Endocytosis and degradation of the yeast uracil permease under adverse conditions.

Yeast uracil permease follows the secretory pathway to the plasma membrane and is phosphorylated on serine residues in a post-Golgi compartment. The protein was found to be rather stable in growing cells, but its turnover rate (half-life of about 7 h) was much faster than that of most yeast proteins. Several adverse conditions triggered the rapid degradation of uracil permease, and so a loss of uracil uptake. Turnover was rapid when yeast cells were starved of either nitrogen, phosphate, or carbon, and as they approached the stationary growth phase. Rapid permease degradation was also promoted by the inhibition of protein synthesis. The degradation of uracil permease in response to several stresses was strikingly slower in the two mutants, end3 and end4, that are deficient in the internalization step of receptor-mediated endocytosis. Thus, internalization is the first step in the permease degradative pathway. Uracil permease is degraded in the vacuole, since pep4 mutant cells lacking vacuolar protease activities accumulated large amounts of uracil permease, which was located within the vacuole by immunofluorescence. We have yet to determine whether adverse conditions enhance permease endocytosis and subsequent degradation or divert internalized uracil permease from a recycling to a degradative pathway.

Yeast uracil permease follows the secretory pathway to the plasma membrane and is phosphorylated on serine residues in a post-Golgi compartment. The protein was found to be rather stable in growing cells, but its turnover rate (half-life of about 7 h) was much faster than that of most yeast proteins. Several adverse conditions triggered the rapid degradation of uracil permease, and so a loss of uracil uptake. Turnover was rapid when yeast cells were starved of either nitrogen, phosphate, or carbon, and as they approached the stationary growth phase. Rapid permease degradation was also promoted by the inhibition of protein synthesis. The degradation of uracil permease in response to several stresses was strikingly slower in the two mutants, end3 and end4, that are deficient in the internalization step of receptor-mediated endocytosis. Thus, internalization is the first step in the permease degradative pathway. Uracil permease is degraded in the vacuole, sincepep4 mutant cells lacking vacuolar protease activities accumulated large amounts of uracil permease, which was located within the vacuole by immunofluorescence. W e have yet to determine whether adverse conditions enhance permease endocytosis and subsequent degradation or divert internalized uracil permease from a recycling to a degradative pathway.
Endocytosis plays a key role in the physiology of eucaryotic cells. In mammalian, it is involved in nutrition, removal of unwanted molecules from the plasma membrane, and in the cell response to hormones. Many polypeptide hormones and their receptors are cleared from the cell surface by endocytosis (1). The intracellular fates of a number of receptors and their ligands have been described in detail. Some receptors undergo endocytosis regardless of their boundfree state, while other receptors only undergo endocytosis when bound to ligand (2). Internalized receptors pass through complex vesicular and tubular structures, defined morphologically and biochemically as early and late endosomes (1). From these compartments, the endocytic pathway branches, allowing either recycling to the cell surface or delivery to the lysosomes for proteolysis. Permanent internalization followed by recycling to the plasma membrane has also been described for the GLUT4 glucose transporter (3). The internalization signals for this protein and for several receptors have been identified (2, 3). In contrast with the knowledge about endocytic pathways or signals, information relative to the endocytic machinery is still rather limited. Biochemical studies have focused mainly on the role of clathrin Scientifique Grant C00031. The costs of publication of this article were * This work was supported by the Centre National de la Recherche defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed. and its associated proteins, adaptins, in the internalization step (4), and more recently on the role of rab proteins in the vesicular traffic (5).
The use of a genetic approach of endocytosis has been developed in recent years in Saccharomyces cerevisiae in order to identify new proteins involved in this complex process and to investigate their in vivo function. Two markers were initially developed. Lucifer yellow was used to follow fluid-phase endocytosis, and a-factor was used to follow receptor-mediated endocytosis (6, 7). a-Factor is a pheromone that binds to its specific receptor, the product of the STE2 gene, and triggers its endocytosis, as judged from the clearance of receptor activity from the cell surface (8). This ligand is internalized and passes through endocytic compartments before being delivered to the vacuole where it is degraded (9). Two genes, END3 and ENDI, necessary for the internalization of a-factor linked to its receptor, have recently been identified (10). A vesiculation step requiring the SEC18 gene product would be involved in the targeting of a-factor from early to late endosomes (11). The delivery of a-factor to the vacuole depends upon the YPT7 gene product, which is homologous with the mammalian rab7p (12). The intracellular fate of a plasma membrane protein when passing along the endocytic pathway was recently described for the first time in yeast, both at the biochemical and morphological level (13). This study showed that the a-factor receptor may undergo two modes of endocytosis depending on the presence or absence of its ligand. Both pathways deliver the receptor to the vacuole for degradation. The occurrence of a recycling pathway in yeast was proposed as a way to explain certain features of the mutant renl impaired in a-factor receptor endocytosis (13).
The number of endocytic markers in yeast presently available remains very limited. Hence, following the fate of other proteins would help both to define hypothetical endocytic pathways alternative to that followed by the pheromone receptors, and to clarify the functions of endocytosis in yeast. The permeases, which catalyze the entry of essential nutrients, are likely to be good candidates for use as tools to probe endocytosis and its role in the turnover of plasma membrane proteins. Nutrient uptake is indeed very sensitive to environmental changes. Yeast cells growing on non-fermentable carbon sources respond to the addition of glucose by inactivating the galactose, maltose, and high affinity glucose transport systems (14,15). Similarly, several amino acid permeases are inactivated by ammonia (16). The observation that these catabolite inactivations are reversible, except after the inhibition of protein synthesis, has led to the suggestion that they result from proteolytic degradation of the permeases (14). But although more than 20 genes encoding plasma membrane transporters have been cloned and sequenced to date in S. cerevisiae, the biochemical information on these proteins is very limited (17,18). Apart from data on the targeting of some permeases at the plasma membrane (19, 201, the intracellular fate of yeast permeases is poorly documented, and the molecular events of their turnover have not yet been analyzed.

Yeast Uracil Permease lhmouer
One of the difficulties in studying permeases at the protein level is that yeast cells contain very little of these proteins.
Overproducing the FUR4-encoded uracil permease (21) in an active form without deleterious effect for yeast cells enabled us to undergo molecular analysis of the intracellular fate of this permease. Newly synthesized uracil permease is delivered to the cell surface via the secretory pathway and is phosphorylated in a post-Golgi compartment on its way to and/or within the plasma membrane (20). We noticed in that study that uracil permease has a short half-life at high temperatures (20). We therefore investigated the turnover of uracil permease to define the situations triggering proteolysis of the permease and the pathway involved in degradation. The data presented here indicate that uracil permease is rather stable in growing cells. However, under adverse circums~nces, uracil permease is selected for degradation. The permease is internalized by endocytosis before it undergoes proteolysis, since proteolysis is severely delayed in the two mutants end3 and end4 that are deficient in receptor-mediated endocytosis. These experiments are the first to directly examine the turnover of a permease in S. cereuisiae at the molecular level and provide an example of endocytosis of a non-receptor plasma membrane protein in yeast.
For all the data reported in the present paper, under conditions of exponential growth (e2 x 10' cells/ml), the uracil uptake activity was raised up to 10-25 nmalrmin/2 x lo' cells for multicopy plasmid-encoded permease, compared with 0.2-1 nmoYminl2 x lo7 cells for chromosomal encoded permease.
East Cell Extracts and ~m~u~a b~o t~~~g -~ x lo1 cells were harvested, and cell extracts were prepared by lysis with 0.5 ml of 0.2 M NaOH, 0.2% mercaptoethanol for 10 min on ice. Trichloroacetic acid was added to a final concentration of 5%, and the samples were incubated for an additional 10 min on ice. Precipitates were Collected by centrifugation at 12,000 x g for 5 min. The pellets were neutralized and dissolved in 35 pl of dissociation buffer (4% sodium dodecyl sulfate, 0.1 M Tris hydrochloride; pH 6.8, 4 m EDTA, 20% glycerol, 2% 2-mercaptoethanol, 0.02% bromphenol blue) and 15 pl of 1 M Tris base, and heated at 37 "C for 10 min. 0.5 to 1 x lo7 cell equivalents were loaded onto a 10% SDS-polyacrylamide gel and subjected to electrophoresis in a Tricine' system (28). Proteins were transferred to a nitrocellulose mem- The abbreviation used is: Tricine, ~-tris(hydroxymethy1)methylglycine.
brane and treated with antibodies as described in Ref. 20. Uracil permease was detected with an antiserum against the C-terminal peptide (provided by M. R. Chevallier) that had been affinity purified by incubation with a nitrocellulose strip containing the same peptide. Antibodies specific for the plasma membrane H+-ATPase (kindly provided by R. Serrano) were used without further purification. Primary antibodies were detected with a horseradish peroxidase-conjugated anti-rabbit I@ second antibody followed by ECL chemiluminescence (Amersham), Labeling and Immunoprecipitation-Yeast cells were grown in YNB medium with glucose as the carbon source. They were labeled by adding [35SSlmethionine directly to the growth medium. Proteins were processed for immunoprecipitation as described previously (20) using rabbit antiserum raised against a peptide corresponding to uracil permease residues 15-30. Immunoprecipitated proteins were analyzed by SDS-polyacrylamide gel electrophoresis and fluorography as described previously (20).
I~~una~uorescence-Cells were grown to a density of 1.5 x lo7 celldml. Freshly prepared paraformaldehyde was added directly to the growth medium to a final concentration of 2%. Sodium azide was also added to give a final concentration of 10 nm. After 10 min, the cells were collected and resuspended at 4 x 10' cellsfml in 3.7% paraformaldehyde in buffer A (0.1 M potassium phosphate, pH 7.5, plus 10 m sodium azide). Cells were fixed for 90 min at room temperature and washed twice in buffer A and once in buffer B (1.2 M sorbitol in buffer A), They were resuspended at the same density in buffer B containing 25 m 2-mercaptoethanol. Zymolyase 20T (Seikagaku Kogyo Go. Ltd., Tbkyo) was added to a final concentration of 350 pg/ml, and cells were incubated at 37 "C for 30 min. Protoplasts were collected by centrifugation (5 min at 3000 xg) and washed once in buffer B. They were resuspended at 5 x 10' cellsfml in PBS (50 m potassium phosphate pH 7.5, 150 m s NaCI) containing 0.1% Triton X-100 and permeabilized by incubation for LO min on ice in the presence of this detergent. Fixed and permeabilized protoplasts were directly applied onto ~lylysine-coated multiwell slides for 2 min. The slides were dipped in solution C (PBS plus 1 mglml bovine serum albumin and 0.1% Tween 20) and incubated for 1 h at room temperature with affinity purified C terminus antibody (2fold diluted with solution C, compared with 1000-fold dilution for Western blots). The slides were then washed three times for 3 min each in solution C and incubated with secondary antibody (affinity purified goat anti-rabbit IgG conjugated to rhodamine (Kirkegaard and Perry Laboratories) diluted 1200 in solution C) for 30 min at room temperature. The slides were mounted with citiffuor (Citifluor Ltd., London).
Confocal Laser Scanning Microscopy-Confocal Laser Scanning Microscopy was performed using a Bio-Rad MRC-600 confocal imaging system (Bio-Rad Microscience Ltd, Hertfordshire, United Kingdom), mounted on an Optiphot I1 Nikon microscope. Images were collected using an oil immersion lens (x 60, NA 1.4 plan Apochromat). For rhodamine excitation, an Helium-Neon ion laser adjusted at 543 nm was used close to the maximum of absorption peak and for phase contrast an argon ion laser adjusted at 488 nm. The emitted signal was digitalized by "photon counting" in order to increase the signal to noise ratio, each section was scanned 90 times, and for phase contrast the signal was treated by Kalman filter (average of eight images) with a scan time of 4 dframe and rastor size of 512 lines. The pinhole ofthe confocal system was adjusted to allow a field depth of about 0.5 pm. Photographs were taken on Kodak T-Max film using a camera mounted on a VM1710 Lucius & Baer film recorder.

RESULTS
LocaEization of Overproduced Uracil Permease-The fate of uracil permease, once it had arrived at the plasma membrane, was investigated in cells which produced this permease from high copy number plasmids bearing the FUR4 gene under the control of its own promoter, or of the inducible GAL10 promoter. The turnover of the permease was followed by checking, in addition to the loss of uracil uptake, the decrease in the amount of permease on Western immunoblots. Pulse-chase experiments showed previously that the protein is phosphorylated in a post-Golgi compartment. As the permease species detected on the immunoblots were mainly phosphorylated species (20), overexpressed permease is likely to be mostly at the plasma membrane. This point was checked by immunofluorescence. Cells disrupted for the FUR4 gene were devoid of any staining Cells harvested during logarithmic growth were fixed and processed for immunofluorescence using a purified specific antibody to uracil permease a s described under "Materials and Methods.'' They were examined for phase contrast and indirect immunofluorescence using a confocal laser microscope. B 1 C decorated by the specific antibodies. Staining seemed to be mainly over the plasma membrane, and immunofluorescence was occasionally seen in vesicle-like bodies that seemed to lie under the inner surface of the plasma membrane ( Fig. 1, B and C). There was no perinuclear-labeled material unlike the situation observed for most proteins that accumulate in the endoplasmic reticulum in yeast (29). The distribution of overproduced uracil permease appeared very similar to that of the few plasma membrane proteins detected so far by immunofluorescence in yeast (30), including non-overproduced H+-ATPase (19, 31). Thus, overproduction probably does not alter the expected location of uracil permease.
lhrnover of Uracil Permease after Inhibition of Protein Synthesis-The turnover of uracil permease was estimated by following uracil uptake and the level of immunodetected permease after inhibiting protein synthesis with cycloheximide. The experiment was performed on cells grown on either glucose (Fig. 2, A and B ) or galactose (Fig. 2, A and C ). The loss of uracil uptake (Fig. 2 A ) followed the time course expected for a firstorder reaction. The calculated half-life was 2.5 h for glucosegrown cells. The loss of uptake was significantly faster (tllP = 1 h) in galactose grown cells.
Extracts from cells withdrawn at different times after adding cycloheximide were analyzed for uracil permease by immunoblots (Fig. 2, B and C). Exponentially growing cells contained several permease species having electrophoretic mobilities corresponding to different phosphorylation levels (Fig. 2B, lane 1, and 2C, lane 2). There seemed to be more highly phospho-rylated permease in glucose grown cells than in galactose grown cells. The drop in immunodetected permease appeared to parallel the loss of uracil uptake whatever the carbon source used, indicating that degradation of the permease is responsible for the drop in uracil uptake, and that permease turnover was faster in galactose grown cells. Adding glucose together with cycloheximide failed to slow the turnover (not shown), indicating that glucose is not involved in degradation per se.
Staining the blots with Ponceau Red showed that there was no significant decrease in the amount of total proteins after cycloheximide treatment (not shown). The behavior of another plasma membrane protein, W-ATPase, was used as a control in galactose grown cells, which have a rapid permease turnover. The overexpression of the permease had no effect on either the amount or the profile of the W-ATPase ( compare lanes 1 and 2,  Fig. 2C). Incubation for 2 h with cycloheximide resulted in only a slight decrease in the ATPase, while the permease had almost completely disappeared (Fig. 2C, lanes 2-5). This behavior of H+-ATPase is in agreement with the half-life of 11 h described for this protein after inhibition of protein synthesis (32).
The turnover of the uracil permease was therefore far higher than that of most of cell proteins and of the control plasma membrane protein, H+-ATPase. This suggests that there is a specific degradation after inhibition of protein synthesis. The rate of permease degradation was found to be dependent upon the carbon source for growth. The degradation required energy metabolism, as the immunodetected permease was not lost when the medium contained sodium azide (Fig. 2C, lane 6 ) . FIG. 2. Loss of uracil uptake and degradation of uracil permease after inhibition of protein synthesis. NC122-sp6 cells transformed with the plasmids pfF or pgF were grown to logarithmic growth phase on glucose or galactose medium, respectively. Cycloheximide (100 pg/ml) was then added to the medium. A, uracil uptake was measured a t different times. Results are percent of initial activities. B , protein extracts were prepared from glucose grown cells removed at the times indicated, and the uracil permease was visualized by Western blot analysis. C , protein extracts were prepared from galactose grown cells and probed with specific antisera for uracil permease and plasma membrane H+-ATPase. Lane 1, control extract from untransformed cells which do not produce permease. Lanes 2 5 , extracts from cells removed at the times indicated after the addition of cycloheximide. Lane 6 , extract from cells to which 10 mM sodium azide was added at the same time as cycloheximide.

Yeast
TLrnover of Uracil Permease in Exponentially Growing Cells-Cells expressing the FUR4 gene under the control of the inducible GAL10 promoter were used to follow the fate of uracil permease in actively growing cells. Glucose was added to galactose growing cells to stop further transcription of the FUR4 gene. As FUR4 mRNA has a very rapid turnover (t,a = 2 min) (33), adding glucose indeed rapidly blocked further synthesis of the permease. It led to a sharp drop in the amount of labeled immunoprecipitable permease within minutes (control not shown). Uracil uptake increased with time in control cells, under conditions of ongoing permease synthesis (Fig. 3A). After glucose repression of permease synthesis in growing cells, uracil uptake exhibited only a slight decrease (-20%) within 4 h (Fig. 3A). The kinetics of this loss of activity followed a firstorder reaction, with a half-time of approximately 7 h. Thereafter, there was a rapid inactivation before cells arrived a t sta- p195gF were grown to 8 x 10' celldml using galactose as a carbon source. Cells were collected, washed with galactose or glucose medium, and resuspended in fresh galactose (0) or glucose medium ( . , A) in the absence (M) or the presence (A) of cycloheximide ( C H X ) (100 pg/ml). A, aliquots were withdrawn a t various times and analyzed for uracil uptake. Results are percent of initial activity. B , proteins extracts were prepared at the same times. Proteins from 0.2 ml of culture were analyzed for uracil permease by Western blots. tionary phase. In contrast to this biphasic loss of uracil uptake during cell growth, cycloheximide immediately triggered a rapid loss of uracil uptake (Fig. 3A), as observed previously (Fig. 2 A ) . The half-life was 30 min in the presence of this inhibitor in the present experiment. The time courses of the amount of immunodetectable permease in the presence or the absence of cycloheximide were qualitatively in agreement with the kinetics of the loss of uracil uptake (Fig. 323). However, degradation seemed to proceed before any significant decrease in uracil uptake, and to be greater, especially in the absence of cycloheximide. Several independent factors might explain such a discrepancy. The uracil permease activity is dependent upon the activity of the H+-ATPase, which is influenced by several environmental changes. For instance it is activated as the cells approach the stationary phase (34). This might explain why the permease degradation was transiently offset by the activation of the remaining protein during this period. A difference in the state of phosphorylation of uracil permease might also influence its specific activity.
This experiment revealed several striking properties of uracil permease turnover. The kinetics of the loss of uracil uptake and of permease degradation appeared very different in growing and non-growing cells. The sharp drop in uracil permease brought on by cycloheximide was probably due to the stress induced by the total inhibition of protein synthesis. The same

Yeast Uracil Permease
Rrnover 9837 observation was made in cells expressing the FUR4 gene under the control of the repressible PH0.5 promoter: repression of FUR4 gene expression with phosphate left the permease activity rather stable in growing cells, whereas it rapidly decreased when cycloheximide was added (data not shown). Uracil permease appeared to be degraded only slowly in early exponential growing cells shifted from galactose to glucose. A rapid degradation was then triggered about one generation before cells arrived at the stationary growth phase (Fig. 3). Similarly, uracil uptake declined and permease was lost during late exponential growth of transformed cells having different genetic background whatever the carbon source for growth. Such a decrease of permease at the approach of the stationary phase cannot be due to the overproduction of the permease, since uracil uptake declined in a similar fashion during the growth of untransformed strains, which carried 20-50-fold less permease activity (data not shown).
TLrnover of Uracil Permease upon Nutrient Starvation-The decline in permease activity during the late exponential growth phase suggested that permease breakdown might also be triggered by the limitation of a required nutrient. To test this possibility cells were starved of nitrogen, phosphorous, or carbon source after shutting off new permease synthesis with glucose. The growth began to slow after 1 h in a nitrogen or phosphorous-free medium compared with that in standard medium, whereas the lack of carbon source led to an immediate arrest of growth. In parallel, uracil uptake was rather stable during the first hour of nitrogen or phosphorous deprivation and thereafter it declined with a half-time of 90 min, whereas it hardly decreased in standard medium (Fig. 4A). The inactivation triggered by phosphate or nitrogen starvation was accompanied by enhanced permease degradation (data not shown). Carbon starvation led to the immediate loss of all permease activity a s did azide (35), since uracil uptake is energydependent. Immunoblots of extracts from glucose-starved cells showed extensive degradation after 15 min of glucose starvation, and the loss of most of the permease within 45 min (Fig.   4B 1. These data indicate that the turnover rate of the permease increased greatly upon nutrient deprivation. As nutrient deprivation triggers an arrest in the G, phase of the cell cycle (36), we checked to see whether the increase in the turnover of uracil permease could be linked to the cell cycle arrest. a-Factor and glucose were added to a cells, which expressed uracil permease under galactose control, and permease activity was followed over time. While a-factor indeed promoted the formation of shmoos, it had no influence on permease turnover (data not shown). The lifetime of uracil permease therefore seems to be sensitive to the stress induced by nutrient starvation, but not to arrest in the G , phase of the cell cycle.
Uracil Permease Degradation Is Subsequent to Znternalization By Endocytosis-Proteolysis of a plasma membrane protein could be achieved by a direct breakdown of a selected protein at the plasma membrane, or by selective internalization and transport to the vacuole for non-specific proteolysis (37). End3 and end4 mutant cells were used to determine whether uracil permease was removed from the cell surface before its degradation. These mutants are thermosensitive for growth (10). They fail to internalize a-factor, its receptor, or to accumulate a fluid-phase marker in the vacuole. The end4 mutant has a thermosensitive endocytic defect: the end4 mutant protein is partially functional a t 24 "C, and irreversibly inactivated a t 37 "C. The end4 mutation affects an early step in the internalization pathway (10). The loss of uracil uptake and the degradation of uracil permease upon inhibition of protein synthesis by cycloheximide were compared in isogenic end3, end4, and wild type cells. In the wild type cells, raising the temperature from 25 to 37 "C accelerated (x 3) the loss of uracil uptake  (control). A, uracil uptake was followed a t various times after resuspension. Results are percent initial activity. B, an aliquot was resuspended at t = 0 in a minimal medium containing a phosphate and a nitrogen sources, but lacking a carbon source. Protein extracts were prepared at the times following resuspension indicated above each lane. Proteins corresponding to 0.2 ml of cultures were analyzed for uracil permease by Western blot.
( Table I). This effect was found in all the wild type strains tested. Thus, heat stress further enhanced the rate of degradation triggered by cycloheximide. A mild heat shock also resulted in an increase in permease turnover, even in growing cells.' The disruption of the END3 gene conferred partial protection against the inactivation of uracil permease (Table I and Fig. 5A 1. The time required to lose 50% of uptake was 2-fold greater than that of wild type cells a t 25 "C and 4-fold greater a t 37 "C (Table I). The coordination between the drop in uracil uptake and the loss of immunodetected permease for both end3 and wild type strain (Fig. 5B) clearly showed that the absence of the END3 gene product slowed permease degradation. The end4 mutation provided a striking protection against both permease inactivation and degradation (Fig. 5, A and B ). This effect was already seen a t 25 "C. The half-time of uracil uptake was twice that of wild type strain (Table I). Increasing the temperature from 25 to 37 "C, which accelerated inactivation in all wild type strains, and even in the end3 strain, almost completely prevented any drop in uracil uptake in the end4 strain (Table I and Fig. 5A). I t was less than 10% during the 2-h treatment with J. M. Galan, personal communication.

Half-life of uracil uptake
Half-life of uracil permease at 25 and 37 "C in wild type and end mutant cells. The experiment was performed as described in the legend to Fig. 5. Cycloheximide was added to the cultures at 25 "C or when the cultures were shifted from 25 to 37 "C.

end3 end4
25 "C 65 130 140 37 "C 25 95 >240 min cycloheximide at 37 "C, compared with over 90% in the wildtype strain (Fig. 5A). In the meantime, the immunodetected permease, which disappeared in the wild type strain, was still present at high level in the end4 mutant strain (Fig. 5B ). However, some degradation still occurred.
These data clearly indicate that the END3 and END4 gene products are involved in the degradation of uracil permease triggered by cycloheximide. Internalization of uracil permease is therefore required prior to its degradation.
We checked that internalization also preceded the degradation of the permease induced by nutritional stress by depriving the same wild type and end cells of nitrogen source a t a restrictive temperature. After 2 h at 37 "C, the level of immunodetected permease had not dropped in end4 cells, while it had dropped to zero in wild type cells (Fig. 6). The behavior of the permease in end3 cells was intermediate (not shown). Therefore, the degradation triggered by nitrogen starvation required the same endocytosis pathway as that induced by an arrest of protein synthesis. In addition, the holding of immunodetected permease at a high level in end4 cells submitted to various kinds of stress further confirmed that a high percentage of the protein was located upstream of the end4 block, i.e. at the plasma membrane, in transformed cells overexpressing uracil permease.
Uracil Permease Is Degraded in the Vacuole-We examined the fate of permease in pep4 cells lacking vacuolar protease activities (38, 39) to determine where the permease was degraded. Pep4 mutants cells overexpressing uracil permease under the control of its own promoter carried the same permease activity as wild type-transformed cells, but about 10 times more permease was detected on immunoblots (Fig. 7A). The amounts of permease labeled in a 10-min pulse with [S351methionine in pep4 mutant cells and wild type cells were not significantly different (Fig. 7B). Therefore, increased permease level in pep4 cells resulted from a decrease in the rate of permease proteolysis. Indirect immunofluorescence in pep4 cells allowed detection of uracil permease at both the plasma membrane and in the vacuole (Fig. 7C). These data would be consistent with vacuolar degradation of uracil permease.

DISCUSSION
Uracil permease has a rather "slow" turnover (half-life of -7 h) in growing cells. The protein appears to reside for several hours at the cell surface before being sorted out for degradation. This relative stability is seen in galactose grown cells transferred to glucose medium. However, the turnover of the "stable" uracil permease in growing cells is much faster than that of the bulk of yeast proteins (t,,, > 150 h for up to 90% of total proteins) (40), or even that of the H+-ATPase (tl,2 > 20 h in glucose-growing cells) (32).
Several adverse conditions can trigger a rapid degradation of uracil permease, and thus a loss of uracil uptake. Turnover is rapid when yeast cells are starved of either nitrogen, phosphate, or carbon source, and as they approach the stationary phase. Permease degradation is also induced by other stresses, such as heat stress, or the inhibition of protein synthesis, two conditions that were also reported to affect the turnover rate of the plasma membrane H+-ATPase (32, 41). Blocking protein synthesis promoted the most spectacular effect on permease degradation: it can induce up to a 15-fold increase in the permease turnover rate. Various stresses can obviously have additional effects: the rate of permease degradation triggered by cycloheximide increases with an upshift in temperature, and as the cells approach the stationary phase. It should be emphasized that the time course of permease inactivation and degradation appear to depend also upon numerous other factors, such as the genetic background of the strains used, or the carbon source for growth.
The acceleration of permease turnover rate by stressful situations is not due to bulk endocytosis, as the receptor-mediated internalization of a-factor is not altered by stress (42). Uracil permease degradation seems to be rather specifically regulated. The permease is entirely degraded after inhibition of protein synthesis for 2 h, whereas the bulk of yeast proteins and the plasma membrane H+-ATPase hardly decrease. Uracil permease is rather stable in growing cells supplemented with glucose, a condition known to promote the catabolite inactivation of several sugar transporters (14, 43,44). In contrast, uracil permease is rapidly lost upon carbon starvation, a stress known to stabilize sugar transporters in yeast (14, 43,44). The physiological relevance of the regulation of uracil permease degradation is not clear. Uracil permease activity is required to provide uracil to pyrimidine-deficient yeast strains or to pump the uracil excreted in the medium (35, 45). Internal uracil is used, via UMP (45) as an alternative pathway to provide pyrimidine for RNA synthesis, i.e. mainly for ribosomal RNA synthesis. The rate of rRNA genes transcription, and more generally ribosome synthesis, is believed to be a sensitive barometer of the cell's prospects (46). It drops during heat stress (47), upon arrival at the stationary phase, or upon inhibition of protein synthesis (48). The release of uracil permease from the cell surface and its subsequent proteolysis under adverse conditions might therefore be correlated with the regulation of ribosome synthesis.
It will be necessary to identify the signals involved in the regulation of uracil permease turnover rate. As observed for several receptors, variations in the phosphorylation state of the permease might regulate its stability. When compared with galactose grown cells, glucose grown cells appear to be enriched in highly phosphorylated permease species. In parallel, the rate of permease degradation triggered by cycloheximide is more rapid in galactose grown cells than in glucose grown cells. Identification of the phosphorylated residues of uracil permease and their in vitro mutagenesis should indicate whether permease phosphorylation is indeed involved in the control of its turnover rate.
On the other hand, uracil permease includes a nine amino acid sequence (RIALGSLTD) that is very similar to the "destruction box" consensus sequence (FEALGXIXN) present in mitotic cyclins (49). It has been demonstrated that this sequence is required for the ubiquitin-dependent proteolysis of a sea urchin B cyclin (50). Although to date, ubiquitin-dependent degradation has been shown to involve cytosolic proteolysis by the proteasome (511, there are rare examples in mammals of the conjugation of ubiquitin to plasma membrane proteins (52,53). Furthermore, ubiquitinated laembrane-bound proteins have been found in the yeast vacuole (54) and mammalian lysosome (55). A role of this nine amino acid sequence in the regulation of permease turnover rate would agree with the enhancement of permease degradation under stress conditions. Such conditions are known to induce transcription of the polyubiquitin UBI4 gene and synthesis of several ubiquitin-conjugating enzymes (56). Current investigations are examining  1C ( e n d l ) , and RH144-3Dend3 (end3) cells which overexpressed the FUR4 gene under the control of its own promoter were grown at 25 "C with glucose as a carbon source to a same cell density (lo7 celldml). At t = 0 cultures were transferred to 37 "C, and cycloheximide was added to the medium. A, uracil uptake was measured a t various times. Results are percent initial activity. B , protein extracts were prepared at the times indicated. Protein extracts from 0.2-ml cultures were analyzed for permease by Western immunoblotting. They were transferred to 37 "C in a medium lacking nitrogen source. Protein extracts were prepared before and after a 2-h deprivation. Proteins from 0.2 ml of culture were analyzed for immunodetectable permease.

Yeast Uracil Permease Turnover
whether the destruction box sequence indeed plays a role in uracil permease degradation. The degradation of uracil permease in response to several stresses is slowed in end3 mutant and severely reduced in end4 mutant strains at 37 "C. Uracil permease is therefore degraded after it has been internalized by endocytosis, a process which requires the End3p and End4p proteins (10). Whatever the function of the End3p, it is partially bypassed, since the defect in permease internalization was only partial in an endd-disrupted strain. The defect in uracil permease internalization is immediate in the end4 thermosensitive mutant upon upshift to a non-permissive temperature, as observed for both a-factor internalization and for clearance of a-factor receptor (10). The permease activity is almost completely maintained in end4 mutant cells at non-permissive temperature although some degradation still occurs. This indicates that end4 deficiency completely blocks internalization and that a certain amount of the protein is present in an internal pool within the endocytic pathway prior to the onset of degradation. This pool, however, would represent only a small percentage of the immunodetected permease in these glucose grown cells. Preliminary experiments let us to show that the degradation of uracil permease is partially inhibited in a secl8 mutant at non-permissive temperature. The Secl8p, the yeast homolog of the mammalian NSF (571, is required in the secretory and endocytic pathways when vesiculation steps proceed. It would be necessary for the transit of internalized a-factor from early up to late endosomes (11). Thus, the partial defect in permease degradation observed in a secl8 mutant implies that at least part of internalized permease requires a vesiculation step to proceed in the degradative pathway.
The degradation of endocytosed uracil permease probably occurs in the vacuole. Pep4 mutant cells lacking vacuolar protease activities (38) accumulate large amounts of uracil permease within the vacuole. However, this accumulation occurs even in the exponential growth phase. This appears to contradict other data indicating that uracil permease has a rather long lifetime in exponentially growing cells, a t least in some experimental conditions (Fig. 3). We have yet to determine whether uracil permease undergoes continuous endocytosis, followed by vacuolar degradation in some strains or under some growth conditions, or whether part of newly synthesized permease is delivered directly to the vacuole due to overproduction. The latter possibility would be consistent with the idea that the default pathway for membrane proteins would be the vacuole rather than the plasma membrane (58). Data presented in this report conclusively show that uracil permease undergoes endocytosis and subsequent proteolysis under several stress situations. This event is dependent upon END3 and END4 gene products, indicating that common proteins are involved in the internalization of receptors and transporters in yeast. Uracil permease appears as an appropriate marker for studies of the endocytic pathway for transporters. Uracil permease can be rather stable, or rapidly degraded, a situation which might be representative of that of other permeases. I t will be important to understand whether stable permeases remain in a stable location at the plasma membrane, or undergo constant recycling through an early-like endosome. Analysis of the mechanisms triggering the degradation of uracil permease which is highly sensitive to environmental changes will expand our understanding of endocytosis in yeast, especially under stress conditions.