The toxicity of an “artificial” amyloid is related to how it interacts with membranes

Abstract

Despite intensive research into how amyloid structures can impair cellular viability, the molecular nature of these toxic species and the cellular mechanisms involved are not clearly defined and may differ from one disease to another. We systematically analyzed, in Saccharomyces cerevisiae, genes that increase the toxicity of an amyloid (M8), previously selected in yeast on the sole basis of its cellular toxicity (and consequently qualified as “artificial”). This genomic screening identified the Vps-C HOPS (homotypic vacuole fusion and protein sorting) complex as a key-player in amyloid toxicity. This finding led us to analyze further the phenotype induced by M8 expression. M8-expressing cells displayed an identical phenotype to vps mutants in terms of endocytosis, vacuolar morphology and salt sensitivity. The direct and specific interaction between M8 and lipids reinforces the role of membrane formation in toxicity due to M8. Together these findings suggest a model in which amyloid toxicity results from membrane fission.

Introduction

Amyloid proteins form a broad group of proteins that share the ability to display a particular quaternary structure: amyloid fibrils. Although unrelated in terms of their primary sequence and function, these proteins share common structural characteristics, such as fibrillar polypeptide aggregates with cross-beta protein conformations. Most studies have converged on amyloid proteins related to various diseases.1 Indeed, the pathological processes associated with Parkinson, Alzheimer or prion diseases are due to the aggregation of α-synuclein (α-syn), Amyloid-β (Aβ) or Prp proteins, respectively. However, amyloids are also used for normal cellular functions.2 This is the case in unicellular cells (in E. coli and S. cerevisiae) in which amyloid proteins allow the fixation of cells onto a glass surface and the formation of biofilms, but amyloid proteins may also form prions that can help the cells to survive in particular conditions.3–6 This is also the case in metazoans with Pmel17, which is involved in mammalian skin pigmentation.7 Also, several endocrine hormones appear to be stored in secretory granules in an amyloid-like state.8

Interestingly, both toxic and non-toxic amyloids share common or closely related oligomeric species. These oligomeric species are produced early during polymerization kinetics and are recognized by the same antibodies. These antibodies are directed against artificially made oligomers, synthesized by coupling an Aβ peptide to the surface of colloidal gold beads via its C-terminus. The assembly of toxic and immunoreactive species reaches a maximal level at the same time during polymerization, leading to the conclusion that these oligomeric species affect cellular viability.9,10 The link between the formation of toxic species and its structural requirements remains poorly understood. This is due in part to the experimental approaches used (typically based on adding large concentrations of proteins to cell cultures), as they do not represent the complexity of amyloid cellular toxicity accurately. We recently set-up a genetic screen based on generating toxic amyloids from an originally harmless amyloid.11 Our model used the yeast S. cerevisiae to monitor the cellular toxicity of amyloids. Although the budding yeast does not allow modeling of complex neuronal diseases, such as Parkinson's or Alzheimer's, it does however pinpoint the role of factors involved in this process. For instance, eight of nine toxicity modifiers identified in yeast had similar effects on α-synuclein toxicity in yeast and in neuronal systems.12 This model may therefore represent an attractive alternative model for studying amyloid toxicity.

The harmless amyloid peptide used in this study was the prion domain of the Het-s protein from P. anserina. The more toxic mutant (Mutant 8, M8) differs from the wild-type by ten mutations. In vivo, this mutant forms smaller dotted aggregates, which differ from the large annular wild-type (WT) aggregates. We then biochemically and biophysically characterized the structural characteristics and differences in this toxic M8 mutant.13 In vitro, the fibers formed by WT and M8 amyloids differ significantly. WT amyloids exhibit typical µm-long fibers when observed through transmission electronic microscopy, whereas M8 amyloids assemble into 80 nm-long filaments. Wide Angle X-ray Scattering (WAXS) and ATR-FTIR spectroscopy confirmed that M8 is also an amyloid, but with a totally different secondary structure based on antiparallel β-sheets. However, one major question continues to persist: what makes an amyloid toxic?

Here, we have used a library of yeast gene deletion strains, known as the Euroscarf Library.14 This yeast KO library has been successfully used to identify genes modulating cellular toxicity of Huntingtin with poly-glutamine (poly-Q) expansions and α-synuclein.15 This collection contains 4,850 viable mutant haploid strains, each lacking a single gene. These strains were manually and individually transformed with constructs allowing conditional expression of the previously isolated toxic mutant M8. From these strains, we isolated 46 gene deletions that enhance M8 toxicity. Interestingly, the isolated genes were different from those previously identified as modulators of α-synuclein or poly-Q expressions. This set of isolated genes clearly highlights the role of traffic vesicles as a key factor in cellular toxicity. High M8 affinity for DMPC (1,2-dimyristoyl-sn-glycero-3-phosphocholine) and changes in cellular traffic due to its expression led us to suggest a model in which M8 toxicity occurs due to a greater propensity for it to be inserted into lipid layers, leading to vesicular trafficking “poisoning,” through a vesicular fission mechanism.

Results

Yeast knockout library screening reveals few toxicity enhancers.

Selecting deleted strains in which M8 toxicity was enhanced involved three steps. First, all strains from an ordered and frozen library were transformed using the microtiter plate transformation method.16 This method was slightly tailored to the materials in our lab (see Materials and Methods).

Isolates of interest were those in which growth defects were specifically correlated with M8 expression. These clones were separated from false positive strains in which the growth defect may be due to the deletion itself or the transformation process. Thus, after the transformation step, we plated yeasts on dextrose medium for plasmid selection and non-selective galactose medium. The dextrose medium ensured the selection of transformed strains and prevented M8 expression. By contrast, the cell suspension grown on non-selective galactose medium was mostly populated with untransformed cells not expressing M8. The comparison between these two conditions allowed us to categorize the Euroscarf strains into three groups (Fig. S1).

Strains unable to grow on galactose medium but forming colonies on the dextrose plate. These strains were unable to grow on medium in which galactose was the sole carbon source. These strains were classified as false positives (as growth inhibition was not due to M8 but due to their incapacity to grow under the experimental conditions outlined). The 271 (Gal-) strains identified are listed in Supplementary Table 1.

Strains unable to grow on both dextrose or galactose media. These were either auxotrophic strains or strains for which the microtiter transformation protocol was ineffective. These strains were then individually re-transformed via a standard acetate lithium protocol.17 Remaining non-transformable strains (auxotrophic or unable to grow on our minimal media) were not analyzed in this study. These remaining 113 strains are listed in Supplementary Table 2.

Strains able to grow on both dextrose or galactose media. These strains were analyzed and screened for M8 toxicity enhancers. At the end of this selection process, 72 strong toxicity enhancers were found for M8. One drawback to the yeast deletion library is that some strains potentially share an additional unexpected mutation.18 As the same putative mutation cannot be present in a strain bearing the same KO but genetically independent from the previous library, we have checked for the toxicity phenotype in the opposite mating type (Mat a) strains obtained independently. We also tested for toxicity specificity using WT Het-s amyloid and finally isolated only 46 of the selected strains as strong M8 toxicity enhancers (Suppl. Table 3). Among these 46 strains, three also increase WT toxicity. Forty-four of these 46 mutants were completely different from those previously identified as a-synuclein and poly-glutamine toxicity enhancers,15 suggesting alternative toxicity pathways for M8 toxicity. We further investigated the mechanism underlying M8 toxicity by classifying the identified genes into functional categories.

Gene ontology classification highlights the role of the vacuole.

Identified genes were classified using the FunSpec algorithm,19 located at the Munich Information Center for Protein Sequences (MIPS) comprehensive yeast genome database and the Gene Ontology (GO) database. These classification methods helped us to identify possible clusters of genes involved in the same mechanism. In our case, it highlighted a group of genes involved in cellular traffic, which were strongly linked to vacuolar pathways (Table 1). Thirty percent of proteins encoded by the selected genes were localized in membranes, mitochondria and vacuoles, and were functionally involved in membrane organization and vesicle-mediated transport. Among these identified genes, three genes (PEP3, PEP5 and VPS16) belong to the class C VPS family. The Vps-C complexes contain the CORVET (class C core vacuole/endosome tethering) and HOPS (homotypic vacuole fusion and protein sorting) complexes. Both are found in endosomes and vacuoles, in which they promote membrane fusion. These complexes are built on a common Vps-C core complex formed by the products of four genes (PEP3 (VPS18), VPS16, PEP5 (VPS11 or END1) and VPS33 (SLP1)). The YPT7 gene is also involved in vacuolar function, but is not part of the Vps-C core complex. We thus characterized, in more detail, the reasons for which M8 expression is more deleterious in these mutants.

M8 selectively inhibits the growth of class C vps mutants.

Defective growth, scored by serial dilution, did not allow the identification of whether the changes occurred during the lag, exponential or stationary phases. We analyzed the toxic effects of M8 in class C vps-deleted strains more precisely; thus, we checked their growth against similar strains expressing WT amyloid or GFP alone (Fig. 1). BY4742 cells expressing WT or GFP reached the stationary phase 48 h after induction. Cells expressing M8 grew more slowly, but reached the same density after 96 h of induction (A). The stationary phase was reached later in pep3, pep5 and vps16 class C vps mutants (D–F) expressing M8, and the plateau was lower, indicating a higher mortality for the yeast cells.

Class C vps-deleted strains exhibited slower growth than wild type strains. Thus, their behavior in the presence of M8 may have been due to the combination of two other growth-slowing factors. We then investigated how other KO strains are affected by WT, M8 or GFP. These other KO strains (cpr7Δ and adk1Δ strains20,21) were selected for their decreased rate of growth during the exponential phase. These strains have a slower growth, but reached the same plateau (stationary phase) whichever protein was expressed (WT, M8 or GFP) (Fig. 1B and C). Thus, toxicity enhancement observed in class C vps-deleted strains induced by M8 is specifically related to the production of M8, and does not result from other unconnected effects that slow growth. One trivial explanation for increasing M8 toxicity may be the capacity for vacuolar mutants to stabilize M8, thus leading to increasing its concentration up to a deleterious threshold level. However, western-blot analysis of M8 expressed in PEP3Δ, PEP5Δ or VPS16Δ does not sustain this hypothesis (Suppl. Fig 2). We then studied how M8 affects vacuolar protein sorting, as mutations in this cellular process increase its toxicity.

M8 interacts with lipids and affects cellular trafficking.

We investigated whether the presence of M8 results in a phenotype mimicking pep3Δ, pep5Δ and vps16Δ phenotypes. These mutants exhibit sensitivity to several metals or salts,22 present some modifications of their vacuolar morphology, and display some defects in vesicular traffic via an alteration of endocytosis and cargo delivery to the vacuole.23

Sensitivity to metals and salts.

We first explored the sensitivity of BY4742 cells expressing GFP, WT or M8 to LiCl, CaCl₂, ZnCl₂, MgCl₂, MnCl₂ and citric acid (Fig. 2). Growth was slower in all strains when plated on medium containing these molecules, but the presence of M8 did indeed increase the sensitivity of the yeast to these salts and metals. As control, we have verified that the three strains pep3Δ, pep5Δ and vps16Δ expressing GFP or WT are sensitive to these molecules (Fig. 2B). It is difficult to measure the extent of salt sensitivity of M8 expression in BY4742 vs. deleted strains since without any chemical compounds these strains do not grew in the same way. However, the sensitivity of strains expressing M8 or bearing deletions appears to be roughly the same.

Vacuolar morphology.

We then analyzed vacuolar morphology by staining cells with lipophilic dye FM4–64. This vital fluorescent dye initially stains the plasma membrane and is then internalized and delivered to the vacuolar membrane.24 Cells were incubated with FM4–64 for 10 min, washed, and subsequently incubated with FM4–64 for up to 45 min at room temperature. The presence of GFP or WT did not disturb vacuolar morphology (Fig. 3A and B). By contrast, several vesicles were placed side by side in M8 cells, suggesting that M8 probably triggers vacuole fragmentation (Fig. 3C).

Lipid binding.

As amyloid toxicity is often related to the ability of these proteins to interact directly with membrane25 we then analyzed the ability of M8 to directly bind to phosphatidylcholine (PC). The purified M8 peptide binds to DMPC (a representative component of PC in yeast membranes)26 (Fig. 4). Under similar experimental conditions, WT was not found to be associated with this phospsholipid, as it was not detected on the filter with the fatty acid. This absence of detection was independent of the method used (direct observation with Ponceau red or immunodection). M8 binds to phospholipids, and thus we also investigated its ability to modify membrane traffic.

Vesicular trafficking.

We monitored endocytic traffic by measuring the transport of 5-fluoro-uracil (5-FU), a toxic analog of uracil that is imported by the permease Fur4p. In wild-type cells, Fur4p is continually removed from the plasma membrane by endocytosis. If endocytosis slows down or is blocked, Fur4p accumulates at the plasma membrane, and allows a significant amount of uracil or 5-FU into the cell, leading to its death.27 We performed spotting assays by plating serial dilutions of BY4742 cells expressing GFP, WT and M8 onto two types of media: containing 5-FU or not (Fig. 5A). After several days, whereas M8 expressing cells no longer display any toxicity phenotype as they reached the stationary plateau on a plate without 5-FU, cells producing M8 and grown on a 5-FU containing medium still display a toxicity phenotype due to their increased sensitivity to 5-FU, indicating a disturbance of endocytosis.

We then investigated the traffic between the Golgi and vacuole. Resident proteins of the vacuole are synthesized and translocated from the cytosol into the lumen of the ER or inserted into the ER membrane. They then move to the Golgi. Two pathways mediate transport between the late Golgi and the vacuole: vacuolar hydrolases, such as carboxypeptidase Y (CPY), are delivered to the vacuole via an endosome-like compartment, but alkaline phosphatases (ALP) bypass the endosome and are directly delivered to the vacuole.28 The intracellular trafficking of proteins along these pathways is accompanied by post-translational modification, glycosylation and proteolytic processing.29 CPY is synthesized as a precursor that is translocated into the lumen of the ER, where it undergoes signal peptide cleavage and N-linked core-glycosylation generating precursor 1CPY (p1:67 kDa). In the Golgi, CPY undergoes additional glycosylation, generating p2CPY (69 kDa). CPY is finally transported via late endosomes to the vacuole, where it is processed to give the 61 kDa mature species (mCPY).29 As expected, vps mutants accumulate p2 CPY (Fig. 5B), and this species is secreted into the medium where it is detected by colony immunoblotting (data not shown). By contrast, we detected no defect in CPY processing in M8 cells: p2CPY does not accumulate and is not secreted in the medium (Fig. 5B). Similar results were obtained during studies of ALP processing (data not shown). Thus, M8 does not prevent the sorting of proteins and their transfer to the vacuole.

Discussion

We have previously carried out a structure-toxicity analysis of an amyloid protein that is well characterized at the biochemical level.30 We aimed to isolate mutants on the basis of their toxicity and to compare their properties with those of the WT amyloid. This unbiased screen isolated a mutant containing 10 single mutations. Indeed, this mutant M8 is quite different from the WT, both in vivo and in vitro. M8 forms smaller aggregates and displays rapid polymerization, involving highly structured intermediates.31 Biochemical characterization is only part of the answer, as information on toxicity results from both biochemical and cellular data. Here, we used a whole-genome approach to determine the mutations that modulate in trans the toxicity due to M8 expression. The yeast deletion library screening procedure isolated 46 KO mutants, of which two were previously shown to be involved in amyloid toxicity. One of these mutants (yol049WΔ) displays a partial loss of viability on the overproduction of exon 1 of human HD protein.15 The corresponding gene encodes for a glutathione synthetase carrying out glutathione synthesis from γ-glutamylcysteine and glycine.32 The corresponding GSH-deficient mutant dies rapidly if the cells are directly exposed to lethal temperatures.33 Thus, this gene may have a general role in protection and may not be specific to amyloid toxicity. The yol108CΔ strain has already being identified as a modulator of α-synuclein toxicity.15 This gene encodes a transcriptional activator of the ENO1 gene,34 an enolase required for vacuole fusion and protein transport to the vacuole.35 Interestingly, oxidative inactivation of human ENO1 (the corresponding ortholog) may lead to the development of Alzheimer disease.36 The link with the vacuole was reinforced by the presence of 3 of the 4 members of the Vps-C core complex. As previously described, knocking out these genes leads to the complete absence of the vacuole.23 This absence may be in part responsible for an M8 toxicity increase, through the relocation of the toxic amyloid to another compartment in which it could have a deleterious effect on the cell. This suggestion is, however, unlikely, as M8 does not appear to be specifically localized to the vacuole.11 Moreover, the identification of YPT7 (a Rab-GTPase involved in binding the Vps-C HOPS complex to the vacuole membrane37) as a modulator of M8 toxicity argues for a role of the Vps-C complex that is independent to the presence of the vacuole. The vacuole is highly fragmented in ypt7Δ cells, and we found a similar phenotype if M8 was expressed in yeast cells. This phenotype indicates that M8 has the capacity to interfere in the equilibrium between fusion and fission of vacuolar vesicles, but does not elicit its toxic effect through vacuolar function alone.

M8 expression failed to block the traffic between the Golgi apparatus and the vacuole, including cargo delivery to the vacuole, as carboxypeptidase Y and alkaline phosphatase are efficiently processed in the presence of M8. Interestingly, M8 impaired but did not suppress endocytosis, as FM4–64 was still internalized in yeast cells. Thus, the presence of M8 in the cell triggers some of the phenotypes observed in class C vps-deleted strains (increased sensitivity to metals or salts and impairment of endocytic traffic), but this effect was significantly attenuated. The synergetic effects between M8 and class-C vps mutants raise the possibility, in addition to their involvement in endosome-to-vacuole transport, that M8 and the class-C VPS complex play additional roles at other transport steps. This suggestion is consistent with the role of the class-C VPS complex in the processes of membrane docking and fusion at both the Golgi-to-endosome and endosome-to-vacuole stages of transport.38 M8 affinity for lipids may lead directly to vesicle fission. In our model (Fig. 6), M8 antagonizes the fusion function of the Vps-C HOPS complex. In the absence of this complex, small vesicle formation is favored, leading to significantly greater growth impairment. The target vesicles may be vacuolar vesicles or other endocytosis vesicles. Mitochondria were excluded, as M8 did not affect mitochondrial networks and its toxicity remained the same in rho° cells (Suppl. Fig. 1). This mechanism echoes the recent finding on α-synuclein toxicity in which a genome-wide analysis has linked its toxicity in yeast to endocytosis of the protein and vacuolar fusion defects.39 It may also be correlated to the disruption of endocytosis which was earlier identified as one of the causes for cellular toxicity induced by aggregated poly-glutamines.40 Our model is also strengthened by a recent study based on in vitro experiments linking disruption of liposome membranes and loss of cell viability to amyloid fibril length.41 Unusual M8 amyloid characteristics (short fibrils of 80 nm), together with the presence of oligomeric intermediates, are consistent with a model in which toxicity results from membrane disruption. Our findings, based on an unbiased screen to isolate a new and artificial toxic amyloid and to search for genes modulating its toxicity through trans interactions, showed that membrane fusion/fission is an essential phenomenon in amyloid toxicity and offers a powerful model for studying this phenomenon at the molecular level.

Materials and Methods

Yeast strains, media and plasmids.

Yeast strains used are isogenic to BY4742 strain (MATα, his3D1, leu2D0, ura3D0). All deletion strains were from the Euroscarf yeast deletion library [14]:

• adk1Δ (MATα, his3Δ1, leu2Δ0,ura3Δ0, YDR226W::kanMX4)

• cpr7Δ (MATα, his3Δ1, leu2Δ0,ura3Δ0, YJR032W::kanMX4)

• pep3Δ (MATα, his3Δ1, leu2Δ0,ura3Δ0, YLR148W::kanMX4),

• pep5Δ (MATα, his3Δ1,leu2Δ0, ura3Δ0, YMR231W::kanMX4)

• vps16Δ (MATα,his3Δ1, leu2Δ0, ura3Δ0, YPL045W::kanMX4).

As specified, yeasts were grown in SD medium (0.67% yeast nitrogen base, 2% dextrose) or SG medium (0.67% yeast nitrogen base, 2% galactose) supplemented with 20 mg/L histidine (H), 20 mg/L lysine (K) and 60 mg/L leucine (L) or 0.67% casaminoacids. We used multicopy (2 µ) yeast expression plasmids with the URA3 selectable marker in this study: they include pYeYGFP2U (GFP), pYecHetsYGFP2U (WT), pYecHetsm8YGFP2U (M8).11 The fusion of the HET-s(PrD) and yeast optimized GFP42 is expressed under control of a GAL10 promoter in a pYeHFN2U.43 pYEF2mtRFP allows expression of mitochondrial RFP. It is derived from pYEF1mtRFP,44 which was constructed using PYX-mtGFP.45 The selectable marker was changed from URA3 (pYEF1mtRFP) to LEU2 (pYEF2mtRFP).

Microtiter plate transformations.

We used the microtiter plate transformation method16 with a few modifications. We used 6 µg of carrier DNA (Ozyme) per well. To lower the cross-contamination risks during supernatant elimination, we also covered the sink used to recover the supernatant with paper. We also used a 48-well replicator to plate solutions onto standard Petri dishes. Moreover, we used pipette tips specific for viscous solutions, i.e., with a larger opening (Dutcher), particularly for solutions with PEG.

Spotting assay.

All spotting assays were performed under identical conditions. Tenfold serial dilutions starting with an equal number of cells (1OD; l = 600 nm) were prepared in sterile water. Three independent fresh transformant samples were pooled for the spotting assays. Five-microliter drops were then plated onto SD or SG medium.

Selecting clones of interest.

After a one-night recovery culture on rich dextrose solid medium (YPD), frozen yeasts from the Euroscarf library were screened using the Gietz microtiter plate transformation protocol. A two-step selection was then used to separate clones of interest from false positive candidates. The transformation mixture was plated onto both selective minimal dextrose media and non-selective minimal galactose media. The comparison between these two types of media allowed us to eliminate three false positive families: isolates unable to grow on minimal medium, isolates that did not undergo transformation and isolates that were unable to grow on minimal galactose medium. Selected transformants were then replica plated onto a selective galactose medium to induce M8 expression. Deletion mutants presenting a growth defect were then selected and processed via a serial dilution assay, to analyze their sensitivity to M8 expression more precisely. Thus, all strains remained in their exact original position during the microtiter plate transformation protocol, such that each strain could be identified by its position on the plate. Moreover, selected deletion mutants presenting a growth defect could be easily identified.

Gene ontology classification.

Genes identified during the library screen and confirmed via fresh transformation of Mat yeast strains were processed using FunSpe19 for MIPS and GO classification. An additional GO classification for the function and the localization was also performed (http://db.yeastgenome.org/cgi-bin/GO/goSlimMapper.pl).

FM4–64 labeling.

1 ml of exponentially growing cells was concentrated into 100 µl, placed on ice and incubated in the presence of 20 µM FM 4–64 (Invitrogen) for 15 min. Cells were then washed twice with 1 ml of cold medium, resuspended in 1 ml of medium (t0) and incubated for 45 min at room temperature.

Microscopy.

Cells were washed in water and resuspended in medium. DNA was stained by adding 2 ng/mL Hoechst 33342 to the mounting solution. Cells were observed using a DMLB (Leica) fluorescence microscope coupled with a ColorView II (Olympus) color camera.

Protein extraction and western blotting.

Alkaline lysis extraction was used for protein extraction. Briefly, 5 OD (l = 600 nm) units of yeast cells in exponential growth were permeabilized with 500 µL of 0.185 M NaOH, 0.2% β-mercaptoethanol. After a 10-min incubation on ice, Trichloroacetic acid (TCA) was added to a final concentration of 5%, and the samples were incubated for an additional 10 min on ice. Precipitates were then collected by centrifugation at 14,000 g for 5 min. Pellets were dissolved in 30 µL of dissociation buffer (4% sodium dodecyl sulfate, 0.1 M Tris-HCl pH 6.8, 4 mM EDTA, 20% glycerol, 2% 2-mercaptoethanol and 0.02% bromophenol blue) and 15 µl of 1 M Tris-base. Yeast proteins were incubated for 5 min at 100°C and separated on a 12% SDS-PAGE. Proteins were electrically transferred onto nitrocellulose membranes (Optitran BA-S83, Schleicher & Schuell) in the presence of transfer buffer (39 mM Glycine, 48 mM Tris-base, 2% EtOH and 0.037% SDS) and were probed with monoclonal 0.5 µg/ml anti-CPY antibodies (Molecular Probes). Peroxidase-conjugated anti-mouse antibodies (Sigma) were used as secondary antibodies. Binding was detected with the SuperSignal reagent (Pierce) and the VersaDoc Imaging system (BioRad).

Protein/lipid interactions in vitro.

WT and M8 histidine-tagged proteins were produced as described previously.13 Proteins in urea were desalted in 10 mM HCl pH 2.0. To start polymerisation, native proteins were prepared at concentrations of 20 µM in sterile water and pH was risen to 7.4 by adding PBS buffer (1X final). For dot blots, 15 µl of DMPC (1 mg/ml 1,2-dimyristoyl-sn-glycero-3-phosphocholine; Aventi Polar Lipids, Inc., Alabaster, AL USA) was allowed to dry on a PVDF transfer membrane (HybondTM-P, GE Healthcare Europe GmbH, Orsay, France) for 15 min. The DMPC membranes were then incubated at 37°C for 24 h with 2 ml of 20 µM WT or M8 proteins. Blots were washed twice with PBS and blocked with PBS 5% dried milk (Régilait, Saint Martin Belle Roche France). To detect the proteins bound to blotted DMPC, Ponceau red (Sigma, St. Louis, MO USA), mouse anti-His-tagged antibodies (Amersham, Piscataway, NJ, USA) and SuperSignal^® West Dura Pierce chemoluminescence kits (Thermo Scientific, Perbio Science France SAS, Brebières, France) were used.

Figures and Tables

Figure 1 Viability assays for WT and KO strains. Growth of BY4742 or mutant cells expressing GFP (▴), WT (◆) or M8 (▪). Cells were grown overnight on dextrose medium supplemented with casaminoacids (0.67%). When cells reached the exponential phase, they were transferred into galactose medium supplemented with casaminoacids (0.05 OD_{600 nm}/ml). Cell concentrations were evaluated by OD_{600 nm} at various times after the induction over a period of 120 h.

Figure 2 Metal and salt sensitivity spotting assay. Ten-fold dilutions of exponentially growing cultures of BY4742 cells expressing GFP, WT or M8 were spotted onto SD agar supplemented with casaminoacids (0.67%) and containing 8 mM MnCl₂, 100 mM MgCl₂, 25 mM LiCl, 100 mM CaCl₂ or 10 mM citric acid (A). pep3Δ, pep5Δ and vps16Δ expressing GFP or WT were tested in the same conditions (B). The cells were incubated at 30°C for 4 days.

Figure 3 Vacuolar morphology observation by microscopy. Exponentially growing cultures of BY4742 cells expressing GFP, WT or M8 were incubated for 15 min on ice with the fluorescent marker FM 4–64 (20 µM). Then, the cells were washed twice and were incubated for 45 min at 30°C. Cells were mounted in medium and observed with a DMRB microscope with a 100X HCX PL fluotar objective.

Figure 4 M8 interacts specifically with lipids in vitro. M8 interacts with DMPC blotted onto a PVDF membrane. M8 is detected directly by Ponceau red and indirectly by an anti-histidine antibody.

Figure 5 Traffic monitoring. (A) Ten-fold dilutions of exponentially growing cultures of BY4742 cells expressing GFP, WT or M8 were spotted onto SG agar with or without 5 µM 5-FU. Petri dishes were observed after 10 days at 30°C to allow the observation of M8 growth. (B) Exponentially growing cultures of BY4742 or mutant cells expressing GFP or M8 were used to prepare total cell extracts. The samples were run on a 12% SDS-PAGE gel, transferred onto a nitrocellulose membrane and exposed to anti-CPY antibodies. m, mature protein; p, precursor protein.

Figure 6 M8 affects vesicle trafficking. The class C HOPS complex promotes small vesicles, like early endosomes, fusion to the vacuole. M8 may prevent this fusion by coating the vacuole with an hydrophobic layer of proteins. Both M8 expression and the absence of the class C HOPS complex will affects vesicle formation leading to an accumulation of small vesicles and to a significantly greater growth impairment. Target vesicles may be vacuolar vesicles or may be other endocytosis vesicles.

Table 1 MIPS and GO classification of verified M8 toxicity enhancers

MIPS functional classification	p-value	Genes in category from cluster	k	f
Vesicular transport	1.50E-03	PEP3 YPT7 PEP5 VPS16	4	72
Vacuole or lysosome	3.43E-03	PEP3 PEP5 VPS16	3	44
Vacuolar/lysosomal transport	3.99E-03	CHC1 PEP3 YPT7 PEP5 VPS16	5	153
MIPS phenotypes	p-value	Genes in category from cluster	k	f
Other vacuolar mutants	1.99E-05	PEP8 PEP3 YPT7 PEP5 VPS16	5	49
Sporulation efficiency	1.01E-03	UME6 PEP3 PEP5 IRA2 VPS16	5	112
Heat-sensitivity	1.26E-03	MDM10 TPD3 TPS2 CHC1 UBI4 PEP3 END3 VPS16	8	313
GO biological process	p-value	Genes in category from cluster	k	f
Golgi to endosome transport	1.09E-04	PEP3 PEP5 VPS16	3	14
Vacuole fusion, non-autophagic	1.57E-04	PEP3 YPT7 PEP5 VPS16	4	40
Mitotic cell cycle spindle assembly checkpoint	7.35E-04	TPD3 BUB1 BUB3	3	26
Late endosome to vacuole transport	1.24E-03	PEP3 PEP5 VPS16	3	31
Negative regulation of protein metabolic process	1.36E-03	TWF1 BUB1 BUB3	3	32
Vacuolar transport	2.25E-03	PEP8 PEP3 PEP5	3	38
Negative regulation of transcription, mitotic	6.97E-03	UME6	1	1
Positive regulation of meiosis	6.97E-03	UME6	1	1
Regulation of membrane potential	6.97E-03	PMP3	1	1
Sequestering of actin monomers	6.97E-03	TWF1	1	1
Vesicle docking during exocytosis	9.17E-03	PEP3 PEP5	2	21

Identified genes were clustered according to their function, phenotype or implied biological process. k is the number of genes from the input cluster in given category and f the total number of genes in given category.

Acknowledgements

We gratefully acknowledge Johanna Coindreau for her technical help and enthusiasm in searching for KO strains showing changes in M8 and WT sensitivity. This work was supported by grants from the CNRS (Programme Interdisciplinaire—Interface Physique Chimie Biologie—Soutien à la prise de Risque) and from the French Research National Agency ANR grant, project ANR-06-MRAR-011-01 «AMYLOI». K.B. received fellowships from the ANR and the Conseil Régional d'Aquitaine.