Periprotein membrane lipidomics and the role of lipids in transporter function in yeast

The yeast plasma membrane is segregated into domains: the Micro-Compartment-of-Can1 (MCC) and Pma1 (MCP) have a different protein composition, but their lipid composition is largely unknown. We extracted proteins residing in these microdomains via stoichiometric capture of lipids and proteins in styrene-maleic-acid-lipid-particles (SMALPs). We purified SMALPs by affinity chromatography and quantitatively analyzed the lipids by mass spectrometry and their role in transporter function. We found that phospholipid and sterol concentrations are similar for MCC and MCP, but sphingolipids are enriched in MCP. Ergosterol is depleted from the periprotein lipidome, whereas phosphatidylserine is enriched relative to the bulk of the plasma membrane. Phosphatidylserine, non-bilayer lipids and ergosterol are essential for activity of Lyp1; the transporter also requires a balance of saturated/unsaturated fatty acids. We propose that proteins can function in the yeast plasma membrane by the disordered state of surrounded lipids and diffuse slowly in domains of high lipid order. Impact statement Membrane protein-specific lipidomics provides information on the organization of the yeast plasma membrane and the functioning of solute transporters


Introduction 36 37
The interior of the cell is separated from the exterior by a lipid bilayer.
(1) Cells can have more than 38 1,000 different lipid species, and the molecular composition at any point on a planar membrane, or 39 in the inner and outer leaflets, differs. Eukaryotic membranes vary along the secretory pathway, and 40 the plasma membrane is enriched in sphingolipids and sterols(2). Furthermore, the inner and outer 41 leaflet of the plasma membrane of eukaryotic cells differ in lipid species: anionic lipids(3) and 42 ergosterol (4)  Yeast and many other fungi tolerate a low external pH(6) and high solvent concentrations (7), which 47 suggests a high degree of robustness of their plasma membranes. This correlates with observations 48 that the lateral diffusion of proteins in the plasma membrane is extremely slow and the permeability 49 for small molecules is low as compared to mammalian or bacterial membranes (8,9), suggesting high 50 lipid order within the plasma membrane(10). In yeast, the lateral segregation of lipids is associated 51 with a differing location of marker proteins. Of these, the Membrane Compartments of Pma1 (MCP) 52 and Can1 (MCC) (11,12) are well studied. The proton-ATPase Pma1 strictly localizes to the MCP, 53 whereas the amino acid transporters Can1 and Lyp1 localize to the MCP or MCC depending on the 54 physiological condition (8,13). Many more proteins are associated with MCC and MCP, but the 55 molecular basis of their partitioning is elusive (14). The MCC is part of a larger complex called the 56 'Eisosome', which stabilizes the MCC membrane invaginations(15), leading to the term 57 'MCC/eisosomes.' In terms of lipid composition, the MCP is believed to be enriched in sphingolipids 58 and the MCC in ergosterol(10,16,17), but evidence for this partition is indirect and mainly based on 59 fluorophore binding(17) and lipid-dependent protein trafficking(13). Attempts to accurately 60 determine the lipids of the yeast plasma membrane after cell fractionation are hampered by 61 impurities from other organellar membranes. 62 63 A recently developed method involving Styrene-Maleic-Acid (SMA) polymers(18) allows extraction of 64 proteins from their native lipid bilayer. SMA-polymers form protein-lipid containing disc-shaped 65 structures, called 'SMALPs', preserve the periprotein lipids or even more distant lipid shells 66 depending on the disc-size. We adopted this method for direct, in situ detection of lipids that 67 surround named membrane proteins, which minimizes contamination of lipids from other 68 membrane compartments. We developed a three-step method whereby very small lipid shells are 69 generated, hereafter called the periprotein lipidomes. Shells containing a named protein are 70 captured, and lipids associated with each named protein are detected by lipidomics in defined 71 domains such as MCC and MCP. Finally, we used the lipidomics data to test the lipid dependence of 72 amino acid transport by purified Lyp1 reconstituted in synthetic lipid vesicles. 73 74

76
Approach to periprotein membrane lipidomics 77 We used SMA to extract transmembrane proteins with surrounding lipids from the plasma 78 membrane to capture periprotein-lipid discs, called SMA-Lipid Particles SMALPs. This approach (19)  79 avoids detergents normally needed to capture membrane proteins and selectively captures lipids 80 within a disc of defined diameter of 9 nm ± 1 nm(18) (Fig. S1), which normally exceeds the area of 81 most membrane proteins (~20 nm 2 ), so we expected less than 5 concentric layers of lipid. Given the 82 density of proteins in cells of ~ 3 per 100 nm 2 (20) versus a SMALP area of ~50 nm 2 (assuming SMA-83 polymer contributes 1 nm to the radius), we reasoned that most SMALPs would contain a single 84 protein and lipid would be in moderate stoichiometric excess. These predictions based on the known 85 cross-sectional area of lipids proteins and SMALPS estimate the SMA:protein:lipid of 1:1:60-120 ( Fig.  86 1). We refer to these lipids as the periprotein lipidome, which includes more than just the annular 87 lipids; the latter are defined as lipids directly contacting the transmembrane domain. 88 89 We determined the lipidomes associated with Pma1, a genuine MCP resident, Sur7, a genuine MCC 90 resident and the amino acid transporters Can1 and Lyp1, which cycle between MCP and MCC. Can1 91 and Lyp1 leave the MCC and are internalized from the MCP when arginine (substrate of Can1) and 92 lysine (substrate of Lyp1) are present in excess (8,21). At low concentrations of arginine and lysine, 93 the Can1 and Lyp1 predominantly localize in the MCC (up to 60% of Can1 and Lyp1 molecules)(8). To 94 trap Can1 and Lyp1 in the MCC and to obtain a better representation of protein-specific MCC lipids, 95 we used a GFP-binding protein (GBP)(22) fused to the MCC resident Sur7 to specifically enrich for 96 Lyp1-YPet and Can1-YPet proteins in the MCC (Fig. S2). The GFP-binding protein binds YPet with high 97 affinity and sequesters YPet-tagged proteins, when Sur7-GBP is present in excess. 98 99 We engineered a C-terminal 10-His-tag to each of the proteins and used metal-affinity (Nickel-100 Sepharose) and size-exclusion chromatography for purification of SMALPs containing either Pma1-101 Ypet, Sur7-Ypet, Can1-YPet or Lyp1-YPet with measurable purity (Supplementary Figure 3A). SDS-102 PAGE analysis shows multiple protein bands (Fig. S3B), presumably due to proteolysis of protein 103 loops during purification. Indeed, 2D native-denaturing gel electrophoresis shows that the vast 104 majority of protein bands are genuine parts of Pma1-Ypet, Sur7-Ypet, Can1-YPet or Lyp1-YPet (Fig.  105 S3C). Each protein migrates as a single band on a native gel and segregates into multiple bands when 106 SDS is included in the 2 nd dimension of the electrophoresis. Furthermore, MS analysis of proteins in 107 SMALPs shows peptide coverage across the full-length amino acid sequence (Fig. S3D). Finally, MS 108 analysis of lipids extracted by SMA polymer from synthetic lipids vesicles shows that the procedure 109 does not bias towards the extraction of specific phospholipids (Fig. S4A), similar to prior 110 observations of sphingolipids and sterols(23,24). 111 112 113 114

122
Lipidomics analysis of periprotein microdomains 123 Next, the SMALP-lipid-protein complexes were treated with a chloroform-methanol-water mixture 124 (1:2:0.8) (25) Figure 2B). The identification of these yeast lipids was confirmed by the CID-160 MS (Fig. 2D). 161 162 The lipidomic profile of SMALP-Sur7 was generated by the same approach as described for SMALP-163 Pma1   to inositol-phosphoceramide (IPC), mannosyl-inositol-phosphoCeramide (MIPC) and ergosterol (Fig. 195 3A and S6A)). We did not initially detect IPC and MIPC in the computerized high throughput analysis 196 of Sur7 lipidomes. These ions were missed by automated peak-picking algorithms, likely due to their 197 very low intensity and nearby background signals. but for Can1-MCP the ratios were not significantly reduced (Fig. S6). These ratios were consistent 206 with the conclusion that MCC-Sur7 domains have reduced sphingolipids as compared to MCP-Pma1, 207 but inconsistent with MCC-Sur7 periprotein domains being enriched in ergosterol (16,17). Separately, 208 we performed ergosterol staining by filipin in our cells using Sur7-Ypet as reporter of 209 MCC/eisosomes and did not find enhanced filipin fluorescence at the MCC domains (Fig. S7). 210 211 Quantitative lipid analysis of SMALPs 212 From the overall lipidomes of MCC-Pma1 and MCP-Sur7, we found that many lipid species are 213 detected in both domains. Next, we investigated whether the protein-associated domains vary in 214 phospholipids whose enrichments and roles in these domains are unknown. We estimated the lipid 215 quantity by comparing the peak areas of ion chromatograms to the external standard curves for PI 216 (34:1), PC (34:1), PE (34:1), PS (34:1), PG (36:1), PA (34:1), CL (72:4), and ergosterol ( Fig. S8A). 217 Ranking the lipids by yield, we found the most abundant lipid species are PC, PI, PE, PS, and 218 ergosterol with minor quantities of PG, PA and cardiolipin (Fig. 4A). 219 220 We calculated the molar ratio of phospholipids and ergosterol associated with each SMALP-protein, 221 assuming 1:1 stoichiometry of SMALP to protein. The estimated average numbers of phospholipids 222 plus ergosterol per SMALP-protein complex are 52, 66, 70, and 87 associated with Sur7, Can1, Lyp1, 223 and Pma1, respectively, which is proportional to the protein sizes (Fig. 4B). From these ratios, we 224 estimate 1-2 rings of lipid around each protein. Since the injection amount for lipid analysis was 225 normalized based on the same molar amount of the input protein, the smaller protein size of Sur7 226 could associate with less lipids and cause the fewer total ions detected in the SMALP-Sur7. However, 227 due to variation among three independently purified SMALP-protein complexes, the protein size 228 dependent lipids association did not reach statistical significance. 229 230 For the lipid class composition, we focused on the 4 major phospholipids and ergosterol regarding 231 their relative abundance (Fig. 4C). To obtain better estimates of lipid quantities, we calculated 232 conversion factors obtained from external standards and the standard addition method, which relies 233 on true internal standards ( Fig. S8 B-D). SMALPs containing Pma1, Can1, and Lyp1 were purified 234 from the Y8000 strain, whereas SMALP-Sur7 were purified from Y5000 strain. Therefore, the total 235 plasma membrane extracts of these two strains were also analyzed separately (Fig. 4D). We found 236 that Y8000 samples, including Pma1-MCP, Lyp1-MCP, Lyp1-MCC, Can1-MCP, and Can1-MCC consist 237 of similar compositions of PC (~40%), PI (~20%), PE (~18%), PS (~16%) and ergosterol (~4%) (Fig. 4C). 238 We also noticed that for the overall (plasma) membrane of strains Y5000 and Y8000, ergosterol (25-239 30 mol%) is 6-fold ( Fig. 4D) higher than in the SMALPs, which suggests that ergosterol is depleted 240 from the periprotein lipidome and more abundant in the bulk lipids surrounding the MCC and MCP 241 proteins. Similarly, we observe 2 to 3-fold higher PS in all SMALP samples compared to overall 242 (plasma) membrane, indicating that PS is enriched in the periprotein lipidome. 243 244 For the fatty acyl chain distribution, we found slight differences between strains. However, for both 245 strains, we observed the similar fatty acyl chain distribution patterns in the four major phospholipid 246 classes, as compared between the SMALP-protein associated lipids and their parent strain overall 247 plasma membrane lipids ( Fig. 4E and S9). We detected 11 forms of PC, 10 PIs, 10 PEs, and 7 PSs. We 248 found C34:2 as major acyl chain for PE and PS and C32:2 and C34:1 for PC and PI, respectively. 249 250 In conclusion, periprotein lipidomes from SMALPs differ from the overall plasma membrane in 251 ergosterol and PS content. Furthermore, the relative abundance of IPC and MIPC vary between the 252 MCC and MCP, whereas differences in the major membrane phospholipids are small and not 253 significant.

275
Phospholipid dependence of transporter function 276 Next, we sought to validate possible functional implications of the lipidomics data by testing the lipid 277 dependence of Lyp1, using membranes formed from synthetic lipids. We previously reconstituted 278 Lyp1 in lipid vesicles composed of yeast total lipid extract(31). The yeast sphingolipids IPC and MIPC 279 and certain headgroup-acyl chain combinations of the phospholipids are not available, but we could 280 design artificial membranes that otherwise mimic those found in cells. In initial experiments with 281 ergosterol present we observed that C34:1 (C18:1 plus C16:0 chains) palmitoyl-oleoyl-sn-282 phosphatidylX (POPX, where X=choline, ethanolamine, glycerol or serine), support a much higher 283 transport activity than the most commonly used C36:2 forms or when C32:0 dipalmitoyl-sn-284 phosphatidylX (DPPX; C32:0 = 2 C16:0 chains) lipids ( Fig. S10A) were used. When we exploit the 285 lipidomics data and prepare vesicles with the dominant acyl chains for each lipid species using C32:2 286 PC and PE and C34:1 PS, which is the second most abundant PS, we also find low transport activity 287 (Fig. S10B). Hence, we started with a mixture of POPS, POPG, POPE and POPC (17.5 mol% each) plus 288 30 mol% ergosterol (Fig. 5C, sample 1) and used that to benchmark the activity of Lyp1 in vesicles 289 against the effects of phospholipids and sterols. 290 291 To drive the import of lysine by Lyp1, we impose a membrane potential (ΔΨ) and pH gradient (ΔpH) 292 by diluting the vesicles containing K-acetate into Na-phosphate plus valinomycin. The magnitude of 293 the driving force is determined by the potassium and acetate gradient (Fig. S11A). Typically, we use a 294 ΔΨ and -ZΔpH of -81 mV each (Z=58 mV), producing a proton motive force of -162 mV. We 295 compared the lipid composition of the Lyp1 vesicles with the starting mixture for the reconstitution, 296 using mass spectrometry, and do not find significant differences (Fig. S4B). This rules out enrichment 297 or depletion of certain lipids during membrane reconstitution, including the possibility that co-298 purified protein lipids contribute significantly to the lipid pool. Furthermore, proton permeability 299 measurements indicate that the transport rates are not skewed by large differences in proton 300 leakage (Fig. S12). Hence, the proton motive force is similar in vesicles of different lipid composition. 301 302 Supplementary Figure S11B shows the import of lysine over time, from which the slope estimates 303 the initial rate of transport. Figure 5C shows that increasing either POPS or POPE increases transport 304 activity (samples 2 and 3), while reducing the fraction of these lipids decreases the activity (samples 305 5 and 6) relative to that in the benchmark mixture (sample 1). Increasing the fraction of POPG or 306 lowering of POPC (sample 4) had no negative effect on transport. This suggests that the non-bilayer 307 lipid POPE and the anionic lipid POPS are a minimal requirement for transport (sample 7). 308 309 310 Figure

317
The role of anionic lipids 318 Next, we simplified the lipid mixture by preparing vesicles composed of POPS, POPE and Ergosterol 319 (Fig. 5C, sample 7) and step-wise reduced the quantity of PS. We found a sigmoidal relationship 320 between Lyp1 activity and POPS concentration (Fig. 6A), which is indicative of cooperativity and 321 suggests that more than one molecule of POPS is needed for activation of Lyp1. The anionic lipid 322 POPG can only partly substitute for POPS (

335
Sigmoidal curves were fitted using the equation: The role of non-bilayer lipids 338 Many phospholipids have a cylindric geometry that allow bilayer formation with a single lipid 339 species. The relatively small headgroup of PE, as compared to the acyl chains, results in cone 340 geometry that does not allow bilayer formation from pure PE. PE and other non-bilayer lipids affect 341 the lateral pressure profile and membrane curvature more drastically than bilayer forming lipids 342 do(32). We determined the PE dependence of Lyp1 by increasing the fraction of POPE at the 343 expense of POPC (Fig. 6B). Lyp1 transport activity increases 3-fold with increasing POPE and 344 saturates at ~10 mol%. To determine whether the ethanolamine headgroup or the geometric shape 345 of the lipid is important, we substituted POPE for POPA (Fig. 6C), a non-bilayer forming conical 346 phospholipid devoid of the headgroup moiety in PE (Fig. 5A). POPA can fully substitute POPE in 347 transporter activity, which suggests that some lipids with non-bilayer properties are important for 348 Lyp1 function. Next, we titrated in DOPE at the expense of POPE to gradually increase the degree of 349 acyl chain unsaturation (Fig. 6D). We observe a 2-fold increase in Lyp1 activity with DOPE at 5 to 10 350 mol% and POPE at 30 to 25 mol%, but a further increase in dioleoyl at the expense of palmitoyl-351 oleoyl chains decreases the activity to zero. These experiments indicate that specific features of the 352 acyl chains (or fluidity) are as important as the geometric shape of non-bilayer lipids.

363
Ergosterol is essential for Lyp1 activity 364 Ergosterol is the major sterol of lower eukaryotes and present in the yeast plasma membrane at 365 concentrations of ≈ 30 mol% (33), but the fraction of ergosterol in the periprotein lipid shell of Lyp1 366 (and Can1, Pma1 and Sur7) appears 6-fold lower than in the bulk membranes (Fig. 4B). We increased 367 the fraction of ergosterol at the expense of POPC and observed an increase in activity up to 10 368 mol%. Without ergosterol, Lyp1 is not active and the activity drops above 25 mol% (Fig. 7A). As with 369 POPS we find a sigmoidal dependence but the apparent cooperativity is much lower. Cholesterol 370 supports less than 15% of Lyp1 activity as compared to ergosterol. Ergosterol differs from 371 cholesterol in that it has two additional double bonds at positions C7-8 and C22-23 and one extra 372 methyl group at C24 (Fig. 7C yellow ovals). To find out which of these is important for transport 373 activity, we tested brassicasterol and dehydrocholesterol and an equal mixture of both (Fig. 7C). The 374 two sterols and the mixture thereof cannot substitute for ergosterol ( Fig 7B). 375 376 In summary, anionic lipids with saturated and unsaturated acyl chains, preferably POPS, and 377 ergosterol are essential for Lyp1 functioning, and both lipid species stimulate transport 378 cooperatively. This suggests that "allosteric" sites on the protein need to be occupied by specific 379 lipids to enable transport. 380 381 Model for protein functioning in a highly ordered yeast PM 382 383 The periprotein lipidomes and published literature lead to a new testable model (Fig. 8) of how 384 proteins may function in a membrane of high lipid order, slow lateral diffusion and low 385 permeability (8,9,14,34). We observe that membrane proteins like Lyp1 require a relatively high 386 fraction of lipids with one or more unsaturated acyl chains that allow sufficient conformational 387 flexibility of the proteins. The proteins with periprotein lipidome are embedded in an environment 388 of lipids that are enriched in ergosterol and possibly saturated long-chain fatty acids such as present 389 in IPC, MIPC and M(IP) 2 C), which yield a highly liquid-ordered state. The ordered state forms the 390 basis for the robustness of the organism to survive in environments of low pH or and high solvent 391 concentration(6,7) and likely explains the slow lateral diffusion and the low permeability of the yeast 392 plasma membrane (Fig. 8). Facilitator Superfamily (Hxt6, Gal2, Ptr2). These proteins undergo conformation changes when 410 transiting between outward and inward conformations, which would be hindered in a highly liquid-411 ordered membrane, where lipids will have to be displaced when the protein cycles between the 412 outward-and inward-facing state. To obtain an estimate of the number of displaced lipids needed 413 for such a conformational change, we analyzed the X-ray structures of the Lyp1 homolog LeuT in 414 different conformations(35). We analyzed structures oriented in the membrane from the OPM 415 database(36), which positions proteins in a lipid bilayer by minimizing its transfer energy from water 416 to membrane. We have used a numerical integration method to estimate the surface area of the 417 outward and inward state of LeuT in the plane of the outer-and inner-leaflet at the water 418 membrane interface (Fig. 9). We estimated the number of lipids in the vicinity of the protein by 419 drawing an arbitrary circle around the protein. With a radius of 35 ångström from the center of the 420 protein we need 43 to 50 lipids per leaflet depending on the conformation. For the inner-leaflet, the 421 inward-to-outward movement of LeuT requires plus three lipids and for the outer-leaflet minus 422 three lipids, which can probably be accommodated by local changes in membrane 423 compressibility(37) and by redistribution of the annular and next shell of lipids even if they are 424 surrounded by membrane that is in a highly liquid-ordered state. We find similar changes in the 425 numbers of lipids when we analyze different conformations of membrane transporters of the MFS 426 (not shown). Although the difference in number of lipids is small (plus or minus three), the 427 projections in figure 9 show that the conformational changes require significant lateral displacement 428 of lipids in both inner-and outer-leaflet. Hence, the degree of acyl chain unsaturation and low 429 ergosterol concentration is in line with the flexibility that is needed for the conformational changes.  transporter Lyp1 reveals the requirements of the protein for specific lipids, and the optimal 453 conditions require a stringent degree of acyl chain saturation, and amounts of anionic lipids (>15 454 mol%), non-bilayer lipids (>10 mol%) and ergosterol (5 mol%), which match the observed lipids in 455 the HPCL-MS analysis. Except for phosphatidylserine we do not find a strong effect of a specific lipid 456 headgroup on the activity of Lyp1. 457 458 When we benchmark our S. cerevisiae lipidomic data to other studies of the yeast plasma membrane 459 we find similar, small quantities of PA, CL and PG(39-42). Thus, the majority of protein is extracted 460 from the plasma membrane rather than from internal membranes. In terms of selectivity, we benefit 461 from the ability to trap Can1 and Lyp1 in MCC/eisosomes, taking advantage of the GFP-binding 462 protein(22) and the fact that we purify proteins using an affinity tag. Thus, our approach for lipid 463 analysis is much less hampered by contamination with internal membranes than conventional 464 fractionation studies. The SMALP technology only identifies protein specific lipidomes and does not 465 report the overall composition of the plasma membrane. 466 467 Reported values for the average degree of acyl chain unsaturation for the plasma membrane of 468 yeast vary. The values we find in our SMALPs are consistent with(41), but somewhat higher than 469 found by others using non-SMALP methods (39,40,43,44). We detected the sphingolipids IPC and 470 MIPC and expected to also find M(IP) 2 C as major lipid species (41), but our M(IP) 2 C signal detection 471 was poor in the overall (plasma) membrane. We therefore do not provide specific conclusions about 472 the quantities of M(IP) 2 C in our SMALP samples. We find similar amounts of ergosterol in MCC and 473 MCP by mass spectrometry analysis, whereas filipin staining in the literature has suggested that 474 ergosterol is enriched in MCC/eisosomes(45). Here both the lipidomics data and filipin staining are 475 consistent with one another(46), but not with prior observations(16,17). 476 477 We find that 30 mol% of ergosterol, corresponding to the overall concentration of this molecule in 478 the plasma membrane, reduces the activity of Lyp1 (Fig. 7A). The observed 3-5 mol% of ergosterol in 479 SMALP-Lyp1 is more in line with a high transport activity. Detailed analysis of the effects of lipids on 480 the activity and regulation of membrane transport is limited to studies on bacterial transporters 481 such as the lactose-proton symporter LacY(47), a leucine-proton symporter(48), the ATP-driven 482 betaine transporter OpuA(49) and other membrane-associated proteins(50,51). Each of these 483 systems requires anionic (PG) and non-bilayer (PE) lipids, and function optimally in lipids with 484 dioleoyl chains. We find that Lyp1 performs poorly in dioleoyl-sn-phosphatidyl-based lipids and has 485 an order of magnitude higher activity in palmitoyl-oleoyl-sn-phosphatidyl lipids. 486 487 Estimates of global membrane order, using fluorescence lifetime decay measurements and mutants 488 defective in sphingolipid or ergosterol synthesis, suggest that the yeast plasma membrane harbors 489 highly-ordered domains enriched in sphingholipids(10,52). This conclusion is consistent with the 490 observation that the lateral diffusion of membrane proteins in the plasma membrane of yeast is 3-491 orders of magnitude slower than has been observed for membranes in the liquid-disordered 492 state(8,13). Accordingly, the passive permeability of the yeast plasma membrane for weak acids is 493 orders of magnitude lower than in bacteria(9). A membrane in the gel or highly liquid-ordered state 494 may provide low leakiness that is needed for fungi to strive in environments of low pH and/or high 495 alcohol concentration, but it is not compatible with the dynamics of known membrane transporters, 496 which undergo relatively large conformational changes when transiting from an outward-to inward-497 facing conformation. In fact, we are not aware of any transporter that is functional when it is 498 embedded in a membrane in the liquid-ordered state. 499 500 The large, ordered lipid domains observed previously(10) might represent the MCP because Pma1 is 501 enriched for sphingolipids relative to Sur7 in MCC. Furthermore, ergosterol is depleted from the 502 periprotein lipidome of the Sur7, Can1, Lyp1 and Pma1 proteins and is 6-fold enriched in the 503 surrounding bulk membranes. Sterols are known to increase the lipid order and interact more 504 strongly with saturated than unsaturated lipids, therefore depletion of ergosterol from the 505 periprotein lipidome is expected to decrease the lipid order. We propose that large parts of the 506 yeast plasma membrane are in the liquid-ordered state but that individual proteins are surrounded 507 by one or two layers of lipids with at least one unsaturated acyl chain to enable sufficient 508 conformational dynamics of the proteins, which might be needed for transporters like 509 Lyp1 (35,53,54). The proteins in these membrane domains would diffuse slowly because they are 510 embedded in an environment with a high lipid order. 511 512 Methods 513 514 Yeast strains and plasmids 515 Saccharomyces cerevisiae strains (Table S1) are derived from Σ1278b (from Bruno Andre (55) washed with 10mL Tris-HCl pH7.5, and SMALPs were eluted in 3 sequential steps using 1mL 50 mM 577 Tris-HCl, 50 mM Imidazole pH7.5 with 10 minutes incubation between each elution. All elution 578 fractions were pooled and concentrated to 500 µL, using a 100 kDa spin concentrator (Merck, DE). 579 Next, the sample was applied onto a size-exclusion chromatography column (Superdex 200 increase  Lipid extraction and mass spectrometry 596 Lipid extraction from the SMALPs and the crude membranes was performed based on the Bligh and 597 Dyer method(25). The lower organic phase was separated from the upper aqueous phase and dried 598 under a nitrogen stream. The extracted lipid residue was re-dissolved in the starting mobile phase A 599 and the injection volume was 10 µl for each HPLC-MS run. The injection concentration for lipid 600 extracts from SMALPs was normalized to 1 µM based on input protein concentration. The injection 601 amount for SMA polymer control was 0.02 µg. The injection amounts for the crude membranes were 602 0.05, 0.25, 1, 2.5, or 10 µg depending on the application. The samples were run on an Agilent 603 Poroshell 120 A, EC-C18, 3 x 50 mm, 1.9 µm reversed phase column equipped with an Agilent EC-604 C18, 3 x 5 mm, 2.7 µm guard column and analyzed using Agilent 6530 Accurate-Mass Q-ToF/ 1260 605 series HPLC instrument. The mobile phases were (A) 2 mM ammonium-formate in methanol /water 606 (95/5; V/V) and (B) 3 mM ammonium formate in 1-propanol/cyclohexane/water (90/10/0.1; v/v/v). 607 In a 20-minute run, the solvent gradient changes as follows: 0-4 min, 100% A; 4-10 min, from 100% A 608 to 100% B; 10-15 min, 100%B; 15-16 min, from 100% B to 100% A; 16-20 min, 100% A. For the 609 lipidomic analysis, three independently purified SMALP-Pma1 (MCP) or SMALP-Sur7 (MCC) 610 complexes were analyzed and compared to the SMA polymers alone. Data were analyzed using Mass 611 Hunter (Agilent) and R package XCMS(59) for lipidomic peak analyses and in house designed 612 software methods(26). 613 614 For the mass spectrometry belonging to figure S4: 10ul of the lipid extraction was injected on a 615 hydrophilic interaction liquid chromatography (HILIC) column (2.6 μm HILIC 100 Å, 50 × 4.6 mm, 616 Phenomenex, Torrance, CA). The mobile phases were (A) acetonitrile/acetone (9:1, v/v), 0.1% formic 617 acid and (B) acetonitrile/H2O (7:3, v/v), 10mM ammonium formate, 0.1% formic acid. In a 6.5-618 minute run, the solvent gradient changes as follows: 0-1 min, from 100% A to 50% A and B; 1-3 min, 619 stay 50% B; 3-3.1 min, 100% B; 3.1-4 min, stay 100% B; 4-6.5 min from 100% B to 100% A. Flowrate 620 was 1.0mL/min The column outlet of the LC system (Accela/Surveyor; Thermo) was connected to a 621 heated electrospray ionization (HESI) source of mass spectrometer (LTQ Orbitrap XL; Thermo) 622 operated in negative mode. Capillary temperature was set to 350°C, and the ionization voltage to 4.0 623 kV. High resolution spectra were collected with the Orbitrap from m/z 350-1750 at resolution 60,000 624 (1.15 scans/second). After conversion to mzXML data were analyzed using XCMS version 3.6.1 625 running under R version 3.6.1. 626 627 To estimate the lipid quantity, the lipid concentration was obtained by external standard curve 628 fitting. Briefly, a series of concentrations of synthetic standard were prepared and analyzed by HPLC-629 MS to determine the response factor and degree of linearity of input lipid to count values. The ion 630 chromatogram peak areas from known concentrations were used to generate the standard curves 631 for determining the unknown concentrations of the extracted lipids. For key applications where the 632 highest quality lipid quantification was needed or when input lipid mixtures were complex, we used 633 internal standards for authentic lipids and the method of standard addition. For phospholipids, 0.25 634 µg crude membrane samples were spiked with a series of known concentrations (0, 0.5, 1.0, 1.5, and 635 2.0 pmol/µl) of synthetic molecules for C34:1 PC, C34:1 PE, C34:1 PS, or C34:1 PI and subjected to 636 HPLC-MS negative ion mode analysis. For ergosterol, 0.05 µg crude membrane samples were spiked 637 with a series of known concentrations of the synthetic standard (0, 1 pmol/µl, 3 pmol/µl, and 5 638 pmol/µl), and the data were acquired in the positive ion mode. The ion chromatogram peak areas of 639 the specific m/z values were plotted against the concentrations of the spiked synthetic standards to 640 extrapolate concentration of natural lipids on the X-axis. 641 642 Liposome formation 643 Phospholipids were obtained from (Avanti polar lipids Inc, Alabaster, AL, USA), and brassicasterol 644 (CarboSynth, UK), 7-dehydrocholesterol (Sigma Aldrich, DE), cholesterol and ergosterol (Sigma 645 Aldrich, DE) were obtained from the indicated vendor. Lipids were dissolved in chloroform and 646 mixed at desired ratios (mol%) to a total weight of 10 mg in a 5 mL glass round bottom flask. 647 Chloroform was removed by evaporation at 40°C and an applied pressure (p) of 370 mBar, using a 648 rotary vaporizer (rotavapor r-3-BUCHI). The resulting lipid film was resuspended in 1 mL of 649 diethylether and the previous step was repeated at atmospheric pressure until a dry film was visible. 650 To remove any residual solvent a pressure of 10 mBar was applied for 20 minutes. The obtained lipid 651 film was hydrated in 1mL of 50 mM NH 4 Acetate pH7.5, 50 mM NaCl by shaking for 5 min and then 652 transferred to a plastic tube compatible with sonication. The lipid suspension was homogenized by 653 tip sonication with a Sonics Vibra Cell sonicator (Sonics & Materials Inc.) at amplitude of 70% for 2 654 minutes with 5 sec pulses and 5 sec pauses between each pulse. The sample was kept at 4 o C in 655 ethanol:water:ice (25:25:50 v/v/v). The lipid suspension (1 mL aliquot at 10 mg of lipid/mL) was 656 transferred to a 1.5 mL Eppendorf tube, snap frozen in liquid nitrogen and thawed at 40 o C. This 657 process was repeated 4 times and stored in liquid nitrogen until further use. 658 659 Proteoliposome preparation 660 Protein purification and membrane reconstitution was performed as described (31). Briefly, n-661 Dodecylmaltoside (DDM) was used to solubilize Lyp1-GFP. The protein was purified by Immobilized 662 Metal Affinity Chromatography (IMAC, using Nickel-Sepharose) and Fluorescence Size-Exclusion 663 Chromatography (FSEC, using a Superdex 200 increase 300/10 GL column attached to an Åkta 900 664 chromatography system (Amersham Bioscience, SE) with in-line Fluorescence detector (1260  665 Infinity, Agilent technologies, US). In parallel, liposomes were thawed at room temperature and 666 subsequently homogenized by 11 extrusions through a 400nm polycarbonate filter (Avestin Europe 667 GMBH, Ger). Next, liposomes were destabilized using Triton X-100 and titrated to a point beyond the 668 saturation point (Rsat) as described (60); the final turbidity at 540 nm was at approximately 60% of 669 Rsat. Purified Lyp1 was mixed with triton x-100-destabilized liposomes of the appropriate lipid 670 composition at a protein-to-lipid ratio of 1:400 and incubated for 15 min under slow agitation at 4°C. 671 Bio-beads SM-200 (Biorad, Hercules, Ca, USA), 100 mg/0.4% Triton X-100 (final concentration of 672 Triton x-100 after solubilization of the liposomes) were sequentially added at 15, 30 and 60 min. The 673 final mixture was incubated overnight, after which a final batch of bio-beads was added and 674 incubation continued for another 2 hours. Protein containing liposomes (proteo-liposomes) were 675 separated from the Bio-beads by filtration on a column followed by ultracentrifugation 444,000 x g 676 at 4°C for 35 min. Proteo-liposomes were suspended in 10 mM potassium-phosphate plus 100 mM 677 potassium-acetate pH 6.0, snap frozen and stored in liquid nitrogen. 678 679 In vitro transport assays 680 Transport assays were performed and the formation of a proton gradient (ΔpH) and membrane 681 potential (ΔΨ) was established as described in (31) Data analysis and transport rates 688 The data were analyzed in Origin (OriginLab, MA). The initial rates of transport were determined 689 from the slope of progress curves (e.g. Supplementary Figure S11B). ) . For this a polar coordinate system from 699 the center of the protein was established (n=80) resulting in an angle (θ) of 4.5° between each 700 coordinate. The distance (r) was determined from the most distant atom between two polar 701 coordinates. This distance was applied at ½ θ. From distance r on ½ θ , a perpendicular line was 702 drawn (b) resulting in a right-angled triangle for which the surface area can be calculated using the 703 tangent, ½ θ and r. The area was calculated for each of the 80 polar coordinates and summated to 704 acquire the total surface area of the protein.          (C) Lipid concentrations were also quantified using internal standards by the method of standard addition. Briefly, the lipid extracts were spiked with known concentrations of the synthetic standard, which has the same total fatty acyl chain length and number of unsaturation as compared to the lipid of interest. The ion chromatogram peak areas were plotted against the concentrations of the synthetic standard and extrapolated to the X-axis. (D) The conversion factors were defined as the ratio of the estimated quantity obtained from the external standard fitting to that from the standard addition method. The conversion factors are 0.71, 0.79, 0.63, 0.35, and 0.73 for PI, PC, PE, PS, and ergosterol, respectively. Final mass values are reported according to the estimated amount from the internal standard curve for one lead lipid in each class. The data are presented as mean ± standard deviation of triplicate measurements (SD).  Figure S11: Generation of proton motive force and lysine transport progress curve. A. Schematic showing the generation of a membrane potential (ΔΨ, red arrow) by a valinomycin-mediated potassium diffusion potential and pH gradient (ΔpH, green) formation by an acetate diffusion potential. Together the ΔΨ and ΔpH form the proton motive force (PMF=ΔΨ-ZΔpH, where Z equals 2.3RT/F and R and F are the gas and Faraday constant, respectively, and T is the absolute temperature. B. Transport of lysine by Lyp1-GFP-containing proteoliposomes and data fitting. Lyp1 activity is obtained from the slopes of such lines and converting the rates of transport into turnover numbers The proton influx was calculated from the slope of a linear fit of the data, where dashed lines in Panel A indicate the start and end points used for the fit. Result is representative of three experiments, and error bars are the standard error of the fit.