J Biol Chem, Vol. 274, Issue 49, 34646-34656, December 3, 1999

The Novel Kinase Peptidylglycine -Amidating Monooxygenase Cytosolic Interactor Protein 2 Interacts with the Cytosolic Routing Determinants of the Peptide Processing Enzyme Peptidylglycine -Amidating Monooxygenase^*

Benjamin D. Caldwell, Daniel N. Darlington^§, Peter Penzes, Richard C. Johnson, Betty A. Eipper, and Richard E. Mains^¶

From the Departments of Neuroscience and Physiology, The Johns Hopkins University School of Medicine, Baltimore Maryland 21205 and the ^§ Departments of Surgery and Physiology, University of Maryland School of Medicine, Baltimore, Maryland 21201

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The cytosolic domain of the peptide-processing integral membrane protein peptidylglycine -amidating monooxygenase (PAM; EC 1.14.17.3) contains multiple signals determining its subcellular localization. Three PAM cytosolic interactor proteins (P-CIPs) were identified using the yeast two hybrid system (Alam, M. R., Caldwel, B. D., Johnson, R. C., Darlington, D. N., Mains, R. E., and Eipper, B. A. (1996) J. Biol. Chem. 271, 28636-28640); the partial amino acid sequence of P-CIP2 suggested that it was a protein kinase. In situ hybridization and immunocytochemistry show that P-CIP2 is expressed widely throughout the brain; PAM and P-CIP2 are expressed in the same neurons. Based on subcellular fractionation, the 47-kDa P-CIP2 protein is mostly cytosolic. P-CIP2 is a highly selective kinase, phosphorylating the cytosolic domain of PAM, but not the corresponding region of furin or carboxypeptidase D. Although P-CIP2 interacts with stathmin, it does not phosphorylate stathmin. Site-directed mutagenesis, phosphoamino acid analysis, and use of synthetic peptides demonstrate that PAM-Ser⁹⁴⁹ is the major site phosphorylated by P-CIP2. Based on both in vitro binding experiments and co-immunoprecipitation from cell extracts, P-CIP2 interacts with PAM proteins containing the wild type cytosolic domain, but not with mutant forms of PAM whose trafficking is disrupted. P-CIP2, through its highly selective phosphorylation of a key site in the cytosolic domain of PAM, appears to play a critical role in the trafficking of this protein.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The targeting of membrane proteins requires sorting signals that govern the interaction of the membrane protein with the proper cytosolic proteins. For integral membrane proteins targeted to trans-Golgi network (TGN),¹ endosomes, lysosomes, or plasma membrane, a number of distinct and often overlapping motifs have been identified in the cytosolic domain (1-11). Tyr-containing motifs can recruit adaptor proteins and coat proteins to membranes (12-15) and phosphorylation can govern individual trafficking steps (3, 4, 7, 9, 11, 16-18). Neurons and endocrine cells store bioactive peptides in a specialized organelle, the large dense core vesicle. Very little is known about the targeting signals governing membrane proteins that function in large dense core vesicles. The peptide amidating enzyme, peptidylglycine -amidating monooxygenase (PAM), functions in the regulated secretory pathway and has been used to study trafficking to large dense core vesicles.

For roughly half of all bioactive peptides, COOH-terminal amidation is the final step of activation following endoproteolytic digestion from inactive precursors (19, 20). PAM is a Type I integral membrane protein with routing determinants within its cytosolic COOH-terminal domain (CD). Membrane forms of PAM are localized primarily in a post-Golgi tubuloreticular compartment overlapping but not identical with the TGN38-containing compartment; PAM that reaches the plasma membrane recycles to the TGN region (21). Truncation of the CD of PAM results in cell surface localization with little internalization (22). Fusion of the transmembrane domain/CD of PAM to the lumenal domain of a plasma membrane protein or to a soluble secretory granule constituent results in TGN localization of both chimerae (21, 23). Thus the CD of PAM contains routing signals which mediate the localization of membrane forms of PAM within the regulated secretory pathway.

We recently reported the use of the yeast two-hybrid system to identify PAM COOH-terminal interactor proteins (P-CIPs) (24). Three cDNAs were identified; one encoded a putative serine/threonine protein kinase followed by a RNA binding motif (P-CIP2) and another encoded Kalirin (formerly P-CIP10), a novel member of the Dbl family of GDP/GTP exchange factors that interacts with Rac1, a small GTP-binding protein of the Rho subfamily (25). Both P-CIP2 and Kalirin were shown to interact directly with membrane PAM in vitro using GST fusion proteins expressed in bacteria. P-CIP2 was also identified independently as an interactor of stathmin, a cytosolic protein involved in microtubule destabilization (26). Stathmin is homologous to the neuronal protein SCG10, which also destabilizes microtubules (27).

In this study we set out to characterize P-CIP2 more fully and to determine whether P-CIP2 is a functional kinase capable of phosphorylating PAM. For the interaction of P-CIP2 and PAM to be of functional significance, the two proteins must be expressed in the same cell; we used in situ hybridization and immunocytochemistry to demonstrate that the sites of P-CIP2 and PAM expression overlapped extensively in the brain. Evidence of a functionally significant interaction of P-CIP2 with PAM was shown by the ability of endogenous P-CIP2 to bind to PAM expressed in AtT-20 corticotrope tumor cells and in test tube assays. In vitro binding assays and co-immunoprecipitation demonstrated that mutant PAM proteins that are misrouted in AtT-20 cells do not interact well with P-CIP2. The site of PAM-CD phosphorylation by P-CIP2 was identified, and the specificity of P-CIP2 for various other substrates was investigated. The results indicate that P-CIP2 phosphorylates a specific region of the cytosolic domain of PAM which contains routing determinants essential for the correct localization of membrane PAM, but phosphorylates few other potential substrates.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cloning of Full-length Rat P-CIP2-- The 1.4-kb cDNA insert isolated from the original yeast two-hybrid screen (24) was ³²P-labeled by random priming (Prime It II, Stratagene) and used to screen a rat hippocampal random primed ZAPII cDNA library (Stratagene). Seven positive overlapping clones were isolated and purified from 6 × 10⁵ plaque forming units. The seven library cDNAs were rescued, the 5' ends were sequenced, and the start codon was identified by its Kozak sequence which matches 7 of 9 bases in an optimal translation initiation site (5'-GCCCGCACCATGG-3', mismatch is underlined) (28). The full sequence has been deposited with GenBank (accession number U70372).

Construction of Expression Vectors-- The 1.4-kb P-CIP2 partial cDNA fragment (I-2) encoding P-CIP2 residues 28-419 (NP-CIP2) (24) was ligated into pGEX5X2 to generate pGEX5X2/NP-CIP2 for the expression of GST/NP-CIP2 fusion protein to be used as an antigen. To construct a full-length P-CIP2 fusion protein expression vector, pGEX5X2/P-CIP2, the cDNA encoding P-CIP2 was modified to generate a SalI site in place of the start codon, and the 1.4-kb fragment was inserted into pGEX5X2 digested with SalI/NotI, yielding pGEX5X2/P-CIP2. Expression of the GST fusion protein with full-length P-CIP2 from pGEX5X2/P-CIP2 was carried out in BL21(DE3) cells (see below).

For mammalian expression, an optimal ribosomal binding site was engineered to precede the initiation codon of P-CIP2 in pBS.KrP-CIP2 (28, 29). Two mammalian expression vectors were used to express P-CIP2 and P-CIP2(1-383) (P-CIP2cc) with the Myc epitope and a His₆-tag at their COOH terminus (pcDNA3.1/mycHis; Invitrogen) or at the amino terminus (created by combining the epitope tag regions of pEAK10-His and pEAK10-Myc from Edge Biosystems, Gaithersburg, MD). Plasmid pBS.KrP-CIP2 was digested with XhoI/EcoRV and inserted into pcDNA3.1/mycHis digested with KpnI (blunted) and XhoI. For expression of P-CIP2cc, pBS.KrP-CIP2 was digested with XhoI/BsaBI and ligated. The full coding sequence of P-CIP2 was excised from pBluescript with NcoI-NotI and inserted into the pEAK10 vectors cut with BspLU11I-NotI. The Stratagene Quikchange site-directed mutagenesis kit was used to create pcDNA.KrP-CIP2cc/K54A using the sense primer 5'-GGCGCCCTCGCGCAGTTCCTGCCTCCGCC-3'. All PCR products and ligations were confirmed by DNA sequencing. The plasmids pcDNA3.1/-galactosidase-mycHis (Invitrogen) and pEGFP-N2 (CLONTECH) were used as transfection controls.

Northern Blotting-- Northern blots were performed as described previously (30) except that total RNA was extracted from adult Harlan Sprague-Dawley rat tissues with RNA Stat-60 (Tel Test). The SalI/NotI 1.4-kb fragment isolated in the original two-hybrid screen was random primed using the Prime-it II kit (Stratagene) with [-³²P]dCTP.

Generation of Antisera-- A glutathione S-transferase fusion protein with NP-CIP2 was expressed in Escherichia coli and purified by glutathione-Sepharose affinity chromatography (24) for use as an antigen to generate anti-P-CIP2 rabbit antisera JH1998 and JH1999 (Covance, Denver, PA). Generation of the rabbit polyclonal antibody JH2004 directed toward a 19-residue peptide near the COOH terminus of P-CIP2 was described previously (24).

In Situ Hybridization and Immunocytochemistry-- In situ hybridization for P-CIP2 and double labeling by in situ hybridization for P-CIP2 with immunocytochemical staining for PAM-1 or P-CIP2 were performed as described previously (24, 25, 31). After completion of the in situ washes, sections were washed in phosphate-buffered saline, blocked with 10% normal goat serum (30 min) and treated with 3% H₂O₂ in phosphate-buffered saline for 30 min. PAM antibodies (Ab JH1761 directed to the amino-terminal third of PAM or Ab 471 directed toward the middle third of PAM) were used at 1:1000 dilutions on sections and coverslipped with Parafilm. For P-CIP2 immunocytochemical staining, Ab JH1998 was used at a dilution of 1:1000. Sections were incubated at 4 °C overnight. Immunocytochemistry was performed using an avidin-biotin kit (Vector, Burlingame, CA) and visualized with horseradish peroxidase substrate. Sections were then dipped in photographic emulsion as described previously (24, 31).

Tissue Extract Preparation and Subcellular Fractionation-- Adult rat brain was separated into soluble and crude particulate fractions or subjected to differential centrifugation. Equal volumes of soluble and particulate fractions were subjected to Western blot analysis with Ab JH2004. For preparation of soluble and crude particulate fractions, tissues were homogenized in a 10-fold volume excess of 0.25 M sucrose in 20 mM TES, pH 7.0, containing 0.3 mg/ml PMSF and protease inhibitor mixture (29). The homogenate was centrifuged at 10,000 × g for 10 min and the pellet discarded; the supernatant was centrifuged at 372,000 × g for 15 min. The pellet was re-suspended in a volume equal to the supernatant (soluble fraction).

Differential centrifugation of parietal cortex was carried out according to Huttner et al. (32) and Bennett et al. (33). Tissue was homogenized in 10 volumes of SHEEP buffer (0.32 M sucrose, 4 mM HEPES-KOH, pH 7.5, 0.1 mM EDTA, 0.1 mM EGTA, 0.3 mg/ml PMSF). The homogenate was centrifuged at 400 × g for 10 min and the pellet was discarded. The supernatant was centrifuged at 800 × g for 10 min; the pellet was washed by re-suspension in the same volume of SHEEP buffer yielding the crude nuclear pellet (P1) and pooled supernatants (S1) that were centrifuged at 9200 × g for 10 min. The crude synaptosomal pellet was washed and resuspended in SHEEP buffer (P2). The pooled supernatants (S2) were centrifuged at 165,000 × g for 15 min; the microsomal pellet was washed as before and resuspended in SHEEP buffer (P3). The final pooled supernatant (S3) is the cytosolic fraction.

Co-immunoprecipitation-- Immunoprecipitations were carried out according to the method of Ratovitski et al. (34). Briefly, the culture medium was removed from a confluent 100-mm dish and the cells were rinsed with phosphate-buffered saline. Extraction buffer (20 mM PIPES buffer, 2 mM Na₂EDTA, 50 mM sodium fluoride, 10 mM Na₃P₂O₇, 1 mM NaV₃O₄, 1% Triton X-100, with 0.3 mg/ml PMSF and protease inhibitor mixture, adjusted to pH 6.8 with NaOH) was added to each dish (2.0 ml) on ice for 15 min. Cells were scraped from the dish, and debris was removed by high-speed centrifugation. Extracts (300 µl) were diluted with 300 µl of extraction buffer, 1 µl of PMSF (30 mg/ml), 5 µl of inhibitor mixture, and 10 µl of P-CIP2 antibody (JH2004 or JH1998) or 10 µl of P-CIP2 antibody pre-blocked for 1 h on ice by incubation with 50 µg of P-CIP2 peptide (PKENPGRGQVFVEYANAGD, underlined in Fig. 1) for JH2004 or 50 µg of GST/P-CIP2 for JH1998. Antibody-extract mixtures were incubated on ice for 1 h, followed by addition of Protein A-agarose (Sigma) and agitation on ice for 1 h. Protein A-agarose beads were washed and boiled into loading buffer for separation by SDS-PAGE (10% polyacrylamide gel). Western blots used antibodies to PAM (Ab 629, 1:2000 dilution, or monoclonal antibody 6E6, 1:20 dilution of medium (35)) and were visualized using the ECL kit (Amersham Pharmacia Biotech).

Tissue Culture and Expression of Proteins in Bacterial and Mammalian Cells-- For mammalian expression of P-CIP2 or its mutants, AtT-20, COS, hEK-293, and pEAK-Rapid (Edge Biosystems) cells were cultured in Dulbecco's modified Eagle's medium/F-12 medium containing 10% NuSerum (omitted for pEAK-Rapid cells) (Collaborative Research, Bedford, MA) and 10% fetal clone serum (Hyclone, Logan, UT) and plated in 75-cm² flasks 1-2 days prior to transfection. Stable cell lines were created as described previously (21), and selected by immunofluorescent staining and Western blotting with myc monoclonal antibody 9E10 (36). Transient transfections were performed using LipofectAMINE or LipofectAMINE Plus (Life Technologies, Inc.). Cells were plated on glass slides and immunostained as described (21, 24) using a Zeiss Axioskop equipped with a Princeton Instruments Micromax digital camera. GST fusion proteins were purified from bacterial cell lysates by affinity chromatography on glutathione-Sepharose (Amersham Pharmacia Biotech) and eluted with glutathione, which was removed by dialysis.

In Vitro Kinase Assays-- Mammalian cells stably transfected with vectors encoding P-CIP2mH, P-CIP2ccmH, or -galactosidase-mH were scraped from culture dishes and extracted with 20 mM Na-TES/10 mM mannitol, pH 7.4, containing PMSF (30 mg/ml) and inhibitor mixture, followed by three cycles of freezing and thawing. Extracts from mammalian cell pellets were diluted with an equal volume of 2× metal chelate resin binding buffer (1× is 10 mM imidazole, 150 mM NaCl, 50 mM Tris, pH 7.4). The diluted extracts were applied batchwise to His-bind metal affinity resin (Novagen) charged with Ni²⁺, or to Ni²⁺-NTA-agarose resin (Qiagen). Samples were incubated with resin for 30-60 min at 4 °C with gentle mixing, washed three times with 40 mM imidazole, 0.5 M NaCl, 50 mM Tris-HCl, pH 7.4, and twice with 5 mM MgCl₂ in Na-HEPES, pH 7.4. P-CIP2 was eluted with 1 M imidazole and required at least a 10-fold dilution before it was assayed.

pEAK-Rapid cells transiently transfected with the pEAK10 vector were extracted into isotonic sucrose buffer containing 20 mM Na-HEPES, pH 7.4, with protease inhibitors, centrifuged at 430,000 × g for 15 min and the pellet was re-extracted with 50 mM HEPES, 1 mM MgCl₂, pH 7.5, containing 1% Thesit and protease inhibitors. Most of the exogenous P-CIP2 was recovered in the pellet fraction following transient expression. P-CIP2 was purified from pEAK-Rapid extracts (100 µg of protein/µg of resin) using the His-bind resin, Talon resin (CLONTECH), or the Ni-NTA resin according to the manufacturers' instructions, and eluates were desalted into 50 mM HEPES, 1 mM MgCl₂, pH 7.5, containing 0.1% Thesit using NICK columns (Amersham Pharmacia Biotech).

Enzyme was added to the reaction mixture (40 or 50 µl) containing recombinant substrate (5 or 10 µg; PAM-CD, GST-PAM-CD, other GST fusion proteins); kinase buffer (50 mM HEPES, 1 mM MgCl₂, pH 7.5, 0.1% Thesit) containing 1 or 2 µCi of [-³²P]ATP (3000 Ci/mmol, Amersham Pharmacia Biotech) was added to each sample. Reactions were incubated at 37 °C for 30 or 60 min. Positive control reactions utilized protein kinase C purified from rat brain (kindly provided by Carol Doherty and Dr. Richard Huganir, Howard Hughes Medical Institutes/Johns Hopkins University School of Medicine), bovine heart protein kinase A (Sigma), or recombinant human casein kinase II (Calbiochem). Reaction supernatants from stable transfections were precipitated with 20 µg of bovine serum albumin and 15% trichloroacetic acid, washed with acetone, boiled into Laemmli loading buffer, loaded onto 15% SDS-PAGE gels, and visualized by autoradiography. For transient expression in pEAK-Rapid cells, the reaction mixtures were loaded directly onto SDS-PAGE peptide gels (37), transferred to nylon membranes, and visualized by autoradiography, direct staining, or Western analyses. Gels were stopped before the dye ran off and a 3-cm region behind the dye was cut off before transfer, to remove the free [-³²P]ATP (38). Phosphoamino acid analysis of phosphorylated products was performed as described (39) using sequential pH 1.9 and pH 3.5 thin layer electrophoresis.

P-CIP2 and commercial kinases were combined with 5 µg of synthetic peptide and 1 µCi of [-³²P]ATP in Hepes-Mg-Thesit for 30 min at 37 ^oC, followed by addition of 5 µg of Trp-Met-Asp-Phe-NH₂ and 100 µg of bovine serum albumin. PAM peptides were synthesized by Dr. Henry Keutmann, Massachusetts General Hospital (24, 39); standard kinase substrate peptides were from Calbiochem. Samples were left on ice for 30 min, centrifuged, and aliquots of the supernatant spotted onto P81 ion exchange paper (Whatman) prewetted with 1% phosphoric acid. One-eighth of each reaction was analyzed; samples were analyzed in duplicate. Samples were washed repeatedly in 1% phosphoric acid, finally rinsed in acetone, air dried, and submitted to Cerenkov counting (38, 40). Data are reported as mean ± S.E.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

P-CIP2: A Putative Ser/Thr Protein Kinase with a RNA Binding Motif-- We previously used the yeast two-hybrid system to identify proteins that interact with the COOH-terminal domain (CD) of PAM-1 (24). One of these PAM-COOH-terminal interactor proteins (P-CIPs) contained an almost complete putative protein kinase domain followed by a putative RNA binding motif and a stop codon. Comparison of the predicted molecular mass (44 kDa) with the band observed in Western blots of endogenous P-CIP2 in AtT-20 corticotrope tumor cell extracts (47 kDa) indicated that the complete cDNA would encode additional residues at the amino terminus.

Subsequent screening of a random primed rat hippocampal cDNA library identified an additional 27 amino acids preceding the known P-CIP2 sequence. Thus, the full-length P-CIP2 cDNA encodes a 419-residue protein with a predicted mass of 46.5 kDa. Consistent with this, in vitro translation of full-length clones of P-CIP2 yielded a 46.5-kDa product (data not shown) and the protein observed in AtT-20 cell extracts by Western blot analysis has an apparent molecular mass (M_r) of 47 kDa (24).

Fig. 1 shows an alignment of the amino acid sequence of P-CIP2 with the most homologous protein kinases according to BLASTP analysis. The kinases shown, most of which are Ser/Thr protein kinases, share 22-30% identity with P-CIP2 within the catalytic core of 250 residues. The greatest sequence variation among these kinases occurs in subdomains X and XI and following subdomain XI. While the sequence Asp¹⁵⁸-Leu-Lys-Pro-Arg-Asn in subdomain VI suggests that P-CIP2 belongs to the serine/threonine protein kinase family, P-CIP2 also shares moderate homology with the dual specificity (Ser/Thr and Tyr) kinase PYT and has some residues normally considered diagnostic for Tyr kinases (40). The non-conserved residues presumably affect the substrate specificity of P-CIP2, which is remarkable (see below).

View larger version (96K): [in this window] [in a new window]   Fig. 1.   P-CIP2 sequence: comparison of kinase catalytic domains. The amino acid sequence of rat P-CIP2 (top sequence) is shown aligned with protein kinases found to be the most homologous to P-CIP2 according to a BLASTP homology search. Multiple sequence alignment was done using the PIMA algorithm (71, 72). Dark gray shading indicates identical amino acid residues; conservative substitutions are shown in light gray shading according to the rules defined by the PIMA algorithm (72). Asterisks (*) indicate residues that are required for Ser/Thr kinase activity. Roman numerals indicate subdomains of the kinase catalytic core (40, 73). Phosphorylation sites for activation of the MAP kinase hog1 are circled (-T-X-Y-) in the activation loop of subdomain VIII. A solid arrow indicates the beginning of the RNA-binding domain (RNA-BD); a dashed arrow indicates the end of the truncated kinase domain (P-CIP2cc). The underlined sequence indicates the peptide antigen used to generate Ab JH2004. The kinases shown include the yeast glycogen synthase kinase Skp1 (accession number L29449), the cdc-related protein kinase CRK1 (accession number X60385), the yeast MAP kinase hog1 (accession number P32485), the phosphotyrosine picked threonine protein kinase PYT (accession number I38144), and the catalytic -subunit of casein kinase II from rat (accession number P19139). The sequence surrounded by the dashed box corresponds to a human expressed tag sequence (accession number AA452389) that is nearly identical to rat P-CIP2.

BLASTP comparison of P-CIP2 versus the Saccharomyces cerevisiae Genome Data base showed that the most homologous yeast protein kinases are members of the MAP/ERK kinase family including SMK1, HOG1 (shown in Fig. 1), SLT2, KSS1, and FUS3; the most similar yeast kinase, HOG1, shows 14% identity and 30% similarity within the catalytic core. These yeast MAP kinases are involved in the mating pheromone response pathway (41, 42), suggesting that P-CIP2 might be involved in signal transduction. MAP kinases have two closely spaced phosphorylation sites (T-X-Y, circled in Fig. 1 for the HOG1 sequence) in subdomain VIII whose phosphorylation is required for activation of the kinase (43). All of the kinases shown, including P-CIP2, share a Thr at the appropriate position in subdomain VIII, whereas only HOG1 contains both sites. Rat CKII shows as much similarity to P-CIP2 as any kinase, which is interesting considering the substrate specificity of P-CIP2 (see below).

As noted previously, the COOH-terminal region of P-CIP2 following subdomain XI shows significant homology with the third ribonucleoprotein consensus sequence (RNCS) of pre-mRNA splicing factor U2AF⁶⁵ (24). A human expressed sequence tag matches 59 of 60 residues following subdomain XI into the ribonucleoprotein consensus sequence region (Fig. 1, dashed box) with the only mismatched residue substituting an aspartate for glutamate. Only one other kinase is known to contain an RNA-binding domain, PKR (for review, see Ref. 44).

P-CIP2 Is a Soluble Protein Found in Nervous and Endocrine Tissues That Contain PAM-- Northern blotting revealed the presence of a single 9-kb transcript in every region of the brain examined (Fig. 2A). PAM transcripts are known to be present in all of the brain tissues tested (45). P-CIP2 mRNA was also found at high levels in anterior and neurointermediate pituitary, adrenal, kidney, and spleen and at lower levels in heart atrium, ventricle, duodenum, liver, and skeletal muscle. Only about 1.5 kb of the 9-kb message encodes protein, and sequence analysis shows that the P-CIP2 transcript contains at least 2 kb of 3'-untranslated sequence. Transcripts with large 3'-untranslated regions are often found in neuroendocrine tissues and the 3'-untranslated regions have been suggested to play a role in the regulation of translation (46, 47).

View larger version (59K): [in this window] [in a new window]   Fig. 2.   A, tissue Northern blot for P-CIP2 mRNA. Total RNA (10 µg, except 2 µg for NIL) from the indicated adult rat tissues was fractionated on a denaturing agarose gel, transferred to a nylon membrane, and hybridized with P-CIP2 cDNA probe (SalI/NotI fragment, 1.4 kb). The ethidium bromide-stained 18 S ribosomal RNA pattern is shown in the lower panel. Similar results were obtained in two additional analyses. B, tissue Western blot for P-CIP2 protein. Soluble (S) and crude particulate (M) fractions prepared from isotonic homogenates of the indicated tissues were subjected to Western blot analysis and visualized with Ab JH2004. In multiple analyses of different liver homogenates, P-CIP2 protein was detectable in liver cytosol only after prolonged exposure. C, subcellular localization of P-CIP2 protein. Parietal cortex was subjected to differential centrifugation, yielding crude nuclear (P1), synaptosomal (P2), and microsomal (P3), and cytosolic fractions (S3). The supernatant from the nuclear pellet (S1) was used to prepare the synaptosomal pellet; the supernatant from the synaptosomal pellet (S2) was used to prepare microsomal pellet (P3). Samples were separated by SDS-PAGE and subjected to Western blot analysis as above. Similar results were obtained in duplicate experiments. Ob, olfactory bulb; Pc, parietal cortex; Hi, hippocampus; Hy, hypothalamus; AP, anterior pituitary; Cb, cerebellum; Ni, neurointermediate pituitary; At, atrium; Ve, ventricle; Ad, adrenal; Ki, kidney; Du, duodenum; Li, liver; Sk, skeletal muscle; Sp, spleen; BS, brain stem.

Expression of P-CIP2 protein in different tissues was compared using Western blotting to examine crude particulate and soluble fractions. Western blot analysis of isotonic extracts from various brain tissues with similar P-CIP2 mRNA levels show that P-CIP2 protein is present at similar levels, and primarily in the soluble fraction (Fig. 2B). Equal amounts of the pellet fractions contain less immunoreactive protein than the soluble fractions. This distribution of P-CIP2 protein is consistent with a cytosolic protein that can interact with membrane proteins. In contrast, although P-CIP2 mRNA was found in the liver (Fig. 2A), relatively little immunoreactive protein was found in soluble or particulate fractions prepared from liver. Longer exposures of liver samples, as well as kidney and spleen extracts, do reveal low levels of P-CIP2 protein (data not shown). Tissue-specific protein stabilization and altered turnover rates have been observed in other cases (48) and may contribute to the lack of P-CIP2 protein in liver, despite the P-CIP2 mRNA visualized in liver total RNA.

The distribution of P-CIP2 protein in subcellular fractions prepared by differential centrifugation was analyzed (Fig. 2C). Subcellular fractionation of parietal cortex showed that most of the P-CIP2 protein resides in the cytosolic fraction (S3). P-CIP2 protein is also associated with the crude synaptosomal fraction (P2); upon hypoosmostic lysis of this fraction, most of the P-CIP2 was recovered in the soluble fraction (not shown). A small but significant amount of P-CIP2 protein is recovered in the microsomal pellet (P3). Despite the presence of a putative RNA binding motif, negligible amounts of P-CIP2 protein were found in the crude nuclear pellet (P1).

In situ hybridization also shows that P-CIP2 mRNA is widely expressed in the rat brain (Fig. 3). Low power magnification of coronal sections from rat brain shows a strong positive signal for P-CIP2 mRNA in all areas of the hippocampus (CA1, CA2, CA3, and dentate gyrus), amygdala, habenula, hypothalamus (paraventricular, dorsomedial, arcuate, and supraoptic nuclei), substantia nigra, ventral tegmental nucleus, and in sensory and motor nuclei of the brain stem (Fig. 3, F and G). Lower levels of P-CIP2 mRNA were found in cortex and thalamus. Hybridization using sense riboprobes showed no discernible signal in the hippocampus or brain stem (Fig. 3B) where antisense probe demonstrated the highest signal levels.

View larger version (128K): [in this window] [in a new window]   Fig. 3.   P-CIP2 localization in the brain. A-G, coronal sections of rat brain from forebrain (A) to brain stem (G) showing location of P-CIP2 mRNA (in situ hybridization) using antisense riboprobes (A, C, D, E, F, and G); sense riboprobe showed no discernible signal (B) in a coronal section similar to A. Abbreviations: paraventricular nuclei (PVN), supraoptic nuclei (SON), dorsomedial nuclei, arcuate nuclei, ventral tegmental area (VTA), substantia nigra compacta (SNc), nucleus Darkschewitsch (DK), various brain stem cranial nerve (CN) nuclei, superior olive, and reticular nuclei. H, P-CIP2 immunocytochemistry (ICC) in neurons of the medulla oblongata. I and J show P-CIP2 mRNA (silver grains) in neurons (mesencephalic division of CN5) that immunostain for P-CIP2 (I) or PAM (J) protein. Bar represents 50 µm for H, I, and J.

Antisera to P-CIP2 were used for immunocytochemistry. Neurons in the lateral reticular nucleus of the medulla oblongata that were visualized by the P-CIP2 antiserum are shown in Fig. 3H; staining could be blocked by preincubation of the antiserum with affinity purified GST-P-CIP2 protein (not shown). P-CIP2 staining filled the cell soma, with little staining extending into neurites; nuclei were not stained. Immunocytochemistry and in situ hybridization were combined to show that the vast majority of the neurons that contained P-CIP2 transcripts could also be immunostained for P-CIP2 (Fig. 3I). To determine whether P-CIP2 colocalizes with PAM, a prerequisite if physiologically significant interactions are to occur, immunocytochemical identification of neurons expressing PAM protein was combined with in situ hybridization of P-CIP2 transcripts (Fig. 3J); neurons expressing PAM protein were frequently found to express P-CIP2, as in the example from the medulla oblongata. Spinal cord motor neurons were examined in a similar manner and were also shown to express P-CIP2 RNA and PAM protein (not shown).

P-CIP2 Is an Active Protein Kinase That Interacts with Itself-- The kinase activity of P-CIP2 was examined in vitro using several forms of recombinant protein. P-CIP2 containing the myc epitope and a polyhistidine (His₆) tag at its COOH terminus (P-CIP2mH) was expressed transiently in COS cells, purified using metal chelate affinity resin, and tested for kinase activity using recombinant PAM CD as substrate (39) (Fig. 4, A and B). Controls included COS cells expressing -galactosidase bearing the same epitope tag (-GalmH) or non-transfected cells (WT COS). P-CIP2mH was expressed, but at relatively low levels (Fig. 4A). The data show that recombinant PAM CD was phosphorylated by P-CIP2mH, but not by samples affinity purified from non-transfected cells or cells expressing -GalmH (Fig. 4B).

View larger version (60K): [in this window] [in a new window]   Fig. 4.   P-CIP2 phosphorylates PAM-CD. A, confluent 100-mm plates of COS cells were transfected transiently with vector encoding PCIP2mH or similarly tagged -galactosidase, -GalmH, or not transfected (WT). Cells were extracted and the tagged proteins purified by metal affinity chromatography. Bound proteins were eluted and subjected to Western blot analysis for the Myc epitope. Arrows indicate expressed epitope tagged protein. B, samples from panel A were diluted into kinase buffer containing 10 µg of purified recombinant PAM CD (rCD), and incubated at 37 °C for 1 h. Kinase reaction products were separated on a 15% SDS-PAGE gel, and visualized by autoradiography. C, AtT-20 cells were stably transfected with vector encoding PCIP2mH or P-CIP2ccmH. Expressed P-CIP2 proteins were purified by metal affinity chromatography. Purified imidazole eluted protein was diluted into kinase buffer with 10 µg of purified recombinant PAM CD, incubated at 37 °C for 1 h, separated by SDS-PAGE, and visualized by autoradiography. D, 32P-labeled recombinant CD bands from panel C were excised from the SDS-PAGE gel and subjected to acid hydrolysis followed by thin layer electrophoresis with the indicated phosphoamino acid standards (39, 57). Standards were visualized with ninhydrin spray, and samples were visualized by autoradiography.

Stably transfected AtT-20 cells expressing P-CIP2mH were generated in an attempt to improve expression levels. In addition, P-CIP2 in which the myc-His₆ tag replaced the putative RNA-binding domain (P-CIP2ccmH) was expressed to determine whether the COOH-terminal domain were essential for the kinase activity of P-CIP2 (Fig. 4C). Extracts of AtT-20 cells expressing PCIP2mH or PCIP2ccmH phosphorylated recombinant PAM CD while extracts of control cells yielded very little activity. The preparations of P-CIP2 from transfected mammalian cells showed negligible autophosphorylation. Phosphoamino acid analysis of recombinant PAM CD phosphorylated by the expressed P-CIP2 proteins demonstrated that phosphorylation was specific for serine (Fig. 4D). Two-dimensional analyses also showed specificity for serine, using P-CIP2 transiently expressed in pEAK-Rapid cells (not shown).

Although stable cell lines expressing P-CIP2mH were established, attempts to improve expression by subcloning lines suggested that higher level expression of P-CIP2 was toxic. We therefore attempted to gain better expression of P-CIP2 using a different transient transfection system. The epitope tag, His₆-myc, was appended to the NH₂-terminal end of P-CIP2 or P-CIP2cc bearing an ATP-binding site mutation (K54A) that should inactivate the kinase. Both proteins were expressed at high levels in pEAK-Rapid hEK-293 cells. The transfected HmP-CIP2 protein was present at a high level and much of it was recovered in the isotonic sucrose pellet; in contrast, the small amount of endogenous P-CIP2 was recovered primarily in the soluble fraction (Fig. 5A). Immunostainable HmP-CIP2 expression was readily detected in over half of the transiently transfected cells. Staining was observed in aggregates which were especially visible in cells which had rounded up and appeared to be detaching from the culture dish (not shown). By comparison, we could not reliably detect stable P-CIP2 expression using the same reagents.

View larger version (51K): [in this window] [in a new window]   Fig. 5.   Purification of P-CIP2. A, a control protein (green fluorescent protein (GFP)) and HmP-CIP2 were transiently expressed in pEAK-Rapid hEK-293 cells, homogenized in isotonic sucrose, and the supernatant (S) and pellet (P) were subjected to Western analysis using the GST-P-CIP2 fusion protein antibody (JH1998); the endogenous P-CIP2-reactive band is indicated (*). B, pEAK-Rapid cells were transiently transfected with pEAK10 vectors encoding HmP-CIP2 (full) or HmP-CIP2cc-K/A (cc-K/A) and the isotonic sucrose pellet was extracted and purified using the Talon metal-chelate protocol. Samples eluted in imidazole were desalted and subjected to Western analysis using the myc monoclonal antibody or the P-CIP2 antibody (JH1998). The small amount of endogenous P-CIP2 that copurified with HmP-CIP2cc-K/A is indicated by an asterisk (*). C, a liquid phase -galactosidase assay was used to quantify the interaction of N-P-CIP2 with itself in the yeast two-hybrid system (24); N-P-CIP-2 was expressed as bait or as both bait and prey. D, samples were prepared as in B, except that the Ni-NTA resin was used and non-transfected cells were analyzed as a control; cells transfected with GFP or -galactosidase vector showed the same pattern as for the NT cells (not shown). Equal portions of all the samples were analyzed; much of the input HmP-CIP2 was bound to and eluted from the Ni-NTA resin, while less was found in the resin flow-through (FT) and washes before elution, and very little remained on the resin after elution. Similar results were obtained in 4 additional experiments of this type.

Both HmP-CIP2 and HmP-CIP2cc-K/A were purified from solubilized sucrose pellets in good yield using a metal-chelate resin (Fig. 5B). Our initial kinase assays unexpectedly indicated that HmP-CIP2cc-K/A had some activity (data not shown). Although the problem of spurious binding of kinase(s) to resins has been reported many times before (38), neither nontransfected cells nor GFP-transfected cells yielded background kinase activity. An explanation for this observation is suggested by the fact that P-CIP2 interacted with itself in the yeast two-hybrid assay to produce yeast growth and -galactosidase activity (Fig. 5C). Visualization of purified HmP-CIP2 and HmP-CIP2cc-K/A using a P-CIP2 antibody instead of the myc antibody identified a small amount of endogenous P-CIP2 in the purified HmP-CIP2cc-K/A (Fig. 5B, asterisk). To eliminate the co-purification of endogenous P-CIP2, a different brand of metal chelate affinity resin was utilized (Fig. 5D). A better purification of HmP-CIP2 (Fig. 5D) was achieved, with no endogenous P-CIP2 detectable. Bacterially expressed P-CIP2 showed very low activity (<1% as much activity as for an equivalent Western blot signal from pEAK-Rapid cells) and was not used in subsequent work.

P-CIP2 Phosphorylates the Region of the PAM Cytosolic Domain Involved in Trafficking-- The ability of purified mammalian recombinant HmP-CIP2 to phosphorylate recombinant PAM-CD and GST-PAM-CDt was compared (Fig. 6A). In GST-PAM-CDt, the PAM protein is truncated at residue 961, deleting the final 15 residues of the cytosolic domain; PAM proteins truncated at this position are known to show normal trafficking in AtT-20 cells (21). P-CIP2 was found to phosphorylate PAM-CD and GST-PAM-CDt to a similar extent. The GST control was not phosphorylated by P-CIP2. The nontransfected cells showed no activity in this assay (Fig. 6) and extracts of pEAK-Rapid cells transfected with GFP vectors were inactive (not shown). Autophosphorylation of the recombinant HmP-CIP2 (at 47 kDa) was not detectable.

View larger version (29K): [in this window] [in a new window]   Fig. 6.   P-CIP2 selectively phosphorylates GST-PAM-CD fusion proteins. Samples from pEAK-Rapid cells expressing HmP-CIP2 (upper) or not transfected (NT; middle) were analyzed for kinase activity. The proteins on the NT gel were visualized with Coomassie Brilliant Blue (Stain; lower). A, phosphorylation of purified PAM-CD, GST control, and GST-PAM-CD was tested, using recombinant proteins. B, phosphorylation of truncated PAM-CD-936s, mutant PAM-CD-K919R, and the three GST Rho-family fusion proteins was also tested. Similar results were obtained in five additional experiments of this type.

Truncation of PAM at residue 936 eliminates all cytosolic trafficking determinants (21), and P-CIP2 was unable to phosphorylate the corresponding GST fusion protein (GST-CD-936s) (Fig. 6B). Mutation of Lys⁹¹⁹ to Arg was shown to eliminate the ability of NP-CIP2 to interact with PAM-CD in the yeast two-hybrid system.² Nevertheless, P-CIP2 phosphorylated the K919R version of CDt. Since Kalirin, another PAM-CD interactor, binds Rac1, the ability of P-CIP2 to phosphorylate members of the Rho subfamily of small GTP-binding proteins was determined (Fig. 6). GST fusion proteins consisting of Rac1, RhoA, and Cdc42 were not detectably phosphorylated by P-CIP2; these GST fusion proteins all functioned as substrates for protein kinase A, protein kinase C, or casein kinase II (not shown). Thus P-CIP2 exhibits specificity for the region of the PAM cytosolic domain involved in trafficking.

P-CIP2 Shows a Strong Preference for Ser⁹⁴⁹ in the PAM-CD-- The substrate specificity of P-CIP2 was tested using GST fusion proteins consisting of the cytosolic domains of two other membrane proteins localized to the TGN region (Fig. 7). Like PAM, the endoprotease furin and the exoprotease carboxypeptidase D are transmembrane proteins with their catalytic regions in the lumen of the TGN, and their carboxyl-terminal regions extending across the membrane and out into the cytosol (8, 49). P-CIP2 did not efficiently phosphorylate the cytosolic domain of either furin or carboxypeptidase D (Fig. 7A). In contrast, the furin and carboxypeptidase D cytosolic domains were substrates for protein kinase A and/or casein kinase II (Fig. 7B) and protein kinase C (not shown).

View larger version (39K): [in this window] [in a new window]   Fig. 7.   P-CIP2 is highly selective for PAM-CD. A, kinase assays were carried out as described in the legend to Fig. 6 using purified HmP-CIP2 and the indicated GST fusion proteins. Expression of GST-fusion proteins was compared by staining the membrane with Coomassie. B, the ability of bovine heart protein kinase A (PKA) and human recombinant casein kinase II -subunit (CKII) to phosphorylate the same GST fusion proteins was compared. C, HmP-CIP2 was used to phosphorylate the indicated GST fusion proteins; the film was deliberately overexposed to determine the level of label in the TS/DD (Thr946-Asp/Ser949-Asp) mutant. Similar results were obtained in five additional experiments of this type.

P-CIP2 was also isolated independently as an interactor protein for P19/stathmin, a small cytosolic phosphoprotein involved in microtubule destabilization (26). SCG10 is a neuronal homolog of stathmin with similar microtubule destabilizing activity (27). We used GST fusion proteins to evaluate the ability of P-CIP2 to phosphorylate P19 and SCG10. Purified mammalian recombinant P-CIP2 did not phosphorylate SCG10 or P19 detectably (Fig. 7A); in contrast, both GST fusion proteins were good substrates for protein kinase A (Fig. 7B). Bacterially expressed GST-P-CIP2 did show some labeling of P19 and SCG10, but only at levels of P-CIP2 protein at least 100-fold higher than the levels used in these assays (data not shown). Other substrates tested included the Exon A region of PAM, casein and histone; none was significantly phosphorylated by P-CIP2 (not shown). Thus, P-CIP2 is highly selective for the cytosolic domain of PAM.

The protein kinase A, protein kinase C, and casein kinase II sites in recombinant PAM CD were identified previously (Fig. 8) (39). GST fusion proteins in which each identified Ser residue was individually mutated to Asp were evaluated as substrates for P-CIP2 (Fig. 7C). Elimination of the protein kinase C sites (Ser⁹³² and Ser⁹³⁷) or the protein kinase A site (Ser⁹²¹) did not affect the ability of P-CIP2 to phosphorylate CD. In contrast, mutation of the sites phosphorylated by sea star casein kinase II (Thr⁹⁴⁶ and Ser⁹⁴⁹) greatly reduced the ability of P-CIP2 to phosphorylate CD. The S937A, S932A, and S921A mutants were also good substrates (data not shown). Since it was established that P-CIP2 primarily phosphorylates Ser residues (Fig. 4D), these data implicate Ser⁹⁴⁹ as the primary target of P-CIP2 phosphorylation. Importantly, the T946D,S949D mutant of PAM (mimicking permanent phosphorylation of the Thr⁹⁴⁶ and Ser⁹⁴⁹ residues) was not trafficked normally in AtT-20 cells (50).

View larger version (22K): [in this window] [in a new window]   Fig. 8.   P-CIP2 phosphorylates Ser949 of PAM. The ability of HmP-CIP2 (filled bars) and recombinant human casein kinase II (open bars) to phosphorylate several synthetic peptides (5 µg) derived from the CD of PAM were compared. The location of each peptide within the 86-amino acid cytosolic domain of PAM is indicated, as are the previously determined PKA, PKC, and sea star CKII sites (39). The protein kinase A substrate and PAM(932-948) were well phosphorylated by protein kinase A; the protein kinase C substrate was phosphorylated by protein kinase C (not shown). Samples were analyzed in duplicate and data are mean ± S.E.; all the results were repeated three or four times with similar outcomes.

To further evaluate the specificity of P-CIP2, synthetic peptides containing various potential phosphorylation sites were tested as substrates (Fig. 8). P-CIP2 phosphorylated PAM(942-953). In contrast, the same synthetic peptide containing phospho-Thr⁹⁴⁶ and phospho-Ser⁹⁴⁹ was not phosphorylated by P-CIP2. PAM(945-961), a peptide in which Ser⁹⁴⁹ is closer to the NH₂ terminus, was not phosphorylated by P-CIP2. The other PAM peptides tested did not serve as substrates for P-CIP2, nor did peptides known to serve as substrates for casein kinase I, protein kinase A, protein kinase C, or calmodulin kinase II (Fig. 8). Although Thr⁹⁴⁶ and Ser⁹⁴⁹ are phosphorylation sites for sea star casein kinase II, they are not the preferred sites for human recombinant casein kinase II. PAM(942-953)/phospho-Thr⁹⁴⁶/phospho-Ser⁹⁴⁹, which contains only one potential acceptor residue at Ser⁹⁴⁵, was efficiently phosphorylated by human recombinant casein kinase II. Loss of phosphorylation by mutagenesis of the putative acceptor site provides strong evidence that the actual site of P-CIP2 phosphorylation is Ser⁹⁴⁹ (51, 52).

Interaction of P-CIP2 with PAM-CD Requires the Correct Routing Determinants-- The yeast two-hybrid screen in which P-CIP2 was identified demonstrated that P-CIP2 recognized PAM-CD truncated at residue 961 but failed to recognize PAM-CD lacking an 18-amino acid motif known to contain key routing determinants (928-945) (24). We used cell extracts prepared from AtT-20 cells expressing wild type membrane PAM (PAM-1) or trafficking deficient mutant membrane PAM proteins to determine whether the ability of PAM to bind to P-CIP-2 correlated with its trafficking ability. Extracts of AtT-20 cells expressing PAM-1, PAM-1/936s, or PAM-1/933-950 were passed over glutathione-Sepharose to which GST/NP-CIP2 was bound (Fig. 9). Control resin with only GST bound showed negligible binding of PAM-1, whereas a significant fraction of the PAM-1 protein bound to the GST/NP-CIP2 resin. PAM-1 truncated at Tyr⁹³⁶ is largely localized to the plasma membrane of transfected AtT-20 cells (21) and showed no detectable binding to the GST/NP-CIP2 resin. PAM-1/933-950, which is also mistargeted (21), did not bind to the GST/NP-CIP2 resin.

View larger version (24K): [in this window] [in a new window]   Fig. 9.   P-CIP2 Binds PAM: in vitro binding of PAM CD mutants to GST/P-CIP2. Extracts of AtT-20 cells expressing wild type or mutant PAM-1 protein were incubated with GST/NP-CIP2 bound to glutathione-Sepharose. GST-glutathione-Sepharose was used as a control. Proteins eluted from the resin were subjected to Western blot analysis and visualized with Ab JH629 directed toward exon A of PAM. Equal proportions of the Input (In), flow-through (FT), and bound (B) fractions were analyzed. The experiment was repeated three times with similar results.

In order to determine if P-CIP2 interacts directly with PAM in cells, co-immunoprecipitation experiments were carried out on non-transfected AtT-20 cells and AtT-20 cells stably expressing various PAM proteins (Fig. 10). The transfected 120-kDa PAM-1 protein is cleaved in a post-TGN compartment by neuroendocrine-specific endoproteases into a soluble peptidylglycine--hydroxylating monooxygenase domain fragment (not shown) and a 70-kDa fragment consisting of PAL (peptidyl--hydroxyglycine--amidating lyase domain) and the transmembrane and cytosolic domains (PALm) (21). Aliquots corresponding to 10 times the amount of input shown were immunoprecipitated with a P-CIP2 antibody; Western blots probed with a PAM antibody revealed intact PAM-1 and PALm in extracts from cells transfected with wild type PAM-1 (wt), but not in nontransfected cells (NT). AtT-20 cells contain a small amount of endogenous PAM, and the endogenous protein can be detected when large amounts of sample are analyzed (Fig. 10, middle, *). Importantly, immunoprecipitations in which the P-CIP2 antibody had been pre-blocked with P-CIP2 bacterial antigen did not show any PAM immunoreactive bands (Fig. 10, right). Co-immunoprecipitation of 70-kDa PALm indicates that P-CIP2 is interacting with PAM proteins that have progressed past the trans-Golgi network. Similar results were obtained using a different P-CIP2 antibody for immunoprecipitation, and with different PAM antibodies for visualization. Since the input sample shown was one-tenth of the cell extract applied to the P-CIP2 antibody, the PAM proteins recovered in the P-CIP2 co-immunoprecipitate represent only a small percentage of the total PAM proteins in the cells.

View larger version (52K): [in this window] [in a new window]   Fig. 10.   PAM co-immunoprecipitates with P-CIP2. Non-transfected AtT-20 cells (NT) or cells transfected with PAM-1 (non-mutated = wt; 933-950; 899t; 928-945) were extracted and immunoprecipitated with P-CIP2 antibody, JH1998. One sample of JH1998 was also pre-blocked with GST/P-CIP2 protein prior to adding cell extract. Western blots were visualized using PAM Ab JH629. Input samples show Western blots of one-tenth the amount of cell extracts used for the co-immunoprecipitation analyses. Similar results were obtained probing Western blots with PAM CD monoclonal antibody 6E6, using cell extracts made with a high salt buffer, and with the anti-peptide directed antibody JH2004 for the initial immunoprecipitation step. Full-length 120-kDa PAM and the endoproteolytic cleavage product 70-kDa membrane PAL (PALm) are indicated. Endogenous AtT-20 PAM is indicated (*).

Three PAM-1 routing mutants were also examined. PAM-1/899t, truncated immediately after the stop transfer signal has a very short cytosolic domain and is largely localized to the plasma membrane in AtT-20 cells. Essentially no binding of PAM-1/899t to P-CIP2 was detected (Fig. 10). Two PAM-1 internal deletion mutants were examined: PAM-1/928-945 and PAM-1/933-950. A significant amount of interaction was detected for PAM-1/933-950, but not for PAM-1/928-945. The inability of PAM-1/928-945 to interact with P-CIP2 is consistent with the results of the initial yeast two-hybrid screen (24). The ability of PAM-1/933-950, which lacks the site phosphorylated by P-CIP2, to interact with P-CIP2 indicates that some determinants governing interaction are distinct from those governing phosphorylation and proximal to residue 933.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Proper targeting of proteins within the secretory pathway is essential for normal function. Complex arrays of overlapping and often redundant sorting signals have been identified in the cytosolic domains of many membrane proteins. While the importance of Tyr motifs, di-Leu motifs, and phosphorylation sites has been established for a number of membrane proteins (1-3, 8, 13, 21), we still know relatively little about the proteins that interact with these motifs. The COOH-terminal cytosolic domain of membrane PAM contains targeting signals that include a tyrosine-based internalization motif and phosphorylation sites. In searching for cytosolic proteins that could interact with membrane PAM, we anticipated finding coat proteins, cytoskeletal proteins, kinases, phosphatases, motor proteins, and/or regulatory factors. The first P-CIPs identified using a yeast two-hybrid screen included a protein kinase (P-CIP2), a GDP/GTP exchange factor for Rac1 (Kalirin) (24, 25), and a novel protein involved in endocytosis (P-CIP1) (53).

Phosphorylation is important in the trafficking of a wide variety of membrane proteins. For example, several steps in the trafficking of furin (2, 8, 11, 54), intracellular cycling of the mannose 6-phosphate receptor (2), transcytosis of the polymeric immunoglobulin receptor (55), retention of TGN38 in the trans-Golgi network (56), differences among NMDA subunits (57), and internalization of epidermal growth factor (58) receptors all involve phosphorylation. The cytosolic domain of PAM is phosphorylated at Ser⁹³⁷ in vivo, in addition to phosphorylation at other Ser residues and Thr (39). The cycle of phosphorylation and dephosphorylation at Ser⁹³⁷ is essential for normal trafficking, and mutagenesis of Ser⁹³⁷ to Ala or Asp demonstrates a role for phosphorylation in avoiding delivery of PAM to lysosomes after internalization (39, 59). In addition, the trafficking of both newly synthesized PAM and PAM internalized from the plasma membrane were altered when Thr⁹⁴⁶ and Ser⁹⁴⁹ were mutated to Ala or Asp (50). The role of protein kinases in regulating cytoskeletal function is also well established (43).

We show here that P-CIP2, a 46.5-kDa cytosolic protein kinase, phosphorylates recombinant PAM CD highly selectively in vitro. This specificity, along with the demonstrated co-expression of P-CIP2 and PAM throughout the nervous system, suggests that P-CIP2 contributes to the phosphorylation and targeting of membrane PAM in situ. Phosphorylation of PAM CD by P-CIP2 is limited to Ser residues. Mutation of potential phosphorylation sites within the CD of PAM and phosphorylation of synthetic peptides identified Ser⁹⁴⁹ as the major site phosphorylated by P-CIP2 (Figs. 6-8). The CD of PAM is phosphorylated in vitro by protein kinase C (at Ser⁹³⁷ and Ser⁹³²) as well as by protein kinase A (at Ser⁹²¹), and at Thr⁹⁴⁶ and Ser⁹⁴⁹ by sea star casein kinase II (39). Unlike sea star casein kinase II, P-CIP2 does not phosphorylate Thr residues in PAM-CD. Interestingly, recombinant human casein kinase II does not phosphorylate Ser⁹⁴⁹. Despite the presence of a potential CKII site closer to the COOH terminus of PAM (Ser⁹⁶¹), P-CIP2 is unable to phosphorylate a synthetic peptide that includes this site. P-CIP2 also fails to phosphorylate a peptide that includes a trio of Ser residues near the transmembrane domain (residues 907-909).

A kinase identical to P-CIP2 was identified by Maucuer et al. (26, 60) when using a yeast two-hybrid system to screen for proteins that could interact with stathmin. A number of extracellular signals associated with cell proliferation and differentiation regulate the phosphorylation of stathmin, a cytosolic protein implicated in cell cycle control (61). The ability of stathmin to destabilize microtubules is regulated by its phosphorylation state (62, 63). We confirmed the reported phosphorylation of stathmin by recombinant P-CIP2/KIS from bacterial sources (60) and tested the ability of recombinant P-CIP2 from mammalian cells to phosphorylate stathmin and SCG10, a membrane-associated neuronal stathmin homologue (27). SCG10 and stathmin were not significantly phosphorylated by recombinant P-CIP2 purified from mammalian sources, although both were good substrates for other kinases (Fig. 7). We found that nanogram amounts of purified P-CIP2 from mammalian sources were sufficient for readily detectable kinase activity (Figs. 4-7), while microgram amounts of P-CIP2 from bacterial sources were required, raising the worry that co-purifying minor contaminants were responsible for some of the apparent activity from the bacterial sources (38, 60).

The specificity of P-CIP2 for Ser⁹⁴⁹ of PAM is quite remarkable. P-CIP2 did not phosphorylate significantly the CDs of carboxypeptidase D or furin, integral membrane proteins whose steady state localization and itineraries resemble those of PAM (5, 8, 11, 49, 54). The phosphorylation sites controlling furin trafficking are typical of the acidic sites recognized by casein kinase II (8, 54). P-CIP2 did not phosphorylate the small GTP-binding proteins Rac1, RhoA, or Cdc42, at least in the nucleotide-free form; these were tested because Rac1 interacts with Kalirin, another P-CIP. P-CIP2 did not phosphorylate any of the standard small peptide substrates used for protein kinase A, protein kinase C, or casein kinases (Fig. 8), histone, casein, or recombinant exon A. The striking specificity of P-CIP2 is undoubtedly key to its function (50). While some protein kinases have a rather wide substrate specificity, others (e.g. myosin light chain kinase) have highly restricted substrate specificity (64). Factors regulating the activity of P-CIP2 have not yet been identified. Homomeric interactions of P-CIP2, interactions of P-CIP2 with Kalirin, stathmin, or RNA as well as phosphorylation/dephosphorylation might regulate the kinase activity of P-CIP2.

The protein kinases most closely related to P-CIP2 (Fig. 1) are involved in a number of intracellular signaling cascades, suggesting a signaling role for P-CIP2. Kalirin, one of the other P-CIPs identified in the initial yeast two-hybrid screen, occurs in a form that includes a putative protein kinase domain homologous to the COOH-terminal region of human Trio² (65). Interestingly, the kinase domain of rat P-CIP2 exhibits 20% identity and 28% similarity to the kinase domain of human Trio. The kinase domain of Trio has not yet been shown to phosphorylate any substrates (66, 67). The homology of rat PCIP2 to rat casein kinase II (18% identity and 33% similarity) is interesting in light of the ability of P-CIP2 to phosphorylate a Ser residue in an acidic motif (Fig. 8).

The putative RNA-binding domain of P-CIP2 is not essential to the expression of kinase activity toward the PAM-CD substrate. Only one other kinase, PKR, contains an RNA-binding domain (44). PKR is involved with the interferon antiviral response pathway and appears to be involved in cell growth and differentiation. PKR contains two RNA-binding domains and specifically binds double-stranded RNA; binding of double-stranded RNA to PKR is required for kinase activity (44). The function of the putative RNA-binding domain of P-CIP2 is still unclear. Several possibilities exist. P-CIP2 may bind RNA and localize it within the cell. P-CIP2 may act to stabilize bound RNA. Alternatively, RNA may act to modulate the kinase activity of P-CIP2, as for PKR (44). Presently, it is not known whether P-CIP2 binds a specific RNA sequence or any RNA at all. However, it has recently been reported that a specific 20-nucleotide sequence in the 3'-untranslated region of PAM mRNA forms a complex with a 46-kDa cytosolic protein (68), and several other short nucleotide stretches in 3'-untranslated regions are known to have intracellular routing information (69, 70).

Identification of PAM as a substrate for P-CIP2 was possible only because PAM and P-CIP2 interacted in a yeast two-hybrid screen. Whether P-CIP2 has additional substrates is not yet clear. The determinants governing the interaction of P-CIP2 with the CD of PAM are not limited to the region surrounding Ser⁹⁴⁹, the site phosphorylated by P-CIP2, and involve regions of the CD that are closer to the transmembrane domain. Thus, P-CIP2 interacts with PAM-CD bearing Ala or Asp at position 949 and fails to interact with PAM-CD bearing mutations at Lys⁹¹⁹. PAM-1/933-950, which lacks the site phosphorylated by P-CIP2 can still interact with P-CIP2. As expected for an interaction that may be involved in regulating a dynamic process such as the targeting of membrane proteins, the interaction of PAM with P-CIP2 may be transient; only a small fraction of the PAM protein in cells can be co-immunoprecipitated with P-CIP2 (Fig. 10). This is consistent with that fact that most of the P-CIP2 protein is cytosolic, with a small fraction associated with membranes (Fig. 2) (24). The fact that 70-kDa PAL is present in the co-immunoprecipitate along with full-length 120-kDa PAM indicates that P-CIP2 can interact with PAM after it has entered the immature secretory granules and undergone endoproteolytic cleavage. Recruitment of P-CIP2 to membranes may involve PAM-CD as well as other soluble and membrane proteins.

In conclusion, P-CIP2 is a highly selective cytosolic protein kinase that interacts with membrane PAM in intact cells and phosphorylates the CD of PAM in vitro at Ser⁹⁴⁹. P-CIP2 and PAM clearly have an opportunity to interact, and the region of PAM required for the interaction with P-CIP2 is critical to its routing. Future studies will examine the specific determinants within the PAM-CD which mediate P-CIP2 interaction and factors regulating P-CIP2 localization and kinase activity.

	ACKNOWLEDGEMENTS

We thank Gary Thomas (Vollum Institute, University of Oregon) for providing GST/furin-CD plasmid, Lloyd Fricker (Albert Einstein College of Medicine) for providing GST-carboxypeptidase D plasmid, David Anderson (California Technical Institute) for supplying SCG10 plasmid, Ulrich Schubart (Albert Einstein) for the P19/stathmin plasmid, Richard Cerione (Cornell University) for the GST-Rac1/RhoA/Cdc42 plasmids, Henry Keutmann (Massachusetts General Hospital) for synthetic peptides, and David Ginty (Johns Hopkins University School of Medicine) for comments and suggestions on experiments and the manuscript. We thank Marie Bell and Cathy Caldwell for general laboratory assistance, Lixian Jin for tissue culture, and Greg Galano for assistance in working out the recombinant P-CIP2 purification.

	FOOTNOTES

^* This work was supported by National Institutes of Health Grant DK-32949.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Current address: Dept. of Chemistry, Missouri Western State College, St. Joseph, MO 64507.

^¶ To whom correspondence should be addressed: Dept. of Neuroscience, Rm. 907 WBSB, The Johns Hopkins University School of Medicine, 725 North Wolfe St., Baltimore, MD 21205. Tel.: 410-955-6938; Fax: 410-955-0681; E-mail: dmains@jhmi.edu.

² R. C. Johnson, R. E. Mains, and B. A. Eipper, unpublished results.

	ABBREVIATIONS

The abbreviations used are: TGN, trans-Golgi network; PAM, peptidylglycine -amidating monooxygenase; CD, COOH-terminal cytosolic domain; P-CIP, PAM COOH-terminal interactor protein; GST, glutathione S-transferase; TES, N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid; PIPES, piperazine-N,N'-bis(2-ethanesulfonic acid); PMSF, phenylmethylsulfonyl fluoride; PAGE, polyacrylamide gel electrophoresis; kb, kilobase(s); PAL, peptidyl--hydroxyglycine--amidating lyase.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES