Ancient Ubiquitous Protein 1 Binds to the Conserved Membrane-proximal Sequence of the Cytoplasmic Tail of the Integrin α Subunits That Plays a Crucial Role in the Inside-out Signaling of αIIbβ3 *

Modification of the cytoplasmic tails of the integrin αIIbβ3 plays an important role in the signal transduction in platelets. We searched for proteins that bind to the αIIb cytoplasmic tail using the yeast two-hybrid assay with a cDNA library of the megakaryocyte-derived cell line and identified a protein, ancient ubiquitous protein 1 (Aup1), that is ubiquitously expressed in human cells. Observation of UT7/TPO cells expressing a red fluorescent protein-tagged Aup1 indicated its localization in the cytoplasm. Immunoprecipitation of UT7/TPO cells by an antibody for Aup1 revealed that ∼40% of αIIb is complexed with Aup1. Binding study with an αIIb cytoplasmic tail peptide and glutathioneS-transferase-Aup1 fusion protein revealed a low affinity (K d = 90 μm). Subsequent yeast two-hybrid assay indicated binding of Aup1 to cytoplasmic tails of other integrin α subunits. Binding study with the purified Aup1 and various glutathione S-transferase-αIIbcytoplasmic tail peptides revealed specific binding of Aup1 to the membrane-proximal sequence (KVGFFKR) that is conserved among the integrin α subunits and plays a crucial role in the αIIbβ3 inside-out signaling. As Aup1 possesses domains related to signal transduction, these results suggest involvement of Aup1 in the integrin signaling.

Integrin ␣ IIb ␤ 3 (GPIIb-IIIa) is one of the receptors on the cellular surface of platelets and megakaryocytes. It binds to various adhesive proteins including fibrinogen, von Willebrand factor, vitronectin, and fibronectin that contain a core amino acid sequence of arginine-glycine-aspartic acids (RGD). Binding of fibrinogen to ␣ IIb ␤ 3 leads to platelet aggregation and finally to thrombus formation at the injured vascular sites. A pivotal role of ␣ IIb ␤ 3 in hemostasis is supported by the clinical observation that the congenital deficiency of ␣ IIb ␤ 3 , Glanzmann's thrombasthenia, results in lifelong bleeding tendency (1). Whereas ␣ IIb ␤ 3 on resting platelets does not bind soluble fibrinogen, once platelets are activated, conformation of the extracellular domains of the ␣ IIb ␤ 3 is altered and its ligandbinding affinity is increased (affinity modulation) (2). This process of the inside-out signaling is considered to be mediated by modification of the short cytoplasmic tails of ␣ IIb and ␤ 3 subunits; however, the mechanism remains to be elucidated.
The nuclear magnetic resonance structural analysis of the ␣ IIb cytoplasmic tail revealed a closed conformation where the highly conserved N-terminal membrane-proximal region forms an ␣-helix followed by a turn, and the acidic C-terminal loop interacts with the N-terminal helix (3). Deletion of almost the entire ␣ IIb -cytoplasmic tail and mutations in its N-terminal sequence (GFFKR) conserved among the integrin ␣ subunits enhance the affinity of ␣ IIb ␤ 3 for ligands (4 -6). The cytoplasmic tail of the ␤ 3 subunit also has an amino acid sequence that is conserved among integrin ␤ subunits: a stretch of 8 amino acids (KLLITIHD) adjacent to the transmembrane domain. In a similar fashion to the ␣ IIb subunit, deletion or mutation in this conserved region induces activation of ␣ IIb ␤ 3 (6,7). These observations suggest that membrane-proximal regions of the cytoplasmic domains of both subunits exert a negative regulatory function and lock ␣ IIb ␤ 3 in a low affinity state. Negative regulation may be mediated by the interaction between ␣ IIb and ␤ 3 cytoplasmic tails, possibly through a salt bridge between Arg-995 in ␣ IIb and Asp-723 in ␤ 3 (6), or binding of intracellular proteins to ␣ IIb and/or ␤ 3 subunits. Two candidates for the modulator proteins have been reported: calcium-and integrinbinding protein (CIB) 1 (8) and ␤ 3 -endonexin (9,10), which bind to ␣ IIb and ␤ 3 cytoplasmic tails, respectively. Although CIB is unlikely to have a regulatory effect on ␣ IIb ␤ 3 ligand binding function (11), ␤ 3 -endonexin fused to GST protein induces the conformational change of ␣ IIb ␤ 3 and activates it when co-transfected with ␣ IIb and ␤ 3 subunits in Chinese hamster ovary cells. Another mechanism of modification has been recently suggested: an interaction between cytoplasmic tails of ␣ IIb ␤ 3 and the actin cytoskeleton. ␣ IIb ␤ 3 and the actin cytoskeleton are physically linked by binding of talin to the ␤ 3 cytoplasmic tail (12), and ␣ IIb ␤ 3 in resting platelets may be constrained in a low affinity state by the actin cytoskeleton (13). An increase in the cytosolic calcium evoked by agonist stimulation initiates actin filament turnover and may lead to relief of the cytoskeletal * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In contrast, binding of fibrinogen to ␣ IIb ␤ 3 leads to platelet shape change, release of granules, and platelet aggregation. These sequential biological phenomena are mediated by the outside-in signaling of ␣ IIb ␤ 3 , calcium mobilization, increase in the cytoplasmic pH, thromboxane A 2 generation, and tyrosine phosphorylation of intracellular proteins including focal adhesion kinase and members of the Src family proteins (1). These signaling proteins are complexed with the actin cytoskeleton and are recruited to the focal contacts. In this process, the ␤ 3 cytoplasmic tail (residues 740 -762) binds to the adaptor proteins Shc and Grb2 when tyrosine residues (Tyr-747 and Tyr-759) are phosphorylated (14). In addition, the ␤ 3 cytoplasmic tail is involved in clot retraction by transmitting the contractile force evoked by the rearrangement of the cytoskeletal proteins to the extracellular matrix (15). Thus, several proteins that bind to the ␤ 3 cytoplasmic tail and are implicated in the ␣ IIb ␤ 3 signaling have been identified; however , little is known about proteins that bind to and modify the ␣ IIb cytoplasmic tail.
In this study, we searched for proteins that bind to the ␣ IIb cytoplasmic tail in the thrombopoietin-dependent acute megakaryocytic leukemia-derived cell line, UT7/TPO (16), by the yeast-two hybrid assay and identified a protein, Aup1, that binds to the conserved membrane-proximal sequence of the cytoplasmic tail of the integrin ␣ subunits.
Amplification of the cDNA Sequence for the Cytoplasmic Domain of Integrin ␣ and ␤ Subunits by PCR-The cDNA sequences for the cytoplasmic domains of various integrin ␣ and ␤ subunits were amplified by reverse transcription-PCR from RNA extracted from UT7/TPO for ␣ IIb , ␣ 2 , ␣ V , and ␤ 3 , HepG2 for ␣ 1 , Raji for ␣ 5 , K562 for ␣ M , and HL60 for ␤ 1 and ␤ 2 , respectively. The cDNA sequences for a mutant ␣ IIb (F992A) and for membrane-proximal (␣ IIb MP; KVGFFKR) and membrane-distal (␣ IIb MD; NRPPLEEDDEEGE) segments of the ␣ IIb cytoplasmic tail were amplified using the normal ␣ IIb cytoplasmic tail cDNA. The nucleotide sequence of each cDNA fragment amplified by PCR was confirmed using the ABI Prism dRhodamine terminator cycle sequencing ready reaction kit (Applied Biosystems Japan, Tokyo, Japan).
Yeast Two-hybrid Assays-A cDNA library was constructed by ligating cDNA synthesized from UT7/TPO RNA to a pAD-Gal4 vector (Stratagene, La Jolla, CA). The cDNA sequence for the cytoplasmic domain of the integrin ␣ IIb subunit was ligated in-frame to a pBD-Gal4 vector (Stratagene). Procedures for screening and the filter lift assay to confirm interactions between the bait and target proteins were according to the manufacturer's instructions. Briefly, yeast YRG-2 cells were transformed with a pAD-Gal4 plasmid encoding the UT7/TPO cDNA and a pBD-Gal4 plasmid encoding the ␣ IIb bait. Then, yeast cells were plated on selective SD agar plates without leucine, tryptophan, and histidine (Leu Ϫ Trp Ϫ His Ϫ ). Colonies grown on the selective plates, indicating interactions between target and bait proteins, were subjected to the filter lift assay to examine the ␤-galactosidase activity to confirm the interaction. Positive yeast colonies were transferred to Whatman filter papers, frozen in the liquid nitrogen, thawed, and incubated with 5-bromo-4-chloro-3-indolyl-␤-D-galactoside (0.3 mg/ml) in Z buffer (per liter, 16.1 g of Na 2 HPO 4 /7H 2 O, 5.5 g of NaH 2 PO 4 /7H 2 O, 0.75 g of KCl, 0.246 g of MgSO 4 /7H 2 O, and 2.7 ml of 2-mercaptoethanol; pH 7.0). Colonies that produced blue color were picked up for the subsequent experiments. To examine interactions between the target protein and cytoplasmic tails of various integrin subunits including ␣ IIb , ␣ 1 , ␣ 2 , ␣ 5 , ␣ M , ␣ V , ␤ 1 , ␤ 2 , and ␤ 3 , yeast two-hybrid assays were performed in a similar fashion.
Quantitative ␤-Galactosidase Assay-To compare interactions between the target protein and the cytoplasmic domains of various integrin subunits, the quantitative ␤-galactosidase assay (18) was performed. Briefly, yeast cells grown in 5 ml of medium at 30°C until the near log phase (OD 600 ϭ ϳ1.0) were resuspended in Z buffer (200 l) with glycerol (50 l). After one cycle of freeze-thawing, 1 mM PMSF and acid-washed glass beads were added to the samples, followed by vigorous vortexing. Then, 50 l of o-nitrophenyl-␤-D-galactoside (4 mg/ml, Sigma-Aldrich Japan, Tokyo, Japan) was added to the supernatants and the samples were incubated at 30°C until a yellow color developed. After addition of 120 l of Na 2 CO 3 (1 M), the OD 420 of each sample was measured. Assays were normalized to the yeast concentration (OD 600 ) of each sample, and the ␤-galactosidase activity was calculated as follows; ␤-galactosidase units ϭ 1,000 ϫ OD 420 /t ϫ V ϫ OD 600 , where t ϭ time of incubation in minutes, V ϭ volume of culture added to Z buffer in ml (5 ml).
Northern Blot Analysis-Approximately 30 g of the total RNA extracted from UT7/TPO cells was electrophoresed in 1.5% agarose/formaldehyde gels, transferred to the nylon membranes (Hybond-Nϩ, Amersham Biosciences, Buckinghamshire, United Kingdom), and hybridized with a full-length Aup1 cDNA fragment labeled with [␣-32 P]dCTP using a random-primed DNA labeling kit (Roche Molecular Biochemicals). To compare the expression of Aup1 transcripts among different human tissues, the Human 12-lane MTN Blot (CLON-TECH Japan, Tokyo, Japan) was hybridized with the same probe.
Preparation of the Synthetic Peptides and Antibody Production-Peptides for Aup1 (RLTPADKAEHMKRQRHPRLR) (Fig.1, A and B), ␣ IIb and ␤ 3 cytoplasmic tails, to which a cysteine residue was added at the N terminus for the antibody production, were synthesized using PSSM-8 (Shimadzu, Kyoto, Japan). Each peptide was coupled to the keyhole limpet hemocyanin (Sigma-Aldrich) and injected subcutaneously to rabbits for immunization.
Immunoblot Analysis-Platelets were isolated from the platelet-rich plasma of the normal peripheral blood, and leukocytes were isolated from the buffy coat after removal of erythrocytes by hypotonic lysis in 0.14 M NH 4 Cl, 20 mM Tris (pH 7.2) at 37°C. Microscopic observation revealed that more than 90% of the prepared leukocytes were neutrophils. To extract proteins, platelets, leukocytes, UT7/TPO, and other cell lines including CMK, HL60, K562, U937, Jurkat, Raji, 293, HepG2, HeLa, MCF7, and A547 were resuspended in the cell lysis buffer (0.15 M NaCl, 10 mM Tris (pH 7.4), 1 mM PMSF, 1.8 g/ml aprotinin, 100 g/ml leupeptin, and 1% Triton X-100). Samples were incubated for 30 min on ice with occasional vortexing, and the cell lysates were subjected to 10% SDS-PAGE, transferred to the nitrocellulose membranes (Trans-blot transfer medium, Nippon Bio-Rad Laboratories, Yokohama, Japan). After blocking with 5% skimmed milk in Tris-buffered saline buffer (10 mM Tris (pH 7.4), 150 mM NaCl) for 1 h, membranes were incubated with a preimmune rabbit serum or the rabbit antiserum for the Aup1 peptide (Aup1-2) for 1 h, and then with the horseradish peroxidase-conjugated goat anti-rabbit immunoglobulins (DAKO Japan, Kyoto, Japan). Signals on membranes were detected with the ECL system (Amersham Biosciences).
Subcellular Localization of Aup1-To study the subcellular localization of Aup1, a full-length Aup1 cDNA was ligated into the pDsRed1-N1 vector that encodes a red fluorescent protein (RFP) (CLONTECH). Then, UT7/TPO cells were transfected with a control vector and a plasmid encoding the Aup1-RFP fusion protein by electroporation using the Gene Pulser II (Bio-Rad). After selection with neomycin, stable cell lines that express RFP (UT7/TPO.VR4-4) and Aup1-RFP fusion proteins (UT7/TPO.Aup1 R23-1) were established. To examine the subcellular localization of Aup1, UT7/TPO.VR4-4 and UT7/TPO.Aup1R23-1 cells were applied to glass coverslips and observed using the LSM 510 laser scanning microscope (Carl Zeiss Microscopy, Jena, Germany) with appropriate filters. For the nuclear staining, cells were treated with 3.7% formaldehyde in phosphate-buffered saline and mounted with a medium containing 4Ј,6Ј-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA).
Immunoprecipitation-UT7/TPO cell extracts (600 g of protein) with the cell lysis buffer containing 1% digitonin and 1 mM Ca 2ϩ were incubated with a preimmune rabbit serum or the Aup1-2 antiserum (40 l) for 1 h, and then with protein-G-Sepharose beads (30 l) (Amersham Biosciences) for 3 h at 4°C with gentle shaking. After washing with the cell lysis buffer, the beads were resuspended in the SDS sample buffer and boiled, and the supernatants were subjected to SDS-PAGE, followed by immunoblot analysis using mouse monoclonal antibodies for the ␣ IIb (SZ22, Cosmo Bio, Tokyo, Japan) and ␤ 3 subunits (SZ21, Cosmo Bio). Signals were detected with the ECL system after incubation with the horseradish peroxidase-conjugated rabbit anti-mouse immunoglobulins (DAKO). To measure how much part of the cellular ␣ IIb is complexed with Aup1, the immunodepletion assay was performed. After preincubation with protein-G beads, UT7-TPO cell lysates (500 g of protein) were immunoprecipitated twice with the control rabbit or the Aup1-2 serum (40 l) and protein-G beads (50 l). Then, aliquots of the supernatants (1-10 l) were subjected to SDS-PAGE, followed by immunoblot analysis using the polyvinylidene difluoride (PVDF) membrane (Immobilon-P, Millipore, Bedford, MA) and the SZ22 antibody. Concentration of the residual (free) ␣ IIb subunit was quantified by densitometry of the respective ␣ IIb bands to calculate the percentage of ␣ IIb bound to Aup1.
GST Pull-down Assay-The cDNA sequences for Aup1, the normal and a mutant (F992A) ␣ IIb cytoplasmic tails, and the membrane-proximal (␣ IIb MP) and membrane-distal segments (␣ IIb MD) were ligated in-frame to the pGEX-4T vector that encodes GST (Amersham Biosciences). For production of GST fusion proteins, DH5␣ cells transfected with the respective plasmids were incubated at 25°C until the culture reached the mid-log phase (OD 600 ϭ ϳ0.8), at which time 25 M isopropyl-␤-D-thiogalactopyranoside was added. After incubation for an additional 2 h, the bacteria were resuspended in phosphate-buffered saline containing 1 mM PMSF and 1 mg/ml lysozyme, followed by sonication. Triton X-100 was then added (1%), and the supernatant was incubated with the GSH-Sepharose beads (Amersham Biosciences) for 30 min at room temperature, followed by washing with phosphate-buffered saline. To examine association between Aup1 and cytoplasmic tails of ␣ IIb and ␤ 3 , beads that bound the GST-Aup1 fusion protein (125 g) and the control GST protein (62.5 g) were incubated with the UT7/TPO cell extract (1 mg of protein) in the GST reaction buffer (200 l, the cell lysis buffer containing 1% Triton X-100) for 3 h at 4°C with gentle shaking. Thereafter, beads were washed with the GST reaction buffer, resuspended in the SDS sample buffer, and boiled, and the supernatant was subjected to SDS-PAGE, followed by immunoblot analysis using the PVDF membrane, the rabbit antiserum for the cytoplasmic tails of ␣ IIb and ␤ 3 , and the horseradish peroxidase-conjugated goat anti-rabbit immunoglobulins. To identify the ␣ IIb sequence to which Aup1 binds, Aup1 cleaved from the immobilized GST-Aup1 fusion protein by thrombin (Sigma-Aldrich) according to the manufacturer's instructions was preincubated with GSH-and GST-Sepharose beads in the GST reaction buffer. Then, the supernatants (50 l containing 0.7 g of Aup1; 0.4 M) was incubated with the immobilized GST-␣ IIb cytoplasmic tail fusion proteins (20 M), including the normal and a mutant (F992A) ␣ IIb , ␣ IIb MP, and ␣ IIb MD segments, followed by SDS-PAGE and immunoblot analysis using the PVDF membrane and the Aup1-2 antiserum.
Estimation of the Affinity of Interaction between Aup1 and the ␣ IIb Cytoplasmic Tail-The synthetic ␣ IIb cytoplasmic peptide was preincubated with the GST-Sepharose beads for 30 min at room temperature, and the supernatants (20 -720 M) were incubated with the immobilized GST-Aup1 fusion protein (1 g, 1.5 M) for 1 h at room temperature in the GST reaction buffer (10 l). Then, various amounts of the supernatants, the original ␣ IIb peptide, and the peptide preabsorbed by GST beads were subjected to 15% SDS-PAGE, followed by immunoblot analysis using the PVDF membrane and the rabbit antiserum for the ␣ IIb cytoplasmic tail. The resulting immunoblots of the ␣ IIb cytoplasmic tail were quantified by densitometry, and the affinity of Aup1 for binding to the ␣ IIb cytoplasmic tail was calculated by Scatchard analysis (19).

Identification of a Protein That Binds to the Integrin ␣ IIb
Cytoplasmic Tail-To search for proteins that bind to the cytoplasmic tail of the integrin ␣ IIb subunit, we constructed a cDNA library from a megakaryocyte-derived cell line, UT7/ TPO. UT7/TPO cells constitutively express ␣ IIb ␤ 3 on the cellular surface, as confirmed by fluorescence-activated cell sorting analysis (data not shown). Screening of the cDNA library by the yeast two-hybrid assay with a bait of the ␣ IIb cytoplasmic tail identified a 900-bp cDNA fragment that encodes a partial Cterminal peptide composed of 173 amino acids (48L21) (Fig.  1A). To obtain a cDNA sequence for the N-terminal portion, we performed PCR using UT7/TPO cDNA and primers from the 5Ј-terminal sequence of the cloning site of pAD-Gal4 (5Ј-AGG-GATGTTTAATACCACTAC-3Ј) and the 5Ј-terminal sequence of the 48L21 (5Ј-GCTGTTGTACACGGAGTGCA-3Ј). A 400-bp cDNA fragment (5Ј-48L21) was amplified, and the nucleotide sequencing revealed the sequence was continuous to the 5Јterminus of 48L21. For the final cloning of a full-length cDNA, PCR with UT7/TPO cDNA and primers from the 5Ј-terminal sequence of 5Ј-48L21 (5Ј-TGCGCCTGGGCGCGAAAATG-3Ј) and 3Ј-terminal sequence of 48L21 (5Ј-GGCTCTGGGTGC-CATCCTGT-3Ј) was performed. Nucleotide sequencing of the amplified PCR product (1.3-kb cDNA fragment) revealed that the protein was composed of 410 amino acids. Subsequent data base searches indicated that the sequence was identical with the short isoform of Aup1 (Refs. 20 and 21; GenBank TM accession no. AF100753) (Fig. 1, A and B).
Aup1 Is Ubiquitously Expressed in Human Cells and Tissues-Northern blot analysis with the total RNA extracted from UT7/TPO cells using a full-length Aup1 cDNA probe revealed a transcript of ϳ1.7 kb (Fig. 2A). Because it was reported that Aup1 is expressed in all mouse tissues (20), we examined the expression of Aup1 transcripts in various human tissues. In concordance with the mouse tissues, Aup1 was expressed in all human tissues examined (Fig. 2B). To examine the expression of the Aup1 protein in UT7-TPO cells, platelets, leukocytes, and other cell lines, a rabbit antiserum was raised against a synthetic peptide for Aup1 (Fig. 1, A and B). Immunoblot analysis using this antiserum (Aup1-2) revealed duplicate bands of ϳ40 kDa at the reducing as well as non-reducing conditions in UT7/TPO cells (Fig. 3A). These bands were observed in other cell lines including CMK, HL60, K562, U937, Jurkat, Raji, HepG2, 293, HeLa, MCF7, and A547 (Fig. 3B). Treatment with the protein phosphatases did not change intensity of these two bands (data not shown). In contrast, only a smaller band was detected in platelets and leukocytes (Fig. 3C).
Aup1 Is Present in Cytoplasm-To examine the subcellular localization of Aup1, stable UT7/TPO cell lines that express the Aup1-RFP fusion protein (UT7/TPO.Aup1R23-1) and the control RFP (UT7/TPO.VR4-4) were established. Overexpression of Aup1 did not affect the expression of ␣ IIb ␤ 3 on the cellular surface, as confirmed by fluorescence-activated cell sorting analysis of UT7/TPO, UT7/TPO.VR4-4, and UT7/TPO.Au-p1R23-1 cells (data not shown). Observation of UT7/T-PO.VR4-4 and UT7/TPO.Aup1 R23-1 cells by confocal microscopy revealed that RFP was distributed evenly throughout the cell; however, the Aup1-RFP fusion protein was observed in the cytoplasm, but not in the nucleus (Fig. 4). Because it was reported that the N terminus of mouse Aup1 resembles the signal peptide of secreted protein, followed by a putative signal cleavage site, we examined whether Aup1 is secreted from cells. Immunoblot analysis with the culture supernatant of UT7/TPO cells using the Aup1-2 antiserum revealed that Aup1 could not be detected in the concentrated (10-fold) culture supernatant (data not shown). These results indicate that Aup1 is a cytoplasmic protein.
Approximately 40% of the ␣ IIb Subunit Is Complexed with Aup1 in UT7/TPO Cells-To examine whether Aup1 is associated with the ␣ IIb cytoplasmic tail in the eukaryotic cells, UT7-TPO cell lysate was immunoprecipitated with the Aup1-2 antiserum. Immunoblot analysis with the precipitates using anti-␣ IIb and -␤ 3 antibodies indicated that Aup1 bound to the Although RFP is distributed evenly throughout the cell (a and c), the Aup1-RFP fusion protein was observed in the cytoplasm without localization in the nucleus (b and d). White bars represent 10 m. ␣ IIb , but not to the ␤ 3 subunit (Fig. 5A). We then measured how much of the cellular ␣ IIb subunit is complexed with Aup1 by the immunodepletion assay. Densitometric analysis of the resulting ␣ IIb bands with the supernatants of the UT7-TPO cell lysate after immunoprecipitation with the Aup1-2 and the control rabbit serum revealed that 41.7 Ϯ 3.2% (mean Ϯ S.D., results from three independent experiments) of the ␣ IIb subunit was complexed with Aup1 (data not shown). Binding of Aup1 to the ␣ IIb cytoplasmic tail was confirmed by the GST pull-down assay, revealing that GST-tagged Aup1 binds to the ␣ IIb , but not to the ␤ 3 subunit (Fig. 5B).
Aup1 Interacts with the ␣ IIb Cytoplasmic Tail with a Low Affinity-We next studied interaction between the synthetic peptide for the ␣ IIb cytoplasmic tail and immobilized GST-Aup1 fusion protein to measure the affinity of interaction. The K d value calculated from the Scatchard plot analysis was 90 M, suggesting a relatively weak affinity of interaction between Aup1 and the ␣ IIb cytoplasmic tail (Fig. 6).
Aup1 Binds to Cytoplasmic Tails of Various Integrin ␣ Subunits-As Aup1 is expressed ubiquitously in human cells and tissues, we examined whether Aup1 binds to cytoplasmic tails of other integrin ␣ as well as ␤ subunits by the yeast two-hybrid assays. Yeast cells were co-transformed with plasmids encoding Aup1 (48L21) and cytoplasmic tails of ␣ IIb , ␣ 1 , ␣ 2 , ␣ 5 , ␣ V , ␣ M , ␤ 1 , ␤ 2 , and ␤ 3 . In selective Leu Ϫ Trp Ϫ His Ϫ plates, only colonies co-transformed with Aup1 and ␣ subunits grew. These colonies were positive for both of the filter lift assay and the quantitative ␤-galactosidase assay (Fig. 7). These results indicate that Aup1 binds to cytoplasmic tails of various integrin ␣ subunits.
Aup1 Binds to the Conserved Membrane-proximal Sequence of the Cytoplasmic Tail of the Integrin ␣ Subunits-The amino acid sequence of the membrane-proximal region of the cytoplasmic tail is highly conserved among the integrin ␣ subunits and plays a crucial role in the inside-out signaling of ␣ IIb ␤ 3 (4 -6).
As the results from the yeast two-hybrid assay suggested binding of Aup1 to this conserved sequence, we examined interaction between the purified Aup1 and immobilized GST fusion proteins of the ␣ IIb cytoplasmic tail, including the normal and a mutant (F992A) ␣ IIb that leads to the high affinity state of FIG. 5. Interaction between Aup1 and the cytoplasmic tail of the integrin ␣ IIb subunit. A, immunoprecipitation. UT7/TPO cell lysate (600 g of protein) coprecipitated with a preimmune rabbit serum (lanes 1-3) and an antiserum for Aup1 (Aup1-2) (lanes 4 -6), and the control lysate (30 g, lanes [7][8][9] were probed with the control mouse immunoglobulin (lanes 1, 4, and 7), the mouse monoclonal antibodies for ␣ IIb (SZ22) (lanes 2, 5, and 8) and ␤ 3 (SZ21) subunits (lanes 3, 6, and 9). Arrowheads indicate the ␣ IIb subunit. Filters (lanes 4 -6) were reprobed with the antiserum for Aup1 (Aup1-2) (strips below lanes 4 -6). B, GST pull-down assay. Control GST protein (lanes 1-3) and the GST-Aup1 fusion protein (lanes 4 -7) that bound to GSH-Sepharose beads were incubated with UT7/TPO cell lysates. These samples (lanes 1-7) and the UT7/TPO cell lysate (100 g of protein, lanes 8 -10) were probed with a preimmune rabbit serum (lanes 1, 4, and 8), rabbit antiserum against ␣ IIb (lanes 2, 5, and 9) and ␤ 3 cytoplasmic tails (lanes 3, 7, and  10). To examine competition, the ␣ IIb cytoplasmic tail peptide (0.8 M) was added to the antiserum against the same peptide (lane 6). Arrowheads indicate ␣ IIb subunits. Strips were reprobed with the goat antiserum for GST (Amersham Biosciences); GST (strips below lanes 1-3) and GST-Aup1 fusion protein (strips below lanes 4 -7). 6. Estimation of the affinity of interaction between Aup1 and the ␣ IIb cytoplasmic tail. A, various amounts of the synthetic peptide for the ␣ IIb cytoplasmic tail were incubated with the immobilized GST-Aup1 fusion protein, followed by SDS-PAGE and immunoblot analysis with the antiserum for the ␣ IIb cytoplasmic tail. The amounts of the ␣ IIb peptide bound to Aup1 were quantified by densitometry. B, Scatchard plot of the same data was linear, giving K d value of 90 M.

FIG. 7.
Quantitative ␤-galactosidase assay indicating associations between Aup1 and cytoplasmic tails of the integrin ␣ subunits. Interactions between Aup1 and cytoplasmic tails of various integrin ␣ and ␤ subunits in the yeast two-hybrid assay are presented as activity of ␤-galactosidase. Yeast cells transfected with Aup1 (Aup1), Aup1 with the control pBD-Gal4 vector (pBD), Aup1 with ␣ subunits (␣ IIb , ␣ 1 , ␣ 2 , ␣ 5 , ␣ V , and ␣ M ), and Aup1 with ␤ subunits (␤ 1 , ␤ 2 , and ␤ 3 ), were grown in Leu Ϫ , Leu Ϫ Trp Ϫ , Leu Ϫ Trp Ϫ His Ϫ , and Leu Ϫ Trp Ϫ selective medium, respectively. Bars represent the mean Ϯ S.D. of three separate experiments, each performed on five independent colonies. ␣ IIb ␤ 3 (6), and the membrane-proximal (KVGFFKR) and membrane-distal (NRPPLEEDDEEGE) sequences. The GST pulldown assay revealed binding of Aup1 to the normal and the membrane-proximal ␣ IIb sequence, but neither to the mutant nor to the membrane-distal sequence (Fig. 8). These results indicate that Aup1 binds to the specific short amino acid stretch that is located at the membrane-proximal region and conserved among the cytoplasmic tails of the integrin ␣ subunits.

DISCUSSION
One of the obstacles to elucidate the integrin ␣ IIb ␤ 3 signaling is the absence of appropriate cell lines that exhibit similar properties to platelets including expression of ␣ IIb ␤ 3 and intracellular signaling molecules, and the response to platelet physiological agonists. Because the ␣ IIb subunit is exclusively expressed in platelets and megakaryocytes, it is suggested that a megakaryocyte-derived cell line would be suitable to search for proteins that bind to the ␣ IIb cytoplasmic tail. In this study, we performed the yeast two-hybrid assay and screened the cDNA library from UT7/TPO cells using the ␣ IIb cytoplasmic tail as bait. Binding of Aup1 to the ␣ IIb cytoplasmic tail was demonstrated by the following results. First, Aup1 bound to the ␣ IIb cytoplasmic tail in the yeast two-hybrid assay. Second, Aup1 is a cytoplasmic protein, as indicated by the observation with the confocal microscopy of UT7/TPO cells that express an RFPtagged Aup1. Third, the ␣ IIb subunit was present in the immunoprecipitate of the UT7-TPO cell lysate by the antiserum for Aup1. Fourth, GST-tagged Aup1 bound to the ␣ IIb subunit in the UT7-TPO cells.
The N terminus of Aup1 is hydrophobic and resembles the signal sequence of secreted proteins, followed by an 11-amino acid sequence with similarity to a prokaryotic lipid attachment site (20) (Fig. 1B). These characteristic amino acid sequences and the present observations with confocal microscopy and immunoblot analysis with the culture supernatant of UT7/TPO cells suggest that Aup1 is a cytoplasmic protein in possible association with the plasma membrane. Immunoblot analysis with the antiserum for an Aup1 peptide revealed duplicate bands of ϳ40 kDa with an estimated molecular mass difference of ϳ3-4 kDa in UT7/TPO and other cell lines. In contrast, only a smaller band was detected in platelets and leukocytes. Although only a single cDNA fragment encoding 410 amino acids was amplified by PCR with UT7/TPO cDNA in the present study, a long isoform of Aup1 composed of 476 amino acids has been reported to be produced by alternative splicing of the Aup1 gene (21). However, considering the difference in the number of amino acids (66) between these two isoforms, it is unlikely that duplicate bands observed in the immunoblot analysis are produced by alternative splicing. Another possible examination is the posttranslational modification including glycosylation, phosphorylation, and cleavage. As the consensus amino acid sequences for the O-and N-linked glycosylation sites are absent in Aup1 and only a smaller band is observed in terminally differentiated platelets and leukocytes, it is hard to explain that the larger protein represents a glycosylated mature protein. With regard to phosphorylation, tyrosine, serine, and/or threonine residues are phosphorylated upon cellular stimulation as observed in a number of intracellular signaling proteins. However, modification by phosphorylation is unlikely because duplicate bands were constitutively expressed and did not exhibit any difference in their intensity after treatment with the protein phosphatase in the immunoblot analysis. On the other hand, the difference in the estimated molecular mass of these two bands (ϳ3-4 kDa) is concordant with that of the postulated signal sequence composed of 37 amino acids. Accordingly, it is conceivable that a larger band represents a precursor protein subjected to cleavage to the smaller mature protein, followed by possible modification with lipid attachment including myristoylation, prenylation, and/or palmitoylation to be associated with the internal leaflet of the plasma membrane (22).
It was reported that mouse Aup1 also consists of 410 amino acids and is expressed in all mouse tissues. In addition, it exhibits an amino acid sequence similar to those of Caenorhabditis elegans and human Aup1 (20). Because of its evolutionary conservation of the amino acid sequence and ubiquitous expression, it appears that Aup1 plays an essential role in cell biology. It was unexpected that Aup1, a ubiquitously expressed protein in various tissues, binds to the cytoplasmic tail of the ␣ IIb subunit that is exclusively expressed in platelets and megakaryocytes. Accordingly, we examined whether Aup1 binds to the cytoplasmic tails of other integrin ␣ as well as ␤ subunits. The yeast two-hybrid assays revealed that Aup1 binds to the cytoplasmic tails of the ␣ 1 , ␣ 2 , ␣ 5 , ␣ M , and ␣ V subunits, but not to the ␤ 1 , ␤ 2 , and ␤ 3 subunits, indicating specific binding of Aup1 to the integrin ␣ subunits. To confirm association between Aup1 and these ␣ subunits, we performed immunoprecipitation using cell lines that express a relatively high level of these ␣ subunits, including IMR32 treated by retinoic acid for ␣ 1 , CCRF-CEM for ␣ 2 , HeLa treated by IL-6 for ␣ 5 , K562 for ␣ M , and RAW264.7 for ␣ V . However, we could co-precipitate Aup1 with these integrin ␣ subunits neither by the Aup1-2 nor by various antiserum for these subunits, probably because the expression level of these proteins is extremely low compared with the ␣ IIb subunit in UT7/TPO cells (data not shown). On the other hand, subsequent GST pull-down assay indicated binding of Aup1 to the membrane-proximal amino acid sequence (KVGFFKR) that is conserved among the cytoplasmic tails of the integrin ␣ subunits. Accordingly, it seems that one of the essential biological functions of Aup1 is to bind to the cytoplasmic tail of the integrin ␣ subunits through the conserved membrane-proximal sequence.
A data base search for the homologous domain structure revealed that Aup1 possesses two domains: CUE and PlsC domains (23). The yeast protein Cue1p is a prototype of CUE domain family and belongs to the integral endoplasmic reticulum membrane proteins. It exhibits a scaffolding activity and recruits the ubiquitin-conjugating enzymes Ubc7p and Ubc6p in the proximity of the translocon pore cytoplasmic exit to deliver proteins for ubiquitination and subsequent digestion by the proteasome (24). The CUE domain is also present in several eukaryotic cytoplasmic proteins. It was suggested that some of the CUE-containing proteins are not associated with endoplasmic reticulum and possess functions different from that of Cue1p (23). Recent studies identified two eukaryote proteins with the CUE domain, Toll- interacting protein (Tollip) and transforming growth factor ␤-activated kinase1 (TAK1)-binding protein 2 (TAB2), that exhibit novel functions in the IL-1 signal transduction pathway. Tollip is present in a complex with the serine/threonine IL-1 receptor (IL-1R)-associated kinase (IRAK) and binding of IL-1 to IL-1R results in the rapid assembly of a membrane-proximal signaling complex that consists of IL-1R, an adaptor protein (myeloid differentiation protein; MyD88), IRAK, and Tollip. Because overexpression of Tollip results in impaired IL-1␤-induced activation of the nuclear transcription factor B and c-Jun N-terminal kinase, it may inhibit IL-1 signaling by silencing components of the signaling cascade including IRAK (25). TAB2 is an adaptor protein that mediates activation of TAK1. IL-1 stimulates translocation of TAB2 from the membrane to the cytosol where it mediates association of TAK1 with the tumor necrosis factor receptorassociated factor 6 (TRAF6), leading to activation of TAK1 (26).
In addition to possessing the CUE domain, the amino acid sequence of Aup1 exhibits a significant similarity to taffazins that belong to the acyltransferase superfamily (27,28), suggesting that Aup1 may exhibit an enzymatic activity. Taffazins are composed of a highly hydrophobic N terminus of 30 amino acids that may serve a membrane anchor and a central hydrophilic domain composed of 72 residues that may serve as an exposed loop interacting with other proteins. Mutations of a gene encoding taffazins (G4.5) lead to a severe inherited (Xlinked) disorder, Barth syndrome, that is characterized by cardiac and skeletal myopathy, short stature, and neutropenia, indicating an essential biological function of taffazins (29). Acyltransferases of the tafazzin superfamily all function in phospholipid synthesis and have either glycerophosphate (GPAT, EC 2.3.15), 1-acylglycerophosphate (AGPAT, EC2.3.1.51), 2-acylglycerophosphate, or 2-acylglycerophosphoethanolamine acyltransferase activity (28). The initial step of phospholipid biosynthesis involves the acylation of glycerol-3-phosphate by GPAT to form lysophosphatidic acid (LPA), followed by acylation of LPA by AG-PAT to form phosphatidic acid (PA). In addition to being the key intermediates in the phospholipid biosynthesis, PA and LPA are the essential lipid messengers in signal transduction. Thrombin stimulation leads to production of LPA, followed by its extracellular release in platelets. Binding of LPA to its G protein-coupled receptor leads to stimulation of phospholipases C and D, inhibition of adenylyl cyclase, activation of Ras and the downstream Raf/mitogen-activated protein kinase pathway, and tyrosine phosphorylation of focal adhesion proteins accompanied by remodeling of the actin cytoskeleton in the integrin signaling pathway (30). In contrast, PA can act as an intracellular as well as an extracellular messenger, activating phospholipase C and the numerous protein kinases involved in the signal transduction of the protein kinase C and Raf/mitogen-activated protein kinase pathways (31,32). Moreover, fibrinogen binding to ␣ IIb ␤ 3 incorporated into PA-containing lipid vesicles is enhanced, indicating that PA can modulate affinity of ␣ IIb ␤ 3 within a membrane environment (33). Implications in signal transduction have also been reported both with GPAT and AGPAT; insulin and epidermal growth factor activate GPAT and increase de novo PA synthesis, which may amplify diacylglycerolprotein kinase C signaling (34). Stimulation with IL-1␤ increases AGPAT activity and leads to an enhanced transcription and synthesis of tumor necrosis factor-␣ and IL-6, suggesting AGPAT may amplify cellular signaling responses from cytokines (32).
Taken together, it is conceivable that Aup1 is involved in the integrin signaling. Binding of Aup1 to the cytoplasmic tail of the integrin ␣ subunits may alter interactions between ␣ and ␤ cytoplasmic tails, including interference of the salt bridge formation as predicted between Arg-995 in ␣ IIb and Asp-723 in ␤ 3 . Otherwise, Aup1 may function as an adaptor protein by its CUE domain and recruit signaling molecules to integrin cyto-plasmic tails, or exert a phosphate acyltransferase activity by its PlsC domain, leading to alternation in the local concentrations of PA and LPA. Consequently, conformation of the integrin extracellular domains may be altered (inside-out signaling), or the sequential biological phenomenon evoked by ligand binding may be modulated (outside-in signaling). The GST pull-down assay indicated binding of Aup1 to the conserved membrane-proximal sequence of the integrin ␣ subunits. As deletions or mutations in this region lead to an increase in the affinity of ␣ IIb ␤ 3 for ligands (4 -6) and a mutation in this region (F992A) prevents binding of Aup1, it is conceivable that binding of Aup1 to this sequence may sustain integrin in a low affinity state. In the inside-out signaling of platelets, thrombin stimulation leads to the rapid activation of tyrosine kinases, including Syk, which is activated within seconds (35), and Src family kinases, and tyrosine phosphorylation of the signaling proteins (36). Considering the remarkable rapidity of this signaling process, Aup1 which binds to the ␣ IIb cytoplasmic tail reversibly as suggested by the relatively low affinity of interaction, may be suitable for one of the modulators in the ␣ IIb ␤ 3 inside-out signaling. However, further studies are necessary to elucidate implication of Aup1 in the integrin signaling and other biological functions.