations of intracellular phospholipid distributions, these results suggest that Sac1p and ER phospholipid flux controls lipid distribution to drive Osh6p-dependent phosphatidylserine/phosphatidylinositol-4-phosphate counterexchange at ER-PM MCSs.
The endoplasmic reticulum (ER) represents the major source of lipid biosynthesis within eukaryotic cells, and the plasma membrane (PM) is the primary destination for many of those lipids.Some lipids are transported to and from the PM by vesicular transport, but significant amounts are transferred by nonvesicular mechanisms, which are facilitated by membrane association at and near ER-PM membrane contact sites (MCSs) where the two membranes are affixed (1)(2)(3)(4).ER-PM MCSs serve not only as an interface for the direct exchange of lipids but also as a nexus to coordinate lipid production with the regulation of PM composition and its expansion during cell growth (5,6).
In budding yeast, roughly half of the PM is associated with the ER (2,5,7).Membrane association is conferred by "primary" membrane tether proteins that are both necessary and sufficient for ER-PM contact under standard growth conditions (3,5,6).These tether proteins staple sections of cortical ER (cER) to the PM to generate MCSs.Six primary tethers are conserved proteins, while another represents a yeast-specific tethering factor (7)(8)(9)(10).All these tethering factors are ER-integral membrane proteins that individually, or with other interacting proteins, reach across from the cER to contact the PM (6).The conserved tethers include (i) the yeast homologues of vesicle-associated membrane proteinassociated protein, Scs2p and Scs22p; (ii) the Extended Synaptotagmins (E-Syts), Tcb1p-3p; and (iii) Ist2p, which is a member of the TMEM16/Anoctamin family of ion channels and phospholipid scramblases (8,9,11,12).Ice2p represents a yeast-specific factor that confers ER attachments with both the PM and lipid droplets (12,13).In dividing cells, Ice2p links cER to the PM to facilitate cER movement along the PM from yeast mother cells into daughter cells (12,13).During stationary phase, however, Ice2p confers ER-lipid droplet association (14).Elimination of ICE2 along with the conserved tether protein genes generates so-called Δ-supertether (Δ-s-tether) cells (tcb1Δ tcb2Δ tcb3Δ scs2Δ scs22Δ ist2Δ ice2Δ) and reduces ER-PM association from 48% to 1.7%, which is below calculated levels of stochastic association between the ER and PM (5).Elimination of these tethers imparts a moderate cell growth defect that can be rescued by supplementing cell cultures with choline, a precursor of phospholipid biosynthesis, although choline treatment does not reestablish ER-PM MCSs (5).These results suggest that yeast ER-PM MCSs have an important role in regulating phospholipid metabolism.
In addition to defects in phospholipid biosynthesis, ER-PM MCSs also impact phosphoinositide distribution within cell membranes (5,7,(48)(49)(50).In yeast, PI4P is an essential lipid that is predominantly localized in the Golgi and exocytic vesicles and in the PM at sites of polarization within budding daughter cells (48,51,52).In the absence of ER-PM MCSs, PI4P distribution spreads throughout the PM in both mother cells and daughter buds (5).This PI4P mislocalization throughout the PM is also detected when SAC1 is deleted (48).SAC1 encodes an ER-localized PI4P phosphatase that dephosphorylates PI4P producing PI (53).Because of the similar defects in PI4P membrane distribution, and because most ER-PM tether proteins physically interact with Sac1p, it is proposed that Sac1p acts at ER-PM MCSs to regulate PI4P levels in both membranes (5,7,48,54).Although SAC1 is not essential for yeast growth, deletion of SAC1 in Δ-s-tether cells is lethal, which confirms a functional interaction (5).
Here, we analyze how Sac1p and ER-PM tether proteins cooperate to regulate phospholipid biosynthesis and lipid distribution through the generation of a conditionally lethal sac1 ts Δ-s-tether mutant.In this mutant, the combined requirements for SAC1 and ER-PM MCSs to maintain ER morphology, cellular lipid composition, and lipid distributions were tested.Lipidomic analysis indicated that phospholipid biosynthesis is disrupted in sac1 ts Δ-s-tether cells and the membrane distributions of specific phospholipids and phosphoinositides are disrupted.Effectors of the overlapping functions of SAC1 and ER-PM MCS were also identified through the isolation of high-copy suppressors of sac1 ts Δ-stether lethality.Several CDP-DAG pathway genes that promote the flux of phospholipid synthesis producing PS are bypass suppressors of sac1Δ Δ-s-tether lethality.These suppressors do not restore normal cER-PM contact, and ER-PM tethering was found to be specifically dependent on established ER-PM tethers.The transcriptomic profile of sac1 ts Δ-s-tether cells indicates that the combined effect of these mutations results in membrane stress and phosphoinositidedependent autophagy dysregulation.This study reveals that ER-PM MCSs maintain cellular membrane lipid distribution by regulating phospholipid flux, contributing to PS and PI4P exchange between the ER and PM thereby generally affecting phosphoinositide distribution and homeostasis.

Eliminating PI phosphate phosphatases increases cER-PM contact
In yeast, the ER-localized PI phosphate phosphatase Sac1p turns over PI4P transferred from the PM and the Golgi (54).In most Saccharomyces cerevisiae strains, SAC1 is not an essential gene unless its deletion is combined with mutations in other homologous PI phosphate phosphatases (i.e., INP52 and INP53) (55).Sac1p also physically interacts with most primary ER-PM tether proteins, which are all ER integral membrane proteins (7).Although the elimination of these ER-PM tether proteins is not lethal, Δ-s-tether cells lacking tethers are inviable if SAC1 is also deleted (5).Based on these genetic results we hypothesized that Sac1p and ER-PM tethers function in parallel but independent pathways, despite that Sac1p physically interacts with many ER-PM tethers.Sac1p can act in trans by dephosphorylating PI4P on closely apposed membranes in vitro, but the genetic interaction between SAC1 and the ER-PM tether genes is also consistent with a model in which Sac1p acts in cis on PI4P transferred to the ER from the PM via ER-PM MCSs (48,54).It is also unknown if these functional interactions between Sac1p and ER-PM tethers has any regulatory impact on the assembly of cER-PM contact sites.
To determine if Sac1p affects cER-PM association, confocal microscopy was used to analyze the juxtaposition of the ER marker DsRed-HDEL relative to the MemBrite-stained cell cortex in sac1Δ cells.As compared with congenic wildtype (WT) cells where 51% of the inner surface of the PM is covered with cER, cER-PM association increases to 59% in sac1Δ cells (Fig. 1, B and C).Because the essential activity of Sac1p overlaps with the other yeast PI phosphate phosphatases, namely, Inp52p and Inp53p, we investigated the combined effect of PI phosphate phosphatase inactivation using the temperature conditional sac1-23 ts allele in cells lacking INP52 and INP53.The mutations that define the sac1-23 ts allele lie within and closely adjacent to the catalytic Sac phosphatase domain of Sac1p, and at elevated temperatures sac1-23 ts mutant cells cannot turn over PI4P in the PM (55).At 37 C for 1 h, the area of the PM covered with cER increases to 79% in temperature-sensitive sac1-23 ts inp52Δ inp53Δ cells, relative to 60% in its congenic WT control strain (Fig. 1, B and C).At 30 C, a temperature at which sac1-23 ts is at least partially functional, 65% of the PM is covered with cER in sac1 ts inp52Δ inp53Δ cells, which still represents a statistically significant increase compared with the congenic WT control.Regardless of temperature, the inp52Δ inp53Δ mutations by themselves have no effect on cER-PM association as compared with WT (Fig. 1C).Although SAC1 deletion leads to moderate increases in cER-PM association, eliminating overlapping activities of the three homologous PI4P phosphatases further spreads cER along the PM, likely due to increased MCSs.
The dsRed-HDEL marker shows general ER morphology, including specific and nonspecific associations of cER with the PM, whereas Tcb3p-GFP is a marker that specifically identifies direct tether attachment sites between the cER and the PM (4).Thus, as a direct measure of membrane contact, we examined the Tcb3p-GFP ER-PM MCSs that are known to proliferate in response to membrane stress and specific lipid transfer mutants (50).In WT cells, Tcb3p-GFP fluorescence was detected on average covering 40% of the PM and the deletion of SAC1 results in a modest increase of Tcb3p MCSs to 46% (Fig. 1, D  and E).The slight increase in Tcb3p ER-PM MCSs in sac1Δ cells, however, is consistent with the observed moderate increase in cER-PM association.
To determine if eliminating multiple PI phosphate phosphatases affects the spread of Tcb3p ER-PM MCSs, the cortical localization of the Tcb3p-GFP ER-PM tether was analyzed in sac1 ts inp52Δ inp53Δ cells at 30 and 37 C (Fig. 1, D and E).In sac1 ts inp52Δ inp53Δ cells cultured at the permissive growth temperature of 30 C, cortical coverage of Tcb3p-GFP increases to a mean of 66%, up from 55% in the congenic WT strain (Fig. 1, D and E).After 1 h at 37 C, however, Tcb3p-GFP fluorescence in sac1 ts inp52Δ inp53Δ cells climbs to 70% of the cortical surface of the PM, as compared with 58% in the congenic WT control under the same conditions (Fig. 1E).Regardless of temperature, in inp52Δ inp53Δ cells no significant change in cortical Tcb3p-GFP is detected, relative to WT (Fig. 1E).These results indicate that all three PI phosphate phosphatases contribute to Tcb3p regulation of cER-PM association, where SAC1 by itself has a modest but otherwise redundant role.After multiple attempts to delete TCB3 in sac1 ts inp52Δ inp53Δ cells, we were unable to obtain viable sac1 ts inp52Δ inp53Δ tcb3Δ transformants and could not directly test if Tcb3p alone is essential for the observed increases in cER-PM association.
The PI-4-kinases Stt4p and Pik1p generate PI4P from PI within the PM and Golgi, respectively, and these PI-4-kinases represent opposing activities to Sac1p and the other PI4P phosphatases (55,56).If ER-PM MCS formation is directly affected by cellular PI4P levels, then stt4 ts and/or pik1 ts mutants are predicted to have the opposite effect on ER-PM association as observed in sac1Δ cells.However, in both stt4 ts and pik1 t cells, cortical dsRed-HDEL increases to 66% along the PM relative to 57% in WT cells, after 1 h at 37 C (Fig. 2, A and B).Concomitant with the increase in cER association with the PM, Tcb3p ER-PM contact sites also proliferate along the PM.In both stt4 ts and pik1 t cells, cortical Tcb3p-GFP fluorescence increases to 58 and 62%, respectively, as compared with 49% observed in the congenic WT control at 37 C (Fig. 2, C and D).Because any change in PI4P metabolism increases ER-PM association, ER-PM MCS assembly is generally induced by phosphoinositide dysregulation.Although it is unclear how PI4P homeostasis affects ER-PM MCSs, it is clear that disruption of SAC1 or other PI4P regulators increases ER-PM MCSs as shown by increased Tcb3p expression at the cell cortex.

The lethal combination of SAC1 mutations in cells lacking ER-PM MCSs disrupts phospholipid and sphingolipid metabolism
Lipidomic assays revealed significant defects in the regulation of phospholipid biosynthesis in Δ-s-tether cells (5).We hypothesized that the lethality of SAC1 deletion in Δ-s-tether might be caused by further exacerbation of this lipid dysregulation.To perform lipidomic analysis on cells lacking both ER-PM tethers and SAC1, a temperature-conditional sac1 ts Δs-tether strain was generated by transforming a plasmid containing the sac1-23 ts mutation into sac1Δ Δ-s-tether cells, which are otherwise inviable.After incubation at 37 C for 1 h, lipidomic analysis revealed significant differences in lipid composition between sac1 ts Δ-s-tether, Δ-s-tether, and sac1 ts inp52Δ inp53Δ strains, as normalized to WT (Fig. 3).In agreement with previous reports, in Δ-s-tether cells many phospholipid levels are significantly reduced including PS and CDP-DAG (Fig. 3A) (5).These reductions are likely due to defects in the biosynthetic utilization of the phospholipid precursor DAG, as indicated by its increased levels and the significant accumulation of triacylglycerol (TAG).When compared with Δ-s-tether cells, reductions in normalized phospholipid levels are even more pronounced in sac1 ts Δ-stether cells incubated at 37 C for 1 h, which exhibit further decreases in PS, PE, and CDP-DAG and greater levels of DAG and TAG (Fig. 3A).Although most phospholipids are unaffected in sac1Δ and sac1 ts inp52Δ inp53Δ cells, PS levels are much reduced, albeit to a lesser degree than in sac1 ts Δ-stether cells.In fact, the combined phospholipid defects of sac1 ts and Δ-s-tether mutations is most evident in the reduction of CDP-DAG and PS levels.
Levels of some single-carbon chain lysophospholipids accumulate in strains lacking ER-PM tethers or the PI phosphatases (Fig. 3B).In Δ-s-tether cells, lysoPC (LPC) levels are comparable with WT levels, whereas lysoPE (LPE) levels are moderately elevated.In sac1Δ cells, LPC levels are equivalent to WT and substantially decreased in LPE.In contrast, the normalized levels of lysophospholipids in sac1 ts inp52Δ inp53Δ cells are 1.2to 1.6-fold higher after incubation at stained with a blue cell surface dye (MemBrite Fix 405/430).Cells were incubated either at 30 C or at 37 C for 1 h, as indicated.C, ratios of cER coverage per total distance of each cell perimeter/PM corresponding to (B) (N ≥ 20 cells/strain).D, representative images of Tcb3p-GFP (pWK092) expressed in sac1Δ (CBY2809) cells and WT (BY4741) control, and sac1 ts inp52Δ inp53Δ (AAY143) cells and WT (SEY6210) control cells, incubated either at 30 C or at 37 C for Coordinate phospholipid regulation by SAC1 and ER-PM MCSs 37 C for 1 h.At 37 C for 1 h, the inactivation of sac1 ts in Δ-stether cells increases the normalized level of LPC and results in a moderate increase in LPE levels.However, the inactivation of PI phosphatase activity in sac1 ts inp52Δ inp53Δ cells has the greatest effect on lysophospholipids and causes considerable LPC and LPE increases.
Because ER-PM MCSs impact sphingolipid biosynthesis (5), levels of ceramide and the complex sphingolipid IPC were analyzed in PI phosphatase and Δ-tether mutants (sphingolipid IPC derivatives, namely, mannose-inositol phosphorylceramide and mannose (inositol-P)2-phosphorylceramide, could not be definitively analyzed).As previously reported, the normalized lipidomic profile of Δ-s-tether cells shows increased ceramide levels and reduced amounts of IPC (Fig. 3C) (5).After 1-h incubation at 37 C, sac1 ts inactivation in Δ-s-tether cells results in comparable changes (Fig. 3C).Inactivation of the PI phosphatases in sac1 ts inp52Δ inp53Δ cells also results in similar ceramide accumulation and lower levels of IPC.Under the conditions used in this study, the deletion of SAC1 by itself reduces the normalized level of IPC.However, unlike previous reports (57), a concomitant accumulation of ceramide is not observed (although SAC1 inactivation in tandem with inp52Δ inp53Δ clearly results in ceramide accumulation).Nevertheless, the disruption of sphingolipid biosynthesis in PI phosphatase and ER-PM tethers mutants suggests that they both participate in regulating ceramide incorporation into complex sphingolipids.
To compare rates of synthesis of phospholipids, logarithmic phased cells were cultured in synthetic medium and lipid synthesis was assayed following 32 P pulse labeling.After preincubation at 30 or 37 C for 1 h, [ 32 P]H 3 PO 4 was added to cultures for 2, 5, 10, and 20 min.Following lipid extraction and separation by thin-layer chromatography, the synthesis of PA, PS, PE, and PI/PC was measured in WT, sac1Δ, Δ-s-tether, sac1 ts Δ-s-tether, and sac1 ts inp52Δ inp53Δ cells (Fig. 3D).The deletion of SAC1 has comparatively little impact on PA and PI/PC synthesis (though PS and PE synthesis is reduced), whereas phospholipid synthesis is markedly reduced in Δ-s-tether cells.At 37 C for 1 h, which greatly increases phospholipid synthesis in WT cells, in sac1 ts inp52Δ inp53Δ cells phospholipid flux is severely inhibited (Fig. 3D).Thus, phospholipid synthesis is dependent on essential combined activities of SAC1, INP52, and INP53.In both Δ-s-tether cells at 30 C and sac1 ts Δ-stether cells at 37 C, the incorporation of 32 P into PA and all other phospholipids assayed is almost blocked when compared with WT (Fig. 3D).Indeed, in sac1 ts Δ-s-tether cells the synthesis of PS and PE is barely detectable even after 20 min.When compared with their WT controls at the same temperature, phospholipid synthetic flux in sac1 ts Δ-stether cells is even further reduced than that in Δ-s-tether cells.These results directly show that phospholipid flux is dependent on the combined functional interaction of ER-PM MCS tethers and Sac1p.

Genes involved in phospholipid biosynthesis suppress sac1 ts Δ-s-tether synthetic lethality
To identify genes that participate in the overlapping functions of SAC1 and the ER-PM tether genes, a dosage suppressor selection was conducted to isolate extragenic suppressors of sac1 ts Δ-s-tether lethality.Following transformation with a high-copy (2 μ) plasmid library derived from Δ-s-tether cells (to avoid reisolation of the seven ER-PM tether genes), surviving sac1 ts Δ-s-tether transformants were selected after growth at 37 C. Of the eight extragenic suppressors isolated, six corresponded to genomic fragments that included SLC1, which encodes an acyltransferase that converts lysophosphatidic acid to PA (Fig. 1A) (18,19).Because SLC1 can rescue growth in either the sac1 ts Δ-s-tether conditional mutant or sac1Δ Δ-s-tether cells, SLC1 represents a bypass suppressor that circumvents the combined essential functions of SAC1 and ER-PM tethers (Fig. 4A).As with all bypass suppressors, the mechanism of SLC1 suppression of sac1Δ Figure 3. Lipidomic profiles of sac1Δ, sac1 ts inp52Δ inp53Δ, Δ-s-tether, and sac1 ts Δ-s-tether cells exhibit defects in phospholipid and sphingolipid biosynthesis.A, phospholipid composition of sac1Δ (CBY2809) and Δ-s-tether (CBY5838) at 30 C, and sac1 ts inp52Δ inp53Δ (AAY143) and sac1 ts Δ-s-tether (CBY6345) cells at 37 C, as a normalized mole percentage relative to WT (SEY6210) (set as 100%) cultured under the same conditions.B, lysophospholipid composition of the mutants in (A), as a normalized mole percentage relative to WT. C, ceramide and IPC composition of the mutants in (A), as a normalized mole percentage relative to WT.The lipidomics data represent the mean ± SEM derived from five independent samples as shown.D, pulse labeling analysis of phospholipid synthetic flux in WT, sac1Δ, and Δ-s-tether at 30 C, as well as in WT, sac1 ts inp52Δ inp53Δ, and sac1 ts Δ-s-tether cells at 37 C. Glycerophospholipids were extracted 2, 5, 10, and 20 min after addition of 32 P to log-phase cells, and lipids were separated via thin layer chromatography and quantified.In WT, label was first incorporated into PA followed by labeling of PE, PS, and PI/PC observed in varying amounts; mutant cells exhibited reduced levels of label incorporation into phospholipids.PI and PC could not be adequately resolved and are presented as a collective measurement.Each time point for each strain represents the average of duplicate independent analyses.CDP, cytidine diphosphate; DAG, diacylglycerol; IPC, inositol phosphorylceramide; LPC, lysophosphatidylcholine; LPE, lysophosphatidylethanolamine; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PIP, phosphatidylinositol phosphate; PS, phosphatidylserine; TAG, triacylglycerol.

Coordinate phospholipid regulation by SAC1 and ER-PM MCSs
Δ-s-tether lethality cannot involve reestablishing physical connections between Sac1p and the tether proteins; interactions between proteins cannot be restored if their respective genes are absent.The selection also twice identified CUE1 as a weak allelic suppressor of sac1 ts Δ-s-tether cell lethality (Fig. 4B).Cue1p encodes an ER-localized ubiquitinbinding protein that recruits the ubiquitin-conjugating enzyme Ubc7p to the ER membrane for ER-associated degradation of misfolded proteins during ER stress (58,59).(UBC7 overexpression was insufficient to suppress sac1 ts Δ-s-tether growth defects.)Given that SLC1 was a strong suppressor with a clear link to the phospholipid defects in PI phosphate phosphatases and Δ-s-tether mutants, we focused exclusively on the mode of SLC1 suppression.Moreover, the low number of CUE1 isolates (N = 2) suggests that the genetic selection was not saturated and other potential suppressors might still be identified.
If SLC1 overexpression suppresses sac1Δ Δ-s-tether lethality by boosting phospholipid levels through increased PA biosynthesis, we hypothesized that other PA-synthesizing genes might also bypass sac1Δ Δ-s-tether lethality (Fig. 4C).Like SLC1, overexpression of either ALE1 or DGK1 on highcopy plasmids rescued sac1Δ Δ-s-tether cell growth (Fig. 4C).ALE1 and DGK1 encode lysophospholipid acyltransferase and diacylglycerol kinase, respectively, both of which lead to PA biosynthesis.The enzymatic activity of the phosphatidate phosphatase Pah1p acts in opposition to Dgk1p by dephosphorylating PA to yield DAG (Fig. 1A) (60).We tested if PAH1 deletion would phenocopy DGK1 overexpression as a bypass suppressor of sac1Δ Δ-s-tether lethality.Unfortunately, after multiple attempts, viable pah1Δ sac1Δ Δ-s-tether transformants could not be isolated.Nonetheless, these results suggest that ER-PM MCSs and SAC1 together affects the availability of PA as a precursor for the synthesis of CDP-DAG, which in turn serves as a substrate for PI and PS generation (Fig. 1A).
For sac1Δ Δ-s-tether suppression, we tested if PI or PS is the relevant phospholipid by overexpressing PIS1 and CHO1 (Fig. 4, C and D).PIS1 encodes PI synthase, which produces the PI precursor for all phosphoinositide synthesis, whereas PS is produced by CHO1, which encodes PS synthase (61,62).Although CHO1 is a strong bypass suppressor of sac1Δ Δ-stether lethality, PIS1 overexpression failed to suppress the growth defect (Fig. 4, C and D).To further delineate the mode of suppression, we tested if high-copy DGK1 or CHO1 rescued PI phosphate phosphatase dysfunction in sac1 ts inp52Δ inp53Δ cells.At 37 C, neither DGK1 nor CHO1 suppressed sac1 ts inp52Δ inp53Δ lethality, suggesting that these high-copy suppressors do not affect PI phosphate phosphatase activity per se (Fig. 4E).However, in the absence of choline the growth defect of Δ-s-tether cells is partially rescued by DGK1 and  SEY6210) and sac1 ts Δ-s-tether (CBY6345) cells transformed with high-copy plasmids containing SCS2 (pSCS2), a genomic fragment containing SLC1 (pCB1350), and the vector control (YEplac195) cultured on selective solid medium for 3 to 5 days at the indicated temperatures.B, tenfold serial dilutions of WT and sac1 ts Δ-s-tether cells transformed with a high-copy plasmid containing a genomic CUE1 fragment and the vector alone control cultured on selective solid medium for 3 to 5 days at the indicated temperatures.C, tenfold serial dilutions of WT and sac1Δ Δ-s-tether cells with high-copy plasmids containing DGK1 (pCB1346), CHO1 (pCB1352), SLC1 (pCB1350), ALE1 (pCB1382), or the corresponding vector control (YEplac181), grown and spotted onto solid synthetic medium at 30 C for 3 days.Strains shown successfully grew after high-copy plasmid suppressors were transformed into WT and sac1Δ Δ-stether cells containing an SCS2 plasmid (pSCS2), followed by SCS2 counter-selection with the addition of 5 0 -fluoroorotic acid.D, tenfold serial dilutions of WT and sac1Δ Δ-s-tether cells containing episomal SCS2 transformed with high-copy PIS1 (pCB1345) in which the SCS2 plasmid was either selected for (+SCS2) or counter-selected against (−SCS2), grown at 30 C for 5 days.E, tenfold serial dilutions of WT (SEY6210) and sac1 ts inp52Δ inp53Δ (AAY143) cells transformed with high-copy DGK1, CHO1, and the vector alone control.F, tenfold serial dilutions of WT and Δ-s-tether (CBY5838) cells containing high-copy ALE1, SLC1, DGK1, CHO1, and the vector alone control cultured at 30 C for 4 days.
CHO1 (Fig. 4F).On the other hand, ALE1 and SLC1 overexpression has little effect on the growth of Δ-s-tether cells (Fig. 4F).Collectively these results suggest that the function of ER-PM contact sites, as relates to Sac1p activity, affects phospholipid flux from PA through the PS branch of the CDP-DAG biosynthesis pathway.The increased expression of these specific phospholipid biosynthetic genes rescues the overlapping lipid defects exacerbated by SAC1 deletion in Δ-stether cells.

Lipid biosynthetic genes rescue PM phospholipid defects in sac1 ts Δ-s-tether cells
Given the altered lipid levels in sac1 ts Δ-s-tether cells (Fig. 3), we hypothesized that high-copy suppressors of sac1 ts Δ-s-tether lethality correct lipid metabolism defects in these cells.In fact, DGK1 and CHO1 dosage suppressors restored levels of most affected phospholipids near to WT levels (Fig. 5A).When compared with sac1 ts Δ-s-tether cells grown at 37 C for 1 h, the normalized lipidomic profiles of sac1Δ Δ-stether cells rescued by high-copy CHO1 or DGK1 show increases in PS and PE levels, and CHO1 and DGK1 suppression confers reductions in DAG and TAG accumulation, respectively.Consistent with previous reports, high-copy CHO1 generally increases both PE and PS levels, and DGK1 overexpression conferred similar effects (27).Otherwise, CHO1 and DGK1 suppression in sac1Δ Δ-s-tether cells has variable effects on other phospholipids, where CHO1 rescued the low level of CDP-DAG but DGK1 did not.The results suggest that Figure 5. Lipidomics analysis of high-copy suppressors of sac1Δ Δ-s-tether lethality.A, phospholipid composition of sac1 ts Δ-s-tether (CBY6345) cells incubated at 37 C for 1 h, and sac1Δ Δ-s-tether cells containing high-copy DGK1 (CBY6508) and high-copy CHO1 (CBY6522) cultured at 30 C, expressed as a normalized mole percentage relative to WT (set as 100%) cultured under the same conditions.B, lysophospholipid composition of the cells in (A), shown as a mole percentage relative to WT. C, ceramide and IPC composition of the cells in (A), as a normalized mole percentage relative to WT.The lipidomics data represent the mean ± SEM derived from five independent samples.D, pulse labeling analysis of phospholipid synthetic flux in DGK1and CHO1suppressed sac1Δ Δ-s-tether cells at 30 C, as compared with WT.As per Figure 3D, glycerophospholipids were extracted 2, 5, 10, and 20 min after addition of 32 P to log-phase cells, and lipids were separated via thin-layer chromatography and quantified.Each time point for each strain represents the average of duplicate independent analyses.CDP, cytidine diphosphate; DAG, diacylglycerol; IPC, inositol phosphoryl-ceramide; LPC, lysophosphatidylcholine; LPE, lysophosphatidylethanolamine; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PIP, phosphatidylinositol phosphate; PS, phosphatidylserine; TAG, triacylglycerol.
suppression of sac1Δ Δ-s-tether lethality correlates with increases in PE and PS levels and decreases in DAG or TAG.
The impact on lysophospholipids by DGK1 and CHO1 suppression of sac1Δ Δ-s-tether lethality is minor.CHO1 overexpression in sac1Δ Δ-s-tether cells does not affect LPC, although LPE levels increased compared with sac1 ts Δ-s-tether cells at 37 C for 1 h (Fig. 5B).DGK1 overexpression elicits relatively modest if any changes.Given the lack of substantial changes in LPC and LPE levels between the suppressed sac1Δ Δ-s-tether strains and sac1 ts Δ-s-tether cells at 37 C, the mechanism of CHO1 and DGK1 suppression is unlikely to involve lysophospholipid regulation.
As a component of the "SPOTS" (serine palmitoyltransferase, Orm1, Orm2, Tsc3, and Sac1) regulatory complex, sac1Δ affects the biosynthesis of sphingolipids from ceramide, as does the deletion of ER-PM tethers (Fig. 3C) (5,63,64).Tricalbins transfer ceramide from the ER to the Golgi for IPC synthesis (64), and the deletion of all tricalbin genes in Δ-stether cells likely impedes the normal sphingolipid synthetic process.As compared with sac1 ts Δ-s-tether cells at 37 C for 1 h, DGK1 or CHO1 overexpression in sac1Δ Δ-s-tether cells shows reduced ceramide accumulations, which correlates with the minor increases in IPC levels observed (Fig. 5C).These results suggest that DGK1 or CHO1 overexpression partially restores sphingolipid biosynthesis in sac1Δ Δ-s-tether cells.Taken together, however, these results suggest that DGK1 and CHO1 overexpression rescues sac1Δ Δ-s-tether lethality by restoring normal phospholipid, DAG, and TAG levels, as opposed to suppressing sphingolipid synthesis defects.
Given that steady-state levels of phospholipids in DGK1-or CHO1-suppressed sac1Δ Δ-s-tether cells are generally returned to WT levels (Fig. 5A), a restoration of phospholipid synthetic flux might also be predicted.To test if phospholipid flux is rescued in suppressed sac1Δ Δ-s-tether cells, extracted lipids from logarithmic phased cells were assayed following 32 P pulse labeling (Fig. 5D).As a proportion of the synthetic rate in WT cells under the same growth conditions, the synthesis of PA, PS, PE, and PI/PC increased in DGK1-or CHO1-suppressed sac1Δ Δ-s-tether cells compared with the lack of phospholipid synthesis in sac1 ts Δ-s-tether at 37 C (Figs. 3D  and 5D).Although phospholipid flux was not completely restored to WT levels in the suppressed strains, the moderate increases in synthetic rates account for the observed increases in steady-state phospholipid levels.Bypass suppressors of sac1Δ Δ-s-tether cells restore limited phospholipid flux irrespective of any physical interaction between Sac1p and ER-PM tethers.

CHO1 and DGK1 suppression of sac1Δ Δ-s-tether lethality through phospholipid synthesis without restoring normal cER-PM association
To determine how the lipid biosynthetic suppressors affect ER-PM association, DsRed-HDEL was expressed in WT and sac1Δ Δ-s-tether cells transformed with high-copy CHO1 or DGK1 plasmids.Consistent with previous reports, DGK1 overexpression in WT cells leads to increased DsRed-HDEL fluorescence in cytoplasmic ER, although without significant changes in cER at the PM (Fig. 6) (50,64,65).DGK1 overexpression in sac1Δ Δ-s-tether cells causes a slight but significant increase in cER-PM association (Fig. 6, A and B).In contrast, high-copy CHO1 in both WT and sac1Δ Δ-s-tether cells significantly increases ER-PM association relative to the vector alone controls (Fig. 6).
As formal possibilities, sac1Δ Δ-s-tether lethality might be suppressed by (i) reestablishing novel cER tethering with the PM; (ii) restoring lipid homeostasis between the ER and PM; (iii) indirectly promoting the cytoplasmic ER proliferation out to the cortex, thereby stochastically increasing ER-PM interaction.With respect to the latter possibility, CHO1 overexpression results in the formation of unusually thick stacks of cytoplasmic ER reaching out toward the cell cortex that are reminiscent of karmellae, which represent an elaboration and proliferation of cytoplasmic ER (66).Karmellae are induced upon overexpression of HMG1, which encodes the isoprenoid synthetic enzyme 3-hydoxy-3-methylglutaryl coenzyme A reductase (66).In both WT and Δ-s-tether cells, HMG1 overexpression results in significant ER expansion as observed using the fluorescent DsRed-HDEL ER-marker (Fig. 6, C and  D).In this regard, the ER proliferation caused by CHO1 overexpression resembles HMG1-induced karmellae.However, induction of cytoplasmic ER by high-copy HMG1 does not rescue Δ-s-tether cells growth defects, indicating that ER proliferation cannot suppress defects in ER-PM MCSs (Fig. 6E).Moreover, this HMG1-induced cytoplasmic ER expansion does not suppress sac1 ts Δ-s-tether cell lethality when HMG1 is overexpressed in sac1 ts Δ-s-tether cells (Fig. 6F).The fact that cytoplasmic ER expansion does not suppress growth defects in cells lacking ER-PM MCSs indicates that random ER association at the PM cannot substitute for directed membrane tethering.
Both Cho1p and Dgk1p have unstructured N-terminal extensions that might adopt the function of a membrane tether in the absence of ER-PM MCSs (Fig. S1A).To determine whether Cho1p has tethering capability or if its enzymatic activity is required, we tested the suppression sac1 ts Δ-s-tether growth defects by the conserved "catalytically dead" CHO1 D127A mutation (67).In the absence of choline in the growth medium, CHO1 D127A is nonfunctional and cannot complement the growth defect of cho1Δ cells (Fig. S1B).High-copy expression of CHO1 D127A cannot suppress sac1 ts Δ-s-tether lethality at 37 C, indicating that Cho1p enzymatic activity is required to confer suppression.Unlike Cho1p or yeast Dgk1p, Escherichia coli diacylglycerol kinase (DgkA) lacks any conceivable tethering region but can enzymatically substitute for its yeast counterpart (Fig. S1A) (68,69).Although the expression of DgkA in sac1 ts Δ-s-tether does not restore growth at 37 C, bacterial DgkA does suppress sac1 ts Δ-s-tether grow defects at the semipermissive growth condition of 34 C (Fig. S1D).This result suggests that yeast Dgk1p does not physically attach membranes, but rather diacylglycerol enzymatic activity is necessary for suppression.Moreover, nonspecific restoration of ER-PM contact is inadequate to suppress sac1Δ Δ-s-tether lethality.ER-PM contact can be reestablished by an "artificial ER-PM Coordinate phospholipid regulation by SAC1 and ER-PM MCSs staple," which can ameliorate some growth defects of Δ-s-tether cells (5).When expressed in sac1Δ Δ-s-tether cells, however, the artificial staple is unable to rescue cell lethality (Fig. S1E).This result suggests that the functional interaction between Sac1p and native ER-PM tethers does not involve nonspecific cER physical attachment to the PM.Suppressors of sac1Δ Δ-s-tether lethality increase both cytoplasmic ER expansion and cER, but ER spread en mass is insufficient for suppression; suppression requires more than nonspecific reestablishment of ER and PM association.Instead, restoration of phospholipid metabolism bypasses the elimination of SAC1 and ER-PM tethers.The distribution of lipids within cellular membranes impacts their levels and metabolism (70).Given the role of MCSs in regulating lipid exchange between cER and the PM, we investigated if the defects in lipid composition revealed by lipidomic analyses are coupled with phospholipid mislocalization.Using specific fluorescent probes, the membrane localizations of DAG, PA, PS, PI4P, and PI(4,5)P 2 were assessed by confocal microscopy in WT and sac1Δ cells as well as in sac1Δ Δ-s-tether cells containing suppressor plasmids, as compared with sac1 ts inp52Δ inp53Δ and sac1 ts Δ-s-tether cells after 1-h incubation at 37 C (Figs. 7 and 8).
Given the significant increases in DAG levels in Δ-s-tether and sac1 ts Δ-s-tether cells, we predicted that DAG distribution might be altered when observed with the DAG-specific lipid probe C1δ-GFP (70).In WT cells, C1δ-GFP fluorescence is primarily observed in the vacuolar membrane and polarized in the PM where it is restricted to budding daughter cells (Fig. 7) (71).However, in sac1Δ at 30 C, and sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ at 37 C for 1 h, C1δ-GFP is distributed all around the cell cortex indicating nonpolarized DAG localization in the mother and bud PM (Figs. 7 and S2A).In sac1Δ cells the PM defect in DAG polarization is most severe, which was a surprise given that the defect in sac1 ts inp52Δ inp53Δ cells (at 37 C for 1 h or even at 30 C, which were equivalent) was substantially less (Figs.7 and S2A).Nonetheless, the changes in DAG levels and PM polarization in these mutant cells suggested further defects in the distribution of the phospholipids requiring DAG as their precursor.
Because PA biosynthetic genes (SLC1, ALE1, DGK1) suppress sac1 ts Δ-s-tether cell growth defects at 37 C, we predicted changes in PA distribution in these and the other mutant cells.Consistent with previous reports, WT cells expressing the GFP-Spo20 51-91 PA-specific lipid probe show fluorescence primarily along the PM with faint nuclear localization at 30 C or 37 C (Fig. 7) (69,72).In sac1 ts Δ-s-tether or sac1 ts inp52Δ inp53Δ cells at 37 C for 1 h, GFP-Spo20 51-91 localization was indistinguishable from WT.Moreover, we observed no differences in GFP-Spo20 51-91 distribution in sac1Δ or Δ-s-tether cells at 30 C, as compared with congenic WT controls (Fig. 7).Despite the PM defects in DAG distribution in sac1 ts Δ-s-tether and other mutants, PA localization and steady-state levels appear unaffected.
To determine if the observed changes in DAG impacted phospholipid distribution other than PA, we analyzed the localization of phospholipids further along the CDP-DAG synthetic pathway (Fig. 1A).To observe PS distribution, the Lact-C2-GFP probe was visualized by confocal fluorescence microscopy in the various mutant cells (Fig. 7).In WT cells, PS is predominantly found in the PM with a polarized distribution somewhat concentrated at the cortex in small/medium buds (73).In many of the tether and PI phosphate phosphatase mutants, however, decreases in PS at the PM were detected.In Δ-s-tether cells, PS Lact-C2-GFP mean fluorescence is reduced at the PM to 47% of WT (N = 20 cells), and a faint fluorescence in internal membranes is detectable.Moreover, in sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells incubated at 37 C for 1 h, the PS Lact-C2-GFP mean fluorescence at the PM is 42% and 40% compared with WT, respectively (N = 20 cells).The fluorescence was less affected in sac1Δ cells, in which PS Lact-C2-GFP was 70% of its congenic WT (N = 20 cells), although an elevation in internal membrane fluorescence is evident.Consistent with reductions in PS levels as shown by lipidomics (Fig. 3A), PS in Δ-s-tether, sac1 ts Δ-stether, and sac1 ts inp52Δ inp53Δ cells is significantly reduced in the PM.
Because ER/PM localization of PS and PI4P are interdependent due to a counter-exchange mechanism (74,75), we analyzed sac1Δ, Δ-s-tether, sac1 ts Δ-s-tether, and sac1 ts inp52Δ inp53Δ cells to determine if the changes in PS levels are coupled with altered PI4P distribution.When expressed in WT cells, the P4M-SidM-GFP lipid probe shows PI4P localization in the Golgi and secretory vesicles and polarized in the PM sites only at the bud tip (76, 77) (Fig. 7).As previously reported, PI4P in sac1Δ and Δ-s-tether cells is not polarized within the PM, where P4M-SidM-GFP fluorescence spreads throughout the PM in both mother and daughter cells (5, 7).The same defect is observed in sac1 ts Δ-s-tether cells at 37 C for 1 h, where PI4P is redistributed uniformly throughout the PM and fewer Golgi puncta are observed (Figs. 7 and S2B).In sac1 ts inp52Δ inp53Δ cells cultured at 30 or 37 C for 1 h (N = 65), P4M-SidM-GFP fluorescence is uniformly concentrated along the PM in irregular puncta, which is suggestive of previous reports of deep phosphoinositide-enriched PM invaginations observed in PI phosphate phosphatase mutants (Fig. 7) (78,79).The PM redistribution of PI4P in these mutant cells is consistent with the interdependence of PI4P and PS transfer between the ER and PM.

Suppressors of sac1Δ Δ-s-tether lethality partially restore intracellular PI4P and PI(4,5)P 2 distribution and cause PS accumulation in intracellular membranes
If sac1Δ Δ-s-tether lethality is due to defects in lipid metabolism, then rescue by high-copy bypass suppressors is predicted to restore at least some aspect of phospholipid distribution and synthesis.When expressed in WT cells, highcopy DGK1 or CHO1 does not appreciably affect the localization of lipid probes detecting PA, PI4P, or PS distribution Coordinate phospholipid regulation by SAC1 and ER-PM MCSs (Fig. 8).Normal PA distribution is also unaffected in sac1Δ Δ-s-tether cells rescued by either the DGK1 or CHO1 suppressors.PI4P and PS distributions are, however, markedly different in the suppressed sac1Δ Δ-s-tether cells (Fig. 8).In WT or sac1 ts Δ-s-tether cells incubated at 37 C for 1 h, PS localization is primarily detected at the cell cortex as shown by the Lact-C2-GFP probe, although, as previously described, PM Lact-C2-GFP fluorescence is considerably less when expressed Coordinate phospholipid regulation by SAC1 and ER-PM MCSs in sac1 ts Δ-s-tether cells (Figs. 7 and 8).In all sac1Δ Δ-s-tether cells, whether suppressed by high-copy DGK1 or CHO1, Lact-C2-GFP cortical fluorescence is nearly restored to WT levels (86% and 91% in DGK1-and CHO1-suppressed cells, respectively, relative to WT) and fluorescence is also evident in internal membranes.The observed increase in PS in intracellular membranes is consistent with PS accumulation in cytoplasmic ER that overproliferates in these cells (Figs. 5A and 8).Because of the dependency of PI4P localization on PS, we predicted that high-copy DGK1 or CHO1 suppression of sac1Δ Δ-stether growth defects would affect PI4P distribution between the ER and PM.As shown above (Figs.7 and 8), P4M-SidM-GFP fluorescence in WT cells indicates PI4P at bud tips and the Golgi, whereas the PI4P distribution in sac1 ts Δ-s-tether cells is uniformly spread along the PM.Although this nonpolarized PI4P distribution is still observed in sac1Δ Δ-stether cells overexpressing DGK1 or CHO1 (albeit with less intensity, particularly in sac1Δ Δ-s-tether cells overexpressing CHO1), all these cells exhibit intense PI4P fluorescence in nonpunctate internal membranes consistent with cytoplasmic ER (Fig. 8).These changes in PI4P distribution correlate with rescue of PI(4,5)P2 defects as shown by GFP-2xPH(PLCδ) fluorescence.The general cytoplasmic GFP-2xPH(PLCδ) fluorescence observed in sac1 ts Δ-s-tether cells at 37 C for 1 h is absent in all sac1Δ Δ-s-tether cells overexpressing DGK1 or CHO1.These suppressors completely restore PI(4,5)P 2 distribution as observed in WT cells (Fig. 8).
Although high-copy CHO1 was effective in returning DAG distribution and levels closer to WT, DGK1 suppression did not have a similar effect (Figs. 5A, 8, and S2A).As shown by

Coordinate phospholipid regulation by SAC1 and ER-PM MCSs
C1δ-GFP fluorescence, uniform DAG localization in CHO1suppressed sac1Δ Δ-s-tether cells is partially rescued, whereas DGK1 suppression exhibits nonpolarized DAG distribution nearly identical to sac1 ts Δ-s-tether cells incubated at 37 C for 1 h (Figs. 8 and S2A).We conclude that defects in DAG distribution are not a primary cause of lethality between sac1Δ and Δ-s-tether mutations because DAG polarization at the PM is not rescued by high-copy DGK1 and only partially suppressed by CHO1 overexpression in sac1Δ Δ-s-tether cells.
Increased expression of OSH6 suppresses sac1 ts Δ-s-tether growth defects Given that PS and PI4P are the most affected of the lipids tested in sac1 ts Δ-s-tether cells, we tested if the PS/PI4P lipid transfer protein Osh6p is an effector of the combined function of SAC1 and ER-PM MCSs.As a member of the oxysterolbinding protein (OSBP)-related protein (ORP) family of soluble lipid transfer proteins, Osh6p represents one of seven yeast ORPs encoded by the OSH1-OSH7 (OSBP homologue) genes (80,81).Osh6p mediates PI4P/PS counter-directional transport in which Osh6p first transfers PS to the PM.At the PM, Osh6p reciprocally exchanges bound PS for PI4P and then returns to the ER with PI4P.The transport cycle is completed when PI4P is exchanged for PS in the ER (74,75).Both Osh6p and Osh7p interact with the ER-PM tether Ist2p to facilitate PI4P/PS transfer between the PM and ER at MCSs (82,83).We hypothesized that the lethality of SAC1 deletion in Δ-stether cells might reflect a perturbation of the Osh6pdependent cycle of PI4P/PS exchange at ER-PM MCSs.If so, Osh6p overexpression might rescue sac1 ts Δ-s-tether growth defects by boosting PI4P/PS counter-directional transport.As shown in Figure 9, a high-copy OSH6 plasmid improves sac1 ts Δ-s-tether growth defects at 30 C and suppresses sac1 ts Δ-stether lethality at 37 C, although growth is poor.High-copy OSH6 was not a bypass suppressor as it could not suppress the complete deletion of SAC1 in Δ-s-tether cells.Osh4p, in contrast to Osh6p, is also a soluble lipid transfer protein but has an affinity for PI4P and sterols, but not PS (84).Unlike OSH6, OSH4 on a high-copy plasmid failed to improve sac1 ts Δ-s-tether cell growth at 30 or 37 C (Fig. 9).Because Osh4p and Osh6p do not share the same lipid affinities, other than binding PI4P, the Osh6p mode of suppression of sac1 ts Δ-stether defects specifically involves PS synthesis and transfer.Osh6p intermembrane lipid transfer, however, is ultimately dependent on levels of PS synthesized in the ER, as well as PI4P generated in the PM by Stt4p and eliminated in the ER by Sac1p.These results are consistent with a model where greater Osh6p expression confers enhanced PS transfer to the PM that partially rescues the reduced PM levels of PS in sac1 ts Δ-stether cells.

The genomic expression profile of sac1 ts Δ-s-tether cells indicates constitutive responses to ER membrane stress and induction of autophagy gene expression
We hypothesized that changes in global gene expression in sac1 ts Δ-s-tether cells might reveal the molecular nature of their growth defects.After incubation at 37 C for 1 h, the transcriptome of sac1 ts Δ-s-tether cells was compared with WT using RNA deep sequencing analysis (RNA-seq) (Fig. 10A).Under these conditions, the transcription of 839 genes are downregulated in sac1 ts Δ-s-tether cells at least 2fold (log 2 ≤ 1) and 948 genes are upregulated 2-fold or more (log 2 ≥ 1) compared with WT. (Significant changes are defined as those involving at least a 2-fold change in transcript levels relative to WT.) Specifically, ER-stress genes represented by unfolded-protein response (UPR)-induced genes (e.g., KAR2, DER1, PDI1) are upregulated in sac1 ts Δ-s-tether cells, as are a subset of autophagy genes (e.g., ATG1, ATG8, ATG31) (Fig. 10B).In addition, expression of many lipid biosynthetic genes under UAS INO transcriptional control, including the key phospholipid synthetic gene INO1 (encoding inositol-3phosphate synthase), are significantly upregulated in sac1 ts Δ-s-tether cells (Fig. S3) (85).Taken together these changes in gene expression are integrated through the broader response elicited by the environmental stress response (ESR) pathway (Fig. 11).The ESR pathway is a global stress response pathway defined by two subsets of genes, which include those induced during general cellular stress (iESR) and a larger subset of transcriptionally repressed (rESR) genes, representing many associated with ribosomal proteins and ribosome biogenesis (86,87).iESR genes are typically involved in protein catabolism, intracellular signaling, autophagy, and stress defense.Thus, defects in sac1 ts Δ-s-tether cells not only cause ER membrane stress but also elicit broader responses to general cellular stress as coordinated through ESR regulation.The validity of the RNA-seq results was confirmed by quantitative PCR analysis of the expression of KAR2 and SIP18, performed in triplicate (Fig. S4).
We compared sac1Δ, Δ-s-tether and sac1 ts Δ-s-tether transcriptomic profiles to discern between transcript changes conferred by SAC1 deletion versus the elimination of ER-PM tethers.To our surprise, the deletion of SAC1 by itself imparts relatively minor effects relative to WT (Fig. 10B).The transcriptomic profile of sac1Δ cells reveals only 36 Coordinate phospholipid regulation by SAC1 and ER-PM MCSs downregulated genes, whereas the expression of 76 genes increased.Overall, these changes mainly involve differences in metabolic gene expression (Fig. 10A).To provide a more comprehensive transcriptomic profile of yeast lacking phosphoinositide phosphate phosphatase activity, RNA-seq was conducted on sac1 ts inp52Δ inp53Δ cells incubated at 37 C for 1 h.Although sac1Δ and sac1 ts inp52Δ inp53Δ cells share similar expression profiles, transcriptional responses to sac1 ts inp52Δ inp53Δ are far more extensive (Fig. 10A).Transcript levels of 892 genes are downregulated in sac1 ts inp52Δ inp53Δ cells and the expression of 1389 genes increase, when compared with WT levels.Repressed transcripts represented lipid biosynthesis (CHO1, PSD1, OPI3) and general metabolism pathways (MTR4, RIX1), while many upregulated Coordinate phospholipid regulation by SAC1 and ER-PM MCSs transcripts represented autophagy (e.g., ATG1, ATG8, ATG31), UPR (e.g., SLT2, KAR2, DER1, PDI1), and osmotic stress pathway genes (e.g., GRE2, GRE3, SSA1) (Fig. 10A).In contrast, the genome expression profile of Δ-s-tether cells revealed significant changes in gene expression affecting lipid biosynthesis genes and UPR regulators, as previously reported (Fig. 10) (50).Comparing the transcriptomic profiles of sac1 ts inp52Δ inp53Δ, Δ-s-tether, and sac1 ts Δ-s-tether cells indicates shared effects on ER stress and UPR activation (e.g., induction of KAR2, DER1, PDI1), but differences in gene expression suggest synergistic amplification of responses when SAC1 and tether gene disruptions are combined.
As confirmation of the UPR induction revealed in the transcriptomic profiles, we tested if sac1Δ, sac1 ts inp52Δ inp53Δ, Δ-s-tether, and sac1 ts Δ-s-tether cells are sensitive to low doses of dithiothreitol (DTT) (Fig. S5A).In the presence of 3 mM DTT, growth is inhibited in mutant cells where the UPR pathway is dysfunctional (88,89).For example, the deletion of IRE1, which encodes a key regulatory kinase for UPR activation, causes cell lethality in the presence of DTT (Fig. S5).Likewise, sac1Δ, sac1 ts inp52Δ inp53Δ, Δ-s-tether, and sac1 ts Δ-s-tether cells are also sensitive to 3 mM DTT (Fig. S5A).These results are consistent with the elevated ER stress in these mutants as determined by RNA-seq transcriptomics (Fig. 10B).
To test if autophagy or its regulation by TORC1 is disrupted in sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells, these strains were cultured on growth medium containing the TOR kinase inhibitor rapamycin and on nitrogen-depleted medium.Although WT cell growth is unaffected by treatment with 1 nM rapamycin, sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells are sensitive to 1 nM rapamycin at 30 C, similar to the rapamycin-sensitive tor1Δ control strain (Fig. S8A).Despite that sac1Δ had a nominal effect on autophagy gene expression (Fig. 10B), sac1Δ cell growth is also sensitive to rapamycin (Fig. S8A), likely due to the critical role of Sac1p in autophagosome-lysosome fusion (91).When cells are nitrogen starved, growth is inhibited in autophagy mutants due to the decrease in intracellular free amino acid pools that limits protein synthesis (92).After incubating at 30 C or 34 C for 3 days in nitrogen-depleted liquid medium, cell cultures were spotted onto rich solid medium to assess recovery from nitrogen starvation.Although Δ-s-tether cells show minor growth defects after nitrogen starvation, sac1 ts Δ-s-tether cells are marginally sensitive as compared with growth in nitrogenreplete medium at 30 or 34 C (Fig. S8B).At 30 C, sac1 ts inp52Δ inp53Δ cell growth is not affected by nitrogen starvation and sac1 ts inp52Δ inp53Δ cells do not grow at 34 C regardless of culture condition (Fig. S8B).At 34 C, the growth of sac1Δ cells is inhibited after culturing in nitrogen-depleted medium (Fig. S8B).In general, autophagy-related growth defects of sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells are consistent with the transcriptomics data.The results suggest that an autophagy regulatory response is elicited when SAC1 is inactivated in cells already affected by PI4P dysregulation.
In sac1 ts Δ-s-tether cells, specific transcriptional changes are also observed that suggest a unique signature of synergistic responses due to the combination of mutations in SAC1 and ER-PM tethers (Figs. 10 and S6).In sac1 ts Δ-s-tether cells incubated at 37 C for 1 h, 480 genes are uniquely downregulated and 230 genes are uniquely upregulated (Fig. S6A).KEGG pathway analysis of these uniquely affected genes indicates that they represent ribosome, ribosome biogenesis, and other metabolic pathways, consistent with synergistic ESR defects that are otherwise less apparent in sac1 ts inp52Δ inp53Δ cells or Δ-s-tether cells (Fig. S6B).

High-copy suppressors partly restore normal gene expression to sac1Δ Δ-s-tether cells
In total, RNA-seq analysis revealed 948 upregulated and 839 downregulated genes in sac1 ts Δ-s-tether cells, relative to WT. Transcriptome analysis of DGK1-supressed sac1Δ Δ-s-tether cells shows only 677 upregulated and 443 downregulated genes, and CHO1-suppressed strains have 585 upregulated and 417 downregulated genes, indicating a partial restoration of the WT genomic expression profile.Specifically, KEGG analysis indicated that DGK1 and CHO1 suppression resulted in a normalization of general metabolism and ribosome biogenesis transcript levels (Fig. S9).These categories generally represented genes uniquely affected in sac1 ts Δ-s-tether cells, as opposed to Δ-s-tether or sac1 ts inp52Δ inp53Δ cells (Fig. S6B).
DGK1 or CHO1 suppression of sac1Δ Δ-s-tether lethality also correlates with a partial mitigation of ESR gene expression (Fig. 11).High-copy DGK1 and CHO1 increases and restores rESR gene expression in sac1Δ Δ-s-tether cells to WT levels, although iESR gene expression is marginally affected by the suppressors (Fig. 11).As predicted, phospholipid biosynthetic gene expression is no longer induced in suppressed sac1Δ Δ-stether cells, which correlates with partial dampening of the rESR response, even though iESR gene expression is still elicited (Figs.11 and S3B).Presumably this suppression is sufficient to reduce the overwhelming level of cellular stress that otherwise results in lethality.
As validation of the genomic responses detected by RNAseq, we showed that DGK1 and CHO1 suppressors confer growth resistance to sac1Δ Δ-s-tether cells when challenged Coordinate phospholipid regulation by SAC1 and ER-PM MCSs with ER stress (Fig. S5B).Although sac1 ts Δ-s-tether cells are inviable in the presence of 3 mM DTT regardless of temperature, multicopy DGK1 or CHO1 suppresses the lethality of sac1Δ Δ-s-tether and confers robust growth on DTTcontaining solid medium (Fig. S5).By alleviating phospholipid defects in sac1 ts Δ-s-tether cells, DGK1 or CHO1 overexpression appears to protect against membrane stresses that elicit ESR.
Unlike membrane stresses, however, multicopy DGK1 or CHO1 does not suppress the growth defects of sac1Δ Δ-stether cells when challenged with 1 nM rapamycin treatment at 30 C (Fig. S8A).Moreover, when compared with growth in rich medium, sac1Δ Δ-s-tether cells expressing DGK1 and CHO1 multicopy plasmids do not fully recover after culturing in nitrogen-depleted medium for 3 days at 30 C (Fig. S8B).Although multicopy DGK1 or CHO1 rescues sac1Δ Δ-s-tether lethality, it does not confer suppression to autophagy-related defects.As such, phosphoinositide dysregulation caused by SAC1 inactivation in Δ-s-tether cells seems to be the primary trigger of autophagy gene expression, and neither DGK1 nor CHO1 restores normal PI4P distribution (Fig. S2B).The autophagy defects are, however, independent of the essential overlapping role of SAC1 and ER-PM tethers as suppressed by DGK1 and CHO1.

Discussion
ER-PM MCSs coordinately regulate lipid metabolism by acting as direct conduits for lipid transport (5,6,(93)(94)(95).Through exchange of lipid precursors between membranes, phospholipids are generated in the mitochondria, ER, and PM where tether proteins interact with different lipid transfer proteins to mediate lipid transfer.Lipid exchange between membranes is driven by concentration gradients maintained by differences in lipid levels and intracellular distribution (96).The ER and PM concentration of PS and PI4P drives their counter-directional transport, which in yeast is partly regulated by Sac1p turnover of PI4P levels in the ER and the generation of PI4P by Stt4p in the PM (96).The lethality of SAC1 deletion in Δ-s-tether cells represents the disruption of intersecting mechanisms promoting PS/PI4P exchange, namely, bringing the ER and PM into close proximity for facilitating intermembrane lipid transfer and PI4P hydrolysis in the ER membrane to generate a PI4P concentration gradient between the ER and PM.Similarly, PS levels in the ER also contribute to the PS/PI4P transport cycle, and we show the dependence of PS levels on the phospholipid flux initiating from DAG biosynthesis.We further report the interdependency of ER-PM MCS formation and lipid metabolic pathways that maintain PS/PI4P homeostasis.The elimination of multiple PI phosphate phosphatases (or to a lesser degree, Sac1p alone), or PI kinases, induces cER association with the PM via an increase in the membrane tether Tcb3p.Moreover, augmenting PS synthesis by increasing phospholipid flux through increased expression of specific phospholipid biosynthetic genes restores growth and partially rescues lipid imbalances in sac1 ts Δ-s-tether cells.As shown by transcriptomic comparisons, defects in sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells lead to the upregulation of membrane stress response pathways that most notably affect the ER, and also elicit the broader ESR pathway.Apart from membrane stress, autophagy gene expression is constitutively activated in sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells.The autophagy response is primarily attributable to phosphoinositide defects, as opposed to membrane defects caused by disruption of ER-PM MCSs.In general, these results underscore the reciprocal relationship between the formation of contact sites and lipid homeostasis, particularly with respect to phosphoinositide regulation and phospholipid synthetic flux between the ER and PM.
Although the lipidomic profiles of Δ-s-tether and sac1 ts Δ-stether cells indicate substantial increases in DAG and TAG levels at the expense of phospholipid synthesis, and DAG polarization in the PM was disrupted, PA distributions were unaffected in all mutants analyzed.The lethality of sac1 ts Δ-stether cells was suppressed by increasing the dosage of phospholipid biosynthetic genes (ALE1, DGK1, CHO1 and SLC1) that promote DAG consumption in phospholipid generation (Fig. 1A).The disruption of phospholipid biosynthetic flux in Δ-s-tether cells is predicted to decrease PI levels, which would be compounded by Sac1p elimination that otherwise contributes to overall cellular PI via PI4P hydrolysis.It was somewhat unexpected then that high-copy PIS1, which generates PI from CDP-DAG (61), does not suppress sac1 ts Δ-s-tether lethality.Previous reports showed decreases in steady-state PI levels in sac1Δ cells (57), although in our hands PI levels are modestly elevated in sac1Δ but unaffected in sac1 ts inp52Δ inp53Δ cells (Fig. 3A).PI is also required for yeast complex inositol sphingolipids, and the mole percentage of inositol sphingolipids (IPC) is unaffected in sac1Δ but reduced in sac1 ts inp52Δ inp53Δ and sac1 ts Δ-s-tether cells (Fig. 3C) (47,57).Suppressors of sac1Δ Δ-s-tether lethality, however, do not restore IPC levels (although ceramide levels are no longer reduced).The CHO1 and DGK1 suppressors do not appear to mitigate sac1Δ defects in PI production or sphingolipid regulation in the context of the SPOTS complex (63).These results point to another phospholipid, other than PI, as central to the overlapping functions of SAC1 and ER-PM MCSs.
Although the lipidomic profiles of Δ-s-tether, sac1Δ, sac1 ts inp52Δ inp53Δ, and sac1 ts Δ-s-tether indicated metabolic defects in several phospholipid species, membrane distributions of DAG, PS, PI4P, and PI(4,5)P 2 are the most perturbed in sac1 ts Δ-s-tether cells.CDP-DAG is also necessary for PS production via the synthase encoded by CHO1, which is a particularly effective suppressor of sac1 ts Δ-s-tether lethality.This result implicates PS metabolism in the overlapping function of SAC1 and ER-PM MCSs.Consistent with this conclusion, PS homeostasis contributes to maintenance of DAG polarization in the PM (97), which was also disrupted in sac1 ts Δ-s-tether cells.Indeed, all suppressors of sac1Δ Δ-stether lethality affected PS and PI4P localization resulting in their partial redistribution, as well as restoration of PI(4,5)P 2 distribution.Although the PS and PI4P distribution is not normal as compared with WT, these suppressors appear to Coordinate phospholipid regulation by SAC1 and ER-PM MCSs reestablish a partial PS/PI4P balance between the PM and ER/ internal membranes.
In WT cells, DGK1 or CHO1 overexpression causes an accumulation of cytoplasmic ER and, while CHO1 overexpression increases cER-PM association, DGK1 overexpression had little effect (Fig. 6) (50,60,65).We therefore tested if increases in nonspecific contact between cytoplasmic ER and the PM generally suppress sac1Δ Δ-s-tether lethality.HMG1 overexpression, which causes cytoplasmic ER expansion, could not rescue sac1 ts Δ-s-tether cells indicating that increasing cytoplasmic ER does not restore contact between the ER and the PM (Fig. 6F).For sac1Δ Δ-s-tether suppression, the enzymatic function of Dgk1p and Cho1p is required, which excludes the possibility of any direct membrane tethering conveyed by either protein.Moreover, additional tethering provided by expression of the artificial ER-PM staple does not suppress sac1Δ Δ-s-tether cells growth defects, indicating that nonspecific ER-PM contact is insufficient (Fig. S1).In other words, the intersecting functions of Sac1p and ER-PM MCSs are dependent on the specific tethering otherwise eliminated in Δ-s-tether cells.
Like the elimination of yeast ORPs, the inactivation of SAC1 combined with the deletion of other PI phosphate phosphatases induces Tcb3p-dependent ER-PM MCSs (50).When yeast ORP genes are deleted, Tcb3p expression increases causing further recruitment of cER to the PM (50).Increased Tcb3p expression and ER-PM association is also observed in response to ER and PM stresses including those induced by lipid dysregulation, such as sterol depletion (5,50).As observed in sac1Δ and sac1 ts inp52Δ inp53Δ cells, the redistribution of PI4P and its accumulation in the PM was also predicted to incur membrane stress (Fig. 7) (98).As such, any perturbation in cellular PI4P levels or distribution might elicit ER or PM stresses that increase Tcb3pdependent ER-PM MCSs, even those caused by stt4 ts and pik1 ts PI-4-kinase defects that impact different intracellular pools of PI4P (99).Combining mutations that eliminate ER-PM MCSs with mutations affecting key lipids regulating ER and PM integrity likely provokes an intolerable cellular stress.
Osh6p is a suppressor (albeit poor) of sac1 ts Δ-s-tether cells growth defects (Fig. 9).These genetic interactions are consistent with models in which the differential concentration of PI4P generated by Sac1p in the ER and Stt4p in the PM drives Osh6p lipid transfer at ER-PM MCSs (5,75,83,84).We propose that increasing the flux of PA and phospholipid synthesis (through either ALE1, CHO1, DGK1, or SLC1 overexpression) also fuels Osh6p-dependent PI4P/PS counterexchange by producing PS in the ER.Even the partial reestablishment of PI4P distribution between the PM and ER through increased phospholipid flux rescues sac1Δ Δ-s-tether growth defects.The Osh protein family shares overlapping functions, but this mode of suppression appears to be Osh6p specific (80).Osh4p represents an Osh6p homologue involved in similar PI4P-dependent transfer between membranes, but Osh4p is primarily localized at the Golgi where it counterexchanges PI4P for sterols, as opposed to PS (84).As such, OSH4 overexpression did not rescue sac1 ts Δ-s-tether lethality (Fig. 9).
From the RNA-seq analysis, sac1 ts Δ-s-tether cells exhibit a transcriptomic profile like other previously reported Δ-s-tether mutants (50).Given the impact of removing ER-PM tethers, it is not surprising that signature UPR genes are induced, indicating substantial ER stress.Combining sac1 ts and Δ-s-tether mutations magnifies these and other membrane stresses eliciting ESR transcriptional changes, like those observed when the yeast ORP OSH4 is inactivated in Δ-s-tether cells (50).Like sac1 ts Δ-s-tether mutations, the lethality of osh4 ts Δ-s-tether cells is also rescued by DGK1, suggesting a related mechanism of suppression that ameliorates membrane defects (50).However, the combination of SAC1 and ER-PM tether mutations elicits unique changes in gene expression.In sac1 ts Δ-stether cells, significant changes in autophagy gene expression are observed that correlate with an upregulation of autophagy activators and a downregulation of autophagy repressors.This distinctive aspect of the sac1 ts Δ-s-tether transcriptomic profile is also observed in sac1 ts inp52Δ inp53Δ cells but not in Δ-s-tether cells.These differences suggest that phosphoinositide defects trigger an autophagy response, as opposed to being caused by membrane effects resulting from ER-PM tether deletion.Previous reports showed that autophagosome fusion with the vacuole is dependent on SAC1 and other related phosphoinositide phosphate phosphatases (91,100).However, to our knowledge, these results are the first to show a phosphoinositide-dependent control of autophagy gene expression.
As a substrate for PI(4,5)P 2 synthesis, PI4P in the PM needs to be tightly controlled to maintain the vital roles of PI(4,5)P 2 in membrane trafficking, cytoskeletal organization, signaling cascades, and cell polarization (101).In sac1 ts Δ-s-tether cells, the perturbation of PI4P had the knock-on effect of disrupting PI(4,5)P 2 distribution, as shown by the cytoplasmic accumulation of the GFP-2xPH(PLCδ) PI(4,5)P 2 probe (Fig. 8).Presumably changes in PI(4,5)P 2 accessibility impede GFP-2xPH(PLCδ) contact with the PM thereby diverting the probe into the cytoplasm.In sac1 ts Δ-s-tether cells, disruptions in PI(4,5)P 2 distribution are somewhat consistent with the observed morphological defects in Δ-s-tether cells, which exhibit failures in bud growth and mother/daughter cell separation (5).Thus, an important axis controlling PI(4,5)P 2 in the PM involves maintaining PI4P distribution between the ER and PM, as driven by PS metabolism and facilitated by Osh6pdependent PI4P/PS exchange at ER-PM MCSs.This appears to be a generally conserved mechanism, given that similar events involving PI(4,5)P 2 in mammalian cells involve ORP5/8 recruitment to ER-PM contact sites for PI4P and PS exchange (102).

Strains, plasmids, microbial techniques
Yeast strains and plasmids used in the study are listed in Tables S1 and S2, respectively.Throughout the experiments, yeast strains were cultured in Yeast extract-Peptone-Dextrose, Coordinate phospholipid regulation by SAC1 and ER-PM MCSs synthetic minimal or synthetic complete medium at 30 C, unless otherwise mentioned.To test growth defects of temperature conditional mutants, sac1 ts Δ-s-tether and sac1 ts inp52Δ inp53Δ cells and WT controls were cultured at permissive growth temperatures (30 C unless otherwise stated) and shifted to 37 C, as specified.Yeast growth on 3 mM DTT (Sigma-Aldrich Canada Co), or 0.1 and 1.0 nM rapamycin (Bioshop Canada Inc), on solid selective medium was assessed after 3 to 5 days.As described, autophagy induction by starvation was tested by culturing yeast in nitrogendepleted liquid synthetic medium for 3 days (91) at the indicted temperatures, and then cultures were spotted onto solid rich synthetic medium and incubated at 30 C for 3 to 4 days.Recovery from nitrogen starvation was assessed compared with strain growth after culturing in rich medium for 3 days.Growth plate assays represent ≥3 trials.
High-copy suppressors of sac1Δ Δ-s-tether lethality were tested by plasmid shuffle involving the transformation of LEU2-marked suppressor plasmids into sac1Δ Δ-s-tether cells containing SCS2 on a URA3-marked plasmid.Bypass suppression of sac1Δ Δ-s-tether growth defects was then evaluated by selection against cells containing the SCS2 URA3marked plasmid on solid synthetic medium containing 1 g/l 5 0 -fluoroorotic acid (5 0 -FOA) (Bioshop Canada).DNA cloning and bacterial yeast transformations were performed using standard techniques (103,104).
Given the inherent genetic instability of sac1 ts Δ-s-tether cells, this strain was maintained by culturing at 30 C on synthetic medium supplemented with 1 mM choline (Sigma-Aldrich Canada) or transformed with a URA3-marked plasmid containing SCS2.For the latter, before use in growth assays or integrations/transformations, cells transformed with the SCS2 plasmid are selected against on 5 0 -FOA medium.Before and after each transformation or growth assay, the temperature sensitivity of sac1 ts Δ-s-tether-derived cells is confirmed by growth inhibition at 37 C.
To generate the plasmid expressing E. coli dgkA from the yeast P GAL1 promoter (pCB1435), a SacI/KpnI fragment containing the P GAL1 -dgkA fusion gene was subcloned from pRS424GAL1pr-DGK (69) into the same restriction sites in pRS426.To generate the catalytically dead cho1 D127A mutation, site-directed mutagenesis was performed on pCB1351 containing the full 1.5-kb CHO1 gene.With the incorporation of the altered codon, the CHO1 gene was PCR amplified with the mutagenic TCCCCACTCCTTCTCAATGT forward primer and the GTGGTTGGCATAGGCAATCC reverse primer.The amplified mutant plasmid product (pCB1427) was enriched over the original template plasmid by DpnI digestion, and the mutagenic change was confirmed by DNA sequencing.

Fluorescence microscopy and live-cell imaging
Superresolution fluorescence microscopy was performed on a Zeiss LSM 880 confocal laser scanning microscope with an Airyscan superresolution GaAsP detector and 63×/1.4oil immersion objective (Zeiss).All fluorophores were acquired using pixel dwell times at approximately 1.31 μsec per pixel.
DsRed-HDEL fluorescence was excited using a 561-nm laser, and GFP fusions were excited using a 488-nm laser.Relative intensities were set to 1.5 for both lasers.Digital gain was set to 900 for the 488-nm laser and 800 for the 561-nm laser.Images were Airyscan processed in Zen Black and deconvolved in Zen Blue (Zeiss).ER association with the PM was determined by tracing the cell cortex in the Zen Blue profile mode, then measuring cER fluorescence intensity at the cortex.ER association with the cell cortex was expressed as a ratio of the total distance of cER fluorescence to the total PM perimeter.Images were exported as 8-bit uncompressed TIFF files then processed in Affinity Photo (Serif Ltd).Contrast enhancement was kept constant for each series of images.Levels of PM Lact-C2-GFP fluorescence were quantified using ImageJ (https:// imagej.nih.gov/ij/index.html) by determining the mean fluorescence of selected areas corresponding to the cell cortex.

Extragenic suppressor selection
To select for extragenic suppressors of sac1 ts Δ-s-tether lethality, a 2μ high-copy yeast genomic plasmid library was transformed into sac1 ts Δ-s-tether cells and cultured at 37 C. To avoid isolation of ER-PM tether genes as dosage suppressors, genomic DNA was purified from Δ-s-tether (CBY5838) cells and partially digested using Sau3AI and 10-kb fragments were size selected from and ligated into the BamHI digested 2μ vector, YEplac195.This 2μ genomic library was then transformed into sac1 ts Δ-s-tether cells and cultured on solid synthetic medium lacking leucine and uracil for 3 to 7 days at 37 C. Surviving colonies from an equivalent of 9000 transformants (6 genomic equivalents) were selected and colony purified.Individual genomic library plasmids were recovered and retransformed back into sac1 ts Δ-s-tether cells to confirm suppression at 37 C. Genes on each suppressor DNA fragment were identified by DNA sequencing.To identify individual suppressing genes from multiple genes on each genomic fragment, each candidate suppressor was individually cloned into YEplac195 and tested for suppression after transformation into sac1 ts Δ-s-tether cells.Bypass suppression was tested by subcloning suppressor genes into YEplac181, the suppressor constructs were then transformed into sac1Δ Δ-s-tether cells containing SCS2 URA3-marked plasmid to maintain viability, and then the URA3 plasmid was counter-selected by streaking transformants onto solid selective synthetic medium containing 5 0 -FOA.

Lipidomics
For lipidomics analysis, cells were first grown in culture to 0.8 A 600 and pelleted cells were mixed with 200 μl methanol and 100 μl deionized water.Resuspended cell pellets were sonicated three times for 15 s on ice, after which 500 μl methanol and 200 μl of chloroform were added to each sample.The sonication was repeated before centrifugation at 21,000g for 15 min.The clarified supernatant was collected and dried under a gentle nitrogen gas flow at room temperature.As a normalization control, protein concentrations of resuspended cell pellets were determined by Bradford analysis Coordinate phospholipid regulation by SAC1 and ER-PM MCSs (Sigma-Aldrich Canada).Dried supernatant residues were dissolved in 20 μl isopropanol per microgram of protein in the original resuspended cell pellet extracts.
Liquid chromatography (LC)-mass spectrometry (MS) lipid analyses from extracted, dried supernatants and resolubilized samples were performed at the University of Victoria Genome BC Proteomics Centre.For the analysis, a quality control sample was prepared by pooling 10 μl of each sample solution.For LC-MS analysis, 10-μl aliquots of sample solutions and the quality control sample were injected in a random order into a Thermo LTQ-Orbitrap Velos Pro mass spectrometer.Using 0.1% formic acid (A) and 0.1% formic acid acetonitrile-isopropanol 2:1 (B), samples were analyzed in duplicate on a C8 LC column (2.1 × 50 mm, 1.7 μm) at 50 C with a flow rate of 0.4 ml/min as described (105).Samples were then analyzed by ultraperformance liquid chromatography-high-resolution MS using full-mass detection within m/z 200 to 1800 with (+) and (−) electrospray ionization.Along with full-mass MS runs, LC-tandem mass spectrometry determinations were acquired using collision-induced dissociation.
For the two full-mass LC-MS datasets acquired, raw data files were converted to a common data format and then processed in R using a customized script for peak detection, retention time, shift correction, peak grouping, and peak alignment in two rounds.Data processing provided m/z ratios, retention time, and peak area of the detected lipids in pairs.Raw data for both ESI+ and ESI-polarities were subject to principal component analysis with m/z ratios and retention time being set as x-variable markers.The peak area for each putative marker was normalized by the total markers area and used for quantification.Mass deisotoping and removal of chemical background peaks were then performed.Using HMDB (https://hmdb.ca/spectra/ms/search) and COMP-DB (www.lipidmaps.org/tools/ms/lm_mass_form.php)databases, lipid identities were assigned based on corresponding lipid m/z ratios; allowable mass error was set to ≤5 ppm in all cases.Lipids that presented a coefficient of variation ≥30% (12.8% of lipids) were excluded from further evaluation.For the (+) ion detection data, ionic forms of (M + H)+ and (M+Na)+ were used, whereas (M-H)-, (M+Na-2H)-, and (M+Cl)-forms were used for the (−) ion detection dataset.As described (106), m/z ratios were matched to specific lipids based on accurate mass matching, class-specific retention time, and adduct-type consistency.Values are represented as lipid content relative to WT.

Phospholipid pulse labeling
Prior to pulse labeling of de novo synthesized phospholipids, WT and mutant cells were cultured at 30 C to a density of A 600 = 1.0 in synthetic growth medium.For temperaturesensitive conditional mutants, strains were then shifted to 37 C for 1 h prior to labeling.To initiate phospholipid labeling, cells were cultured with 50 μCi/ml [ 32 P]H 3 PO 4 and equal samples were removed after 2, 5, 10, and 20 min of incubation.Labeling was terminated by resuspending cells in 5% trichloroacetic acid and then chilling on ice for 30 min prior to lipid extraction (107).Phospholipid extraction and lipid separation by one-dimensional paper thin-layer chromatography (TLC), was essentially performed as described (36).Before separation, samples were spotted and dried onto TLC silica gel 60 plates (MilliporeSigma), which were then imaged and analyzed on an Amersham Typhoon IP phosphoimager (GE Healthcare Life Sciences).

Genomic expression analysis
Transcriptomic analysis by RNA-seq was performed as described (50).Mid-log-phase cells were cultured at 30 C in synthetic minimal medium, or temperature-sensitive cells were cultured at 30 C before incubation at 37 C for 1 h, prior to poly(A) mRNA isolation and cDNA library generation.Read quality controls, alignments, read counting, gene ontology, and statistical analysis were performed as described (50).RNA-seq data were validated by quantitative PCR as described (50), whereby transcript levels were quantified by comparing SIP18 and KAR2 cDNA amounts with amplified ACT1 cDNA as the internal control.

Figure 2 .
Figure 2. Inactivation of the phosphatidylinositol kinases Pik1p or Stt4p increases cER-PM association.A, representative images of WT (BY4741), stt4 ts (CBY5090), or pik1 ts (CBY5092) cells expressing the ER marker DsRed-HDEL (pRS416-DsRed-HDEL) and stained with a blue cell surface dye (MemBrite Fix 405/ 430).Cells were incubated at either 30 C or 37 C for 1 h, as indicated.B, ratios of cER coverage per total perimeter distance of the PM corresponding to (A) (N = 20 cells/strain).C, representative images of Tcb3p-GFP (pWK092) expressed in WT, stt4 ts or pik1 t cells, incubated at either 30 C or 37 C for 1 h.D, ratios of Tcb3p-associated cER coverage per total distance of the cell perimeter/PM corresponding to (C) (N = 20 cells/strain).Arrowheads indicate cER associated with the PM, and asterisks indicate cells with nearly absolute coverage of the cortex with ER.The scale bar represents 5 μm.***p ≤ 0.00015.cER, cortical endoplasmic reticulum; ER, endoplasmic reticulum; PM, plasma membrane.

Figure 4 .
Figure 4. Lipid biosynthesis genes and CUE1 are dosage suppressors of the lethal inactivation of SAC1 in Δ-s-tether cells.A, tenfold serial dilutions of WT (SEY6210) and sac1 ts Δ-s-tether (CBY6345) cells transformed with high-copy plasmids containing SCS2 (pSCS2), a genomic fragment containing SLC1 (pCB1350), and the vector control (YEplac195) cultured on selective solid medium for 3 to 5 days at the indicated temperatures.B, tenfold serial dilutions of WT and sac1 ts Δ-s-tether cells transformed with a high-copy plasmid containing a genomic CUE1 fragment and the vector alone control cultured on selective solid medium for 3 to 5 days at the indicated temperatures.C, tenfold serial dilutions of WT and sac1Δ Δ-s-tether cells with high-copy plasmids containing DGK1 (pCB1346), CHO1 (pCB1352), SLC1 (pCB1350), ALE1 (pCB1382), or the corresponding vector control (YEplac181), grown and spotted onto solid synthetic medium at 30 C for 3 days.Strains shown successfully grew after high-copy plasmid suppressors were transformed into WT and sac1Δ Δ-stether cells containing an SCS2 plasmid (pSCS2), followed by SCS2 counter-selection with the addition of 5 0 -fluoroorotic acid.D, tenfold serial dilutions of WT and sac1Δ Δ-s-tether cells containing episomal SCS2 transformed with high-copy PIS1 (pCB1345) in which the SCS2 plasmid was either selected for (+SCS2) or counter-selected against (−SCS2), grown at 30 C for 5 days.E, tenfold serial dilutions of WT (SEY6210) and sac1 ts inp52Δ inp53Δ (AAY143) cells transformed with high-copy DGK1, CHO1, and the vector alone control.F, tenfold serial dilutions of WT and Δ-s-tether (CBY5838) cells containing high-copy ALE1, SLC1, DGK1, CHO1, and the vector alone control cultured at 30 C for 4 days.

Figure 6 .
Figure 6.High-copy DGK1, CHO1, or HMG1 increase cER, but HMG1 overexpression does not suppress sac1 ts Δ-s-tether lethality.A, representative images of endoplasmic reticulum-stained DsRed-HDEL in WT (SEY6210) and sac1 ts Δ-s-tether (CBY6345), or WT and sac1Δ Δ-s-tether cells expressing highcopy DGK1 (CBY6508) or CHO1 (CBY6522) and counterstained with a blue cell surface dye (MemBrite).Cells were incubated at 30 C or at 37 C for 1 h, as indicated.B, quantification of cell ratios of cER coverage per total distance of the PM perimeter, as corresponding to images shown in (A) (N = 20 per strain; *** p < 2 × 10 −5 ).C, representative images of endoplasmic reticulum-marked DsRed-HDEL in WT and Δ-s-tether (CBY5838) cells expressing high-copy HMG1 (+HMG1; pCB1402) or vector alone control (-HMG1; YEplac195) treated with the blue cortical dye and cultured at 30 C. D, quantification of cell ratios of cER coverage per total distance of the PM perimeter corresponding to the images in (C) (N = 20 per strain; ***p < 1 × 10 −9).E, tenfold serial dilutions of WT and Δ-s-tether cells expressing high-copy HMG1 or the vector control spotted on solid synthetic medium without supplemental 1 mM choline and cultured for 3 days at 30 C. F, tenfold serial dilutions of WT and sac1 ts Δ-s-tether cells expressing high-copy HMG1 or the vector control spotted on solid synthetic medium and cultured for 3 to 4 days at 30 C or 37 C.The scale bars represent 5 μm.cER, cortical endoplasmic reticulum; PM, plasma membrane.

Figure 10 .
Figure 10.Transcriptomic profiles of sac1 ts Δ-s-tether, Δ-s-tether, sac1Δ, and sac1 ts inp52Δ inp53Δ.A, volcano plots showing relative transcript abundance in Δ-s-tether (CBY5898) at 30 C, and sac1 ts Δ-s-tether (CBY6345) and sac1 ts inp52Δ inp53Δ (AAY143) incubated at 37 C for 1 h.Transcript changes are shown relative to WT (SEY6210) incubated under the same respective conditions.sac1Δ (CBY2809) was cultured at 30 C relative to its congenic WT control (BY4741).Plots show the statistical significance of the difference in expression (negative log 10 -p value) versus log 2 -fold transcript changes, with examples of downregulated genes in blue and upregulated genes in red.B, heatmap analyses of autophagy and unfolded protein response (UPR) gene transcript changes relative to congenic WT controls for sac1 ts Δ-s-tether and sac1 ts inp52Δ cells inp53Δ at 37 C for 1 h, and for Δ-s-tether and sac1Δ cells at 30 C. Autophagy genes were curated using the Saccharomyces Genome Database (SGD), and UPR genes were curated from Kimata et al. (120) and SGD.Upregulated genes for each strain are shown in red, and downregulated genes are indicated in green.