Cycloheximide treatment affects the abundance of 33 PM proteins

To identify new PM proteins that require α-arrestins for endocytosis, we performed a quantitative proteomic screening. We analyzed and compared the PM proteome for each artΔ strain to a wild-type (WT) strain after a 90-min cycloheximide (CHX) treatment. CHX is a translation inhibitor and can trigger the endocytosis of several transporters in a substrate-independent manner [6,7]. We searched for proteins that underwent CHX-induced endocytosis in a WT strain but that remained stabilized at the PM in an artΔ strain (Tables 1 and 2). We identified 93 different PM proteins (S1 Table), which account for approximately 39% of the total number of PM proteins reported in databases (245 in Saccharomyces Genome Database and 234 in UniProtKB). The first step of the analysis consisted in the comparison of protein abundance in a CHX-treated WT strain (WT+CHX) and a mock-treated WT strain (WT ctrl) in order to identify proteins affected by the treatment (Table 1 and S2 Table). We observed that 24 integral PM proteins, mostly transporters, were less abundant in the WT+CHX than in the WT ctrl (in blue on Table 1), suggesting that CHX stimulates their endocytosis. In comparison, 9 proteins were more abundant in the WT+CHX than in the WT ctrl (in red on Table 1), suggesting that these proteins accumulate at the PM in response to CHX.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. CHX treatment affects the abundance of 33 PM proteins.

Fold-change ratios of the abundance of PM proteins identified in the proteomic screening after CHX treatment. The first two columns indicate the UniProtKB accession numbers and names of proteins. The third column corresponds to the WT+CHX/WT ctrl ratio, specifying the protein relative abundance in the CHX-treated WT divided by its relative abundance in the mock-treated WT. The same total amount of proteins was analyzed in both conditions. The blue and red colors specify whether the ratio is significantly lower or higher than one after hypothesis testing (i.e., with a p-value < 0.05), respectively. A protein with a blue ratio (WT+CHX/WT ctrl < 1) is less abundant in the WT+CHX than in the WT ctrl, indicating a decrease in abundance upon CHX treatment. Inversely, a protein displaying a red ratio (WT+CHX/WT ctrl > 1) is more abundant in the WT+CHX than in the WT ctrl, suggesting an increase of the protein abundance upon CHX treatment. Proteins are classified according to their fold change after CHX treatment of WT strain, from the greatest decrease (Ptr2) to the greatest increase in abundance (Zeo1). Displayed ratios correspond to the geometric means from 6 independent experiments.

https://doi.org/10.1371/journal.pbio.3000512.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Impact of α-arrestin deletion on the abundance of PM proteins affected by the CHX treatment.

Fold-change ratios of the abundance of PM proteins identified in Table 1 between the CHX-treated artΔ and WT strains. The first two columns indicate the UniProtKB accession number and name of proteins. The remaining columns correspond to artΔ+CHX/WT ctrl ratio, specifying the protein relative abundance in the CHX-treated arrestin-knockout strains divided by its relative abundance in the CHX-treated WT. The same total amount of proteins was analyzed in all conditions tested. The red and blue colors specify whether the ratio is significantly higher or lower than 1 after hypothesis testing (i.e., with a p-value < 0.05), respectively. A protein displaying a red ratio (artΔ+CHX/WT+CHX > 1) is more abundant in the artΔ+CHX than in the WT+CHX condition. We interpret an increase in surface protein upon α-arrestin deletion (i.e., indicated in red) as reflecting a defect in internalization of that PM protein. Inversely, a protein with a blue ratio (artΔ+CHX/WT+CHX < 1) is less abundant in the artΔ+CHX than in the WT+CHX. White ratios are not significantly different from one upon hypothesis testing, regardless of their absolute values. Proteins are classified as in Table 1. Displayed ratios correspond to the geometric means from 3 independent experiments.

https://doi.org/10.1371/journal.pbio.3000512.t002

To determine whether CHX-induced endocytosis is α-arrestin dependent, we compared the effect of α-arrestin deletion (artΔ+CHX vs WT+CHX) on PM protein abundance of the same 24 proteins that were depleted upon CHX treatment (Table 2 and S2 Table). We interpreted an increase in surface protein upon α-arrestin deletion (indicated in red) as reflecting a defect in internalization of that PM protein. It appeared that the endocytosis of some proteins was affected by only one α-arrestin, such as peptide transporter 2 (Ptr2). For example, in the absence of Bul1 (bul1Δ), CHX treatment failed to reduce the abundance of Ptr2 (Table 2). In contrast, the abundance of other proteins was affected by several α-arrestins. For example, the catalytic subunit of the β-1,3-D-glucan synthase Fks1 appears to be stabilized in several artΔ strains. These data suggest that several α-arrestins are involved in the regulation of Fks1 abundance at the PM in a nonredundant manner, since the lack of each one independently results in an increased Fks1 abundance. As expected, our data are consistent with previously described interactions between model transporters and α-arrestins, such as the Bul1 dependency of Ptr2 endocytosis after CHX treatment [33], the Art1 dependency of Can1 endocytosis after substrate and CHX treatment [6], and the regulation of Hxt1 and Hxt3 by Art4 and Art7 after 2-deoxyglucose treatment [14]. However, our data did not show some of the previously described effects of α-arrestins on some transporters. For example, the role of Art8 in Hxt7 endocytosis upon glucose depletion [16] was not observed in our current working condition, most likely because cells were grown in glucose-containing medium.

Unexpectedly, proteins such as Hxt4, Hxt7, Pma1, or Gas1 were less abundant at the PM in some artΔ strains after CHX treatment (Table 2). This could be due to increased protein degradation in the absence of the α-arrestin, the involvement of a stimulated compensatory α-arrestin-independent endocytic mechanism, or reduced gene expression as recently illustrated for Hxt6 and Hxt7 in an art8Δ strain [16]. Of note, the glycosylphosphatidylinositol (GPI)-anchored protein Gas1 does not have any cytosolic part and is therefore unable to interact directly with α-arrestins. Yet, its abundance is decreased in the art9Δ and bul2Δ backgrounds compared to the WT condition. To our knowledge, this result is the first example of an effect of α-arrestin on the abundance of GPI-anchored proteins.

Among the nine proteins that were more abundant after the CHX treatment than in the control condition, we primarily identified peripheral proteins or proteins tightly associated with the PM. Indeed, the eisosome component Nce102 was the only identified intrinsic PM protein. These nine proteins are involved in either the regulation of PM lipid composition (as the ER–PM tether components Tcb1/3 and Ist2), PM organization (Nce102), cell integrity pathway (Rho1, Zeo1), or polarized secretion (Rho3). Moreover, we identified the guanosine triphosphate (GTP)-binding protein Ras2, previously shown to physically interact with Art1, Art4, and Art7 [34] and with two other identified proteins of our screening, the ATP-binding cassette (ABC) transporters Yor1 and Pdr5 [35]. In our proteomic analysis, Ras2 abundance is decreased in art9Δ and bul2Δ conditions.

CHX- and thiamine-induced endocytosis of Thi7 is impaired in the art2Δ strain

The abundance of the high-affinity thiamine transporter Thi7 was highly affected by CHX treatment. In the CHX-treated WT, Thi7 abundance is reduced to 14% of its value after mock treatment. Moreover, following CHX treatment, Thi7 was 4 and 11 times more abundant at the PM of art2Δ and art9Δ strains than of the WT, suggesting that its CHX-induced endocytosis is mediated by Art2 and Art9. These results indicate that these two α-arrestins are not fully redundant, as the presence of ART9 in art2Δ does not rescue the down-regulation of Thi7 upon CHX treatment and vice versa.

To assess protein trafficking, we engineered N- and C-terminal green fluorescent protein (GFP) fusions of Thi7 expressed under the THI7 promoter and observed protein localization following CHX treatment. The same was carried out for the two low-affinity thiamine transporter homologs Nrt1 and Thi72 (84% and 87% identity with Thi7, respectively). Phenotypic growth tests confirmed the activity of C-terminal GFP fusions (S1 Fig). The growth defect triggered by a toxic analog of thiamine, oxythiamine, was used as readout of transporter activity. The thi7Δnrt1Δthi72Δ strain, which is defective for thiamine transport and expresses native N- and C-terminal GFP proteins, was grown on thiamine-free oxythiamine-supplemented medium. As expected, the thi7Δnrt1Δthi72Δ strain transformed with an empty plasmid was resistant to oxythiamine. We observed that THI7- and THI72-expressing strains were sensitive to oxythiamine, whereas the NRT1-expressing strains were less sensitive. The C-terminal GFP fusion proteins displayed the same phenotype as the corresponding native protein. Because thiamine is the physiological substrate of Thi7, thiamine uptake by the fusion proteins was also probed by testing growth on low thiamine concentration. The thi7Δnrt1Δthi72Δthi4Δ strain, which is deficient for thiamine transport and synthesis and expresses GFP-fused transporters, was grown on medium containing 1 nM to 100 μM thiamine. As expected, the thi7Δnrt1Δthi72Δthi4Δ strain that was transformed with the empty plasmid only grew in the presence of the highest concentration of thiamine because of unspecific thiamine entry (S1 Fig). In contrast, strains that expressed a transporter grew on all media regardless of thiamine concentration. Growth level of strains expressing the C-terminal GFP fusion proteins was similar to those expressing native transporters.

To confirm that the high abundance of Thi7 following CHX treatment in art2Δ and art9Δ strains resulted from the impairment of Thi7 endocytosis, we analyzed Thi7-GFP fluorescence in WT and the different artΔ strains. In addition, we included the analysis of Thi7-GFP endocytosis in the mutant strain rsp5, which is deficient for RSP5 expression [5,36]. Strains were grown in thiamine-free medium to allow THI7-GFP expression and localization at the PM before mock- or CHX treatment [37]. Specific immunoblot signal for Thi7-GFP appears as two major bands in the 100-kDa range (Fig 1A). Moreover, free GFP, which is more resistant to vacuolar degradation, is detected at about 27 kDa, suggesting that our constructs can be targeted to the vacuole. CHX treatment decreased Thi7-GFP abundance in the WT strain compared to the mock-treated condition. Reduced Thi7-GFP abundance was attenuated in art2Δ and rsp5 backgrounds but not in art9Δ (Fig 1A). Free GFP signal is less intense in art2Δ and almost absent in rsp5 strains. Of note, the measured endocytosis rates of Thi7 are lower in immunoblot experiments than in the proteomic screening because of two major differences in the methodology. On the one hand, mass spectrometry was performed on PM-enriched fractions to focus on the abundance of Thi7 at the PM, whereas immunoblots were performed on total protein extracts. In total extracts, the whole pool of Thi7 is identified (including Thi7 en route to the PM or the vacuole), lessening the ratio of Thi7 degradation. On the other hand, the protein extracts used for immunoblots were performed on cells growing in thiamine-free synthetic medium, maximizing THI7 expression, whereas mass spectrometry was performed in YD medium (containing thiamine), in which THI7 expression is low. This may affect the relative proportion of endocytosed Thi7. In parallel, we assessed Thi7-GFP localization after CHX treatment and observed that the construct was blocked at the PM in art2Δ and rsp5 backgrounds (Fig 1B). By contrast, Thi7-GFP localized to the vacuolar compartment in WT and art9Δ strains (Fig 1B). The art2Δart9Δ strain has the same phenotype as the art2Δ strain (Fig 1B). Together, these results demonstrate that CHX-induced endocytosis/degradation of Thi7 is Rsp5 and Art2 dependent.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Art2 and Rsp5 are required for CHX- and thiamine-induced Thi7 endocytosis.

(A) The WT, art2Δ, art9Δ, and rsp5 strains expressing THI7-GFP under their endogenous promoter were grown in thiamine-free selective medium up to early-log phase and harvested after 120 min with CHX (final concentration; 50 μg/ml) or EtOH (top panel) and with thiamine (“Th.,” final concentration: 100 μM) or water (H₂O) (bottom panel). Total protein extracts were immunoblotted with anti-GFP and anti-Pma1 as a loading control. Free GFP is the results of the degradation of Thi7-GFP in the vacuole. Thi7-GFP band intensity is normalized by the intensity of the Pma1-corresponding band. Representative of 3 experiments. (B) Localization of Thi7-GFP in the WT, art2Δ, art9Δ, rsp5, and art2Δart9Δ strains after addition of thiamine or CHX or mock-treated with water. The vacuolar membrane is stained with FM4–64. Quantification shows the ratio of internal-over-PM fluorescence intensity as described in Materials and methods section (n > 30 cells) (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05). (C) Expression of 3xHA-ART2 restores thiamine-induced endocytosis of Thi7-GFP in the art2Δ background. THI7-GFP and either 3xHA-ART2 or its corresponding e.v. are expressed in the art2Δ strain, grown in thiamine-free selective medium, and complemented with thiamine (final concentration: 100 μM). Scale bar represents 5 μm. The numerical data are included in S1 Data. CHX, cycloheximide; EtOH, ethanol; e.v., empty vector; GFP, green fluorescent protein; ns, nonsignificant; PM, plasma membrane; Pma1, plasma membrane ATPase 1; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000512.g001

To assess trafficking at a condition that was more physiologically relevant, we measured localization of Thi7-GFP following thiamine treatment of cells precultured in thiamine-free medium (Fig 1A). Thiamine is the natural substrate of Thi7, and when added in excess, it inhibits THI7 gene expression [37]. In WT and art9Δ strains, thiamine treatment decreased Thi7-GFP protein level by 67%–76%, indicating that these transporters were degraded. By contrast, thiamine treatment of art2Δ and rsp5 strains partially inhibited Thi7-GFP endocytosis. Thi7-GFP protein levels were reduced by 47% and 29% in art2Δ and rsp5 strains, respectively Consistent with these observations, free GFP level was reduced in art2Δ and rsp5 compared to WT and art9Δ strains. We also monitored the localization of Thi7-GFP after thiamine treatment in WT, art2Δ, art9Δ, art2Δart9Δ, and rsp5 strains by microscopy. GFP signal appeared primarily targeted to the vacuole after 3-h thiamine treatment of WT and art9Δ strain (Fig 1B) as in the remaining artΔ strains (S2 Fig). However, this is not observed in art2Δ, art2Δart9Δ, and rsp5 strains, and the GFP signal appeared to be stabilized at the PM. Finally, expression of 3xHA-ART2 in art2Δ fully rescued thiamine-induced Thi7 endocytosis (Fig 1C). These results demonstrate that thiamine-induced endocytosis/degradation of Thi7 is Rsp5 and Art2 dependent.

The deletion of ART2 and ART9 impairs endocytosis of the two putative low-affinity thiamine transporters Nrt1 and Thi72

We extended our study to examine the effect of α-arrestin on endocytosis of low-affinity thiamine transporters. We examined their endocytosis after thiamine treatment (Fig 2). Thiamine reduced Nrt1-GFP and Thi72-GFP protein abundance by 60% in the WT strain. This reduction was partially inhibited in the art2Δ, art9Δ, rsp5, and art2Δart9Δ strains (Fig 2A). Of note, the effect of ART9 deletion on Nrt1-GFP and Thi72-GFP endocytosis is different from the one observed on Thi7-GFP (Fig 1A). We also calculated the ratio of free GFP normalized to Nrt1-GFP or Thi72-GFP abundance in the different strains to estimate the proportion of transporters targeted to vacuolar degradation. The free GFP abundance is 2-fold higher in thiamine-treated than in the mock-treated WT for both transporters. In contrast, free GFP level is 2-fold lower in art2Δ after thiamine treatment than after mock-treatment, suggesting that degradation is strongly impaired. Although the effect observed in the art9Δ strain is less important than with art2Δ, it nevertheless suggests a lower degradation in art9Δ compared to the WT. Nrt1-GFP and Thi72-GFP, which under thiamine-free conditions are localized to the PM, were targeted to the vacuole in the WT strain treated with thiamine (Fig 2B). The deletion of ART2 reduced the extent of Nrt1 and Thi72 degradation by stabilizing the transporters at the PM. In the art9Δ strain, we observed an intermediate phenotype: GFP signal appeared to localize within the PM of approximately 50% of cells, whereas it could be clearly observed in the vacuoles of the remaining cells. Of note, ART9 deletion had no previous effect on the Thi7 transporter. Finally, we observed Nrt1-GFP and Thi72-GFP localization after thiamine treatment in all the other artΔ strains, and no difference was observed with WT strain (S2 Fig). Taken together, these results emphasize a general role of Art2 in the endocytosis of the three thiamine transporters, whereas Art9 function seems to be restricted to the endocytosis of the low-affinity homologs, Nrt1 and Thi72.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The endocytosis of the low-affinity thiamine transporters Nrt1 and Thi72 is also mediated by Art2.

(A) Strains expressing NRT1-GFP (top panel) and THI72-GFP (bottom panel) under THI7 endogenous promoter were grown in thiamine-free medium up to early-log phase and harvested after 120-min incubation with thiamine (“Th.”) at a concentration of 100 μM for total protein extractions. Extracts were immunoblotted with anti-GFP and anti-Pma1 as a loading control. Values are quantification of band intensity. (B) Localization of Nrt1-GFP or Thi72-GFP in WT, art2Δ, art9Δ, rsp5, and art2Δart9Δ strains after addition of an excess of thiamine into culture grown in thiamine-free medium. The vacuolar membrane is stained with FM4–64. Scale bar represents 5 μm. Quantification shows the ratio of internal-over-PM fluorescence intensity as described in Materials and methods section (n > 50 cells). (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05) The numerical data are included in S2 Data. GFP, green fluorescent protein; Nrt1, nicotinamide riboside transporter 1; ns, nonsignificant; PM, plasma membrane; Pma1, plasma membrane ATPase 1; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000512.g002

Thi7 is ubiquitylated within the cytosolic C terminus in an Art2-dependent manner

We designed various lysine-to-arginine (KR) mutants of Thi7-GFP to identify which lysine residues are essential for Rsp5-dependent ubiquitylation and endocytosis (Fig 3). We performed phenotypic growth tests on oxythiamine-containing medium of the thi7Δnrt1Δthi72Δ strain expressing the KR mutants to ensure that mutations do not impair transport activity (Fig 3A). Thiamine treatment had no effect on the endocytosis of the Thi7^KR-GFP mutant, in which all cytosolic N- and C-terminal lysine residues were mutated to arginine (Fig 3B). Similarly, endocytosis is completely abrogated upon mutation of C-terminal lysine residues (Thi7^CterKR-GFP mutant), suggesting ubiquitylation takes place at the C terminus. Indeed, endocytosis occurred if lysine residue mutations were restricted to the cytosolic N-terminal tail (Thi7^NterKR-GFP mutant). To confine our search to 6 C-terminal lysine residues, which were predicted to be ubiquitylated [38], we engineered the Thi7^6KR-GFP mutant. Similar to the Thi7^KR-GFP and Thi7^CterKR-GFP mutants, the Thi7^6KR-GFP mutant is not endocytosed after thiamine treatment, suggesting that at least one of these lysine residues is essential for Thi7 endocytosis (Fig 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The Art2-dependent ubiquitylation of Thi7-GFP C terminus is required for its endocytosis.

(A) GFP-tagged KR-mutated THI7 was expressed in a thi7Δnrt1Δthi72Δ strain and grown in a thiamine-free medium. Serial 10-fold dilutions of yeast cells in early log phase were spotted on thiamine-free oxythiamine-containing medium. The expression of pRS416 (empty vector) and the nontagged version were used as controls. Oxythiamine was transported by all the KR Thi7 mutants (B) Localization of the different KR mutant versions of Thi7-GFP in the WT strain after thiamine treatment (final concentration: 100 μM): Thi7^KR-GFP, Thi7^NterKR-GFP, Thi7^CterKR-GFP, and Thi7^Cter6KR-GFP. (C) Top: THI7-GFP and THI7^KR-GFP were expressed in end3Δ and end3Δart2Δ strains and grown in thiamine-free medium to early log phase before a 30-min incubation with 100 μM thiamine. Bottom: THI7-GFP, Thi7^KR-GFP, Thi7^NterKR-GFP, Thi7^CterKR-GFP, or Thi7^Cter6KR-GFP were expressed in art2Δ strain (complemented or not with 3xHA-ART2) and grown in thiamine-free medium to early log phase before a 30-min incubation with 100 μM thiamine. After solubilization, IP of transporters was performed using the GFP-Trap kit. Samples were immunoblotted with anti-GFP and anti-Ub. Quantification is based on the intensity of the Ub band and normalized to the intensity of the corresponding GFP signal. GFP, green fluorescent protein; IB, immunoblotting; IP, immunoprecipitation; KR, lysine-to-arginine; Ub, ubiquitin; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000512.g003

To examine the role of Art2 in Thi7 ubiquitylation, we immunoprecipitated Thi7-GFP and Thi7^KR-GFP from end3Δ and end3Δart2Δ backgrounds, in which endocytosis is impaired and ubiquitylated proteins remain stabilized within the PM. Using an anti-ubiquitin antibody, we detected a higher molecular weight signal than Thi7-GFP in the end3Δ strain, corresponding to a ubiquitylated form of the construct (Fig 3C, upper panel). The intensity of this signal was reduced (68%) in the end3Δart2Δ background, suggesting that Art2 is involved in the ubiquitylation of Thi7-GFP. However, the residual signal indicates that some ubiquitylation is Art2-independent. The absence of a ubiquitylated signal within the Thi7^KR-GFP fusion protein, regardless of the background (Fig 3C, upper panel), supports the specificity of our model. Additionally, we tried to rescue the ubiquitylation process by expressing 3xHA-ART2 and the different constructs in art2Δ (Fig 3C, lower panel). In the absence of Art2, Thi7-GFP is not ubiquitylated. Whereas Thi7^KR-GFP, Thi7^CterKR-GFP, and Thi7^6KR-GFP remained nonubiquitylated in both conditions, expression of 3xHA-ART2 in art2Δ restored ubiquitylation of Thi7-GFP and Thi7^NterKR-GFP. Therefore, our data demonstrate that Thi7 is ubiquitylated in its cytosolic C-terminal tail and that its ubiquitylation is mainly Art2 dependent.

An oxythiamine-based screening revealed specific transport-defective Thi7 mutants

We focused next on the underlying mechanism of Thi7 endocytosis in response to thiamine transport. First, we found that Thi7-GFP is endocytosed after oxythiamine treatment of WT cells (S3 Fig), suggesting that substrate transport triggers Thi7-GFP endocytosis regardless of TPP-dependent enzyme activity. To address the underlying mechanism of Thi7 endocytosis in response to thiamine transport, an oxythiamine-based screening was completed in an attempt to isolate transport-defective Thi7 mutants. Thirteen single-point mutants were identified in this screen: Thi7^G59R, Thi7^T80K, Thi7^D85G, Thi7^S130N, Thi7^N133K, Thi7^M247I, Thi7^P286Q, Thi7^P291Q, Thi7^S338L, Thi7^N350K, Thi7^V398F, Thi7^M399R, and Thi7^A447P. In addition, we generated a Thi7^T287N mutant that in a previous study reduced thiamine uptake activity by 75%–80% [39]. All but one mutation appears to be localized within transmembrane segments (Fig 4A). The thi7Δnrt1Δthi72Δ strain was retransformed with each isolated Thi7 mutant, and phenotypic growth tests confirmed that their expression induced resistance to varying extents to oxythiamine (Fig 4B). The strains expressing Thi7^G59R, Thi7^D85G, Thi7^N133K, Thi7^P286Q, Thi7^N350K, or Thi7^M399R mutants appeared to be as resistant as the thi7Δnrt1Δthi72Δ strain up to 15 μM oxythiamine, suggesting a loss of function. At this concentration, growth defects are observed when thi7Δnrt1Δthi72Δ cells express Thi7^T80K, Thi7^S130N, Thi7^M247I, Thi7^T287N, Thi7^P291Q, Thi7^S338L, Thi7^V398F, and Thi7^A447P, indicating an intermediate functionality. Small variations of abundance between the mutants and WT Thi7 proteins were observed, but they were not correlated to resistance, confirming that the phenotypes resulted from partial functionality rather than impaired protein expression (S4 Fig). Finally, we verified that all mutants were correctly localized at the PM, except Thi7^T80K and Thi7^S338L, which were partially retained in the ER (S5 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Selection of 14 single-point Thi7 mutants with reduced transport activity and structural modelling of Thi7.

(A) Scheme of Thi7 topology generated by Protter software [43], representing the 14 single-point Thi7 mutants isolated by oxythiamine-based screening (red) or by side-directed mutagenesis (blue). The lysine residues mutated to generate Thi7^KR mutants are shown in purple. (B) Phenotypic growth test of thi7Δnrt1Δthi72Δ strain expressing single-point THI7-GFP mutants, WT THI7-GFP, or an e.v. on thiamine-free selective medium (SC-U-B1) supplemented or not with oxythiamine (final concentrations; 3, 6, and 15 μM). Representative of 5 experiments. (C) Phenotypic growth test of thi7Δnrt1Δthi72Δthi4Δ strain expressing single-point THI7-GFP mutants, WT THI7-GFP, or an e.v. on thiamine-free selective medium (SC-U-B1) or supplemented with thiamine (final concentrations; 10 nM, 1 μM, and 100 μM). Representative of 4 experiments. (D) A 3D model of Thi7 in an occluded state. (E) Proposed thiamine binding site in Thi7. Residues predicted to be at a distance equal to or less than 6 Å from the thiamine molecule are shown. Docking experiments predict that thiamine is stabilized by two hydrogen bridges with S55 and N137 (in purple) and one π-π interaction with Y346 (in green). Eight Thi7 mutants obtained in the oxythiamine-based screening result from mutation of specific residues surrounding the binding site (encircled in black: G59, D85, N133, M247, P286, N350, V398, M399). (F) Superimposition of the 3D model of WT Thi7 (residue N133 in yellow) in a thiamine-bound occluded state with the 3D model of Thi7^N133K (residue K133 in red) in an occluded state displays how the N133K mutation could affect the binding site and the stabilization of thiamine. (G) Superimposition of the 3D model of WT Thi7 (residue M399 in yellow) in a thiamine-bound occluded state with the 3D model of Thi7^M399R (residue R399 in green) in an occluded state displays how the M399R mutation could disturb the binding site and stabilization of thiamine. 3D, three-dimensional; C-ter, C terminus; e.v., empty vector; GFP, green fluorescent protein; KR, lysine-to-arginine; N-ter, N terminus; TM, transmembrane span; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000512.g004

The thi7Δnrt1Δthi72Δthi4Δ strain has impaired thiamine biosynthesis and transport. Therefore, this strain can be used to confirm whether the above mutants are deficient for thiamine transport by performing a phenotypic growth test. This strain was complemented with each Thi7 mutant and grown on solid medium containing a broad range of thiamine concentrations to determine which thiamine concentration sustains growth (Fig 4C, S6 Fig). As shown previously, thi7Δnrt1Δthi72Δthi4Δ cells grew only in the presence of the highest thiamine concentration (100 μM) because of unspecific thiamine entry. The growth of the thi7Δnrt1Δthi72Δthi4Δ strain complemented with Thi7^G59R, Thi7^T80K, Thi7^D85G, Thi7^S130N, Thi7^M247I, Thi7^P286Q, Thi7^T287N, Thi7^P291Q, Thi7^S338L, Thi7^V398F, or Thi7^A447P was restored at relatively low thiamine concentrations (10 nM), suggesting that their thiamine transport is sufficient to restore growth. The growth of the thi7Δnrt1Δthi72Δthi4Δ strain complemented with Thi7^N133K or Thi7^N350K was rescued at 1 μM thiamine, suggesting a much reduced thiamine transport (Fig 4C). Finally, the Thi7^M399R mutant was the most affected mutant and required at least 5 μM thiamine to restore growth of the thi7Δnrt1Δthi72Δthi4Δ strain (Fig 4C, S6 Fig). The inability of Thi7^N133K, Thi7^N350K, and Thi7^M399R mutants to sustain growth on low thiamine concentration is consistent with their strong resistance on oxythiamine-containing medium (Fig 4B).

We sought to gain insight into the structure of Thi7 to understand how these mutations could impact Thi7 transport. Thi7 belongs to the nucleobase-cation-symport-1 (NCS1) family of secondary active transporters, which includes the hydantoin transporter membrane hydantoin transport protein 1 (Mhp1) from Microbacterium liquefaciens [40]. The crystallized structure of Mhp1 displays 12 transmembrane helices. The transport mechanism, described by the alternating access model, implies structural changes from outward-facing (OF) to inward-facing (IF) conformations with different intermediate states adopted by the protein during the transition, among which at least one must be occluded [40–42]. Therefore, we generated multiple three-dimensional (3D) models of native and mutated Thi7. We modelized transporters in the OF open and occluded conformations as well as in the IF conformation. We also simulated thiamine docking within these models to predict the thiamine binding site (Fig 4D, S7 Fig). It appears that 8 mutants result from mutations of residues forming or surrounding the thiamine binding pocket: Thi7^G59R, Thi7^D85G, Thi7^N133K, Thi7^M247I, Thi7^P286Q, Thi7^N350K, Thi7^V398F, and Thi7^M399R (Fig 4E). Therefore, the partial functionality of these mutants can be explained by disturbances localized in the binding pocket that prevent the interaction and stabilization of thiamine. For instance, this is directly visible for the mutants Thi7^N133K and Thi7^M399R (Fig 4F and 4G). The remaining mutants are localized outside of the binding pocket and could result in structural changes altering the stability of transmembrane spans and loops as well as the dynamics occurring from an OF to an IF state and therefore impacting transport. Any interaction with proteins or lipids could be affected.

Thi7 mutants have impaired thiamine-induced endocytosis compared to WT Thi7

To correlate the partial functionality of these mutants with their endocytic behavior, we examined thiamine-induced endocytosis of the PM-localized Thi7 mutants in a thi7Δnrt1Δthi72Δ background. In thiamine-free medium, the Thi7 mutants are PM-localized with the appearance of a few intracellular punctate foci for Thi7^M247I and Thi7^A447P (Fig 5). After thiamine treatment, the majority of the mutants remained partially or totally within the PM compared to the WT Thi7, suggesting endocytosis defect (Fig 5). Thiamine-induced endocytosis of the most functionally affected mutants, Thi7^M399R and Thi7^N350K, was totally abolished (Fig 5). Other mutants, including Thi7^D85G and Thi7^P291Q, also resisted thiamine-induced endocytosis, as indicated by the low amount of intracellular signal after thiamine treatment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Single-point Thi7 mutants display thiamine-induced endocytic defects compared to WT Thi7.

Localization of Thi7 mutants in a thi7Δnrt1Δthi72Δ strain after thiamine addition (final concentration: 100 μM) for 3 h into culture grown in thiamine-free selective medium. Scale bar represents 5 μm. Quantification shows the ratio of internal-over-PM fluorescence intensity of all single-point Thi7 mutants before and after thiamine addition (100 μM) as described in the Materials and methods section (n > 69 cells per condition in 3 replicates). The upper graph is data not normalized, whereas the bottom graph is data normalized for each mutant by the ratio at time 0. (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05) The numerical data are included in S3 Data. GFP, green fluorescent protein; ns, nonsignificant; PM, plasma membrane; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000512.g005

Thiamine uptake through Thi7^6KR is sufficient to trigger endocytosis of Thi7^M399R and Thi7^N350K

Transporter endocytosis often relies on the accumulation of its substrate in the cytosol [26]. The resistance of the most functionally affected mutant Thi7^M399R to thiamine-induced endocytosis could be due to the low amount of thiamine in the cytosol. To address this question, we coexpressed Thi7^M399R-GFP with the active but endocytosis-resistant Thi7^6KR mutant (fused to DsRed) in thi7Δnrt1Δthi72Δ cells. When solely expressed, Thi7^6KR and Thi7^M399R failed to be endocytosed upon thiamine treatment (Fig 6A). However, coexpression of Thi7^6KR and Thi7^M399R led to endocytosis of Thi7^M399R after thiamine treatment, whereas Thi7^6KR remained localized at the PM. Neither Thi7^M399R nor Thi7 appears to interact with Thi7^6KR, as we did not detect vacuolar targeting of Thi7^6KR while endocytosed (Fig 6A and 6B). Similarly, although Thi7^N350K was functionally affected compared to the WT Thi7 and blocked at the cell surface when expressed alone, its coexpression with Thi7^6KR rescued its endocytosis (Fig 6C). Together, these results suggest that the increase of thiamine intracellular pool is necessary to induce Thi7 endocytosis in response to thiamine transport.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Thiamine uptake through Thi7^6KR leads to endocytosis of the inactive Thi7 transporters.

(A) Localization of Thi7^M399R-GFP and Thi7^6KR-mDsRed after incubation with thiamine (final concentration: 100 μM) in a thi7Δnrt1Δthi72Δ strain coexpressing THI7^M399R-GFP and THI7^6KR-mDsRed, or THI7^M399R-GFP with an empty vector (pRS313; Φ) or THI7^6KR-mDsRed with an empty vector (pRS316; Φ). Relative quantification of the internal-over-PM fluorescence intensity ratio of Thi7^M399R-GFP before and after incubation with thiamine (100 μM) when coexpressed with Thi7^6KR-mDsRed or an empty vector (pRS313; Φ). (B) Localization of Thi7-GFP and Thi7^6KR-mDsRed after incubation with thiamine (final concentration: 100 μM) in a thi7Δnrt1Δthi72Δ strain coexpressing THI7-GFP with an empty vector (pRS313; Φ) or THI7-GFP with THI7^6KR-mDsRed. Relative quantification of the internal-over-PM fluorescence intensity ratio of Thi7^M399R-GFP before and after incubation with thiamine (100 μM) when coexpressed with Thi7^6KR-mDsRed or an empty vector (pRS313; Φ). (C) Localization of Thi7^N350K-GFP and Thi7^6KR-mDsRed after incubation with thiamine at a final concentration of 100 μM in a thi7Δnrt1Δthi72Δ strain coexpressing THI7^N350K-GFP and either THI7^6KR-mDsRed or an empty vector (pRS313; Φ). Relative quantification of the internal-over-PM fluorescence intensity of Thi7^N350K-GFP before and after incubation with thiamine (100 μM) when coexpressed with either Thi7^6KR-mDsRed or an empty vector (pRS313; Φ). For quantification, the median with 95% confidence interval is presented (n > 94 cells per condition in 3 biological replicates) (****p < 0.0001, Wilcoxon test). Scale bar represents 5 μm. The numerical data are included in S4 Data. GFP, green fluorescent protein; ns, nonsignificant; PM, plasma membrane.

https://doi.org/10.1371/journal.pbio.3000512.g006

Increased thiamine intracellular pool is not sufficient to down-regulate Thi7^D85G and Thi7^P291Q

Thi7^M399R and Thi7^N350K resist thiamine-induced endocytosis because of the absence of intracellular accumulation of thiamine. Yet, Thi7^D85G and Thi7^P291Q, individually expressed, also resist thiamine-induced endocytosis although they seem to still transport thiamine, albeit with reduced functionality (Figs 4C, 5, 7A and 7B). Even when coexpressed with the active Thi7^6KR-DsRed in thi7Δnrt1Δthi72Δ cells, which fully restores the uptake of thiamine, no endocytosis was observed for these mutants (Fig 7A and 7B). Hence, in contrast to Thi7^M399R and Thi7^N350K mutants, the impaired thiamine-induced endocytosis of Thi7^D85G and Thi7^P291Q cannot be explained by only the lower intracellular pool of thiamine resulting from their reduced transport activity. Rather, these data suggest that these mutants adopt a specific conformation that might mask putative binding sites required for ubiquitylation and/or endocytosis. Interestingly, CHX treatment induces endocytosis of Thi7^M399R and Thi7^N350K as well as Thi7^D85G and Thi7^P291Q mutants (Fig 7C), suggesting another thiamine-independent pathway involved in the degradation of Thi7.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Uptake of thiamine through Thi7^6KR is not sufficient to induce endocytosis of Thi7^D85G and Thi7^P291Q.

(A) Localization of Thi7^D85G-GFP and Thi7^6KR-mDsRed after incubation with thiamine (final concentration: 100 μM) in a thi7Δnrt1Δthi72Δ strain coexpressing THI7^D85G-GFP and either THI7^6KR-mDsRed or an empty vector (pRS313; Φ). Relative quantification of the internal-over-PM fluorescence intensity of Thi7^D85G-GFP before and after incubation with thiamine (100 μM) when coexpressed with either Thi7^6KR-mDsRed or an empty vector (pRS313; Φ). (B) Localization of Thi7^P291Q-GFP and Thi7^6KR-mDsRed after incubation with thiamine (final concentration: 100 μM) in a thi7Δnrt1Δthi72Δ strain coexpressing THI7^P291Q-GFP and either THI7^6KR-mDsRed or an empty vector (pRS313; Φ). Relative quantification of the internal-over-PM fluorescence intensity of Thi7^P291Q-GFP before and after incubation with thiamine (100 μM) when coexpressed with either Thi7^6KR-mDsRed or an empty vector (pRS313; Φ). (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05) (C) Localization of Thi7^M399R-GFP, Thi7^N350K-GFP, Thi7^D85G-GFP, Thi7^P291Q-GFP, and Thi7-GFP after incubation with CHX 50 μg/ml for 3 h or EtOH in a thi7Δnrt1Δthi72Δ strain grown in thiamine-free medium. For quantification, the median with 95% confidence interval is presented (n > 101 cells per condition in 3 biological replicates) (*p < 0.05; Wilcoxon test). Scale bar represents 5 μm. The numerical data are included in S5 Data. CHX, cycloheximide; EtOH, ethanol; GFP, green fluorescent protein; KR, lysine-to-arginine; ns, nonsignificant; PM, plasma membrane.

https://doi.org/10.1371/journal.pbio.3000512.g007

Thiamine-induced endocytosis of Thi7 requires TORC1 and Sit4 but is not inhibited by Npr1

Results within this study suggest that thiamine-induced endocytosis of Thi7 is Art2 dependent. Although increased intracellular pool of thiamine partially explains Thi7 endocytosis in response to thiamine transport, how Art2-mediated ubiquitylation is stimulated by thiamine remains unclear. In the current model, transporter endocytosis often relies on the activation of α-arrestin via dephosphorylation [13,19,20,23]. The TORC1-Npr1-Sit4 pathway was shown to regulate Art1 and Bul α-arrestins in response to variations in nitrogen availability [29]. Moreover, CHX, which stimulates TORC1 activity, induces Art2-mediated endocytosis of thiamine transporters in the absence of thiamine. Therefore, we tested the role of this pathway in Thi7 endocytosis (Fig 8). In the sit4Δ strain grown in thiamine-free proline medium (low TORC1 activity) as well as thiamine-free ammonium-containing complete medium (high TORC1 activity), both the CHX- and thiamine-induced endocytosis of Thi7 was inhibited, indicating that the two processes are Sit4 dependent (Fig 8A). If thiamine-induced endocytosis of Thi7 is inhibited by Npr1, the deletion of NPR1 should lead to constitutive Thi7 endocytosis. Such a phenotype was observed for Gap1 in the npr1Δ mutant grown in proline medium and resulted from permanently dephosphorylated Bul1/2 [13]. However, the deletion of NPR1 did not lead to constitutive endocytosis of Thi7, which remains sensitive to CHX- and thiamine-induced down-regulation (Fig 8A). This suggests that Thi7 endocytosis is Npr1 independent. Yet, the rate of thiamine-induced endocytosis of Thi7 was higher in the WT than in the npr1Δ mutant, especially in thiamine-free ammonium-containing complete medium, suggesting that thiamine has a positive influence on the Npr1 kinase and subsequently on Thi7 endocytosis. Therefore, we examined the consequence of thiamine on Npr1 phosphorylation profile. Previous studies have shown that TORC1 activation leads to hyperphosphorylation and subsequent inactivation of Npr1 [44]. In WT cells grown in thiamine-free proline medium, the hemagglutinin (HA)-Npr1 construct migrated as two close bands below 100 kDa, regardless of thiamine presence. When cells were grown in ammonium-containing thiamine-free complete medium, HA-Npr1 is phosphorylated. Interestingly, addition of thiamine or oxythiamine led to apparent dephosphorylation of HA-Npr1, although only observable after 3 h of thiamine addition and not at early time points (Fig 8B, S8 Fig). In this growth medium, CHX treatment led to an upward shift to the HA-Npr1 protein, indicating a higher level of phosphorylation. Together, those results are consistent with the idea that thiamine might dephosphorylate Npr1, stimulate Art2, and induce Thi7 endocytosis in response to transport. Finally, we tested the influence of rapamycin, an inhibitor of TORC1 activity, on Thi7 localization. Previous studies reported that several permeases such as Gap1, Can1, lysine-specific permease 1 (Lyp1), Fur4, and Jen1 were endocytosed upon rapamycin treatment, although the underlying mechanisms of this stress-induced down-regulation remain poorly understood [21,45]. In thiamine-free medium, rapamycin treatment (100 and 200 ng/ml) induced Art2-dependent endocytosis of Thi7, yet the endocytosis rate was similar regardless of rapamycin concentration and always lower than the one observed upon thiamine addition (Fig 8C). A simultaneous treatment with rapamycin and thiamine also led to a lower endocytosis rate of Thi7, corresponding to the rate of rapamycin treatment alone. With the rate of Thi7 endocytosis being similar upon rapamycin treatment, regardless of thiamine addition, we suggest that the observed endocytosis was due to rapamycin only and that the thiamine-induced pathway that initiates Thi7 endocytosis was partially inhibited by rapamycin. Therefore, our data suggest that the thiamine-induced pathway could be TORC1 dependent. Taken together, these results indicate that thiamine-induced endocytosis of Thi7 requires the Sit4 phosphatase and might be TORC1 dependent. Interestingly, Npr1 does not inhibit Art2-dependent endocytosis. Our data rather suggest a positive role of Npr1 on Thi7 endocytosis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Thiamine-induced endocytosis of Thi7 appears to require TORC1 and Sit4 but is not inhibited by Npr1.

(A) Localization of Thi7-GFP in WT, sit4Δ, and npr1Δ strains complemented with the pFL36-HIS3-LYS2 or pFL36-HIS3-MET15 plasmid grown in thiamine-free medium containing proline (10 mM) as the sole nitrogen source (“Pro–Thiamine”) or on ammonium-(38 mM) containing thiamine-free complete medium (“Am + a.a.–Thiamine”). Cells were incubated for 3 h with or without thiamine (100 μM) or CHX (50 μg/ml). Relative quantification of the internal-over-PM fluorescence intensity of Thi7-GFP in the different conditions (n > 52 cells). (B) Addition of thiamine or oxythiamine leads to de-phosphorylation of HA-Npr1. A WT strain expressing HA-NPR1 and complemented with the pFL36 plasmid was grown up to early log phase in thiamine-free medium containing proline (10 mM) as the sole nitrogen source and incubated or not for 3 h with thiamine (100 μM), on thiamine-free medium containing ammonium (38 mM) as sole nitrogen source and incubated for 3 h with or without thiamine (100 μM) and ammonium-containing thiamine-free complete medium incubated for 3 h with thiamine (100 μM), oxythiamine (100 μM), CHX (50 μg/ml), or EtOH before being harvested. Cell extracts were immunoblotted with anti-HA and anti-Pma1 antibodies. Normalized intensities; distribution of HA-Npr1 in ammonium-containing thiamine-free complete medium with or without incubation with thiamine 100 μM, oxythiamine 100 μM, CHX 50 μg/ml, or EtOH. (C) Localization of Thi7-GFP in WT and art2Δ strains grown in ammonium-containing thiamine-free complete medium (“Am + a.a.–Thiamine”) and incubated for 3 h with or without DMSO, rapamycin 100 ng/ml, rapamycin 200 ng/ml, rapamycin 200 ng/ml + thiamine 100 μM, or thiamine 100 μM. Relative quantification of the internal-over-PM fluorescence intensity of Thi7-GFP in the different conditions. For quantification in (A) and (C), the median with 95% confidence interval is presented (n > 32 cells). Scale bar represents 5 μm (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05). The numerical data are included in S6 Data. CHX, cycloheximide; EtOH, ethanol; GFP, green fluorescent protein; HA, hemagglutinin; ns, nonsignificant; PM, plasma membrane; Pma1, plasma membrane ATPase1; TORC1, target of rapamycin complex 1; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000512.g008

Yeast strains and growth conditions

Yeast strains used in this study (S3 Table) are derivatives of Saccharomyces cerevisiae auxotrophic strain BY4742. The npi1–1 (rsp5) mutant strain, obtained by insertion of a selective marker in the RSP5 promoter and resulting in an attenuated RSP5 gene expression by a factor of 10, was provided by Pr. Bruno André (ULB, Belgium). The art2Δart9Δ strain was obtained by yeast crossing and tetrad dissection. The end3Δart2Δ strain was generated by homologous recombination using a natNT2 deletion cassette on an end3Δ background. The strains lacking genes encoding the thiamine transporters Thi7, Nrt1, and Thi72 and the gene encoding the thiazole synthase Thi4 were received from Pr. Jürgen Stolz (TUM, Germany). For proteomic analysis, yeast strains were grown in YD medium (2% glucose and 2% KAT yeast extract). For Thi7 endocytosis experiments, transformed yeast strains were grown in selective thiamine-free complete medium (2% glucose, 0.69% yeast nitrogen base [YNB] without amino acids and thiamine [ForMedium, Norfolk, United Kingdom], Drop-Out powder, supplemented with required amino acids [pH 5.2]). To study the influence of thiamine on TORC1/Npr1/Sit4 pathway, yeast strains were grown in minimal medium (2% [w/v] glucose, 10 mM proline [Sigma-Aldrich, Schnelldorf, Germany] as unique nitrogen source, 0.19% [w/v] YNB without amino acids, ammonium sulfate and thiamine). In case of ammonium-containing medium, 0.69% (w/v) YNB without amino acids and thiamine was used.

Plasmids

The plasmids used in this study are listed in S4 Table. We received from Pr. Jürgen Stolz (TUM, Germany) a high-copy plasmid expressing THI7-GFP under the control of the THI7 promoter from which we transferred the cassette into a pRS316 centromeric plasmid. Plasmids encoding N- and C-terminally GFP-fused versions of Thi7, Nrt1, and Thi72 were designed using Gibson assembly technology (NEB) [61]. KR mutants of THI7 were engineered using synthetic oligonucleotides (Integrated DNA Technologies, Leuven, Belgium) and Gibson assembly. Plasmids encoding monomeric DsRed-fused versions of thiamine transporters were designed using Gibson assembly. The plasmid expressing 3xHA-ART2 under control of the endogenous promoter was provided by Pr. Hugh Pelham (MRC, UK) and is based on the YCplac111 vector. In this study, THI7-GFP mutants were isolated from an oxythiamine-based screening. All these genes are controlled by the THI7 promoter and the ADH1 terminator. The QuikChange site-directed mutagenesis method was used to design THI7^T287N-GFP. All plasmids were verified by sequencing.

Oxythiamine-based screening

The thi7Δnrt1Δthi72Δ strain expressing THI7-GFP was grown overnight in a thiamine-free medium, plated on solid medium containing oxythiamine 3 μM and 6 μM, and grown for 3 days, and 156 spontaneous growing colonies were analyzed by epifluorescence microscopy based on protein expression and localization. Plasmids were extracted, purified in Escherichia coli, sequenced, and retransformed into the thi7Δnrt1Δthi72Δ and thi7Δnrt1Δthi72Δthi4Δ strains for characterization.

Proteomic analysis

WT and 12 knockout strains (artΔ) were grown at 28 °C to an optical density (OD_600nm) of 7 (3 × 10⁷ cells/ml) in 2 L of YD. Cultures of the 12 artΔ strains were incubated with CHX (Sigma-Aldrich, Schnelldorf, Germany) (solubilized in 100% ethanol) at a final concentration of 50 μg/ml for 90 min. As a control, an additional WT strain was mock treated with the same volume of 100% ethanol. The final concentration of ethanol in the medium was 0.5%. Cells were harvested and homogenized, and PM enrichment was performed. The PM enrichment protocol was previously developed in our laboratory [62–65]. Twenty micrograms of proteins were digested with trypsin and labeled with isobaric tags for absolute and relative quantification (iTRAQ, ABSciex) as previously described [63]. Labeled peptides were resuspended in 78 μl of a 2% (v/v) acetonitrile (ACN) and 0.1% (v/v) formic acid solution in order to obtain a final concentration of 2 μg/μl. Peptide separation was performed on an ultrahigh-performance liquid chromatography (UHPLC) (Eksigent 2D nano-ultra, Sciex). Two micrograms of labeled peptides were diluted in 5 μl of a 2% ACN–0.5% formic acid solution and injected on a C18 pre-column (Acclaim PepMap 100 C18, 2 cm, 100 μm, Thermo Scientific). Eluted peptides were immediately subjected to reverse phase (RP) chromatography on a C18 analytical column (Acclaim PepMap 100 C18, 75 μm id × 25 cm, Thermo Scientific) for 150 min at a flow rate of 300 nl/min using a gradient of ACN. The iTRAQ-labeled peptides were analyzed by electrospray ionization (ESI) TripleTOF 5600+ System (AB Sciex). The experiment was performed with three biological replicates of each artΔ strain and six biological replicates of the CHX- and mock-treated WT.

Peptide identification and quantification were performed using ProteinPilot 5.0 software, Paragon search algorithm (AB Sciex), and the S. cerevisiae UniProtKB protein database. The PM proteome of the CHX- treated WT strain was compared with the one of the mock-treated WT strain to select PM proteins whose abundance is affected upon CHX addition. The abundance of these proteins was also compared between the CHX-treated artΔ and WT strains to determine whether the α-arrestin gene deletion disrupts the effect of CHX. A protein abundance ratio between two conditions corresponds to the fold change of its proportion in the total amount of PM proteins between the two conditions because the same amount of protein of each condition was analyzed. The logarithmic mean of the replicates of each protein ratio and the corresponding pooled p-value were computed using R software. Briefly, the weighted average of p-value-corresponding z-scores were calculated (considering a normal distribution) using the number of peptides used for protein quantification as weights [66,67]. The average z-scores were converted to their original corresponding probability and corrected using the Bonferroni method. The significance of each ratio is taken into account during hypothesis testing (H₀: ratio ≠ 1), which is associated to a p-value. Red and blue ratios are significantly different from 1 (i.e., characterized by a p-value < 0.05). The false discovery rate (FDR) of the mass spectrometry analysis is lower than 1%, and all the proteins (Tables 1 and 2) were identified with an Unused ProtScore of at least 2 in the ProteinPilot software.

Total protein extraction

Yeast cells were grown at 28 °C to early exponential phase in 5 or 50 ml of thiamine-free selective medium. Endocytosis was triggered by addition of thiamine, oxythiamine (Sigma-Aldrich) (final concentration of 100 μM), or CHX (final concentration 50 μg/ml). After 2 h of incubation, cells were harvested by centrifugation at 3,000g for 5 min, washed with cold distilled water, and resuspended in 200 μl lysis buffer (250 mM sorbitol, 50 mM imidazole, 1 mM MgCl₂.6H₂O [pH 7.5]) containing phenylmethylsulfonyl fluoride (PMSF) at a final concentration of 1 mM, 0.1% of a cocktail of protease inhibitors (PICs) (Roche), and N-ethylmaleimide (NEM) at a final concentration of 10 mM. Cells were lysed with glass beads. Cell debris and unbroken cells were eliminated by centrifugation at 3,000g for 20 s. The supernatant was mixed with an equal volume of sample buffer (100 mM Tris HCl [pH 6.8], 4 mM EDTA, 4% SDS, 20% glycerol, and 0.002% bromophenol blue) containing 1% dithiothreitol (DTT) for immunoblotting. The protein concentration was determined using a bicinchoninic acid (BCA) assay [68].

Immunoprecipitation of Thi7-GFP

Yeast cells were grown at 28 °C to an OD_600nm of 1 in 500 ml of thiamine-free selective medium. Total protein extraction was performed as described above but using a Triton X-100 containing lysis buffer (1% Triton X-100, 50 mM Tris, 150 mM NaCl, 1 mM PMSF, 0.1% PIC, 10 mM NEM [pH 7.4]). Protein concentration of the extracts was quantified by a BCA assay. Total protein extracts were incubated with 5 mM n-Dodecyl β-D-maltoside (DDM) for 3 h at 4 °C. After centrifugation at 100,000g for 1 h, the supernatant was recovered, and immunoprecipitation was then carried out according to GFP-Trap manufacturer’s protocol (Chromotek, Planegg-Martinsried, Germany). Equilibration and washing buffers have the same composition as the lysis buffer, supplemented with 5 mM and 0.5 mM DDM, respectively. A volume of 25 μl of GFP-Trap magnetic agarose beads were mixed with 500 μl of equilibration buffer and separated from the supernatant, which was discarded. The protein lysate was added to the equilibrated beads and incubated for 2 h at 4 °C. Beads were harvested and the supernatant (flow-through) kept for immunoblotting. Beads were washed five times in washing buffer. Elution was carried out using a 0.2 M glycine solution (pH 2.5), supplemented with 1 mM PMSF, 0.1% PIC, 10 mM NEM, and 0.5 mM DDM. The eluate was immediately neutralized with 5 μl of 1 M Tris-containing neutralization solution (pH 10.4), supplemented with 1 mM PMSF, 0.1% PIC, 10 mM NEM, and 0.5 mM DDM.

Immunoblotting and antibodies

Proteins were separated by SDS-PAGE, transferred to a polyvinylidene fluoride membrane (Millipore) using the Trans-Blot Turbo Transfer System (Bio-Rad), and probed with a mouse monoclonal anti-GFP (Roche), rabbit polyclonal anti-GFP (Abcam), rat anti-HA (Roche), mouse anti-ubiquitin (Thermo Fisher Scientific), or rabbit monoclonal anti-Pma1 [64], followed by incubation with horseradish peroxidase–coupled anti-rat, anti-rabbit, or anti-mouse IgG antibodies (dilution 1:10,000) and chemiluminescence detection (Roche Diagnostics). Chemiluminescence was captured using an Amersham Imager 600 (GE Healthcare). Signals were quantified using Fiji software [69].

Growth tests on oxythiamine-containing medium

Transformed thi7Δnrt1Δthi72Δ strains were grown overnight in thiamine-free selective (SC-U-B1) liquid medium. Cultures were adjusted to an OD_600nm of 0.4, and four 10-fold serial dilutions were spotted onto SC-U-B1 solid medium containing oxythiamine at different concentrations.

Growth tests on thiamine-containing medium

Transformed thi7Δnrt1Δthi72Δthi4Δ strains were grown overnight in thiamine-free liquid medium supplemented with 10 nM of thiamine. Cultures were adjusted to an OD_600nm of 0.4 and resuspended in thiamine-free selective (SC-U-B1) liquid medium. Then, 10-fold serial dilutions were spotted onto SC-U-B1 solid medium containing thiamine at different concentrations.

Fluorescence microscopy

Yeast cells were grown at 28 °C up to exponential phase in thiamine-free selective medium. The addition of thiamine, oxythiamine (final concentration of 100 μM), CHX (final concentration of 50 μg/ml), or rapamycin (100 or 200 ng/ml) triggered endocytosis. After 0 and 180 min, 5 μl cellular suspension was dropped onto a microscope slide covered by agarose 1.5% (w/v). The fluorescence analysis was performed with a Leica DMR microscope equipped with a 100× oil-immersion objective and a Hamatsu Orca- ER digital camera. Cells were observed at room temperature using appropriate filter sets. For FM4–64 vacuolar staining, cells were incubated at 28 °C for 45 min with 8 μM FM4–64 dye (Thermo Scientific), washed twice with thiamine-free selective medium, and incubated for endocytosis experiments. For each sample, a corresponding Nomarski picture was taken. Images were processed with HOKAWA! Imaging pf07 2.1 software. In some experiments, fluorescence quantification was performed using Fiji software [69].

Fluorescence quantification

The fluorescence intensity of Thi7-GFP and mutants was quantified using the Fiji platform from ImageJ. Two ellipses were manually drawn, with the first one surrounding the cell and the second one delimiting the inside, excluding the PM. The median with 95% confidence interval of the ratio of internal-over-PM intensity for each condition is presented. We measured internal instead of vacuolar fluorescence intensities to consider internalized transporters localizing in nonvacuolar intracellular compartments. Wilcoxon tests were used to assess the significance of internal-over-PM fluorescence ratio. In case of multiple tests with a same set of data, p-values were adjusted using the Holm method.

Structural modeling of Thi7 and docking of thiamine

A BLAST search performed on the Protein Data Bank (PDB) sequences identified the benzyl-hydantoin transporter Mhp1 sequence as similar to Thi7. An alignment between Thi7 and Mhp1 sequences was performed with PROMALS3D, also using information from the crystal structure of Mhp1 [70]. Five 3D models of Thi7 were built by comparative modeling with Modeller9v11 [71] using the OF open, occluded (substrate-bound), and IF open crystal structures of Mhp1 (PDB codes 2JLN, 4D1B, and 2X79) as templates [40–42] based on the PROMALS3D alignment. In each case, the modelization of Thi7 was performed from residue A22 to residue Y532. Docking of thiamine (PubChem code 1130) was performed in the occluded models of Thi7 with Glide [72]. Thiamine input file was prepared with the LigPrep module of Schrödinger. Prior to docking, transporter preparation was carried out using the Protein Preparation Wizard protocol (Schrödinger software). A grid receptor box with the dimensions adapted to the benzyl-hydantoin was generated. Hydroxyls of serine, threonine, or tyrosine and thiol of cysteine residues in the binding site were specified as rotatable groups. The Glide SP function was chosen to score the docking poses of the ligand. The remaining parameters were set at their default values. The Schrödinger maestro graphics interface was used to analyze all poses and models.