The Saccharomyces cerevisiae vacuolar acid trehalase is targeted at the cell surface for its physiological function


  • Susu He,

    1.  University of Toulouse, INSA, UPS, INP & INRA, France
    2.  INRA-UMR 792 Ingénierie des Systèmes Biologiques et procédés, Toulouse, France
    3.  CNRS-UMR 5504, Toulouse, France
    Search for more papers by this author
  • Kerstin Bystricky,

    1.  Laboratoire de Biologie Moléculaire Eucaryote, University of Toulouse, France
    2.  CNRS-UMR5099, Toulouse, France
    Search for more papers by this author
  • Sebastien Leon,

    1.  Institut Jacques Monod, UMR7592 CNRS/Université Paris Diderot, France
    Search for more papers by this author
  • Jean M. François,

    1.  University of Toulouse, INSA, UPS, INP & INRA, France
    2.  INRA-UMR 792 Ingénierie des Systèmes Biologiques et procédés, Toulouse, France
    3.  CNRS-UMR 5504, Toulouse, France
    Search for more papers by this author
  • Jean L. Parrou

    1.  University of Toulouse, INSA, UPS, INP & INRA, France
    2.  INRA-UMR 792 Ingénierie des Systèmes Biologiques et procédés, Toulouse, France
    3.  CNRS-UMR 5504, Toulouse, France
    Search for more papers by this author

J. M. François, University of Toulouse, INSA, UPS, INP & INRA, 135, Avenue de Rangeuil, F-31077, Toulouse, France
Fax: +33 5 6155 9400
Tel: +33 5 6155 9492


Previous studies in the yeast Saccharomyces cerevisiae have proposed a vacuolar localization for Ath1, which is difficult to reconcile with its ability to hydrolyze exogenous trehalose. We used fluorescent microscopy to show that the red fluorescent protein mCherry fused to the C-terminus of Ath1, although mostly localized in the vacuole, was also targeted to the cell surface. Also, hybrid Ath1 truncates fused at their C-terminus with the yeast internal invertase revealed that a 131 amino acid N-terminal fragment of Ath1was sufficient to target the fusion protein to the cell surface, enabling growth of the suc2Δ mutant on sucrose. The unique transmembrane domain appeared to be indispensable for the production of a functional Ath1, and its removal abrogated invertase secretion and growth on sucrose. Finally, the physiological significance of the cell-surface localization of Ath1 was established by showing that fusion of the signal peptide of invertase to N-terminal truncated Ath1 allowed the ath1Δ mutant to grow on trehalose, whereas the signal sequence of the vacuolar-targeted Pep4 constrained Ath1 in the vacuole and prevented growth of this mutant on trehalose. Use of trafficking mutants that impaired Ath1 delivery to the vacuole abrogated neither its activity nor its growth on exogenous trehalose.


endoglycosidase H


glycolsyl hydrolase


green fluorescent protein


multivesicular body




Trehalose [alpha-d-glucopyranosyl (1→1) alpha-d-gluocopyranoside] is a nonreducing disaccharide found in many organisms including yeasts, fungi, bacteria, plants and insects. Trehalose is one of the major storage carbohydrates in the yeast Saccharomyces cerevisiae, accounting for > 25% of cell dry mass depending on the growth conditions and the life-cycle stage of the yeast [1–3]. The accumulation of intracellular trehalose has two potential functions. First, it constitutes an endogenous storage of carbon and energy during spore germination and in resting cells. Second, trehalose acts as a stabilizer of cellular membranes and proteins [4–6].

In S. cerevisiae, trehalose is hydrolyzed to glucose by the action of two types of trehalase: ‘neutral trehalases’ encoded by NTH1 and NTH2 [3,7], which are optimally active at pH 7, and ‘acid trehalases’ encoded by ATH1, which show optimal activity at pH 4.5 [8]. Although fungal acid trehalases, including those of the yeast Candida albicans [9] and Kluyveromyces lactis [10], have been reported to be localized at the cell surface, the localization of the S. cerevisiae acid trehalase remains a matter of controversy. In 1982, Wiemken and co-workers [11] first identified this protein in a vacuole-enriched fraction obtained by density gradient centrifugation of a yeast protoplast preparation. The vacuolar localization of acid trehalase was very recently supported by in vivo imaging analyses using green fluorescent protein (GFP)–Ath1 fusion constructs under the strong and constitutive TPI1 promoter [12]. Furthermore, Huang et al. used various trafficking mutants to show that this acid trehalase reaches its vacuolar destination via the multivesicular body (MVB) pathway. However, this localization contrasts with the fact that this enzyme allows yeast to grow on exogenous trehalose [13] and with measurable Ath1 activity at the cell surface [14].

The purpose of this study was to revisit this controversy regarding the localization of Ath1 in light of its biological function, combining cell biology and biochemical approaches. To this end, we investigated the localization of Ath1 using strains expressing the red fluorescent protein mCherry fused to the C-terminus of Ath1. Integration of this construct at the ATH1 locus had the advantage of expressing the protein at levels comparable with those in wild-type cells, because it is reported that overexpression may cause the mislocalization of proteins into the vacuoles [15], and also to investigate the fusion protein under physiological conditions. The domain responsible for targeting Ath1 at the cell surface and the role of the single transmembrane (TM) domain at the N-terminus of this protein were investigated. The functional localization of Ath1 was further assessed by constructing various Ath1 hybrid proteins bearing different targeting signal peptides. Together, our results demonstrated that the localization of Ath1 at the cell periphery is required for growth on trehalose, whereas the vacuolar localization of this protein is not compatible with growth on this carbon source.


Ath1 is localized at the cell periphery

In a previous report, the localization of S. cerevisiae Ath1 was visualized using a pGFPATH1 construct that expressed a GFP fused to the N-terminus of Ath1 under the strong TPI1 promoter [12]. We obtained a comparable result with a GFP–Ath1 construct that was expressed under the control of the methionine-repressible MET25 promoter in a glucose medium lacking methionine (Fig. 1A). However, western blotting using a GFP antibody on extracts from cells expressing GFP–Ath1 revealed a major band migrating at a position corresponding to ∼ 30 kDa, instead of bands migrating at > 150 kDa (Fig. 1B). Fluorescence in the vacuole may therefore be caused by free GFP which accumulated in this organelle because it has been reported that targeting of GFP-fusion proteins to the vacuolar lumen leads to their degradation by vacuolar proteases. However, this degradation process is usually delayed, leading to the transient accumulation of GFP-containing proteolytic fragments of ∼ 30 kDa, and a sustained luminal vacuolar fluorescence [16]. Note that a similar result was reported by Huang et al. [12], although they were also able to detect a band corresponding to the native GFP–Ath1.

Figure 1.

 Cellular localization of Ath1. Yeast cells in the exponential phase (D600 ∼ 1.0) expressing fusions of Ath1 to GFP or mCherry were collected for live cell microscopy (A,C,D) and western blotting (B,E). Bar length = 2 μm. (A,B) BY4742 ath1Δ transformed with pGFP–ATH1 under the MET25 promoter in YN glucose without methionine. (A) (left) fluorescence, (right) DIC. (B) Immunoblot with anti-GFP of crude extract before (−) and after (+) N-deglycosylation by EndoH; (C) BY4741 bearing ATH1–GFP integrated at the ATH1 locus in YN trehalose. (D,E) BY4741 bearing ATH1–mCherry integrated at the ATH1 locus in YN trehalose. (D) Microscopy and (E) immunoblot with the DsRed polyclonal antibody of crude extract before (−) and after (+) N-deglycosylation by EndoH. M, Molecular mass marker; arrowhead, expected full-length fusion protein at 150 kDa; asterisk, degradation product.

This proteolytic problem, coupled with the fact that overexpression under a strong promoter has been reported to mislocalize some proteins into vacuoles [15], prompted us to re-examine the localization of Ath1 by fusing of GFP to its C-terminus, and expressing the corresponding ATH1–GFP fusion gene under the native promoter after integration at ATH1 locus. Under this condition, we were able to observe a green signal at the cell periphery, although most of the signal was still localized in the lumen of the vacuole (Fig. 1C). Similar results were obtained using the red fluorescent protein mCherry, which was also integrated at the ATH1 chromosomal locus, as well as with the tag fused at the N-terminus of Ath1 (data not shown). As for Ath1–GFP or GPF–Ath1 (see above), the Ath1–mCherry fusion protein was fully functional as indicated by the growth of this recombinant strain on trehalose and by enzymatic measurement (see below). Under live cell fluorescence microscopy, we observed a strong signal in the vacuolar compartment together with a clearly discernable signal at the cell periphery (Fig. 1D). These results indicated that Ath1 may have two localizations, one in the vacuole, in agreement with previous studies [11,12], and another at the cell periphery, in accordance with its ability to hydrolyze exogenous trehalose [14]. We then verified the Ath1–mCherry fusion protein by western blot. This analysis made on extracts from yeast cells expressing the chimeric protein revealed a band at a size > 200 kDa with the rabbit anti-DsRed sera (Fig. 1E). Because this signal disappeared upon endoglycosidase H (EndoH) treatment, the glycosylation that was reported for this protein [7] may explain this migration property at an apparent size much higher than expected. However, the expected band at a size of 164 kDa (Ath1 + mCherry) was barely detected upon EndoH treatment, and instead, a relatively strong band migrating at around 65 kDa could be identified (Fig. 1E).

As a second, independent way to support the localization of Ath1 at the cell periphery, we used the invertase secretion system. Invertase is a secreted protein with a classical signal peptide at its N-terminus (amino acids 1–19) for secretion at the cell periphery. Deletion of this signal peptide (suc2ic allele) prevents secretion and results in the accumulation of the truncated form of the enzyme in the cell, impairing the ability of S. cerevisiae to grow on sucrose or raffinose as the sole carbon source. We generated an inframe fusion of full-length ATH1 and suc2ic (pSC1–ATH1), leading to the chimeric Ath1–Suc2 protein expressed under the ATH1 promoter. As shown in Fig. 2, suc2Δ mutant expressing this gene construct recovered growth on sucrose, like the positive control expressing the full-length secreted invertase under its own promoter (pLC1), whereas suc2Δ mutant transformed with pSC1 lacking of signal peptide grew very poorly on sucrose, probably using amino acids present in the medium (Fig. 2B). Consistent with this, these cells also recovered invertase activity in both crude extract and intact cells (Fig. 3), albeit five times lower than that measured in suc2Δ mutant transformed with pLC1. Replacing the ATH1 promoter in pSC1–ATH1 with the stronger SUC2 promoter resulted in an invertase activity similar to that in pLC1 (data not shown). We noticed that the invertase activity in a crude extract of cells transformed with pLC1 was lower than that in intact cells. This may be caused by incomplete lysis of the cells or partial denaturation of proteins during extraction and vortexing with glass beads

Figure 2.

 Complementation of the S. cerevisiae SEY6210 strain (suc2Δ mutant) with different Ath1–invertase chimera. (A) Schematic representation of the different gene fusion constructs. pSC1, negative control (invertase without signal peptide); pLC1, positive control (full-length invertase); for the remaining constructs, the Suc2 signal peptide has been replaced by full-length ATH1 sequence (pSC1–ATH1) or N-terminal sequence variants of ATH1 with decreasing size; (B) growth on YP medium with sucrose (complementation test) or glucose (control) for 5 days.

Figure 3.

 Invertase activity in the S. cerevisiae SEY6210 strain (suc2Δ mutant) transformed with the various Ath1–invertase constructs. The constructs are those shown in Fig. 2. Transformed cells were cultivated in YP sucrose medium until the stationary phase before measurement of invertase activity on intact cells and in crude extract as described in Experimental procedures. Histograms show the results of two independent experiments (mean ± SD).

In addition to the cell biology data, we also revalidated our enzymatic assay of acid trehalase. Our current method is based on the measurement of the activity in intact cells according to the procedure employed to measure secreted invertase [17], in which NaF is added to the incubation mixture to block glucose uptake. We verified that the use of NaF did not cause any enzymatic artifact, for example, cell lysis or the release of intracellular glucose. First, incubation of intact cells from an exponential culture grown on glucose that do not express acid trehalase because of glucose repression [18] in a reaction mixture optimal for neutral trehalase activity and containing NaF did not lead to any glucose production from trehalose (data not shown). This excluded the possibility of cell leakage and the release of proteins or intracellular glucose under NaF treatment. Further validation of our assay was the successful measurement of acid trehalase activity on intact cells from a mutant completely defective for glucose uptake (hxt1-17Δ strain) [19] cultivated on glycerol and ethanol as the carbon source, which allowed ATH1 expression, even in the absence of NaF (data not shown). These elements demonstrated that the glucose measured in intact cells resulted from cleavage of the disaccharide by an acid trehalase localized at the cell surface.

Searching for the minimal domain of Ath1 for invertase secretion

Full-length Ath1–invertase fusion protein was targeted at the cell surface, suggesting the existence of a secretion sequence in Ath1. As shown in Fig. 4, domain prediction using the smart program [20,21] did not reveal any classical signal peptide for secretion at the N-terminus of Ath1. This in silico analysis only revealed a short 23 amino acid TM domain near the N-terminus, followed by three ‘glycosyl hydrolase’ (GH) domains (amino acids 132–415, 474–845 and 849–904) that together may constitute the catalytic domain of Ath1 [22]. To map the minimal domain of Ath1 that allows the secretion of this protein, various DNA fragments of ATH1 were fused inframe with the suc2ic allele (Fig. 2A). A series of plasmids, namely pSC1–N that carried a fusion to the first 131 N-terminal amino acids of Ath1, pSC1–TM bearing a fusion to the first 69 amino acids of Ath1, which includes the TM domain, and pSC1–tm that only bears the first 46 amino acids of Ath1 excluding the TM domain, were introduced into the suc2Δ mutant SEY6210. Transformants were tested for growth recovery on sucrose (Fig. 2B) and for invertase activity (Fig. 3). As shown in Fig. 2B, suc2Δ mutant cells transformed with pSC1–N or pSC1–TM were able to grow on YP sucrose as readily as pSC1–ATH1, whereas cells transformed with pSC1–tm poorly grew on sucrose, as did cells bearing the negative control pSC1.

Figure 4.

 Ath1 predicted functional domain using the smart program. Theoretical glycosylation sites (yellow triangles); N-terminal transmembrane segment (TM); N-term (GH_65N), central (GH_65m) and C-term (GH_65C) domains from the CAZy glycoside hydrolase family 65. The latter three domains likely constitute the catalytic core of trehalase.

Invertase activity was measured in intact cells and crude extracts from suc2Δ mutant transformed with these various constructs, compared with growth efficiency on sucrose (Figs 2 and 3). Cells transformed with pSC1–N showed an activity nearly twofold higher than that in cells expressing a fusion to the full-length Ath1 (pSC1–ATH1). One explanation might be that the full size Ath1 fused to internal invertase somehow impairs folding of the invertase domain and/or the catalytic efficiency on its substrate. Despite this difference, as for pSC1–ATH1, the activity in intact cells was comparable with that in cell extract, and both transformed cells showed similar qualitative growth on sucrose. The activity measured in intact cells expressing pSC1–TM was four times lower than that in the crude extract, and two to four times lower than in intact cells transformed with pSC1–ATH1 and pSC1–N. Bearing in mind this low activity, pSC1–TM transformed cells were found to grow slightly more slowly on sucrose than cells transformed with pSC1–ATH1. Further reduction at the N-terminus (i.e. with pSC1–tm) resulted in residual invertase activity in the crude extract, together with an inability to grow on sucrose. Taken together, these results showed that a minimal fragment of 69 amino acids encompassing the unique TM domain of Ath1 was needed to promote correct expression of the internal invertase, but was not sufficient for efficient protein secretion, which was achieved with a 131 amino acid N-terminus of Ath1.

Removal of the N-terminus of Ath1 caused a strict vacuolar localization

Because the 131 amino acid N-terminus of Ath1 appeared to be sufficient for invertase secretion, we further investigated the targeting properties of this fragment by using a mCherry fusion that was expressed under the control of the ATH1 promoter (pN–mCherry). Figure 5A shows a fluorescent signal at the cell periphery and a stronger signal in the vacuole, similar to that observed using full-length Ath1 fused to mCherry (compare Figs 5A and 1D). This result confirmed that the N-terminal part of Ath1 was sufficient to target the recipient protein to these two cellular compartments.

Figure 5.

 The N-terminal domain of Ath1 is needed for the localization of mCherry at the cell periphery. Live cell microscopy of exponentially growing cells in YN trehalose medium of the BY4741 strain transformed with pN–mCherry (A) or with path1ΔN–mCherry (B).

Reciprocally, we analyzed the consequences of deleting the first 100 codons of the ATH1 sequence (path1ΔN) on red protein localization. When expressed in a wild-type strain grown on trehalose, the Ath1ΔN–mCherry fusion protein led to a fluorescent signal exclusively in the vacuole (Fig. 5B). No discernable signal could be detected at the cell periphery, even after 10-fold longer exposure times. From this result, we first verified that a BYath1Δ mutant transformed with the centromeric plasmid pATH1 carrying the wild-type ATH1 gene recovered wild-type characteristics, i.e. growth on trehalose as the sole carbon source (not shown), and acid trehalase activity in both intact cells and cell crude extracts (Fig. 6). However, when this ath1Δ mutant was transformed by path1ΔN it was not able to grow on trehalose (data not shown) and had no Ath1 activity (Fig. 6). From these data, we were able to confirm that the 131 amino acid N-terminal fragment contains important information for cell-surface targeting, and we suggest that there may be vacuolar targeting determinants in the catalytic domain, as in the case of acid phosphatase [23].

Figure 6.

 Acid trehalase activity of ath1Δ mutant cells expressing various ATH1 constructs. The BY4741 ath1Δ mutant strain transformed with gene constructs expressing different Ath1 variants was cultivated in YN glucose to late exponential phase (D600 ∼ 8) with plasmid selection. The cells were then collected and transferred to YPD medium until the stationary phase (D600 ∼ 20) to allow ATH1 derepression. Acid trehalase activity was measured in intact cells and crude extracts as described in Experimental Procedures. Histograms show the results of two independent experiments (mean ± SD).

Substitution of the N-terminus of Ath1 by the invertase signal peptide restored acid trehalase activity and growth on trehalose

The exclusive, strong vacuolar signal observed in the absence of the 100 amino acid N-terminus of Ath1, together with the subsequent loss of catalytically active trehalase (Ath1ΔN variant), suggested that the vacuolar fraction consisted mainly of inactive Ath1. We therefore asked whether targeting of Ath1 to the cell periphery could restore trehalase activity. We made use of the invertase secretion property by fusing the signal peptide of this protein to the N-terminus of the Ath1ΔN variant Fig. 7A). When transformed in ath1Δ mutant cells, the resulting plasmid pSPSUC2–ATH1ΔN did allow recovery of the growth ability on trehalose and the acid trehalase activity in both cell crude extract and intact cells (Fig. 6). Moreover, the ath1Δ mutant strain bearing this plasmid grew about two times faster than wild-type BY4741 strain on synthetic trehalose medium (μ = 0.10 versus 0.047; Fig. 8). Localization of this hybrid protein was verified by C-terminal fusion to mCherry. Setting our exposure time as in Fig. 1, we found that the intensity of the fluorescent signal at the cell periphery was significantly higher than that of the full-length Ath1–mCherry protein (compare Figs 7B and 1D). However, the bulk of the fluorescent signal still resided in the vacuolar compartment, which substantiated the idea that the catalytic domain of Ath1 contains some targeting signal for the vacuole. Using western blot analysis, we found a 65kDa proteolytic fragment that was already obtained with the Ath1–mCherry fusion protein (Fig. 1C), but also a clearly detectable band corresponding to the SPSuc2–Ath1ΔN–mCherry chimeric protein after EndoH treatment (173 kDa, Fig. 7C), indicating better stability for this construct than for native Ath1. Overall, these results suggest that secretion of Ath1 at the cell periphery is associated with the stabilization and physiological function of this protein.

Figure 7.

 Cellular localization of the Ath1 catalytic core fused to different signal peptides. (Panel I, A) Schematic representation of SUC2, PEP4 and ATH1 nucleotide sequences with emphasis on the 5′-end containing targeting information. Nucleotides 1-111 and 1-267 of SUC2 and PEP4, respectively, replaced nucleotides 1-300 of ATH1 giving the SPSUC2–Ath1ΔN and SPPEP4–Ath1ΔN chimeras. (Panel II) BY4741 cells were transformed with pSPSUC2–ath1ΔN–mCherry and cultivated in YN trehalose medium. Exponential growing cells were collected for live cell microscopy (B), and immunoblot on crude extract before (−) and after (+) deglycosylation with EndoH using DsRed polyclonal antibody (C). (D) BY4741 cells transformed with pSPPEP4–ath1ΔN–mCherry were collected for live cell microscopy. M, molecular mass markers; arrowhead, expected full-length fusion protein; asterisk, degradation product.

Figure 8.

 Growth complementation of the ath1Δ mutant strain on trehalose. After a preculture on a selective YN glucose medium, the BY4741 ath1Δ mutant strain expressing the catalytic domain of Ath1 (amino acids 301 to 1211) fused to signal sequence of Suc2 and Pep4, respectively, were transferred in YN trehalose medium to evaluate growth complementation. BY4741, positive control; ath1Δ, negative control.

Constraining Ath1 to the vacuole impaired growth on trehalose

Although Ath1 can be targeted to the cell periphery, the vacuolar localization appeared to be the major destination for this protein, as illustrated by the strong vacuolar signal obtained using fluorescence microscopy. To check the possible function of the vacuolar pool of acid trehalase for growth on trehalose, we sought a strategy to constrain all Ath1 in this intracellular compartment. To this end, we fused the signal peptide of the vacuolar protein Pep4 [24] to the N-terminus of the truncated Ath1ΔN variant. Very interestingly, when transformed in ath1Δ mutant cells, the plasmid pSPPEP4–ATH1ΔN did not allow recovery of the growth on trehalose (Fig. 8), although the cells did exhibit acid trehalase activity in the crude extract, which accounted for ∼ 50% of the activity measured in cells expressing SPSuc2–Ath1ΔN (data not shown). As shown in Fig. 7D, microscopy analysis confirmed that the SPPep4–Ath1ΔN–mCherry chimeric protein was exclusively targeted to the vacuole when expressed in the wild-type strain. This strongly indicated that the vacuolar pool of acid trehalase has no role in trehalose assimilation for cell growth.

As a complementary approach, we used mutants of genes involved in the vacuolar sorting pathway, like VPS4 which encodes a protein implicated in the delivery of proteins from the prevacuolar compartment to the vacuole [25]. As shown in Fig. 9A, the intracellular red fluorescent signal derived from Ath1–mCherry was totally mislocalized in a vps4Δ mutant, being completely excluded from the lumen of vacuole. However, the fluorescent signal at the cell periphery was still visible in this vps4Δ mutant and the relative Ath1 activity between intact cells and crude extract was identical to that of wild-type cells (Fig. 9B). The presence of the Ath1–mCherry fusion protein was also monitored in this mutant using the rabbit anti-DsRed sera. In untreated extract, a band migrating at ∼ 200 kDa was relatively comparable in this mutant and the wild-type (data in Figs 9C and 1C can be compared because similar amount of protein were loaded). After EndoH treatment, the expected 164 kDa band was visible, whereas the abundance of the 65 kDa band was drastically reduced compared with that in Fig. 1C, indicating significantly decreased proteolysis of this protein when preventing vacuolar targeting. Together, these results confirmed that trehalase in the vacuole is likely prompted to partial degradation and is not required for cell growth on trehalose.

Figure 9.

 Localization and activity of Ath1 in vps4Δ mutant. Cells cultivated in YN trehalose medium to the exponential phase were collected for live cell microscopy (A), and to test the activity of acid trehalase in intact cells or cell crude extracts (B), as described in Experimental Procedures. Histograms show the results of two independent experiments (mean ± SD). (C) Crude extract from vps4Δ cells expressing Ath1–mCherry immunoblotted with the DsRed polyclonal antibody, before (−) and after (+) deglycosylation with EndoH. M, molecular mass markers. Bar = 2 μm.

The TM domain is indispensable for Ath1 function

Previous studies have indicated that the short TM domain located at the N-terminus of Ath1 contained sufficient signaling information to deliver Ath1 to the vacuole via the MVB pathway [12]. As already observed when studying invertase fusions, the requirement for a minimal N-terminal fragment encompassing the TM domain indicated the importance of this domain in protein expression and secretion (see the minimal construct pSC1–tm in Figs 2 and 3). We confirmed this by studying pSC1–ath1ΔTM and pSC1–NΔTM, in which the TM domain was specifically deleted in the full-length ATH1–SUC2 gene fusion and in its truncated variant, respectively. When transformed in suc2Δ mutant, these constructs did not restore growth on sucrose or invertase activity (Fig. 10A). Similarly, when using BYath1Δ mutant as a recipient strain for functional complementation by various Ath1 variants, the plasmid path1ΔTM, which expressed an Ath1 protein lacking the TM domain, was not able to complement growth deficiency of this mutant on trehalose or yield measurable Ath1 activity in this strain (data not shown). Finally, when using the pNΔTM–mCherry plasmid that expressed a fusion of the N-terminal fragment lacking the TM domain to mCherry, the fluorescence was observed in cytoplasmic patches distinct from the vacuole (Fig. 10B).

Figure 10.

 Role of the single transmembrane (TM) domain in protein expression. (A) Left, schematic view of chimera proteins Ath1ΔTM–invertase and NΔTM–invertase, respectively. Right, complementation tests of Suc2Δ mutant by these two constructions on YP sucrose agar for 5 days, and invertase activity (IA). (B) BY4741 cells transformed with plasmid pNΔTM–mCherry were cultivated in YN trehalose medium to the exponential phase and collected for live cell imaging. (C) The HA–ATH1 or HA–ATH1ΔTM gene constructs expressing Ath1 with or without the TM sequence tagged with HA under the GAL1 promoter were transformed into ath1Δ mutant cultivated in YN galactose. EndoH-treated crude extracts were immunoblotted with the anti-HA IgG. Lane 1, wild-type Ath1 (negative control); lane 2, HA–Ath1; lane 3, HA–Ath1ΔTM. M, molecular mass markers.

The absence of Ath1 activity in crude extracts when TM was deleted from the protein prompted us to verify whether removal of this short TM domain may hamper expression of these constructs. For this purpose, Ath1 and its ath1ΔTM variant were tagged with 3HA at their N-terminus. The ath1 mutant transformed with pHA–ATH1 expressing the Ath1–HA fusion protein recovered growth on trehalose, although the chimeric protein could not be detected by western blotting, probably because of its very low expression level. We therefore replaced the ATH1 promoter with the strong, inducible GAL1 promoter, leading to very high Ath1 activity in cells transformed with pPGAL1–HA–ATH1 (data not shown). By contrast, no activity was measured in cells transformed with pPGAL1–HA–ath1ΔTM, although the gene construct was expressed (data not shown). These results were confirmed by western blot analysis using anti-HA IgG, which revealed a band at ∼ 130 kDa (wild-type Ath1) after EndoH treatment of protein extracts from cells expressing pPGAL1–HA–ATH1; no band was detected when the TM domain was missing from the protein (Fig. 10C). These results supported the idea that absence of the TM domain may lead to a deficiency in protein production, which likely occurred during the early steps of endoplasmic reticulum protein synthesis and/or during folding.


Vacuolar Ath1 is also found at the cell surface

Controversy concerning the localization of Ath1 has been raised in two recent papers. In a previous study, we suggested a localization for Ath1 at the cell surface based on enzymatic data because most Ath1 activity could be measured in intact cells [14], in a manner similar to that for the secreted invertase [17]. However, Huang et al. [12] provided several arguments for a strict vacuolar localization of Ath1, identifying the MVB pathway as the main transport route for sorting this protein into the vacuole. In this paper, we used two independent methodologies, fluorescence microscopy and gene fusion to invertase, which together provided evidence that Ath1 is also targeted to the cell surface. Using the GFP or the red fluorescent protein mCherry fused to the C-terminus of Ath1, we clearly observed a localization of Ath1 at the cell periphery, although the bulk fluorescent signal was still seen in the vacuole. A possible reason for the failure of Huang et al. [12] to find Ath1 at the cell periphery may be that these authors used a GFP–Ath1 construct that was expressed from the strong constitutive TPI1 promoter, because we obtained similar results using GFP–Ath1 expressed from another strong MET25 promoter. However, we examined the localization of Ath1 in cells expressing either Ath1–GFP or Ath1–mCherry cultivated on trehalose, whereas Huang et al. [12] investigated this localization problem using exponentially growing cells on glucose. It can be proposed that the correct localization of Ath1 is dependent on the substrate (in this case, trehalose), as shown for the control of Fur4 permease by uracile [26]. When expressed under its own promoter, as in our study, ATH1 is repressed by glucose [18] and the localization of Ath1 can be examined only in the stationary phase. Thus, the use of a glucose medium to study the localization of Ath1 can be cautioned because it is not physiologically relevant for this protein. Further evidence for a cell-surface localization of Ath1 was obtained by showing that expression of the Ath1–Suc2 protein fusion allowed recovering suc2Δ mutant to grow on sucrose, indicating that the full-length Ath1 protein was able to drive the yeast internal invertase to the cell surface. These cell biology data were further supported by the revalidation of our enzymatic assay of acid trehalase on intact cells, confirming that glucose measured in NaF-treated intact cells results from the cleavage of the disaccharide at the cell surface by an extracellular ‘acid trehalase’ pool [14].

The cell-surface localization accounts for growth on trehalose

It is known that Ath1 hydrolyzes exogenous trehalose to grow on this carbon source. Based on an exclusive vacuolar localization for this protein, two models have been proposed [27]. The first suggested that Ath1 is transported to the plasma membrane where it binds to trehalose located at the cell surface; both trehalose and trehalase are then internalized by endocytosis into the vacuole where hydrolysis takes place. According to the results of Huang et al. [12], this model may be discarded because transport of Ath1 via the MVB pathway en route to the vacuole bypasses the plasma membrane. The second model considered that trehalose alone is delivered to the vacuole by endocytosis, where it is hydrolyzed by the resident Ath1. However, this model requires the identification of a trehalose endocytosis process and this is difficult to reconcile with mono- and disaccharides entering the cell by sugar permeases [19], and yeast cells possessing a high-affinity trehalose transporter encoded by AGT1 [28]. Instead, we provide arguments that support a more simple model [14], in which trehalose can be assimilated by either a Agt1–Nth1 pathway, implicating the uptake and intracellular hydrolysis by neutral trehalase, or by direct hydrolysis of trehalose by the extracellular acid trehalase encoded by ATH1 into glucose, which is thereafter taken up by the cells. These two pathways only function in a MAL-positive strain such as the CEN.PK background because expression of AGT1 is MAL dependent. Because the sequenced BY4741 strain is mal-negative, the assimilation of exogenous trehalose can rely only on the Ath1-dependent pathway [14]. Moreover, this model is consistent with what has been shown for fungal and plant acid trehalases, which are all localized at the cell surface or cell wall [22,29,30]. In addition to these data, other results support this model. First, constraining acid trehalase in the vacuole by replacing its 100 amino acid N-terminal fragment with the signal sequence of the vacuolar Pep4 [24], a protein known to be specifically targeted to the vacuole, prevented growth on trehalose. Second, impairment of Ath1 delivery to the vacuole using vps4Δ mutants defective in the MVB pathway did not abrogate growth on trehalose or the activity of Ath1 on intact cells.

Although Ath1 is present at the cell periphery, our data,together with those from Huang et al. [12], showed an apparent large accumulation of this protein in the vacuole, as monitored by the fluorescence intensity from GFP- or mCherry-tagged protein. However, this result contrasted with enzymatic data showing that Ath1 activity measured in crude extract was only 20–40% higher than that measured in intact cells. One explanation for this discrepancy can be found from western blot analysis in which full-length Ath1 fused to reporter mCherry was barely detected, whereas a partially proteolysed Ath1 fragment was predominantly observed. Also, use of a vps4Δ strain impaired Ath1 delivery to the vacuole and significantly reduced its proteolysis. Similar observations were obtained with the vps1Δ strain (S. He, unpublished), which was initially identified as a protein involved in transport from the late-Golgi complex to the prevacuolar compartment [31] in the vacuole protein-sorting pathway. To summarize, these results demonstrated that the vacuole is not the obligate functional destination for Ath1, and that partial proteolysis of Ath1 could take place in this subcellular compartment. In contrast, targeting this enzyme at the cell surface is indispensable for growth of yeast cells on trehalose.

Ath1 domains relevant for cell-surface targeting and protein function

The finding that Ath1 could be targeted at the cell periphery raised questions about secretion determinants because domain-predicting tools did not identify any sequence feature to explain Ath1 intracellular trafficking. Klionsky and co-workers [12] showed that the short TM domain located at the N-terminus of Ath1 contained sufficient information to deliver Ath1 to the vacuole via the MVB pathway. They reached this conclusion using a chimeric construct in which only the TM domain was fused to GFP. Alternatively, we specifically removed the unique TM domain from full-length Ath1 or from the 131 amino acid N-terminal fragment fused to Suc2, and found that absence of this TM domain abrogated the activity of invertase and growth on sucrose. More remarkably, removal of TM in Ath1 led to a complete loss of enzyme activity and the inability of a HA antibody to detect the HA–Ath1ΔTM construct. Because we were able to verify that the absence of Ath1 protein was not caused by inefficient ATH1 transcription (not shown), these results suggested a critical function for the TM domain in the translation and/or stabilization of Ath1 during early secretion steps. This also fits with the mislocalization of the NΔTM–mCherry chimera in cytosolic patchy bodies, whose origin is currently unknown.

As indicated by hybrid Ath1–invertase fusions, a 131 amino acid N-terminal fragment was needed to recover normal invertase secretion, whereas reducing this N-terminal fragment to only 69 amino acids decreased the secretion and activity of invertase at the cell surface. Taking this result together with those using the reporter protein mCherry, the minimal information for correct targeting to the cell surface is likely localized between amino acids 69 (after the TM domain) and 131 of Ath1 protein sequence. Several intracellular enzymes in yeast, in particular the glycolytic enzymes glyceraldehyde dehydrogenase [32], 3-phosphoglycerate mutase [33] and enolase [34,35], were found to be secreted at the cell surface although they did not harbor any classical signal sequence for secretion. Nombela et al. [36] proposed that these signalless proteins could be exported by nonclassical export systems, such as those identified in mammals and parasites, which involve membrane blebbing (bubble formation) and secondary-structure elements that might also contribute to export [37]. A common feature between these glycolytic enzymes and S. cerevisiae Ath1 is the lack of a classical secretion sequence. However, because Ath1 is not a cytosolic protein, these modes of secretion remain unknown. By contrast, the classical secretion pathway cannot be excluded because it was reported that mutations that cause accumulation of secretory proteins in the endoplasmic reticulum (sec18) or in the Golgi apparatus (sec7) led to diminished Ath1 activity [38,39]. Also, previous findings of co-purification of Ath1 with cell-surface secreted proteins such as invertase [7,40] and Ygp1 [41] further supported this mode of secretion. In conclusion, the secretion pathway for Ath1 needs to be thoroughly reinvestigated using specific mutants altered in various secretion processes.

Experimental procedures

Strains, media and culture conditions

BY4741 (MAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ0), BY4742 (MAT α his3-Δ1 leu2-Δ0 lys2-Δ0 ura3-Δ0) and SEY6210 (MAT a his3-Δ200 leu2-3,112 lys2-801 trp1-901 ura3-52 suc2-Δ9) were used as recipient strains for various gene constructs, as described in Table 1. Yeast transformation was performed according to the lithium acetate method, as described in Woods & Gietz [42]. The vps4Δ mutant used in this study was derived from the Euroscarf deletion collection (BY background). The ath1Δ null mutant was constructed by replacing the gene of interest with selective cassettes KanMX4 using in vivo homologous recombination. Unless otherwise stated, yeast cells were cultured in yeast nitrogen base (YN) synthetic medium (0.17% w/v yeast nitrogen base without amino acid and without ammonium, supplemented with 0.5% ammonium sulfate w/v, buffered to pH 4.8 with sodium succinate/NaOH and with the auxotrophic amino acids when required). A carbon source glucose, galactose, sucrose or trehalose was added up to 2% (w/v). Cultures were carried out at 30 °C in shaking flasks at a shaking speed of 170 rpm·min−1.

Table 1.   Strains used in this study. Euroscarf, Institute of Molecular Biosciences, Johann Wolfgang Goethe-University Frankfurt, Germany; H. Bussey, McGill University, Québec, Canada.
BY4741MAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ0Euroscarf
BY4742MATα his3-Δ1 leu2-Δ0 Lys2-Δ0 ura3-Δ0Euroscarf
SEY6210MAT a his3-Δ300 leu2-3,112 lys2-801 trp1-901 ura3-52 suc2-Δ9Gift of H. Bussey
BY4741 ath1ΔMAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ0 ath1Δ::KanMX4This study
BY4741 vps4ΔMAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ vps4Δ::KanMX4Euroscarf
BY4741 ATH1_GFPMAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ0 ATH1-GFP-His3MX6This study
BY4741 ATH1_mCherryMAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ0 ATH1-mCherry-His3MX6This study
BY4741 vps4Δ ATH1_mCherryMAT a his3-Δ1 leu2-Δ0 ura3-Δ0 met15-Δ0 vps4Δ::KanMX4 ATH1-mCherry-His3MX6This study
hxt1-17ΔMATα hxt1-17 gal2[19]

Plasmids construction

To construct N-terminal truncated versions of Ath1, primers ATH1_−1000_BH and ATH1_+508 (for primers list see Table 2) were used to amplify a DNA fragment carrying the ATH1 gene and its promoter and terminator from extracted genomic DNA of BY4741. This PCR product was first cloned in pGEM-T-easy vector and a cut BamHI/PstI fragment was inserted into centromeric YCplac33 (linearized by BamHI and PstI) to construct pATH1 (for plasmids list see Table 3). Mutagenesis of the TM domain of Ath1 was carried out using the four nucleotides recombinant PCR method [43]. Use of ATH1_A and ATH1_D external primers, together with the internal mutagenic primers ATH1_B and ATH1_C, led to the deletion of nucleotides 139–207 that encode the TM domain of Ath1. The recombinant PCR fragment was cloned into the pGEM-T-easy vector and cut by AgeI/AflII digestion to replace the AgeI/AflII fragment in pATH1, which yielded path1ΔTM. The same method was used to construct path1ΔN, with the primers ATH1_E, ATH1_F, ATH1_G and ATH1_D that lead to the deletion of nucleotides 1-300 of ATH1 sequence.

Table 2.   Primer sequences for PCR. Restriction sites are shown in bold, underlined and homologue recombination region in italics.
NameOligo sequence
Table 3.   Plasmids used in this study. GFP, green fluorescent protein; TM, transmembrane.
YCplac33Centromeric plasmid[45]
pLC1Centromeric plasmid, containing SUC2 ORF[46]
pSC1Centromeric plasmid, containing suc2ic allele without sequences coding signal peptide[46]
pFA6a–GFP–His3MX6As a PCR template for amplifying GFP sequence[47]
pFA6a–mCherry–His3MX6As a PCR template for amplifying mCherry sequenceGift of S. Bachellier-Bassi
pFA6a–3HA–KanMX6As a PCR template for amplifying 3HA sequence[47]
pFA6a–KanMX6–PGAL1As a PCR template for amplifying the promoter GAL1[47]
pUG36For a N-terminal GFP fusion construction under promoter MET25Gift of Hegemann JH
pATH1ATH1 ORF with its promoter and terminator cloned in YCplac33This study
pGFP–ATH1To overexpress the chimeric protein GFP–Ath1This study
path1ΔNATH1 variant without 5′-end 300 nucleotides cloned in YCplac33This study
path1ΔTMATH1 variant without 5′-end 139-207 nucleotides coding TM domain (aa 47-69) cloned in YCplac33This study
pSPSUC2–ath1ΔNTo express a chimeric protein with the signal peptide of invertase fused to Ath1ΔNThis study
pSPPEP4–ath1ΔNTo express a chimeric protein with the signal peptide of Pep4 fused to Ath1ΔNThis study
pPGAL1–HA–ATH1To express a chimeric protein with HA tag in the N-terminus of Ath1This study
pPGAL1–HA–ath1ΔTMTo express a chimeric protein with HA tag in the N-terminus of Ath1ΔTMThis study
path1ΔN–mCherryBearing mCherry at the 3′-end of ath1ΔNThis study
pSPSUC2–ath1ΔN–mCherryBearing mCherry at the 3′-end of SPSUC2-ath1ΔNThis study
pSPPEP4–ath1ΔN–mCherryBearing mCherry at the 3′-end of SPPEP4-ath1ΔNThis study
pSC1–ATH1ATH1 ORF fused to 5′-end of suc2ic alleleThis study
pSC1–NATH1 5′-end 395 nucleotides fused to 5′-end of suc2ic alleleThis study
pSC1–TMATH1 5′-end 209 nucleotides fused to 5′-end of suc2ic alleleThis study
pSC1–tmATH1 5′-end 140 nt fused to 5′-end of suc2ic alleleThis study
pSC1–ath1ΔTMath1ΔTM fused to 5′-end of suc2ic alleleThis study
pSC1–NΔTMATH1 5′-end 395 nucleotides with a gap of 139-207 nucleotides coding the TM domain (amino acids 47-69) fused to 5′-end of suc2ic alleleThis study
pN–mCherryATH1 5′-end 395 nucleotides fused to 5′-end of mCherryThis study
pNΔTM–mCherryATH1 5′-end 395 nucleotides with a gap of 139-207 nucleotides coding TM domain (amino acids 47-69) fused to 5′-end of mCherryThis study

To fuse the signal peptide of Suc2 to the catalytic domain of Ath1, the following constructions were carried out using the centromeric plasmid pLC1 containing the SUC2 gene (1602 bp) flanked by its own promoter [44]. The pSPSUC2–ath1ΔN plasmid was constructed by replacing the fragment coding the catalytic domain of invertase, which starts from the 112th nucleotide to the stop codon (remaining 5′-end fragment including the region coding signal peptide of Suc2) of SUC2 in pLC1 by the ath1 allele without its 5′-end 300 bp (Fig. 7). To construct pSPPEP4–ath1ΔN, the SUC2 ORF in pLC1 was replaced by the PEP4 (1218 bp) ORF, which was amplified using the primers PEP4_D and PEP4_R. The 3′-end (951 bp) fragment encoding the catalytic domain of Pep4 (amplified by using primers ATH1_pep4 and ATH1_pLC1), was removed and replaced by the ath1 allele without its 300 bp 5′-end in order to yield pSPPEP4–ath1ΔN.

Plasmids bearing Ath1-truncated fusion proteins inframe to the intracellular invertase encoded by suc2ic allele were constructed by using another centromeric plasmid pSC1 containing a suc2ic allele lacking its signal sequence [44]. Using the plasmid pATH1 as the template, PCR fragments containing 1000 bp of ATH1 promoter sequence and part of the 5′-end of ATH1 coding sequence were obtained using ATH1_−1000_BH as the forward primer and the following reverse primers: ATH1_3633_BH for amplification of full-length ATH1 coding sequence (without the stop codon); ATH1_395_BH for the ‘N’ construct that carries an allele version of ATH1 that stops just before the catalytic domain of Ath1 (amino acid 131); ATH1_209_BH for the ‘TM’ construct that stops just after the TM domain of Ath1 (at amino acid residue 69); and ATH1_140_BH for the ‘tm’ construct that stops just before the TM domain (amino acid residue 41); Similarly, using path1ΔTM as template, ATH1_−1000_BH forward primer together with ATH1_3633_BH and ATH1_395_BH were used to obtain ‘ath1ΔTM’ and ‘NΔTM’ constructs, respectively. In order to achieve in frame fusion with suc2ic allele, all these PCR fragments were cloned in pGEM-T-easy vector and excised by BamHI digestion for subcloning into the BglII site of pSC1 to produce plasmids pSC1–ATH1, pSC1–N, pSC1–TM, pSC1–tm, pSC1–ath1ΔTM and pSC1–NΔTM, respectively.

Ath1 was tagged with 3HA at the N-terminal end by inserting 3HA after the start codon ATG of ATH1. For this purpose, two rounds of the recombinant PCR were successively carried out. First, we fused ATH1 promoter (primers ATH1_1 and ATH1_2) and the 3HA (primers HA_D and HA_R, using pFA6a–3HA–KanMX6 as template), using ATH1_1 and HA_R as external primers. Second, this recombinant PCR product was fused to an ATH1 5′-end PCR product (primers ATH1_3 and ATH1_4) using ATH1_1 and ATH1_4 as external primers. This final HA-tagged PCR fragment was cloned into the pGEM-T-easy vector and was then excised by SnaBI/AgeI to replace the SnaBI/AgeI fragment in pATH1 and path1ΔTM, respectively, to obtain pHA–ATH1 and pHA–ath1ΔTM.

Using the plasmid pFA6a–KanMX6–PGAL1 as the template, the primers PGAL_D and PGAL_R were used to amplify a GAL1 promoter PCR cassette that was co-transformed into yeast cells with SnaBI-linearized plasmids pHA–ATH1 and pHA–ath1ΔTM, respectively. Cells carrying recombinant plasmids pPGAL1–HA–ATH1 or pPGAL1–HA–ath1ΔTM, which express the HA-tagged versions of Ath1 under the strong promoter GAL1 instead of the native promoter, were selected in the absence of uracil.

Construction of fluorescent fusion proteins

ATH1 was amplified from the plasmid pATH1 using the primers ATH1_pUG36_D and ATH1_pUG36_R. This PCR product was first cloned in pGEM-T-easy vector and a cut SpeI/SmaI fragment was inserted into plasmid pUG36 (linearized by SpeI and SmaI) to construct pGFP–ATH1. The GFP–His3MX6 or mCherry–His3MX6 cassette that contains the gene encoding GFP or mCherry was amplified from plasmid pFA6a–GFP–His3MX6 or pFA6a–mCherry–His3MX6 (kind gift of S. Bachellier-Bassi, Institut Pasteur, Paris, France). Primers F2_ATH1 and R1_ATH1 were used to amplify the GFP–His3MX6_ATH1 and mCherry–His3MX6_ATH1, which were integrated into the genome of the wild-type strain BY4741 or vps4Δ mutant by homologous recombination, for C-terminal tagging of Ath1 with GFP or mCherry. The path1ΔN–mCherry vector was constructed by in vivo homologous recombination after co-transformation of yeast cells with the mCherry–His3MX6_ATH1 PCR cassette together with plasmids path1ΔN, and selection of the recombinant plasmid in the absence of both uracil and histidine. Similarly, co-transformation of a mCherry–His3MX6 PCR cassette that was obtained from primers F2_ATH1 and R1_pLC1, together with pLC1 derivative plasmids described above, led to mCherry–tagged versions of Ath1 chimeric variants, i.e. pSPSUC2–ath1ΔN–mCherry and pSPPEP4–ath1ΔN–mCherry.

The two plasmids pN–mCherry and pNΔTM–mCherry were obtained by replacing the suc2ic allele sequence by mCherry in plasmids pSC1–N and pSC1–NΔTM. This was carried out by co-transformation into yeast cells of a mCherry PCR cassette obtained from primers mCherry–pSC1_D and mCherry–pSC1_R, together with AgeI-linearized pSC1–N and pSC1–NΔTM, respectively.

Western blotting

Crude cell extract was prepared in the same way as crude extract for trehalase activity measurement [14] with additional protease inhibitor (Roche, Basel, Switzerland, NO.11836170001). Crude extract containing tagged proteins was first treated with EndoH for 3 h at 37 °C. Western blots were performed using the primary mouse mAb anti-HA IgG (Roche, No. 11583816001) at a dilution of 1/2000 or mouse mAb anti-GFP IgG (Roche, NO. 11814460001) at a dilution of 1/1000 or rabbit living colors DsRed polyclonal antibody (Clontech, Palo Alto, CA, USA, NO.632496) at a dilution of 1/1000, and the secondary antibody horseradish peroxidase-conjugated goat anti-mouse or rabbit IgG at a dilution of 1/20000 supplied in SuperSignal West Pico Complete Mouse (Pierce, Rockford, IL, USA, NO. 34081) or rabbit (Pierce, NO. 34084) IgG Detection Kit.

Fluorescence and microscopy

Fluorescent protein tagged cells were cultivated in YN trehalose or glucose medium to reach the exponential phase, and then cells were collected by centrifugation (3000 g, 5 min). Images were captured on a Metamorph driven Olympus IX81 wide-field microscope equipped with a Coolsnap HQ camera and a Polychrome V (Till Photonics, Munich, Germany). A 100×/1.4 Oil Plan-Apochromat objective from Zeiss was used. Exposure times were 500 ms for GFP (excitation λ = 490 nm) and 2000 ms for mCherry (excitation λ = 590 nm). Images were minimally adjusted for brightness and contrast using photoshop.

Assay of trehalase and invertase activity

Yeast cells (D ∼ 100) were harvested by centrifugation (3000 g, 5 min) and washed twice. Activity of acid trehalase and invertase on intact cells and in crude extract was measured as described in [14]. The activity was expressed as nmol of glucose released from either trehalose or sucrose per minute and per D600.


We thank S. Bachellier-Bassi for generously providing us with the pFA6a–mCherry–His3MX6 plasmid. Microscopy analyses were performed at the RIO microscopy facility in Toulouse, France. This work was partially supported by ANR Blanc grant n° 05-2-42128 to JMF. SH holds a fellowship for PhD students from the Research Grants China Scholarship Council.