A Family of Yeast Proteins Mediating Bidirectional Vacuolar Amino Acid Transport*

Seven genes in Saccharomyces cerevisiae are predicted to code for membrane-spanning proteins (designated AVT1–7) that are related to the neuronal γ-aminobutyric acid-glycine vesicular transporters. We have now demonstrated that four of these proteins mediate amino acid transport in vacuoles. One protein, AVT1, is required for the vacuolar uptake of large neutral amino acids including tyrosine, glutamine, asparagine, isoleucine, and leucine. Three proteins, AVT3, AVT4, and AVT6, are involved in amino acid efflux from the vacuole and, as such, are the first to be shown directly to transport compounds from the lumen of an acidic intracellular organelle. This function is consistent with the role of the vacuole in protein degradation, whereby accumulated amino acids are exported to the cytosol. Protein AVT6 is responsible for the efflux of aspartate and glutamate, an activity that would account for their exclusion from vacuoles in vivo. Transport by AVT1 and AVT6 requires ATP for function and is abolished in the presence of nigericin, indicating that the same pH gradient can drive amino acid transport in opposing directions. Efflux of tyrosine and other large neutral amino acids by the two closely related proteins, AVT3 and AVT4, is similar in terms of substrate specificity to transport systemh described in mammalian lysosomes and melanosomes. These findings suggest that yeast AVT transporter function has been conserved to control amino acid flux in vacuolar-like organelles.

Caenorhabditis elegans UNC-47 (1) and the vertebrate homologues from rat (VGAT, 1) and mouse (VIAAT, 2) are synaptic vesicular transporters that are expressed exclusively in inhibitory neurons (see also Ref. 3) and are specific for the neurotransmitters ␥-aminobutyric acid (GABA) 1 and glycine. These proteins differ in sequence, structure, and bioenergetics from the previously characterized family of vesicular transporters that package monoamines and acetylcholine (for current reviews, see Refs. 4 and 5). Common to all of these transporters is the fact that the movement of substrate from the cytosol into synaptic vesicles is driven by a proton electrochemical gradient that is generated by the action of a vacuolar-type H ϩ -ATPase and involves the exchange of lumenal protons.
A search for other vertebrate proteins related to the GABAglycine vesicular transporters has led recently to the isolation and expression of cDNAs that, surprisingly, code for Na ϩcoupled plasma membrane carriers. The system N transporters from rat (SN1, 6) and mouse (NAT1, 7) are expressed in the liver, kidney, and heart and within brain astrocytes and are specific for glutamine, histidine, and to a lesser degree, asparagine. Amino acid transport by these proteins is unique in that it also involves an exchange of protons (6,8), the mechanism utilized by vesicular neurotransmitter transporters. In addition, two transporters with ϳ50% identity to SN1/NAT1 have now been identified and categorized as system A due to their ability to transport ␣-methylaminoisobutyric acid (9 -12). One protein (termed SA1, ATA2, or SAT2) is expressed in most tissues and, compared with system N, has a wider substrate specificity that includes glutamine as well as most of the smaller neutral amino acids. The second system A carrier, referred to as GlnT, transports a similar set of amino acids but is highly expressed in the brain, specifically within neurons that release the excitatory neurotransmitter glutamate (11).
Other characterized members of this growing protein family include the group of plant amino acid/auxin permeases, of which 12 proteins have now been identified in Arabidopsis thaliana (13). These permeases actively transport amino acids into the plant cell and presumably utilize an amino acid/H ϩ symport mechanism.
The vacuole of Saccharomyces cerevisiae is a highly complex organelle that is involved in both the enzymatic degradation of proteins and the homeostatic control of a vast assortment of solutes including ions and amino acids (for a review, see Ref. 14). Like synaptic vesicles, vacuoles maintain an acidic internal environment through the action of a vacuolar-type H ϩ -ATPase that generates a proton electrochemical gradient of 180 mV (⌬pH of 1.7 units) across the vacuolar-membrane (15). In mammals, the equivalent to vacuoles and the major site for protein degradation is the lysosome. Both organelles support a variety of carrier-mediated transport systems, many of which rely on the imposed pH gradient as a driving force. Although several amino acid transport systems have been described in purified vacuoles (16 -18) and lysosomes (for a review, see Ref. 19), the proteins responsible for these activities have yet to be identified.
Sequence analysis of the complete S. cerevisiae genome has predicted seven yeast proteins of unknown function that are related to the UNC-47/VGAT/VIAAT family. Because these proteins do not resemble any of the characterized permeases required for the cellular uptake of amino acids, we have investigated the role of this new family of yeast proteins in vacuolar amino acid transport.  Table II for specific gene loci) with the corresponding known homologues from Schizosaccharomyces pombe (GenBank™ accession numbers are shown). Also included are the GABA vesicular transporters from C. elegans (UNC-47; GenBank™ accession number AF031935) and rat (VGAT; GenBank™ accession number AF030253), as well as the system N (SN1; GenBank™ accession number AF273025) and system A (GlnT; GenBank™ accession number AF075704) plasma membrane transporters. Overall identities for the more closely related pairs are 52% for AVT5 and AVT6, 50% for SN1 and GlnT, 38% for AVT3 and AVT4, and 35% for UNC-47 and VGAT. AVT7 shares 35% sequence identity with both AVT5 and AVT6. B, dendrogram of proteins from C. elegans and Drosophila that show the highest BLAST alignment scores to both AVT3 and AVT4, compared with UNC-47. The GanBank™ accession numbers for hypothetical nematode proteins T27A1.5 and Y43F4B.7 are AAB71045.1 and T26845, respectively, whereas those for the predicted fly proteins are indicated. C, sequence alignment of AVT1-7 with VGAT showing identical (black) and conserved (gray) residues. D, sequence alignment of the proteins shown in B, excluding UNC-47.  Table I. Yeast strains were grown at 30°C in either YPD or selective SD medium (20). To knock out each of the AVT genes, the entire coding region was replaced with the appropriate selective marker gene. Marker genes were polymerase chain reaction-amplified from the plasmids pRS423 (HIS3), pRS424 (TRP1), pRS425 (LEU2), or pRS426 (URA3) (gifts of Joachim Lee) with 60-mer oligonucleotides of which 40 nucleotides corresponded to either the 5Ј-or 3Ј-flanking portion of the AVT gene to be disrupted. Transformation of the appropriate haploid yeast strains (21) was carried out with gel-purified polymerase chain reaction products followed by selection on SD medium. Correct homologous recombination events were confirmed by chromosomal polymerase chain reaction analysis of the isolated transformants.
Purification of Yeast Vacuoles-Yeast vacuoles were isolated using a modified version of the procedure described by Kakinuma et al. (15). Briefly, 500-ml cultures were grown in YPD medium to an A 600 of 3.5-4.0, and the cells were collected by centrifugation at 4,400 ϫ g for 5 min. Cells were washed once with 30 ml of distilled water at room temperature and converted to spheroplasts by resuspending them in 30 ml of 1 M Sorbitol and adding 500 l of a 4 mg/ml zymolyase 100T (ICN) solution in 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 50% glycerol. The culture was shaken gently for 1 h at 30°C, and spheroplasts were collected at 2,200 ϫ g for 5 min at 4°C and washed twice in 20 ml of ice-cold 1 M Sorbitol. All subsequent manipulations were carried out at 4°C. Spheroplasts were lysed by resuspension in 16 ml of Buffer A (10 mM PIPES, pH 6.9, 0.1 mM MgCl 2 , 12% Ficoll 400, and 0.42 mM Pefabloc SC) and homogenized by 15 strokes in a 15 ml Dounce with tight pestle. The lysate was cleared by centrifugation at 2,200 ϫ g for 10 min, and 12 ml of the supernatant was divided between two 89 ϫ 14-mm polyallomer tubes. Each sample was overlaid with another 6 ml of Buffer A and centrifuged at 60,000 ϫ g for 30 min in a Beckman SW41 Ti rotor. The floating vacuoles were collected and completely suspended in 400 l of 2ϫ Buffer C (1ϫ Buffer C ϭ 10 mM PIPES, pH 6.9, 5 mM MgCl 2 , and 12.5 mM KCl) and 0.42 mM Pefabloc SC before diluting with an equal volume of 1ϫ Buffer C and 0.42 mM Pefabloc SC. The final vacuole suspension, which was typically at a concentration of 1.25-1.5 g/l as determined by the Bradford assay, was either used immediately or frozen at Ϫ80°C in 200-l aliquots.
Amino Acid Transport Assays-Standard reaction conditions included 10 mM PIPES, pH 6.9, 4 mM MgCl 2 , 4 mM KCl, 4 mM ATP, and 50 M unlabeled amino acid with 1 Ci of 3 H-amino acid and 10 -20 g of purified vacuoles per 100 l of reaction volume. Where indicated, nigericin was also included at a final concentration of 6.5 M. Generally, reaction mixtures minus vacuoles were preincubated for 1 min at 30°C. After the addition of vacuoles, duplicate (or quadruplicate in the case of aspartate and glutamate) 100-l samples were removed at the appropriate times. Vacuoles were recovered quickly by suction on a membrane filter (Millipore HA; 0.45 M) and washed immediately three times with 2 ml of ice-cold 1ϫ Buffer C. Filters were dried, and the incorporated radioactivity was measured by scintillation counting in 5 ml of Filtron-X (National Diagnostics). All radiolabeled amino acids were obtained from New England Nuclear Laboratories.

RESULTS
A Family of Fungal Genes Related to Vesicular GABA Transporters-A homology search of the complete S. cerevisiae genome with VGAT sequences identified seven open reading frames that we have designated AVT1-7 for amino acid vacuolar transport (see below). Along with VGAT and its orthologues, the yeast AVT proteins belong to a large family of proteins that includes the amino acid/auxin permease family of plants (13), ϳ12 other predicted open reading frames from both C. elegans and Drosophila, and the mammalian system A and system N Na ϩ -dependent plasma membrane transporters (6 -12). Like other members of this family, the yeast proteins contain a core region of ϳ400 -450 amino acids that harbors 9 -11 predicted hydrophobic transmembrane domains. In addi- Mata ⌬avt3ϻHIS3 ⌬avt4ϻURA3 leu2-3,112 ura3-52 trp1-289 his3-⌬200 tion to this apparent conservation in topological features, a sequence alignment (see Fig. 1C) reveals a number of amino acid positions (usually within transmembrane domains) where invariant or highly conserved amino acid residues are utilized. Sizes of the predicted yeast proteins range from 448 to 713 amino acids, with the greatest disparity due to extended and divergent hydrophilic N-terminal sequences in the case of AVT1, AVT3, and AVT4. The yeast AVT proteins can be further subdivided into four main branches defined by AVT1, AVT2, AVT3/4, and AVT5/6/7 (Fig. 1A). As shown, some of these yeast proteins are more closely related to known transporters from distant species such as worms and mammals than they are to other yeast branch members. Interestingly, AVT3 and AVT4 appear to be more highly evolutionarily conserved because these proteins show significant homology (greater than 25% identity) to a number of predicted proteins from C. elegans and Drosophila melanogaster, two organisms whose entire genome sequences have now been determined. The relationship of these predicted proteins to AVT3 and AVT4 as compared with UNC-47 is shown in Fig. 1B, whereas their amino acid sequences have been aligned in Fig. 1D.
The AVT Genes Are Not Involved in GABA Metabolism or the Vacuolar Accumulation of Basic Amino Acids-The identification of several genes closely related to the GABA vesicular transporters was interesting, given that many proteins in S. cerevisiae, including UGA4, a GABA-specific permease (22), are involved in the use of GABA as a nitrogen source. To investigate the potential role of the AVT genes in GABA flux and to establish a genetic system to study AVT protein function, we decided to test strains carrying deleted AVT genes for GABAspecific growth defects. Preliminary knockout experiments in diploid strains followed by sporulation and tetrad analysis indicated that the AVT genes were not essential for growth. This allowed the creation, by direct transformation (see "Experimental Procedures" for details), of a variety of isogenic haploid strains carrying both the single and multiple gene deletions shown in Table I. We tested these strains for their ability to grow on 1 mM GABA as the sole nitrogen source and found no discernible difference from the wild type. Furthermore, we disrupted the UGA1 gene (23), which codes for GABA transaminase that converts GABA to glutamate, in each of the single AVT deletion strains and observed no differential growth on YPD medium containing high levels (50 mM) of GABA compared with the ⌬uga1 strain alone. Finally, we have observed that GABA is not actively taken up by purified yeast vacuoles, even at high (500 M) concentrations ( Fig. 2A, see below).
Seven amino acid uptake systems have been described in yeast vacuoles (16 -18), including three that are specific for basic amino acids, the most abundant amino acids found in this organelle (24,25). To determine whether any of the AVT proteins were involved in these processes, we tested vacuoles purified from the created ⌬avt strains for the uptake of radiolabeled amino acids. Shown in Fig. 2A is the ATP-dependent uptake of lysine (10-fold) and arginine (5-fold) in 4 min by vacuoles purified from wild type yeast. Both activities are extremely sensitive to nigericin, a K ϩ /H ϩ antiporter that disrupts the pH gradient without affecting the electrical gradient. However, robust lysine, arginine, and histidine transport were also observed in vacuoles obtained from any of the AVT gene-deleted strains tested. The results obtained for lysine and arginine with the ⌬avt1, ⌬avt2, ⌬avt3, ⌬avt4 quadruple mutant and the ⌬avt5, ⌬avt6, ⌬avt7 triple mutant are shown in Fig. 2B. However, the finding that lysine influx is unaffected by AVT protein function allowed us to accurately quantify the activity of purified vacuolar fractions, greatly facilitating the comparative analysis described below.
AVT1, AVT3, and AVT4 Are Involved in the Bidirectional Movement of Large Neutral Amino Acids across the Vacuolar Membrane-Distinct transport systems have also been described for tyrosine, glutamine/asparagine, isoleucine/leucine, and phenylalanine/tryptophan (17). Like lysine, tyrosine was also concentrated ϳ10-fold in the presence of ATP (Fig. 3B).
This uptake was decreased dramatically in the presence of nigericin and was not further reduced by the addition of valinomycin, a K ϩ ionophore that would dissipate any remaining electrical gradients (data not shown). Our first indication that some of the AVT proteins might be involved in vacuolar amino acid transport came from the observation that tyrosine uptake was completely eliminated in the ⌬avt1 mutant (Fig. 3B), whereas high lysine activity was retained (Fig. 3A). Although tyrosine uptake was shown to be blocked in these vacuoles in vitro, high levels of tyrosine (0.5 and 5 mM) were not toxic to the ⌬avt1 strain (data not shown). Surprisingly, AVT1 was also required for the active uptake of both glutamine and isoleucine, which were previously thought to be transported by separate carriers. The accumulation of these amino acids was reduced ϳ5-fold (Fig. 4A) and 8-fold (Fig. 4B), respectively, in vacuoles purified from ⌬avt1 compared with the wild type.
Further testing of vacuoles purified from the other AVT deletion strains resulted in the finding that large neutral amino acids could accumulate more rapidly and to higher levels when compared with wild type. As shown in Fig. 4, tyrosine, glutamine, and isoleucine uptakes were all increased in the ⌬avt3, ⌬avt4 double mutant. The effect was more dramatic for tyrosine and isoleucine (greater than 2-fold) but was also significant for glutamine. One interpretation of these observations was that there exists an activity that exports these amino acids to the outside. Thus, AVT3 and AVT4, which share ϳ40% sequence identity, appear to act synergistically in this process because single mutants had only partial phenotypes (Fig. 4). We note that for each amino acid, the removal of AVT4 appeared to have a more pronounced effect, and, in fact, the uptake of isoleucine was unaltered in vacuoles purified from ⌬avt3. We conclude from these observations that AVT4 may be a more efficient transporter than AVT3 under the given in vitro assay conditions. One possibility is that AVT4 possesses a higher affinity for the various substrates. As expected, the accumulation of large neutral amino acids in the ⌬avt3, ⌬avt4 double mutant was completely eliminated with the further removal of AVT1 (⌬avt1, ⌬avt3, ⌬avt4 triple mutant; data not shown).
Sato et al. (17) observed that unlabeled tryptophan or phenylalanine did not compete with tyrosine or leucine for vacuolar uptake and therefore must be actively transported by a protein or proteins other than AVT1. In our hands, uptake assays with [ 3 H]tryptophan appeared to result in some degree of vacuolar accumulation, but over very high background levels due to nonspecific filter binding (data not shown). This has made it difficult to adequately interpret the data and to assess relatively small strain-specific differences. However, we have routinely observed a lack of ATP-dependent tryptophan uptake in vacuoles isolated from ⌬avt1 but cannot yet offer an explanation for this apparent discrepancy with the previously published competition studies.
Previous reports (16,17) have also shown that many amino acids, including methionine, serine, glycine, cysteine, proline, valine, threonine, and alanine, are not actively taken up by vacuoles in vitro. With the discovery that the inactivation of the amino acid effluxers AVT3 and AVT4 resulted in higher accumulation of some neutral amino acids, we retested glycine and proline in ⌬avt3, ⌬avt4 vacuoles and observed no ATP-dependent uptake (data not shown).
AVT6 Is Required for the Efflux of Glutamate and Aspartate-Vacuoles purified from wild type cells accumulated less glutamate and aspartate in the presence of ATP (Fig. 5, A and  B). As could be demonstrated for amino acid uptake, this apparent ATP-dependent efflux of acidic amino acids was sensitive to nigericin. To distinguish between efflux, the actual displacement of substrate from the lumen of the vacuole to the outside, and an alternative phenomenon such as decreased inward diffusion when a proton electrochemical gradient was imposed on the system, we carried out the assays shown in Fig.  6. Vacuoles were preloaded with [ 3 H]glutamate in the absence of ATP and then monitored for amino acid content after the addition of ATP or ATP plus nigericin. Consistent with an active efflux mechanism, glutamate levels decreased steadily in the presence of ATP alone. Again, efflux was not observed in the presence of nigericin (Fig. 6A) nor from vacuoles prepared FIG. 6. AVT6, but not AVT5 or AVT7, is required for glutamate and aspartate transport out of vacuoles. Efflux assays using vacuoles purified from either (A) wild type or (B) ⌬avt6 strains. Vacuoles were preloaded with [ 3 H]glutamate for 5 min at 30°C. Samples (0 min) were removed immediately before or 1 min and 4 min after the addition of ATP or ATP plus nigericin. Accumulation in 4 min of (C) glutamate and (D) aspartate in the absence or presence of ATP or ATP plus nigericin in vacuoles isolated from the indicated mutant strains is shown. from the ⌬avt6 mutant (Fig. 6B, see below).
We further tested various AVT deletion strains for ATP-dependent efflux of either glutamate (Fig. 6C) or aspartate (Fig.  6C) and observed that both activities were lost only in vacuoles purified from ⌬avt6. Even though AVT5 and AVT7 are 50% and 35% identical, respectively, to AVT6, wild type activity was retained in the ⌬avt5 and ⌬avt7 mutant strains. In addition, glutamate and aspartate levels were not further altered in a ⌬avt5, ⌬avt6 double mutant (Fig. 6) or in a ⌬avt5, ⌬avt6, ⌬avt7 triple mutant (data not shown). The potential inability of the ⌬avt5, ⌬avt6 mutant strain to pump acidic amino acids from vacuoles in vivo did not alter its growth (data not shown) on medium containing high glutamate (0.5 M, pH 7.0).
Localization of AVT2 and AVT7 by Indirect Immunofluorescence-We have been unable to assign any vacuolar amino acid transport role to proteins AVT2, AVT5, or AVT7. To determine (a) if these proteins are associated with vacuoles and (b) whether transporters with a definite vacuolar function are localized exclusively to this organelle, we have carried out preliminary indirect immunofluorescence using HA-tagged molecules (data not shown). To summarize briefly, AVT1HA, AVT4HA, and AVT6HA (all proteins with vacuolar function) display a staining pattern that is indistinguishable from that of VMA2, the 60-kDa B subunit of the vacuolar ATPase. A similar pattern was obtained with AVT7HA, but in addition, strong immunoreactivity was observed at the plasma membrane in a large fraction of cells. On the other hand, AVT2HA expression was distinct and resulted in punctate staining reminiscent of proteins that reside in the endoplasmic reticulum. Thus, AVT2 may function in a different region of the cell, although we cannot rule out the possibility that the observed localization pattern may have been due to the retention of misfolded derivatized molecules.

DISCUSSION
In this study, we have described a variety of vacuolar amino acid transport activities that can be attributed to members of the AVT protein family of S. cerevisiae (see Table II for a summary). These observations are consistent with the known homology of these proteins to other amino acid transporters found in organisms ranging from plants to mammals. One protein, AVT1, provides the sole molecular basis for what was described as three separate vacuolar transport systems almost 20 years ago and is responsible for the uptake of tyrosine, glutamine, asparagine, isoleucine, and leucine. More importantly, the ability to easily disrupt genes in yeast has made it possible to identify, for the first time, proteins that transport amino acids out of an acidic compartment, a process that seems essential to remove protein degradation products from organelles such as vacuoles and lysosomes. Two closely related proteins, AVT3 and AVT4, display the same specificity as AVT1 but are synergistically involved in the movement of amino acids from the vacuole to the cytosol. The fourth protein, AVT6, is responsible for the efflux of aspartate and glutamate from vacuoles and requires a proton electrochemical gradient. The action of AVT6 in vivo may explain the observation that although glutamate is the most abundant amino acid found in yeast, only a small percentage of glutamate is detected in vacuolar pools (24,25).
Other than the use of amino acids as substrates, it is difficult to categorize the UNC-47 family of transporters in terms of energetics, cellular localization, or specificity. Like the GABAspecific vesicular transporters, AVT1 is associated with an acidic cellular compartment whose outwardly directed pH gradient is used to drive amino acid uptake. This type of amino acid/H ϩ antiport mechanism is also thought to exist for the vertebrate system N transporter SN1/NAT1, a glutamine/histidine transporter (actually more closely related to AVT2; see Fig. 1A) that resides in plasma membranes and requires, in addition, Na ϩ cotransport (6,8). Amino acid transport by the related system A is sensitive to pH, but it is not known whether proton transport is required for function (9,11).
In the case of AVT6 function, transport of acidic amino acids out of the vacuole likely involves the cotransport of protons. This mechanism is also used by plasma membrane transporters from both the related family of plant amino acid/auxin permeases (13) and the large unrelated major facilitator superfamily of amino acid permeases from bacteria and yeast (26). Interestingly, one of the latter transporters from S. cerevisiae, DIP5 (27), has the same specificity as AVT6 because it mediates the high-affinity transport of glutamate and aspartate into the cell. Both transporters contain 10 -11 transmembrane domains and possess the same specificity and general mode of action. Despite this, they lack any sequence similarity, perhaps indicative of the fact that they occupy unique cellular environments. On the other hand, AVT5 and AVT7 are closely related to AVT6 (50% and 35% identity, respectively) and, in the case of AVT7 at least, appear to be localized to vacuoles based on immunofluorescence studies (see "Results"). Nonetheless, we have been unable to detect any acidic amino acid transport functions attributable to these proteins. For AVT5 and AVT7, as well as AVT2, it is possible that these proteins transport other organic compounds or that these proteins require additional cofactors that were not included in the standard conditions used throughout this study.
Both AVT3 and AVT4 are involved in amino acid efflux, with ligand specificities that appear identical to AVT1. Thus, vacuoles deficient for AVT3 and AVT4 accumulate to elevated levels those amino acids that are actively taken up by AVT1. These include aromatic neutral compounds such as tyrosine and other neutral amino acids with bulky side chains such as isoleucine and glutamine. It is likely that tryptophan, another neutral aromatic amino acid, is also imported by AVT1 (see "Results") and may therefore also be a substrate for AVT3 or AVT4. The carrier-mediated transport of a similar set of amino acids, referred to as system h, has been described in lysosomes isolated from rat FRTL-5 thyroid cells (28) as well as murine  (17) have shown that asparagine and glutamine and leucine and isoleucine utilize the same vacuolar uptake system. melanosomes (29), two intracellular organelles that are similar to the fungal vacuole. Because these findings are based on countertransport assays after the preloading of methyl esterderivatized amino acids or the uptake of amino acids at relatively high concentrations, it is still not clear in which direction these carriers normally function in vivo. In either case, it would appear that transport by system h does not require a proton electrochemical gradient. Studies in human fibroblast lysosomes (30) have further identified a system d that transports glutamate and aspartate and therefore may represent an activity related to AVT6 function.
In terms of their mechanism of action, we have been unable to ascertain whether amino acid transport by AVT3 and/or AVT4 requires a proton motive force or any other type of electrochemical gradient. To address this problem, we have carried out experiments using ⌬avt1 vacuoles that do not actively take up glutamine or isoleucine. Although it could be demonstrated that these amino acids are exported from preloaded vacuoles in an ATP-dependent manner, this process, in contrast to AVT6 efflux, was resistant to both nigericin and valinomycin. More importantly, a similar activity was also detected in vacuoles that were lacking AVT3 and AVT4 and therefore cannot be attributed to these proteins. The inability, thus far, to detect specific efflux under these conditions may mean that AVT3 and AVT4 are active only at the higher internal amino acid concentrations that exist when AVT1 is active. Alternatively, we cannot entirely rule out the possibility that AVT3 and AVT4 are negative modulators of AVT1 function.
For S. cerevisiae growing under normal laboratory conditions, basic amino acids become highly concentrated in the vacuole. This is due to three transport systems that have been described in purified vacuoles: (a) an arginine/histidine exchanger, (b) an arginine/lysine-specific transporter, and (c) a histidine-specific transporter (17,18). Arginine is a good source of nitrogen and, unlike lysine and histidine, can be degraded in the cytoplasm. During periods of starvation, vacuolar arginine stores are mobilized (24), a process that may be regulated. Surprisingly, none of the AVT proteins are involved in the vacuolar transport of these amino acids. We have demonstrated, however, that the steady-state levels reached by tyrosine, glutamine, and isoleucine in vitro are determined by the opposing actions of AVT1 and AVT3/AVT4. This type of system offers many potential targets for regulation in vivo and may provide a means to precisely control amino acid levels or to change the overall direction of amino acid flux. Interestingly, Harper et al. (31) have found that exposure of FRTL-5 cells to thyroid-stimulating hormone has a dramatic effect on lysosomal tyrosine countertransport, a process that may be mediated by a cAMP signal. Also, because a protein (AVT6) has now been identified that pumps glutamate out of acidic organelles, it is tempting to speculate that such an activity could modulate neurotransmitter loading of synaptic vesicles in glutamatergic neurons.
In summary, the conserved function of the UNC-47/VGAT/ VIAAT family in amino acid transport across biological membranes, generally in response to an imposed pH gradient, is found in diverse applications ranging from nutrient uptake by plant tissues to neurotransmitter packaging and glutamine cycling in the vertebrate brain. As demonstrated by the yeast members of this family, this function now extends to the bidirectional transport of amino acids in vacuoles. The existence of close yeast AVT homologues in worms and flies, together with the identification of vacuolar transport systems that resemble those found in mammalian lysosomes, strongly suggests that some of these AVT functions have been conserved.