RESEARCH ARTICLE 311 Protein phosphatase 2A binds to the cytoplasmic tail of D and regulates post-trans-Golgi network trafficking

Oleg Varlamov, Elena Kalinina, Fa-Yun Che and Lloyd D. Fricker* Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY 10461, USA *Author for correspondence (e-mail: [email protected])

Accepted 20 November 2000 Journal of Cell Science 114, 311-322 © The Company of Biologists Ltd

SUMMARY

Carboxypeptidase D (CPD) is a transmembrane protein surface of AtT-20 cells is decreased 45% by okadaic acid, that processes proteins in the trans-Golgi network (TGN). a PP2A inhibitor. Microinjection of the CPD tail into AtT- A 20-residue region within the cytoplasmic tail of CPD 20 cells inhibits the transition of CPD from endosomal binds protein phosphatase 2A (PP2A). PP2A also binds to compartments to the TGN. However, okadaic acid does not the cytoplasmic tails of other secretory pathway proteins: affect the rate of budding of CPD from the TGN into peptidylglycine-α-amidating mono-oxygenase, the cation- nascent vesicles or the rate of uptake from the cell surface independent mannose-6-phosphate receptor and TGN38. into endosomal compartments. These results are consistent The CPD tail is phosphorylated on Thr residues in the AtT- with the model that PP2A is involved in the trafficking of 20 cell line. The CPD tail can also be phosphorylated by proteins between a TGN recycling loop and a cell-surface purified protein kinase A, protein kinase C and casein recycling loop, but is not involved in the individual kinase II. Both the in vitro and the in vivo phosphorylated recycling loops. CPD tail can be dephosphorylated by purified PP2A. The binding of CPD tail peptide to PP2A does not influence phosphatase activity. Key words: Peptide processing, Furin, Protein trafficking, Protein The rate of transport of CPD from the TGN to the cell phosphatase 2A

INTRODUCTION slower than that of a soluble form of CPD lacking the cytoplasmic and transmembrane domains (Varlamov et al., Many bioactive peptides and proteins are produced from larger 1999b), indicating that these domains are important for the precursors by proteolytic processing at sites containing basic retention of full-length CPD in the TGN. The tail contains amino acids. A number of that cleave proteins several motifs including a ‘FHRL’ sequence which resembles at basic amino acids have been described, including furin the ‘YxxL’ motif involved in both endocytosis and TGN and prohormone convertases (Zhou et al., 1999; Seidah and localization of CPD (Eng et al., 1999). Mutations of consensus Chretien, 1997). Following the action of endopeptidases, the sites for casein kinase II (CKII) within the CPD tail processing intermediate contains basic amino acids attached significantly reduce the half-life of CPD and alter its TGN to the C terminus. These residues are then removed by a retention (Eng et al., 1999). In addition to CKII sites, the carboxypeptidase present in the secretory pathway. Two cytoplasmic tail of CPD contains potential phosphorylation have been identified in the secretory sites for protein kinase A (PKA) and protein kinase C (PKC) pathway of mammalian cells; (CPE) is (Kuroki et al., 1995; Xin et al., 1997; Tan et al., 1997), raising primarily active in the late secretory pathway of the possibility that intracellular trafficking of CPD may be neuroendocrine cells, whereas carboxypeptidase D (CPD) is controlled by phosphorylation. enriched in the trans Golgi network and is present in many cell Studies on furin trafficking have demonstrated the existence types (Song and Fricker, 1996; Xin et al., 1997; Varlamov and of two recycling loops; one between the TGN and immature Fricker, 1998; Dong et al., 1999). Like furin, CPD is a type I vesicles and another between the cell surface and endosomes membrane protein that cycles between the cell surface and the (Molloy et al., 1999). The localization of furin to these loops TGN (Kuroki et al., 1995; Xin et al., 1997; Tan et al., 1997; requires casein kinase II (CKII)-mediated phosphorylation and Varlamov and Fricker, 1998). PACS-1 binding (Wan et al., 1998; Molloy et al., 1998; Molloy The cytoplasmic tail is important for the TGN localization et al., 1999). The transition of furin from the cell-surface and endocytosis of CPD from the cell surface; truncation of the recycling loop to the TGN recycling loop has been proposed to C-terminal 45 residues of CPD causes the protein to remain on be regulated by protein phosphatase 2A (PP2A) (Molloy et al., or near the cell surface (Eng et al., 1999). The rate of CPD 1998; Molloy et al., 1999). Several trimeric forms of PP2A have movement from the TGN into nascent vesicles is 2-3 times been characterized; all of them contain a 36 kDa catalytic C 312 JOURNAL OF CELL SCIENCE 114 (2) subunit, a 65 kDa A subunit and one of several B subunits (Wera Typically, two to four confluent 100 mm plates of AtT-20 cells were and Hemmings, 1995). Although the role of PP2A in metabolically labeled with 35S-Met for 2-4 hours, sonicated in the lysis dephosphorylation of many cytoplasmic and nuclear proteins is buffer (1% NP-40, 50 mM Tris-HCl, 0.15 M NaCl, 2 mM MgCl2, pH evident (Wera and Hemmings, 1995; Price and Mumby, 1999), 7.2) and then centrifuged for 30 minutes to remove insoluble material. the mechanism of PP2A targeting to specific intracellular To scale up PP2A preparation, 10-15 mouse brains were homogenized locations is poorly understood. It has been reported that a pool in 30 ml of 25 mM Hepes, 25 mM KCl, 2.5 mM MgAc, pH 7.2 and centrifuged for 30 minutes at 14,000 g. The supernatant was of PP2A binds microtubules and neurofilaments controlling recentrifuged for 1 hour at 100,000 g to yield the cytoplasmic fraction their stability (Price and Mumby, 1999; Sontag et al., 1999). A containing the soluble pool of PP2A. The cytoplasm was dialyzed trimeric complex containing the B subunit of PP2A has been against the homogenization buffer, aliquoted and snap-frozen on dry localized to interphase Golgi membranes; Golgi localization is ice. The typical protein concentration of the cytoplasm was 4-8 mg/ml. lost during the early stages of mitosis indicating a cell cycle- In the large-scale GST pull-down experiments 3-6 ml of the mouse dependent regulation of PP2A localization (Lowe et al., 2000). brain cytosol containing 1% NP-40 was mixed with 0.4 mg of the PP2A may control different aspects of organelle trafficking, GST-CPD tail protein and incubated first at 4oC for 3 hours, then 30 including sorting in endosomes (Molloy et al., 1999), minutes at 37oC with gentle agitation. Then, 0.1-0.2 ml of glutathione- melanosome aggregation (Reilein et al., 1998), TGN Sepharose (Sigma) were added and the samples mixed for an fragmentation (Horn and Banting, 1994), protein secretion additional 40 minutes at room temperature. The beads were recovered by centrifugation, washed six times with the homogenization buffer (Krautheim et al., 1999) and endocytosis (Slepnev et al., 1998). containing 1% NP-40, and then either boiled in the SDS gel-loading This report describes the finding that PP2A is associated buffer (for analysis by western blot and mass spectrometry), or eluted with the cytoplasmic tails of CPD and other integral membrane in 5 mM glutathione (for dephosphorylation studies). proteins, providing a possible link between PP2A and The GST-tail bound material was resolved on denaturing 10% intracellular organelles. Furthermore, PP2A efficiently polyacrylamide gels, stained with Coomassie Blue, and then specific dephosphorylates the CPD cytoplasmic tail. Cell culture data bands were excised from the gel, extensively washed with NH4HCO3 suggest that PP2A controls post-TGN steps of CPD trafficking. (pH 8.0) and water, and incubated overnight at 37oC with trypsin (50 ng per µg of protein). After incubation, the samples were centrifuged, the supernatants were collected and the gel was extracted with 60% acetonitrile/0.1% trifluoroacetic acid. The resulting supernatant was MATERIALS AND METHODS combined with the first supernatant, the samples were dried under vacuum and then analyzed by matrix-assisted laser desorption Plasmid construction, protein expression and ionization (MALDI) mass spectrometry. recombinant protein production In small-scale GST pull-down experiments, 50 µg of the GST- Polymerase chain reaction (PCR) fragments encoding the 58-amino fusion protein was incubated with 100 µl of the mouse brain cytosol acid cytoplasmic domain of rat CPD and various CPD tail deletions for 40 minutes at 37oC in 25 mM Hepes, pH 7.2, 25 mM KCl, 2.5 were ligated into the BamHI/EcoRI sites of the pGEX2T vector mM MgAc. In competition studies, approximately 20 µg of the CPD (Pharmacia). Deletions shorter than 55 nucleotides were generated tail peptide was added to the incubation mixture. Following the using synthetic oligonucleotides that were annealed and then incubation, 40 µl of the glutathione-Sepharose was added and subcloned into BamHI/EcoRI sites of pGEX2T. Glutathione S- incubated for 40 minutes at room temperature. The beads were (GST)-fusion proteins containing the cytosolic tail of recovered by centrifugation, washed five times with the incubation TGN38 and peptidyl-glycine α-amidating monooxygenase (PAM, buffer containing 0.1% NP-40, boiled in SDS gel-loading buffer and residues 891-961) were obtained from Dr Sharon Milgram (University analyzed by western blot using the antisera to the C, A or B / subunits of North Carolina); the cation-independent mannose 6-phosphate of PP2A (Santa Cruz Biotechnology) at 1:500 dilution. receptor (M6PR) from Dr Suzanne Pfeffer (Stanford University); and furin from Dr Gary Thomas (Vollum Institute, Oregon Health Co-immunoprecipitation studies, membrane preparation Sciences University). The fusion proteins were expressed in E. coli, and immunofluorescence analysis purified according to the manufacturer’s instruction (Pharmacia) and To isolate PP2A associated with the cytoplasmic tails of the dialyzed against 50 mM Tris-HCl, pH 7.4 containing 150 mM NaCl. transmembrane proteins, 5 mouse brains were homogenized in 10 ml Human albumin C-terminally fused with the transmembrane and of 50 mM Tris-HCl, pH 7.4 containing 1% NP-40, 0.15 M NaCl, 2 o the cytoplasmic domains of duck CPD was constructed using the mM MgCl2 and 1 mM phenylmethylsulfonyl fluoride, mixed at 4 C pcDNA3 plasmid containing the coding region of human albumin for 60 minutes, and then centrifuged in the microcentrifuge for 15 (Alb/pcDNA3); this plasmid contains a Bsu36I site near the C minutes at 4oC. 1 ml of the supernatant was combined with 20-50 µl terminus of the coding region of albumin and an ApaI site in the of the polyclonal antisera to either the CPD tail (Song and Fricker, 3′ non-coding region (Mitra et al., 1994). Two complementary 1996), TGN38 (gift of Sharon Milgram), furin (gift of Ruth Angeletti) oligonucleotides were ligated and subcloned into the Bsu36I/ApaI or control rabbit IgG, and then incubated overnight at 4oC with sites of Alb/pcDNA3. This linker contains internal XhoI and AflII sites mixing. Following the incubation, the samples were combined with and encodes a cleavage site for thrombin. The transmembrane domain 50-100 µl of protein A Sepharose (Sigma) and incubated for 4 hours. of CPD was generated by PCR. The PCR product was digested with The beads were washed six times with homogenization buffer, boiled XhoI/AflII and subcloned into the XhoI/AflII sites of the Alb/pcDNA3 in the SDS gel-loading buffer and analyzed by western blotting using plasmid containing the linker described above. The PCR product the antiserum to the catalytic C subunit of PP2A. encoding the cytoplasmic tail of CPD was subcloned into the The dual immunofluorescent analysis of AtT-20 cells expressing AflII/ApaI sites of Alb/pcDNA3 containing the linker and the albumin fused with the transmembrane and the cytoplasmic domains transmembrane domain of CPD. The constructs were transfected into of CPD was performed as described (Varlamov and Fricker, 1998) AtT-20 cells using the Lipofectin transfection kit (GIBCO BRL). using the antisera to human albumin and syntaxin 6 (gift of Richard Stable cell lines were selected using 1 mg/ml Geneticin (G418). Scheller) (dilution 1:1000). In some experiments, cells were incubated with 5 µg/ml brefeldin A (BFA) for 2 to 30 minutes and then subjected Purification of protein phosphatase 2A to immunofluorescence with antisera to albumin or Cab45 (a gift from Either AtT-20 cells or mouse brains were used as a source of PP2A. Philipp Scherer) along with syntaxin 6. PP2A binds to carboxypeptidase D 313

Total microsomes were prepared as described (Varlamov et al., buffer pH 1.9, and then subjected to two-dimensional electrophoresis 1999a), except that the high-speed centrifugation step was repeated (Boyle et al., 1991). two more times to wash the membranes prior to the immunoisolation procedure. Trafficking assays To measure transport from the TGN to the plasma membrane, 60 In vitro phosphorylation and dephosphorylation studies mm plates of AtT-20 cells expressing the albumin-CPD tail fusion GST-CPD tail (100 µg) was subjected to in vitro 32P phosphorylation protein were metabolically labeled for 20 minutes with 35S-Met, using PKC (mixture of isoforms, Panvera), PKA (Sigma) or CKII washed with Dulbecco’s modified Eagle’s medium, and incubated (Sigma), according to the manufacturer’s instructions. The reactions for 90 minutes at 18.5oC to allow the labeled proteins to accumulate were dialyzed against dephosphorylation buffer (25 mM Tris-HCl, pH in the TGN. Then, the cells were placed on ice, washed 5 times with 32 o 7, 0.2 mM MnCl2 and 1 mM dithiothreitol). The P-labeled GST- cold PBS, and incubated for 30 minutes at 37 C in PBS containing CPD tail peptide (2-5 µg) was combined with PP2A, incubated for 2 1 mg/ml EZ-Link Sulfo-NHS-SS-Biotin (Pierce). In some hours at 37oC, boiled in the SDS gel-loading buffer, and analyzed on experiments, the biotinylation buffer was supplemented with 200 a denaturing 12% polyacrylamide gel. Either brain PP2A purified nM okadaic acid. The biotinylation reaction was stopped by washing on the GST-CPD tail affinity column, muscle PP2A (Upstate with ice-cold 0.2% bovine serum albumin in PBS. The cells were Biotechnology), or purified PP2A (Aα Bα C; a gift of Dr Marc lysed and subjected to immunoprecipitation (Milgram and Mumby) were used. The PP2A concentration was measured by a Mains, 1994) using antibodies to human albumin. To elute the quantitative western blot. immunoprecipitated material, 50 µl of the protein A beads were To analyze the sites of phosphorylation in vitro, the GST-CPD tail treated three times with 30 µl 3 M KSCN/0.5% NP-40. The was phosphorylated by different kinases, as described above, except supernatants were pooled, diluted ten times with 10 mM Tris-HCl using unlabeled γ-ATP. To separate GST from the CPD tail peptides, pH 8, 0.15 M NaCl, 1 mM EDTA, 0.1% NP-40 at 4oC, and then the phosphorylated GST fusion proteins were first immobilized on the incubated for 4 hours at 4oC with 50 µl streptavidin-agarose glutathione-Sepharose and then digested with thrombin (Pharmacia) (Sigma). The streptavidin-bound albumin represents the cell surface- according to the manufacturer’s instructions. Nonbound material derived material that typically constitutes 5-7% of the total pool of containing the CPD tail peptides were subjected to proteolytic albumin-CPD tail in AtT-20 cells. Following the incubation, the digestion using trypsin (Promega) or Staphylococcus aureus V8 streptavidin beads were first washed four times with 50 mM Tris- (Promega), according to the manufacturer’s instructions. HCl, pH 8, 0.5 M NaCl, 1 mM EDTA, 1% NP-40, 0.3 M KSCN, After digestion, the peptide mixture was desalted using the ZipTip C18 twice with the same buffer lacking KSCN, and then boiled in SDS reverse phase cartridges (Millipore), and analyzed by MALDI or gel-loading buffer and analyzed on a denaturing polyacrylamide gel. by liquid chromatography electrospray ionization collision-induced The rate of endocytosis was measured as described (Volz et al., dissociation mass spectrometry. 1995). Briefly, AtT-20 cells expressing albumin-CPD tail were labeled with 35S-Met for 4 hours, surface labeled with NHS-SS-biotin (Pierce) Labeling of AtT-20 cells with 32P and analysis of protein at 4oC, and then recultured at 37oC. At different time points, the biotin phosphorylation in vivo label was removed by treating cells with glutathione at 4oC. One 60 mm plate of cells expressing albumin fusion proteins was Control dishes were frozen immediately after biotinylation without rinsed with phosphate-free Dulbecco’s modified Eagle’s medium glutathione treatment to determine the total amount of biotinylated (GIBCO BRL), and then incubated for 3 hours at 37oC in 1 ml of the albumin-CPD tail. The biotinylated Alb-CPD tail was 32 same medium containing 0.5 mCi/ml [ P]-H3PO4 (New England immunoabsorbed and isolated on streptavidin agarose as described Nuclear). The cells were chilled on ice, washed with cold phosphate above. buffered saline (PBS) and then subjected to lysis in buffer containing To measure transport between the ER and the TGN, cells were 1% Triton X-100, 1% deoxycholate, 0.1 M NaCl and 50 mM Tris- labeled with 35S-Met and then chased with unlabeled Met at 37 oC. HCl, pH 7.4 and a mixture of phosphatase inhibitors: 5 mM NaF and CPD was immunoprecipitated and analyzed on denaturing 7% 0.1 mM Na vanadate, and a 1:100 dilution of the phosphatase inhibitor polyacrylamide gels. Previously, the transition of CPD from 170 kDa cocktail (Sigma). The cells were sonicated, centrifuged for 15 minutes to 180 kDa was found to occur in the late Golgi or TGN in AtT-20 at 4oC and subjected to immunoprecipitation using antibodies to cells (Varlamov et al., 1999b). human albumin (Milgram and Mains, 1994). The beads containing the The in vitro vesicle budding assay was performed as described 32P-labeled albumin-fusion proteins were washed several times with (Varlamov et al., 1999b). The resulting fractions were subjected to the lysis buffer without phosphatase inhibitors, and then either directly immunoprecipitation (Milgram and Mains, 1994) using antibodies to boiled in SDS gel-loading buffer or first incubated for 90 minutes at human albumin and CPE (Varlamov et al., 1996). room temperature with 2 U of thrombin. The samples were analyzed Antibody uptake experiments were performed as described (Eng et on a tricine gel and subjected to autoradiography. al., 1999) using the antibody to human albumin (1:150 dilution). The To study the effect of brain PP2A on the in vivo phosphorylated cells were fixed, stained with the antibody to syntaxin 6, and then with albumin fusion proteins, protein A beads containing the fluorescent secondary antibodies to rabbit and mouse IgG. Transferrin immunoprecipitated material were extensively washed with uptake was determined using Cy3-transferrin. dephosphorylation buffer lacking phosphatase inhibitors and then incubated for 2 hours at 37oC with the indicated amounts of PP2A. Microinjection of AtT-20 cells The degree of dephosphorylation was estimated by gel AtT-20 cells expressing the Alb-tail construct were cultured on autoradiography. poly--coated coverslips and microinjected with 3-5 mg/ml GST- For phosphoamino acid analysis of in vivo phosphorylated CPD tail fusion protein in 50 mM Tris-HCl pH 7.4 and 0.15 M NaCl albumin-CPD tail, the immunoprecipitated 32P-labeled albumin-CPD using an Eppendorf semiautomated microinjection system as tail was resolved on a tricine gel and transferred to nitrocellulose described (McIlroy et al., 1997). Two GST-CPD tail fusion proteins membranes. The 32P-labeled band corresponding to the location of were used; one contained the full-length CPD tail (58 amino acids) immunoprecipitated Alb-CPD tail was excised, washed with 50 mM and the other contained only the C-terminal 49 residues of the CPD NH4HCO3 pH 8 and then incubated at room temperature overnight tail. To allow the detection of the injected cells, 2 mg/ml of mouse with 1 µg of trypsin (Promega). The peptide mixture was dried, IgG was added to the microinjection solutions. Approximately 300 resuspended in 0.5 ml 6 N HCl and hydrolyzed for 70 minutes at cells were microinjected on each coverslip. After microinjection, the 110oC. The sample was dried, resuspended in the first dimension cells were incubated in culturing media for 3-5 hours at 37oC and then 314 JOURNAL OF CELL SCIENCE 114 (2)

Fig. 1. Purification of PP2A on the GST-CPD tail affinity column. (A) AtT-20 cells were metabolically labeled with 35S-Met, extracted and incubated with either GST or GST-CPD tail. The beads were washed with the extraction buffer, boiled and analyzed on a 10% polyacrylamide gel. Asterisks indicate the positions of specific bands. (B) The 65 kDa band was excised from the gel, incubated with trypsin, and subjected to MALDI mass spectrometry. The analysis yielded 14 peptide masses (Obs. masses) that matched the calculated peptide masses of the A subunit of mouse PP2A using MS-Fit program. (C) Mouse brain extract was incubated with the GST-CPD tail proteins as described in Materials and Methods and the CPD-tail bound material was analyzed on a western blot using antisera to the A, B and C subunits of PP2A. In panels A and C, the positions of prestained protein standards (in kDa) are indicated. subjected to the antibody uptake procedure described above using the antiserum to human albumin. The cells were washed, fixed, stained with fluorescently labeled secondary antibodies to mouse and rabbit IgG, and then examined using a Nikon microscope. Approximately 10-20% of the injected cells survived the microinjection procedures, and all surviving cells showed a similar pattern of staining.

Fig. 2. Mapping of the PP2A-binding region in the CPD cytoplasmic tail. (A) The deletion and point mutations of the CPD cytoplasmic tail fused to the C terminus of GST. Residues discussed later in the text are indicated, and include consensus sites for CKII (*), PKC (circle) and PKA (#). Point mutations were generated in ∆39L, the shortest CPD construct that binds PP2A with a near wild-type efficiency. The asparagine in the sixth position of the CPD tail was replaced with leucine to facilitate the construction of the mutants; this N to L substitution had little effect on PP2A binding. (B,C) Quantitation of binding. The constructs were subjected to the GST pull-down assay described in Materials and Methods and analyzed on western blots using antisera to the A and C subunits of PP2A. The efficiencies of binding of the deletion mutants (B) and point mutants (C) were estimated using a semi-quantitative western blot analysis. The entire binding experiment was repeated three to four times with comparable results. Error bars show standard error of the mean. PP2A binds to carboxypeptidase D 315

Fig. 3. Binding of PP2A to the cytoplasmic tails of the integral membrane proteins and microsomes. (A) GST fused with the cytoplasmic tails of CPD, PAM, the cation- independent M6PR, TGN38 or furin were incubated with mouse brain extract, processed as described in Materials and Methods, and analyzed on a western blot using an antiserum to the A subunit of PP2A. (B) The pull-down assay was performed in the presence (+) or absence (−) of 20 µg of the CPD tail peptide. (C) Co-immunoprecipitation of the integral membrane proteins with PP2A. Mouse brain extract was subjected to immunoprecipitation using antisera to either CPD, TGN38, furin or control rabbit IgG as described in Materials and Methods and analyzed on a western blot using an antiserum to the C subunit of PP2A (upper panel) or the indicated proteins (lower panels). The positions of the specific bands are indicated by asterisks. PP2A, the aliquot of the brain extract analyzed directly on a western blot. (D) Association of PP2A with microsomes. Cell fractionation was performed as described in Materials and Methods. PNS, post-nuclear supernatant; 100K g, first 100,000 g supernatant; wash 1 and 2 represent the subsequent washes of the initial pellet (S, supernatants; P, pellets). The positions and molecular weights (in kDa) of prestained protein standards are indicated.

RESULTS To identify the region of the cytoplasmic tail that interacts with PP2A, we generated a series of deletions within the CPD Identification of PP2A as a CPD-binding protein tail and tested PP2A binding. Removal of the C-terminal 44 The 58-amino acid cytoplasmic domain of CPD was used as a amino acids of CPD (∆44) completely abolishes PP2A binding, bait to identify interacting proteins. Initial experiments used AtT- whereas smaller deletions (∆ 22 to ∆ 38) are still able to bind 20 cells that were metabolically labeled with 35S-Met. Although PP2A (Fig. 2). The deletion 41 shows a reduced ability to bind the control GST beads demonstrate some nonspecific binding, PP2A (Fig. 2). The importance of the N-terminal 20 amino acid several additional proteins bind the CPD tail-containing beads region of the CPD tail for PP2A binding was verified by N- (Fig. 1A, asterisks). The binding of these proteins is independent terminal deletion analysis. All N-terminal deletions are unable of the phosphorylation state of the CKII phosphorylation sites, to bind PP2A (Fig. 2, +49 to +22). To elucidate the role of since mutation of the TDT to ADA or EDE did not influence individual amino acids within the PP2A binding region of CPD, binding (Fig. 1A). To identify the CPD-interacting proteins, point mutations were introduced in the C-terminal 39L deletion mouse brain extract was used as a source of proteins. Two major construct (Fig. 2). A substitution of the cysteine at the N bands migrating at 65 kDa and 36 kDa were isolated, subjected terminus of the CPD tail for alanine abolished PP2A binding to limited trypsin proteolysis, and analyzed by MALDI mass (Fig. 2, CSI/ASI). In contrast, replacement of the adjacent spectrometry. Computer searches of protein and GenBank serine with alanine does not have a significant effect on PP2A databases revealed a single major hit for each protein: fourteen binding, although replacement with glutamic acid slightly tryptic peptide masses of the 65 kDa protein matched with reduced binding (Fig. 2, CSI/CAI and CSI/CEI). A replacement predicted tryptic fragments of the A subunit of mouse PP2A of phenylalanine within the FxxL motif with tyrosine only (Fig. 1B) and six of the tryptic fragments from the 36 kDa moderately affected PP2A binding, whereas alanine in this protein matched with predicted peptide masses of the catalytic position substantially reduced binding of PP2A (Fig. 2, C subunit of mouse PP2A (data not shown). FxxL/YxxL and FxxL/AxxL). The last three basic amino acids To verify that PP2A binds to the CPD tail, we subjected the within the ∆39L construct are also crucial for PP2A binding; GST-tail bound material to western blot analysis using substitution of these residues for alanine greatly reduced PP2A antibodies to various subunits of PP2A. Proteins bound to the binding (Fig. 2, RQHH/AQAA). GST-CPD tail-containing beads are immunoreactive with antibodies to all three subunits of PP2A, with apparent Interaction of PP2A with the cytoplasmic domains of molecular weights of 65 kDa for the A subunit, 55 kDa for the the transmembrane proteins B subunit and 36 kDa for the C subunit (Fig. 1C). These To test whether the cytoplasmic domains of several other proteins did not bind to control resin containing GST alone (Fig. proteins that are enriched in the TGN and/or post-TGN 1C). Since the predominant form of PP2A found in vivo is a compartments also bind PP2A, brain extract was incubated with complex of the A, C and B subunits (Wera and Hemmings, fusion proteins containing various cytoplasmic tails attached to 1995), it is likely that PP2A binds to the CPD tail as a trimeric the C-terminus of GST. PP2A binds strongly to the cytoplasmic complex. tails of PAM, M6PR and TGN38 but not furin (Fig. 3A). 316 JOURNAL OF CELL SCIENCE 114 (2)

Fig. 4. In vitro phosphorylation and dephosphorylation of the cytoplasmic tail of CPD. (A) The GST-CPD tail was subjected to in vitro 32P phosphorylation using PKC, PKA or CKII as described in Materials and Methods. Dephosphorylation reactions containing 2-5 µg of the 32P-labeled GST-CPD tail peptide were carried out in the presence of the indicated amount of purified mouse brain PP2A. The samples were incubated for 2 hours at 37oC, boiled in SDS gel- loading buffer, analyzed on a 12% polyacrylamide gel and subjected to autoradiography. (B) The efficiency of dephosphorylation of PKC- phosphorylated CPD tail by either brain PP2A purified on the GST- CPD tail affinity column or muscle PP2A (Upstate Biotechnology). (C) Dephosphorylation of the CKII-phosphorylated full-length CPD tail (wt tail) and the truncated CPD tail without the PP2A-binding domain (+49 tail) by purified PP2A. Dephosphorylation reactions containing 2-5 µg of the 32P-labeled GST-CPD tail peptides were carried out in the presence of the indicated amount of PP2A (provided by Dr Marc Mumby). The positions and molecular weights (kDa) of prestained protein standards are indicated.

The subcellular distribution of PP2A The finding that PP2A binds the cytoplasmic domains of proteins suggests that PP2A may be associated with the intracellular membranes in vivo, using the cytoplasmic domains of the transmembrane proteins as receptors for organelle targeting. To determine whether PP2A is associated with the membranes, the total microsomal fraction was prepared from AtT-20 cells. The 100,000 g supernatant contains the bulk of PP2A immunoreactivity (Fig. 3D, compare PNS with 100K g, S) indicating that the majority of phosphatase is a soluble cytoplasmic protein. Typically, only 10-20% of the PP2A is associated with the microsomal membranes (Fig. 3D, 100K g, P). To confirm that PP2A is present in the microsomal membranes, the initial pellet was washed twice and each fraction was analyzed for PP2A. Although some PP2A was recovered in the supernatant of the first wash, a large amount of PP2A-immunoreactive material was detected in the pellet fraction of the second wash (Fig. 3D) indicating a tight association of PP2A with the membranes. Dephosphorylation of the cytoplasmic tail of CPD by purified PP2A The CPD tail is phosphorylated by PKC, PKA or CKII in vitro, and the phosphorylated tail is dephosphorylated by purified PP2A (Fig. 4A). With exception of the PKA- phosphorylated tail that demonstrates partial resistance to PP2A, dephosphorylation reactions are complete and When the CPD cytoplasmic tail peptide was included in the concentration dependent (Fig. 4A). To test whether the CPD incubation mixture, binding of PP2A to the GST-CPD tail tail modulates PP2A activity, the PKC-phosphorylated CPD protein is dramatically reduced (Fig. 3B), indicating that PP2A tail was tested with either purified PP2A/CPD tail complex binds CPD in a specific manner. The CPD tail peptide also from mouse brain or a commercial PP2A from rabbit muscle competes with PP2A for binding to the cytoplasmic tails of (which does not bind to the CPD tail peptide). Both TGN38 and PAM (Fig. 3B) suggesting that PP2A may use the preparations of PP2A show similar efficiencies (Fig. 4B), same region to bind these proteins. suggesting that interaction with CPD-tail neither inhibits nor We also explored the ability of PP2A to bind the cytoplasmic activates PP2A activity. In addition, comparable amounts of tails of these proteins in co-immunoprecipitation experiments. PP2A are required to dephosphorylate either CKII- PP2A co-immunoprecipitates with CPD or TGN38 but not phosphorylated full-length CPD tail (58 residues) or the with furin or control IgG (Fig. 3C, upper panel). The deletion mutant ‘+49’ that does not bind PP2A in vitro, immunoprecipitated material was also probed with antisera to indicating that PP2A binding does not influence the catalytic CPD, TGN38 and furin to confirm that the differential binding activity (Fig. 4C). The PP2A/CPD-tail complex from mouse of PP2A to these proteins is not due to inefficient recovery (Fig. brain is inhibited by tautomycin and okadaic acid with IC50 3C, lower panel). values of 200 nM and 1 nM, respectively. The sensitivity to PP2A binds to carboxypeptidase D 317

Fig. 5. Alb-tail is localized to the TGN in AtT-20 cells. (Top) Schematic diagram of the construct containing the transmembrane and cytoplasmic domains of CPD attached to the C-terminus of human albumin. The diagram is not to scale. (Bottom) Dual immunofluorescent analysis using antisera to human albumin and syntaxin 6 (left panels) or Cab45 and syntaxin 6 (right panels) was performed as described in Materials and Methods. BFA: Cells were treated with 5 µg/ml BFA for 2 min prior to analysis. Insets show enlarged portions of cells; arrowheads indicate representative structures stained with both antisera. Scale bar: 10 µm.

Fig. 6. Labeling of AtT- 20 cells with 32P and analysis of protein phosphorylation in vivo. (A) AtT-20 cells expressing albumin fused with either the transmembrane and cytoplasmic domain or only the transmembrane domain of CPD were incubated for 3 hours at 37oC in the presence of 32 [ P]-H3PO4, lysed and then subjected to immunoprecipitation using antibodies to human albumin. The beads containing the 32P- labeled albumin-fusion proteins were washed several times with the lysis buffer and then either directly boiled in SDS gel-loading buffer (untreated) or incubated with thrombin. The samples were analyzed on a tricine gel and subjected to autoradiography. Lane 1, Alb-tail; lane 2, Alb-transmembrane domain; lane 3, wild type cells. (B) Dephosphorylation of the in vivo phosphorylated Alb-tail by brain PP2A. Protein A beads containing the immunoprecipitated Alb-tail were washed with dephosphorylation buffer, incubated for 2 hours at 37oC with PP2A, and analyzed on a polyacrylamide gel followed by autoradiography. (C) Phosphoamino acid analysis of the in vivo 32P-labeled tail of CPD. The 32P-labeled Alb-tail was subjected to phosphoamino acid analysis as described in Materials and Methods. The products were separated by thin layer electrophoresis first at pH 1.9 and then in a second dimension at pH 3.5. (Left panel) phosphoserine, phosphothreonine and phosphotyrosine standards were visualized with ninhydrin. (Right panel) autoradiography of the 32P-labeled hydrolysate of the albumin fusion protein. In A,B, the positions and molecular weights (kDa) of prestained protein standards are indicated. 318 JOURNAL OF CELL SCIENCE 114 (2)

Fig. 7. Effect of okadaic acid on the intracellular trafficking of CPD. (A) AtT-20 cells expressing Alb-tail were labeled with 35S-Met and chased in EZ-Link Sulfo-NHS-SS-Biotin in the absence (C) or presence of 200 nM okadaic acid (OA). The biotinylated material was recovered by immunoprecipitation followed by streptavidin-agarose to select the cell surface-derived material. (Top) A representative autoradiogram. (Bottom) Quantitation of the effect of okadaic acid on the rate of the Alb-CPD tail trafficking to the cell surface. Error bars show standard error of the mean from three independent experiments. (B) Endocytosis of Alb-tail from the cell surface of AtT-20 cells. Metabolically labeled AtT-20 cells were subjected to cell surface biotinylation at 4oC, and then chased for 0 or 30 minutes at 37oC. The cells were treated with (+) or without (−) glutathione (GSH) to remove the cell surface-associated biotin, lysed, and then processed to select the biotinylated pool of Alb-tail molecules. C, control cells; OA, cells treated with 200 nM okadaic acid. Graph shows quantitation of the internalization rate of Alb-tail calculated as the percentage that was glutathione insensitive (the difference between the 30 minute timepoint and the 0 timepoint in the presence of glutathione divided by the 30 minute point in the absence of glutathione). Error bars show standard error of the mean from three experiments. (C) Trafficking of CPD from the ER to the late Golgi compartments. AtT-20 cells were labeled with 35S-Met and chased with unlabeled Met at 37 oC for the indicated time in the absence (control) or presence of 200 nM okadaic acid. CPD was immunoprecipitated and analyzed on denaturing 7% polyacrylamide gels. (D) Packaging of Alb-tail and CPE into post-TGN secretory vesicles. The in vitro vesicle budding assay was performed as described (Varlamov et al., 1999b) using AtT-20 cells expressing Alb-tail. Permeabilized cells were incubated in the absence (C-) or presence (C) of an energy-generating system either alone or in the presence of 200 nM okadaic acid. Following incubation, the samples were centrifuged to generate supernatants (S) containing nascent secretory vesicles and pellets (P) containing the TGN, and then subjected to immunoprecipitation using either antibodies to human albumin to detect Alb-tail or antibodies to CPE. Graph shows quantitation of the TGN budding assay after subtraction of the budding in the absence of energy; error bars represent standard error of the mean, n=3. okadaic acid further suggests that the phosphatase present in domain in order to distinguish phosphorylation of albumin from brain extracts that bound to the CPD tail is PP2A. phosphorylation of the cytoplasmic domain. When expressed in AtT-20 cells, Alb-tail is predominantly co-localized with Phosphorylation of CPD in AtT-20 cells syntaxin 6, a TGN marker (Fig. 5). To confirm that the Alb-tail To examine whether the CPD tail undergoes phosphorylation is present in the TGN and not the Golgi, which also shows a in cells, we generated a fusion protein containing the perinuclear distribution in the AtT-20 cell line, the cells were transmembrane and cytoplasmic domains of CPD attached to treated with BFA. This treatment causes Golgi proteins to the C terminus of albumin (Alb-tail) (Fig. 5). A cleavage site redistribute to the ER while TGN proteins remain associated for thrombin was introduced upstream of the transmembrane with a perinuclear-localized microtubule-organizing center PP2A binds to carboxypeptidase D 319

Fig. 8. Effect of PP2A binding on endocytosis of the Alb-tail construct. (A) To follow the internalization, AtT-20 cells expressing the Alb-tail constructs containing either the full- length CPD tail (CSI) or the mutant CPD tail that does not bind PP2A (AAI) were incubated at 4oC in the presence of antibodies to albumin (Alb) and then chased at 37oC for 15 minutes. The cells were fixed, stained with antibodies to syntaxin 6 (Syn 6), and then incubated with the fluorescent secondary antibodies to mouse and rabbit IgG. The right panels represents merged images, with co-staining appearing as yellow. (B) AtT-20 cells expressing the AAI mutant Alb-tail construct were pre-incubated with antiserum to albumin and then chased at 37oC for 15 minutes with Cy3-transferrin. The cells were fixed and stained with fluorescein isothiocyanate- coupled secondary antibodies to rabbit IgG. Co-localization of internalized Alb-tail and transferrin to vesicular structures is indicated by arrowheads. Scale bar: 10 µm.

Mapping phosphorylation sites of the CPD cytoplasmic tail in vivo and in vitro Phosphoamino acid analysis of the 32P-labeled CPD tail from the Alb-tail construct revealed that phosphorylation predominantly occurs at threonine residues (Fig. 6C), consistent with the potential phosphorylation sites for CKII (Fig. 2). To map the phosphorylation sites in vitro, the CPD tail phosphorylated by either CKII, PKA or PKC was first treated with either trypsin or V8 protease and then subjected to mass spectrometry. MALDI analysis of the tryptic fragments of the CKII- phosphorylated tail showed two peptides with observed masses (2666.6 and 2746.9) that match the predicted masses of the monophosphorylated (2667.2) and diphosphorylated (2747.2) forms of the C-terminal tryptic fragment Ser36-Lys57 of the CPD tail. These two peptides were not detected in the absence of CKII. The collision-induced dissociation mass spectrometry (Klausner et al., 1992; Reaves and Banting, 1992). When the analysis revealed a fragmentation pattern of the 2667 Da cells were treated with BFA for 2 minutes (Fig. 5) or 30 minutes peptide that is consistent with a mixture of two peptides, each (not shown), the Alb-tail remains co-localized with syntaxin 6 containing a single phosphate on either of the threonine in the perinuclear region. In contrast, the Golgi protein Cab45 residues within the CKII site. To confirm that CKII exclusively rapidly redistributes to a diffuse ER-like pattern in AtT-20 phosphorylates these threonine residues, they were replaced cells (Fig. 5), as previously reported in the 3T3-L1 cell line with glutamate and the resulting construct was subjected to (Scherer et al., 1996). These results suggest that Alb-tail is CKII phosphorylation using 32P-ATP. No phosphorylation of predominantly localized to the TGN. Furthermore, the half-life the mutant form of the CPD tail was detected, indicating that of the Alb-tail protein in AtT-20 cells (5 hours) is very similar the two threonine residues within the predicted CKII site are to that of CPD (when CPD is overexpressed at similar levels as the only residues within the CPD tail that are phosphorylated the Alb-tail construct), indicating that the cytoplasmic tail is by this kinase (Fig. 2, asterisks above sequence). sufficient for correct routing of the protein (data not shown). The MALDI analysis of the CPD tail that was Upon labelling of the cells with 32P, the 80 kDa Alb-tail protein phosphorylated by PKC in vitro revealed a tryptic peptide with becomes phosphorylated (Fig. 6A, lane 1, left). When Alb-tail a molecular mass of 2803.2 (data not shown). This mass, which was digested with thrombin, the 32P is associated with the 10 was not present in the absence of PKC, matches the predicted kDa CPD tail fragment and not the 67 kDa albumin fragment monophosphorylated tryptic fragment Ser36-His58 (2804.2). (Fig. 6A). When the PKC-phosphorylated tail was digested with V8 Purified PP2A is able to completely dephosphorylate the in protease, peptides of 3107.3, 3924.3 and 4004.8 were vivo phosphorylated Alb-tail (Fig. 6B), although with a lower observed, which match the predicted monophosphorylated V8 efficiency compared to the dephosphorylation of the GST-CPD protease fragment Glu25-Glu49 (3107.3) and the tail (Fig. 4). This could be explained either by the presence of monophosphorylated and diphosphorylated V8 protease trace amounts of phosphatase inhibitors used during the fragment Ile26-His58 (3923.8 and 4003.8). This region of the immunoprecipitation, or by the fact that the Alb-tail substrate CPD tail contains three potential sites for PKC (Fig. 2, dots was immobilized on a solid support that could influence the above sequence). Similar analysis of the PKA-phosphorylated rate of hydrolysis. CPD tail by MALDI along with consideration of the consensus 320 JOURNAL OF CELL SCIENCE 114 (2) proteins to accumulate in the TGN, and finally incubated at 37oC for 30 minutes in the presence of biotin. During the 37oC incubation the metabolically labeled proteins that appear at the cell surface undergo biotinylation. Alb-tail was detectable at the cell surface after 30 minutes of incubation (Fig. 7A). The PP2A inhibitor okadaic acid inhibited by 45% the TGN-to-plasma membrane trafficking of Alb-tail (Fig. 7A). To test whether okadaic acid affects endocytosis of Alb-tail from the plasma membrane, metabolically labeled AtT-20 cells were subjected to cell surface biotinylation at 4oC and then chased at 37oC to allow Alb-tail to internalize. Upon internalization, the biotinylated pool of Alb-tail becomes resistant to the reducing action of glutathione, whereas the cell-surface biotinylated proteins can be de-biotinylated by glutathione. When cells are treated with glutathione immediately after biotinylation, Alb-tail molecules can be efficiently de- biotinylated indicating that this pool of Alb-tail resides at the cell surface (Fig. 7B, 0′). When cells are incubated at 37oC for 30 minutes, 40% of the biotinylated Alb-tail becomes resistant to glutathione, indicating that a significant fraction of Alb-tail undergoes internalization (Fig. 7B, 30′). Treatment of AtT-20 cells with okadaic acid during the internalization step does not significantly alter the amount of internalization from the cell surface (Fig. 7B), suggesting that PP2A-mediated dephosphorylation is not involved in the uptake of Alb-tail from the cell surface to a glutathione-resistant compartment. The effect of okadaic acid on the rate of ER-to-TGN trafficking of CPD was also examined. CPD is synthesized in the ER as a 170 kDa form and is modified to a 180 kDa form in the TGN or early post-TGN vesicles, presumably due to sialylation of N-linked carbohydrates (Varlamov et al., 1999b). The rate of conversion of the 170 kDa form into the 180 kDa form reflects the efficiency of the ER-to-Golgi and intra-Golgi Fig. 9. Microinjection of AtT-20 cells with the CPD tail. AtT-20 cells trafficking of CPD. In control AtT-20 cells, CPD undergoes post- expressing Alb-tail were microinjected with GST fused either with translational modification within the first 30 minutes of chase the full-length CPD tail (+58) or with the truncated form of the CPD (Fig. 7C). Similarly, cells treated with 200 nM okadaic acid tail that does not bind PP2A (+49). To detect the microinjected cells, mouse IgG was added to the microinjection mixture. After demonstrate the same rate of conversion (Fig. 7C), indicating microinjection, the cells were first incubated for 3-6 hours at 37oC, that this low concentration of okadaic acid does not affect the then at 4oC in the presence of antibodies to albumin and finally early secretory pathway. In addition, treatment of the AtT-20 chased at 37oC for 30 minutes. The cells were fixed and stained with cells with 200 nM okadaic acid does not affect the intracellular the fluorescent secondary antibodies to mouse and rabbit IgG as distribution of Alb-tail or syntaxin 6 (not shown), indicating no described in Materials and Methods. (Right) Uptake of antibodies to effect on general TGN structure. albumin. (Left) The injected cells visualized using mouse IgG. A TGN budding assay with semi-intact AtT-20 cells Several examples of cells injected with the full length (+58) peptide expressing Alb-tail was used to test whether okadaic acid alters µ are shown. Scale bar: 10 m. the rate of vesicle formation from the TGN. Budding of vesicles containing Alb-tail is energy dependent (Fig. 7D, compare C- sites suggested that PKA phosphorylation is likely to occur at with C). The budding efficiency of the Alb-tail-containing the potential PKA site Ser36 (Fig. 2, # above sequence). vesicles from the TGN is fairly low compared with the relatively Taken together, the cytoplasmic tail of CPD can be high budding efficiency of vesicles containing soluble CPE (Fig. phosphorylated in vitro on Thr by CKII and on Ser by PKC and 7D). This is consistent with our previous results that CPD exits PKA. In contrast, in vivo 32P-labeling in AtT-20 cells revealed the TGN more slowly than soluble proteins and hormones phosphorylation only on Thr, which presumably represents the (Varlamov et al., 1999b). Formation of Alb-tail-containing CKII sites. Furthermore, PP2A acts as a broad specificity vesicles is not affected by okadaic acid added to the budding phosphatase that may dephosphorylate different phosphorylation mixture (Fig. 7D, OA) indicating that PP2A is not involved in sites in the CPD tail. this event. Effect of the PP2A inhibitor on intracellular trafficking Intracellular trafficking of Alb-tail point mutants that of CPD do not bind PP2A To test whether TGN-to-plasma membrane trafficking is To examine whether the mutation of the PP2A-binding region regulated by PP2A, AtT-20 cells were briefly labeled with 35S- affects endocytic trafficking of Alb-tail, cells expressing the Alb- Met, incubated at 18.5oC to allow the metabolically labeled tail construct containing the CSI/AAI mutation (Fig. 2) were PP2A binds to carboxypeptidase D 321 incubated at 4oC in the presence of antibodies to albumin, and proximal cysteine residue is crucial for PP2A binding is then chased at 37oC. Within 15 minutes of incubation, the wild unexpected because no cysteine residues are found in the type form of Alb-tail is internalized to the TGN and is generally comparable positions of the cytoplasmic tails of the other PP2A- co-stained with syntaxin 6 (Fig. 8A, CSI). Although the mutant binding proteins. It is possible that the cytoplasmic tails of form of Alb-tail demonstrates a high degree of overlap with unrelated proteins share the same structural elements that form syntaxin 6 after 15 minutes of incubation, a detectable amount the PP2A-binding domain. of albumin immunoreactivity is seen in the large endosome-like The observation that the CPD cytoplasmic tail can be structures located in the cytoplasm and near the plasma phosphorylated in vitro on serine by PKA and PKC, and on membrane (Fig. 8A, AAI). Co-uptake of antibodies to albumin threonine by CKII was expected, based on the presence of and Cy3-transferrin reveals a partial colocalization of the consensus sites for these kinases. However, in vivo data revealed albumin-immunoreactive cytoplasmic vesicles with internalized phosphorylation only on threonines (Fig. 6), consistent with the transferrin indicating that a fraction of the mutant form of Alb- model that CKII is important in the trafficking of CPD. In the tail is endosomal (Fig. 8B). present study, PP2A efficiently dephosphorylates the CPD cytoplasmic tail, raising the possibility that it regulates the Effect of the CPD peptide on endocytic trafficking of phosphorylation and sorting of CPD in vivo, analogous to the Alb-tail proposed regulation of the phosphorylation and sorting of furin Since the CPD tail peptide competes with PP2A for binding to by PP2A (Molloy et al., 1998). It is not clear as to whether the the cytoplasmic tails of different TGN/endosomal proteins (Fig. binding of PP2A to the CPD tail and other proteins is regulated. 3B), we examined the effect of the CPD tail peptide on endocytic Although the presence of a phosphorylated form of the CPD tail trafficking of Alb-tail in vivo using a microinjection approach. in AtT-20 cells would seem to suggest that PP2A is not AtT-20 cells expressing the wild-type form of Alb-tail were permanently bound to the CPD tail, bound PP2A may be microinjected with either the full-length 58-residue CPD tail or sterically constrained from dephosphorylating the same the truncated (+49) form that does not bind PP2A (Fig. 2). The molecule of CPD, unless the PP2A binds through its catalytic cells were incubated with antibodies to albumin and then site (which is unlikely because the binding of PP2A to the CPD subjected to immunofluorescence. The microinjection of the tail is not dependent on phosphorylation). Thus, the most likely truncated CPD tail has no effect on TGN accumulation of Alb- scenario is that CPD-bound PP2A functions in the tail; both uninjected and injected cells show a perinuclear dephosphorylation of other molecules and the increased local distribution of albumin immunoreactivity after 30 minutes of concentration of these various molecules during the sorting internalization (Fig. 9, +49). In contrast, when the cells were process would lead to their dephosphorylation. microinjected with the full-length CPD tail containing the intact The finding that a low concentration of okadaic acid blocks PP2A-binding domain, Alb-tail fails to internalize into the the delivery of Alb-tail to the cell surface is evidence that PP2A perinuclear location; albumin immunoreactivity is found in the regulates CPD trafficking between the TGN and the plasma dispersed vesicles throughout the cytoplasm (Fig. 9, +58). The membrane. Although previous studies demonstrated that dramatic effect of the full length (+58) peptide on the distribution okadaic acid induces an early block in ER export by preventing of internalized Alb-tail was apparent in all injected cells. This protein entry into the COPII-coated vesicles, these studies used result suggests that TGN targeting of Alb-tail is blocked in the 1 µM okadaic acid (Davidson et al., 1992; Pryde et al., 1998) presence of the PP2A-binding peptide. and treatment of AtT-20 cells with the same dose of okadaic acid results in a change of morphology and cell death (our unpublished observations). In contrast, 200 nM okadaic acid DISCUSSION used in the present study did not have an effect on the cell shape and organelle morphology, on the ER-to-Golgi trafficking of A major finding of the present study is that PP2A binds the CPD, on the budding of CPD into nascent vesicles or on the cytoplasmic tails of CPD, PAM, M6PR and TGN38. The finding uptake of CPD from the cell surface to a glutathione-insensitive that binding of PP2A to the cytoplasmic tails of CPD, PAM and compartment, indicating that the inhibition of the TGN-to- TGN38 can be competed by the CPD tail suggests the same plasma membrane transport is not due to a general effect on region of the PP2A holoenzyme interacts with the cytoplasmic membrane trafficking. tails of these proteins. Interestingly, the PP2A-binding region of It has been proposed that the TGN localization of furin is CPD contains a YxxL-like motif (FHRL) that was previously achieved by retrieval from immediate post-TGN organelles, and found to be important for endocytosis and TGN localization of that this step is regulated by CKII phosphorylation of Ser CPD (Eng et al., 1999). A similar Y/FxxV/L is found in the residues within the cytoplasmic tail (Dittie et al., 1997) followed cytoplasmic tails of TGN38, PAM and the M6PR. A substitution by the binding of PACS-1 to the phosphorylated protein (Wan of the phenylalanyl residue within the FHRL sequence of CPD et al., 1998). A similar cycling loop operates between the cell for alanine (but not for tyrosine) abolishes PP2A binding, surface and early endosomes; once in early endosomes, the indicating that a large bulky hydrophobic residue in this position CKII-phosphorylated furin is retrieved by the plasma membrane, is required for PP2A binding. The other distinguishing feature while the dephosphorylated form of furin is sorted into the TGN of the PP2A-binding region of CPD is that almost half of the recycling loop (Jones et al., 1995; Molloy et al., 1998). Furin region is composed of basic amino acids that are important for undergoes dephosphorylation by isoforms of PP2A, which is PP2A binding (Fig. 2). Stretches of basic residues are also found necessary for sorting of furin from early endosomes to the TGN in the cytoplasmic tails of TGN38, PAM and M6PR, raising the (Molloy et al., 1998). Although furin does not appear to directly possibility that electrostatic interactions coordinate the high- bind PP2A, it is likely that the binding of PP2A to the affinity binding of PP2A. The finding that the membrane cytoplasmic tails of CPD and other transmembrane proteins 322 JOURNAL OF CELL SCIENCE 114 (2) provides an efficient and targeted dephosphorylation of McIlroy, J., Chen, D., Wjasow, C., Michaeli, T. and Backer, J. M. (1997). neighboring proteins during their sorting in the secretory and Specific activation of p85-p110 phosphatidylinositol 3′-kinase stimulates endocytic pathways. This model is supported by the relatively DNA synthesis by ras- and p70 S6 kinase-dependent pathways. Mol. Cell. Biol. 17, 248-255. small effect that elimination of the PP2A- within the Milgram, S. L. and Mains, R. E. (1994). Differential effects of temperature CPD tail has on the trafficking of CPD-tail containing proteins. blockade on the proteolytic processing of three secretory granule-associated According to this model, the presence of wild-type CPD and proteins. J. Cell Sci. 107, 737-745. other PP2A-binding proteins in the cell is sufficient for the Mitra, A., Song, L. and Fricker, L. D. (1994). The C-terminal region of carboxypeptidase E is involved in membrane binding and intracellular routing trafficking of proteins that do not directly bind PP2A. However, in AtT-20 cells. J. Biol. Chem. 269, 19876-19881. when the binding of PP2A to the various proteins is competed Molloy, S. S., Thomas, L., Kamibayashi, C., Mumby, M. C. and Thomas, by injection of the CPD tail (Fig. 9), the rate of movement of G. (1998). Regulation of endosome sorting by a specific PP2A isoform. J. CPD from the cell surface to the TGN is greatly slowed. Taken Cell Biol. 142, 1399-1411. together with previous studies implicating PP2A in furin Molloy, S. S., Anderson, E. D., Jean, F. and Thomas, G. (1999). Bi-cycling the furin pathway: from TGN localization to pathogen activation and trafficking, it is likely that PP2A plays an important role in the embryogenesis. Trends Cell Biol. 9, 28-35. intracellular trafficking of CPD and a variety of other proteins. Price, N. E. and Mumby, M. C. (1999). Brain protein serine/threonine phosphatases. Curr. Opin. Neurobiol. 9, 336-342. This work was primarily supported by National Institutes of Pryde, J. G., Farmaki, T. and Lucocq, J. M. (1998). Okadaic acid induces Health grant RO1-DK-51271 and also by grants R01-DA-55711 and selective arrest of protein transport in the rough endoplasmic reticulum K02-DA-00194 (L.D.F.). Mass spectrometry was performed in the and prevents export into COPII-coated structures. Mol. Cell. Biol. 18, Laboratory of Macromolecular Analysis of Albert Einstein College of 1125-1135. Medicine, which is supported in part by the Cancer Center Core grant Reaves, B. and Banting, G. (1992). Perturbation of the morphology of the CA13330 and by the Diabetes Research Training Center Core grant trans-Golgi network following brefeldin A treatment: Redistribution of a DK20541. The DNA sequencing facility of Albert Einstein College of TGN-specific integral membrane protein, TGN38. J. Cell Biol. 116, 85-94. Medicine is supported in part by Cancer Center grant CA13330. Special Reilein, A. R., Tint, I. S., Peunova, N. I., Enikolopov, G. N. and Gelfand, V. I. (1998). Regulation of organelle movement in melanophores by protein thanks to Dr Lin Yan for performing the mass spectrometry for the initial kinase A (PKA), protein kinase C (PKC), and protein phosphatase 2A (PP2A). identification of PP2A subunits and to Dr Olga Varlamova for help with J. Cell Biol. 142, 803-813. the phosphorylation assay. Antisera and/or GST fusion constructs Scherer, P. E., Lederkremer, G. Z., Williams, S., Fogliano, M., Baldini, G. containing the cytoplasmic tails of M6PR, TGN38, PAM and furin were and Lodish H. F. (1996). Cab45, a novel Ca2+ binding protein localized to generously provided by Drs Suzanne Pfeffer, Sharon Milgram, Ruth the Golgi lumen. J. Cell Biol. 133, 257-268. Angeletti, Gary Thomas, Richard Scheller and Philipp Scherer. Special Seidah, N. G. and Chretien, M. (1997). Eukaryotic protein processing: thanks to Dr Marc Mumby for providing purified PP2A. endoproteolysis of precursor proteins. Curr. Opin. Biotechnol. 8, 602-607. Slepnev, V. I., Ochoa, G. C., Butler, M. H., Grabs, D. and Camilli, P. D. (1998). Role of phosphorylation in regulation of the assembly of endocytic REFERENCES coat complexes. Science 281, 821-824. Song, L. and Fricker, L. D. (1996). Tissue distribution and characterization of Boyle, W. J., van der Geer, P. and Hunter, T. (1991). Phosphopeptide mapping soluble and membrane-bound forms of metallocarboxypeptidase D. J. Biol. and phosphoamino acid analysis by two-dimensional separation on thin-layer Chem. 271, 28884-28889. cellulose plates. Methods Enzymol. 201, 110-149. Sontag, E., Nunbhakdi-Craig, V., Lee, G., Brandt, R., Kamibayashi, C., Davidson, H. W., McGowan, C. H. and Balch, W. E. (1992). Evidence for the Kuret, J., White, C. L., III, Mumby, M. C. and Bloom, G. S. (1999). regulation of exocytic transport by protein phosphorylation. J. Cell Biol. 116, Molecular interactions among protein phosphatase 2A, tau, and microtubules. 1343-1355. J. Biol. Chem. 274, 25490-25498. Dittie, A. S., Thomas, L., Thomas, G. and Tooze, S. A. (1997). Interaction of Tan, F., Rehli, M., Krause, S. W. and Skidgel, R. A. (1997). Sequence of furin in immature secretory granules from neuroendocrine cells with the AP-1 human carboxypeptidase D reveals it to be a member of the regulatory adaptor complex is modulated by casein kinase II phosphorylation. EMBO J. carboxypeptidase family with three tandem domains. Biochem. J. 16, 4859-4870. 327, 81-87. Dong, W., Fricker, L. D. and Day, R. (1999). Carboxypeptidase D is a potential Varlamov, O., Leiter, E. H. and Fricker, L. D. (1996). Induced and candidate to carry out redundant processing functions of carboxypeptidase E spontaneous mutations at Ser202 of carboxypeptidase E: Effect on based on comparative distribution studies in the rat central nervous system. expression, activity, and intracellular routing. J. Biol. Chem. 271, Neuroscience 89, 1301-1317. 13981-13986. Eng, F. J., Varlamov, O. and Fricker, L. D. (1999). Sequences within the Varlamov, O. and Fricker, L. D. (1998). Intracellular trafficking of cytoplasmic domain of gp180/carboxypeptidase D mediate localization to the metallocarboxypeptidase D in AtT-20 cells: Localization to the trans-Golgi trans-Golgi network. Mol. Biol. Cell. 10, 35-46. network and recycling from the cell surface. J. Cell Sci. 111, 877-885. Horn, M. and Banting, G. (1994). Okadaic acid treatment leads to a Varlamov, O., Eng, F. J., Novikova, E. G. and Fricker, L. D. (1999a). fragmentation of the trans-Golgi network and an increase in expression of Localization of metallocarboxypeptidase D in AtT-20 cells: Potential role in TGN38 at the cell surface. Biochem. J. 301, 69-73. prohormone processing. J. Biol. Chem. 274, 14759-14767. Jones, B. G., Thomas, L., Molloy, S. S., Thulin, C. D., Fry, M. D., Walsh, K. Varlamov, O., Wu, F., Shields, D. and Fricker, L. D. (1999b). Biosynthesis A. and Thomas, G. (1995). Intracellular trafficking of furin is modulated by and packaging of carboxypeptidase D into nascent secretory vesicles in the phosphorylation state of a casein kinase II site in its cytoplasmic tail. pituitary cell lines. J. Biol. Chem. 274, 14040-14045. EMBO J. 14, 5869-5883. Volz, B., Orberger, G., Porwoll, S., Hauri, H. and Tauber, R. (1995). Selective Klausner, R. D., Donaldson, J. G. and Lippincott-Schwartz, J. (1992). reentry of recycling cell surface glycoproteins to the biosynthetic pathway in Brefeldin A: Insights into the control of membrane traffic and organelle human hepatocarcinoma HepG2 cells. J. Cell Biol. 130, 537-551. structure. J. Cell Biol. 116, 1071-1080. Wan, L., Molloy, S. S., Thomas, L., Liu, G., Xiang, Y., Rybak, S. L. and Krautheim, A., Rustenbeck, I. and Steinfelder, H. J. (1999). Phosphatase Thomas, G. (1998). PACS-1 defines a novel gene family of cytosolic sorting inhibitors induce defective hormone secretion in insulin-secreting cells and proteins required for trans-Golgi network localization. Cell. 94, 205-216. entry into apoptosis. Exp. Clin. Endocrinol. Diabetes 107, 29-34. Wera, S. and Hemmings, B. A. (1995). Serine/threonine protein phosphatases. Kuroki, K., Eng, F., Ishikawa, T., Turck, C., Harada, F. and Ganem, D. Biochem. J. 311, 17-29. (1995). gp180, a host cell glycoprotein that binds duck hepatitis B virus Xin, X., Varlamov, O., Day, R., Dong, W., Bridgett, M. M., Leiter, E. H. and particles, is encoded by a member of the carboxypeptidase gene family. J. Fricker, L. D. (1997). Cloning and sequence analysis of cDNA encoding rat Biol. Chem. 270, 15022-15028. carboxypeptidase D. DNA Cell Biol. 16, 897-909. Lowe, M., Gonatas, N. K. and Warren, G. (2000). The mitotic phosphorylation Zhou, A., Webb, G., Zhu, X. and Steiner, D. F. (1999). Proteolytic processing cycle of the cis-Golgi matrix protein GM130. J. Cell Biol. 149, 341-356. in the secretory pathway. J. Biol. Chem. 274, 20745-20748.