AGP2 Encodes the Major Permease for High Affinity Polyamine Import in Saccharomyces cerevisiae*[boxs]

Polyamines play essential functions in many aspects of cell biology. Plasma membrane transport systems for the specific uptake of polyamines exist in most eukaryotic cells but have been very recently identified at the molecular level only in the parasite Leishmania. We now report that the high affinity polyamine permease in Saccharomyces cerevisiae is identical to Agp2p, a member of the yeast amino acid transporter family that was previously identified as a carnitine transporter. Deletion of AGP2 dramatically reduces the initial velocity of spermidine and putrescine uptake and confers strong resistance to the toxicity of exogenous polyamines, and transformation with an AGP2 expression vector restored polyamine transport in agp2Δ mutants. Yeast mutants deficient in polyamine biosynthesis required >10-fold higher concentrations of exogenous putrescine to restore cell proliferation upon deletion of the AGP2 gene. Disruption of END3, a gene required for an early step of endocytosis, increased the abundance of Agp2p, an effect that was paralleled by a marked up-regulation of spermidine transport velocity. Thus, AGP2 encodes the first eukaryotic permease that preferentially uses spermidine over putrescine as a high affinity substrate and plays a central role in the uptake of polyamines in yeast.

nized as a major contributing factor to polyamine homeostasis in vivo and is receiving considerable attention, especially from the perspective of cancer therapy based on anti-polyamine strategies (3).
Plasma membrane carriers responsible for the influx or excretion of polyamines have been described in considerable detail in bacteria (4). On the other hand, much less is known about the molecular identity of polyamine carrier proteins in eukaryotic cells. During the preparation of this manuscript, a putrescine-preferential plasma membrane permease catalyzing the influx of putrescine and spermidine has been identified in the parasite Leishmania major (6). In Saccharomyces cerevisiae, four H ϩ gradient-dependent Tpo permeases (Tpo1p, Tpo2p, Tpo3p, and Tpo4p) belonging to the H ϩ /drug antiporter subfamily with 12 transmembrane spans (7) catalyze the efflux of polyamines as well as a large variety of organic substrates (e.g. quinidine, cycloheximide, mycophenolic acid, etc.) at the plasma membrane level (8 -10). A fifth Tpo carrier, Tpo5, has recently been described that belongs to the group of L-type amino acid transporters within the APC 1 family (11,12) and is thought to function as a H ϩ -coupled symporter in the efflux of polyamines into Golgi-derived vesicles (13). The same group has also reported that the Gap1p permease accepts putrescine and, to a lesser degree, spermidine, as substrates in S. cerevisiae (14). Gap1p has thus far been characterized as an amino acid permease belonging to the YAT group within the APC family (11,12) that recognizes a wide array of amino acid substrates (15,16). However, deletion of the GAP1 gene elicits only modest reductions of putrescine and spermidine uptake, raising questions about the identity of the permease responsible for the major component of polyamine influx in yeast.
In yeast, spermidine and spermine are thought to be imported across the plasma membrane via both high and low affinity transport systems (17), whereas putrescine transport occurs only via an apparently nonsaturable, low affinity pathway (17,18). An important determinant of polyamine transport is the Ser/Thr protein kinase Ptk2p, which pleiotropically acts as an activator of plasma membrane permeability to various cations, including polyamines (17,19). Via the glucose-dependent activation of the Pma1p H ϩ -ATPase, Ptk2p activity increases the proton motive force that energizes the transport of a variety of cationic compounds (20). Sky1p, a SR protein kinase, is another important factor that probably regulates the rate of polyamine transport via modulation of the Trk1p-Trk2p K ϩ carrier system, which tends to dissipate the proton gradient established by Pma1p pumping (21).
Using genome-wide screening techniques, we have recently identified Agp2p, another member of the YAT family, as a major component of the membrane transport system responsible for the uptake of the A 5 isoform of the anticancer drug bleomycin (22), whose structure includes a spermidine substituent at the terminal amine position (23). Interestingly, among the four other genes that the same screen identified as conferring sensitivity to bleomycin A 5 were PTK2 and SKY1 (22), which are major regulators of polyamine transport, as mentioned above. Although AGP2 was initially characterized as a gene encoding a carnitine transporter (24), we now present evidence that Agp2p is largely responsible for high affinity spermidine transport, contributes to a substantial fraction of total putrescine uptake, and is the major polyamine permease in S. cerevisiae.

EXPERIMENTAL PROCEDURES
Strains, Media, Transformation, and Reagents-The S. cerevisiae strains used in the present work are listed in Table I. Yeast cells were grown at 30°C in either YPD or minimal synthetic medium (Difco) (25,26). Yeast cells were transformed by the lithium acetate method (27). Fluorescein isothiocyanate (FITC)-labeled spermine was synthesized as described (23). [  Survival Assays and Growth Tests-Survival assays and standard spot tests were performed as previously described (28). Polyamine-free H medium (2% (w/v) glucose, 0.17% (w/v) amino acid-free yeast nitrogen base, and 0.5% ((NH 4 ) 2 SO 4 ) was prepared as described (29). All media and water were filter-sterilized and dispensed in 50-or 15-ml plastic sterile polypropylene tubes. H medium was supplemented with the required amino acid (L-histidine, L-leucine, and L-methionine, each at 20 g/ml) or uracil (1 g/ml) according to strain auxotrophy. Cells were grown overnight on a solid minimal medium plate. The next day, cells were scraped from the plate, resuspended in 1 ml of filter-sterilized water in a 1.5-ml sterile Eppendorf tube, and conditioned for 40 min. A 600 was adjusted to 0.04 in H medium. Two-ml aliquots of cell suspension were then distributed to 15-ml sterile plastic Falcon tubes containing the indicated concentration of putrescine. The loosely capped tubes were shaken for 21 h at 30°C, and the A 600 was then determined.
Gene Disruption-Deletion of the SPE1 gene in the indicated strain was carried out by one-step gene targeting using universal upstream and downstream primers. These primers contain 45 nucleotides from the targeted gene sequence and 20 nucleotides matching the universal regions present in a set of selective marker genes including HIS3, URA3, TRP1, and LEU2 (30). Deletion of the AGP2 gene in the indicated strain was made using endogene C-tap tagging of the first Nterminal 400 base pairs of the gene, thus disrupting the function of the gene (31). Deletions were verified using PCR. The truncated allele of AGP2 expressed an inactive fragment of the protein and confirmed by Western blot using anti-protein A antibody (1:3000; Sigma) (see below).
Spermidine, Putrescine, and Leucine Uptake Analysis-Prior to the polyamine uptake assay, cells were grown to the midlogarithmic phase, washed three times in uptake buffer A (50 mM sodium citrate, pH 5.5, 2% D-glucose), and resuspended in 100 l of the same buffer at 2 ϫ 10 7 cells/ml. The uptake assay was initiated by the addition of [ 14 C]spermidine, [ 3 H]putrescine, or [ 3 H]leucine at the indicated concentration followed by incubation in a water bath at 30°C with shaking. The reaction was stopped at predetermined intervals by adding 1 ml of ice-cold uptake buffer. Cells were washed three times with uptake buffer and resuspended in 100 l of this buffer. Five ml of scintillation mixture (Amersham Biosciences) were added to each sample, and the retained radioactivity was determined by liquid scintillation spectrometry.
Fluorescence Microscopy-Cells were grown to a density of 2 ϫ 10 7 cells/ml, washed twice in water, and resuspended in uptake buffer A. Briefly, a 100-l aliquot of the cell suspension was incubated with either F-spermine or FITC (each at 1 g/ml) for 1 h at 30°C in the dark with mild shaking. Cells were washed three times with 1 ml of phosphatebuffered saline and processed for epifluorescence microscopy (23).
Immunoblot Analysis-Immunoblot analysis was performed as previously described (32). Membranes were blocked for 2 h in buffer B (100 mM Tris, pH 7.5, 150 mM NaCl, 10 mM EDTA), containing 5% (w/v) milk powder and 0.1% (v/v) Tween 20, washed twice (5 min each) in buffer B, and incubated with either anti-GFP monoclonal antibody (BD Biosciences Clontech, Mississauga, Ontario) at a dilution of 1:5000 or anti-protein A antibody at a dilution of 1:3000 in buffer B, and the secondary antibodies were anti-mouse and anti-rabbit coupled to horseradish peroxidase (Amersham Biosciences), respectively. Protein bands were revealed by using PerkinElmer Chemiluminescence Reagent Plus followed by exposure to Kodak double emulsion film.
Northern Blot-Total RNA was prepared by the rapid method (33). Typically, 15 g of RNA/lane was analyzed by electrophoresis in an agarose/formaldehyde gel, transferred to a nylon Nϩ membrane (Amersham Biosciences) and incubated 1 h at 42°C in 9 ml of prehybridization buffer UltraHyb (Ambion). Membrane was probed overnight at 42°C with 32 P-labeled AGP2 DNA (1 ϫ 10 6 cpm) derived from the open reading frame of AGP2 using PCR and a Ready-to-Go DNA labeling Kit (Amersham Biosciences). The membrane was washed two times in 1ϫ SSC, 0.1% SDS for 5 min at 42°C and subjected to another two washes in 0.2ϫ SSC, 0.1% SDS at 42°C for 5 min, following autoradiography.
Topological and Phylogenetic Analysis-Membrane protein topology was analyzed using the HMMTOP algorithm (34). Multiple sequence alignments were performed using the ClustalW algorithm (35) 36), we assessed whether the other three mutants in the bleomycin-resistant panel exhibited the same resistance. Exponentially growing cells were spotted onto solid YPD plates containing the indicated concentrations of spermine, spermidine, and putrescine ( Fig. 1). Among these mutants, agp2⌬ and fes1⌬ were as resistant to polyamine toxicity as either ptk2⌬ or sky1⌬. In contrast, brp1⌬ exhibited an intermediate level of resistance to spermine and spermidine (Fig. 1). These mutants were also resistant to the polyamine analog paraquat (Fig. 1). As expected from our previous observation that END3 deletion up-regulates polyamine transport activity (37), end3⌬ mutants exhibited a marked increase to the toxic effects of high concentrations of putrescine, spermidine, and spermine, confirming that the assay accurately reflected changes in polyamine transport activity. These data suggest that, like Ptk2p and Sky1p, the L-carnitine transporter Agp2p, as well as Brp1p and Fes1p, may also play a role in polyamine transport. agp2⌬, ptk2⌬, sky1⌬, brp1⌬, and fes1⌬ Single Mutants Are Defective in the Vacuolar Accumulation of F-SPM-Using a fluorescently (FITC)-labeled form of spermine (F-SPM), we previously demonstrated that it accumulates in the vacuoles of the parental strains, but not in the ptk2⌬ mutant, consistent with a reduction in the rate of polyamine uptake and accumulation in the latter (22). We therefore examined whether the other mutants (agp2⌬, sky1⌬, brp1⌬, and fes1⌬) were also defective in the vacuolar accumulation of F-SPM. In the parent, F-SPM accumulated into the vacuole, which was visualized with the FM® 4-64 dye that stains vesicles of the endocytic pathway as well as the vacuolar membrane ( Fig. 2) (38,39). In contrast, the bleomycin-resistant mutants showed either a modest (e.g. sky1⌬) or no significant (e.g. agp2⌬) accumulation of F-SPM in the vacuoles ( Fig. 2 and data not shown). These data and the fact that ptk2⌬ and sky1⌬ mutants are defective in polyamine transport (17,19,36) suggested that the inability of the agp2⌬ mutant, in particular, to efficiently accumulate F-SPM in the vacuoles might result from a defect at the level of transport. Moreover, the molecular defect in the agp2⌬ mutant is likely to affect polyamine transport at the plasma membrane rather than vacuolar level, since deficient vacuolar accumulation would have translated into increased sensitivity to polyamines rather than the marked resistance observed here. This is supported by the fact that Agp2p has been previously shown to be a plasma membrane transporter of L-carnitine (24). On the other hand, neither Brp1 (i.e. YGL007W, an open reading frame of unknown function) or Fes1p, a protein that binds to a complex involved in the regulation of protein translation, has any documented role in transport function (17,22,24,36). Thus, the most likely candidate gene having a role in polyamine transport among the mutants screened was clearly AGP2.
Agp2p Is a Major Effector of High Affinity Polyamine Transport-To address the hypothesis that Agp2p is a polyamine transporter, we monitored the uptake of 14 C-labeled spermidine in the agp2⌬ mutant and its parental strain as described (17). As shown in Fig. 3A, spermidine accumulation after the first 5 min was at least 18-fold lower in the agp2⌬ mutant, and net accumulation of the polyamine stopped after 10 min. Kinetic analysis showed that spermidine uptake was saturable up to about 100 M (V max ϭ 172 Ϯ 10 pmol/10 7 /min) with an apparent K m of 15 Ϯ 3 M (Fig. 3B), which is consistent with the high affinity component of spermidine transport previously reported in yeast (17,36,40). Spermidine uptake in the agp2⌬

FIG. 2. Decreased vacuolar internalization of F-SPM in agp2⌬
and sky1⌬ mutants. Cells were incubated with F-SPM (1 g/ml for 1 h) and then photographed at a magnification of ϫ100 with a Leica epifluorescence microscope equipped with a digital camera (Retiga GX 32-002TB-303). FITC accumulation (top panel, right) was used as a negative control. The FM® 4-64 dye was used as a positive control to localize the vacuole. The data are representative of three independent experiments. mutant was nonsaturable, and its linear relationship with substrate concentration was consistent with the low affinity component of spermidine uptake detected in S. cerevisiae (17). The loss of saturable spermidine transport in the agp2⌬ mutant strongly suggests that Agp2p acts as a high affinity polyamine permease of major importance in yeast. Furthermore, putrescine uptake was also reduced by disruption of AGP2 expression (Fig. 3C). Although putrescine transport is a relatively low affinity process as compared with spermidine uptake in yeast, it is coordinately regulated by Ptk2p (17) and Sky1p (36). The latter data thus indicate that Agp2p might also function as a putrescine permease in yeast. In control experiments, the uptake of [ 3 H]leucine was not altered in the agp2⌬ mutant, as compared with the parent (Fig. 3D). Likewise, neither serine or lysine could compete with [ 14 C]spermidine uptake (data not shown). These results are in agreement with previous reports that Agp2p has a negligible contribution to the transport of the common amino acids under nitrogen-replete conditions (24,41).
Since Agp2p was first reported as a high affinity L-carnitine transporter (K m ϳ5 M) (24), we examined the ability of this quaternary amino acid to compete for [ 14 C]spermidine uptake. Quite unexpectedly, L-carnitine (0.01-10 mM) did not interfere with the uptake of spermidine, even after allowing a 5-min period of preincubation before the addition of the polyamine (data not shown; cf."Discussion").
Decreased AGP2 Expression by Osmotic Stress Coincides with a Reduction in Spermidine Uptake-It has recently been shown that osmotic stress imposed by high concentrations of NaCl, KCl, or sorbitol decreases AGP2 expression (42). Indeed, we confirmed by Northern blot analysis that parent cells exposed for 1 h to 1 M NaCl, KCl, or sorbitol showed a Ն3-6-fold reduction in AGP2 mRNA levels, whereas actin (ACT1) expression remained unaffected (Fig. 4A). We therefore examined the effect of these hyperosmotic stress conditions on [ 14 C]spermidine uptake. Cells treated for 1 h with either NaCl, KCl, or sorbitol (each at 1 M) showed a marked reduction of [ 14 C]spermidine uptake (Fig. 4B). Among the agents used, sorbitol at 1 M was the most effective in reducing [ 14 C]spermidine uptake (an approximately 6-fold decrease after 30 min), in agreement with the Northern blot data. These data clearly support the notion that spermidine transport is dependent on AGP2 expression.
Agp2 Overexpression Enhances Spermidine Transport and Toxicity-We next assessed whether enhanced expression of AGP2 would increase spermidine transport. A multicopy plasmid (pAGP2-GFP) was thus engineered to overproduce Agp2p as a GFP fusion protein (Agp2-GFP) under the control of its own promoter. We have demonstrated that the resulting Agp2-GFP fusion protein is physiologically active in bleomycin uptake (22). Western blot analysis revealed that a total cell extract prepared from either the parent or agp2⌬ mutant strain carrying the plasmid pAGP2-GFP expressed the expected 93-kDa fusion protein (Fig. 5A). The overproduction of Agp2-GFP stimulated [ 14 C]spermidine uptake into the parent by about 40% after 30 min (Fig. 5B). Likewise, Agp2-GFP overexpression in the agp2⌬ mutant stimulated [ 14 C]spermidine uptake to the level observed in the parental strain. Thus, Agp2 overexpression results in a marked enhancement of the velocity of spermidine influx, a finding consistent with its identity as a polyamine permease.
We next assessed whether an enhanced level of AGP2 expression increases polyamine cytotoxicity as a result of an increased rate of polyamine accumulation. Spot test analysis revealed that Agp2-GFP overexpression significantly increased the sensitivity of cell proliferation to exogenous polyamines After total RNA was extracted from the samples, the transcript levels were determined by Northern analysis. Actin mRNA was measured for normalization of RNA levels. B, suppression of [ 14 C]spermidine uptake by different osmotic stress agents. BY4741 cells growing exponentially in the YPD medium were treated for 1 h with 1 M NaCl, KCl, or sorbitol, and [ 14 C]spermidine uptake was determined as described under "Experimental Procedures." (Fig. 5C). In control experiments, the overproduction of Agp2-GFP did not enhance the sensitivity of the parent cells to a variety of genotoxic and cytotoxic agents, including the DNAdamaging agent 4-nitroquinoline-1-oxide and calcofluor white, an inhibitor of cell wall synthesis (data not shown). These data further support the hypothesis that AGP2 encodes a polyamine permease and that its expression level correlates with cellular sensitivity to exogenous polyamines.
Growth of spe1⌬ Mutants Depends on the Transport of Exogenous Polyamines Mediated by Agp2p-In yeast, the endogenous synthesis of spermine from ornithine and S-adenosylmethionine requires four biochemical steps (43). Interruption of any of these steps makes cell growth dependent on an exogenous source of polyamines (44). To investigate whether Agp2p is absolutely required for the transport of polyamines, by permitting cell growth in the absence of endogenous polyamine synthesis, we blocked the first step of the polyamine biosynthetic pathway by deleting the SPE1 gene. SPE1 encodes ornithine decarboxylase, which converts ornithine to putrescine (44). Uptake experiments revealed that SPE1 deletion did not interfere with [ 14 C]spermidine transport when compared with parental cells (Fig. 6A). Likewise, deletion of the SPE1 gene in the agp2⌬ mutant resulted in the spe1⌬agp2⌬ double mutant showing the same low, basal level of [ 14 C]spermidine uptake as the single agp2⌬ mutant (Fig. 6A). Therefore, cells devoid of endogenous polyamine synthesis do not significantly derepress its transport pathway, although this would be expected, since polyamines accumulated from the rich YPD medium probably exert feedback regulation on the expression of the polyamine transport system (37).
To directly test whether the growth of spe1⌬ mutant is dependent upon transport of exogenous polyamines via Agp2p, the parent and isogenic mutant strains were inoculated at a low cell density (A 600 ϳ0.07) in polyamine-free H medium (cf. "Experimental Procedures") and supplemented with increasing concentrations of putrescine. Both the parent and the agp2⌬ mutant grew in H medium without putrescine and reached a similar maximal density after 21 h (A 600 ϳ0.7) (Fig. 6B). In the absence of putrescine, the spe1⌬ single mutant and the spe1⌬agp2⌬ double mutant A 600 of 0.07 under these conditions, suggesting that these two mutants require exogenous polyamines for exponential growth (Fig. 6B). Indeed, supplementation with putrescine restored growth in the spe1⌬ mutant in a concentration-dependent manner, with an EC 50 ϳ9 M. In contrast, the spe1⌬agp2⌬ double mutant reached half-maximal growth only in the presence of a ϳ12-fold higher concentration (ϳ120 M) of putrescine. These data clearly indicate that Agp2p plays a crucial role in the influx of exogenous polyamines under conditions that prevent their biosynthesis.

Enhancement of Spermidine Uptake in Endocytosis-deficient Mutants Correlates with an Increased Abundance of Agp2p-
We have previously shown that defects in an early step of the endocytic pathway enhance the maximal velocity of spermidine uptake, which we postulated to be the result of a decreased turnover of the plasma membrane polyamine transporter (37). We therefore examined whether basal Agp2p abundance is increased in end3⌬ mutants, which lack a cytoskeletal protein that plays a key role in the early steps of internalization of plasma membrane proteins, and are thus defective in both fluid phase and clathrin-dependent endocytosis (45,46). To carry out this experiment, we used the parent and end3⌬ mutant strains expressing Agp2p that was endogenously tagged on the C terminus with the c-TAP tag. Western blot analysis of total cell extracts derived from either the parent or the end3⌬ mutant showed that the level of the Agp2-cTAP fusion protein (expected molecular mass ϳ88-kDa) (Fig.   7A, upper panel) was increased about 2-fold by END3 disruption (Fig. 7A, upper panel, lane 9 versus lane 6), a value that could be underestimated due to protein degradation as judged by the appearance of a faster migrating polypeptide of ϳ30 kDa (Fig. 7A, lane 9). The higher level of Agp2-cTAP in the end3⌬ mutant was not due to enhanced gene expression, since the AGP2 mRNA level was not altered in the mutant when compared with the parent (Fig. 7B, lane 4 versus lane 1). Thus, the previously reported increased rate of spermidine transport in end3⌬ mutants could be explained, at least in part, by an enhanced abundance of Agp2p. To confirm this, we monitored the uptake of [ 14 C]spermidine in the parent, in the end3⌬ and agp2⌬ single mutants, and in the end3⌬agp2⌬ double mutant (Fig. 7C). As shown previously, the end3⌬ mutant exhibited a sharp, ϳ2.5-fold increase in the level of spermidine uptake, relative to parental cells (Fig. 7C). It is noteworthy that AGP2 deletion in the end3⌬ mutant did not reduce the velocity of spermidine uptake to the level observed in the agp2⌬ single mutant but led to a transport rate similar to that observed in the parental strain). The latter data indicate that another A, wild-type cells and their spe1⌬, agp2⌬, and spe1⌬agp2⌬ mutant derivatives were grown to exponential phase in YPD medium and then incubated with 10 M [ 14 C]spermidine and processed as described under "Experimental Procedures." B, growth of spe1⌬ mutants depends on exogenous transport of polyamine by Agp2. Experimental conditions are described under "Experimental Procedures." Cells were grown for 21 h in a polyamine-free medium in the presence of the indicated concentrations of putrescine before the A 600 value was determined.
transporter for spermidine influx might be subject to decreased End3p-dependent turnover, in addition to Agp2p. Taken together, these results strongly suggest that an increased abundance of Agp2p molecules at the plasma membrane level are largely responsible for the up-regulation of spermidine transport noted in end3⌬ mutants, although at least another transporter that is subjected to steady-state endocytosis contributes to a lesser extent to spermidine uptake in yeast. DISCUSSION The molecular identity of the plasma membrane transporters responsible for polyamine influx in eukaryotic cells has long been elusive. We have demonstrated that AGP2 encodes the major component of the high affinity spermidine transport system in yeast, and its deletion leaves only a nonsaturable spermidine influx activity with negligible velocity at submillimolar polyamine concentrations. Agp2p is the first eukaryotic plasma membrane permease that preferentially recognizes spermidine as a substrate, in contrast with LmPot1 (6) and Gap1p (14), which are both more active toward putrescine uptake. Agp2p also contributes significantly to putrescine uptake, which is a nonsaturable activity in yeast as previously reported (17). The quantitative importance of Agp2p in controlling polyamine influx is clearly seen from the very strong resistance exhibited by agp2⌬ mutants to toxic concentrations of natural polyamines as well as paraquat. Another dramatic illustration of the importance of Agp2p in polyamine transport is that a defective AGP2 gene entails a Ͼ10-fold increase in the concentration of putrescine required to restore growth in yeast cells deficient in polyamine biosynthesis. This result must be especially appreciated from the fact that even trace amounts of polyamines (Ͻ5 nM) in growth media allow cell proliferation in yeast cells deficient in polyamine biosynthesis (29).
The identification of Agp2p as a polyamine permease has emerged from our previous studies, showing that it directs the transport of bleomycin A 5 , an isoform that contains a spermidine residue as its terminal amine substituent (22,23). However, AGP2 had initially been characterized as a high affinity plasma membrane carnitine transporter (24). The carnitine uptake system in yeast depends on induction by oleate as a carbon source and exhibits a rather narrow specificity, being restricted to the transport of L-and D-carnitine, and O-acetylcarnitine while excluding all other amino acids examined (24). There is also indirect evidence that Agp2p may function as a low affinity transporter for certain amino acids, including branched-chain amino acids, phenylalanine, and threonine (41). However, the contribution of Agp2p to the total uptake of these amino acids is extremely low under normal conditions, becoming significant only when the two major general amino acid permeases, namely Gap1p and Agp1p, are completely defective and in the presence of an inducer such as leucine (41). Since the carnitine and amino acid transport functions of Agp2p are negligible in wild-type cells grown under standard conditions, the present data demonstrate that the role of Agp2p as the main agency of polyamine influx is clearly predominant under basal conditions.
During the preparation of this manuscript, overexpression of the general amino acid permease Gap1p was shown to increase putrescine and spermidine uptake activity, whereas deletion of GAP1 brings up modest (20 -25%) decreases of putrescine and, to a lesser extent, spermidine influx rates (14). It is noteworthy that Gap1p-dependent putrescine transport was found to be a low affinity but saturable process, unlike the lack of saturability noted here for uptake of the diamine in cells expressing GAP1 at normal levels. As observed by the authors of the latter study, Gap1p represents only a minor fraction of total polyamine uptake, and due to the comparable affinity that it displays toward basic amino acids and spermidine, it is most likely inefficient at polyamine accumulation under standard growth conditions (14). As shown by the complete suppression of saturable spermidine transport in agp2⌬ mutants, Agp2p is the most likely candidate for the major permease responsible for polyamine uptake activity in wild type as well as in gap1 deletion mutants. In addition, a clear indication that Agp2p might be a more likely candidate as the major polyamine permease comes from the fact that polyamine transport is strongest in cells growing on a rich nitrogen source (e.g. NH 4 ϩ ) (i.e. under conditions where Gap1p activity is strongly repressed) (37).  lanes 1-3), wild type carrying the endogenous Agp2-cTAP (lanes 4 -6), and the end3⌬ mutant carrying the Agp2-cTAP (lanes 7-9). Cells were grown overnight in YPD and processed for Western blot analysis (upper panel)(cf. "Experimental Procedures"). The lower panel was stained with Coomassie Blue to monitor the amount of protein. B, disruption of END3 does not affect AGP2 mRNA expression. Cells growing exponentially in YPD were harvested, and total RNA was extracted and processed as described under "Experimental Procedures." C, effect of END3 disruption on the time course of [ 14 C]spermidine uptake.
Agp2p was previously shown to localize mainly at the plasma membrane by expression of a HA epitope-tagged construct of the gene, using either immunogold electron microscopy or sub-cellular fractionation (37). A minor fraction of the Agp2p-HA is also recovered in the endoplasmic reticulum and vacuole, although the significance of this subcellular localization is uncer- FIG. 8. Alignment of the primary structure of Agp2p with other prokaryotic and eukaryotic diamine or polyamine permeases. The amino acid sequence of Agp2p (residues 90 -596; gi6319608) was aligned with those of S. cerevisiae Gap1p (residues 92-595; gi6322892), L. major LmPot1 (95-639; gi8744996), and Escherichia coli PotE (residues 11-439; gi16128668) and CadB (residues 9 -444; gi16131958) using the ClustalW algorithm (35). All five protein sequences bear the conserved motifs characteristic of the APC superfamily (11,12), including the invariant residues indicated with black triangles under each block of the alignment. Amino acid residues important for the putrescine uptake and/or excretion activity of PotE (47) are indicated with inverted black triangles over the PotE sequence. The 12 transmembrane regions predicted for PotE (47) or Agp2p (using the HHMTOP algorithm) (34) are delineated next to the respective sequences with a series of asterisks within a pair of parentheses. Note that residues 504 -532 inclusively in the LmPot1 sequence, which include predicted transmembrane region 8 and adjacent loop do not align with the other sequences shown and have been omitted for the purpose of the illustration. tain. A population of Agp2p molecules may exist in the endoplasmic reticulum and vacuolar compartments, since we have observed that Agp2p abundance is dependent on the expression of NPR1, 2 a protein kinase gene related to PTK2. Npr1p is involved in the targeting of several amino acid permeases, including GAP1, from the Golgi to the plasma membrane, and defective NPR1 results in the routing of permeases from the Golgi to the vacuolar degradation pathway (47).
Along with LmPot1 from L. major (6) and Gap1p from S. cerevisiae (14), Agp2p is the first eukaryotic permease reported exhibiting high affinity polyamine import across the plasma membrane. As expected from the fact that they both belong to the YAT permease family, Agp2p and Gap1p share extensive homology (28% identity, 59% similarity) (Fig. 8). Multiple sequence alignments also clearly demonstrate that Agp2p is similar to LmPot1 (10.9% identity, 23.0% similarity) as well as to two APC transporters from Gram-negative bacteria, namely PotE (10.3% identity, 33.5% similarity) and CadB (9.0% identity, 28.9% similarity), which catalyze putrescine/ornithine (48) and cadaverine/lysine antiport (49), respectively. Interestingly, pairwise alignment indicates that several of the residues shown to be crucial for putrescine transport activity in PotE (47) (Fig. 8). Interestingly, the only permease among the YAT family possessing an aromatic residue at the position equivalent to Trp 292 and Trp 289 in the PotE and CadB sequences, respectively, is Agp2p (i.e. Tyr 400 ), in view of the fact that this residue is crucially important in forming hydrophobic contact with the butyl moiety of putrescine. Transmembrane regions and associated loops of the Na ϩ -K ϩ -2Cl Ϫ cotransporter NKCC2 (50), as well as a few members of the APC and L-type amino acid transporter families, are the closest counterparts to Agp2p among known mammalian proteins (data not shown), but such phylogenetic relationships are expected from the conserved motifs that are common to all of these transporters (51). Thus, we have not yet identified specific structural characteristics distinguishing Agp2p from other YAT permeases that might help to identify potential mammalian candidates among the several APC-related members of the APC superfamily.
Although Agp2p is obviously a major agent in polyamine transport, the behavior of the various substrates attributed to this permease is rather complex. We were unable to demonstrate competitive inhibition of spermidine transport even by a Ͼ1000-fold molar excess of exogenous L-carnitine, despite the fact that L-carnitine has an affinity for its transport system comparable with spermidine (24). One possibility is that Agp2p has distinct, functionally independent substrate binding sites that recognize determinants specific for carnitine and polyamines, respectively, without cross-inhibiting the uptake of the other substrate (Fig. 9). Although a rare occurrence, such an arrangement is thought to account for the very poor crosscompetition between sucrose and biotin for H ϩ -sucrose symporters that exhibit dual specificity for these substrates in plants (52). Furthermore, studies with chimeric transporters have shown that two functional substrate binding sites with nonoverlapping specificity can coexist in different regions of a single permease and efficiently catalyze translocation of their cognate substrate (53). Thus, it is theoretically conceivable that Agp2p behaves as a permease with a dual specificity as a carnitine and polyamine transporter, with both functions being carried out virtually independently one from another (Fig. 9).
Alternatively, Agp2p might act as a common plasma membrane sensor for signals that induce the downstream expression of distinct permeases, respectively, responsible for the specific transport of polyamines, carnitine and other substrates, in a manner analogous to Ssy1p, another member of the YAT family (54 -56) (Fig. 9). Ssy1p possesses the distinctive structural features of permeases of the YAT family but has no significant transport function per se. Instead, it interacts with two peripheral membrane proteins, namely Ptr3p and Ssy5p, to form the SPS sensor system that detects extracellular amino acids and regulates the transcription of an array of genes related to nitrogen metabolism (57). A similar role for Agp2p as a plasma membrane sensor of extracellular signals would provide an explanation for the complete lack of competition for transport between the various substrates found to be under the control of an active AGP2 gene (Fig. 9).
A third, unifying hypothesis to account for the multiplicity of noncompeting substrates for Agp2p might be a mixed sensor/ transporter model (Fig. 9), as in the case recently described for Gap1p (58). In addition to its well established function as a general amino acid permease, Gap1p apparently transduces signals down the fermentable growth medium pathway that leads to rapid protein kinase A-dependent activation of several intracellular processes (58). Specific regions of Gap1p such as the C terminus are involved in this signal transducing activity independently of its transport function, indicating that the sensor activity does not merely result from accumulation of signaling amino acids (59). It is therefore intriguing to postulate that Agp2p, which is a close phylogenetic relative of Ssy1p (13), might also represent another example of a sensor/transporter. Accordingly, Agp2p might function as a plasma membrane permease that specifically imports polyamines and as a sensor regulating the downstream expression of a set of plasma membrane permeases different from that controlled by the FIG. 9. Schematic illustration of three different models for the mechanism of action of Agp2p. A, the dual specificity permease model. Agp2p is a classic permease with at least two independent binding sites specific for carnitine and polyamines, respectively. Both types of substrates are transported via Agp2p. B, the sensor model. Agp2p is a plasma membrane sensor similar to Ssy1p. A signaling molecule (e.g. spermidine) binds to Agp2p, resulting in the activation of a signaling cascade (triggered by an effector site (Eff)) that up-regulates the expression of substrate-specific permeases (of unknown identity), including a polyamine-specific and a carnitine-specific permease. Agp2p itself may not transport the signaling molecules. C, the mixed permease/sensor model. Agp2p is a permease (both putrescine-and spermidine-specific) that transports polyamines. Either a region of the permease (Eff) or the transported polyamines activate the regulatory pathway that up-regulates the expression of a carnitine permease of unknown identity.
Ssy1p/Ptr3p/Ssy5p system. The main transport function of Agp2p, if it acts as a permease, is most likely polyamine influx, since carnitine uptake activity is observed almost exclusively upon induction of lipid ␤-oxidation (24), and its postulated amino acid transport activity is significant only in cells deficient in GAP1 and AGP1 expression (41). Such a model would obviously challenge the previously published assignment of Agp2p as a carnitine transporter (24), since the sensor model would predict that Agp2p would in fact regulate the expression of an oleate-inducible carnitine permease of unknown identity. Likewise, Agp2p would signal the expression of an as yet undefined low affinity permease that catalyzes the uptake of certain amino acids when Gap1p and Agp1p function is suppressed. The latter permease might in fact be Agp3p, which would account for the fact that both Agp2p and Agp3p are required for the activity of this low affinity transport system (41). A role for polyamines as inducers of heterologous permeases would not be unprecedented in yeast. Indeed, spermidine has been shown to be a sensitive inducer of choline transport in yeast (60), which is carried out by the Hnm1p permease, a close relative of Tpo5p and Uga4p within the APC family (61).
The definitive elucidation of the substrate specificity and mechanism of action of Agp2p in the polyamine, carnitine, and amino acid transport and its possible sensing function is a current subject of investigation. Nevertheless, the present data demonstrate that Agp2p qualifies well as the major polyamine permease in S. cerevisiae and raise important questions as to its dual connection with polyamine homeostasis and lipid ␤-oxidation. One intriguing possibility is that an increased rate of polyamine utilization might be required to provide the high amounts of acetyl-CoA required for the metabolic induction of peroxisomal proliferation and lipid ␤-oxidation. Indeed, pantothenate, the precursor of acetyl-CoA, is derived from ␤-alanine, which is a by-product of spermine oxidation by the amine oxidase encoded by FMS1 (43). Thus, Agp2p might salvage putrescine and spermidine massively excreted by the Tpo5p efflux permease (13) to supplement polyamine utilization for acetyl-CoA synthesis. Such a physiological response could contribute to explain a role for polyamine sensing through Agp2p in regulating the carnitine uptake system.