A Calcineurin-dependent Switch Controls the Trafficking Function of (cid:1) -Arrestin Aly1/Art6

Background: In response to nutrient signals, (cid:1) -arrestins selectively regulate trafficking of membrane transporters. Results: Aly1 is a substrate of the phosphatase calcineurin, and dephosphorylation triggers Aly1-dependent internalization of the permease Dip5. Conclusion: Endocytic function of (cid:1) -arrestins is stimulated by removal of inhibitory phosphorylation. Significance: These insights define a molecular mechanism controlling the function of an (cid:1) -arrestin in endocytosis, which is critical for cellular adaptation. shown, to share (18). We located 22 phospho-sites in Aly1, identified a subset of these as regulated by calcineurin, and delineated the specific P X I X IT-dock-ing motif in Aly1 needed for its interaction with and dephosphorylation by calcineurin. We further show that Aly1 mutants that cannot be dephosphorylated by calcineurin or that mimic persistent phosphorylation are unable to reduce cell surface levels of Dip5. Thus, these studies identify a new role for calcineurin in membrane trafficking; calcineurin promotes vacuolar trafficking of Dip5 by dephosphorylating Aly1, thereby stimulating its endocytic function. By contrast, we found that dephosphorylation of Aly1 is not required for its role in the intracellular sorting of Gap1. Our data add to the growing body of evidence that (cid:1) -arrestin-mediated trafficking is strictly con-trolled by a phosphorylation-dependent switch wherein phosphorylation blocks and dephosphorylation promotes the function of an (cid:1) -arrestin in endocytosis. This study further identifies the first phosphatase responsible for direct regulation of an (cid:1) -arrestin.

Cellular adaptation to environmental changes requires tight regulation of cell surface proteins. Specific ligands, excess nutrients, or stress factors induce endocytosis of plasma membrane receptors and permeases, thereby impacting nearly every aspect of cell physiology. Thus, a dynamic interplay exists between such extrinsic signals and endocytosis of specific membrane proteins, whose removal from the cell surface can alter intracellular signaling (1). One of the clearest examples of this interplay is the regulation of G-protein-coupled receptors (GPCRs) 4 by the ␤-arrestin class of trafficking adaptors. In mammalian cells, agonist-stimulated GPCRs initiate intracellular signaling that leads to feedback phosphorylation of the receptor (2,3) and, in several cases, dephosphorylation of ␤-arrestins (4 -6). Dephosphorylated ␤-arrestins associate with the plasma membrane and bind both GPCRs and clathrin to stimulate GPCR endocytosis (4 -10). GPCR removal from the plasma membrane dampens signaling. Although ␤-arrestin function is clearly regulated by phosphorylation, in many cases the kinases and phosphatases responsible for this regulation have not been identified.
How do ␣-arrestins achieve this signal-induced cargo selectivity? Recent reports suggest that phosphorylation blocks ␣-arrestin-mediated endocytosis and, conversely, that dephosphorylation releases this inhibition (17,34,35). For example, Art1-mediated endocytosis of the arginine permease Can1 is impaired when Art1 is phosphorylated by Npr1 (34), a kinase that is activated during nitrogen starvation, a condition known to inhibit permease internalization (35)(36)(37). Similarly, in response to nitrogen starvation, Npr1-dependent phosphorylation of Bul1 and Bul2, recently identified as yeast ␣-arrestins, impairs Bul-mediated endocytosis of the general amino acid permease Gap1 (35). In addition, phosphorylation of ␣-arrestin Rod1/Art4 by Snf1 (mammalian ortholog is the AMP-and ADP-activated protein kinase) blocks its ability to internalize the lactic acid permease Jen1, whereas dephosphorylation of Rod1 promotes endocytosis of Jen1 (17). In mammalian cells similar phospho-inhibition was recently demonstrated; phosphorylation of ␣-arrestin TXNIP by AMP-and ADP-activated protein kinase induces ␣-arrestin degradation, thereby impeding endocytic turnover of the glucose transporter GLUT1 (32). Hence, identification of the kinases and phosphatases that modify ␣-arrestins provides important mechanistic insights into the regulation of ␣-arrestin-mediated trafficking. Although dephosphorylation of at least some ␣-arrestins appears to be required for their function in permease endocytosis, direct dephosphorylation by a specific phosphatase has not yet been shown for any ␣-arrestin.
One signal-regulated phosphatase that is a good candidate for ␣-arrestin regulation is calcineurin (also called phosphoprotein phosphatase 2B or PP2B), a calcium and calmodulindependent phosphoprotein phosphatase conserved across eukaryotes and the target of the immunosuppressant drugs, FK506 and cyclosporin A. In mammals, calcineurin is abundant in the brain, where it regulates synaptic vesicle endocytosis, neural outgrowth, and synaptic plasticity; in other tissues, it impacts a wide range of processes, playing critical roles in T-cell differentiation and muscle development (38,39). Although not essential in yeast under standard growth conditions, calcineurin is required for survival under various stressful conditions, such as in the presence of high salt, at alkaline pH, or in the presence of cell wall perturbants (40). Calcineurin promotes survival under stress conditions by binding to substrates that contain a conserved docking motif (PXIXIT and variants thereof) (39,41) and dephosphorylating these targets to facilitate their function in stress adaptation. For example, dephosphorylation of the transcription factor Crz1 in response to cell wall stress induces expression of cell wall maintenance genes (42), whereas dephosphorylation of Hph1, an endoplasmic reticulum resident protein involved in protein translocation, assists in adaptation to alkaline pH (43,44). In this regard, the first indication that calcineurin might also regulate endocytosis in yeast was the observation that calcineurin dephosphorylates two plasma membrane-localized phosphatidylinositol 4,5-bisphosphate-binding proteins (Slm1 and Slm2) to promote heatinduced internalization of the uracil permease Fur4 (45).
Here, we show that ␣-arrestin Aly1/Art6 is a new substrate of calcineurin and demonstrate that dephosphorylation of Aly1 by calcineurin is required for Aly1-mediated internalization and delivery to the vacuole of a nutrient permease. Aly1/Art6 and its closely related paralog Aly2/Art3 regulate intracellular sorting of the general amino acid permease Gap1 in response to nitrogen supply (24). Aly2 also promotes endocytosis of the aspartic acid/glutamic acid permease Dip5, a function that Aly1 was suggested, but not previously shown, to share (18). We located 22 phospho-sites in Aly1, identified a subset of these as regulated by calcineurin, and delineated the specific PXIXIT-docking motif in Aly1 needed for its interaction with and dephosphorylation by calcineurin. We further show that Aly1 mutants that cannot be dephosphorylated by calcineurin or that mimic persistent phosphorylation are unable to reduce cell surface levels of Dip5. Thus, these studies identify a new role for calcineurin in membrane trafficking; calcineurin promotes vacuolar trafficking of Dip5 by dephosphorylating Aly1, thereby stimulating its endocytic function. By contrast, we found that dephosphorylation of Aly1 is not required for its role in the intracellular sorting of Gap1. Our data add to the growing body of evidence that ␣-arrestin-mediated trafficking is strictly controlled by a phosphorylation-dependent switch wherein phosphorylation blocks and dephosphorylation promotes the function of an ␣-arrestin in endocytosis. This study further identifies the first phosphatase responsible for direct regulation of an ␣-arrestin.

EXPERIMENTAL PROCEDURES
Yeast Strains and Growth Conditions-Yeast strains used in this study and their construction are described in supplemental Table 1 (84,85). Synthetic complete (SC) medium was prepared as in Refs. 24, 46 with 2.5 g/liter (NH 4 ) 2 SO 4 used routinely as the nitrogen source. Minimal (MIN) medium was made in the same way as SC, except that only the amino acids required for growth of auxotrophic strains were provided. Cells were grown at 30°C unless otherwise indicated. For growth assays on agar plates, 5-fold serial dilutions of stationary phase cultures with a starting density of ϳ1 ϫ 10 7 cells/ml were plated onto the indi-cated medium and grown for 3-6 days at 30°C. Rapamycin (LC Laboratories, Woburn, MA), azetidine 2-carboxylic acid (AzC) (Sigma), and canavanine (Can) (Sigma) were added to either SC or MIN ϩ 0.5% (NH 4 ) 2 SO 4 at the concentrations indicated. Yeast two-hybrid analyses employed yeast strain PJ69-4a (47) containing pGBT9-derived plasmids bearing Gal4 DNA-binding domain (DBD) fusions and pACT2-derived plasmids bearing Gal4 transcriptional-activation domain (TAD) fusions. Transformants were plated as serial dilutions onto SC medium lacking leucine and tryptophan, as a positive control for cell growth, or additionally lacking histidine or adenine, where growth is a read-out of GAL1prom-HIS3 or GAL2prom-ADE2 reporter activation, respectively. A competitive inhibitor of His3, 3-aminotriazole (Sigma), was added where indicated to the SC medium lacking histidine to increase the amount of GAL1prom-HIS3 reporter expression needed to allow growth. Yeast cells were transformed using the lithium-acetate method (48).
Plasmids and DNA Manipulations-Plasmids used in this study and their construction, where applicable, are described in supplemental Table 2 (86 -90). Plasmids were generated using standard recombinant DNA methods (49) and propagated in Escherichia coli strain DH5␣. Site-directed mutagenesis, used to generate Aly1 plasmids with altered CN-binding sites or mutated serines/threonines, was performed with either the QuikChange II site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) or using a similar methodology with Phusion high fidelity DNA polymerase (New England Biolabs, Ipswich, MA) and appropriately mutated primers. All constructs generated by PCR were verified using Sanger sequencing (50).
Yeast Protein Extraction, Purification, and Immunoblotting-Yeast protein extracts were generated by growing cells to midexponential phase at 30°C (A 600 nm ϭ 0.5-1.0) and harvesting an equal number of cells by centrifugation. In some cases, cell cultures were pretreated with either 2 g/ml FK506 (LC Laboratories) for 30 min to inhibit CN or with 200 mM calcium chloride for 10 min to stimulate CN-mediated dephosphorylation. Cells were then lysed using sodium hydroxide, and proteins were precipitated using trichloroacetic acid (TCA) (51). Precipitated protein was solubilized by resuspension in SDS/urea sample buffer (40 mM Tris (pH 6.8), 0.1 mM EDTA, 5% SDS, 8 M urea, and 1% ␤-mercaptoethanol) and heating to 37°C for 15 min. An equal amount of extract was resolved by SDS-PAGE and specific proteins identified by immunoblotting (for a list of antibodies used see below). Immunoblotting with anti-Pgk1, anti-Pma1, or anti-Zwf1 (glucose-6-phosphate dehydrogenase, hereafter referred to as G6PD) was used to assess protein loading.
To assess Dip5-GFP stability, MKY1800 (aly1⌬ aly2⌬ DIP5-GFP) cells containing pRS315, pRS315-Aly1, pRS315-Aly1 ⌬PILKIN , pRS315-Aly2, pRS315-Aly1 5E , or pRS315-Aly1 5A were grown to mid-exponential phase at 30°C in MIN ϩ 0.5% (NH 4 ) 2 SO 4 and treated with 200 g/ml aspartic acid and glutamic acid to trigger Dip5 endocytosis. Cell samples were taken at times indicated post-aspartic/glutamic acid addition, and total protein was extracted using the sodium hydroxide lysis and TCA precipitation method described above. Dip5-GFP sig-nal intensity, normalized for loading using G6PD, was quantified using ImageJ (National Institutes of Health), and the percentage of Dip5 remaining at each time point post-aspartic acid/glutamic acid addition was plotted. A representative data set from at least three independent experiments is shown.
Yeast extracts for pulldowns using GST fusion proteins were prepared by growing either strain BJ5459 or JR11 containing the pKK212-derived Aly1 or Aly2 expression plasmid indicated to mid-exponential phase in SC medium lacking tryptophan and inducing expression of CUP1 promoter-driven GST or GST-Aly fusions with 200 M CuSO 4 for 60 min. Where indicated, cells were also treated with either FK506 or CaCl 2 as described above. Cells were harvested by centrifugation, washed, frozen in liquid N 2 , and stored at Ϫ80°C. Cell pellets were then resuspended in co-IP buffer (50 mM Tris-HCl (pH 7.4), 15 mM EGTA, 100 mM NaCl, 0.2% Triton X-100, 5 mM N-ethylmaleimide, with phosphatase inhibitors (5 mM NaF, 5 mM Na 2 MoO 4 , 5 mM EDTA, 5 mM EGTA, 1 mM sodium orthovanadate, 2.5 mM ␤-glycerophosphate, 10 mM sodium pyrophosphate, and 0.4 mM sodium metavanadate), and protease inhibitors (Complete protease inhibitor mixture tablets, Roche Applied Science)) and lysed at 4°C using glass beads and vigorous vortexing. GST fusion proteins were purified from equal concentrations of clarified lysates by incubation with glutathione-Sepharose beads (GE Healthcare). Pulldowns were washed three times in 500 l of co-IP buffer, aspirated to dryness, eluted in Laemmli buffer (52), resolved by SDS-PAGE, and assessed by immunoblotting. GST-Aly1 proteins treated with calf intestinal alkaline phosphatase (CIP) (New England Biolabs) were purified as described above, washed with CIP buffer, and incubated with 30 units of CIP for 30 min at 37°C prior to elution in Laemmli buffer (52) and further analyses.
In Vitro Dephosphorylation Assays-GST-fused Aly1, Aly1 ⌬PILKIN , Aly2, and Crz1 were purified on glutathione-Sepharose beads from BJ5459 cell extracts as described above. Under these conditions, GST-arrestins purify with several interacting proteins (24), including protein kinases. 5 To allow in vitro phosphorylation (and radiolabeling) of arrestins by these copurifying protein kinases, bead-bound proteins were washed and resuspended in "kinase" buffer (50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 0.1 mM DTT, 0.1 mM unlabeled ATP, aprotinin, and leupeptin) and incubated at 30°C for 60 min in the presence of 75 nM [␥-32 P]ATP (PerkinElmer Life Sciences). Unincorporated 32 P and copurifying proteins were removed from the glutathione-immobilized arrestins by repeated washing of beads with co-IP buffer containing an additional 650 mM NaCl, 2 mM EDTA, and 0.8% Triton X-100. Bead-bound proteins were then resuspended in kinase buffer (listed above) with either -phosphatase (New England Biolabs) or recombinant CN-trunc, a mutant version of calcineurin that lacks its autoinhibitory domain (which is constitutively active in the absence of calcium and calmodulin), and incubated at 30°C. Samples were removed at the times indicated, beads were aspirated to dryness, and bound proteins were eluted in Laemmli buffer (52) and resolved by SDS-PAGE. Gels were stained with Gel Code Blue (Thermo Scientific), dried, exposed to a Phosphor-Imager TM screen, and imaged with a Typhoon scanner (GE Healthcare). ImageJ software was used to quantify changes in 32 P signal intensity (normalized for protein loading using quantification of Gelcode Blue staining) after incubation with phosphatase. Data were quantified from three replicates (representative data show), and the mean percentage of the original phosphorylation signal is plotted Ϯ S.E.
Mass Spectroscopy Analysis-GST-fused Aly1 or Aly1 ⌬PILKIN was purified from BJ5459 cells that were treated with either CaCl 2 (to stimulate calcineurin-mediated dephosphorylation) or FK506 (to block calcineurin-mediated dephosphorylation), as described above. Samples were resolved by SDS-PAGE and stained with Gelcode Blue (Thermo Scientific), and the bands corresponding to Aly1 or Aly1 ⌬PILKIN were excised. Proteins in each gel slice were then trypsinized and extracted as described in the "Enzymatic Digestion of Proteins from Gel Bands" protocol provided by the California Institute for Quantitative Biosciences. Mass spectrometry of extracted peptides was performed by the Vincent J. Coates Proteomics/ Mass Spectrometry Laboratory at University of California at Berkeley. A nano-LC column was packed in a 100-m inner diameter glass capillary with an emitter tip. The column consisted of 10 cm of Polaris c18 5 m packing material (Varian), followed by 4 cm of Partisphere 5 SCX (Whatman). The column was loaded by use of a pressure bomb and washed extensively with buffer A (see below). The column was then directly coupled to an electrospray ionization source mounted on a Thermo-Fisher LTQ XL linear ion trap mass spectrometer. Data collection was programmed so that neutral loss of phosphate would trigger the collection of an MS3 spectrum of the neutral loss peak. An Agilent 1200 HPLC equipped with a split line so as to deliver a flow rate of 30 nl/min was used for chromatography. Peptides were eluted using a four-step MudPIT procedure (54). Buffer A was 5% acetonitrile, 0.02% heptafluorobutyric acid; buffer B was 80% acetonitrile, 0.02% heptafluo-robutyric acid. Buffer C was 250 mM ammonium acetate, 5% acetonitrile, 0.02% heptafluorobutyric acid; buffer D was same as buffer C but with 500 mM ammonium acetate. The programs SEQUEST and DTASELECT were used to identify peptides and proteins from a database composed of the Aly1 sequence and common contaminant proteins. A minimum XCORR of 1.5, 2.2, and 3.5 was required for charge states 1, 2, and 3, respectively. To assess enrichment of phosphorylated peptides in FK506-treated versus Ca 2ϩ -treated samples, two-tailed Z-tests were preformed, and p values are reported in Table 1.
Uptake of 14 C-Labeled Amino Acids-Citrulline uptake assays were performed as described previously (24,55). In brief, BY4741 cells were made prototrophic by transformation with pCK283 (56) and either pRS426, pRS426-Aly1, pRS426-Aly2, or pRS426-Aly1 ⌬PILKIN . Cells were grown to early exponential phase (ϳ3-4 ϫ 10 6 cells/ml) in MIN ϩ 0.5% (NH 4 ) 2 SO 4 medium, collected, washed by filtration, and resuspended in nitrogen-free medium, and 20 M [ 14 C]citrulline (PerkinElmer Life Sciences) was added to cells to initiate uptake assays. Cell aliquots were removed every 30 s over four time points, collected, washed by filtration, and [ 14 C]citrulline incorporation was measured using a Beckman Coulter LS6500 Multipurpose Scintillation Counter (Indianapolis, IN). Multiple replicate assays were performed (minimum of three), and the rate of citrulline uptake (dpm/min/A 600 nm ) was determined. Aspartic acid uptake assays were performed as described for the citrulline uptake assays with the exception that 40 M [ 14 C]aspartic acid (PerkinElmer Life Sciences) was employed to initiate the uptake assays. In each case, plotted data represent the rate of amino acid uptake relative to the wild-type control.
Fluorescence Microscopy-Cell imaging of MKY1800 (aly1⌬ aly2⌬ DIP5-GFP) containing pRS315, pRS315-Aly1, pRS315-Aly1 ⌬PILKIN , pRS315-Aly2, pRS315-Aly1 5E , or pRS315-Aly1 5A was performed at the Center for Biological Imaging (Pittsburgh, PA) using an Olympus IZ81 inverted microscope (Olympus America Inc., Center Valley, PA) with a 100 ϫ 1.40 NA objective, and images were captured using a QImaging Retiga EXi CCD camera (QImaging, Surrey, British Columbia, Canada) and MetaMorph 7.5 Imaging System software (Molecular Devices, Sunnyvale, CA). Exposure times and microscope settings were kept constant during Dip5-GFP imaging. All Dip5-GFP images presented were processed equivalently using Adobe Photoshop software where intensity levels were adjusted and an unsharp mask filter applied (radius of 5 pixels). Pixel intensity for Dip5-GFP at the plasma membrane (PM) was measured by outlining PMs manually using a 3-pixel wide stroke with ImageJ software (National Institutes of Health) for a minimum of 140 cells per sample. Mean background fluorescence for each image was subtracted from the mean pixel intensity for each region of interest. The mean PM intensity (in arbitrary units) for each sample is plotted Ϯ S.E., and statistical significance of changes in PM intensity was assessed using oneway ANOVA and Tukey's post hoc test using Prism software (GraphPad Software Inc., San Diego). To assess the ratio of PM to vacuolar fluorescence, PM intensities were determined as defined above, and each cell was assigned a specific number identifier. Vacuolar regions of interest were assigned using the corresponding differential interference contrast images and then overlaid onto the fluorescent images. Only cells where clear vacuolar morphology was evident on the differential interference contrast image were used in subsequent analyses. A minimum of 40 cells was assessed this way per sample, and the mean PM/vacuolar fluorescence intensity is presented. Error bars represent Ϯ S.E. and statistical significance of changes in PM/vacuolar fluorescence intensity ratio assessed using oneway ANOVA and Tukey's post hoc test in Prism software.

C-terminal PXIXIT Motif in Aly1 Mediates Its Association
with Calcineurin-Yeast two-hybrid analysis successfully identified several calcineurin substrates (43,45,57,58). To identify additional potential substrates, we conducted a yeast two-hybrid screen using a Gal4-DBD fusion to Cna1, a catalytic subunit of calcineurin, as the bait. The plasmid that showed the strongest interaction with this bait encoded the C-terminal half (amino acids 493-915) of Aly1 fused in-frame to the Gal4-TAD (Fig. 1A). Further yeast two-hybrid analysis using the PJ69-4a reporter strain (47) revealed that full-length Aly1 interacts with yeast Cna1 (Fig. 1, A and B) and even with the catalytic subunit (CNA) of Homo sapiens calcineurin (Fig. 1B). Neither yeast Cna1 nor human CNA interacted with Aly2, the Aly1 paralog (39% identity; 64% similarity), or with any other yeast ␣-arrestin (Fig. 1B), even though every ␣-arrestin tested was expressed in the yeast two-hybrid strain (Fig. 1C).
The original library screen identified residues 493-915 of Aly1 as sufficient for robust interaction with Cna1 (Fig. 1A). To narrow down the region of Aly1 that interacts with calcineurin, various deletions and truncations were tested using the same yeast two-hybrid method. An even shorter segment of the C-terminal portion of Aly1, residues 610 -915, was sufficient for Cna1 binding. Further truncations established that Aly1 residues 822-915, but not 838 -915 or 610 -831, interacted with Cna1 ( Fig. 1A), suggesting that the residues responsible for the Aly1-Cna1 association are contained within amino acids 831-838. Indeed, residues 832-837 correspond to PILKIN, a sequence related to the consensus PXIXIT-docking motif found in many calcineurin substrates (43, 45, 59 -62). As a critical test of the role of this sequence element, a precise 6-residue deletion (Aly1 ⌬PILKIN ) was constructed and failed to interact with Cna1 (Fig. 1B), even though it was expressed (as a Gal4- For each treatment, numbers indicate the number of phosphorylated peptides identified over the total number of peptides identified for that region. Phospho means phosphorylated. Note: in addition, Ser-188, Ser-214, Ser-215, Ser-216, Ser-220, Ser-228, and Ser-568 were identified as phosphorylated in PhosphoPep. Two-tailed Z-tests were used to assess the significance of differences in phosphorylated peptide populations between Ca 2ϩ -and FK506-treated samples for both Aly1 and Aly1 ⌬PILKIN , and the p values from these tests are presented below. Tests results are significant if p Ͻ 0.05. If less than 30 total peptides were identified for a given region, the Z-tests were not determined (ND) as the sample size is prohibitively small for this analysis. If no phosphorylated peptides were identified in a particular comparison, the Z-test comparison is not applicable (NA). Green highlighting identifies the phospho-sites significantly enriched in Aly1 FK506-treated samples compared with Aly1 Ca 2ϩ -treated samples based on Z-test analyses (p values listed). It should be noted that there are approximately 5-fold fewer total peptides identified for the Ca 2ϩ -treated Aly1 ⌬PILKIN compared with the FK506-treated Aly1 ⌬PILKIN in these regions, which confounds comparison of enrichment in phosphorylation at these sites for Aly1 ⌬PILKIN . However, the enrichment for wild-type Aly1 between Ca 2ϩ and FK506 treatments is apparent, and these same sites are heavily modified in the FK506-treated Aly1 ⌬PILKIN samples as well.
The four serines and one threonine mutated to alanine to generate Aly1 5A or glutamic acid to generate Aly1 5E are underlined.
To confirm that the interaction detected by the yeast twohybrid method is physiologically relevant, we overexpressed either GST alone or fused to Aly1 or Aly1 ⌬PILKIN from the copper-inducible promoter CUP1 in yeast cells coexpressing GFPtagged Cna1 from its chromosomal locus. Reassuringly, Cna1-GFP strongly associated with Aly1 but did not associate with Aly1 ⌬PILKIN at a level any higher than seen with the negative control (GST alone) (Fig. 1D), demonstrating that Aly1 interacts with calcineurin in vivo and that this interaction requires the PILKIN motif near the C-terminal end of Aly1.
Aly1 Is a Substrate of Calcineurin-Next, we tested whether Aly1 was dephosphorylated by calcineurin in vitro. For many calcineurin substrates, loss of the PXIXIT-docking motif abrogates phosphatase binding and prevents dephosphorylation (43,45,59,60). To prepare phosphorylated versions of Aly1, Aly1 ⌬PILKIN (lacks the calcineurin-binding site), and Aly2 (does not bind calcineurin), we purified these proteins (as GST fusions) from yeast under low stringency conditions to retain associated protein kinases. Indeed, when these preparations were incubated with [␥-32 P]ATP, radioactivity was readily incorporated into all three proteins but not into the GST alone control purified in the same fashion (data not shown). After stringent washing to remove the kinases and unincorporated [␥-32 P]ATP, the radiolabeled proteins were then incubated with either purified calcineurin or -phosphatase ( Fig. 2A, t ϭ 0 lanes). Recombinant Crz1, a known calcineurin substrate (60,61), radiolabeled by phosphorylation with protein kinase Hrr25 (63), was used as a positive control. Upon incubation with calcineurin, both Aly1 and Crz1, but neither Aly1 ⌬PILKIN nor Aly2, exhibited the increased electrophoretic mobility and decreased 32 P label indicative of dephosphorylation (Fig. 2, A and B), whereas all four substrates were dephosphorylated by -phosphatase ( Fig. 2A). Thus, Aly1, but not Aly2, is a direct substrate for calcineurin in vitro, and the PILKIN sequence, which represents the PXIXIT motif in Aly1, is required for calcineurin to interact with and dephosphorylate Aly1.
Calcineurin Dephosphorylates a Specific Subset of Phosphosites in Aly1-We noted that calcineurin-mediated dephosphorylation of Crz1, a substrate with a high affinity for calcineurin, was more rapid and complete than dephosphorylation of Aly1 ( Fig. 2A; compare t ϭ 30 min for Aly1 and Crz1). Incubation of Crz1 with either calcineurin or -phosphatase resulted in a similar shift in electrophoretic mobility and equivalent loss of 32 P signal. In contrast, Aly1 was not dephosphorylated as extensively by calcineurin as it was by -phosphatase ( Fig. 2A), suggesting that only a subset of Aly1 phospho-sites are regulated by calcineurin.
To assess whether Aly1 is a calcineurin substrate in vivo, we examined the electrophoretic mobility of Aly1 or Aly1 ⌬PILKIN extracted from yeast treated with the following: (a) FK506, which inhibits calcineurin (64); (b) Ca 2ϩ , which stimulates calcineurin; or (c) a combination of FK506 followed by Ca 2ϩ , to control for any Ca 2ϩ -stimulated calcineurin-independent effects. Similar analyses were conducted with Aly1 PVIVIT , a mutant in which PILKIN was converted to a known PXIXIT variant that displays high affinity calcineurin binding (59,65), and with Aly1 AAAAAA , in which the PILKIN sequence was abrogated by six Ala substitution mutations. Upon calcineurin activation, only Aly1 and Aly1 PVIVIT migrated as a discrete doublet with the faster migrating band (below the 150-kDa marker) distinctly more prevalent than the slower migrating species (above the 150-kDa marker) (Fig. 2C). In contrast, Aly1 ⌬PILKIN and Aly1 AAAAAA exhibited a diffuse banding pattern even when calcineurin was activated (Fig. 2C, compare Ca 2ϩ -treated lanes). The same diffuse banding pattern was displayed by all versions of Aly1 when calcineurin was inhibited by treating cells with FK506 (Fig. 2C), consistent with a lack of dephosphorylation and a concomitant increase in phospho-Aly1 isoforms.
In agreement with this conclusion, a diffuse Aly1 migration pattern was also observed in cells where calcineurin activity was abolished by other means, specifically through loss of its regulatory subunit (in a cnb1⌬ mutant) or loss of both of its catalytic subunits (a cna1⌬ cna2⌬ double mutant) (Fig. 2D). Moreover, in contrast to the effect observed in wild-type cells, in the absence of functional calcineurin the addition of Ca 2ϩ failed to generate a defined Aly1 doublet with a prominent faster migrating species (Fig. 2D, compare Ca-treated lanes).
To confirm that the diffusely migrating species represent phosphorylated isoforms, purified GST-Aly1, Aly1 ⌬PILKIN , Aly1 PVIVIT , and Aly1 AAAAAA were treated with CIP. In all cases, treatment with CIP collapsed Aly1 species into a crisp doublet in which the fastest migrating band was markedly more prominent (Fig. 2E), demonstrating that the diffuse migration of the Aly1 doublet is due to phosphorylation. As observed in vitro ( Fig. 2A), in vivo activation of calcineurin did not sharpen the Aly1 doublet to the same extent as incubation with CIP (Fig. 2E, compare Ca 2ϩ -treated Aly1 or Aly1 PVIVIT Ϫ/ϩ CIP), further suggesting that calcineurin is responsible for dephosphorylation of only a subset of phospho-sites in Aly1.
Calcineurin-mediated Dephosphorylation Does Not Influence Aly1 Ubiquitinylation or Stability-Because Aly1 was identified as an in vitro substrate for the ubiquitin ligase Rsp5 (66), the GST-Aly1 species isolated from cell extracts was probed with an anti-GST antibody and an anti-ubiquitin antibody to allow simultaneous detection of ubiquitinylated and nonubiquitinylated forms on immunoblots. This analysis revealed that the slower migrating band in the Aly1 doublet is ubiquitinylated (band above 150-kDa marker; Aly1-Ub, Fig. 2, C-E). As noted above, both bands in the Aly1 doublet were sharpened upon addition of CIP (Fig. 2E), suggesting that both Aly1 and Aly1-Ub are phosphorylated. Importantly, there was no apparent defect in ubiquitinylation of Aly1 when the calcineurin-binding site was mutated or in response to altered calcineurin activity (Fig. 2E).
It has been reported that the phosphorylation status of yeast ␣-arrestins Bul1, Bul2, and Rod1 impacts their ubiquitinylation and/or interaction with Rsp5 (17,35) and that phosphorylation regulates the stability of the mammalian ␣-arrestin, TXNIP (32). We therefore tested whether dephosphorylation of Aly1 by calcineurin affected any of these properties. Dephosphory-lation of Aly1 by calcineurin did not alter Aly1 stability; in cells treated with Ca 2ϩ to activate calcineurin and cycloheximide (CHX) to inhibit protein synthesis, the degradation profiles of Aly1 and Aly1 ⌬PILKIN were indistinguishable (Fig. 3, A and B). Furthermore, neither ubiquitinylation of Aly1, nor its interaction with Rsp5 appeared to be regulated by calcineurin. Aly1 and Aly1 ⌬PILKIN were both ubiquitinylated (Fig. 2E), and equivalent amounts of Rsp5 copurified with Aly1 and Aly1 ⌬PILKIN from yeast extracts (Fig. 3C). As a control in these experiments, we included mutants (Aly1 Y686G or Aly2 Y703G ) in which puta-  A, GST-tagged Aly1, Aly1 ⌬PILKIN , Aly1, and Crz1 were purified on glutathione-Sepharose beads yeast extracts, incubated with [␥-32 P]ATP, and phosphorylated by copurifying kinases. Glutathione-bound proteins were washed to squelch further phosphorylation, incubated with CN-trunc or -phosphatase for the indicated times, and assessed by SDS-PAGE. Gels were stained for total protein or imaged to detect 32 P. Representative data from one of four replicates are shown. B, values plotted are the mean percent decrease in phospho-signal upon addition of CN-trunc (normalized for loading) for four replicate experiments performed as in A, and error bars represent the standard deviation of the means. C, BJ5459 cells expressing GST-Aly1, GST-Aly1 ⌬PILKIN (a deletion of the CN-binding site), GST-Aly1 PVIVIT (a high affinity CN-binding site, Aly1 PVIVIT ), or GST-Aly1 AAAAAA (mutation of the CN-binding site to alanines) were treated with nothing (Ϫ), 200 mM calcium chloride (Ca), 2 g/ml FK506 (calcineurin inhibitor; FK), or a combination of FK506 followed by Ca ( FK / Ca ). WCEs generated by TCA extraction were resolved by on 4% acrylamide gels and assessed by immunoblotting. The black line between lanes in C indicates where samples were run on two separate gels. D, WCEs from cells treated as in C were TCA-extracted from BJ5459 cells lacking either the regulatory subunit of CN (cnb1⌬) or the two catalytic subunits of CN (cna1⌬ cna2⌬). A white line indicates lane removal. E, treatment of GST-purified Aly1 and Aly1 mutants with calf intestinal alkaline phosphatase assists in identifying Aly1 phospho-species. BJ5459 cells expressing either GST-Aly1, GST-Aly1 ⌬PILKIN , GST-Aly1 PVIVIT , or GST-Aly1 AAAAAA (each expressed from pKK212-derived plasmids) were treated with nothing (Ϫ), Ca, FK, or both FK and Ca, and WCEs were made by glass bead lysis. GST-tagged arrestins were purified from these lysates and treated with calf intestinal alkaline phosphatase (ϩ CIP) or mock incubated in CIP buffer (Ϫ CIP) at 37°C for 30 min. Arrestins were then resolved on 4% acrylamide gels and analyzed by immunoblotting using a Li-COR Odyssey imaging station, allowing for detection of both the anti-ubiquitin (Ub) antibody (with anti-mouse IRDye 680) and the anti-GST antibody (with anti-rabbit IRDye 800) simultaneously. The white asterisks denote a proteolysis product generated during protein extraction.
tive Rsp5-binding sites in Aly1 or Aly2 (66) were changed from a canonical PPXY motif to PPXG. Introducing these mutations into Aly1, Aly1 ⌬PILKIN , or Aly2 reduced the association of each protein with Rsp5 to the background levels observed with the GST control (Fig. 3C). In addition, introducing this mutation into Aly1 resulted in loss of ubiquitinylation and an increased level of this protein (Fig. 3, C and D). However, in contrast to Aly1, Aly1 ⌬PILKIN , and Aly2, whose overexpression confers . CN regulation does not alter Aly1 stability or association with Rsp5. A, BJ5459 cells expressing either GST-Aly1 or GST-Aly1 ⌬PILKIN (from pKK212-derived plasmids) were treated with 200 mM calcium for 10 min (to stimulate CN activity and dephosphorylation of the WT Aly1 protein) and then incubated with 50 g/ml CHX. Culture samples were removed at the times indicated post-CHX addition; WCEs were prepared by TCA extraction, and Aly1 mobility and levels were assessed on 4% acrylamide gels followed by immunoblotting. Representative data from three replicate experiments are shown. B, Aly1 and Aly1 ⌬PILKIN protein levels (assessed as in A) were quantified, and the mean percentage of Aly1 or Aly1 ⌬PILKIN protein remaining (three replicates, normalized for loading) after CHX addition is plotted with error bars representing Ϯ S.D. C, GST, GST-Aly1, GST-Aly1 ⌬PILKIN , GST-Aly1 5A , GST-Aly1 5E , GST-Aly1 Y686G , GST-Aly1 ⌬PILKIN,Y686G , GST-Aly2, or GST-Aly2 Y703G (expressed from pKK212-derived plasmids) were extracted and purified using glutathione-Sepharose beads. Copurification of endogenous Rsp5 was assessed by immunoblotting. 4% of the WCE used as input for pulldowns is shown. A red asterisk denotes residual signal from the anti-Rsp5 antibody detection in the anti-GST pulldown blots. A horizontal white line denotes where the GST input blot has been cropped to conserve space. D, indicated GST fusions purified from cells treated with nothing (Ϫ), Ca, FK, or FK followed by Ca, resolved on 4% acrylamide gels, and immunoblotted for detection of ubiquitin (Ub) (with anti-mouse IRDye 680) and GST (anti-rabbit IRDye 800) simultaneously. The white asterisks denote a proteolysis product generated during protein extraction. For Aly1, both darker and lighter exposures are shown because the levels of Aly1 are significantly lower than those for the Y686G mutants. The darker exposure for wild-type Aly1 allows visualization of the calcineurin-dependent band shift and serves as a ubiquitinylated control for the Y686G mutant proteins. E, growth of serial dilutions of BY4741 cells containing pRS426 (vector), pRS426-Aly1, pRS426-Aly1 Y686G , pRS426-Aly1 ⌬PILKIN , pRS426-Aly1 ⌬PILKIN,Y686G , pRS426-Aly2, or pRS426-Aly1 Y703G on SC medium lacking uracil (Ura Ϫ control medium) or SC-Ura Ϫ with rapamycin. resistance to rapamycin, a TORC1 inhibitor that mimics nitrogen starvation (Fig. 3E) (24,67), overexpression of PPXG mutant Aly alleles failed to improve growth under these conditions, suggesting that these mutations abrogate the function of these proteins in vivo (Fig. 3E). Thus, ubiquitinylation of Aly1 resulted in an observed, slower migrating form of the protein, and neither the ubiquitinylation, stability, nor ability of Aly1 to bind Rsp5 was altered by its calcineurin-mediated dephosphorylation.
Mapping of Calcineurin-regulated Phospho-sites in Aly1-To determine which residues of Aly1 are phosphorylated and which are susceptible to dephosphorylation by calcineurin, we performed LC-MS 3 analysis of purified GST-tagged Aly1 and Aly1 ⌬PILKIN extracted from cells in which calcineurin was activated (by treatment with Ca 2ϩ ) or in which calcineurin was inhibited (by treatment with FK506). Peptides covering Ͼ50% of the protein were recovered (Fig. 4A), and a total of 22 phosphorylated residues in Aly1 were identified ( Fig. 4B and Table  1), several of which were also earmarked as phospho-sites in various yeast global phosphoproteomic analyses (Table 1) (68,69). Statistical analysis showed that of these 22 phospho-sites, two (Ser-252 and Ser-573) were significantly enriched in FK506-treated Aly1 (and also identified in FK506-treated Aly1 ⌬PILKIN ), making them good candidate sites for regulation by calcineurin (Fig. 4B and Table 1). Two other sites (Thr-250 and Ser-569) near Ser-252 and Ser-573, respectively, occasionally showed a modest enrichment, and a fifth site, Ser-568, near Ser-573, although not detected in our work, was reported to be an in vivo phospho-site in PhosphoPep (66). All these sites (except Ser-568) fit an -(S/T)P-consensus, and all lie in predicted solvent-exposed regions. Thus, based on these analyses, we mutated five candidate sites (Thr-250, Ser-252, Ser-568, Ser-569, and Ser-573) to Ala (generating mutant Aly 5A ). The Aly 5A mutant should mimic Aly1 that has been permanently dephosphorylated by calcineurin. We also generated Aly1 5E to mimic Aly1 persistently phosphorylated at these same sites.
Phenotypic Effects of Calcineurin Regulation of Aly1-We showed previously that Aly1 overexpression confers an increase in resistance to rapamycin (Fig. 3E) (24). Although Aly1 and all the Aly1 phospho-mutants tested elevated resistance to rapamycin (Fig. 4C), we noted that cells expressing the Aly1 variants capable of being dephosphorylated by calcineurin (Aly1) or that mimic the dephosphorylated state (Aly1 5A ) grew slightly better on rapamycin than cells expressing Aly1 variants that cannot be dephosphorylated by calcineurin (Aly1 ⌬PILKIN ) or that mimic the persistently phosphorylated state (Aly1 5E ) (Fig. 4C). This modest difference in phenotype was not due to differences in protein level as all of these GST fusions were expressed equivalently to wild-type Aly1 (Figs. 3C, 4D, and 7A) and behaved very similarly to wild-type Aly1 in other assays (Fig. 5, A and E, and data not shown).
In response to activation or inhibition of calcineurin activity, the electrophoretic migrations of Aly1 5A and Aly1 5E were indistinguishable from that of wild-type Aly1 (Fig. 4D), suggesting that additional calcineurin-regulated phospho-sites exist in Aly1 that were not pinpointed by our MS analyses (Fig. 4, A and  B). However, further phenotypic characterization of Aly1 5A and Aly1 5E indicated (see last section under "Results") that the sites altered in these mutants are functionally important (Figs.  4C and 6, B-G).
When nitrogen is limiting, Gap1 mediates entry of a wide range of amino acids, including proline (70). Uptake of two amino acid analogs, AzC, a toxic proline mimetic, and citrulline, an arginine mimetic taken up exclusively through Gap1, serve as readouts for Gap1 activity at the plasma membrane (55,70). We showed previously that overexpression of either Aly1 or Aly2 results in hypersensitivity to AzC and an increase in the rate of citrulline uptake (24). Increased Gap1 activity at the cell surface is due to increased Gap1 retrieval from endosomes (likely via an endosome-to-Golgi route); impeding Gap1 trafficking to the vacuole (a degradative organelle) increases Gap1 levels in the cell and leads, ultimately, to an increased level of Gap1 at the plasma membrane (24). Thus, we tested whether the phosphorylation status of Aly1 influences Gap1 trafficking. We found that overexpression of Aly1, Aly1 ⌬PILKIN , Aly1 5A , or Aly1 5E each conferred an equivalent increase in sensitivity to AzC (Fig. 5A) and Gap1 protein levels (Fig. 5C). In addition, overexpression of either Aly1 or Aly1 ⌬PILKIN caused a similar increase in the rate of citrulline uptake (Fig. 5D). To ensure that overexpression of these ALY1 alleles was not masking subtle phenotypic differences between them, we examined the ability of Aly1 phospho-mutants expressed from low copy plasmids to complement the aly1⌬ aly2⌬ AzC resistance phenotype, and we found that all alleles restored AzC sensitivity equally (Fig. 5B).
Our earlier studies demonstrated that Aly2, but not Aly1, requires a number of factors to promote Gap1 recycling, including the following: Npr1, a nitrogen-regulated protein kinase that promotes Gap1 trafficking to the plasma membrane (35)(36)(37); Lst4, a factor known to stimulate nutrient-dependent recycling of Gap1 (22,55), and AP-1, an adaptor complex that recruits clathrin to endosome-derived vesicles trafficking to the Golgi (24,71,72). Deletion of any of these factors cripples Aly2stimulated Gap1 recycling and may thereby create a sensitized assay to detect subtle defects in Aly1-mediated Gap1 trafficking that might arise from loss of calcineurin regulation (24). By every metric employed (rate of citrulline uptake, growth using citrulline as sole nitrogen source, and sensitivity to AzC), overexpression of Aly1, Aly1 ⌬PILKIN , Aly1 5A , and Aly1 5E caused similar increases in Gap1 activity at the plasma membrane irrespective of the genetic background used (npr1⌬, lst4⌬, apl2⌬) (Fig. 5, E and F, and data not shown). Together, these findings indicate that dephosphorylation of Aly1 by calcineurin does not regulate Aly1-mediated Gap1 recycling.
Dephosphorylation of Aly1 by Calcineurin Is Required for Aly1-mediated Endocytosis of Dip5-A previous study by another group demonstrated that Aly2 promotes endocytosis of Dip5 in response to excess amounts of the amino acids transported by this permease (aspartic acid or glutamic acid), but it did not address directly whether Aly1 may also play a role in this process (18). Therefore, we tested whether regulation of Aly1 by calcineurin might alter its ability to stimulate Dip5 traffick-ing to the vacuole. We found that the absence of either Aly1 or Aly2 caused a detectable increase in the steady-state level of Dip5, and an even more pronounced increase was observed in cells lacking both Aly1 and Aly2 (Fig. 6A), as was observed previously (18). This is consistent with the idea that each of these

N Arr C Arr Tail
T54 S157 S174 S182   (82,83)). The calcineurin docking-site (PXIXIT motif) is indicated in yellow. Positions of phospho-sites identified using MS analysis of Aly1 or Aly1 ⌬PILKIN are shown as lines below the protein with the single letter code and sequence position provided for the amino acid. Sites regulated by calcineurin are indicated in boldface red type. C, growth of serial dilutions of BY4741 cells containing pRS426 (vector), pRS426-Aly1, pRS426-Aly1 ⌬PILKIN , pRS426-Aly1 5A , or pRS426-Aly1 5E on SC medium lacking uracil (Ura Ϫ control medium) or SC-Ura Ϫ with rapamycin (Rapa). D, WCE from BJ5459 cells expressing GST-Aly1, GST-Aly1 5A or GST-Aly1 5E were treated with nothing (Ϫ), 200 mM calcium chloride (Ca), 2 g/ml FK506, or a combination of FK506 followed by Ca ( FK / Ca ) were generated by TCA extraction, resolved on 4% acrylamide gels, and assessed by immunoblotting. A red asterisk (within the blot) and line denoted with Ub (adjacent to blot) are used to help identify the ubiquitinylated species of Aly1.
␣-arrestins contributes to the endocytosis of Dip5. Therefore, as an independent means to assess the impact of Aly2 and Aly1 variants on Dip5 activity and trafficking, all subsequent analyses were performed in an aly1⌬ aly2⌬ background containing either a vector control or a low copy plasmid expressing the ALY allele of interest from its endogenous promoter. We measured Dip5 function at the cell surface by monitoring the rate of radiolabeled aspartic acid uptake and found that expression of Aly1, Aly1 5A , or Aly2 each significantly reduced the rate of aspartic acid uptake compared with the vector control (Fig. 6B), consistent with diminished Dip5 level and/or activity at the PM in these cells. By contrast, significantly higher Dip5 activity and/or levels were observed in cells expressing Aly1 ⌬PILKIN or Aly1 5E , which maintains Aly1 in a hyperphosphorylated or phospho-mimetic state, compared with cells expressing either Aly1 or Aly1 5A (Fig. 6B), their dephosphorylated counterparts. These data support a role for Aly1 and Aly2 in basal turnover of Dip5 as the aspartic acid uptake assays are done over a very short time course (2 min) and so likely reflect differences in the steady-state levels of Dip5 operating at the plasma membrane.
To determine whether the observed effects were due to alterations in the level of Dip5 at the cell surface, steady-state localization of Dip5-GFP was examined in cells expressing various ALY alleles. For these experiments, cells containing DIP5-GFP integrated at its chromosomal locus were grown under conditions equivalent to those used for the aspartic acid uptake assays. Images (Fig. 6C) were quantified, and reported differences in Dip5-GFP fluorescence intensity were assessed statistically, both as the pixel count at the PM (Fig. 6D) and as the ratio of the pixel count at the PM to that in the vacuole of the same cell ( Fig. 6E; see also "Experimental Procedures"). Despite some cell-to-cell variation in these cell populations, these analyses showed that expression of Aly2 (as a positive control), as well as Aly1 and Aly1 5A , significantly reduced the amount of Dip5 at the PM and increased the amount of Dip5 in the vacuole, compared with the aly1⌬ aly2⌬ controls cells carrying the vector alone (Fig. 6, C-E). In marked contrast, expression of Aly1 ⌬PILKIN or Aly1 5E in the aly1⌬ aly2⌬ cells neither reduced the amount of Dip5 at the PM nor increased the amount of Dip5 in the vacuole (Fig. 6, C-E). It should be noted, however, that due to background cytoplasmic fluorescence the PM-to-vacuolar ratios presented in Fig. 6E are lower than expected for the vector control (and likely for the Aly1 ⌬PILKIN and Aly1 5E alleles as well) based on the micrographs presented in Fig. 6C. These findings are consistent with the interpretation that phosphorylation blocks the ability of Aly1 to stimulate Dip5 internalization because Aly1 (which can be dephosphorylated by CN) and the Aly1 5A variant (which mimics its dephosphorylated state)  FIGURE 5. CN regulation is not required for Aly1-mediated Gap1 recycling. A, growth of serial dilutions of BY4741 cells containing pRS426 (vector), pRS426-Aly1, pRS426-Aly1 ⌬PILKIN , pRS426-Aly2, pRS426-Aly1 5A , or pRS426-Aly1 5E on MIN 0.5% (NH 4 ) 2 SO 4 Ϯ AzC. B, growth of serial dilutions of aly1⌬ aly2⌬ (D2-6A) cells containing pRS315 (vector), pRS315-Aly1, pRS315-Aly1 ⌬PILKIN , pRS315-Aly2, pRS315-Aly1 5A , or pRS315-Aly1 5E on MIN 0.5% (NH 4 ) 2 SO 4 Ϯ AzC. C, BY4743 cells containing pRS426 (vector, pRS426-Aly1 or pRS426-Aly1 ⌬PILKIN , pRS426-Aly1 5A , pRS426-Aly1 5E , or pRS426-Aly2) and pCKB230 (Gap1-GFP) were grown in MIN 0.5% (NH 4 ) 2 SO 4 . WCEs were resolved using SDS-PAGE, and Gap1 levels were assessed by immunoblotting. Gap1 levels relative to the vector control extract are presented below the immunoblot. D and F, prototrophic BY4741 (WT) or npr1⌬ cells with pCK283 and pRS426, pRS426-Aly1, pRS426-Aly1 ⌬PILKIN , or pRS426-Aly2 were assayed for [ 14 C]citrulline uptake. The mean uptake rate for seven replicates (in C) or three replicates (in E) is shown as a percentage relative to the WT containing vector, and error bars represent Ϯ S.D. E, growth of serial dilutions of npr1⌬ cells with pRS425, pRS425-Aly1, pRS425-Aly1 ⌬PILKIN , pRS425-Aly2, pRS425-Aly1 5A , or pRS425-Aly1 5E on MIN 0.5% (NH 4 ) 2 SO 4 Ϯ AzC.  both promoted basal endocytosis of Dip5 in aly1⌬ aly2⌬ cells, whereas the Aly1 ⌬PILKIN variant (which cannot bind CN) and the Aly1 5E phospho-mimetic mutant were unable to promote Dip5 endocytosis. Because these assays were done in the absence of substrate (no aspartic acid/glutamic acid ligand), the differences in Dip5 cell surface levels represent changes in the basal trafficking of Dip5 to the vacuole in the presence of these ALY alleles.

A D
We also carried out kinetic analysis of Dip5 stability in response to exogenous aspartic acid and glutamic acid, which induces endocytosis of the Dip5 permease. After addition of aspartic acid and glutamic acid, degradation of Dip5-GFP was appreciably faster in cells containing Aly2, Aly1, or Aly1 5A than in cells expressing Aly1 ⌬PILKIN , Aly1 5E , or the control vector (Fig. 6, F and G). Moreover, even at the zero time point, cells expressing Aly2, Aly1, and Aly1 5A contained less Dip5 than cells containing Aly1 ⌬PILKIN , Aly1 5E , or a control vector (Fig.  6F), consistent with faster turnover of Dip5 when Aly2 or dephosphorylated Aly1 are present. Together, these data demonstrate that the phospho-mimetic Aly1 5E protein and Aly1 ⌬PILKIN , which cannot be dephosphorylated by calcineurin, fail to internalize and degrade Dip5 as effectively as wild-type Aly1 or the dephosphorylation-mimetic Aly1 5A . Thus, we conclude that dephosphorylation of Aly1 by calcineurin is required to promote Aly1-mediated trafficking of Dip5 to the vacuole.
Our data for Aly1 are consistent with recent findings that ␣-arrestins Rod1 and Bul1 must be dephosphorylated to promote nutrient-induced endocytosis of the Jen1 and Gap1 permeases, respectively. Dephosphorylation of Rod1 and Bul1 by unidentified phosphatases results in loss of binding to the yeast 14-3-3 proteins (Bmh1 and Bmh2), suggesting that it is their association with 14-3-3 proteins that may be responsible for inhibiting their endocytic function (17,35). Because previous MS analyses revealed enrichment of Bmh1 and Bmh2 peptides in purified preparations of Aly1 and Aly2, 5 we examined whether calcineurin-mediated dephosphorylation of Aly1 alters its interaction with Bmh1 and Bmh2. We found that Bmh1 and Bmh2 copurified from yeast extracts with GST-Aly1 and GST-Aly2 but not with the GST control (Fig. 7), confirming our previous MS analyses. Although significantly higher levels of Bmh1 and Bmh2 copurified with Aly2 than with Aly1, the amount of Bmh1 and Bmh2 associated with Aly1, Aly1 ⌬PILKIN , FIGURE 6. Calcineurin-mediated dephosphorylation of Aly1 is required for Aly1-dependent trafficking of Dip5 to the vacuole. A, WCEs from BY4741, aly1⌬, aly2⌬, or aly1⌬ aly2⌬ cells containing a chromosomally integrated Dip5-GFP grown in MIN ϩ0.5% (NH 4 ) 2 SO 4 were resolved by SDS-PAGE, and Dip5 levels were assessed by immunoblotting. Dip5 levels were quantified and are presented relative to the wild-type extract below the immunoblot. B, cells lacking both ALY1 and ALY2 containing pRS315 (vector, pRS315-Aly1, pRS315-Aly1 ⌬PILKIN , pRS315-Aly1 5A , pRS315-Aly1 5E , or pRS315-Aly2) or cells lacking DIP5 were assayed for [ 14 C]aspartic acid uptake. The mean uptake rate for a minimum of four replicates is shown as a percentage relative to aly1⌬ aly2⌬ with vector and error bars represent Ϯ S.D. Unpaired Student's t tests were performed to assess the significance of these data; *** indicates a p value Ͻ0.0001. It should be noted that Aly1, Aly1 5A , and Aly2 were all significantly less than the vector control (p value Ͻ0.001) but were not significantly different from one another (p value Ͼ0.01). Similarly, Aly1 ⌬PILKIN and Aly1 5E were not significantly different from one another (p value Ͼ0.01). C, Dip5-GFP in aly1⌬ aly2⌬ cells containing pRS315, -Aly2, -Aly1, -Aly1 ⌬PILKIN , -Aly1 5A , or -Aly1 5E was visualized by fluorescence microscopy, and two panels of representative cell images are presented for each. Aly2, Aly1, and the dephosphorylation mimetic (Aly1 5A ) restore endocytosis of Dip5, as evidenced by vacuolar fluorescence in Ͼ46% of cells, although cells containing vector or mutants of that mimic the phosphorylated form of Aly1 have predominantly PM-localized Dip5 (Ͻ27% of cells displayed dim vacuolar fluorescence). D, plasma membrane fluorescence intensities for Dip5-GFP (as shown in C) were normalized to background and quantified for Ͼ140 cells per strain (see "Experimental Procedures"). The mean plasma membrane intensity is plotted Ϯ S.E. (in arbitrary units). A one-way ANOVA with Tukey's post hoc comparison was used to assess the statistical significance of fluorescence differences compared with the vector control (***, p value Ͻ0.0001; ns, not significant p value Ͼ0.01). E, ratio of plasma membrane fluorescence to vacuolar fluorescence for Dip5-GFP (as shown in C) was measured for a minimum of 40 cells per condition (see "Experimental Procedures"). A one-way ANOVA with Tukey's post hoc comparison was used to assess the statistical significance for each of these ratios compared with the pRS315 vector control (***, p value Ͻ0.0001; ** p value Ͻ0.001; *, p value Ͻ0.01, and ns ϭ not significant p value Ͼ0.01). F, aly1⌬ aly2⌬ cells containing pRS315, -Aly2, -Aly1, -Aly1 ⌬PILKIN , -Aly1 5A , or -Aly1 5E and a chromosomally integrated Dip5-GFP were grown in MIN ϩ0.5% (NH 4 ) 2 SO 4 medium, and 200 g/ml aspartic acid and glutamic acid were added to trigger internalization and degradation of Dip5. Cells were harvested at the times indicated post-Asp/Glu addition; WCEs were prepared using TCA extraction and analyzed by SDS-PAGE and immunoblotting. Data representing one of the three replicate experiments performed are presented with the quantification for the percentage of Dip5 remaining in each lane relative to the t ϭ 0 point and normalized for alterations in the G6PD loading control is indicated. G, Dip5-GFP band intensities from three replicate experiments (representative panel in F) were measured, normalized to the G6PD loading control, and the mean percentage of Dip5 remaining post Asp/Glu addition Ϯ S.E. is plotted. DIC, differential interference contrast. Aly1 5A , or Aly1 5E was not detectably different (Fig. 7). These findings suggest that phosphorylation of Aly1 at the CN-sensitive sites delineated in this study is not the major determinant of the association with Bmh1 and Bmh2. Indeed, there are two sites in Aly1 that fit the predicted 14-3-3-binding sites consensus, each of which lies in peptides not recovered in our MS analyses and neither of which map to any calcineurin-regulated phospho-site identified here (Fig. 4A). Thus, although dephosphorylation of other ␣-arrestins may serve as a regulatory mechanism to release these adaptors from the grasp of 14-3-3 proteins, and thereby promote ␣-arrestin-mediated endocytosis of permeases, that does not appear to be the underlying mechanism by which CN-mediated dephosphorylation stimulates the trafficking function of Aly1.

DISCUSSION
Phospho-regulation of Aly1-mediated Trafficking-We show here that ␣-arrestin Aly1 interacts with calcineurin through a C-terminal docking motif and identified up to five possible CNtargeted phospho-sites in Aly1. Furthermore, by monitoring the amino acid permease Dip5 as its cargo, we demonstrated that Aly1 must be dephosphorylated by calcineurin to mediate Dip5 trafficking to the vacuole (Fig. 8). This study is the first to identify a phosphatase that directly regulates ␣-arrestin function. Although phospho-sites have been identified for various ␣-arrestins in global proteomic screens (68,69), aside from our work on Aly1, a comprehensive phospho-map exists for only one other ␣-arrestin, namely Ldb19/Art1 (34). Given that only a handful of the 22 phospho-residues detected in Aly1 appear to be regulated by calcineurin (and there appear to be additional calcineurin-regulated sites in Aly1 not identified in our study), Aly1 action may be under the control of multiple protein kinases and phosphatases. Indeed, other ␣-arrestins are highly modified, many with Ͼ10 phospho-sites identified to date (68,69), underscoring the potential complexity of their regulation.
Both Aly1 and the related Aly2 promote retrieval of the general amino acid permease Gap1 from endosomes, and Npr1-dependent phosphorylation of Aly2 is needed for its function in this Gap1 recycling (24). We show here that dephosphorylation of Aly1 by calcineurin is not required for such Gap1 recycling, suggesting that different subsets of phospho-sites may regulate distinct ␣-arrestin trafficking functions. It was demonstrated that Aly2 is necessary for efficient endocytosis of the Dip5 transporter (18), but whether Aly1 might function semi-redundantly in this process was never directly tested. We show here that Aly1, when dephosphorylated by calcineurin, does stimulate Dip5 internalization and turnover, providing evidence that Aly1, like other ␣-arrestins, mediates the endocytosis of nutrient transporters.
Thus, our results indicate that Aly1 action is under control of a phosphorylation-dephosphorylation switch; calcineurin-mediated dephosphorylation stimulates Aly1 function in trafficking at least one membrane permease to the vacuole (Fig. 8), and phosphorylation, by an as yet unidentified protein kinase, inhibits it. Because excess aspartate or glutamate stimulates Dip5 internalization, and Aly1 can only mediate Dip5 downregulation when dephosphorylated by calcineurin, influx of these acidic amino acids may stimulate calcineurin-dependent signaling through unknown mechanisms. In turn, dephosphorylated Aly1 promotes Dip5 endocytosis to the vacuole (Fig. 8), perhaps to prevent the cytotoxicity that an excess of these amino acids can cause (73). Indeed, in mammalian cells, amino acid overload increases the level of Ca 2ϩ -calmodulin (74), a requirement for calcineurin activation.
Arrestin Dephosphorylation Is a Conserved Mechanism That Promotes Arrestin-mediated Endocytosis-Our data and other recent findings suggest a general model for phospho-regulation of ␣-arrestins (Fig. 8) (75). As shown here for Aly1/Art6, phosphorylation of Rod1/Art4 (17), Ldb19/Art1 (34), and two newly recognized ␣-arrestins Bul1 and Bul2 (35) blocks their ability to mediate endocytosis of specific nutrient permeases. In cells grown on lactate, the protein kinase Snf1 phosphorylates Rod1 to retain the lactic acid permease Jen1 at the plasma membrane; however, upon glucose addition, which inactivates Snf1, endocytosis of Jen1 ensues (17). Rod1 dephosphorylation appears to be PP1-dependent, but direct dephosphorylation of Rod1 by PP1 has not been demonstrated. Moreover, because PP1 action (Glc7-Reg1 complex) is necessary for inactivation of Snf1, the lack of apparent Rod1 dephosphorylation in reg1⌬ cells may arise indirectly from lack of Snf1 down-regulation. For Ldb19, phosphorylation by the protein kinase Npr1 impairs Ldb19mediated endocytosis of the arginine permease Can1, whereas the dephosphorylated protein associates with the plasma membrane and promotes Can1 endocytosis (34). However, the phosphatase responsible for Ldb19 dephosphorylation is not known. For Bul1 and Bul2, phosphorylation by Npr1, which is activated on poor nitrogen sources (e.g. proline), inhibits Bul-mediated endocytosis of Gap1. When NH 4 ϩ is the nitrogen source, Npr1 is inactive, and the Bul proteins are dephosphorylated in a man- Dip5 ? FIGURE 8. Model for phospho-regulation of ␣-arrestin-mediated trafficking. Unidentified kinases (purple rectangle) maintain Aly1 (red hexagon) in its fully phosphorylated state (phosphosites represented as gray and green circles), which is unable to stimulate internalization of the aspartic acid/glutamic acid transporter Dip5. When calcineurin (CN; blue oval) is activated by calcium, as may occur in response to excess aspartic acid/glutamic acid (indicated gray dashed line), it dephosphorylates Aly1, and this dephosphorylation at specific sites (removal of CN-regulated phosphosites; loss of green circles) is required for optimal Aly1-mediated endocytosis of Dip5. Dephosphorylated Aly1 promotes endocytosis and/or trafficking of Dip5 to the vacuole (indicated by dashed black arrow and trafficking route denoted by red line).
ner that depends on the phosphatase Sit4, promoting Gap1 internalization (35). However, direct dephosphorylation of Bul1 or Bul2 by Sit4 has not been demonstrated.
Thus, in the case of Aly1 and three other yeast ␣-arrestins, phosphorylation blocks and dephosphorylation promotes their endocytic function. Indeed, this phospho-inhibition of ␣-arrestin-mediated endocytosis is conserved in mammals. For example, phosphorylation of ␣-arrestin TNXIP by AMPK results in reduced stability of this ␣-arrestin, diminishing endocytic turnover of the GLUT1 glucose transporter regulated by TXNIP (32). However, the evidence to date indicates that phosphorylation can modulate ␣-arrestin function through distinct mechanisms. First, in the case of yeast Rod1, dephosphorylation of the ␣-arrestin promotes its endocytic function and is required for its ubiquitinylation (17). In contrast, dephosphorylation does not seem to regulate the ubiquitinylation of either Aly1, as we have shown here, or Bul1 (35). Similarly, under conditions that do not promote Ldb19-mediated internalization of Can1, this ␣-arrestin is ubiquitinylated, further supporting the idea that ␣-arrestin ubiquitinylation and its endocytic function need not be coupled (20,34). Second, phosphorylation reportedly regulates the interactions of Rod1, Bul1, and Bul2 with cytoplasmic 14-3-3 proteins, resulting in sequestration of the ␣-arrestin (or the ␣-arrestin⅐Rsp5 complex) in the cytosol, and thereby inhibiting ␣-arrestin activity at and/or recruitment to the plasma membrane (17,32,35). Other ␣-arrestins are among the 271 proteins demonstrated to associate with Bmh1 and Bmh2 in a phosphorylation-dependent manner (76); therefore, interaction with 14-3-3 proteins may broadly regulate ␣-arrestin function. However, neither of these mechanisms is adequate to explain phospho-regulation of Aly1, because we demonstrated here that the calcineurin-dependent dephosphorylation of Aly1 does not alter its ubiquitinylation, stability, or interaction with either Rsp5 or 14-3-3 proteins. Dephosphorylation of mammalian ␤-arrestins is required for their association with clathrin and AP-2 and subsequent stimulation of endocytosis (6,10). Both Aly1 and Aly2 have been shown to interact with clathrin adaptors (24); therefore, phosphorylation might similarly regulate the endocytic function of Aly1 by altering its interaction with these components of the trafficking machinery.
Conserved Interaction of Calcineurin with ␣-Arrestins and Its Implications-A Caenorhabditis elegans ␣-arrestin, CNP-1/ ArrD-17, interacts with the catalytic subunit of calcineurin and is a calcineurin substrate in vitro (57). Although no calcineurindocking site was identified in the nematode protein, we note that CNP-1/ArrD-17 contains near its C terminus a variant PXIXIT motif (PIVIGS) that is conserved in its mammalian ␣-arrestin relatives ARRDC2-4 and TNXIP (as PLVIGS or PLVIGT). Interestingly, calcineurin regulation of ArrD-17 in C. elegans also impacts response to starvation, although a link between specific ␣-arrestins and trafficking of any nutrient transporter has not yet been established.
To our knowledge, our work is the first demonstration that calcineurin dephosphorylates an ␣-arrestin to regulate the trafficking function of this class of adaptor proteins. However, calcineurin action influences membrane protein trafficking in other ways. For example, heat stress-induced calcineurin-me-diated dephosphorylation of the TORC2-and eisosome-associated components Slm1 and Slm2 stimulates internalization of the uracil permease Fur4 (45,62,77). Studies of synaptic vesicle retrieval in mammalian cells indicate that calcineurin dephosphorylates a suite of endocytic regulators, including amphiphysins, synaptojanin, epsins, and dynamin (78,79). Calcineurin activation also stimulates receptor-mediated endocytosis of AMPA and transferrin (80,81). Collectively, and combined with our demonstration here that calcineurin regulates ␣-arrestin-mediated trafficking to the vacuole in yeast, it is clear that calcineurin has an important physiological function as a global regulator of membrane protein trafficking.