Glucose induces rapid changes in the secretome of Saccharomyces cerevisiae

Experimental conditions to study the effects of glucose on protein levels

Glucose has profound effects in regulating proteins levels. For instance, glucose up-regulates Lia1p involved protein synthesis, while down-regulating enzymes involved in gluconeogenesis. There are different ways to examine glucose effects in regulating protein levels. We used wild-type cells grown either in YNB-based media (yeast nitrogen base with amino acids) or YP based media to study glucose effects in up-regulation of Lia1p and down-regulation of Fbp1p (Figure 1A). In Experiment I, wild-type cells expressing Lia1p-GFP were grown in YNB-based media containing 2% glucose for 3d followed by the addition of 2% glucose directly to the existing culture media for 0, 2, and 4 hours. In experiment II, cells were grown in YNB media containing 2% glucose for 3d. Cells were pelleted and resuspended in fresh YNB with 2% glucose for 0, 2, and 4 hours. In experiment III, cells were grown in YP-based media containing 0.5% glucose (YPKG) for 3d followed by the addition of 2% glucose directly to the existing YPKG media for 0, 2, and 4 hours. In experiment IV, cells were grown in YPKG for 3d. Cells were pelleted and resuspended in YPD media containing 2% glucose for 0, 2, and 4 hours. Equal amounts of cells were harvested at each time point, processed and examined for changes in Lia1p and Fbp1p levels (Figure 1A). In experiments I and II where cells grown in YNB-based media, the level of Liap1 was low with an increase at t = 4 hours. Lia1p levels were higher in cells grown in YPKG. However, Lia1p was not up-regulated in experiment III but was up-regulated rapidly in experiment IV. Fbp1p levels were low in cells grown in YNB at t = 0 and did not show significant changes at the 2 and 4 hour time points (experiments I and II). Fbp1p levels were higher in cells grown in YPKG (experiments III and IV). Additionally, Fbp1p levels decreased slowly in experiment III but decreased rapidly in experiment IV. Tpi1p (triose phosphate isomerase) is a glycolytic enzyme and levels of this protein were similar under all four experimental conditions. Because experiment IV produced a faster increase in Lia1p levels and a faster decrease in Fbp1p levels, this condition was used to study glucose effects for most experiments described in this study unless otherwise specified.

To further determine whether or not changes in Lia1p and Fbp1p levels were due to the presence of glucose in the media, wild-type cells expressing Lia1p-GFP were grown in YPKG and transferred to YPD or YP (without added glucose) for 0, 2, and 4 hours (Figure 1B). Again, when cells were transferred from YPKG to YPD, levels of Lia1p increased and levels of Fbp1p decreased (Figure 1B, left panels). However, when cells were transferred from YPKG to YP, Lia1p levels did not increase and Fbp1p levels did not decrease (Figure 1B, right panels). Therefore, the presence of 2% glucose in the YPD media is critical in up-regulating Lia1p and down-regulating Fbp1p.

We next examined whether or not cells grown in YPKG media for 3 days were able to re-grow in YPD media containing 2% glucose (Figure 1C). Wild-type cells were grown in YPKG for 3d to OD₆₀₀ = 4-5/ml. Cells were diluted to OD₆₀₀ = 0.2/ml in YPD containing 2% glucose or YP media without glucose for 0–8 hours and changes in cell densities were measured at OD₆₀₀ using a spectrophotometer. When cells were diluted in YPD, cells increased their densities after a lag. Likewise, when cells were diluted in YPKG, cells densities also increased after a lag. Hence, cells are able to resume growth when diluted in YPD or YPKG. In contrast, there was little increase in density when cells were cultured in YP media without glucose, suggesting that the presence of glucose in the media is required for cells to re-grow.

Fbp1p is re-distributed from the periplasm/extracellular fraction to the cytoplasm when cells are transferred from YPKG to YPD

We next determined the effects of glucose on the distribution of Fbp1p at the ultra-structural level using immuno-TEM. Wild-type cells were grown in YPKG for 3d and transferred to YPD for 30 min. Cells were processed and thin sections of cells were incubated with affinity purified anti-Fbp1p antibodies followed by goat anti-rabbit secondary antibodies conjugated with 10 nm gold particles. The distribution of Fbp1p was then observed by TEM. When cells were grown in YPKG for 3d, substantial amounts of Fbp1p were in the periplasm (Figure 2A, arrow). Because Fbp1p is degraded in the vacuole following glucose addition, extracellular Fbp1 should be internalized upon a transfer or cells from YPKG to YPD. Indeed, Fbp1p was found in intracellular structures that contained clusters of small vesicles at the t = 30 min time point (Figure 2B, arrow). Thus, transferring cells from low to high glucose causes a rapid redistribution of Fbp1p from the periplasm to the cytoplasm, resulting in a rapid decline in Fbp1p levels in the extracellular fraction.

Extraction of extracellular proteins

We next used an extraction procedure to determine whether or not this protocol could detect a decline in Fbp1p levels in the extracellular fraction. This protocol utilizes the combination of reducing agents such as β-mercaptoethanol (βME) or dithiothreitol (DTT) and high pH to release proteins in the extracellular fraction[13, 63]. This method of extraction has been utilized to study the secretion of mammalian galectin-1 expressed in Saccharomyces cerevisiae[64]. A similar method has been used to identify proteins associated with the cell wall in Candida albicans[16, 38].

We first determined whether or not cells that were subjected to the extraction procedure retained the ability to exclude trypan blue from cells (Figure 3A-D). Trypan blue dye permeates dead cells and is therefore an indicator of cell viability. Wild-type cells were grown in YPKG for 3d (t = 0) or transferred to YPD for 30 min (t = 30). Cells were then treated with or without Triton X-100 and incubated with trypan blue for 30 min. Cells were then observed by light microscopy. In t = 0 and t = 30 cells, the dye did not stain cells (Figure 3A and C, left panels). However, when Triton X-100 was added to t = 0 and t = 30 cells followed by incubation with trypan blue, about 60-90% of the cells were stained (Figure 3A and3C, right panels). We also used t = 0 and t = 30 cells that were extracted and then incubated with trypan blue. The majority of extracted cells were not stained (Figure 3B and3D, left panels). In contrast, when Triton X-100 was added to the extracted cells followed by incubation with trypan blue, about 60-90% of the cells were stained (Figure 3B and3D, right panels).

We also examined whether or not cells that were extracted retained the ability to internalize exogenous molecules into the cells. FM 4–64 (FM) is a vital dye that is internalized into live cells and actively transported to the vacuole via the endocytic pathway[65]. Wild-type cells were grown in YPKG for 3d (t = 0) or transferred to glucose for 30 min (t = 30). Cells were treated with or without Triton X-100 and then incubated with FM for 3 hours. The same amounts of cells were also subjected to the extraction protocol and treated in the absence or presence of Triton X-100 followed by incubation with FM for 3 hours. The distribution of FM was then observed using fluorescence microscopy (Figure 3E-3L). In t = 0 and t = 30 non-extracted cells, this dye was transported to the vacuole membrane as ring-like circles (Figure 3E and3G). In contrast, when these cells were treated with Triton X-100 and then incubated with FM, this dye was not on the vacuole membrane (Figure 3F and3H). In t = 0 and t = 30 cells that were extracted and then incubated with FM, this dye was also transported correctly to the vacuole membrane (Figure 3I and3K). However, when extracted cells were treated with Triton X-100 and incubated with FM, this dye was not on the vacuole membrane (Figure 3J and3L). Thus, FM is targeted correctly to the vacuole membrane in both non-extracted and extracted cells. Furthermore, the endocytic pathway is functional in cells that were either grown in YPKG or transferred to YPD for 30 min.

We next used the extraction procedure to examine whether or not this protocol could detect a decline in Fbp1p levels in the extracellular fraction when cells were transferred from YPKG to YPD. Wild-type cells expressing GFP-tagged proteins were grown in YPKG media for 72 hours and aliquoted. The same amounts of cells were then transferred to YPD media containing 2% glucose (Figure 4A) or to YP media containing no glucose (Figure 4B) for 0, 15, and 30 min. Following the extraction procedure, the distribution of GFP-tagged proteins in the intracellular and extracellular fractions was determined by Western blotting with anti-GFP antibodies (Figure 4A). In this study, proteins that were released into the supernatant following the extraction procedure were precipitated with TCA, washed and solubilized in SDS sample buffer to form the extracellular fraction. Proteins from the cell-associated fraction were solubilized in SDS sample buffer and this was called the intracellular (I) fraction in this study. Fbp1p was in the extracellular fraction at t = 0 min. Levels of this protein in the extracellular fraction decreased rapidly by t = 30 min (Figure 4A). Other gluconeogenic enzymes are also degraded in the vacuole in response to glucose re-feeding[62, 66]. Therefore, they should show distribution patterns similar to those observed for Fbp1p. These gluconeogenic enzymes include malate dehydrogenase (Mdh2p which catalyzes the interconversion of malate and oxaloacetate), isocitrate lyase (Icl1p which produces succinate and glyoxylate from isocitrate), phosphoenolpyruvate carboxykinase (Pck1p which coverts oxaloacetate to phosphoenolpyruvate), and malate synthase (Mls1p which synthesizes malate from glyoxylate). Indeed, Mdh2p, Icl1p, Pck1p, and Mls1p were in the extracellular fraction at t = 0 min and their levels were reduced at the t = 30 min time point (Figure 4A). In contrast, when cells were transferred from YPKG to YP, levels of extracellular Fbp1p, Mdh2p, Icl1p, Pck1p, and Mls1p did not decrease (Figure 4B). These results suggest that the decline of these proteins in the extracellular fraction is dependent on the presence of glucose in the media.

Identification of extracellular proteins using iTRAQ

The findings that multiple gluconeogenic enzymes showed reduced levels in the extracellular fraction after a transfer of cells to YPD prompted us to hypothesize that changes in the secretome induced by glucose were not limited to gluconeogenic enzymes. To identify other extracellular proteins that changed levels in cells that were transferred from low to high glucose media, we used the iTRAQ technique that we previously employed to identify proteins in the total proteome that change their relative amounts under the same conditions[55]. Because a significant reduction in protein levels in the extracellular fraction was observed at the t = 30 min time point for gluconeogenic enzymes, our iTRAQ experiments were performed using wild-type cells that were transferred to YPD for 30 min. For these iTRAQ experiments, gluconeogenic enzymes that were shown to be in the extracellular space were considered as positive controls that should be identified in the extracellular fraction. Their levels should decrease in the extracellular fraction at the t = 30 min time point. Additional controls for extracellular proteins included isoforms of glyceraldehyde-3-phosphate dehydrogenase Tdh1p, Tdh2p, and Tdh3p that have been shown to be on the cell-surface of Saccharomyces cerevisiae[11, 38, 67]. Ssa1p and Ssa2p are on the cell surface of Saccharomyces cerevisiae as indicated by immunofluorescence microscopy[15] and served as additional controls for extracellular proteins. Invertase Suc2p and the constitutively expressed acid phosphatase Pho3p are secreted into the periplasm[68, 69] and should also be identified in the extracellular fraction.

Wild-type cells were grown in YPKG for 72 hours after which equal amounts of cells (OD₆₀₀ =10) were either harvested at t = 0 min or transferred to YPD for 30 min (t = 30 min). Extracellular proteins were extracted from t = 0 and t = 30 cells, precipitated with TCA and digested with trypsin. The resulting tryptic peptides were then labelled with different iTRAQ tags and subjected to 2D-LC/MALDI-MS/MS (Figure 5A). In the first iTRAQ experiment (called iTRAQ1 in this study), peptides from replicates of t = 0 cells were labelled with the 113 and 114 tags, whereas peptides from replicates of t = 30 cells were labelled with 115 and 116 tags. We then used the 115/113 (t30_a/t0_a) and 116/114 (t30_b/t0_b) ratios to identify proteins that changed levels after a shift to YPD. In the iTRAQ1 experiment, Mdh2p, Icl1p, Pck1p, Mls1p, Tdh1p, Tdh2p, Tdh3p, Ssa1p, Ssa2p, Suc2p, and Pho3p were identified with multiple peptides having more than 95% confidence (Additional file1: Table S1), although for unknown reasons, Fbp1p was not identified in the iTRAQ1 experiment. In a repeat of the same iTRAQ experiment using similar approaches (called iTRAQ2 in this study), peptides from t = 0 cells were labelled with the 113 iTRAQ tag, whereas peptides from t = 30 cells were labelled with the 114 iTRAQ tag. We then used the 114/113 (t30_c/t0_c) ratios to quantitate level changes in the identified proteins (Figure 5A). In the iTRAQ2 experiment, all 12 positive control proteins including Fbp1p were identified with multiple peptides having at least 95% confidence (Additional file1: Table S1). The identification of these proteins known to be in the extracellular fraction thus validated our iTRAQ studies.

Overall, we have confidently identified 318 extracellular proteins in iTRAQ1 and 392 extracellular proteins in ITRAQ2 with an estimated local False Discovery Rate (FDR) of less than 5%. Proteins from which only a single peptide had been identified were manually inspected and only those that were identified in both experiments were included. (Proteins that were identified with a single peptide with 95% confidence in only one of the two iTRAQ experiments were not included). After removing those single-peptide proteins, a total of 347 extracellular proteins remained including 264 extracellular proteins from iTRAQ1 and 324 proteins from iTRAQ2. The name of each identified protein with their accession number and the number of peptides identified with at least 95% confidence are listed in Additional file1: Table S1. In general, both iTRAQ1 and iTRAQ2 identified similar sets of proteins, but iTRAQ2 identified additional proteins that nonetheless belonged to the same biological functions. For instance, there are four Ssa proteins in the Hsp70 family in Saccharomyces cerevisiae. Three of the four Ssa proteins (Ssa1p, Ssa2p, and Ssa4p) were identified in iTRAQ1, whereas all four Ssa proteins were identified in iTRAQ2. As seen in the Venn diagram (Figure 5B), 241 identified proteins were seen in both iTRAQ experiments, whereas 23 proteins were uniquely identified in iTRAQ1 and 83 proteins were uniquely identified in iTRAQ2.

Classification of extracellular proteins identified by iTRAQ

According to our iTRAQ data, a large number of extracellular proteins that we have identified were involved in the metabolisms of carbohydrates, amino acids, lipids, and nucleotides (Additional file1: Table S1). For the metabolism of carbohydrates, enzymes involved in glycolysis/gluconeogenesis, the pentose pathway, glycogen biosynthesis, the TCA cycle, and alcohol production were identified as present in the extracellular fraction (Additional file1: Table S1). The secretion of these enzymes in response to different stimuli has been reported in many studies; for example, glycolytic enzymes are secreted as immunogens during infection by Candida albicans[16, 38, 70], and GAPDH and enolase may be secreted during invasion by pathogens, as they bind to various mammalian proteins such as lysozyme, fibronectin, actin, myosin, and plasmin[16, 38, 70]. Elevated levels of glycolytic enzymes are secreted from various cancer cells[7, 26–29, 71]. Furthermore, enolase, GAPDH, 3-phosphoglycerate kinase, alcohol dehydrogenase, pyruvate kinase, fructose-1,6-bisphosphate aldolase, phosphoglycerate mutase and triose phosphate isomerase are commonly identified in the secretomes from bacteria, fungi, parasites and mammalian cells[6, 7, 22–24, 27, 37, 70]. In this current study, we have also identified enzymes involved in glycolysis, the pentose pathway, glycogen biosynthesis, alcohol metabolism, as were a large number of proteins involved in the metabolisms of amino acids and purines/pyrimidines.

Anti-oxidant proteins are important for detoxifying potentially damaging reactive oxygen and nitrogen intermediates. The anti-oxidant proteins superoxide dismutase and thioredoxin peroxidase have been reported in various secretomic studies[20, 27, 31, 36, 72]. Many of these anti-oxidant proteins were identified in our study. Heat shock proteins Hsp70, Hsp90, and Hsp97 constituted the largest protein family in the secretome from adult worms of Schistosoma japonicum[73]. The immunodominant 47-kDa antigen in Candida albicans was the breakdown product of Hsp90 and was detected in patients with systemic candidiasis[38, 74]. The Hsp70s play important roles in protein folding, protein translocation across membranes, gene regulation, and protection from heat shock[75, 76]. Multiple protein-folding ATPase, heat shock proteins, and co-chaperones were also identified in the current study. Protein disulfide isomerases and cyclophilins may act with heat shock proteins to assist in protein folding and are frequently observed in many secretomic studies from different species[6, 26, 27, 36]. The current study also identified proline isomerase, proline rotamase, and protein disulfide isomerase.

Validation of extracellular proteins identified by iTRAQ

We used the extraction protocol and Western blotting to validate the extracellular location of 10 candidate proteins identified from the iTRAQ experiments (Figure 6). Wild-type cells expressing GFP-tagged proteins were grown in YPKG and subjected to the extraction procedure. Levels of GFP tagged proteins in the intracellular and extracellular fractions were examined using anti-GFP antibodies. To rule out the possibility that GFP tagging caused non-specific appearance of proteins in the extracellular fraction, we first examined the distribution of proteins known to be in the intracellular fraction that were fused with GFP. For example, signalling molecules involved in the inactivation and degradation of gluconeogenic enzymes are distributed mostly in the intracellular fraction under the same experimental conditions[77]. Vps34p is the major phosphatidylinositol 3-kinase in yeast[78] and is involved in the degradation of Fbp1p[77]. Tor1p is a subunit of the TORC1 complex[79] required for the degradation of gluconeogenic enzymes[66]. Tpk1p is a subunit of the protein kinase A that phosphorylates Fbp1p in response to glucose replenishment[57–59]. GFP fusion to these proteins should not change the distribution of these proteins in the intracellular fraction.

We next confirmed that proteins that are known to be present in the extracellular fraction/cell-surface were indeed distributed in the extracellular fraction. Scw4p and Scw11p are soluble cell wall proteins[80], and the majority of Scw4p and Scw11p was found in the extracellular fraction (Figure 6). We then determined the distribution of enzymes involved in carbohydrate metabolism. Tdh1p is a cell-surface protein[11, 38, 67], and was in the extracellular fraction (Figure 6). Pgk1p is 3-phosphoglycerate kinase that transfers high-energy phosphoryl groups from 1,3-bisphosphoglycerate to ADP during glycolysis and the reverse reaction during gluconeogenesis[81]. Zwf1p is glucose-6-phosphate dehydrogenase that catalyzes the first and irreversible step of the pentose phosphate pathway[82]. Pdc1p is the major pyruvate decarboxylase that decarboxylates pyruvate to acetaldehyde for ethanol production[83]. Consistent with our iTRAQ results, Pgk1p, Pdc1p, and Zwf1p were detectable in the extracellular fraction (Figure 6). We also examined the distribution of proteins involved in oxidative stress. Sod1p is the copper-zinc superoxide dismutase that detoxifies superoxide[84], while Trx1p is thioredoxin that protects cells against oxidative and reductive stress[85]. Both Sod1p and Trx1p were in the extracellular fraction (Figure 6). Finally, we examined the distribution of heat shock proteins. As mentioned above, Ssa1p is a cell-surface protein[15] and was in the extracellular fraction (Figure 6). Hsc82p is required for folding of nascent polypeptides as well as for the refolding of denatured proteins[86]. Hsp104p refolds and reactivates previously denatured and aggregated proteins[87]. Hsc82p and Hsp104p were also present in the extracellular fraction (Figure 6). In contrast, Vps34p, Tor1p, and Tpk1p were distributed mostly in the intracellular fraction (Figure 6). Thus, proteins participate in many different biological functions including those involved in glycolysis/gluconeogenesis, the pentose pathway, alcohol production, oxidative stress, and protein folding are all present in the extracellular faction.

Validation of the t30/t0 ratios

We next used the t30_a/t0_a and t30_b/t0_b ratios from iTRAQ1 and the t30_c/t0_c ratio from iTRAQ2 to identify proteins that changed levels following a transfer to YPD (Additional file2: Table S2). The Protein Pilot Program calculates a p-value for each protein t30/t0 ratio based on the mean of the t30/t0 ratios of all the peptides belonging to that protein, and the scatter of those peptide ratios. We highlighted in bold the protein t30/t0 ratio whose calculated p-value was less than 0.05 (Additional file2: Table S2). Also highlighted (Additional file2: Table S2) are a total of 230 proteins that contained at least one of the three t30/t0 ratios having p-values smaller than 0.05. Because gluconeogenic enzymes decrease their levels after cells are transferred to YPD, they should have reduced t30/t0 ratios. Consistent with this idea, the t30/t0 ratios for these gluconeogenic enzymes were reduced. In a similar manner, significantly decreased t30/t0 ratios were observed for enzymes involved in glycolysis, the pentose pathway, alcohol production, glycogen biosynthesis, and the TCA cycle. Moreover, proteins involved in oxidative stress, in protein folding, and in other functions also exhibited a significant reduction in t30/t0 ratios (Additional file2: Table S2).

According to our iTRAQ data, t30/t0 ratios were reduced for Tdh1p, Pgk1p, Zwf1p, and Pdc1p (Additional file2: Table S2). Our iTRAQ results also indicated that Sod1p, Trx1p, Ssa1p, Hsc82p, and Hsp104p decreased their t30/t0 ratios following a transfer of cells to YPD. To confirm that these extracellular proteins reduced their levels following glucose addition in YPD, wild-type cells expressing GFP tagged proteins were grown in YPKG media and transferred to YPD for 0, 15, and 30 min. Cells were subjected to the extraction procedure and the distribution of GFP-tagged proteins in the intracellular and extracellular fractions was determined by Western blotting using anti-GFP antibodies (Figure 7). At t = 0 min, Tdh1p, Pgk1p, Zwf1p, and Pdc1p were detectable in the extracellular fraction. Following a transfer of cells from YPKG to YPD for up to 30 min, their levels in the extracellular fraction decreased rapidly. In a similar manner, Sod1p, Trx1p, Ssa1p, Hsc82p, and Hsp104p were in the extracellular fraction when cells were grown in YPKG. Levels of these proteins in the extracellular fraction decreased after a shift of cells to YPD for 30 min. Taken together, these proteins that are involved in different biological functions show very similar distribution characteristics. They are all present in the extracellular fraction during growth in YPKG and their levels in the extracellular fraction are all reduced following a transfer of cells to YPD for 30 min.

Small Vesicles disappear in response to glucose re-feeding

The extraction procedure employed in the current work has been used to study aspects of protein composition in the extracellular fraction[2, 64, 77]. However, the contents of the extracted materials had not previously been visualized at the ultra-structural level. Small vesicles called exosomes are released from a variety of mammalian cells. In yeast, Fbp1p is associated with small vesicles that are 30–50 nm in diameter[88]. Because of the size of these vesicles, they can only be visualized by EM techniques. We thus performed TEM studies to examine whether or not small vesicles were present in the extracellular fraction in cells grown in YPKG. Wild-type cells were grown in YPKG for 3d and harvested (t = 0 min), or transferred to YPD for 30 min (t = 30 min) and subjected to the same extraction procedure. Extracted materials were centrifuged at 3,000 × g and then at 200,000 × g. The 200,000 × g pellets from the t = 0 and t = 30 cells were stained with uranyl acetate followed by visualization using TEM (Figure 8A and8B). When total extracts were prepared from t = 0 cells, at least two types of structures were observed. Small vesicles of 30–50 nm in diameter and large structures of 100–300 nm in diameter were identified (Figure 8A, arrows). Quantification of these structures indicated that total extracts from cells grown in YPKG consisted of approximately 93.9% small vesicles and 6.1% large structures. However, when total extracts were isolated from t = 30 cells, very few small vesicles were observed, whereas the 100–300 nm large structures still remained (Figure 8B, arrows). The number of small vesicles in total extracts was 218.3 ± 13.5 per μm² at t = 0 min and was 4.5 ±1.3 per μm² at t = 30 min (Figure 8C). The number of 100–300 nm large structures was 14.3 ± 1.7 per μm² at t = 0 min and was 10.3 ±1.1 per μm² at t = 30 min (Figure 8D). Therefore, approximately 98% of the small vesicles disappeared within 30 min following glucose replenishment in YPD. In contrast, only about 28% of the large structures had disappeared at the 30 min time point. This rapid disappearance of small vesicles from the extracellular fraction at t = 30 min demonstrates another example of rapid changes in the secretome/extracellular fraction during transition from glucose-deficient to glucose-rich media.

Finally, we determined whether or not these extracellular proteins were distributed in the vesicle-enriched fraction. Wild-type cells expressing GFP fusion proteins were grown in YPKG for 3d and extracted. Total extracts were first centrifuged at 3,000 × g and then at 200,000 × g. The 200,000 × g pellet fraction was resuspended in PBS and incubated in the absence or presence of 2% SDS for 30 min. Following incubation, samples were re-centrifuged at 200,000 × g for 2 hours and the distribution of proteins in the 200,000 × g supernatant (S) and pellet (P) fractions was then examined. In the absence of detergent, most of the proteins were in the 200,000 × g pellet fraction (Figure 9, left panels), suggesting that these proteins are associated with the small vesicles. However, when membranes were disrupted by SDS, the majority of these proteins were in the 200,000 × g supernatant fraction (Figure 9, right panels). Thus, the distribution of these proteins in the vesicle-enriched fraction is dependent on membrane integrity.

Cell culture, media and antibodies

Wild-type (BY4742, MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) and wild-type cells expressing GFP tagged proteins were inoculated from -70°C frozen stock into 2 ml YPD (1% yeast extract, 2% peptone, and 2% glucose) in sterile glass tubes at 30°C in an environmental shaker at 250–300 rpm for 16 hours. Cell densities were measured using a Beckman DU640B spectrophotometer. Cells (50 μl) were diluted to OD₆₀₀ = 0.01/ml in 10 ml YNB (0.67% yeast nitrogen base with amino acids and 2% glucose) or in 10 ml YPKG (1% yeast extract, 2% peptone, 1% potassium acetate, and 0.5% glucose), in sterile glass tubes for 72 hours at 30°C until cell density reached OD₆₀₀ = 4-5/ml. The YNB media was supplemented with 0.1 ml of the 100 × of solution containing 200 mg uracil, 200 mg L-histidine, 300 mg L-leucine, and 300 mg L-lysine. Cells containing OD₆₀₀ = 10/ml were aliquoted and used for each time point. In experiment I, 2% glucose was added directly to the existing YNB culture and incubated for 0, 2, and 4 hours at 30°C. In experiment II, cells were pelleted by a low speed centrifugation at 3,000 × g for 10 min and resuspended in fresh YNB-based media containing 2% glucose for 0, 2, and 4 hours at 30°C. In experiment III, 2% glucose was added directly to the existing YPKG culture for 0, 2, and 4 hours at 30°C. In experiment IV, cells were pelleted and resuspended in YPD containing 2% glucose for 0, 2, and 4 hours at 30°C. Cells grown in YPKG also pelleted and resuspended in YP (without added glucose) for 0, 2, and 4 hours at 30°C. Cells (OD₆₀₀ = 10/ml) were harvested at each time point and processed. Levels of Lia1p and Fbp1p, and Tpi1p were examined by Western blotting. For re-growth experiments, wild-type cells were grown in 10 ml YPKG for 3d at 30°C in an environmental shaker until OD₆₀₀ = 4-5/ml. Cells were pelleted and diluted to OD₆₀₀ = 0.2/ml in 10 ml YPD (1% yeast extract, 2% peptone, 2% glucose), 10 ml YPKG (1% yeast extract, 2% peptone, 1% potassium acetate, and 0.5% glucose) or 10 ml YP (1% yeast extract, 2% peptone) at 30°C. Cell densities were measured at OD₆₀₀ for 0, 2, 4, 6, and 8 hours using a Beckman DU640B spectrophotometer.

Extraction and Western blotting

Extraction of extracellular proteins was performed as described[2, 64, 77]. Wild-type cells and wild-type cells expressing GFP tagged proteins were inoculated from -70°C frozen stock into 2 ml YPD culture in sterile glass tubes in an environmental shaker at 250–300 rpm at 30°C for 16 hours. Cells (50 μl) were cultured into 10 ml of YPKG for 72 hours at 30°C. Cell densities were measured and cells (OD₆₀₀ = 10/ml) were harvested (t = 0) or transferred to YPD containing glucose in an environmental shaker at 250–300 rpm at 30°C for 15, and 30 min (t = 15 and t = 30). At each time point, 10 mM NaN₃ (final concentration) was added to stop the reactions. Cells were harvested by a low speed centrifugation at 3,000 × g for 10 minutes. Following centrifugation, the culture media was removed and cell pellets were resuspended in 100 μl of extraction buffer that contained 0.1 M Tris pH 9.4 and 10 mM β-mercaptoethanol. Samples were placed in a 37°C water bath for 15 min at 200 rpm. Following the extraction procedure, cells were pelleted at 3,000 × g for 10 minutes. The extracellular proteins (100 μl) that were released into the supernatants were precipitated in 15% trichloroacetic acid (TCA), washed, and solubilized in SDS-PAGE sample buffer. The remaining cell-associated fraction was lysed and solubilized in SDS-PAGE sample buffer. The distribution of intracellular and extracellular proteins was examined by Western blotting using Western Lighting Plus ECL kit (Perkin Elmer). Wild-type cells expressing GFP tagged proteins were purchased from Invitrogen. Anti-GFP antibodies were purchased from Abcam. Anti-TPI antibodies were from Proteintech Group.

Trpan blue staining and FM4-64 uptake

Wild-type cells were grown in 10 ml YPKG for 3d at 30°C in an environmental shaker. Cells were either harvested at t = 0 min or transferred to YPD for 30 min at 30°C. Cells were resuspended in 200 μl PBS buffer containing 140 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.4 and incubated in the absence or presence of 2% Triton X-100 for 30 min. This was followed by incubation with 20 μl of 0.4% trypan blue in PBS for 30 min at room temperature. The same amounts of t = 0 and t = 30 cells were subjected to the extraction procedure and pelleted. Extracted cells were resuspended in 200 μl PBS and incubated in the absence or presence of 2% Triton X-100 for 30 min. Treated and non-treated cells were then incubated with 20 μl of 0.4% trypan blue for 30 min. Cells were observed using a Zeiss microscope.

For FM uptake experiments, wild-type cells were grown in 10 ml YPKG for 3d at 30°C. Cells were either harvested at t = 0 min or transferred to YPD media for 30 min (t = 30) at 30°C. Half of the cells were pelleted, resuspended in 200 μl PBS, and incubated with 1 μl FM (16 mg/ml) for 3 hours at 30°C. Another half of the cells were extracted, resuspended in 200 μl PBS and then incubated with 1 μl FM for 3 hours at 30°C. The distribution of FM was visualized using a Zeiss fluorescence microscope.

iTRAQ labelling and data analysis

Two TRAQ labelling procedures were performed following the instructions from Applied Biosystems with minor modifications. Wild-type yeast cells were grown in YPKG and either harvested (t = 0 min) or the same amounts of cells (OD₆₀₀ =10/ml) were transferred to media containing 2% glucose for 30 min (t = 30 min). Extracellular proteins were extracted as described above and then precipitated in 15% TCA. Protein concentrations were determined using Bio-Rad DC assay. Extracted proteins were reduced with 5 mM of tris(2-carboxyethyl) phosphine at 60°C for 60 min, and alkylated with 4 mM iodoacetamide for 30 min at room temperature in the dark. Samples were digested with 10 μg of trypsin at 48°C for 16 hours. Peptides were labelled with iTRAQ tags, combined, dried and resuspended in 500 μl of 10 mM ammonium formate, pH 3.6, in 20% acetonitrile/80% water. Labelled peptides were subjected to SCX separation by loading onto a passivated Waters 600E HPLC system with a 4.6 × 250 mm PolySULFOETHYL Aspartamide^TM column (PolyLC, Columbia MD) followed by elution at a flow rate of 1 ml per minute with a 1 hour gradient going from 10 mM→666 mM ammonium formate in 20% acetonitrile, pH 3.6. After evaporating and resuspending the resulting 15 SCX fractions 3 times in a SpeedVac (final resuspension in 9 ml of 2% acetonitrile, 0.1% TFA), reverse phase C18 nanoflow-LC separation of the 15 resuspended SCX fractions was accomplished by loading each fraction onto a Chromolith CapRod column (150 × 0.1 mm, Merck) from a 5 ml sample loop on an LC-Tempo MALDI separation/plate spotting system at 2 ml per minute in 6.8% acetonitrile, 0.1% TFA, then eluting with a 50 minute gradient from 6.8% to 79% acetonitrile, 0.1% TFA at a flow rate of 2.5 ml per minute. A 3 ml per minute flow of MALDI matrix solution was added in-line post-column (7 mg/ml recrystallized CHCA (a-cyano-hydroxycinnamic acid), 2 mg/ml ammonium phosphate, 0.1% trifluoroacetic acid, 80% acetonitrile), and the resulting eluant was automatically spotted onto a stainless steel MALDI target plate every 6 seconds (0.55 μl per spot), for a total of 370 spots per original SCX fraction. Each target plate was analyzed in a data-dependent manner on either an ABI 5800 MALDI TOF-TOF or an ABI 4800 MALDI TOF-TOF. The MS spectra for an entire 2D-LD experiment were thus taken from 5500 MALDI spots, averaging 500 laser shots per spot at laser power 3400. In a data-dependent manner, 10,000-15,000 MS/MS spectra were taken of the strongest example of each unique m/z mass observed across each target plate, using up to 2500 laser shots per MS/MS spectrum at laser power 4100, with CID gas at 1.2 to 1.3 × 10^-6 Torr.

Protein identification was accomplished using the Paragon Algorithm[96] as implemented in ProteinPilot^TM 4.0 software (ABSciex). Search parameters were set as cysteine alkylation: iodoacetamide, ID focus: biological modifications, search effort: thorough. Combined spectra were searched against the species-specific (S. cerevisiae) NCBI RefSeq database concatenated with a reversed “decoy” version of the same database plus 156 common lab contaminants as described by Keller et al.[97]. To reduce the potential number of false positive protein identifications, the Paragon Algorithm search results were further filtered through the use of a very stringent Local False Discovery Rate (FDR) estimation calculated from the Proteomics System Performance Evaluation Pipeline (PSPEP)[98, 99]. We accepted protein IDs with an estimated local False Discovery Rate (FDR) estimate of <5%. A total of 318 extracellular proteins were identified in iTRAQ1 (Additional files3 and4) and 392 extracellular proteins were identified in iTRAQ2 (Additional files5 and6) using these criteria for acceptance. We next removed proteins that were identified by only a single confidently identified peptide in only one of the two iTRAQ experiments, (proteins that were identified with a single peptide with 95% confidence, but in both experiments, were included). From the identifiers (gi numbers) obtained from searching against the NCBI RefSeq database, we used the Saccharomyces Genome Database (Stanford University) and the Gene Ontology Slim Mapper to search for the function for each protein. BioVenn analysis was used to compare proteins identified from different studies[100].

Transmission electron microscopy

Immuno-TEM was performed as described previously[77, 101]. Wild-type cells were grown in 2 ml YPKG for 3d and harvested (t = 0 min) or transferred to YPD for 30 min and harvested (t = 30 min). The cells were fixed with 3% para-formaldehyde and 0.2% glutaraldehyde for 16 hours. Cells were then washed by centrifugation at 4,500 × g sequentially in 1 ml of 0.5 M sorbitol in 0.08 M potassium phosphate buffer (pH 6.7), 0.25 M sorbitol in 0.08 M potassium phosphate buffer (pH 6.7), and 0.08 M potassium phosphate buffer (pH 6.7). Cells were resuspended in 1 ml of sodium metaperiodate (1% w/v), incubated at room temperature for 20 min, re-suspended in 1 ml of 50 mM ammonium chloride, and then incubated for 15 min at room temperature. Cells were centrifuged at 4,500 × g and dehydrated by 5 min incubation in 1 ml 50% cold ethanol, 70% cold ethanol and 80% cold ethanol followed by 10 min incubation in 85% cold ethanol, 90% cold ethanol, 95% cold ethanol and 100% cold ethanol. Finally, 1 ml 100% ethanol was added to the cell pellet and incubated for 5 min at room temperature. After dehydration, cells were sequentially incubated in 1 ml LR White 2:1, ethanol:LR White at room temperature rotator for 2 hours, 1:1 ethanol:LR White resin at room temperature rotator overnight, 1:2 ethanol:LR White resin at room temperature rotator for 2 hours and 100% resin at room temperature rotator for 2 hours three times. Cells were transferred to gelatin capsules, dried and cut into 10 nm thin sections. Sections were placed onto grids and incubated with affinity purified Fbp1p antibodies followed by goat anti-rabbit antibodies conjugated with 10 nm gold particles (Ted Pella) and viewed on a JEOL JEM-1400 electron microscope.

For negative staining and TEM, wild-type cells were grown in 200 ml of YPKG in sterile glass flasks for 72 hours and split into equal aliquots based on cell density. Cells (100 ml) were transferred to YPD media containing 2% glucose for 30 minutes (t = 30 min). Cells (100 ml) from the t = 0 and t = 30 time points were harvested and resuspended in 4 ml extraction buffer containing 0.1 M Tris pH 9.4 and 10 mM β-mercaptoethanol at 37°C for 15 minutes. Total extracts from the t = 0 and t = 30 cells were centrifuged at 3,000 × g for 10 minutes at 4°C. The 3,000 × g supernatant was further centrifuged at 200,000 × g for 2 hours at 4°C. The resulting 200,000 × g pellets were resuspended in 30 μl of PBS. Samples were fixed with 3% para-formaldehyde and 0.2% glutaraldehyde for two hours at 4°C. Nickel grids were incubated with 10 μl of poly-lysine buffer for 5 minutes and washed with distilled water for 3 minutes. Total extracts (10 μl) were incubated with poly-lysine coated grids for 30 minutes. The grids were blotted with 0.1% BSA in PBS buffer. The grids were washed once with PBS buffer and twice with distilled water. The grids were then blotted with Whatman filter papers and stained with 2% uranyl acetate for 10 minutes in the dark. The grids were dried by blotting with Whatman filter papers and samples were viewed on a JEOL JEM-1400 electron microscope.